Thesis

UvA-DARE (Digital Academic Repository)
Volcanic ash soils in Andean ecosystems : unravelling organic matter distribution and
stabilisation
Tonneijck, F.H.
Link to publication
Citation for published version (APA):

Tonneijck, F. H. (2009). Volcanic ash soils in Andean ecosystems : unravelling organic matter distribution and
stabilisation Amsterdam: Universiteit van Amsterdam
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),
other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating
your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask
the Library: http://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,
The Netherlands. You will be contacted as soon as possible.
UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)
Download date: 30 Oct 2017

Femke H. Tonneijck Volcanic ash soils in Andean ecosystems
Femke H. Tonneijck
distribution and stabilisation
Unravelling organic matter
in Andean ecosystems
Volcanic ash soils
Graag nodig ik u uit voor het bijwonen van

de openbare verdediging van mijn proefschrift
VOLCANIC ASH SOILS IN ANDEAN ECOSYSTEMS
UNRAVELLING ORGANIC MATTER

DISTRIBUTION AND STABILISATION
Vrijdag 6 februari 2009 10.00 in de Agnietenkapel

Oudezijds Voorburgwal 231 Amsterdam (tramhalte Spui)
Ook op de receptie na afloop bent u van harte welkom
Femke Tonneijck
Volcanic ash soils
in Andean ecosystems

Unravelling organic matter
distribution and stabilisation

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor
aan de Universiteit van Amsterdam
op gezag van de Rector Magnificus
prof. dr. D.C. van den Boom
ten overstaan van een door het
college voor promoties ingestelde commissie,
in het openbaar te verdedigen in de Agnietenkapel
op vrijdag 6 februari 2009, te 10:00 uur

door

Femke Harriët Tonneijck

geboren te Utrecht
Promotiecommissie

Promotor Prof. dr. J.M. Verstraten

Co‐promotor Prof. dr. J. Sevink

Overige leden Prof. dr. ir. W. Bouten
dr. P. Buurman
Prof. dr. H. Hooghiemstra
dr. A.G. Jongmans
Prof. dr. J. van der Plicht

Faculteit der Natuurwetenschappen, Wiskunde en Informatica

Tonneijck, F.H.

Volcanic ash soils in Andean ecosystems – Unravelling organic matter distribution and stabilisation/
F.H. Tonneijck
Thesis Universiteit van Amsterdam – With ref. – With summaries in English, Dutch and Spanish.

ISBN: 978‐90‐9023803‐6

Printed by Koninklijke Wöhrmann, Wöhrmann Print Service, Zutphen
Cover design Jan van Arkel, Amsterdam
Design Marjo Oligschläger, Utrecht

This research was carried out at the Centre for Geo‐ecological research (ICG) and the Institute for Biodiversity
and Ecosystem Dynamics (IBED), Earth Surface Processes and Materials (ESPM), Faculty of Science,
Universiteit van Amsterdam, The Netherlands. This research was funded by NWO/WOTRO (WAN 75‐405).

Voor Kasper

Index

Introduction 7

1 Description of the research area 11

2 Radiocarbon dating of soil organic matter fractions in Andosols in 19
northern Ecuador

3 Tephra stratification of volcanic ash soils in northern Ecuador 39

4 The influence of bioturbation on the vertical distribution of soil 67
organic matter in volcanic ash soils: a case study in northern Ecuador

5 Carbon stocks and stabilisation mechanisms in volcanic ash soils in 89
natural Andean ecosystems of northern Ecuador

6 Modelling vertical organic matter distribution in a volcanic ash soil 109
in the Ecuadorian páramo: impact of soil volume change

Synthesis 129

References 137

Summary 151

Samenvatting 155

Resumen 159

Appendix I List of RUFLE publications 163

Appendix II Soil profile descriptions 164

Appendix III Soil properties 174

Dankwoord 179

Introduction

Framework

The Ecuadorian Andes harbours fragile tropical montane forest and grassland (páramo)
ecosystems of outstanding biodiversity and a high degree of endemism, making them
globally leading biodiversity ‘hot spots’ of key importance for conservation (Myers et al.
2000; Keese et al. 2007). Moreover, both ecosystems are highly valued for their supply and
storage of water and protection against soil erosion (Hofstede 1995a; Poulenard et al. 2001;
Keese et al. 2007). While current montane forests mostly occur as scattered patches, they
are believed to once have covered much larger stretches of land up to much higher altitudes
(Ellenberg 1979; Laegaard 1992; Wille et al. 2002). The reduction in montane forest cover is
often ascribed to human interference such as deforestation, grazing and burning (Ellenberg
1979; Laegaard 1992; Wunder 1996; Wille et al. 2002). Currently, Ecuadorian forests suffer
the highest annual rate of deforestation in the whole of South America (FAO 2006) and
deforestation in montane areas has only slowed down because of the inaccessibility of the
scarce forests left (Wunder et al. 1996; Keese et al. 2007). Similarly, the páramo ecosystem
is seriously degrading due to grazing and burning (Hofstede 1995a; López‐Sandoval 2004).
These land‐use changes affect soil formation.

Because soil organic matter (SOM) constitutes the largest pool of terrestrial organic carbon,
soils are potential sources or sinks of the greenhouse gas CO2 that are of global importance
(Eswaran et al. 1993; Batjes 1996; Swift 2001; Lal 2004). In international agreements such as
the ‘Kyoto Protocol’ (Dumanski 2004; IPCC 2007), this has clearly been acknowledged.
Whether soils become a net carbon source or sink with climate and/or land‐use change
depends on the stability of SOM against decomposition. This stability is still not fully
understood, as it is influenced by the composition of SOM itself, by its protection against
decomposition through various soil constituents and by environmental conditions (Sollins
1996; Baldock & Skjemstad 2000). Moreover, SOM stability may change with soil depth
(Jobbágy & Jackson 2000). More detailed knowledge on the mechanisms of stabilisation and
vertical distribution of SOM is therefore urgently needed to accurately estimate potential
carbon gains or losses from soils with climate and/or land‐use change (Jobbágy & Jackson
2000; Swift 2001; Dumanski 2004; Lal 2004). Volcanic ash soils, the dominant soil type in the
high Ecuadorian Andes, contain very large stocks of SOM per unit area (Eswaran et al. 1993;
Batjes 1996; Lal 2004). Therefore, the impact of climate change and land‐use change on
SOM in volcanic ash soils may be particularly strong (Veldkamp 1994; Rhoades et al. 2000).
For a proper assessment of these impacts on volcanic ash soils it is vital to understand their
functioning under undisturbed natural conditions.
7
Over the last few decades, both national and international incentives have been developed
to promote forestation activities in the high Andean region of Ecuador (Morris 1997; Neira
et al. 2006; Farley 2007). Since the position of the upper forest line (UFL) in the absence of
human interference is still subject to scientific debate (Wille et al. 2002; López‐Sandoval
2004; Farley 2007; Di Pasquale et al. 2008), ecologically sustainable reforestation in the
montane forest and páramo ecosystem zone is severely hindered. To elucidate the current
natural position of the UFL, knowledge of past UFL dynamics is a prerequisite because
currently existing UFLs in Ecuador are mostly artificial. Precisely because of their
characteristic accumulation of organic matter (including paleoecological proxies) in
combination with their stratigraphic nature, volcanic ash soils provide great opportunities
for paleoecological reconstructions of their vegetation (Bakker & Salomons 1989). However,
the use of paleorecords contained in terrestrial soils is a challenge because stratification of
SOM is not straightforward due to redistributing soil forming processes, notably
bioturbation and leaching (Dimbleby 1957; Havinga 1963; Keatinge 1983; Kelso 1994;
Davidson et al. 1999). Nevertheless, a major advantage of analysing paleorecords in these
soils is that the investigation is not restricted to the accidental peat or sedimentary deposit
traditionally used in paleoecological analyses (Dimbleby 1957). Thus, paleorecords
contained in volcanic ash soils potentially enhance the spatial resolution of paleoecological
investigations and may greatly increase understanding of UFL dynamics.

The ‘Reconstruction of the Upper Forest Line in Ecuador’ ‐ or ‘RUFLE’ ‐ project investigates
the UFL dynamics in the Ecuadorian Andes during the Holocene to clarify what the natural
location of the UFL would have been in the absence of human interference. RUFLE
furthermore aims to increase understanding of SOM distribution and stabilisation in
volcanic ash soils under (semi‐) natural conditions. In this way, RUFLE supports conservation
and sustainable management of the tropical montane forest and páramo ecosystems and
explores their role in the global carbon cycle. The multidisciplinary RUFLE project is
organised in three integrated projects that investigate in short: 1) the current vegetation,
and the past vegetation as derived from fossil pollen records contained in peat bogs and
volcanic ash soils; 2) the past vegetation as derived from fossil biomarker (plant specific
organic molecules) records preserved in peat bogs and volcanic ash soils and 3) the vertical
distribution (stratification) and stabilisation of SOM in volcanic ash soils under (semi‐)
natural vegetation. The current PhD thesis concerns the latter topic. A list of publications as
a result of the RUFLE project up to 2008 other than those contained in this PhD thesis are
presented in Appendix I.

8
Research objectives and outline

The main objective of this PhD research is to gain insight into the processes affecting the
vertical distribution and stabilisation of SOM in volcanic ash soils in tropical montane forest
and páramo ecosystems in the northern Ecuadorian Andes. This study thus supports the
reconstruction of the natural position of the UFL during the Holocene and contributes to
sustainable management of volcanic ash soils as carbon sinks.

Boundary conditions for the vertical distribution of SOM in volcanic ash soils are set by
geogenetic processes, specifically the amount and timing of tephra depositions. However,
the actual vertical distribution of SOM depends on subsequent pedogenetic processes: e.g.
aboveground and belowground litter input, organic matter decomposition, bioturbation and
leaching. SOM accumulation in volcanic ash soils is generally related to the formation of
organo‐mineral and/or organo‐metallic complexes, each protecting SOM from
decomposition through mechanisms that are still subject of debate (Mizota & van Reeuwijk
1989; Shoji et al. 1993; Dahlgren et al. 2004; Buurman et al. 2007). Therefore, the research
aim was further specified in two research questions:

1. What is the influence of geogenetic and pedogenetic processes on the vertical
distribution of SOM in the volcanic ash soils studied?

2. What are the mechanisms of SOM stabilisation and how much SOM accumulates in
these volcanic ash soils?

A general description of the physiography of the research area is provided in Chapter 1. The
research questions are addressed in subsequent chapters. In Chapter 2 radiocarbon dating
of different fractions of SOM is discussed, to obtain an accurate time‐line for paleorecords
and to offer an additional proxy clarifying the vertical distribution of SOM. In Chapter 3 the
influence of geogenetic processes is investigated by unravelling the tephra stratigraphy of
the volcanic ash soils. In Chapter 4 the role of pedogenetic processes, notably bioturbation,
is studied with micromorphological techniques. In Chapter 5 soil organic carbon stocks are
quantified and mechanisms of soil organic carbon stabilisation are examined. In Chapter 6
the processes affecting the vertical distribution of SOM are further explored by integrating
knowledge obtained in the previous chapters in a modelling exercise. Finally, in the
Synthesis the conclusions of these chapters are placed in a broader framework and
combined to answer the research questions.

9
1 Description of the research area

Location of study area and study sites

The Ecuadorian Andes consists of two mountain ranges trending northeast‐southwest,
called the Western Cordillera and the Eastern Cordillera, separated by the Inter Andean
valley (Sauer 1971). Our study sites are located in the nature protection area of Estacion
Biologica Guandera in northern Ecuador near the border with Colombia, Figure 1.1 and
Table 1.1. Guandera is situated in the Eastern Cordillera on the western flanks facing the
Inter Andean valley and has a (semi‐) natural UFL at approximately 3650 m a.s.l. Since pre‐
Columbian times, the Inter Andean valley has been the most densely populated part of the
country and deforestation in the Inter Andean valley has therefore been particularly severe.
Guandera contains unique remnants of native tropical montane forest and páramo
ecosystems representative of the ‘Tropical Andes’ biodiversity hotspot (Myers 2000) and its
continued preservation is of key importance.

Figure 1.1 Map of study area and study sites. § = study site; = village or town (courtesy of J.P. Lesschen).

11
Table 1.1 Description of study sites

Site Altitudea Coordinatesb Slope Aspect Soil type Current vegetation
[m a.s.l.] [°] [°] WRB
G11 3330 N 0°35’12’’/ W 77°42’37’’ 8 260 Andosol Forest patch
G1 3501 ± 9 (n = 20) N 0°35’27’’/ W 77°42’1’’ 3 270 Histosol Forest
G2 3520 ± 9 (n = 14) N 0°35’27’’/ W 77°42’1’’ 8 290 Histosol Forest
G4 3616 ± 12 (n = 12) N 0°35’53’’/ W 77°41’ 41’’ 9 250 Andosol Forest just below UFL
G5a 3697 ± 9 (n = 9) N 0°35’41’’/ W 77°41’36’’ 13 280 Cambisol Forest patch
G5b 3694 ± 13 (n = 9) N 0°35’41’’/ W 77°41’35’’ 8 280 Andosol Páramo
G6 3790 N 0°35’46’’/ W 77°41’30’’ 10 280 Andosol Páramo
a
= Altitudes from altimeter, n = 1 unless stated otherwise
b
= GPS coordinates in WGS 1984

We selected an altitudinal transect from 3330 to 3990 m a.s.l. intersecting the current UFL
(Figure 1.2), starting with a site in a forest patch surrounded by agricultural lands (G11),
followed by sites currently covered by continuous forest (G1 and G2), a site just below the
UFL (G4), a site in a forest patch above the UFL (G5a) and 5 sites currently covered by
páramo vegetation (G6, G7, G9, G10) of which one was located next to the forest patch
(G5b). Furthermore, a small mire (G15) and a peat bog (G8) were selected specifically for
paleoecological analysis, which was the topic of a separate PhD thesis within the RUFLE
project and will not be further discussed here. Two main findings of the RUFLE project are
that site G5a has been only relatively recently (~50 years) colonised by forest vegetation
(Jansen et al. 2008), while G1 and G2 have been forested for longer time periods of at least
500 years (Bakker et al. 2008).

Figure 1.2 Altitudinal transect of study sites at the
4000 .
nature protection area of Guandera Biological G9
Station (courtesy of M. Moscol Olivera).
Sites G8 and G15 were specifically selected
G10 3900 .
for pollen analysis and not used in
G7
this thesis.
G8
G6 3800 .
Forest patch Peat bog

G5a G5b 3700 .

G4 3640 ma.s.l. UFL
Low forest 3600 .

Páramo island
Altitude (m a.s.l.)
G2
G1
3500
.
Mire
Clusia forest
G15
3400
.
+ 3300 .
Forest patch
12 G11
Climate

Guandera has a typical humid tropical alpine climate. Equatorial temperatures do not show
large seasonal variation (Sarmiento 1986). Large daily air temperature fluctuations are
typical for tropical alpine environments and are poetically described as “winter every night,
summer every day” by Hedberg (1964). Night frost may occur year round, with frequencies
depending on altitude. Precipitation patterns vary with the movement of the Intertropical
Convergence Zone and depend on distance to the Pacific Ocean, altitude and exposition.
Mean annual precipitation in Guandera is around 1900 mm. In general, seasonal variation in
precipitation is small (Bader 2007) and therefore the soil climate is perudic. Soil
temperature at 30 cm, representative of the mean annual temperature, was on average 9.1
± 0.5 °C at forest sites and 9.7 ± 0.4 °C at páramo sites (isomesic), which is similar to data
reported by Bader (2007). Soil temperature did not show a lapse rate with altitude, probably
because forest vegetation and high air humidity in the páramo buffer the temperature. Daily
air temperature fluctuations at the UFL in Guandera are much larger in páramo (up to 30 °C)
than in forest (up to 15 °C), as discussed by Bader (2007).

Geology

Volcanism in Ecuador and southern Colombia is part of the ‘Northern Volcanic Zone’,
extending from 5˚N tot 2˚S (Stern 2004). In the neighbouring Amazonian basin, back‐arc
volcanoes are situated (Barragan et al. 1998). The northern part of the Ecuadorian Andes is
characterised by Cenozoic volcanic rocks (Hörmann & Pichler 1982). The volcanic rocks of
the Western Cordillera belong to a calc‐alkaline andesite‐dacite series, while the volcanic
rocks in the Eastern Cordillera are members of the andesite‐dacite‐rhyodacite series
(Hörmann & Pichler 1982; Stern 2004). Volcanoes in the region of the study area that have
been active over the last thousands of years are Cerro Negro, Chiles, Soche and Azufral,
while Cumbal erupted during historic times (Droux & Delaloye 1996; Instituto Geofisico
2006). Additionally, explosive eruptions of the more distant volcanoes Cuicocha and
Quilotoa might have reached the area (personal communication Hall 2006).

Eruptions of andesitic magma generally produce limited amounts of volcanic ash that are
distributed locally by the dominant S‐SE‐E winds of the Inter Andean valley, while more
siliceous magma types generate large quantities of ash that may be transported over much
larger distances eastwards in the stratosphere (Hall & Mothes 1994). Tephra deposits show
lateral and vertical variations in particle size and mineralogy. The largest and heaviest
particles fall nearest to the source, while smaller and lighter particles fall at increasing
distances (Shoji et al. 1993). Therefore, the mineralogical assemblage of volcanic ash is
often not representative of the source magma. However, Hall & Mothes (1994)
demonstrated that the mineralogical assemblage of pumice in contrast is representative of
the source magma, and that most Ecuadorian volcanoes possess a typical mineralogical
assemblage that varies little between eruptions. The mineralogy of pumice encountered in
13
the different (rhyo‐) dacitic tephra deposits in which the studied volcanic ash soils were
formed closely resembles that of the Soche volcano with at least one known eruption,
containing feldspar, amphibole, quartz and biotite. The occurrence of pumice of large size
(up to 3 cm) implies a nearby source and because the study area is directly west of Soche
volcano, it is likely that the tephra deposits in the study area are of Soche origin (personal
communication M. Hall 2006).

Vegetation

The vegetation in Guandera is described in detail by Moscol Olivera & Cleef (2008a and
2008b). The tropical montane forest has a structure and composition characteristic for
mountainous equatorial regions, being characterised by a reduced tree stature, canopy
trees with dense compact crowns, small coriaceous leaves and a high proportion of
epiphytic biomass of bryophytes, ferns, bromeliads, orchids and other vascular species
(Keese et al. 2007). The transition from forest to páramo is abrupt, probably indicating at
least some human influence (Laegaard 1992), but the forest still contains trees with a
considerable diameter at breast height (dbh) of 70 cm. Although occupying only 10 % of
Ecuador, the tropical montane forests contain half of the country’s species (Keese et al.
2007). The dominant species in the Guandera forest include Clusia flaviflora Engl.,
Weinmannia cochensis Hieron. and Ilex colombiana Cuatrec. Aboveground biomass of a
similar Clusia forest at 3059 m a.s.l. in southern Ecuador was estimated between 7.4 and
15.6 kg m‐2 (Leuschner et al. 2007) with a leaf litter production of 263 ± 44 g m‐2 y‐1 and a
root litter production of 2084 ± 177 g m‐2 y‐1 (Röderstein et al. 2005).

Páramo refers to the tropical montane grasslands encountered between the UFL and the
snow limit in the equatorial Andes. The páramo forms an altitudinal belt extending from
northern Peru to northern Costa Rica (Luteyn 1992) and occupies 5 % of the land area in
Ecuador thus representing 25 % of the total Andean páramo surface (Mena Vasconez et al.
2001). The Guandera páramo is characterized by bunchgrasses of Calamagrostis effusa
(Kunth) Steud. and stem rosettes of Espeletia pycnophylla Cuatrec., the latter only occurring
from northern Ecuador ‐ including our study area ‐ and further north. Aboveground biomass
in a similar páramo ecosystem at 4100 m a.s.l. in Colombia was 3.5 kg m‐2 with
approximately 1.7 kg m‐2 leaf litter and a belowground biomass of 1.2 kg m‐2 (Hofstede &
Rossenaar 1995).

Soils

Authorative reviews concerning the properties of volcanic ash soils are presented by Theng
(1980), Mizota & Van Reeuwijk (1989), Shoji et al. (1993) and Dahlgren et al. (2004).
Characteristic chemical and physical soil properties and soil classification according to the
World Reference Base (WRB) for soil resources developed by the FAO (2006) of the soils in
the study area are discussed in more detail in this section.
14
Soil profile descriptions according to the FAO (1990) guidelines are presented in Appendix II.
Figure 1.3a and 1.3b show typical soil profiles under forest and páramo vegetation
respectively. The forest soils characteristically had ectorganic horizons with a combined
thickness of up to 100 cm, classifying as Folic horizons according to the World Reference
Base (FAO 2006). Roots were generally concentrated in these ectorganic horizons. Such
thick ectorganic horizons (Wilcke et al. 2002) with rooting (Soethe et al. 2006) are
characteristic for tropical montane forest. Contrary, ectorganic horizons barely exceeded 1
cm in the páramo profiles and litter was mostly present as hanging dead material within the
tussock grasses or on the stem rosettes, while roots were concentrated mainly in the
mineral topsoil. These traits are indeed typical for páramo vegetation (Hofstede &
Rossenaar 1995). C/N ratios of forest litter (~88 mass ratio) and páramo litter (~87) were
similar, C/N ratios in the other ectorganic horizons of the forest sites were lower (~39).

Based on the mineral soil horizon sequence, we generally identified at least two tephra
deposits in the field (on average ~50 cm thick), recognizable as a ‘current soil’ and a
‘paleosol’. Beneath this paleosol an ‘unweathered ash’ was present that could either belong
to the paleosol or represent a separate preceding tephra deposition and thus could
constitute an additional immature or strongly truncated paleosol. Most soils had a placic
horizon, which may be ascribed to the strong textural contrast hindering drainage of the
otherwise well drained soils.
Figure 1.3 a) Forest soil profile (G1) b) Páramo soil profile (G7)
15
The horizon sequence based on the field observations can be summarised as O – Ah – AB or
Bw – 2Ahb – 2Bsb – 2 or 3 BCb. The number of tephra deposits is often underestimated in
the field as soil formation following tephra deposition may overprint preceding deposits
(Buurman et al. 2004). In Chapter 3 the conclusive tephra stratigraphy and corresponding
soil horizonation is unravelled. Therefore, Chapter 2 refers to field horizons while in Chapter
4 to 6 horizon naming is according to this unravelled tephra stratigraphy.
Chemical and physical soil properties are presented in Appendix III and in Chapter 3 and 5,
the methodology is described in these chapters, unless specifically mentioned here. In
general, organic carbon contents were very high (> 6 mass %) and soil pH was acidic (pHCaCl2
= 4.2 ± 0.3) throughout the current soil and paleosol, while the unweathered ash contained
less organic carbon (0.9 ± 1.2 %) and had a higher pHCaCl2 (5.1 ± 0.4). The C/N ratio was on
average 21 ± 4 throughout the soil profiles. The crystalline clay fraction throughout the soil
profile contained predominantly Al‐hydroxy interlayered minerals and clay sized primary
minerals (albite, amphibole, cristobalite and quartz) and only trace amounts (if any) of
kaolinite, illite and halloysite. Allophane was absent in the current mineral topsoils, but
allophane content in the current subsoils was on average 4.4 ± 2.6 % and increased to 9.5 ±
4.0 % in the paleosol. The pH as measured in 1 M NaF (1:50 w/v) exceeded 9.5 throughout
the mineral soil, indicating the presence of (amorphous) hydroxide groups (Fieldes & Perrott
1966). In the current soil and paleosol, pHKCl was always higher than pHCaCl2 demonstrating
the presence of dominant positive variable charge rather than permanent negative charge,
which is typical for volcanic ash soils and related to amorphous constituents. The effective
CEC measured at ambient soil pH in 1 M KCl or 1 M NH4Cl (w/v 1:5) is 67 ± 44 mmolckg‐1 in
the current soil and 41 ± 10 mmolckg‐1 in the paleosol, with a base saturation of only 13 ± 7
% in the current soil and 17 ± 10 % in the paleosol. Exchangeable acidity (H+ and Al3+
extractable with 1M KCl) amounted to 60 ± 44 mmolckg‐1 in the current soil and 35 ± 31
mmolckg‐1 in the paleosol.
Dry bulk density was low (<< 0.90 gcm‐3) throughout the current soil and paleosol and higher
in the unweathered ash (> 0.90 gcm‐3). Porosity was high (76 ± 8 %) in representative
páramo topsoil samples of site G7 and G9 and in a forest subsoil sample of site G11. The
density of the solid phase in these samples was 1.9 ± 0.1 gcm‐3 (as determined with a
pycnometer) reflecting the high SOM content. Gravimetric water content was 163 ± 57 % at
saturation point (15 kPa or pF 2.2) and still 46 ± 13 % at wilting point (1500 kPa or pF 4.2).
Soil structure was massive, consistency was (very) friable when moist and non to slightly
sticky when wet. Moreover, the soils showed thixotropic behaviour, i.e. the soil changes
from a plastic solid into a liquid under pressure.
Mineral topsoils of both forest and páramo sites generally showed Vitric properties rather
than Andic properties according to the WRB classifaction (FAO 2006), because the (Alo + ½
Feo, subscript ‘o’ indicates oxalate extractable) content, indicative for the degree of
weathering, did not exceed 2.0. Since all other diagnostic criteria for Andic properties were
fulfilled, e.g. low dry bulk densities (≤ 0.90 gcm‐3) and high P retention (≥ 85 %), and because
16
the (Alo + ½ Feo) content did approach or exceed 2.0 in the subsoils, we consider the Vitric
properties on the verge of maturing into Andic properties. In accordance we decided to
apply the prefix qualifier Andic rather than Vitric in the soil classification. More specifically,
the current soils possess Aluandic properties because Alp/Alo (subscript ‘p’ indicates
pyrophosphate extractable) exceeded 0.5, indicating the dominance of Al‐humus
complexes. Contrary, the paleosol generally showed Silandic properties with an Alp/Alo ratio
below 0.5, pointing towards the predominance of amorphous minerals. Andosols with
Aluandic properties in the mineral topsoil are also referred to as ‘non‐allophanic’, while
those with Silandic properties in the mineral topsoil are called ‘allophanic’ (Takahashi &
Shoji 2002). The unweathered ash beneath the paleosol had Vitric properties. Additionally,
all páramo soil profiles contained an Umbric surface horizon because of their low base
saturation.
Shoji et al. (1993) mentioned the influence of vegetation in creating either Fulvic (forest) or
Melanic (grassland) Andosols and this vegetation effect was also present in our study area.
At the forest sites, the diagnostic criteria for Fulvic horizons were not met sensu strictu,
because of the absence of unambiguous Andic properties in the mineral topsoils. However,
we argue that classification as Fulvic horizon is still justified because of the clear
development towards Andic properties. In order to fulfil the cumulative thickness
requirement of 30 cm, Ah and Bw horizons were grouped, which is allowed because both
horizons fulfil the supplementary diagnostic criteria of either having a melanic index greater
than 1.7 or having a Munsell colour chroma exceeding 2. In most páramo sites, the
diagnostic criteria for Melanic horizons were clearly met, despite the Vitric nature of the
mineral topsoils, because average (Alo + ½ Feo) content of the Ah horizon did closely
approach or exceed 2.0. Accordingly, we added the qualifier Melanic to the páramo soils.
The forest patch soil (G5a) has a Melanic horizon, although at the top of the Ah horizon the
melanic index slightly exceeded 1.7 indicating transformation towards a Fulvic horizon,
probably as a result of the relatively recent recolonisation by forest.
Forest soils with thick ectorganic horizons (> 25 cm) keyed out as Histosol or Cambisol
rather than Andosol in the WRB classification, but to enable easy comparison we used the
list of Andosol qualifiers also for these soil types during classification. Additional specifiers
‘Endo‐’ or ‘Bathy‐’ indicate that the diagnostic feature started at a depth greater than 50 cm
or 100 cm respectively, while specifier ‘Thapto‐’ indicates the presence of buried layers with
a cumulative thickness of 30 cm within 100 cm from the soil surface. Concluding, the forest
patch surrounded by agricultural land (G11) is a Folic, Hemic, Aluandic, Fulvic Andosol
(Dystric, Placic, Thixotropic). Forest soils (G1, G2) classify as Folic, Hemic,
Endo/Bathyaluandic, Endo/Bathyfulvic Histosols. The forest site just below the UFL (G4) has
a Folic, Hemic, Aluandic, Fulvic, Thaptaluandic, Andosol (Thixotropic, Colluvic, Dystric), and
the forest patch soil (G5a) is a Folic, Hemic, Aluandic, Endomelanic Cambisol (Hyperhumic,
Thixotropic, Dystric). The páramo soils (G5b, G6, G7, G9, G10) classify as Aluandic, Melanic,
Umbric Andosols (Thixotropic).
17
2 Radiocarbon dating of soil organic matter
fractions in Andosols in northern Ecuador
Published in Radiocarbon 48(3): 337–353, 2006,

Tonneijck FH, Van der Plicht J, Jansen B, Verstraten JM & Hooghiemstra H.

Abstract

Volcanic ash soils (Andosols) may offer great opportunities for paleoecological studies, as
suggested by their characteristic accumulation of organic matter (OM). However,
understanding of the chronostratigraphy of soil organic matter (SOM) is required.
Therefore, radiocarbon dating of SOM is necessary, which is unfortunately not
straightforward. Dating of fractions of SOM obtained by alkali–acid extraction is promising,
but which fraction (humic acid or humin) renders the most accurate radiocarbon dates is
still subject to debate. To determine which fraction should be used for radiocarbon dating of
Andosols and to evaluate if the chronostratigraphy of SOM is suitable for paleoecological
research, we measured radiocarbon ages of both fractions and related calibrated ages to
soil depth for Andosols in northern Ecuador. We compared the time frames covered by the
Andosols with those of peat sequences nearby to provide independent evidence. Humic acid
(HA) was significantly older than humin, except for the mineral soil samples just beneath a
forest floor (ectorganic horizons) where the opposite was true. In peat sections, radiocarbon
ages of HA and humin were equally accurate. In the soils, calibrated ages increased
significantly with increasing depth. Age inversions and homogenisation were not observed
at the applied sampling distances. We conclude that in Andosols lacking a thick ectorganic
horizon, dating of HA renders the most accurate results, since humin was contaminated by
roots. Contrary, in mineral soil samples just beneath a forest floor, humin ages were more
accurate, because HA was then contaminated by younger HA illuviated from the ectorganic
horizons. Overall, the chronostratigraphy of SOM in the studied Andosols appears to be
suitable for paleoecological research.

Keywords
radiocarbon dating, volcanic ash soils, peat, pollen analysis, soil organic matter fractions
19
Introduction

In the Ecuadorian Andes, a rise in deforestation, grazing and burning as a result of the
increase in population pressure during the last 500 years severely disturbed the Upper
Forest Line (UFL) (Ellenberg 1979; Hofstede 1995b). Therefore the natural altitudinal
position of the UFL is subject to debate (Laegaard 1992; Wille et al. 2000). The last decade,
negative impacts of deforestation were recognized and efforts made to forest degraded
lands. However, forestation of tropical alpine grasslands (páramo) that under natural
circumstances would not be forest, damages that ecosystem. To support sustainable
forestation, we need to investigate the natural dynamics of the UFL by investigating the
local vegetation history.

Traditionally, vegetation history is reconstructed by analysing fossil pollen records in peat
bogs or lake sediments (e.g. Erdtman 1943; Birks & Birks 1980; Moore et al. 1991; Bradley
1999). To increase understanding of UFL dynamics, our investigation would ideally be
concentrated around the hypothetical UFL. Unfortunately, conventional polliniferous
deposits are not necessarily located there. Alternatively, Andosols, the dominant soil type in
the high Andes, could provide great opportunities for paleoecological studies including
pollen analyses (Bakker & Salomons 1989). Andosols accumulate vast amounts of OM (Shoji
et al. 1993), which is related to low temperatures, acidic soil pH and the formation of
organo‐metallic or organo‐mineral complexes that resist decomposition (Shoji et al. 1993;
Torn et al. 1997). These conditions also favour pollen preservation (Moore et al. 1991),
which has been demonstrated in Andosols above 3000 m a.s.l. in Colombia. (Salomons
1986).

However, to link soil pollen records to a time scale, understanding of the stratification of
SOM in the soil and its age‐depth relationship in particular is required (Keatinge 1983;
Davidson et al. 1999). Furthermore, since pollen analysis in terrestrial soils is not a widely
applied method yet, verification of the results is necessary e.g. by comparing the pollen
records from soils to those from nearby peat bogs. For such a comparison an accurate
chronology of either record is crucial. Since peat bogs and Andosols are both rich in organic
carbon, measuring of 14C activities offers a suitable dating method.

Unfortunately, radiocarbon dating of SOM is not straightforward (e.g. Mook & Streurman
1983; Scharpenseel & Becker‐Heidmann 1992; Pessenda et al. 2001). SOM consists of a
range of complex organic molecules, varying in decomposition phase. The continuous input
of fresh OM into soils causes the measured 14C ages of bulk SOM to be somewhat younger
than the depositional age (Wang et al. 1996). In addition, input of young roots into older soil
horizons, leaching of mobile organic compounds through the soil profile, bioturbation and
deposition of allochtonous OM may also contaminate SOM ages (Mook & Streurman 1983;
Pessenda et al. 2001). The relative importance of these pedogenetic processes, which
determine the measured age of SOM, depends on e.g. soil properties and vegetation type.
20
Therefore, interpretation of radiocarbon dates must be tailored to the specific soil
ecosystem under study.
If macro remains of vegetation are lacking in the soil, dating of chemical fractions of SOM
(humic substances) obtained by the standard alkali–acid extraction (Schnitzer 1982) instead
of dating bulk SOM, offers the opportunity to select the fraction yielding a more accurate 14C
age as based on a qualitative evaluation on the validity of dating (Orlova & Panychev 1993;
Kristiansen et al. 2003). However, which fraction provides the most accurate dates is not
obvious. Radiocarbon dating of humic substances rendered more accurate results in a range
of soil types (Van Mourik et al. 1995; Dalsgaard & Odgaard 2001; Kovda et al. 2001;
Pessenda et al. 2001), but Andosols were not yet investigated. Dating peat bog sediments by
14
C analysis is generally more straightforward than dating terrestrial soils, as peat sequences
are much less disturbed.
The aims of this current paper are 1) to determine which fraction of SOM (HA or humin)
renders more accurate radiocarbon dates in Andosols and 2) to evaluate whether the age‐
depth relationship of SOM in the Andosols of our study area is suitable for paleoecological
research such as the reconstruction of the natural position of the UFL. The results are
compared to 14C ages of similar OM fractions from nearby peat bogs, to assess whether the
pollen records from those peat sequences can be used to verify the soil pollen records. By
fulfilling these aims, the current paper will make an indispensable first step towards
reconstruction of the UFL in northern Ecuador.

Descriptive background
The study sites are located in three nature protection areas in northern Ecuador near the
border with Colombia, Table 2.1 and Figure 2.1.

Table 2.1 Description of study sites
Soil type (FAO) Altitude Current vegetation Coordinatesa Cordillera Site

[m a.s.l.]
Andosol 3860 Páramo N 0°35’48’’/ W 77°41’25’’ Eastern G7
Andosolb 3697 ± 9 (n=9) Forest patch N 0°35’41’’/ W 77°41’36’’ Eastern G5a
Andosol 3694 ± 13 (n=9) Páramo N 0°35’41’’/ W 77°41’35’’ Eastern G5b
Andosol 3570 Forest patch N 0°39’31’’/ W 77°52’33’’ Western A7A
Andosol 3570 Páramo N 0°39’31’’/ W 77°52’36’’ Western A7B
Histosol 3869 Xyris bog N 0°35’42’’/ W 77°41’14’’ Eastern G8
Histosol 3740 Oreobolus bog N 0°41’16’’/ W 77°52’46’’ Western A11
Histosol 3418 Juncus bog N 0°38’40’’/ W 77°53’39’’ Western A2
a
= GPS coordinates in WGS 1984, altitudes from altimeter, GPS altitude only used for site G8, n = 1 unless
stated otherwise.
b
= This profile classifies as (Andic) Cambisol instead of Andosol, but since soil properties are similar to the
other soils and for the purpose of simplicity we refer to the soil as Andosol in the context of this paper.
21
The Guandera Biological Station lies in the Eastern Cordillera (mountain range) and
preserves a (semi‐)natural UFL at approximately 3650 m a.s.l. The transition from forest to
páramo is abrupt, probably indicating at least some human influence (Laegaard 1992), but
the forest still contains trees with a considerable diameter at breast height (dbh) of 70 cm.
Some forest patches occur above the current UFL. The dominant species in the forest are
Clusia flaviflora ENGL., Weinmannia cochensis HIERON. and Ilex colombiana CUATREC. and the
páramo is characterized by Calamagrostis effusa KUNTH (STEUD.) bunch‐grass and Espeletia
pycnophylla CUATREC. stem rosette.

Figure 2.1 Maps of Ecuador (province of Carchi
indicated in grey, with approximate location of
the study area indicated by a square) and of the
study area, including the location of the National
Parks (in grey) and of the study sites.

z = Sites of transect Western Cordillera;
S = Sites of transect Eastern Cordillera;
= village or town

22
El Angel Ecological Reserve and Los Encinos Biological Station are located close to each
other in the Western Cordillera and were heavily deforested before conservation, the few
remaining trees rarely exceed 30 cm dbh. El Angel and Los Encinos are now covered with
páramo vegetation and some patches of remnant forest and there is no proper UFL.
However, the zonal forest patches that seem to maintain more of the original taxonomic
composition of the natural montane forest are located between 3450 and 3600 m a.s.l. and
dominated by different woody species of the genus Weinmannia and Miconia, as well as
Clethra crispa C. GUST. and Hedyosmum cumbalense H. KARST. Above this altitude, the
páramo vegetation is characterized again by C. effusa and E. pycnophylla.

The study area in the Eastern Cordillera receives almost double the precipitation annually
than the study areas in the Western Cordillera (annual means respectively 1900 mm and
1000 mm), but mean annual temperatures are similar (from 12 °C at 3400 m a.s.l. to 4 °C at
4000 m a.s.l.). We did not observe volcanic emanations of (fossil) CO2 in the field.

In both Cordilleras we selected a forest patch and a site with páramo next to it, to
determine the effect of the current vegetation on soil profile development and its SOM age‐
depth relationship. The floristic composition of the forest patch in Guandera is similar to
that of the continuous forest belt just below the UFL, the subalpine rainforest, and similarly
the forest patch in Los Encinos appears to maintain the original taxonomic composition of
the now disappeared forest (Moscol Olivera et al. 2008a and b). In Guandera we additionally
selected a peat bog at 3869 m a.s.l. and a site with páramo next to it. Furthermore, in El
Angel two peat bogs were investigated, one at 3740 m a.s.l. and one at 3418 m a.s.l. The El
Angel peat bog at higher altitude is characterized by Oreobolus obtusangulus GAUDICH., and
the peat of Guandera by Xyris subulata RUIZ & PAV., both small herbs, whereas the El Angel
peat bog at lower altitude is dominated by the tall rush species Juncus balticus ssp. andicola
(HOOK.) SNOGERUP. The sites selected for radiocarbon dating belong to two transects with
sites at altitudinal intervals of 100 m from 3000 to 4000 m a.s.l.

Materials and methods

Sampling procedures

Soil pits of approximately 1.5 m2 surface area and a depth of 1.5 m to 2 m, depending on the
soil profile, were excavated and soil profiles were described according to the FAO guidelines
(FAO 1990) and classified according to the World Reference Base (FAO 2006). In both study
areas one paleosol (buried soil) was generally observed beneath the current soil, indicating
at least two main events of tephra deposition. The current soil was considered sufficiently
thick to contain a period without intensive human activities. Therefore, the main focus was
on the current soil profile. We took an undisturbed vertical soil sample using a metal gutter
of 75 x 5 x 4 cm and, if the current soil was not yet completely covered, a second vertical
sample was collected with a small overlap. The gutters were packed in aluminium foil and
23
then plastic foil and finally securely tied with adhesive tape to avoid dislocation and drying
during transport. In addition, we took bulk soil samples for chemical analyses
(approximately 1 kg field moist material) and ring samples (with a volume of 100 cm3) for
bulk density determination in the same soil pit at regular depths but respecting horizon
boundaries. Each bulk soil sample was obtained from a vertical interval of approximately 5
cm, the middle of which was noted as the depth of the sample, in order to gather sufficient
material. The gutters and soil samples were stored at 2 °C under field moist conditions prior
to analyses. The peat bogs were sampled to their maximum depths with hand operated
Russian augers of 75 mm and 50 mm diameter and a standard length of 50 cm from
Wittander Co. (Lund, Sweden). The smaller diameter auger is used when sediments are
compact and resistant. The auger is brought to the correct depth by 1‐m long extension
rods. The auger segments were wrapped in halved plastic tubes and then packed and stored
like the gutter samples.

For radiocarbon dating, at least three samples per soil profile were cored from the gutter
samples using a corer with 0.5 cm diameter to minimise the time span covered by the
sample. Two samples were located in the current soil: one just below the mineral soil
surface in the Ah horizon and one at its bottom in the Ah2 or Bw horizon. A third sample
was taken at the top of the paleosol, the 2Ahb horizon. Additionally, we took two samples
from the thick ectorganic horizon in the Guandera forest profile (one in the middle and one
at the boundary with the Ah horizon) and one extra sample at the top of the páramo soil
profile located next to it. The vertical distance between samples was on average 40 cm. The
samples for radiocarbon dating were taken at the same depths as the samples for chemical
analyses (except for the sample at the boundary between ectorganic horizon and Ah). In the
peat cores three samples at regular spacing were taken using the same 0.5 cm diameter
corer. Fresh roots were removed from the peat sample by hand‐picking, using a binocular.

Laboratory procedures

Chemical fractionation of SOM by alkali‐acid extraction followed the procedure of Schnitzer
(1982), which is the conventional method for OM fractionation (Stevenson 1994). The
organic fraction that is soluble in alkali solution and immobile in acid solution is called HA,
the residual fraction (immobile in both alkali solution and acid solution) is referred to as
humin. Fulvic acid, the OM fraction that is soluble in both alkali solution and acid solution,
was discarded because of its inherent mobility in soils. Humin is generally considered to be
the most humified (decomposed) OM fraction. Strictly spoken, the division of OM in humic
substances is not meaningful in the case of anaerobic peat, since decomposition is negligible
and consequently the composition of OM in peat is not much altered. However, for a proper
comparison of soil and peat records we fractionated the peat samples as well, applying the
same procedure.

24
Soil pH was measured with a glass electrode (Consort SP10B, Turnhout, Belgium) in a 0.01 M
CaCl2 soil solution (w/v: 1:5 mineral soil samples and w/v 1:10 organic samples, using
fieldmoist samples) as a measure of actual acidity. Total carbon was measured with a
VarioEL (Elementar Analysensysteme GmbH, Hanau, Germany) CNS auto‐analyser. Total
carbon equals organic carbon, since carbonates are not present. Dry bulk density was
determined by weighing oven‐dried samples (105 °C, 44 h mineral soil samples and 70 °C,
48h organic samples) and then dividing the oven‐dry weight by the known volume of the
sampling ring (100 cm3).

Radiocarbon dating

Radiocarbon analysis was performed by the AMS facility of Groningen University, the
Netherlands (laboratory code GrA). The AMS (Accelerator Mass Spectrometer) enables the
measurement of 14C concentrations in graphite, made from milligram size sample material.
After pre‐treatment as described above, the isolated fraction of the original sample is
combusted into CO2 and purified using an Elemental Analyser (EA) (Aerts et al. 2001). The
EA provides also quality check parameters such as the organic carbon content and the δ13C
value of the sample. The CO2 is then collected cryogenically for later graphitisation. The
graphite powder is pressed into targets which are placed in the sample carousel of ion
source of the AMS. The AMS system measures the isotopic ratios 14C/12C and 13C/12C of the
graphite (Van der Plicht et al. 2000). Typical measurement errors are 0.4 % (Meijer et al.
2006). From these measured isotopic ratios, the 14C activities are calculated (including the
correction for isotopic fractionation) and the 14C ages in BP.

Telford et al. (2004) explained that mathematical age‐depth relationships only make sense
when using calibrated radiocarbon ages. Therefore, to evaluate the OM age‐depth
relationship for the soil and peat sequences, we transformed 14C ages to a calendar year
probability distribution (cal AD/BC) with the WinCal25 calibration software (Van der Plicht
2005) using the presently recommended IntCal04 calibration dataset (Reimer et al. 2004).
Some 14C measurements yield activities > 100 %, which means the samples are modern,
containing 14C above natural level produced by atmospheric nuclear explosions during the
late 1950's and early 1960's, the so‐called ‘bomb peak’. The atmospheric 14C content for
these ‘modern samples’ is derived from direct measurements of 14C in atmospheric CO2, for
a variety of locations (e.g., New Zealand: Manning et al. 1990; Norway: Nydal & Lövseth
1983; the Netherlands: Meijer et al. 1995). These records can be used to transfer the 14C
activities to calendar ages. Our data contain just a few such modern measurements; it is not
relevant which of these atmospheric datasets are used for our purposes.

Calibrated radiocarbon ages are expressed as the range of calendar years of the 1 sigma
peak with the largest relative area under the probability distribution, except for the modern
samples where the most probable range was selected based on expert knowledge. Since the
25
ranges are small, calculations are performed with the mean and the dots displayed in the
graphs represent the whole range of calendar years at the scale of the graph.

Statistical analysis
Differences in means were considered significant when p < 0.01 unless stated otherwise, as
determined with a 2 tailed Paired Samples Test performed with the SPSS Compare Means
procedure (SPSS Inc. 2001). Pearsons correlations (bivariate) between variables were
calculated using the SPSS Correlate procedure and considered significant when p < 0.01
unless stated otherwise. We performed a linear regression in Graphpad Prism (Graphpad
Software Inc. 1999) to evaluate the relationship between calibrated age and depth.

Results and discussion

Soil properties are reported in Table 2.2. The soils in both areas consist of Andosols (FAO
2006) with upper mineral horizons characterised by a very large organic carbon content (9 %
to 21 %), acidic pH (pHCaCl2 3.7 to 4.5) and low dry bulk densities (0.4 to 0.6 gcm‐3). The dry
bulk densities are similar for all samples and therefore the effect of compaction on the age‐
depth relationship as described by Hetier et al. (1983) is minimal. The horizon sequence can
be summarised as Ah1 ‐ Ah2/Bw ‐ 2Ahb ‐ 2Bsb ‐ 2BCb, common for Andosols (Shoji et al.
1993). The forest profiles in addition have ectorganic horizons (a forest floor) overlying the
mineral horizons. All soils show a multisequum, containing at least one paleosol. Roots
concentrate in the first Ah horizon of the páramo soils and in the relatively nutrient rich
ectorganic horizons of the forest soils.

14
C ages of HA and humin in Andosols
The results of 14C dating of the HA and humin fraction of soils collected in the Eastern and
Western Cordilleras of northern Ecuador are presented in Figure 2.2 and Table 2.3. The δ13C
values (Table 2.3) indicate a dominant C3 vegetation, which is to be expected in a humid
and cold climate. Humic acid and humin contributed equal proportions (both approximately
30 % organic carbon, see Appendix III) to the total bulk soil organic carbon, underlining that
the impact of contamination of either fraction on bulk SOM age will be considerable.
The HA fraction was significantly (p = 0.001) older than the humin fraction, except for the
mineral soil samples just beneath a forest floor where the opposite was found. The
difference between the 14C ages of the fractions (hereafter referred to as ‘age difference’)
can be used as a measure of contamination with younger material (e.g. by roots). The
absolute age difference ranges from 33 14C yr to 995 14C yr and increases significantly with
increasing 14C age (r = +0.85, p < 0.001 with 14C age HA and r = +0.79, p = 0.001 with 14C age
humin; mineral soil samples just beneath a forest floor excluded). The 14C age increases
significantly with depth for both fractions (r = +0.93 with 14C age HA and +0.94 with 14C age
humin, both with p < 0.001, modern samples excluded).
26
Table 2.2 Selected soil properties for the Andosols (mean and standard deviations calculated if at least 2
samples were taken within the horizon at different depths).

Site Horizona Depth b.d.b Ctotal pHCaCl2 Rootsc

[cm] [gcm‐3] [ % ] [‐] [‐]
O 0 to 1 n.d. n.d. 4.4 n.d.
Ah1 1 to 40 0.35 18.20 ± 2.7 4.0 ± 0.1 common v.f., few f., very few m.
Ah2 40 to 70 0.42 ± 0.01 13.07 ± 0.6 4.3 ± 0.0 few v.f., very few f.
G7 2Ahb 70 to 130 0.36 ± 0.04 17.62 ± 2.9 4.2 ± 0.1 few v.f., very few f.
2Bwb 130 to 147 0.71 5.30 4.6 few v.f., very few f.
2Bsb 147 to 150 n.d. n.d. n.d. n.o.
2BCb 150 to 170+ 0.93 0.38 5.4 n.o.
O1 0 to 1 n.d. n.d. 3.9 n.d.
O2 1 to 35 0.10 50.11 ± 0.2 2.5 ± 0.1 common v.f. to f., few m., very few c.
Ah 35 to 75 0.45 ± 0.00 15.58 ± 3.0 3.9 ± 0.3 very few v.f. to c.
Bw 75 to 105 0.57 ± 0.01 9.51 ± 0.9 4.2 ± 0.0 very few v.f. to m.
G5a 2Ah1b 105 to 125 0.48 13.34 4.1 very few v.f. to f.
2Ah2b 125 to 140 0.44 12.46 4.1 very few v.f. to f.
2Ah3b 140 to 175 0.42 12.01 4.3 very few v.f. to f.
2Bsb 175 to 177 n.d. n.d. n.d. n.o.
2BCb 177 to 190+ 0.92 0.35 5.1 n.o.
O 0 to 1 n.d. n.d. 4.1 n.d.
Ah 1 to 45 0.37 ± 0.05 17.71 ± 3.7 4.1 ± 0.2 common v.f., few f., very few m. to c.
Bw1 45 to 70 0.54 9.15 4.4 few v.f., very few f.
Bw2 70 to 115 0.57 ± 0.02 8.48 ± 0.6 4.4 ± 0.0 few v.f., very few f.
G5b
2Ahb 115 to 175 0.48 ± 0.07 10.92 ± 1.0 4.3 ± 0.1 very few v.f. to f.
2Bwb 175 to 192 0.61 7.13 4.6 n.d.
2Bsb 192 to 194 n.d. n.d. n.d. n.o.
2BCb 194 to 200+ 0.94 0.29 5.2 n.o.
O1 0 to 1 n.d. n.d. 3.8 n.d.
O2 1 to 5 n.d 42.87 3.7 many v.f. to f., common m., very few
c.
Ah1 5 to 40 0.42 ± 0.06 20.21 ± 6.6 3.7 ± 0.4 few v.f. to f., very few m.
A7a
Ah2 40 to 80 0.51 ± 0.07 9.45 ± 2.1 4.2 ± 0.1 few v.f. to f., very few m.
2Ahb 80 to 165 0.41 ± 0.03 13.9 ± 2.1 4.2 ± 0.1 very few v.f. to f.
2ABb 165 to 175 0.66 5.60 4.5 n.o.
2BCb 175 to 195+ 0.97 0.70 5.2 n.o.
O 0 to 1 n.d. n.d. 4.6 n.d.
Ah1 1 to 25 0.43 16.97 4.1 common v.f. to f., few m., very few c.
Ah2 25 to 70 0.52 ± 0.05 9.21 ± 1.6 4.5 ± 0.1 few v.f. to f. and very few m.
A7b
2Ahb 70 to 135 0.40 ± 0.04 14.61 ± 2.1 4.3 ± 0.0 few v.f. to f. and very few m.
2ABb 135 to 155 0.66 6.43 4.7 very few v.f. to f.
2BCb 155 to 177 0.99 0.78 5.2 n.o.
a
= horizons named according to FAO guidelines (1990), horizons sampled for 14C analysis indicated in bold
b
= dry bulk density
c
= as described in the field according to FAO guidelines (1990), v.f. = very fine roots, f. = fine roots, m. = medium
roots, c. = coarse roots, n.d. = not determined; n.o. = not observed.
27
14 14
Figure 2.2 C age (BP) and location of the sample in the soil profile, showing that the C age of HA is
14
generally older than the C age of humin and that the age difference increases with depth.
z = humic acids U = humin

28
Contamination with roots may well explain the observed younger 14C ages of humin.
Although the humin fraction is often considered as the OM fraction containing the most
resistant and immobile organic molecules (Stevenson 1994; Pessenda et al. 2001), some
authors report that it is also the fraction that is rejuvenated considerably by the inclusion of
relatively young roots (Orlova & Panychev 1993). In accordance, Nierop & Buurman (1999)
and Nierop et al. (1999) demonstrated with pyrolysis GC/MS that organic molecules of fresh
roots accumulate in the humin fraction. The larger age differences with older 14C ages may
be explained by the prolonged exposure to contamination by roots. Additionally, the
rejuvenating impact of fresh roots is higher on samples with low 14C activity than on samples
with high 14C activity (Mook & Van de Plassche 1986).

14 13
Table 2.3 C ages (BP) and δ C (‰) values of HA and humin in the Andosols.

Humic acid Humin
14
Site Horizon Sample depth CIO code C agea δ13C CIO code 14
C agea δ13C
[cm] [BP] [‰] [BP] [‰]
Ah1 10.0 to 10.5 GrA ‐28130 180 ± 35 ‐25.24 GrA ‐28117 45 ± 35 ‐25.67
G7 Ah2 46.5 to 47.0 GrA ‐28134 1690 ± 40 ‐25.22 GrA ‐28131 1405 ± 35 ‐25.66
2Ahb 89.5 to 90.0 GrA ‐28126 4910 ± 45 ‐24.98 GrA ‐28135 4130 ± 40 ‐25.47
F 14.5 to 15.0 GrA ‐ 30109 ‐623 ± 35 ‐27.46 GrA ‐ 30108 ‐656 ± 35 ‐28.83
(108.1 ± 0.4) (108.5 ± 0.4)
F 34.0 to 34.5 GrA ‐ 30112 ‐28 ± 35 ‐26.39 GrA ‐ 30110 ‐218 ± 35 ‐26.93
G5a (100.4 ± 0.4) (102.8 ± 0.4)
Ah 45.0 to 45.5 GrA ‐28102 490 ± 35 ‐25.67 GrA ‐28101 775 ± 35 ‐25.93
Bw 84.5 to 85.0 GrA ‐28104 2170 ± 35 ‐25.92 GrA ‐28103 2005 ± 35 ‐26.56
2Ah1b 114.5 to 115.0 GrA ‐28108 4355 ± 40 ‐25.10 GrA ‐28106 3360 ± 35 ‐26.13
Ah 14.5 to 15.0 GrA ‐30138 590 ± 35 ‐25.33 GrA ‐30136 425 ± 35 ‐25.69
Ah 34.5 to 35.0 GrA ‐28111 1350 ± 35 ‐25.45 GrA ‐28109 1270 ± 35 ‐26.08
G5b
Bw2 75.0 to 75.5 GrA ‐28113 3465 ± 35 ‐25.51 GrA ‐28112 2615 ± 35 ‐26.09
2Ahb 124.5 to 125.0 GrA ‐28116 5730 ± 40 ‐24.91 GrA ‐28115 4945 ± 40 ‐25.16
Ah1 30.0 to 30.5 GrA ‐28129 445 ± 35 ‐26.08 GrA ‐28127 555 ± 35 ‐25.95
A7a Ah2 64.5 to 65.0 GrA ‐30153 2050 ± 35 ‐25.24 GrA‐ 30154 1505 ± 35 ‐24.94
2Ahb 99.5 to 100.0 GrA ‐28140 4715 ± 40 ‐24.17 GrA ‐28139 3930 ± 40 ‐24.56
Ah1 15.0 to 15.5 GrA ‐28144 595 ± 35 ‐24.68 GrA ‐28141 310 ± 35 ‐25.02
A7b Ah2 54.5 to 55.0 GrA ‐28146 2540 ± 40 ‐25.15 GrA ‐28145 2320 ± 35 ‐25.60
2Ahb 85.0 to 85.5 GrA ‐28149 4060 ± 45 ‐24.55 GrA ‐28147 3470 ± 40 ‐24.95
a
= For samples with measured activities higher than the standard, the 14C ages expressed in BP are negative
numbers. In these cases, we also show the activity ratios in %.

However, how accurate are the 14C ages of the HA fractions? It is highly unlikely that HA is
mobile in the mineral part of Andosols, because of the high metal/SOM ratio combined with
the very low pH in these soils. Therefore, contamination by illuviation of younger HA is
negligible. In addition, undisturbed Andosols show strong resistance to water erosion, due
to rapid rainfall infiltration and high aggregate resistance to dispersion (Shoji et al. 1993). In
the field we did not observe signs of erosion or deposition on our sites. Hence, deposition of
allochtonous OM is not expected to have a dominant influence. Some earthworm species
are known to burrow up to 1 m depth (Lee & Foster 1991) and Barois et al. (1998) observed
29
earthworm casts up to 60 cm depth in Mexican Andosols at 3100 m a.s.l. (mean annual
temperature 5 ˚C to 10 ˚C and pHKCl at 20 cm soil depth = 4.6). However, the strong
correlation between age and depth in the studied soils indicates that bioturbation did not
homogenise SOM at the scale of the applied vertical sampling distances. We thus conclude
that the 14C ages of the HA fractions are more accurate than the 14C ages of the humin
fractions for all samples except the mineral soil samples just beneath a forest floor.

Two factors may explain the younger 14C age of HA compared to that of humin in the
mineral soil samples just beneath a forest floor. Firstly, contrary to the mineral horizons, HA
may be mobile in the ectorganic horizons overlying them in the forest. Upon entering the
mineral horizon these younger mobile HA form complexes with the abundant metals in
Andosols (Al and Fe) and become immobile. Consequently, the 14C age of the HA fraction in
that mineral topsoil is lowered. Secondly, roots concentrate in the relatively nutrient rich
ectorganic horizons rather than in the mineral horizons (Table 2.2), resulting in less
contamination of the humin fraction of the mineral horizon. Accordingly, in the ectorganic
horizon, HA was again older than or of equal age as humin. Possibly, mobile (younger) HA
disappears from the ectorganic horizons by leaching while the residual HA remains
immobile and uncontaminated. Therefore, we conclude that in the mineral soil samples just
beneath a forest floor the humin fraction is more accurate for dating, if roots are removed
adequately. In the ectorganic horizons themselves HA still seems more accurate for dating,
but more data is necessary to support this conclusion.

Páramo fires are common in the northern Andes (Laegaard 1992; Hofstede 1995b; Ramsay
& Oxley 1996), and they may occur even in protected areas. Fire can cause complex
transformations in the humic substances, e.g. from FA to HA and from HA to humin and
black carbon (Gonzalez‐Perez et al. 2004). Since the original humic substances may differ in
radiocarbon age, such fire induced transformations could contaminate the measured 14C age
of a specific OM fraction. However, transitions due to fire are much less if OM is complexed
with metals or minerals (Gonzalez‐Perez et al. 2004) as is the case in Andosols. Furthermore,
during a páramo fire, soil temperatures 2 cm belowground remain relatively low, not
exceeding 65 ˚C (Ramsay & Oxley 1996), making such transitions unlikely (Gonzalez‐Perez et
al. 2004). Therefore, we regard the contaminating effect of fire on radiocarbon age as
minimal.

Our results show that both HA and humin can be used for radiocarbon dating, but accuracy
of either depends on e.g. soil properties and vegetation type, stressing the necessity for
tailor‐made research. Additionally, the 14C ages of both HA and humin remain an average,
influenced by the continuous input of fresh OM (Wang et al. 1996), and must thus be
interpreted as minimum ages.
30
14
C ages of HA and humin in peat

The results of radiocarbon dating of the HA and humin fraction of peat, collected in the
Eastern and Western Cordilleras of northern Ecuador, are presented in Table 2.4. Also here
the δ13C values (Table 2.4) are those of a C3 vegetation, which is to be expected in peat.

14 13
Table 2.4 C ages (BP) and δ C (‰) values of HA and humin of the peat sequences.

Humic acid Humin
14
Site Sample depth CIO code C age δ13C CIO code 14
C age δ13C
[cm] [BP] [‰] [BP] [‰]
146.0 to 147.0 GrA ‐28093 2110 ± 35 ‐25.19 GrA ‐28092 2110 ± 35 ‐25.30
G8 196.0 to 197.0 GrA ‐28097 6075 ± 40 ‐24.95 GrA ‐28095 5980 ± 40 ‐25.11
246.0 to 247 GrA ‐28099 8010 ± 50 ‐25.06 GrA ‐28098 8290 ± 50 ‐24.92
100.0 to 101.0 GrA ‐28057 1640 ± 35 ‐26.20 GrA ‐28055 1610 ± 35 ‐26.33
A11 160.0 to 161.0 GrA ‐28059 3450 ± 35 ‐26.06 GrA ‐28058 3330 ± 35 ‐26.21
222.0 to 222.5 GrA ‐28082 7515 ± 45 ‐26.53 GrA ‐28081 7625 ± 45 ‐26.89
66.0 to 66.5 GrA ‐28086 290 ± 35 ‐26.96 GrA ‐28083 185 ± 35 ‐26.46
A2 176.5 to 177.0 GrA ‐28088 1500 ± 35 ‐27.42 GrA ‐28087 1525 ± 35 ‐26.58
290.0 to 290.5 GrA ‐28091 2630 ± 35 ‐26.78 GrA ‐28089 2940 ± 35 ‐26.46

Contrary to the soils, the difference between the 14C ages of HA and that of humin is
generally small and not significant (p= 0.459), although Shore et al. (1995) demonstrated
that large age differences (up to 1210 14C yr) between these fractions can exist in peat
samples. Several factors explain the similarity in 14C ages of the humic substances in peat.
Firstly, bioturbation is of minor importance. Secondly, the time span of contamination with
roots is limited due to the continuous shifting upwards of the active rooting zone during OM
accumulation (peat growth). Additionally, the depth of the top samples (varying from 60 cm
to 150 cm) is probably already beyond the currently active rooting zone. Finally, although
HA mobility cannot be ruled out, its vertical movement may be limited in stagnant water.

Only in two cases (GrA ‐28099 and GrA ‐28091) was the age difference somewhat larger, the
differences amounting to 310 14C yr and 280 14C yr respectively and humin rendering older
ages. These age differences are small compared to the differences occurring in the soils
(differences up to 995 14C yr) as well as compared to the differences reported by Shore et al.
(1995) for peat sequences (differences up to 1210 14C yr). At this moment we have no
information on the basis of which we can explain the somewhat larger differences for the
two samples, therefore we treat our samples as one statistical population and regard the
differences as not significant.

Our results show that 14C ages of both HA and humin are equally accurate for the peat bogs,
although ages are still averages and should be interpreted as a minimum (Wang et al. 1996).
However, in the active rooting zone the humin ages may become less accurate due to
contamination by fresh roots.
31
SOM age‐depth relationship in Andosols

To evaluate if the chronostratigraphy of SOM in the studied Andosols is suitable for
paleoecological research such as the reconstruction of the natural UFL, we investigated the
relationship between age and depth. Calibrated radiocarbon ages and depth are shown in
Table 2.5. We used the calibrated radiocarbon ages of HA for the age‐depth relationship in
all cases, except for the mineral soil samples just beneath a forest floor where we used
calibrated radiocarbon ages of humin instead. There are no age inversions and no signs of
homogenisation at the applied sampling distances (see Table 2.5).

Table 2.5 Calibrated radiocarbon ages (range and mean, year AD/BC) of the Andosols.

Calibrated agea
Site Horizon Sample depth Range Mean
[cm] [cal AD/BC] [cal AD/BC]
Ah1 10 to 10.5 cal AD 1761 to 1789 cal AD 1775
G7 Ah2 46.5 to 47 cal AD 329 to 407 cal AD 368
2Ahb 89.5 to 90 cal BC 3712 to 3643 cal BC 3678
F 14.5 to 15 cal AD 1996 to 2002 cal AD 1999
F 34 to 34.5 cal AD 1956 to 1957 cal AD 1957
G5a Ah 45 to 45.5 cal AD 1247 to 1271 cal AD 1259
Bw 84.5 to 85 cal BC 352 to 293 cal BC 323
2Ah1b 114.5 to 115 cal BC 3015 to 2944 cal BC 2980
Ah 14.5 to 15 cal AD 1311 to 1359 cal AD 1335
Ah 34.5 to 35 cal AD 648 to 682 cal AD 665
G5b
Bw2 75 to 75.5 cal BC 1780 to 1740 cal BC 1760
A7a Ah2 64.5 to 65 cal BC 105 to 29 cal BC 67
A7b Ah2 54.5 to 55 cal BC 797 to 760 cal BC 779
2Ahb 85 to 85.5 cal BC 2634 to 2562 cal BC 2598
a
= ages derived from humin 14C age instead of humic acid 14C age are depicted in bold font

Since we measured only a few samples per soil profile, we performed linear regressions on
the relationship between calibrated age and depth for all soils grouped together, for the
soils grouped according to Cordillera (Eastern or Western) and for the soils grouped
according to current vegetation type (forest or páramo). Therefore, our linear regressions
cannot be used to interpolate dates in between measured dates for a given soil profile.
However, they can be used to evaluate whether rates of SOM accumulation differ between
the Cordilleras and between current vegetation types. Shoji et al. (1993) stressed the
influence of vegetation on Andosol formation.
32
The linear regressions according to Cordillera are shown in Figure 2.3 and the results of all
linear regressions are additionally presented in Table 2.6. Because of the different nature of
the forest floor we excluded the organic samples from the regressions, but the samples are
included in the figures. All these linear regressions have a good fit (r2 ≥ 0.85) and a narrow
confidence interval (C.I.). However, Telford et al. (2004) demonstrated that, when the
dataset is small, no age‐depth model provides a good fit to reality and that calculated
confidence intervals are too optimistic.

Figure 2.3 Age‐depth relationships Eastern Cordillera and Western Cordillera. 95 % confidence interval
(dashed lines) of linear regression (solid line) included. Regression equations presented in Table 2.6.
= páramo □ = forest (mineral soil samples) U = forest (ectorganic samples)

Table 2.6 Linear regressions of SOM age‐depth relationships of the Andosols, showing that the 95 %
confidence intervals of the slopes are overlapping.

Group Y intercept [cm] X intercept [cal AD] Slopea ± 95 % C.I. r2 nb
All soils 49.77 ± 3.645 3135 ‐0.01588 ± 0.00358 0.87 16
Western Cordillera 47.57 ± 4.606 3097 ‐0.01536 ± 0.00657 0.91 6
Eastern Cordillera 51.16 ± 5.462 3186 ‐0.01606 ± 0.00544 0.85 10
Forest 62.95 ± 5.647 4292 ‐0.01467 ± 0.00781 0.87 6
Páramo 41.78 ± 2.461 2515 ‐0.01661 ± 0.00248 0.97 10
a
= Slope deviates significantly from zero (p < 0.005)
b
= Samples from ectorganic horizon excluded from regressions

The general increase of age with depth (slopes deviate significantly from zero) gives rise to
two possible scenarios concerning the development of the current soil profile. The first
scenario assumes that the current soil consists of only one tephra deposit. The observed
age‐depth relationship can then be explained by continuous burying of OM by younger OM
much like in a peat bog. This scenario is possible with and without (shallow) bioturbation,
because the zone of active bioturbation could shift upwards during SOM accumulation
33
resulting in a crude stratification. The second scenario assumes that the current soil consists
of multiple tephra depositions. SOM is then intermittently buried by tephra and
sedimentation rate mainly determines the age‐depth relationship. Again this scenario is
possible with and without (shallow) bioturbation.

The slopes of the linear regressions for both Cordilleras were surprisingly uniform
(overlapping confidence intervals), suggesting that rates of tephra deposition and SOM
incorporation have been similar. The similarity (overlapping confidence intervals) of the
slopes according to vegetation type shows that the current vegetation type did not strongly
influence the vertical OM distribution in the mineral horizons at the applied sampling
distances. However, during soil profile development vegetation could have been different
from today. The most notable influence of the current vegetation on soil profile
development is the development of ectorganic horizons (dated at ~1957 cal AD) in the
forest. Although we did not include the organic samples in the regression, it is clear that the
age‐depth relationship in this layer is different, according to expectations. Additionally, the
Y intercept of the linear regression of the forest soils (mineral horizons) is located at
somewhat greater depth than the Y intercepts of the other regressions as a result of the
relatively voluminous ectorganic horizons.

Changes in SOM quantity due to fire may affect soil dry bulk density and therefore the age‐
depth relationship. However, Hofstede (1995) demonstrated that soil dry bulk density was
not measurably affected by fire alone in Colombian páramos. As mentioned earlier, the dry
bulk densities in the studied soils were similar for all samples. Additionally, the linear
regressions of the age‐depth relationship were similar for both Cordilleras and similar for
the current vegetation types, giving no reason to suspect that probable differences in e.g.
fire frequency between Cordilleras or vegetation types caused changes in the age‐depth
relationship.

We conclude that the chronostratigraphy of the studied Andosols is very promising for the
use in paleoecological reconstructions, but more research concerning the tephra
stratigraphy is necessary to explain the observed age‐depth relationship. The presence of
(shallow) bioturbation cannot be ruled out. We recommend applying higher resolution
dating of the soil profiles to enable interpolation between dates.

Comparison of time frame soils and peat sequences

Calibrated radiocarbon ages (of HA) of the peat cores are reported in Table 2.7. The
calibrated radiocarbon ages of the Andosols were discussed in the previous section. Both
the current soils (at least up to 1760 cal BC) and the paleosols (at least up to 4565 cal BC)
were found to be of Holocene age. The time frames must be considered as a minimum,
because we did not date the exact events of tephra deposition or the base of the paleosol.
Poulenard et al. (2003) reported similar 14C ages of the bulk SOM fraction < 50 µm of soils in
34
northern Ecuador and dated the base of the paleosol at at least 7838 BP (equalling
approximately 6683 cal BC). Similarly, the three peat bogs are all of Holocene age, enabling
verification of pollen records from soil by comparing them to pollen records from peat bogs.
To enhance comparison, the tephra stratigraphy of the peat bogs should ideally be linked to
the tephra stratigraphy of the soils. The peat sequences of Guandera and El Angel cover a
larger time span (at least up to 6941 cal BC) than the peat of Los Encinos (at least up to 809
cal BC). This independent evidence supports the time frame we report for the studied
Andosols.

Table 2.7 Calibrated radiocarbon ages (range and mean, year AD/BC)
of the peat sequences.

Calibrated age
Site Sample depth Range Mean
[cm] [cal AD/BC] [cal AD/BC]
146.0 to 147.0 cal BC 180 to 91 cal BC 136
G8 196.0 to 197.0 cal BC 5041 to 4939 cal BC 4990
246.0 to 247 cal BC 6882 to 6827 cal BC 6855
100.0 to 101.0 cal AD 379 to 434 cal AD 407
A11 160.0 to 161.0 cal BC 1776 to 1734 cal BC 1755
222.0 to 222.5 cal BC 6438 to 6363 cal BC 6401
66.0 to 66.5 cal AD 1522 to 1577 cal AD 1550
A2 176.5 to 177.0 cal AD 542 to 603 cal AD 573
290.0 to 290.5 cal BC 820 to 797 cal BC 809

Conclusions

Our results show that in Andosols, radiocarbon dating of the HA fraction renders more
accurate results than radiocarbon dating of the humin fraction, except for mineral soil
samples just beneath a forest floor where the opposite holds true. These results
demonstrate that the accuracy of the radiocarbon age of a specific OM fraction depends on
e.g. soil properties and vegetation, underlining the necessity of tailor‐made radiocarbon
dating of soils. In the peat sequences both OM fractions yield accurate results. The
chronostratigraphy of SOM in the studied Andosols appears to be suitable for
paleoecological research such as reconstruction of the natural position of the UFL, because
age inversions and homogenisation were lacking at the applied vertical sampling distances
and age increased with depth. However, higher resolution dating of the soil profiles and
more research concerning the tephra stratigraphy is necessary to be able to interpolate
dates using the age‐depth relationship. The time frame covered by the studied Andosols is
similar to that of peat sequences nearby enabling verification of the soil pollen records.
With this research, we made an indispensable first step towards reconstruction of the
natural position of the UFL in northern Ecuador.

35
Acknowledgements

We gratefully acknowledge the Ecuadorian Ministerio del Ambiente for issuing the necessary
permits for conducting our research in northern Ecuador and for supporting us in El Angel
Ecological Reserve. Jatun Sacha and Grupo Randi Randi are thanked for allowing us to
perform the research in Guandera Biological Station and Los Encinos Biological Station
respectively and for helping us in the field. Ecopar is acknowledged for facilitating our
research in the form of office assistance. Furthermore, we wish to thank Leo Hoitinga, Leen
De Lange, Ton Van Wijk and Niko de Wilde‐de‐Ligny for their assistance in the laboratory.
Finally, we acknowledge Marcela Moscol, Antoine Cleef, Jan Sevink, Robert Hofstede, Emiel
Van Loon, Jan Van Mourik, Klaas Nierop, Marjolein Van der Linden and Maartje Van
Meeteren for their valuable suggestions leading to the improvement of this manuscript. This
research was funded by WOTRO (WAN 75‐405) and generously sponsored by Fjällraven in
the form of clothing and gear.
36
37
3 Tephra stratification of volcanic ash soils in
northern Ecuador
Published in Geoderma 144: 231–247, 2008,
Tonneijck FH, Hageman JA, Sevink J, Verstraten JM.

Abstract

We combined proxies traditionally used in stratigraphic research (mineral assemblages,
grain size distribution, and chemical element ratios) with soil organic carbon contents and
radiocarbon dating both at a high vertical resolution, to unravel the tephra stratigraphy in
volcanic ash soils. Our results show that soil profiles along an altitudinal transect
intersecting the upper forest line in northern Ecuador were formed in three distinct tephra
deposits. Although the deposits contained a similar assemblage of minerals, we were able to
differentiate these deposits because of their characteristic organic carbon distribution, grain
size distribution and typical SrO to Na2O, CaO and crystalline Al2O3 ratios. Unravelling the
tephra stratigraphy improved understanding of the vertical distribution of soil organic
carbon, including paleoecological proxies, in the studied soils. We demonstrated that
bioturbation likely plays an important role in current pedogenesis, resulting in overprinting
(merging, mixing) of the paleosol. Surprisingly, in spite of bioturbation, a linear age depth
relationship exists, leading to the hypothesis that the active zone of bioturbation shifted
upwards during soil formation. Therefore, we conclude that paleoecological proxies are
stratified in our soils, albeit probably somewhat more crudely than in undisturbed peat bogs
or lake sediments.

Keywords
volcanic ash soil, paleoecology, stratification, soil organic matter, element ratios, grain size
distribution
39
Introduction

Volcanic ash soils (Andosols) offer great opportunities for paleoecological analyses (Bakker
& Salomons 1989; Tonneijck et al. 2006), due to their characteristic accumulation of soil
organic matter (SOM, Shoji et al. 1993; Dahlgren 2004), including paleoecological proxies
such as pollen and biomarkers. However, a thorough understanding of the soil
chronostratigraphy is required for correct interpretation of paleoecological records that are
contained in soils (Keatinge 1983; Davidson et al. 1999).

Boundary conditions for the vertical distribution of SOM (including paleoecological proxies)
in Andosols are set by volcanic activity, specifically through the amount and timing of tephra
deposition. However, the actual vertical distribution of SOM depends on subsequent
pedogenetic processes; e.g. aboveground and belowground biomass input, organic matter
decomposition and bioturbation. In the field, the number of tephra layers is often
underestimated since soil formation following tephra deposition may overprint preceding
(buried) deposits (Buurman et al. 2004) and hence mineralogical, chemical and physical
analyses are needed to unravel the tephra stratification. Overprinting occurs when the
thickness of a tephra deposit is not sufficient to fully interrupt SOM accumulation and
bioturbation of the preceding soil. Since overprinting may cause underestimation of
radiocarbon ages and erroneous interpretation of paleoecological records, it is particularly
important to recognize overprinted paleosols. On the other hand, distinguishing (very) thin
tephra layers (< 5 cm) if possible at all is less important, because they do not interrupt SOM
incorporation nor cause underestimation of radiocarbon ages.

Unfortunately, the application of standard mineralogical, chemical and physical analyses for
the identification of tephra layers is not straightforward in soils in which overprinting may
have occurred. Overprinting is common in soils developed in distal facies with complex
tephra chronology. Ratios of stable lithogenic elements (SLE), e.g. Al, Fe(III), Ti, Zr, are often
used to distinguish tephra layers, because contents of SLE change during weathering while
their ratios do not (Brimhall & Dietrich 1987; Chadwick et al. 1990; Buurman et al. 2004).
However, successive tephra layers may be chemically very similar and therefore ratios of SLE
are not necessarily the most suitable for differentiation (Buurman et al. 2004).
Unfortunately, ratios of mobile lithogenic elements (MLE), e.g. Si, Na, Ca, Fe(II) to SLE reflect
weathering rather than lithology and therefore cannot be used without a thorough
understanding of the weathering processes that have taken place. Since reference, i.e.
unweathered, tephra layers are frequently lacking in soil profiles and thick tephra deposits
are often far from homogeneous, mass balance equations to clarify and quantify element
behaviour upon weathering (Chadwick et al. 1990) are not reliable in many cases. SLE are
not necessarily inert, but may be stabilised in situ after release upon weathering, instead of
being leached like MLE (Chadwick et al. 1990). To overcome the above problems, SLE should
be corrected for (amorphous) secondary constituents, thus eliminating the opposing
behaviour of MLE versus SLE upon weathering. In relatively young volcanic ash soils, such as
40
the studied soils, the clay fraction is dominated by amorphous constituents (Torn et al.
1997) and therefore correction for amorphous constituents offers a reliable approximation
of primary SLE. Furthermore, when amorphous secondary constituents are dominant and
CEC is very low, many MLE behave similarly upon weathering. Ratios between MLE and of
MLE to primary SLE then reflect element ratios within primary minerals and these ‘primary
elemental ratios’ enhance recognition of tephra deposits.

In addition, Buurman et al. (2004) demonstrated that in some cases differences in grain size
distribution (GSD) enable distinction between chemically similar tephra deposits. During
weathering, much of the characteristics of the primary GSD (i.e. on a clay‐free basis) remain
intact (Buurman et al. 2004). Obtaining reliable GSDs for volcanic ash soils is difficult due to
aggregate resistance to dispersion, but it is not impossible if optimal pH adjustment is
achieved during the sieve‐pipette method (Shoji et al. 1993; Nanzyo 2002).

The aim of the current paper was to unravel the tephra stratigraphy of soil profiles along an
altitudinal transect intersecting the upper forest line (UFL) in northern Ecuador, thus
increasing understanding of the vertical distribution of SOM (including paleoecological
proxies) in volcanic ash soils. To reach our aim, we used proxies traditionally used in tephra
stratigraphy (mineral assemblages, chemical element ratios, GSDs) in combination with the
vertical distribution of SOM and high resolution radiocarbon dating.


Geological setting

The Ecuadorian Andes consists of two mountain ranges trending northeast‐southwest,
called the Western Cordillera and the Eastern Cordillera, separated by the Inter Andean
valley (Sauer 1971). The northern part of the Ecuadorian Andes is characterised by Cenozoic
volcanic rocks (Hörmann & Pichler 1982). Those of the Western Cordillera belong to a calc‐
alkaline andesite‐dacite series, while in the Eastern Cordillera they are members of the
andesite‐dacite‐rhyodacite series (Hörmann & Pichler 1982; Stern 2004). Our study area in
northern Ecuador was covered by at least two volcanic ash deposits of Holocene age
(Tonneijck et al. 2006). Several volcanoes near the study area have been active during the
Holocene (Droux & Delaloye 1996; Instituto Geofisico 2006).

Study area

The study sites are located within the nature protection area of Guandera Biological Station
in northern Ecuador near the border with Colombia, Figure 1.1 and Table 1.1. Guandera is
situated in the Eastern Cordillera and has a (semi‐)natural UFL at approximately 3650 m
above sea level (a.s.l.). Mean annual precipitation is around 1900 mm and mean annual
temperature ranges from 12 °C at 3000 m a.s.l. to 4 °C at 4000 m a.s.l. We selected an
41
altitudinal transect from 3330 to 3990 m a.s.l. intersecting the UFL, starting with a site in a
forest patch surrounded by agricultural lands (G11), followed by sites currently covered by
continuous forest (G1, G2), a site in a forest patch just above the UFL (G5a) and 5 sites
currently covered by tropical alpine grassland (páramo) vegetation (G6, G7, G9, G10) of
which one was located next to the forest patch (G5b). For a proper tephra stratigraphic
analysis, colluvial and related transported materials were excluded from the current
research. Naturally, we avoided areas with clear erosional/depositional features in the field.
One soil profile (G4) without visual signs of disturbance in the field was still recognizable as
colluvial by anomalous clustering results caused by deviating GSD and element ratios and
subsequently discarded for further research.

Sampling procedures

Soil pits of approximately 1.5 m2 surface area and a depth of 1.5 m to 2 m depending on the
soil profile, were excavated and soil profiles were described according to the FAO guidelines
(FAO 1990) and classified according to the World Reference Base (FAO 2006). In the field,
horizon boundaries were identified based on changes in colour and/or morphology. At site
G11 we used an exposure along a roadcut instead of a soil pit, after setting it back for at
least 50 cm. We took bulk soil samples for chemical analyses (approximately 1 kg field moist
material) and ring samples (with a volume of 100 cm3) for bulk density determination in the
same soil pit at regular depths but respecting horizon boundaries. Bulk and ring samples
were taken over a vertical interval of approximately 5 cm, the middle of which was noted as
the depth of the sample. Additionally, we took undisturbed box samples (with dimensions of
7h x 5w x 4d cm) for micromorphological analysis in a selection of soil pits (notably G1, G5a,
G5b, G6, G7 and G9) at the same depths as bulk and ring samples. In these selected pits, we
also took undisturbed vertical soil samples (monoliths) using one or two metal gutters of
75h x 5w x 4d cm. Monoliths were packed in aluminium foil and then plastic foil and finally
tightly sealed with adhesive tape to avoid dislocation and drying during transport. For
radiocarbon dating and additional measurement of organic carbon contents (further
referred to as monolith organic carbon contents), samples were cored from the monoliths
using a corer with 0.5 cm and 0.75 cm diameter respectively. All samples were stored at 2 °C
under field moist conditions prior to analysis.


Soil pH was measured with a glass electrode in a 0.01 M CaCl2 soil solution (w/v: 1:5 mineral
samples and w/v 1:10 organic samples, both on an air dry basis, using fieldmoist samples).
Total carbon was measured with a VarioEL (Elementar) CNS auto‐analyser. Total carbon
equals organic carbon, since carbonates were found to be absent. Dry bulk density was
determined by weighing oven‐dried ring samples (105 °C, 44 h mineral samples and 70 °C,
48 h organic samples).

42
In addition to the radiocarbon data already published (Tonneijck et al. 2006), we performed
dating for profiles G5a and G5b to obtain a higher vertical resolution and acquired a first set
of 14C dates for profile G9. Radiocarbon dating and calibration followed procedures outlined
in Tonneijck et al. (2006). In short, fractions of SOM, as obtained by alkali and subsequent
acid extraction according to Schnitzer (1982), were radiocarbon dated with an Accelerator
Mass Spectrometer. Fulvic acid was discarded because of its inherent mobility in most soils.
Radiocarbon dating of the humic acid fraction rendered more accurate results in all cases
except for mineral soil samples just beneath the forest floor where humin gave more
accurate results (Tonneijck et al. 2006). Since age‐depth relationships only make sense
when using calibrated radiocarbon ages (Telford et al. 2004) we transformed 14C ages to a
calendar year probability distribution (cal AD/BC) with the WinCal25 calibration software
(Van der Plicht 2005). Calibrated radiocarbon ages are expressed as the range of calendar
years of the 1 sigma peak with the largest relative area under the probability distribution.
Since the ranges are small, calculations are performed with the mean and the dots displayed
in the graphs represent the whole range of calendar years at the scale of the graph. Since
our focus was on tephra stratigraphy and not on ectorganic soil horizons which owe their
existence to superficial OM accumulation, we set the zero depth at the mineral soil surface
instead of at the ectorganic soil surface as was done in Tonneijck et al. (2006).
GSDs of the fine earth fraction were estimated with the sieve‐pipette method, after SOM
removal with H2O2 (Gee & Or 2002). To avoid irreversible aggregation upon drying (Kubota
1972; Shoji et al. 1993), we used fieldmoist samples. We applied either NaOH or HCl for pH
adjustment to achieve maximum chemical dispersion, according to Shoji et al. (1993) and
Wada (2002). To ensure equally sized classes on a logarithmic scale, we grouped classes 63 –
75 μm and 75 ‐ 105μm together. The high r2 (0.75, excluding unweathered ash samples) of
the linear regression between clay percentage and the sum of calculated allophane and
ferrihydrite contents suggests that dispersion was effective (Shoji et al. 1993) and the sieve‐
pipette method with pH adjustment thus rendered reliable GSDs. Only one sample (G10A20)
flocculated both at low and high pH values rendering clay dispersion impossible. To
approach the primary GSDs as much as possible, we present the results on a clay free basis.
Acid ammonium oxalate (AAO) extractable Al, Si and Fe (Alo, Sio and Feo) and sodium
pyrophosphate extractable Al and Fe (Alp and Fep) were determined in fieldmoist samples
(fine earth fraction) as described by Van Reeuwijk et al. (2002). Allophane contents were
then calculated according to the formula proposed by Parfitt & Wilson (1985) and modified
by Mizota & Van Reeuwijk (1989):

Equation 3.1
%Sio
%Allophane = 100
(23.4 − 5.1x)
with x = (Alo‐Alp)/Sio representing the atomic ratio of allophane. As recommended by Mizota
& Van Reeuwijk (1989) we applied a maximum (Alo‐Alp)/Sio atomic ratio of 2.5, assuming
that an excess in Al should be allocated to hydroxy interlayered Al.
43
Ferrihydrite content was estimated using Feo, according to Parfitt & Childs (1988) and Childs
et al. (1991):

%Ferrihydrite = 1.7%Feo Equation 3.2

For the estimation of total element concentrations in ground and dried samples (fine earth
fraction), a Perkin Elmer Optima 3000 XL ICP‐OES (ICP) was used after destruction of the
sample in a hot HF/H2SO4 mixture (Jackson et al. 1986). Fe(II) content of the sample in a cold
HF/H2SO4/H3BO3 solution was determined by colorimetric detection after complexation with
1.10‐phenantroline (Begheijn 1979). In the same sample Fe(II) was determined again, after
reduction of all Fe(III) to Fe(II) using hydroquinone (Begheijn 1979). The total Fe that was
colorimetrically determined was scaled to the total Fe measured by ICP and Fe(III)
concentrations were adjusted accordingly. Loss on ignition (LOI) was determined at 950 °C.
The oxide equivalents of the total element concentrations were transformed to LOI‐free
weight percentages. Total SiO2 content was then calculated by subtracting the sum of the
weight percentages of the other oxides from 100 %.

To correct SLE for secondary constituents, we subtracted amorphous oxide contents of Al,
Fe(III), Si and Ti as measured by AAO (addition ‘o’) from their total oxide contents (addition
‘t’) rendering crystalline oxide contents (addition ‘c’). To verify the assumption that the clay
fraction of the studied soils was dominated by amorphous constituents, we analysed the
clay fraction of samples with a clay content > 20 % (n = 11). Amorphous Al and Fe as
measured with AAO in the bulk sample represented > 80 % of the Al and Fe as measured in
the clay fraction. Amorphous Si as measured with AAO (bulk sample) was on average 30 %
of Si in the clay fraction, but since the vast majority of Si was present in primary constituents
(including volcanic glass), correction for either amorphous or crystalline secondary Si
rendered similar results. We therefore concluded that correction for amorphous
constituents yielded a reliable approximation of primary SLE.

Heavy liquid separation of minerals was performed for one representative soil profile (G5a)
on the sand fraction (63 µm to 2 mm) separated during the sieve‐pipette method, using
polytungstate diluted with demineralised water to a density of 2.80 gcm‐3. We performed X‐
ray diffraction of both low and high density minerals with a Guinier camera for
identification, in addition to X‐ray diffraction scanning of the total silt and sand fractions.
Moreover, the magnetic fraction was separated from the sand fraction using a magnet and
then X‐ray diffracted and identified.

Fieldmoist undisturbed box samples were impregnated after replacing water by acetone
according to Miedema et al. (1974). Thin sections were prepared according to Murphy
(1986). Minerals were identified in these thin sections based on their optical properties
44
according to Nesse (2004) with a petrographic microscope to supplement the X‐ray
diffraction analysis when necessary.

Grain size statistics were computed with the Gradistat program (Blott & Pye 2001) applying
the method of moments (Krumbein & Pettijohn 1938). Results are reported in the phi scale
(Krumbein 1934), where φ = ‐log2 d, with d = grain diameter in mm and using descriptive
terms as suggested by Blott & Pye (2001). As a measure of sorting (Blott & Pye 2001) we
additionally report the standard deviation (σ).

Pearsons correlations (bivariate) between variables were calculated using the SPSS Correlate
procedure and considered significant when p < 0.01. Linear regressions were performed in
Graphpad Prism (Graphpad Software Inc. 1999). Differences in means were considered
significant when p < 0.05, as determined with an analysis of variance, carried out with the
SPSS One‐way ANOVA procedure (SPSS Inc. 2001). If equal variances could be assumed we
applied the Tukey post hoc multiple comparison to find out which groups differed from each
other. In all other cases we used Tamhane’s T2.

Several (combinations of) data were used as input in a hierarchical clustering procedure
using Matlab 7.1 (The Mathworks, Inc. 1984–2003). A hierarchical cluster tree was created
with Ward’s method (Ward 1963), using the Euclidean distance as the distance measure.
Ward’s method is a minimum variance method that aims at finding compact, spherical
clusters. Before clustering, the data was autoscaled. After autoscaling, absolute values do
not matter anymore, but objects that have co‐varying behaviour will be clustered together.

Results

Field description & soil properties

Soil properties are reported in Table 3.1. The soils changed along the altitudinal transect
from Histosols with andic properties at sites currently covered by forest, via an Andic
Cambisol in the forest patch above the UFL, to Andosols at sites currently covered by
páramo vegetation (FAO 2006, Table 1.1). The forest profiles had ectorganic horizons with a
combined thickness of up to 100 cm, while ectorganic horizons barely exceeded 1 cm in the
páramo profiles. All soils showed a multisequum as based on field descriptions, containing
at least one paleosol, except for site G10 at 3905 m a.s.l. Soil structure was massive. Most
soil profiles had a thin placic horizon at the field boundary between the paleosol and the
unweathered ash. In general, the upper mineral horizons were characterised by a very high
organic carbon content (8 % to 22 %), acidic pH (pHCaCl2 3.2 to 4.4) and low dry bulk density
(0.3 to 0.7 gcm‐3).
45
Table 3.1 Selected soil properties and re‐horizonation of samples as based on carbon contents and clustering
on GSD or primary SrO ratios. Final horizonation presented in last column.
Soil properties Rehorizonation
Site Field Sample Deptha C pHb b.d.c Allod Ferre Sandf Clayg Ch GSD Primary Final
[‐] horizons [‐] [cm] [%] [‐] [gcm‐3] [%] [%] [%] [%] SrO ratios horizons
Ah1 G11 B20 10 7.8 4.22 0.67 6.7 1.1 18.7 20.0 1 1 1 Ah1
Ah1 G11 C40 30 7.9 4.30 0.67 8.8 2.0 28.9 22.4 1 1 2 1/2Ah1
G11 2Ahb G11 D57 47 8.3 4.30 0.62 10.7 2.5 44.6 17.9 2 2 2 2Ahb
2?BCb G11 E 110 100 0.27 5.23 0.99 2.9 0.4 82.2 3.1 3 3 3 3BCb
2?Cb G11 F 170 160 0.08 5.32 0.89 1.8 0.3 90.1 1.2 3 3 3 3Cb
Ah G1 E82 7 21.0 3.18 0.43 0.0 0.8 13.6 3.9 1 1 1 Ah
Bw G1 F100 25 9.1 4.01 0.67 3.2 0.5 25.2 13.2 * 1 1 1/2Bw
2Ahb G1 G125 50 13.4 4.07 0.50 8.6 3.8 28.9 22.0 2 1 2 2Ahb
G1
2?BChb Q1 I 150 75 4.0 4.45 n.d. 9.4 2.2 66.3 8.9 3 3 3 3BChb
2?BCb Q1 J 160 85 0.67 5.05 n.d. 6.0 0.6 86.7 1.4 3 3 3 3BCb
2?Cb Q1 K 240 167 0.09 5.20 n.d. 2.2 0.2 90.0 0.4 3 3 3 3Cb
Ah G2 E105 7 19.8 3.76 0.38 0.0 0.6 14.1 5.2 1 1 1 Ah
G2 Bw G2 F135 37 8.2 4.12 0.69 1.4 0.2 23.1 11.2 1 1 1 Bw
2Ahb G2 G170 72 20.3 3.85 0.48 0.0 0.8 32.1 9.6 2 1 2 2Ahb
Ah G5a D45 10 17.7 3.67 0.45 0.0 1.3 21.0 8.2 1 1 1 Ah
Ah G5a E60 25 13.5 4.05 0.45 3.6 1.5 20.3 14.7 1 1 1 Ah
Bw G5a F85 50 8.9 4.19 0.58 3.9 1.5 27.3 16.5 1 1 1 Bw
Bw G5a G95 60 10.1 4.15 0.56 5.4 2.1 31.4 16.3 * 1 2 1/2Bw
G5a
2Ah1b G5a H115 80 13.3 4.05 0.48 4.8 2.1 37.8 16.7 * 2 2 1/2Ah1b
2Ah2b G5a I130 95 12.5 4.11 0.44 6.7 2.1 43.2 18.5 2 2 2 2Ah2b
2Ah3b G5a J160 125 12.0 4.26 0.42 11.2 2.0 40.3 23.1 2 2 2 2Ah3b
2?BCb G5a L185 150 0.35 5.11 0.92 2.7 0.4 81.4 3.0 3 3 3 3BCb
Ah G5b A15 15 20.3 3.95 0.34 0.0 1.1 18.1 12.1 1 1 1 Ah
Ah G5b B35 35 15.1 4.17 0.41 1.8 1.2 20.9 12.3 1 1 1 Ah
Bw1 G5b C55 55 9.1 4.44 0.54 5.6 1.3 29.4 15.8 1 1 1 Bw1
Bw2 G5b D75 75 8.0 4.36 0.58 4.7 1.3 51.7 12.0 * 2 2 1/2Bw2
Bw2 G5b E95 95 8.9 4.41 0.55 6.2 2.3 42.2 16.0 * 2 2 1/2Bw2
G5b
2Ahb G5b F125 125 10.3 4.26 0.55 5.4 2.4 51.3 13.7 2 2 2 2Ahb
2Ahb G5b G145 145 10.3 4.35 0.48 7.8 2.0 50.5 17.6 2 2 2 2Ahb
2Ahb G5b H160 160 12.1 4.40 0.41 10.4 2.2 37.1 25.0 2 2 2 2Ahb
2Bwb G5b I180 180 7.1 4.64 0.61 11.0 1.3 45.5 20.3 2 2 2 2Bw1b
2?BCb G5b J 199 199 0.29 5.24 0.94 2.4 0.3 84.4 2.0 3 3 3 3BCb
Ah G6 A15 15 21.9 3.90 0.33 0.0 1.2 18.1 9.9 1 1 1 Ah
Ah G6 B30 30 19.2 4.05 0.36 0.0 1.3 19.5 8.2 1 1 1 Ah
AB G6 C55 55 11.6 4.40 0.43 5.5 1.9 22.7 17.9 1 1 1 AB
AB G6 D70 70 14.2 4.38 0.41 6.4 1.9 22.7 20.9 * 1 1 1/2AB
G6
2Ahb G6 E100 100 20.3 4.18 0.34 3.1 2.0 29.7 18.2 * 1 2 1/2Ahb
2Ahb G6 F120 120 15.0 4.29 0.37 8.4 1.8 36.6 24.5 2 2 2 2Ahb
2Bwb G6 H 160 160 9.4 4.52 0.51 10.1 1.8 43.4 22.3 2 2 2 2Bw1b
2?BCb G6 I170 170 1.2 4.90 0.97 3.5 1.1 78.9 6.6 3 3 3 3BCb
Ah1 G7 B10 10 20.1 3.92 0.35 0.0 1.4 17.7 12.5 1 1 1 Ah1
Ah1 G7 C30 30 16.3 4.10 n.d. 1.3 1.5 18.1 12.4 1 1 1 Ah1
Ah2 G7 D47 47 12.7 4.30 0.42 4.9 1.9 20.3 17.0 1 1 1 Ah2
Ah2 G7 E63 63 13.5 4.34 0.41 6.3 2.0 22.4 19.6 * 1 1 1/2Ah2
G7 2Ahb G7 F90 90 19.6 4.16 0.34 5.1 2.2 27.9 19.3 * 1 2 1/2Ahb
2Ahb G7 G110 110 15.6 4.24 0.39 9.3 2.1 38.4 23.2 2 2 2 2Ahb
2Bwb G7 H140 140 5.3 4.56 0.71 9.7 0.7 55.2 15.0 2 2 2 2Bw1b
2?BChb Q7 K154 154 2.7 n.d. n.d. 7.1 1.1 77.8 4.2 3 3 3 3BCb
2?BCb G7 I170 170 0.38 5.36 0.93 2.8 0.7 79.1 3.4 3 3 3 3BCb
Ah G10 A20 19 19.6 4.03 0.32 0.4 1.5 20.8 n.d. 1 n.d. 1 Ah
Ah G10 B50 49 11.0 4.43 0.47 5.1 1.7 22.6 15.6 1 1 1 Ah
G10 Ah G10 C 80 79 15.6 4.32 0.39 5.3 1.6 24.9 17.3 2 1 2 2Ah
Ah G10 D110 109 13.0 4.39 0.45 8.1 0.7 38.3 24.2 2 2 2 2Ah
BC G10 E135 134 0.25 5.61 n.d. 2.3 0.4 76.3 2.6 3 3 3 3BCb
Ah G9 A10 10 17.0 4.05 0.33 0.5 1.2 20.2 8.6 1 1 1 Ah
Ah G9 B25 25 12.7 4.15 0.44 1.6 1.1 21.7 10.9 1 1 1 Ah
Bw1 G9 C45 45 9.5 4.28 0.47 4.1 0.4 26.3 14.2 * 1 1 1/2Bw1
G9
Bw2 G9 D60 60 12.8 4.34 0.39 5.2 0.3 26.0 17.3 * 1 1 1/2Bw2
2Ahb G9 E75 75 14.9 4.27 0.46 5.0 0.2 39.2 17.9 2 2 2 2Ahb
2?BCb G9 G90 90 0.27 5.47 0.98 2.8 0.5 85.2 2.6 3 3 3 3BCb
a
= Depth of bulk sample, vertical sample interval 5 cm, the mineral soil surface was set at 0 cm; b = pHCaCl2; c = dry bulk density; d = allophane
content; e = ferrihydrite content; f = sand fraction 63 μm ‐ 2 mm; g = clay fraction < 2 μm; h = * indicates zone with ‘anomalous increase’ in
organic carbon content; n.d. = not determined
46
Radiocarbon dating

Calibrated radiocarbon ages and depths were reported in Tonneijck et al. (2006). We
combined these earlier data with the additional data obtained as part of the present study
in Table 3.2 and Figure 3.1 (vertical sample interval 0.5 cm, the vertical distance between
samples was ~15 cm in profiles G5a and G5b and ~40 cm in other profiles). Only new
findings are discussed in the current paper. Age increased linearly with depth (r2 of linear
regression, excluding ectorganic horizons = 0.92). We found only one age inversion, at the
upper part of profile G5a (sample GrA‐30113). The paleosol upper boundary, as identified in
the field, was dated to 1999 ‐ 2372 cal BC at G5a, to 4053 ‐ 4304 cal BC at G5b and to 2779
cal BC at the bottom of the current soil of G9.

Table 3.2 Conventional (year BP) and calibrated (year AD/BC) radiocarbon ages of the humic acid fraction.

Calibrated age
Site Horizona Depthb Codec 14
C age δ13C Range Mean Ref.
[cm] [BP] [‰] [cal AD/BC] [cal AD/BC]
Ah 10.0 GrA ‐28101 775 ± 35c ‐25.93c cal AD 1247 to 1271c cal AD 1259c *
Ah 24.5 GrA ‐30113 355 ± 35 ‐26.03 cal AD 1467 to 1523 cal AD 1495
G5a Bw 49.5 GrA ‐28104 2170 ± 35 ‐25.92 cal BC 352 to 293 cal BC 323 *
1/2Bw 69.0 GrA ‐30133 3650 ± 40 ‐25.71 cal BC 2041 to 1956 cal BC 1999
1/2Ah1b 70.5 GrA ‐30134 3880 ± 40 ‐25.62 cal BC 2406 to 2337 cal BC 2372
1/2Ah1b 79.5 GrA ‐28108 4355 ± 40 ‐25.10 cal BC 3015 to 2944 cal BC 2980 *
Ah 34.5 GrA ‐28111 1350 ± 35 ‐25.45 cal AD 648 to 682 cal AD 665 *
Bw1 49.5 GrA ‐30139 2070 ± 40 ‐26.45 cal BC 116 to 43 cal BC 80
Bw1 64.5 GrA ‐30107 3045 ± 40 ‐25.79 cal BC 1323 to 1266 cal BC 1295
G5b 1/2Bw2 75.0 GrA ‐28113 3465 ± 35 ‐25.51 cal BC 1780 to 1740 cal BC 1760 *
1/2Bw2 94.5 GrA ‐30238 4270 ± 40 ‐25.81 cal BC 2913 to 2879 cal BC 2896
2Ahb 115.5 GrA ‐30234 5405 ± 40 ‐25.20 cal BC 4327 to 4280 cal BC 4304
2Ahb 124.5 GrA ‐28116 5730 ± 40 ‐24.91 cal BC 4614 to 4515 cal BC 4565 *
Ah1 10.0 GrA ‐28130 180 ± 35 ‐25.24 cal AD 1761 to 1789 cal AD 1775 *
G7 Ah2 46.5 GrA ‐28134 1690 ± 40 ‐25.22 cal AD 329 to 407 cal AD 368 *
1/2Ahb 89.5 GrA ‐28126 4910 ± 45 ‐24.98 cal BC 3712 to 3643 cal BC 3678 *
G9
2Ahb 69.0 GrA ‐30236 4190 ± 40 ‐25.09 cal BC 2810 to 2748 cal BC 2779
a
= depth of sample, vertical sample interval 0.5 cm, the mineral soil surface was set at 0 cm, samples taken
from soil monoliths
b
= Laboratory code of the Centre for Isotope Research, Groningen University, the Netherlands
c
= Humin fraction used for dating instead of HA fraction
* = Tonneijck et al. (2006)

47
Figure 3.1 Calibrated radiocarbon ages (cal AD/BC) versus sample depth (cm), zero depth was set at the
2
mineral soil surface. Linear regression with r = 0.92 (smooth line) and 95 % confidence interval included
(dotted lines).

Vertical distribution of SOM
The vertical distribution of SOM in the soil profiles is presented in Table 3.1 and Figure 3.2.
Generally the organic carbon content of the bulk samples (vertical sample interval 5 cm, the
vertical distance between samples was on average ~15 cm) decreased with depth until the
field boundary with the paleosol, where the bulk organic carbon content increased and then
decreased again with depth down to the unweathered ash. In all cases, except G2, the
increase of bulk organic carbon content already started slightly above the field boundary
between the current soil and paleosol. The monolith organic carbon contents (vertical
sample interval 0.5 ‐ 0.75 cm, the vertical distance between samples was 5 cm) of G5a, G6
and G7 showed that there was a gradual increase (further referred to as the ‘anomalous
increase’ in organic carbon content) starting just above the field boundary between the
current soil and the paleosol and ending just below it. The same ‘anomalous increase’ in
monolith organic carbon contents was present in profile G1, G5b and G9, but in these cases
the field boundary coincided with the maximum organic carbon content.
Grain Size Distributions
GSDs (on a clay free basis) of samples grouped according to the current soil, paleosol and
unweathered ash are presented in Figure 3.3. A reference line, representing a sample with
evenly distributed grain sizes, was added to the figures for easy comparison. Most samples
of the current soil plotted below this reference line and had an average grain size of 5.2 ±
0.5 φ (coarse silt). Most paleosol GSD curves concentrated around the reference line and
showed an average grain size of 4.3 ± 0.5 φ. The GSDs of the unweathered ash samples
were located well above the reference line with an average grain size of 2.1 ± 0.5 φ (fine
sand). Although all samples were poorly sorted (σφ > 1.0), the paleosol samples were more
poorly sorted (on average σφ > 2, i.e. very poorly sorted) than the current soil samples and
the unweathered ash samples (on average σφ < 2). Samples deviating from these general
trends were indicated with a symbol in Figure 3.3.
48
Figure 3.2 Organic carbon contents (%) at high vertical resolution versus sample depth (cm). Field boundaries
indicated with horizontal lines. = bulk sample (sample interval 5 cm), × = monolith sample (sample interval
14
0.75 cm), = monolith sample, used for C dating (sample interval 0.5 cm).

49
Figure 3.3 Grain size distributions (on a clayfree basis) of current soil samples, paleosol samples and
unweathered ash samples. Deviating samples indicated with symbols and sample codes.

When focussing on individual soil profiles it became apparent that the current soil, paleosol
and unweathered ash all showed an upward fining trend. Mean size also increased
significantly with depth (φ decreases, r = ‐0.78, p < 0.001, Figure 3.4) when all profiles were
grouped. Only soil profile G5b represented an exception. This profile showed the upward
fining trend only in the current soil, with the exception of sample G5bD75 which was
coarser than the underlying sample G5bE95. The GSDs of both these samples were relatively
coarse and overlapped with the GSDs from typical paleosol samples. On the other hand, the
GSD of sample G5bH160 (paleosol) of site G5b was too fine in view of its position in the
paleosol, although it did show a typical paleosol GSD.
50
Figure 3.4 Mean size (φ) versus sample depth (cm). = current soil, = paleosol, × = unweathered ash.

Mineral assemblage of silt and sand fractions

Results of the X‐ray diffraction analysis of the minerals in the sand and silt fraction of profile
G5a are reported in Table 3.3. All samples contained abundant albite (ideal composition
NaAlSi3O8) and kaersutite (ideal composition ((Na, K), Ca2)(Mg2,5, Fe(II),
Fe(III)0.5)TiSi6Al2O22(OH)2) and a trace of quartz in the sand fraction. Samples G5aI130 and
G5aL185 also contained traces of a 12Å (mixed layer illite and an Al‐hydroxy interlayered
mineral) and/or 14Å mineral (an Al‐hydroxy interlayered mineral) in the sand fraction. X‐ray
diffraction analysis of the high density part of the sand fraction revealed the presence of
some magnetite (ideal composition Fe(II)Fe2(III)O4) and hematite (ideal composition Fe2O3)
in addition to kaersutite, with somewhat higher magnetite contents in the paleosol. Analysis
of the magnetic fraction showed the presence of magnetite and hematite in equal amounts
and some kaersutite.

Table 3.3 Mineral assemblage of samples G5a, for the sand (63 μm ‐ 2 mm) and silt (2 – 63 μm) fraction.
Site Horizon Sample Sand fraction Silt fraction

[‐] [‐] [‐] 14Å 12Å Kaersutite Albite Quartz 14Å 12Å Kaersutite Albite Quartz
Ah G5a D45 x(x) xx tr x xx x
Ah G5a E60 x(x) xx tr x xx x
Bw G5a F85 x(x) xx tr x xx x
Bw G5a G95 x(x) xx tr x xx x
G5a
2Ah1b G5a H115 x(x) xx tr tr x xx x
2Ah2b G5a I130 tr x(x) xx tr tr x xx x
2Ah3b G5a J160 x(x) xx tr tr x xx x
2?BCb G5a L185 tr tr x(x) xx tr tr tr x xx tr

The silt fraction showed a similar mineral assemblage, although kaersutite decreased a little
in abundance, while quartz increased somewhat and the 14Å mineral now occurred
throughout the paleosol. Magnetite and hematite could not be detected in the silt fraction
due to overlapping reflections with kaersutite and albite. Optical analyses additionally
51
showed the presence of trace amounts of biotite throughout the current soil, paleosol and
unweathered ash. Although the presence of goethite was not confirmed by X‐ray diffraction,
it may have been present in low crystalline form or very small amounts.

Elements and element ratios as indicators of lithology

Chemical element contents are presented in Table 3.4. We only mention correlations larger
than ± 0.50 and significant at the 99 % confidence level. The principal component plot is
shown in Figure 3.5, the cumulative variance amounted to 65 %. We distinguished four
groups of elements that were closely correlated to each other and were additionally located
closely together in the component plot.

‐1
Table 3.4 Major elements (as oxide weight % on a LOI free basis) and trace elements (μgg on a LOI free
basis). Ectorganic and placic horizons excluded.

a b c d
Current soil Overprinted zone Paleosol Unweathered ash
min max mean st.dev. min max mean st.dev. min max mean st.dev. min max mean st.dev.
SiO2c 61 70 65 3 58 66 61 3 54 69 58 4 57 63 59 2
SiO2o 0.05 2.05 0.79 0.64 1.22 2.46 1.71 0.38 0.26 3.71 2.52 0.88 0.58 2.58 1.03 0.63
Al2O3c 11.40 14.75 12.69 0.80 9.88 15.36 12.71 1.39 10.77 16.05 12.99 1.59 15.46 19.11 17.72 1.30
Al2O3o 2.83 6.65 4.83 1.08 4.65 11.39 7.29 2.21 6.69 12.18 9.26 1.73 0.90 6.48 2.36 1.73
Fe2O3c 0.39 3.10 1.68 0.71 1.04 4.11 2.43 0.99 0.69 4.36 2.54 1.11 2.28 4.68 3.08 0.66
Fe2O3o 0.17 2.19 1.56 0.48 0.38 3.06 1.87 0.94 0.22 4.54 2.01 1.03 0.21 2.12 0.63 0.54
FeO 1.45 2.23 1.75 0.24 1.00 2.20 1.48 0.36 0.88 2.04 1.24 0.31 0.99 1.75 1.43 0.23
Na2O 3.09 3.95 3.46 0.21 2.51 3.98 3.35 0.41 2.80 4.57 3.54 0.48 4.34 5.39 5.01 0.37
CaO 2.43 4.16 3.47 0.42 2.74 4.50 3.72 0.45 2.81 4.84 3.93 0.51 4.68 5.94 5.36 0.45
MgO 1.18 2.23 1.90 0.26 1.72 2.18 1.91 0.15 1.50 2.27 1.92 0.20 1.42 2.32 1.89 0.27
K2O 1.09 2.40 1.52 0.33 0.91 1.29 1.14 0.11 0.77 1.23 1.02 0.13 0.96 1.65 1.17 0.18
SrO 0.035 0.050 0.041 0.004 0.037 0.069 0.049 0.011 0.043 0.083 0.063 0.012 0.090 0.112 0.104 0.006
MnO2 0.065 0.095 0.077 0.007 0.067 0.083 0.073 0.005 0.057 0.084 0.070 0.007 0.056 0.087 0.068 0.008
ZnO 0.006 0.013 0.009 0.002 0.008 0.011 0.009 0.001 0.007 0.010 0.009 0.001 0.009 0.019 0.013 0.003
CuO 0.003 0.017 0.008 0.004 0.003 0.011 0.006 0.003 0.003 0.010 0.005 0.002 0.001 0.009 0.004 0.003
Cr2O3 0.003 0.006 0.005 0.001 0.004 0.006 0.005 0.000 0.004 0.005 0.005 0.000 0.003 0.005 0.004 0.001
TiO2c 0.409 0.572 0.465 0.036 0.480 0.752 0.605 0.092 0.573 0.938 0.714 0.103 0.494 0.926 0.649 0.106
TiO2o 0.079 0.141 0.107 0.014 0.093 0.261 0.160 0.054 0.093 0.267 0.175 0.040 0.015 0.116 0.036 0.029
P2O5 0.163 0.313 0.227 0.050 0.147 0.327 0.203 0.053 0.114 0.328 0.213 0.052 0.099 0.259 0.165 0.043
Niµgg‐1 6 11 10 1 7 11 8 1 6 11 8 1 5 15 9 3
Coµgg‐1 11 17 15 2 14 19 16 2 15 22 17 2 13 21 16 2
Vµgg‐1 51 120 95 17 91 134 114 14 87 168 122 25 73 141 100 19
Liµgg‐1 15 21 16 2 15 18 16 1 14 18 16 1 14 22 16 2
Addition ‘c’ for crystalline oxide contents as calculated by subtracting the amorphous oxide contents as measured by
AAO (addition ‘o’), from total oxide contents (addition t).
a
= with n = 20; b = with n = 12; c = with n = 16; d = with n = 12, except for CuO where n = 8

Group I contained SiO2t, SiO2c, K2O and FeO. All group I elements, except FeO, showed a
positive correlation with the silt fraction (respectively +0.79, +0.72 and +0.57) , while SiO2t
and SiO2c were in addition correlated negatively with the sand fraction (respectively ‐0.62
and ‐0.52). Group II consisted of Na2O, CaO, Al2O3c, SrO and ZnO, with correlations
exceeding +0.90, except for correlations with ZnO that were somewhat weaker. All group II
52
elements were correlated positively to the sand fraction (> + 0.85, except for ZnO: +0.63)
and negatively to the silt fraction (< ‐ 0.75 for all group II elements, except for ZnO: ‐0.55) .
Na2O and Al2O3c were additionally correlated negatively to the clay fraction (‐0.67 and ‐
0.63). Interestingly, Na2O, CaO, SrO and ZnO were not correlated significantly to Al2O3t.
Group III was formed by Al2O3o, Fe2O3o and TiO2o. All group III elements showed a positive
correlation with the clay fraction (+0.88, +0.60 and +0.81). Finally, group IV contained TiO2t,
TiO2c, Al2O3t, Fe2O3t, Fe2O3c, SiO2o, Co and V, although SiO2o and V also showed positive
correlations with group III elements (e.g. +0.80 and +0.64 with Al2O3o). MgO plotted
between groups II, III and IV on the PCA figure and correlated only with Co and V (+0.60 and
+0.53).

We compared the means of primary elemental ratios and SLE ratios to distinguish significant
differences between the current soil, paleosol and unweathered ash. To facilitate
interpretation we only considered ratios between primary elements belonging to a specific
element group (intra‐group primary elemental ratios) as defined above, even though many
inter‐group primary elemental ratios showed significant differences between the current
soil, paleosol and/or unweathered ash as well. Most intra‐group primary elemental ratios
yielded significant differences, except for primary elemental ratios CaO/Na2O and FeO/K2O
and SLE ratios Fe2O3t/Al2O3t and Fe2O3t/TiO2t that were similar for all these deposits.

Figure 3.5 Principal component analysis of oxides, and clay, silt and sand contents. Different groups of oxides
marked with roman numbers.

53
The current soil differed significantly from the paleosol with regard to group I ratios
FeO/SiO2c and K2O/SiO2c; group II ratios CaO/Al2O3c, SrO/Al2O3c, SrO/Na2O, SrO/CaO; and
SLE ratio TiO2t/Al2O3t, while group IV ratio Fe2O3c/TiO2c and group II ratio Na2O/Al2O3c did
not differ significantly. Similarly, the current soil was distinct from the unweathered ash
with regard to all ratios except for group I ratios FeO/SiO2c and K2O/SiO2c and SLE ratio
TiO2t/Al2O3t. Finally, the paleosol and unweathered ash were different with regard to most
ratios except for CaO/Al2O3c, FeO/SiO2c, K2O/SiO2c and Fe2O3c/TiO2c. In summary,
SrO/Al2O3c, SrO/Na2O and SrO/CaO differed between all deposits, while the other ratios
showed differences with at least one deposit.

Cluster analyses
We performed a cluster analysis to identify objectively discontinuities in lithology on each of
the following sets of variables: GSD classes (on a clay free basis), mean grain size and sorting
(both clay free), intra‐group primary elemental ratios showing significant differences
between at least two deposits for every element group separately, and finally primary
elemental ratios SrO/Al2O3c, SrO/Na2O and SrO/CaO (further referred to as primary SrO
ratios). Figure 3.6a shows the dendrogram of the clustering on GSD classes and Figure 3.6b
shows the dendrogram of the clustering on primary elemental ratios SrO/Al2O3c, SrO/Na2O
and SrO/CaO.

Figure 3.6 a) Dendrogram of clustering based on grain size distributions.
3BC / G11 F170

3BC / Q1 K 240
3BC / Q1 J 160
3
3BC / G6 I170
3BC / G10 E 135
3BC / G11 E 110
3BC / G9 G90
3BC / G5b J 199
3BC / G5a L185
3BC / G7 I170
3BC / Q7 K 154
3BC / Q1 I 150
2A / G5b I180
2A / G7 H140
2A / G5b G145
2A / G5b F125
1A / G5b D75
2
2A / G7 G110
2A / G5b H160
2A / G5a J160
2A / G11 D57
2A / G5a I130
2A / G6 H160
2A / G6 F120
1A / G10 D110
2A / G9 E75
1A / G5b E95
2A / G5a H115
1A / G5a G95
1A / G5b C55
1A / G5a F85
2A / G7 F90
2A / G6 E 100
2A / G2 G170
1A / G11 C40 1
2A / G1 G125
1A / G2 E 105
1A / G1 E82
1A / G5b A15
1A / G11 B20
1A / G7 C30
1A / G7 B10
1A / G6 A15
1A / G9 B25
1A / G7 D47
1A / G5b B35
1A / G9 A10
1A / G6 B30
1A / G5a E60
1A / G5a D45
1A / G10 C80
1A / G9 D60
1A / G10 B50
1A / G9 C45
1A / G6 D70
1A / G1 F100
1A / G2 F135
1A / G7 E63
1A / G6 C55
0 1 2 3 4 5 6 7 8
Euc lidean dis tance
54
Figure 3.6 b) Dendrogram of clustering based on primary elemental ratios SrO/Na2O, SrO/CaO, SrO/Al2O3c.
3BC / Q1 K 240
3BC / G9 G90
3BC / G11 F170
3BC / G10 E 135
3BC / G11 E 110
3BC / G7 I170
3BC / G6 I170
3
3BC / G5b J 199
3BC / G5a L185
3BC / Q1 J 160
3BC / Q7 K 154
3BC / Q1 I 150
2A / G11 D57
2A / G5b G145
2A / G5b F125
2A / G7 H140
2A / G5b I180
2A / G5a I130
2A / G7 G110
2A / G6 F120
2A / G6 H160
2A / G5b H160
2A / G5a J160
1A / G10 D110 2
2A / G9 E75
1A / G5b E95
1A / G5b D75
2A / G5a H115
2A / G7 F90
2A / G6 E 100
1A / G10 C80
1A / G5a G95
2A / G2 G170
1A / G11 C40
2A / G1 G125
1A / G2 E 105
1A / G1 E82
1A / G10 B50
1A / G9 D60
1
1A / G6 D70
1A / G11 B20
1A / G5b C55
1A / G10 A20
1A / G9 B25
1A / G9 A 10
1A / G7 B 10
1A / G5b A15
1A / G5a F85
1A / G6 B30
1A / G5a D45
1A / G7 C30
1A / G6 A 15
1A / G7 E63
1A / G5b B35
1A / G2 F135
1A / G5a E60
1A / G1 F100
1A / G9 C45
1A / G7 D47
1A / G6 C55
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

E uc lidean dis tance

Clustering on GSD classes and mean grain size and sorting yielded virtually the same results,
separating the unweathered ash completely from the current soil and paleosol and
separating the latter two with a few exceptions (G1G125, G2G170, G5bD75, G5bE95,
G6E100, G7F90, and G10D110). Clustering on group I primary elemental ratios roughly
separated the current soil from the paleosol with overlap, but the unweathered ash samples
were encountered with both current soil samples and paleosol samples. Group II primary
elemental ratios enabled clustering of the current soil, paleosol and unweathered ash in
separate groups, although some overlap between the groups occurred. Group IV primary
elemental ratios did not separate the current soil, paleosol and unweathered ash in
different clusters. Clustering of primary SrO ratios showed very similar results to the
clustering on GSD, separating the current soil, paleosol and unweathered ash from each
other with only some overlap between the current soil and paleosol (notably G5aG95,
G5bD75, G5bE95, G10C80, G10D110, G11C40).

Tephra stratigraphy

On the right side of Table 3.1, we present the re‐horizonation of the samples for every soil
profile based on the observed ‘anomalous increase’ of organic carbon and the clustering on
GSD and primary SrO ratios. In the last column we present the final horizon designation
reflecting the tephra stratigraphy. In most cases, the field descriptions, organic carbon
contents and clustering results were in agreement and the field horizon designation was not
55
changed. For the organic carbon contents, we marked the zone with the ‘anomalous
increase’ in organic carbon content with an asterisk. The GSDs and primary SrO ratios
indicated that samples in the zone with the ‘anomalous increase’ in organic carbon content
either belonged to the current soil (5 times), to the paleosol (2 times) or showed
characteristics of both (G5aG95, G6E100 and G7F90). In only a few other cases the proxies
were not in agreement, notably G1G125, G2G170, G5bD75, G10C80 and G10D110.

Chronostratigraphy

In Figure 3.7 the soil catena including calibrated radiocarbon ages (cal BC as negative
numbers) is presented, correlating the tephra deposits of the different soils to each other. In
Figure 3.7, the paleosol field boundary was replaced by an overprinted zone, starting within
the current soil at the minimum organic carbon content and ending at the maximum organic
carbon content at the boundary with the paleosol. Radiocarbon dating close (within 5 cm) to
the maximum organic carbon content yielded a calibrated age of 2779 cal BC (G9), 2980 cal
BC (G5a), 3678 cal BC (G7) and 4053 cal BC (G5b).

Discussion

Field description

Based on the horizon sequence, we identified at least two tephra deposits in the field (on
average > 50 cm thick) in every soil profile except G10, recognizable as a current soil and a
paleosol. Beneath the paleosol an unweathered coarse ash was present which could either
belong to the paleosol or represent a separate preceding deposition. The thin placic horizon
was ascribed to a strong textural contrast hindering drainage of the otherwise well drained
volcanic soils.

Vertical distribution of SOM

The fluctuating organic carbon contents through the soil profiles (Figure 3.2) generally
confirm the field observations and the existence of at least one paleosol. However, the
gradual increase in organic carbon content around or above the boundary is unexpected.
We ruled out the possibility of the increase being an artefact caused by relatively large bulk
sampling intervals (5 cm), since the trend remains visible with the smaller sampling intervals
(0.5 – 0.75 cm) of the monolith samples.

We hypothesise that either bioturbation at the onset of current soil formation or burial of
forest type ectorganic horizons could have caused the ‘anomalous increase’ in organic
carbon contents. We consider recent bioturbation reaching the paleosol unlikely, since it
would have caused considerable rejuvenation of the radiocarbon ages. The age‐depth
curves did not show any anomalous younger ages around the paleosol boundary. However,
56
if the active zone of bioturbation shifted upwards during SOM accumulation in the current
soil, then the ‘anomalous increase’ could be the result of past bioturbation when this active
zone still reached the paleosol. The age difference between SOM in the current soil at the
onset of soil formation and that of the paleosol would have been considerably less and
consequently have less impact on the radiocarbon ages. While burial of a páramo
vegetation (aboveground biomass ~3500 gm‐2, Hofstede et al. 1995) would probably not
result in elevated organic carbon contents, one can imagine that burial of a forest floor with
a thickness of ≥ 75 cm (profile G1 and G2), corresponding to ~85000 gm‐2, could.

We discarded two remaining possibilities that theoretically could have caused the
‘anomalous increase’ in organic carbon contents. First, illuviation of modern humic
substances at the paleosol boundary would have caused rejuvenation of the radiocarbon
ages, which we did not observe. Furthermore, leaching of humic substances is highly
unlikely in the studied Andosols because of the high (Alp+ Fep)/SOM ratio. Second, an
increased abundance of roots at the paleosol boundary could not explain the ‘anomalous
increase’ in organic carbon content, since roots were concentrated in the top horizons of
the current soils and not around the paleosol boundary. This is in accordance with data of
Hofstede & Rossenaar (1995) for Colombian páramo ecosystems.

Grain Size Distributions

The current soil, paleosol and unweathered ash each showed characteristic GSDs and
average grain sizes, indicating that these are indeed different tephra deposits. Furthermore,
each of these tephra deposits shows an obvious upward fining trend, indicative of one thick
tephra deposit rather than a sequence of several thinner ones.

Mineral assemblage of silt and sand fractions

The mineral assemblage showed little variation throughout the current soil, paleosol and
unweathered ash, except for a somewhat higher content of magnetite and the occurrence
of traces of an Al‐hydroxy interlayered mineral in the paleosol and unweathered ash. The
origin or formation of 2:1 layer silicates in Andosols is subject to debate and several
hypotheses have been formulated (Shoji et al. 1993). The presence of an Al‐hydroxy
interlayered mineral in the sand and silt fraction suggests that in our case it was likely
inherited from primary minerals, as also supported by sharp X‐ray diffraction reflections.
Similarly to data of Shoji et al. (1993), albite showed a uniform distribution throughout the
sand and silt fraction, while kaersutite content decreased with smaller particle size.

Elements and element ratios as indicators of lithology

We conclude that group I elements (SiO2t, SiO2c, K2O and FeO) had a dominant volcanic
glass origin. Optical mineralogy confirmed the presence of volcanic glass. Of course, Si is
57
present in almost all primary and secondary minerals occurring within this tephra sequence,
but since most of these minerals show contrasting behaviour, Si was only correlated to
elements of group I. Other possible origins include biotite and quartz. Both biotite, due to its
platy form, and quartz (dominantly present in the silt fraction, Table 3.3) could represent
eolian additions of tephra, as supported by the correlations of group I elements with the silt
fraction. The positive correlations of K2O and FeO to both SiO2t and SiO2c indicated not only
that their origin was similar, but also that these elements showed similar behaviour upon
weathering, i.e. leaching.

The correlations between group II elements (Na2O, CaO, Al2O3c, SrO and ZnO) point to their
presence in albite and/or kaersutite minerals, as confirmed by the correlations with the
sand fraction indicative of their primary origin. During fractional crystallisation, Sr tends to
concentrate in plagioclase and hornblende (Deer et al. 1993). The correlations with ZnO are
less pronounced, probably because the vertical distribution of ZnO is differentially
influenced by biocycling; being immobilized in SOM complexes in surface horizons while
being leached from (buried) subsurface horizons (Rahman et al. 1996). The absence of a
correlation with Al2O3t for all group II elements indicates that these elements behaved
opposingly during weathering, with Al remaining immobile while Na, Ca, Sr and Zn were
leached.

Group III elements (Al2O3o, Fe2O3o and TiO2o) represented the amorphous minerals
allophane, ferrihydrite and probably some leucoxene, since even TiO2t contained a
considerable amorphous component (on average ~20 % in the current soil and paleosol).
The secondary origin of these elements was supported by the positive correlation with the
clay fraction.

Finally, group IV (TiO2t, TiO2c, Al2O3t, Fe2O3t, Fe2O3c, SiO2o, Co and V) probably represented
mainly the minerals magnetite, hematite and possibly goethite (except for Al2O3t).
Magnetite may contain up to 27 % of TiO2 (Deer et al. 1993). Vanadium occurs in
titaniferrous magnetite and hematite and is often sorbed onto iron oxides (Adriano 2001).
Similarly, cobalt is often associated with iron oxides (Adriano 2001), although it is also
influenced by biocycling (Rahman et al. 1996). The addition of SiO2o to this group is
unexpected, but SiO2o did still show positive correlations to group III, related to the
formation of allophane. Apparently, sorption of Si onto iron oxides could play a role as well,
which may be related to desilication in the topsoils as was demonstrated in Podzols
(Sommer et al. 2006). The absence of a correlation between e.g. TiO2c and Al2O3c and Al2O3c
and Fe2O3c, while the total contents of these elements did show positive correlations
demonstrates that although of different origins, the accumulating behaviour of these
elements upon weathering was similar.

Magnesium could occur in kaersutite (group II), magnetite (group IV) and in secondary
minerals (group III), explaining its intermediate position on the PCA plot (Figure 3.5).
58
To identify discontinuities in tephra deposition, we reviewed intra‐group primary elemental
ratios and ratios between SLE, because these do not change upon weathering. Since most
intra‐group primary elemental ratios differed significantly between at least two of the
deposits, we conclude that the current soil, paleosol and unweathered ash represent
individual tephra deposits. The composition of albite and/or kaersutite, as represented by
the ratios of SrO to Na2O, CaO and Al2O3c, appeared to be characteristic for each tephra
deposit. Due to the dominance of amorphous secondary constituents and very low CEC with
high adsorbed acidity, these MLE behaved similarly upon weathering. Therefore, ratios
between these MLE indeed represent primary elemental ratios, as supported by their high
positive correlations with the sand fraction. The SLE ratios on the other hand were less
suitable for distinguishing tephra discontinuities, Fe2O3t/Al2O3t and Fe2O3t/TiO2t even being
similar for all three deposits. These results underline that primary elemental ratios enhance
identification of tephra stratification, even when SLE are similar between deposits. There
was no evidence for a finer lithological stratification within the current soil, paleosol or
unweathered ash, in accordance with GSD results.

Cluster analyses

Based on the organic carbon contents, GSD and geochemical data, we concluded that the
current soil and paleosol constitute different tephra deposits. The current soil and
unweathered ash evidently must also be different deposits. The latter two cannot have
been mixed by bioturbation and therefore should be well separated from each other by
cluster analysis. For that reason, we disregarded clustering results where the current soil
and unweathered ash were not separated in different clusters, the variables used for
clustering not being discriminating. Clustering on GSD and on primary SrO ratios rendered
the best results, separating the unweathered ash (cluster 3) completely from the other
deposits and separating the current soil (cluster 1) and paleosol (cluster 2) with only a few
explicable exceptions that will be discussed in the next section. The similarity between these
two clustering results, although based on different variables, confirms the validity of the
results since obtaining such similar clustering by chance is extremely unlikely.

Tephra stratigraphy

Combining mineralogical, physical and geochemical data with field descriptions and organic
carbon contents, we were able to unravel the tephra stratigraphy of the studied soil
profiles, thus increasing understanding of the vertical distribution of SOM.

Disagreement in the data occurred only at the boundary between current soil and paleosol
and consequently also within the zone of the ‘anomalous increase’ in organic carbon
content. If the zone with the ‘anomalous increase’ in organic carbon content was caused by
the burial of ectorganic horizons, then the lithology should be that of the tephra deposit
burying it, i.e. that of the current soil. Contrary, we found that this zone showed either
59
characteristics of the current soil, of the paleosol or of both, supporting the hypothesis that
past bioturbation (belonging to the current pedogenesis) mixed the bottom of the current
soil with the paleosol. At the top of this zone the lithological characteristics of the current
soil are often dominant, while at the bottom paleosol features more frequently dominate, in
accordance with an upward shifting bioturbation zone. As a result, the lithology of this zone
cannot be denominated straightforwardly, whereas pedogenesis clearly belongs to the
current phase of soil formation, overprinting the paleosol. In the soil horizon designations
we decided to use ‘1/2’ as a prefix, to draw attention to the overprinted nature of this zone.

In a few cases, other processes than overprinting may explain disagreement of the data. The
GSDs of G1G125 and G2G170 were somewhat finer than other paleosol samples, which is
not unlikely since weathering proceeds from the earth surface. Since these samples
clustered as paleosol according to all variables except GSD, we conclude that they indeed
belong to the paleosol and did not change the field designation. Both G5bD75 and G5bE95
exhibited lithological characteristics of the paleosol, while the organic carbon contents were
related to current pedogenesis. The most likely explanation would be a locally even less
sorted paleosol deposit that was probably influenced by colluvial processes prior to SOM
incorporation, of which the top was overprinted by current soil formation resulting in its
field classification as current soil. Another G5b paleosol sample (G5bH160) also showed an
anomalous GSD, this time being finer than expected on the basis of its position in the
profile, supporting a colluvial origin. The field horizon designation was changed from Bw2 to
1/2Bw2 for G5bD75 and G5bE95. We were able to identify the paleosol in profile G10,
starting at least at 80 cm (G10C80), even though this was not possible in the field. Because
we did not have a monolith from profile G10 we were not able to identify the overprinted
zone. Since the GSD clustering and primary SrO ratios were not in agreement, sample
G11C40 probably belongs to the overprinted zone and was designated 1/2Ah1 accordingly.
Finally, we changed the prefixes of the horizon designations of the unweathered ash from
‘2’ to ‘3’ based on our results.

Summarising, the soil profiles were formed within three distinct tephra deposits of similar
mineralogical composition, containing abundant albite and kaersutite and small quantities
of quartz, magnetite, hematite and biotite. The current soil was formed in a coarse silt sized
deposit showing a clear upward fining trend, with a relatively low SrO content with regard
to Na2O, CaO and Al2O3c in the primary minerals albite and kaersutite. The paleosol was
formed in a very poorly sorted deposit of very coarse silt size, with less obvious upward
fining, and a somewhat higher SrO content in the albite and kaersutite minerals. The top of
the paleosol was overprinted by current soil formation, which caused a gradual increase in
organic carbon content instead of an abrupt increase, and resulted in transitional
lithological features (i.e. GSD and primary SrO ratios). Finally, an immature or strongly
truncated paleosol occurs in the third tephra deposit, which is fine sand sized and contains
the highest SrO content within the albite and kaersutite minerals. We did not find evidence
for a finer stratification within each of these deposits.
60
In general the field descriptions were reasonably reliable, although in order to identify the
overprinted part of the paleosol, GSDs, geochemical analysis and also organic carbon
contents at high vertical resolution proved to be indispensable. Similarly, only based on
GSDs and geochemical analysis we were able to decide whether the unweathered ash
belonged to the paleosol or formed a second, truncated or immature paleosol, which could
not have been done in the field.

Radiocarbon dating

For a detailed discussion on the radiocarbon ages we refer to Tonneijck et al. (2006). Our
results are in line with dating of charcoal fragments in soil profiles from the same research
area by Di Pasquale et al. (2007), confirming the reliability of dating the HA fraction. The
anomaly at G5a is probably due to local disturbance.

In our view, two scenarios may explain the observed linear increase of age with depth
(Tonneijck et al. 2006). The ‘traditional’ scenario assumes that the current soil consists of
multiple tephra depositions that could not be observed in the field due to subsequent soil
formation. SOM would then have been intermittently buried by tephra, tephra deposition
rate mainly determining the age‐depth relationship. However, we demonstrated that
neither of the three deposits contained a finer tephra stratification. On the other hand we
did show that bioturbation must play an important role in current pedogenesis, even
resulting in the overprinting of the paleosol. Still, the linear increase of age with depth
shows that bioturbation did not homogenise SOM at the scale of the applied vertical
sampling distances (on average 15 cm in profiles G5a and G5b, in other soil profiles on
average 40 cm) and thus recent bioturbation could not have reached the paleosol.

Therefore, the most probable scenario explaining the linear age depth relationship is ‘peat‐
like’ stratified SOM accumulation in the presence of an upward shifting bioturbation zone. A
linear age depth relationship in the presence of bioturbation is only possible if the zone of
active bioturbation shifts upward during SOM accumulation (Tonneijck et al. 2006),
assuming that bioturbation results in homogenisation of radiocarbon ages (Davidson et al.
1999). While such a scenario might seem uncommon for soils in general, Andosols are
known to strongly accumulate SOM (Shoji et al. 1993; Dahlgren et al. 2004), causing the soil
to thicken (Barois et al. 1998).

61
Chronostratigraphy

Based on the tephra stratigraphy outlined above in combination with radiocarbon dating,
we were able to correlate the soil profiles (Figure 3.7). The thickness of the current soil
clearly depends on OM input into the mineral soil, which is less in the forest where OM
accumulates on the soil surface and which decreases with altitude (temperature) as a result
of lower biomass and litter production.

Figure 3.7 Catena showing the tephra stratigraphy of the studied soil profiles, calibrated radiocarbon ages
(cal BC as negative numbers) shown where available. Dotted lines indicate that exact boundary depth is not
known.
62
Due to overprinting of the paleosol by the current soil, it has become impossible to obtain
an accurate timing of the event of tephra deposition with radiocarbon dating. Normally, the
exact depth of the top of a paleosol can be identified by an abrupt increase in organic
carbon content and the position for dating exact events of tephra deposition would then be
just above and below the maximum organic carbon content. However, the position of the
maximum organic carbon content is now dependent on e.g. the depth of the bioturbation
zone extending into the paleosol, which may have been variable in different soil profiles.
Therefore, the maximum organic carbon content would represent an overestimation of the
timing of last tephra deposition. Still, we were able to narrow the large gap (2000 14C yr)
between the radiocarbon datings of the different paleosol field boundaries to an age
difference of ~1000 14C yr between the different positions dated close to the end of the
overprinted zone (maximum organic carbon content). The last tephra deposition would
have occurred at maximum between 2779 cal BC (G9) and 4053 cal BC (G5b). Because either
timings would constitute an overestimation, we consider 2779 cal BC to be the most likely
estimate.

Despite the difficulties in dating the event of tephra deposition, soil formation did not result
in an obliteration of the stratification of SOM within the tephra deposit of the current soil,
as evidenced by the linear age‐depth relationship. As mentioned previously, stratification in
the presence of bioturbation is only possible if the zone of active bioturbation shifts
upwards during SOM accumulation. However, knowledge on the depth of the active
bioturbation zone is needed to find out how crude or fine the stratification of e.g.
paleoecological proxies is. Therefore, prior to using these soils for the reconstruction of past
vegetation with paleoecological analysis, we need to increase our understanding of
bioturbation in these soils.

Conclusion

Our results show that the studied soil profiles along an altitudinal transect intersecting the
UFL in northern Ecuador were formed in three distinct tephra deposits. For the
identification of these deposits, mineralogical, physical and geochemical data in
combination with organic carbon contents at high vertical resolution proved to be
indispensable. Unravelling the tephra stratigraphy improved understanding of the vertical
distribution of SOM in the studied volcanic ash soils. Bioturbation most likely plays an
important role in current pedogenesis, even resulting in the overprinting of the paleosol.
Surprisingly, despite bioturbation, a linear age depth relationship exists, suggesting that the
active zone of bioturbation must have shifted upwards during soil formation. Therefore,
paleoecological proxies are stratified in our soils, albeit probably somewhat more crudely
than in undisturbed peat bogs or lake sediments. To further increase understanding of the
vertical distribution of paleoecological proxies in soils, detailed research on bioturbation is
recommended, e.g. in the form of micromorphological analysis and/or modelling.

63
Acknowledgements

We acknowledge the Ecuadorian Ministerio del Ambiente for issuing the necessary permits,
Jatun Sacha for their support in Guandera Biological Station and Ecopar for their office
assistance. We thank Minard Hall (Escuela Politecnica Nacional, Ecuador) and Manfred Van
Bergen (University of Utrecht, the Netherlands). Furthermore, we thank Leo Hoitinga, Leen
de Lange, Ton Van Wijk, Niko de Wilde de Ligny and Bert de Leeuw (University of
Amsterdam, the Netherlands) for their laboratory assistance. Finally, we acknowledge all
participants in the RUFLE project and Mirjam Vriend (University of Amsterdam) and Hans
Van der Plicht (University of Groningen, the Netherlands) for their valuable input leading to
the improvement of this manuscript. This research was funded by WOTRO (WAN 75‐405)
and the University of Amsterdam and generously sponsored by Fjällraven in the form of
clothing and gear.

64
65
4 The influence of bioturbation on the vertical
distribution of soil organic matter in volcanic ash
soils: a case study in northern Ecuador
Published in European Journal of Soil Science 59: 1063–1075, 2008,
Tonneijck FH & Jongmans AG. With additional photographs.

Abstract

Soil faunal bioturbation (‘bioturbation’) is often cited as a major process influencing the
vertical distribution of soil organic matter (SOM). The influence of bioturbation on vertical
SOM transport is complex because it is the result of interaction between different groups of
soil faunal species that redistribute SOM through the soil profile in distinct ways. We
performed a semi‐quantitative micromorphological analysis of soil faunal pedofeatures and
related their occurrence to the vertical distribution of SOM and high resolution radiocarbon
dating in volcanic ash soils under montane forest and grassland (páramo) vegetation in the
northern Ecuadorian Andes. The páramo soil data suggest that bioturbation was largely
responsible for the vertical distribution of SOM, while illuviation and root input were of
minor importance. Bioturbation was caused by endogeic species, which typically mix the soil
only over short vertical distances. Short vertical distance mixing was apparently enhanced
by upward shifting of bioturbation as a result of soil thickening due to SOM accumulation. A
change from páramo to forest vegetation was accompanied by a change from endogeic to
epigeic species. These latter species do not redistribute material vertically, which eventually
resulted in the formation of thick ectorganic horizons in the forest.

Keywords
bioturbation, organic carbon, vertical distribution, radiocarbon dating, volcanic ash soils
67
Introduction
The vertical distribution of SOM is principally determined by 1) the input ‐ above and
belowground ‐ of litter, 2) the output of organic matter through decomposition and 3) its
vertical transport. With respect to the latter, soil faunal bioturbation (further referred to as
‘bioturbation’) is often cited as a major process influencing the vertical distribution of SOM
(e.g. Anderson 1988; Lavelle 1988; Rasse et al. 2006). Elzein & Balesdent (1995)
demonstrated in a modelling study that bioturbation is the dominant mode of vertical
transport, relative to leaching, in a range of soil types.
Soils formed in volcanic ash contain the largest amounts of SOM per unit area of all mineral
soil orders (Dahlgren et al. 2004), making them potential carbon sinks for the greenhouse
gas CO2, which is of global importance. We previously investigated the age and vertical
distribution of SOM in such soils under montane forest and grassland (páramo) vegetation
in the northern Ecuadorian Andes (Tonneijck et al. 2006; Tonneijck et al. 2008a). These soil
profiles consist of a current soil in a thick tephra deposit superimposed on a SOM‐rich
paleosol in a preceding tephra deposit. We observed a progressive downward increase in
SOM from the relatively SOM‐poor subsoil of the current soil into the SOM‐rich top of the
paleosol, further referred to as the ‘overprinted zone’. Vertical transport of dissolved OM by
leaching can be assumed insignificant in volcanic ash soils, because of the large metal‐to‐
SOM ratios that strongly limit its mobility (Aran et al. 2001; Dahlgren et al. 2004).
Furthermore, rooting is typically superficial in Andean montane forests (Soethe et al. 2006)
and páramo (Hofstede & Rossenaar 1995). Therefore, the occurrence of the overprinted
zone strongly suggested that bioturbation plays an important role during soil formation.
The influence of bioturbation on vertical SOM transport is complex because it is the result of
interaction between different groups of soil faunal species. These may typically be divided in
three main groups according to their behaviour in the soil (Anderson 1988), similar to the
classification for ecological groups of earthworms (Lee 1959): epigeic, anecic and endogeic
species. These groups of species redistribute SOM through the soil profile in distinct ways
(Anderson 1988; Lavelle 1988; Lee & Foster 1991). Epigeic species inhabit the ectorganic
layers and do not actively redistribute material. Anecic species feed mainly on OM from the
earth surface but inhabit deeper soil layers in which they create (semi‐)permanent vertical
burrows and therefore are responsible for transport of OM over large vertical distances that
may exceed 1 m. By contrast, endogeic species ingest both mineral soil and organic matter
and create extensive networks of subhorizontal burrows thus transporting SOM only over
short vertical distances. Some endogeic species are typically concentrated around 10 – 20
cm depth, other endogeic species occur deeper in the soil profile up to 50 cm depth.
To increase understanding of the influence of bioturbation on the vertical distribution of
SOM in volcanic ash soils in northern Ecuador, we performed a semi‐quantitative
micromorphological analysis of soil faunal pedofeatures and related their occurrence to the
vertical distribution of SOM and high resolution radiocarbon dating.

68
Description of study area and sites

The study sites are located in the nature protection area of Guandera Biological Station in
northern Ecuador, near the border with Colombia, on the west facing slopes of the Eastern
Cordillera (Figure 1.1 and Table 1.1, 0° 35’ north and 77° 41’ west). Guandera has a (semi‐)
natural upper forest line at approximately 3650 m above mean sea level (a.s.l.). There are
some forest patches above the current upper forest line. Dominant species in the forest are
Clusia flaviflora ENGL., Weinmannia cochensis HIERON. and Ilex colombiana CUATREC. and
the páramo is characterized by bunch‐grass Calamagrostis effusa KUNTH (STEUD.) and stem
rosette Espeletia pycnophylla CUATREC. The forest and páramo belong to the globally
leading Tropical Andes biodiversity ‘hotspot’ (Myers et al. 2000). Because deforestation in
the Inter Andean valleys has been particularly severe (Keese et al. 2007), the Guandera
forest constitutes a unique remnant. Mean annual precipitation is around 1900 mm and
mean annual temperature ranges from 12 °C at 3000 m a.s.l. to 4 °C at 4000 m a.s.l. Both
precipitation and temperature show little seasonal variation. The soil climate is isomesic and
perudic.

We selected a site covered by forest (G1) at 3501 m a.s.l., a site in a forest patch above the
current upper forest line (G5a) at 3697 m a.s.l., a site just next to it with páramo vegetation
(G5b) and finally a site at 3860 m a.s.l. in the páramo (G7). We were able to identify soils
with erosional/depositional features by anomalous grain size and geochemical data
(Tonneijck et al. 2008a) and only investigated undisturbed soils. The soil profiles at our sites
were each formed in three distinct tephra deposits of Holocene age (Tonneijck et al. 2008a).
Related to this tephra stratigraphy all soils contained a multisequum consisting of a current
soil, a paleosol and a second heavily truncated or immature paleosol, each at least 40 cm
thick. Between the current soil and the first paleosol an overprinted zone with transitional
organic carbon contents, grain size distribution and element ratios was present. In the soil
horizon designations we used ‘1/2’ as a prefix to indicate this overprinted zone.

The mineral horizon sequence can be summarised as Ah1 – 1/2Ah2 or 1/2Bw ‐ 2Ahb ‐ 2Bsb –
3BCb. In addition, the forest profiles had ectorganic horizons (LFH) with a combined
thickness of up to 75 cm. By contrast, ectorganic horizons (L) barely exceeded 1 cm in the
páramo profiles. A thin placic horizon was present at the boundary between the first and
second paleosol. This placic horizon is ascribed to a strong textural contrast hindering
drainage of the otherwise well drained soils. The soils classified as a Histosol overlying a
mineral soil with andic properties (G1), as an Andic Cambisol (G5a) and as an Andosol (G5b
and G7) according to the World Reference Base (FAO 2006). The ectorganic horizons
classified as resimor (G1) and mor (G5a) in the forest and as mull in the páramo (G5b and
G7), according to Green et al. (1993). We observed that the ectorganic horizons changed
abruptly from mor to mull at the boundary between forest (patch) and páramo.

69

Sampling procedures

Soil pits of approximately 1.5 m2 surface area and a depth of at maximum 2 m were dug. We
took bulk soil samples for physical and chemical analyses (n = 35, for humic substances n =
28), ring samples for bulk density determination (n = 35) and undisturbed box samples (with
dimensions of 7 cm height x 5 cm width x 4 cm depth) for the preparation of thin sections (n
= 30, up to the top of the first paleosol), in the same soil pit at regular depths but respecting
horizon boundaries. Bulk and ring samples were taken over a vertical interval of 5 cm, the
middle of which was noted as the depth of the sample. Additionally, we took undisturbed
vertical soil samples (monoliths) using one or two metal gutters of 75 cm height x 5 cm
width x 4 cm depth. For radiocarbon dating (n = 35) and additional estimation of organic
carbon contents (n = 182) and humic substances (n = 19, being a selection of the samples
used for radiocarbon dating), samples were cored from the monoliths by means of a corer
of 0.5 cm and 0.75 cm diameter respectively. Since the continuous input of fresh OM into
soils causes the measured 14C ages of bulk SOM to be always somewhat younger than the
depositional age (Wang et al. 1996) it is necessary to use such small sampling intervals for
radiocarbon dating. All samples were stored at 2 °C under field moist conditions prior to
analysis.

Laboratory procedures and statistical analysis

Laboratory procedures for the measurement of total carbon content, dry bulk density,
pHCaCl2 and allophane content were described in Tonneijck et al. (2006) and Tonneijck et al.
(2008a). Total carbon equals organic carbon, since carbonates were absent. Humic
substance fractions of SOM were obtained by alkali and subsequent acid extraction
according to Schnitzer (1982). After each extraction, carbon content in the extract was
measured with a Shimadzu TOC Analyser 5000 (Kyoto, Japan), to obtain the organic carbon
content present in humic acids (HA). Pearson and Spearman correlations (bivariate)
between total organic carbon and carbon in HA were calculated (n = 47) using the SPSS
Correlate procedure and considered significant when p < 0.01. Field‐moist undisturbed box
samples were impregnated after replacing water by acetone, according to Miedema et al.
(1974). Thin sections were prepared according to Murphy (1986).

Radiocarbon dating

In addition to the radiocarbon data already published (Tonneijck et al. 2006; Tonneijck et al.
2008a), we performed dating for profiles G5a, G5b and G7 to further increase the vertical
resolution and acquired a first set of 14C dates for profile G1. Radiocarbon dating followed
procedures outlined in Tonneijck et al. (2006). In short, humic substance fractions of SOM
were radiocarbon dated with an Accelerator Mass Spectrometer. The fulvic acid fraction
70
was discarded because of its inherent mobility in most soils. We previously discussed
(Tonneijck et al. 2006) that in Andosols lacking a thick ectorganic horizon, dating of HA gave
the most accurate results since roots tend to accumulate in the residual (humin) fraction
rather than in the HA fraction and HA was immobile due to the large metal/SOM ratios and
acidic pH. Contrarily, in mineral soil samples just beneath thick ectorganic horizons, humin
ages were more accurate, because HA was then contaminated by younger HA illuviated
from the ectorganic horizons while roots were concentrated in the ectorganic horizons
rather than in the mineral topsoil. Consequently, we dated HA in all cases, except for
samples just beneath a thick ectorganic horizon where we dated both the HA and humin
fractions. Radiocarbon ages were calibrated, according to procedures outlined in Tonneijck
et al. (2006), to a calendar year probability distribution (cal AD/BC). Calibrated radiocarbon
ages are expressed as the range of calendar years of the 1 sigma peak with the largest
relative area under the probability distribution. Since the ranges are small, calculations are
performed with the mean and the dots displayed in the graphs represent the whole range of
calendar years at the scale of the graph.
Micromorphological analysis
Thin sections were analysed using a Leitz M420 makrozoom microscope (for a magnification
of 8x) and a Leitz Wetzlar petrographic microscope (for magnifications of 25x and 63x). Thin
sections were described following the micromorphological terminology of Stoops (2003).
Abundance classes were as follows: very few (< 5 %), few (5 – 15 %), common (15 – 30 %),
frequent (30 – 50 %), dominant (50 – 70 %) and very dominant (> 70 %). To display these
classes in bar graphs we used class midpoints. Rather than using Stoops’ grades of
coalescence of micro‐aggregates, we preferred the more illustrative terminology of welding.
Our single excrements correspond to Stoops’ very porous micro‐aggregates and our very
strongly welded excrements to very dense micro‐aggregates. We grouped the terms ‘organ
residues’, ‘tissue residues’ and ‘cell residues’ of Stoops in one class of ‘fragmented OM’, to
increase readability. Fragmented OM includes both fresh and more decomposed (less
birefringent) OM.
We defined soil faunal pedofeatures, i.e. excrements, loose infillings and dense infillings, as
‘bioturbation features’. In addition, we included zones with a vughy microstructure (i.e.
zones containing frequent vughs) in our definition, since these zones were interpreted to
result from very strong welding of loose infillings, which is in accordance with observations
by Pawluk (1987).

Results
Soil properties
General soil properties were reported in Tonneijck et al. (2008a). Briefly, the upper mineral
horizons were characterised by a very large organic carbon content (8 % to 22 %), acidic pH
71
(pHCaCl2 3.2 to 4.4) and small dry bulk density (0.3 to 0.7 gcm‐3). The mineral topsoils did not
contain allophane, while the subsurface horizons contained up to 5 % allophane.
Vertical distribution of SOM and HA

The vertical distribution of total organic carbon and carbon in the HA fraction is presented
for each soil profile in Figure 4.1a. As explained in the introduction, organic carbon content
decreased with depth in the current soils, but instead of an abrupt increase in organic
carbon content at the boundary with the paleosol, organic carbon content increased
gradually. Carbon in the HA fraction followed a similar distribution with depth as organic
carbon and both Pearson and Spearman correlations between carbon in the HA fraction and
total organic carbon were large (+0.82 and +0.86 respectively, p < 0.001 in both cases).

Radiocarbon dating

Conventional and calibrated radiocarbon ages are presented in Table 4.1 and Figure 4.1c
additionally shows calibrated age versus depth for each soil profile. We used the calibrated
radiocarbon ages of HA for the age‐depth relationship in all cases, except for the mineral
soils just beneath the ectorganic horizon of the forest profiles where we used calibrated
radiocarbon ages of humin instead, according to Tonneijck et al. (2006). The r2 value of the
linear regression was large in all cases (> 0.84). The slope of the age‐depth relationship in
the mineral soil of forest site G1 was significantly different from the slopes of the other soil
profiles, while forest patch profile G5a resembled the páramo profiles.

Figure 4.1 Vertical distribution of:
a) Total organic carbon content (%): = bulk sample (5 cm interval), = monolith sample (0.75 cm interval),
14
= C sample (0.5 cm interval) and organic carbon content in humic acids (%) : × = bulk sample (5 cm
interval), × = monolith sample (0.75 cm interval).
b) Bioturbation features: + = very few, ++ = few, +++ = common, ++++ = frequent, +++++ = dominant, ++++++ =
very dominant.
c) Calibrated radiocarbon ages (cal AD/BC) for the mineral soil profiles.

72
Figure 4.1 continued
73
Table 4.1 Conventional (BP) and calibrated (cal AD/BC) radiocarbon ages of the humic acid fraction (except
for samples located just beneath a thick ectorganic horizon where the humin fraction was dated, as indicated
in bold font).

Calibrated age
Site Codea Horizonb Depthc 14
C age δ13C Range Mean Referenced
[cm] [BP] [‰] [cal AD/BC] [cal AD/BC]
GrA‐30094e H ‐30.5 40 ± 35 ‐25.73 cal AD 1877 to 1912 cal AD 1895
GrA‐30092 H ‐1 380 ± 35 ‐25.68 cal AD 1449 to 1517 cal AD1483
G1 GrA‐34933 Ah 0 1225 ± 40f ‐28.15 cal AD 769 to 830 cal AD 800
GrA‐34934 Ah 10 2475 ± 40 ‐27.18 cal BC 727 to 606 cal BC 667
GrA‐34935 2Ahb 39 6155 ± 45 ‐25.31 cal BC 5135 to 5049 cal BC 5092
GrA‐30109 F ‐20.5 ‐623 ± 35 ‐27.46 cal AD 1996 to 2002 cal AD 1999 *
GrA‐30112 F ‐1 ‐28 ± 35 ‐26.39 cal AD 1956 to 1957 cal AD 1957 *
GrA‐34938 Ah 5 325 ± 35g ‐26.91 cal AD 1538 to 1602 cal AD 1570
GrA‐28101 Ah 10 775 ± 35h ‐25.93 cal AD 1247 to 1271 cal AD 1259 *
GrA‐34939 Ah 20 755 ± 35 ‐26.51 cal AD 1251 to 1281 cal AD 1266
GrA‐30113 Ah 24.5 355 ± 35 ‐26.03 cal AD 1467 to 1523 cal AD 1495 **
G5a
GrA‐30114 Ah 39.5 1855 ± 40 ‐26.65 cal AD 124 to 217 cal AD 171 **
GrA‐28104 Bw 49.5 2170 ± 35 ‐25.92 cal BC 352 to 293 cal BC 323 *
GrA‐30133 1/2 Bw 69 3650 ± 40 ‐25.71 cal BC 2041 to 1956 cal BC 1999 **
GrA‐30134 1/2 Ah1b 70.5 3880 ± 40 ‐25.62 cal BC 2406 to 2337 cal BC 2372 **
GrA‐28108 1/2 Ah1b 79.5 4355 ± 40 ‐25.10 cal BC 3015 to 2944 cal BC 2980 *
GrA‐34943 2Ah1b 85 4870 ± 40 ‐25.57 cal BC 3667 to 3638 cal BC 3653
GrA‐30138 Ah 14.5 590 ± 35 ‐25.33 cal AD 1311 to 1359 cal AD 1335 *
GrA‐28111 Ah 34.5 1350 ± 35 ‐25.45 cal AD 648 to 682 cal AD 665 *
G5b GrA‐30139 Bw1 49.5 2070 ± 40 ‐26.45 cal BC 116 to 43 cal BC 80 **
GrA‐30107 Bw1 64.5 3045 ± 40 ‐25.79 cal BC 1323 to 1266 cal BC 1295 **
GrA‐28113 1/2 Bw2 75 3465 ± 35 ‐25.51 cal BC 1780 to 1740 cal BC 1760 *
GrA‐30238 1/2 Bw2 94.5 4270 ± 40 ‐25.81 cal BC 2913 to 2879 cal BC 2896 **
GrA‐30232 1/2 Bw2 114 5260 ± 40 ‐24.81 cal BC 4071 to 4034 cal BC 4053 **
GrA‐30234 2Ahb 115.5 5405 ± 40 ‐25.20 cal BC 4327 to 4280 cal BC 4304 **
GrA‐28116 2Ahb 124.5 5730 ± 40 ‐24.91 cal BC 4614 to 4515 cal BC 4565 *
GrA‐28130 Ah1 10 180 ± 35 ‐25.24 cal AD 1761 to 1789 cal AD 1775 *
GrA‐28134 Ah2 46.5 1690 ± 40 ‐25.22 cal AD 329 to 407 cal AD 368 *
G7
GrA‐28126 1/2 Ahb 89.5 4910 ± 45 ‐24.98 cal BC 3712 to 3643 cal BC 3678 *
GrA‐34944 2Ahb 98 5985 ± 45 ‐25.47 cal BC 4935 to 4860 cal BC 4898
a = Laboratory code of the Centre for Isotope Research, Groningen University, the Netherlands
b = horizon nomenclature of ectorganic horizons according to Green et al. (1993)
c = depth of sample, vertical sample interval 0.5 cm, the mineral soil surface was set at 0 cm, samples taken
from soil monolith
d = * Tonneijck et al. (2006), ** Tonneijck et al. (2007)
e = in this case the humin fraction was dated because HA was lost
14
f = C age of HA was 840 ± 35 BP
14
g = C age of HA was 255 ± 35
14
h = C age of HA was 490 ± 35 BP
74
Mineral and organic components

Throughout the mineral horizons of all soil profiles mineral components (data not shown)
were represented by plagioclase, amphibole, biotite, vesicular rhyolitic volcanic glass and
rhyolitoid rock fragments. Opaque mineral grains (e.g. magnetite and hematite) were
difficult to recognize due to the presence of very fine (< 20 μm) opaque OM fragments. In
general, minerals and rock fragments showed only initial pellicular and/or dotted
weathering. Fine mineral components in the mineral horizons of all soil profiles consisted of
dull light to dark brown, isotropic material with undifferentiated b‐fabric. The presence of
very fine (< 20 μm) opaque OM resulted in a dotted appearance.

Distribution of fresh roots and fragmented OM (as area %) as observed in the thin sections is
presented in Figure 4.2. In the ectorganic horizons of the forest profiles, OM was mainly
present as common fresh roots, frequent to common fragmented OM (both roots and
leaves) and fine organic groundmass. In the mineral horizons of all soil profiles, the
abundance of fresh roots and fragmented OM decreased quickly with depth from (very) few
to common in the mineral topsoils to very few further down the soil profiles. Very few intact
and fragmented sclerotia occurred throughout the mineral soil profiles. Additionally, very
few irregular shaped opaque OM fragments of various sizes were present, mostly without
recognizable cell structure.

Type and distribution of bioturbation features

The vertical distribution of bioturbation features (as area %), type of bioturbation features
(as proportions of total) and microstructures (as area %) are presented in Figure 4.2 and
described in detail for each soil profile in the following sections. Figure 4.1b also shows
bioturbation features versus depth in combination with carbon contents and 14C ages.

• G7 páramo
Within the first 3 cm of the Ah1 horizon, bioturbation features were frequent and occurred
mainly as loose discontinuous infillings containing clustered spherical brown matric
excrements (50 ‐ 100μm ∅) in varying stages of welding. Welding resulted in transformation
to mammilated excrements, occurring again in clusters (Figure 4.3). In addition, very few
organic/matric dark brown mammilated excrements up to ~8 mm ∅ were present near the
surface. This zone had a crumb microstructure and showed an irregular tonguing boundary
along which massive dense complete matric infillings (~5mm ∅) were observed, composed
of fine brown groundmass with (very) few mineral grains and very few OM fragments. These
dense infillings were similar in size and composition to the ones occurring at greater depth
but had a mammilated outline owing to the greater porosity of the topsoil.

75
Figure 4.2 Bar graphs of bioturbation features (area %), type of bioturbation features (as proportions of
total), microstructures (area %) and organic matter (area %) versus sample depth (cm, top of mineral soil set
at zero) for all soil profiles.

76
Bioturbation features decreased with depth to few in the Ah horizon (at 30 cm depth),
initially mainly occurring as loose discontinuous infillings with clustered spherical to
mammilated excrements similar to the ones in the topsoil. These infillings were often
welded to such an extent that a vughy microstructure developed (Figure 4.3). In addition,
very few loose continuous infillings with welded matric spherical excrements (~100 μm ∅)
occurred in clustered and banded basic distribution patterns. From 30 cm depth onwards,
bioturbation features were mainly present as brown dense complete infillings instead of
loose infillings.
Figure 4.3 Current soil with páramo vegetation: Loose discontinuous infillings with welded spherical (example
encircled in the middle of the graph) and mammilated excrements (example encircled at top), eventually
grading into a vughy microstructure.
Strikingly, dense complete infillings increased to dominant in the 1/2Ah2 and 1/2Ahb
horizons (the overprinted zone), while loose discontinuous infillings were absent. The colour
contrast between different dense infillings increased and they were now composed of
either dark brown fine groundmass or brown to light brown fine groundmass (Figure 4.4a
and b). The mineral grains sometimes showed a crescent distribution pattern following the
outline of the dense infilling. In the overprinted zone a channel microstructure (Figure 4.5)
became dominant. The randomly distributed channels ranged from 60 – 600 μm ∅ and did
not contain excrements, some channels did contain roots.
Figure 4.4 Current soils with páramo vegetation and overprinted zones of all soil profiles: a) Brown dense
complete infilling and b) Dark brown dense complete infilling with coarser texture; dense infillings resulted in a
massive microstructure.
A B
77
Figure 4.5 Overprinted zones and top of paleosol: Figure 4.6 Ectorganic profile forest: Ellipsoidal

Channel microstructure. organic excrements (example encircled at the top
N.B. dotted appearance due to abrasive powder. of the graph) and mammilated organic excrements
(example encircled left bottom corner).

• G5b páramo next to forest patch
The type and distribution of bioturbation features in profile G5b was very similar to profile
G7 (see Figure 4.2). Within the first 2 cm of the Ah horizon, bioturbation features were
frequent and the microstructure was crumb with few massive zones. Similar to G7,
bioturbation features decreased with depth to very few at the bottom of the Ah horizon.
Loose discontinuous infillings remained present until 35 cm. Microstructure changed to
massive, with very few vughy zones where loose infillings became so strongly welded that
they were no longer recognizable as such. The proportion of dense infillings was somewhat
higher than that of loose infillings throughout the Ah horizon. As in profile G7, from the Bw
horizon onwards bioturbation features increased and occurred only as light brown to dark
brown dense infillings. However, the frequency of bioturbation features in the overprinted
zone was somewhat lower than in G7. Colour contrast between the dense infillings
increased with depth throughout the profile. At depth, a similar channel microstructure as
in profile G7 became dominant.

• G1 forest
Within the first 5 cm of the thick (75 cm) ectorganic profile of G1, bioturbation features
were frequent and represented by both single and welded excrements. Microstructure was
crumb. We encountered small spherical/ellipsoidal organic excrements, ranging from 50 –
200μm ∅. Such excrements were composed of light brown to reddish brown organic fine
groundmass without recognizable OM fragments, often associated with leaf and root tissues
(Figure 4.6). In addition, larger mammilated organic excrements (up to 800 μm ∅)
composed of light brown to reddish brown organic fine groundmass with common OM
fragments occurred. Welding increased quickly with depth.

78
Further down the ectorganic horizons, the mammilated organic excrements became so
strongly welded that they turned into fine organic groundmass mostly without recognizable
excrements (Figure 4.7a and b). The microstructure became massive in large parts of the
thin sections (Figure 4.7c). These massive zones did not qualify as ‘bioturbation feature’
according to our definition, because (welded) excrements were no longer recognizable as
such. Bioturbation features decreased with depth to very few. Still, very few single to
strongly welded organic spherical/ellipsoidal excrements were present even at the bottom
of the ectorganic horizons, often associated with roots and leaf fragments.

Figure 4.7 Ectorganic profile G1: a) Mammilated organic excrements; b) becoming more and more welded
with depth; c) Until they grade into an organic fine groundmass with massive microstructure. N.B. dotted
appearance due to abrasive powder.
A B C

In the Ah horizon, very few bioturbation features were present as loose discontinuous
infillings composed of (very strongly) welded bright reddish brown spherical organic/matric
excrements, and microstructure was massive. The fine groundmass was of similar colour as
these excrements and the boundaries between infillings and fine groundmass were diffuse.
The transition between ectorganic and mineral horizons was gradual, the Ah horizon still
containing common fresh roots and fragmented OM.

In the 1/2Bw horizon (overprinted zone), bioturbation changed drastically with regard to
type of activity, and increased in abundance to very dominant. Bioturbation features were
only present as dense complete matric infillings of the same types as encountered in profile
G7 and G5b. Bioturbation features in the 2Ahb horizon were very similar to that in the
overlying 1/2Bw horizon.

• G5a forest patch
In the ectorganic horizons of profile G5a, similar types of bioturbation features were present
as in forest profile G1. We observed that the mammilated organic excrements were often
composed of smaller excrements, which was not clear in profile G1 probably due to stronger
welding there. Contrary to G1, we did not observe zones with massive microstructure and
excrements remained recognizable. Bioturbation features remained frequent to dominant
throughout the ectorganic horizons.

79
The mineral topsoil of forest patch G5a resembled páramo profiles G7 and G5b rather than
forest profile G1. At the top of the Ah horizon, bioturbation features were frequent,
occurring as loose discontinuous infillings similar to the ones found in páramo profiles G7
and G5b. They were again often welded to such an extent that a vughy microstructure
developed. Contrary to profile G1, the transition between ectorganic and mineral horizons
was abrupt, the Ah horizon containing only very few fresh roots and fragmented OM and
lacking the reddish staining. With depth, the loose infillings decreased in abundance until
they disappeared in the 1/2 Bw horizon. Instead, very few dense complete matric infillings
similar to the ones encountered in profile G7 appeared at the bottom of the Ah horizon.
These dense infillings became dominant in the Bw horizon and very dominant in the 1/2Bw
and 1/2Ah1b horizons (overprinted zone). Simultaneously, microstructure changed to
massive with depth. Finally, in the 2Ah2b horizon dense infillings decreased somewhat in
abundance to frequent and a channel microstructure developed. The colour contrast
between dense infillings increased with depth.

Discussion

Mineral and organic components

Micromorphological observations of the mineral components confirmed the mineralogical
assemblage as determined by X‐ray diffraction and presented by Tonneijck et al. (2008a).
The dull light to dark brown colour of the fine groundmass indicated staining by OM, since
organic carbon contents were high (> 6 %) in all mineral horizons. A high organic carbon
content is typical for Andosols and can be related to the formation of amorphous organo‐
mineral/metal complexes (Dahlgren et al. 2004). Our conclusion with respect to staining by
OM was confirmed by the undifferentiated b‐fabric of the fine mineral components that
indicated masking by humus and/or the presence of amorphous materials (Stoops 2003).
The red colouring of the fine groundmass in the mineral topsoil of forest site G1 was most
likely due to OM staining and not to the presence of iron, since amorphous Fe contents
were low there and similar to the other topsoils lacking this red colouring (Tonneijck et al.
2008a). The irregular shaped opaque OM fragments may well be charcoal fragments, since
páramo vegetation in the study area is known to be subject to burning and charcoal
fragments were found by Di Pasquale et al. (2007) in soils from the same study area.

Soil faunal groups

Based on the distribution and composition of bioturbation features, we deduced which soil
faunal group probably created them. This is of great importance, since it may clarify the
vertical distribution of SOM (Anderson 1988), as explained in the Introduction. Both in the
mineral soil profiles and in the ectorganic horizons of the forest sites, the heterogeneity of
the bioturbation features encountered indicated the activity of several soil faunal species.
The small organic excrements occurring in the ectorganic horizons of the forest profile
80
obviously were formed by epigeic species. The matric composition and random distribution
of the small (50 – 100 μm ∅) excrements occurring as loose infillings in the Ah horizons
clearly suggest that these were produced by small endogeic species. Similarly, the larger (~5
mm ∅) matric dense infillings were created by larger endogeic species, as indicated by their
composition and dominant occurrence in the subsoil. Since we did not observe permanent
vertical burrows in the thin sections nor in the field, anecic species must have been scarce
or absent. The dominant soil fauna is most likely represented by Lumbricidae, Acari,
Collembola, Enchytraeidae, and Nematoda, similar to Colombian (van der Hammen &
Beglinger 1989) and Mexican Andosols (Barois et al. 1998). Detailed classification of soil
faunal species is not possible due to the morphological similarity of soil faunal pedofeatures
(Pawluk 1987).

Vertical distribution of roots

The dominant occurrence of both fresh roots and fragmented OM (including roots and
leaves) in the ectorganic horizons and/or mineral topsoil confirmed our field descriptions of
superficial root distribution (Tonneijck et al. 2006) and are in accordance with data from
Soethe et al. (2006) for Ecuadorian montane forest and data from Hofstede & Rossenaar
(1995) for Colombian páramo ecosystems. Páramo ecosystems have the highest
aboveground:belowground biomass ratio (typically 1:0.35) of a wide range of grassland
ecosystems (Hofstede & Rossenaar 1995).

Since the channel microstructure that was present in the overprinted zone and top of the
paleosol (except in profile G1) was not accompanied by an increase in fresh roots (Figure
4.2) and channels were often empty, we conclude that the root system there is no longer in
use. Because the channel microstructure occurred in both part of the current soil and the
paleosol, it was probably formed after the last tephra deposition. Possibly, roots have
shifted upwards over time. Alternatively, a deeper rooting vegetation type in the past could
have created the channel microstructure, although this seems unlikely in light of the
superficial root systems of both current forest and páramo vegetation.

Nierop et al. (2007) found with analytical pyrolysis techniques that the subsoils of our soil
profiles exhibited more of a root than of a leaf ‐derived SOM type. Root litter may have
been transported by bioturbation to the subsoil instead of being deposited there in situ
and/or prior to upward shifting roots may have reached the subsoil.

Vertical distribution of HA

The strong correlation between carbon in the HA fraction and total organic carbon in the
mineral soil profiles demonstrates that HA was transported through the soil profile by the
same mechanisms as bulk SOM. Radiocarbon dating of HA therefore offers an additional
clue regarding the vertical transport of bulk SOM. Radiocarbon dating of HA matched very
81
well with dating of charcoal fragments in mineral soil profiles from the same research area
by Di Pasquale et al. (2007), confirming the reliability of our datings. In a previous study
(Tonneijck et al. 2006) we already indicated that HA in these soils was most likely not
rejuvenated by roots or illuviation of younger OM. Only the mineral topsoils just beneath
the thick ectorganic horizons of the forest profiles formed an exception, as the younger age
of HA than of the humin fraction there did indicate some illuviation of younger HA from the
ectorganic horizons. Because of this, in those specific cases humin was used for dating.

Vertical distribution of SOM: ectorganic horizons

OM accumulation on top of the mineral soil surface demonstrated that bioturbation did not
transport OM vertically in the forest sites (G1 and G5a). Indeed, the soil faunal pedofeatures
in the ectorganic horizons were produced by epigeic species and these do not redistribute
OM (Anderson 1988). In the ectorganic horizons of forest profile G1, the strong increase in
welding with depth to the extent that a massive microstructure developed, suggested that
bioturbation is no longer active at depth. Still, the frequent occurrence of fragmented OM
tissue and the decrease in size of fragmented OM showed that bioturbation must once have
been more active. Apparently, bioturbation mainly affects fresh OM continuously supplied
at the surface, rather than older OM (up to 1483 cal AD, Table 4.1) remaining at depth.
Additionally, the very acidic pH (pHCaCl2 < 2.5) and periodically wet and anaerobic conditions
in the thick ectorganic horizons are unfavourable for soil faunal activity. At forest patch G5a,
the ectorganic horizon was thinner (35 cm) than at G1 and much younger, its base dated to
1957 cal AD (Table 4.1). Because OM was still relatively fresh, bioturbation features were
more frequent than at G1. The site also had a better drainage, resulting in less welding of
excrements as compared to profile G1. Bioturbation throughout the ectorganic horizons of
G5a is probably still active, although it does not result in vertical transport of OM.

Vertical distribution of SOM: current soils

Pessenda et al. (2001) and Rumpel et al. (2002) found a similar increase of age with depth
accompanied by a decrease in organic carbon content as in our current mineral soil profiles
and related this to a larger input of younger material to the topsoil than to the subsoil. In
our case this could have been caused by the decrease in bioturbation (loose infillings) with
depth combined with dominant OM input from the soil surface (aboveground litter) and
topsoil (superficial rooting). Additionally, at the forest sites (G1 and G5a) some OM
illuviation in the mineral topsoil occurred as explained previously.

The rejuvenating impact of fresh OM increases with increasing age difference (Mook & Van
de Plassche 1986). Apparently, bioturbation seldom transports fresh OM directly to the
older subsoil, even though some bioturbation features were found there. Indeed, fresh OM
was concentrated in the topsoil and loose and dense infillings in the current soils were
mainly composed of matric rather than organic material. The endogeic species observed to
82
be dominant in the studied soil profiles feed on SOM rather than fresh OM from the earth
surface (Anderson 1988; Lavelle 1988; Lee & Foster 1991). Mixing of SOM with small age
differences over short distances does not prevent a linear age‐depth relationship even if
bioturbation extends to considerable depth. Thus, downward SOM transport must have
occurred largely in the chronological order in which it was supplied at the surface, gradually
moving deeper into the soil.

Vertical distribution of SOM: overprinted zones

In the overprinted zone, the unusual increase of organic carbon with depth suggested that
soil fauna mixed the subsoil of the current soil with the topsoil of the paleosol (Tonneijck et
al. 2008a), which was confirmed by our micromorphological observations. Material with a
relatively low organic carbon content and relatively young age (bottom of the current soil)
was mixed into material with a higher organic carbon content and relatively high age
(paleosol), as suggested by the occurrence of dense infillings of contrasting colour. Current
roots were mainly concentrated in the topsoil (páramo) and ectorganic horizons (forest), but
past roots may have contributed to the increase in organic carbon in the overprinted zone,
as suggested by the channel microstructure. However, root input alone cannot readily
explain the gradual nature of this increase and therefore bioturbation is probably of
overriding importance.

We discarded two remaining possibilities that in theory could have caused the unusual
distribution of organic carbon contents in the overprinted zone (Tonneijck et al. 2008a).
First, while burial of a páramo vegetation would probably not result in elevated organic
carbon contents, one can imagine that burial of a thick ectorganic horizon could. However,
in that case the lithology should be that of the tephra deposit burying it, which was not
supported by grain size and geochemical data. Moreover, buried ectorganic horizons would
be recognizable as such with micromorphological analysis. Second, as mentioned before,
leaching and subsequent illuviation of OM is highly unlikely in Andosols because of the large
metal‐to‐SOM ratios (Aran et al. 2001; Dahlgren et al. 2004) and indeed we did not observe
any signs of OM illuviation (e.g. coatings) in our micromorphological analysis. Rumpel et al.
(2002) presented data on a Podzol profile with leaching as the dominant mode of OM
transport and illuviation of younger OM resulted in a radiocarbon age inversion. Such age
inversions were not observed in the overprinted zone.

In view of the magnitude of bioturbation in the overprinted zone, one would expect the
entire overprinted zone to have been homogenised, which clearly contradicts the observed
gradual increase in organic carbon content. In addition, if bioturbation indeed had resulted
in complete homogenisation of the overprinted zone, one would not expect to find a linear
age‐depth relationship but rather one average age. At the very least one would have
expected to find some age inversions, which was not the case. Apparently, analogous to
overlying horizons, mixing in the overprinted zone occurred only over short vertical
83
distances. The observed increase in contrast between dense infillings with depth indeed fits
short vertical distance mixing; since the colour contrast between the bottom of the current
soil and top of the paleosol must have been greater than the contrast between different
positions within the current soil itself. Indeed, the endogeic species dominant in the profiles
studied mainly move horizontally and are not responsible for transport over large vertical
distances (Anderson 1988).

In the top of the paleosol, a channel microstructure was superimposed on the dense
infillings. Because the channel microstructure was related to the former presence of roots
as explained previously, this implies that the dense infillings are not recently made anymore.
The simultaneous occurrence of channel microstructures and massive microstructures in the
overprinted zone between the current soil and paleosol, suggest that either the root system
was less extensive or that part of the dense infillings were formed relatively recently. The
latter explanation seems more likely since roots generally decrease rather than increase
with depth. Possibly the zone with the dense infilling type of bioturbation has shifted
upwards over time.

Short vertical distance mixing

Both in the current soils (mineral horizons) and the overprinted zones, our results suggest
that mixing by endogeic species occurred only over short vertical distances. Even though
bioturbation differed greatly in magnitude throughout the soil profiles, being much more
extensive in the overprinted zone, the slope of the linear age‐depth relationship was not
much affected. This implies that the vertical interval of mixing, i.e. the age differences
mixed, were small and similar throughout the soil profile. Since both radiocarbon ages and
organic carbon contents were not homogenised over vertical intervals of 5 cm, we conclude
that mixing must have occurred over even smaller distances.

An additional process explaining mixing over short vertical distances could be an upward
shift of the soil fauna, rendering deeper positions in the soil out of reach of bioturbation
over time. The channel microstructure superimposing the dense infillings and vice versa
indeed suggested upward shifting. Upward shifting may have been caused by soil thickening
due to SOM accumulation (Tonneijck et al. 2008a). Such soil thickening has been observed
in Mexican volcanic ash soils as well (Barois et al. 1998). Another explanation for upward
shifting could be that soil fauna itself migrated upwards in response to changes in hydrology
over time, e.g. such as caused by the formation of an impermeable placic horizon as
encountered in the soils studied.

Spatial variability of bioturbation features

Micromorphological studies are typically not suited to account fully for the spatial variability
of pedofeatures, due to the small amount of samples and number of replicates involved.
84
However, we do not expect great variability of bioturbation features within the soil
horizons, since we observed in the field that the variability of soil properties within horizons
(e.g. very fine to medium root distribution, soil moisture, porosity, structure, and colour)
was small. Our micromorphological results indeed showed that the variability of
bioturbation features in the overprinted zone was small, since bioturbation features
occupied a high proportion of the thin sections in all soil profiles studied (see Figure 4.2).
The spatial variability of bioturbation features in the current soils may be somewhat greater
because of their lower abundance. However, the cumulative effect of localised bioturbation
features over time decreases the spatial variability if they are randomly distributed, which is
likely because they were formed by endogeic species. Still, the possibility exists that we
missed bioturbation features that were less abundant, but these scarce features would then
probably not exert a strong influence on the vertical distribution of SOM. Overall, the
general trends in the vertical distribution of SOM, i.e. decreasing in the current soil profiles
and increasing in the overprinted zones, can largely be explained by our micromorphological
observations of bioturbation features.

Vegetation type and soil fauna

Vegetation type indirectly influences the soil fauna due to the palatability of its litter. (Swift
et al. 1979). The sites studied were located along an altitudinal transect intersecting the
upper forest line. Altitudinal fluctuations of this upper forest line result in a complex
vegetation history at our sites (Di Pasquale et al. 2007; Bakker et al. 2008; Jansen et al.
2008), which is reflected by changes in soil faunal pedofeatures.
At the páramo sites above the current upper forest line (G5b and G7) both loose infillings
and dense infillings produced by endogeic species were present up to the soil surface, while
they were absent at the surface of the forest sites. This suggests that endogeic species are
still active today in the páramo ecosystem while they are not active anymore in the forest.
Instead, epigeic species dominate the forest soils, eventually resulting in the formation of
thick ectorganic horizons because epigeic species do not redistribute OM (Anderson 1988).
Site G5a was only relatively recently colonised by forest vegetation (Jansen et al. 2008),
likely explaining why the mineral soil profile of G5a resembled the mineral soil profiles of
páramo (G5b and G7) rather than forest (G1) with regard to the soil faunal pedofeatures
encountered and the slope of the age‐depth relationship.
The deviating slope of the age‐depth relationship in the mineral soil of forest site G1 was
likely related to dominant accumulation of OM on the surface of the soil instead of in the
mineral soil for a prolonged period of time. Indeed, the age‐depth relationship in the forest
profile of Di Pasquale et al. (2007), with an equally thick ectorganic layer, had a similar
deviating slope. Still, the sudden appearance of dense infillings at depth in forest profile G1
could indicate the former presence of páramo vegetation even at this low altitude site. In
view of the thickness of the ectorganic profile and the gradual transition from ectorganic to
mineral horizons in G1, such páramo vegetation would predate that of site G5a. Di Pasquale
85
et al. (2007) indeed suggested that páramo was present before ~3900 cal BC at their
currently forested site at similar low altitude (3540 m a.s.l.) in the same study area.
Similarly, pollen analysis of a small mire in our study area indicated that the upper forest
line was located at a low altitude (between 3100 and 3300 m a.s.l.) both before ~4300 cal
BC and between ~210 cal BC and ~1430 cal AD (Bakker et al. 2008). Thus, site G1 could well
have been covered by páramo vegetation during these periods.

Implications
Vertical transport of SOM by bioturbation is often modelled with diffusion equations that
use uniform diffusion coefficients with depth (e.g. Elzein & Balesdent 1995; Bruun et al.
2007). However, in our case this would have resulted in a vertical distribution of SOM with
smooth turning points rather than in the observed sharp turning points. Modelling of the
vertical distribution of SOM in our soils, and probably in volcanic ash soils in general, may be
greatly enhanced by applying diffusion coefficients changing with depth as based on
observed bioturbation features, in combination with soil thickening and subsequent upward
shifting of bioturbation.

The studied soil profiles contain paleoecological proxies (e.g. pollen and biomarkers) that
are used to reconstruct the vegetation history in the study area (Jansen et al. 2008).
Because the use of paleoecological proxies contained in soils depends inter alia on their
preservation in chronostratigraphic order, the degree and type of bioturbation is of key
importance. The results from the present study suggest that as long as paleoecological
proxies were transported through the soil profile similarly to bulk SOM and HA, they will be
distributed in a (crude) chronostratigraphic order. However, a major consequence of the
present research with respect to paleoecological investigations is also that any sample taken
inevitably would produce a mixed signal.

Conclusions
The páramo soil data suggest that bioturbation was largely responsible for the vertical
distribution of SOM, while illuviation and root input were of minor importance. In the
mineral topsoils, a decrease in bioturbation by small endogeic species with depth combined
with dominant OM supply from the soil surface and topsoil resulted in a decrease in organic
carbon content and an increase of age with depth. In the overprinted zone, our data suggest
that extensive bioturbation by larger endogeic species mixed the initially relatively SOM‐
poor subsoil with the SOM‐rich underlying paleosol, which resulted in a gradual increase in
organic carbon content with depth. Our data additionally suggest that mixing throughout
the soil profile occurred only over short (< 5 cm) vertical distances. Short vertical distance
mixing was apparently enhanced by upward shifting of bioturbation as a result of soil
thickening due to SOM accumulation. A change from páramo to forest vegetation was
accompanied by a change from endogeic to epigeic species. These latter species do not
redistribute material vertically, which eventually resulted in the formation of thick
ectorganic horizons in the forest.
86
Acknowledgements

assistance. Furthermore, we acknowledge the fellow members of the RUFLE program: Jan
Sevink, Koos Verstraten, Boris Jansen, Marcela Moscol Olivera, Henry Hooghiemstra and
Antoine Cleef for their valuable contributions to this research. In this respect we also thank
Emiel van Loon, Misha Velthuis, Mirjam Pulleman, Jan Peter Lesschen and Kasper de Rooy.
Finally, we wish to thank Leo Hoitinga for his assistance in the chemical laboratory and Frans
Backer and Wijnanda Koot for thin section preparation. This research was funded by
WOTRO (WAN 75‐405) and the University of Amsterdam and generously sponsored by
Fjällraven in the form of clothing and gear.

87
5 Carbon stocks and stabilisation mechanisms in
volcanic ash soils in natural Andean ecosystems of
northern Ecuador
Submitted to Global Biogeochemical Cycles by
Tonneijck FH, Jansen B, Nierop KGJ, Verstraten JM, Sevink J & De Lange L.

Abstract

Volcanic ash soils contain very large stocks of soil organic matter (SOM) per unit area.
Consequently, they constitute potential sources or sinks for the greenhouse gas CO2.
Whether soils become a net carbon source or sink with climate and/or land‐use change
depends on the stability of SOM against decomposition, which is influenced by stabilisation
mechanisms in the soil. To clarify the role of chemical and physical carbon stabilisation
mechanisms in volcanic ash soils, we applied selective extraction techniques, performed X‐
ray diffraction analysis of the clay fraction and estimated pore size distribution of soils under
natural upper montane forest and grassland (páramo) in the Ecuadorian Andes. Our results
show that organic carbon stocks under both vegetation types roughly doubled global
averages for volcanic ash soils. The carbon stabilisation mechanisms involved are 1) direct
stabilisation of SOM in organo‐metallic (Al‐humus) complexes; 2) indirect protection of SOM
(notably aliphatic compounds) through low soil pH and toxic levels of Al and 3) physical
protection of SOM due to a very high microporosity. Moreover, in the case of the forest
soils, inherent recalcitrance of OM itself was responsible for substantial accumulation in
ectorganic horizons. Both vegetation types contributed to soil acidification, thus increasing
SOM accumulation and inducing positive feedbacks. Most types of land‐use change will
result in immediate and substantial carbon loss from these volcanic ash soils to the
atmosphere. Our results stress the urgent need to protect the Tropical Andes ‘hotspot’ from
destructive land‐use change, not only for the sake of preserving its outstanding biodiversity
but also for its function as carbon sink.

Keywords
volcanic ash soils, carbon sink, land‐use change, organo‐metallic complex, organo‐mineral
complex, pore size distribution
89
Introduction

Because soils contain the largest pool of terrestrial organic carbon, in the form of soil
organic matter (SOM), they are potential sources or sinks of the greenhouse gas CO2 of
global importance (Eswaran et al. 1993; Batjes 1996; Lal 2004). Whether soils become a net
carbon source or sink with climate and/or land‐use change depends on the stability of SOM
against decomposition. This stability is still not fully understood, being influenced by the
composition of SOM itself, by its protection against decomposition through various soil
constituents and by environmental conditions (Baldock & Skjemstad 2000). Knowledge of
the mechanisms of soil organic carbon stabilisation is therefore urgently needed (Dumanski
2004; Lal 2004). Volcanic ash soils (Andosols) contain very large stocks of SOM per unit area
(Eswaran et al. 1993; Batjes 1996; Lal 2004). They cover approximately 0.84 % of the earth’s
surface, but contain approximately 5 % of global soil carbon (Eswaran et al. 1993; Dahlgren
et al. 2004). Consequently, the impact of land‐use change on SOM in volcanic ash soils may
be particularly strong (Veldkamp 1994; Rhoades et al. 2000).

The Ecuadorian Andes, with abundant volcanic ash soils, harbours fragile tropical montane
forests and grasslands (páramo) of outstanding biodiversity, ranking them as globally
leading biodiversity ‘hot spots’ of key importance for conservation (Myers et al. 2000; Keese
et al. 2007). Currently, Ecuadorian forests suffer from the highest annual rate of
deforestation in the whole of South America (FAO 2006) and deforestation in montane
areas has only slowed down because of the inaccessibility of the scarce forests left (Keese et
al. 2007). Similarly, the páramo is seriously degrading due to grazing and burning (Hofstede
1995a; López‐Sandoval 2004). To assess the impact of land‐use change on these montane
ecosystems, it is vital to first understand their functioning under natural conditions thus
minimising the amount of confounding factors.

SOM stabilisation in Andosols is generally related to 1) the formation of organo‐mineral
complexes and/or 2) the formation of organo‐metallic (Al‐humus) complexes, each
protecting SOM from decomposition through mechanisms that are still subject of debate
(Shoji et al. 1993; Baldock & Skjemstad 2000; Dahlgren et al. 2004). Two types of Andosols
are discerned accordingly (Mizota & Van Reeuwijk 1989; Parfitt & Kimble 1989; Shoji et al.
1993; Dahlgren et al. 2004). The ‘allophanic Andosols’ contain short‐range order amorphous
aluminosilicates, e.g. allophane, imogolite, imogolite‐like allophane and ferrihydrite, and
tend to form at relatively high soil pH (5 ‐ 7) and relatively low SOM contents. At lower soil
pH (< 4.5) and higher SOM contents the ‘non‐allophanic’ Andosols predominate, presumably
because the formation of organo‐metallic complexes inhibits Al hydroxide polymerisation
and thus the formation of allophane (Shoji et al. 1993). Consequently, the non‐allophanic
Andosols are suggested to accumulate more soil organic carbon than the allophanic ones
(Shoji et al. 1993). Both types of Andosols are distributed throughout the world (Dahlgren
2004 and references therein).

90
By influencing soil pH, vegetation affects the formation of either allophanic or non‐
allophanic Andosols, thus influencing soil organic carbon accumulation. Plant roots remove
nutrients from the soil solution in the form of cations and anions. An excess uptake of
cations leads to exudation of protons, causing acidification in the vicinity of roots. By
returning nutrients to the soil through mineralisation of litter this acidification may be
counterbalanced, while on the other hand the production of organic acids in the litter layer
may increase acidification. Interaction of precipitation with the vegetation canopy also
alters its pH, leading to acid neutralisation or increased acidity in the throughfall (Ugolini et
al. 1988; Veneklaas 1990). In the case of volcanic ash soils, forest vegetation is often
reported to cause a stronger acidification than páramo vegetation, due to slower recycling
of base cations and production of organic acids in the forest floor (Ugolini et al. 1988;
Dahlgren et al. 1991).

The aim of the current paper was to quantify organic carbon stocks and clarify mechanisms
of organic carbon stabilisation in volcanic ash soils with natural upper montane forest and
páramo vegetation in a unique protected area in northern Ecuador, representative of the
Tropical Andes biodiversity ‘hotspot’ as defined by Myers et al. (2000). To reach our aim, we
applied selective extraction techniques to distinguish between aluminium species,
performed X‐ray diffraction analysis to ascertain the crystalline clay mineralogy and
estimated the pore size distribution of representative horizons to clarify the role of both
chemical and physical organic carbon stabilisation mechanisms. We combined our results
with previous research on the molecular composition of SOM in the same study area
(Nierop et al. 2007).


Study area

The study sites are located within the nature protection area of the Guandera Biological
Station in northern Ecuador near the border with Colombia, Figure 1.1 and Table 1.1.
Guandera is situated in the Eastern Cordillera and has a (semi‐)natural upper forest line
(UFL) at approximately 3650 m a.s.l. Mean annual precipitation is around 1900 mm and
mean annual temperature ranges from 12 °C at 3000 m a.s.l. to 4 °C at 4000 m a.s.l. We
selected an altitudinal transect from 3330 to 3990 m a.s.l. intersecting the current UFL,
starting with sites currently covered by continuous forest (G1, G2), a site in a forest patch
just above the UFL (G5a) and 5 sites currently covered by páramo vegetation (G6, G7, G9,
G10) of which one was located next to the forest patch (G5b). Site G5a has been only
relatively recently (~50 years ago) colonised by forest vegetation (Jansen et al. 2008), while
G1 and G2 have been forested for longer time periods of at least 500 years (Bakker et al.
2008). Site G11 concerns a forest patch at lower altitude surrounded by agricultural land,
where a roadcut exposed the soil profile.

91
All soil profiles consist of a current soil in a thick tephra deposit (~ 2800 cal BC) overprinting
a paleosol in a preceding tephra deposit (Tonneijck et al. 2008a) as a result of soil faunal
bioturbation (Tonneijck & Jongmans 2008). For this research we focused on the current soil
profiles and excluded the overprinted zone to avoid features introduced by mixing of the
subsoil of the current soil with the topsoil of the paleosol. Data for the soil profiles up to 200
cm were used only for calculation of the organic carbon stocks, to allow comparison with
global estimates. The forest profiles had ectorganic horizons (LFH) with a combined
thickness of up to 100 cm, while ectorganic horizons (L) barely exceeded 1 cm in the páramo
profiles. Páramo litter was mostly present as hanging dead material in the tussock grasses or
on the stem rosettes rather than in a litter layer. The soils changed along the altitudinal
transect from Histosols with andic properties at sites currently covered by continuous
forest, via an Andic Cambisol in the forest patch above the UFL, to Andosols at sites
currently covered by páramo vegetation (Table 1.1).

Sampling procedures

Forest and páramo litter was collected from a surface area of 50 x 50 cm. Soil pits of
approximately 1.5 m2 surface area and a depth of 1.5 m to 2 m were excavated and soil
profiles were described according to the FAO guidelines (FAO 1990) and classified according
to the World Reference Base (FAO 2006). We took bulk soil samples for chemical analyses
(approximately 1 kg fieldmoist material, with a vertical sample interval of 5 cm and covering
the width of the soil pit horizontally) and ring samples (with a volume of 100 cm3) for bulk
density determination in the same soil pit at regular depths but respecting horizon
boundaries. We took additional ring samples (with a volume of 200 cm3) in a few
representative soil horizons for determination of the water retention curve and pore size
distribution. Bulk and ring samples were taken over a vertical interval of approximately 5
cm, the middle of which was noted as the depth of the sample. Additionally, we took
undisturbed vertical soil samples (monoliths) using one or two metal gutters of 75 cm height
x 5 cm width x 4 cm depth at all sites except G2, G6, G10 and G11. For additional estimation
of organic carbon contents, samples were cored from the monoliths by means of a corer of
0.75 cm diameter. All samples were stored at 2 °C under fieldmoist conditions prior to
analysis.


Soil pH was measured with a glass electrode in demineralised H2O, in a 0.01 M CaCl2 soil
solution and in a 1 M KCl soil solution (w/v: 1:5 mineral samples and w/v 1:10 organic
samples, both on an air dry basis, using fieldmoist samples). Total carbon (C) and total
nitrogen (N) was measured with a VarioEL (Elementar) CNS auto‐analyzer. C equals organic
carbon, since carbonates were absent. We assumed that SOM content equals twice the
amount of total organic carbon content according to Van Reeuwijk (2002) and we use both
92
terms interchangeably. Soil dry bulk density was determined by weighing oven‐dried ring
samples (105 °C, 44 h mineral samples and 70 °C, 48 h organic samples).

Organic carbon stocks (kgm‐2) were calculated for every soil profile up to the maximum
depth of the current soil (excluding overprinted zone), representing organic carbon
accumulation due to current soil forming processes. We also calculated carbon stocks up to
15 cm, to minimise the effect of fluctuations in vegetation type. Furthermore we calculated
organic carbon stocks up to 30 cm, 100 cm and 200 cm depth, i.e. crossing lithological and
horizon boundaries, to allow comparison with global estimates. When a soil profile did not
reach those depths we used the maximum depth instead. We used the following equation:

i=k Equation 5.1
TCSd = ∑ b
i=1
ρs ,iCiDi (1 − Si )

With k = the number of layers up to depth d (m); TCSd = total organic carbon stock over
depth d (m); bρs,i = soil dry bulk density of layer i (kgm‐3); Ci = fraction of carbon of layer i
(kgkg‐1); Di = thickness of layer i (m) and Si is the fraction of fragments > 2 mm in layer i
(kgkg‐1). Si was negligible in the soil profiles studied. At the sites where soil monoliths were
taken, sampling intervals for carbon content were much smaller than sampling intervals for
bulk density and because of the strong relationship (r2 > 0.95) between these variables we
interpolated dry bulk density from carbon fractions using the following equation:

Equation 5.2
b
ρ = ae(bC i )
s ,i

With a = 0.9348 and b = ‐4.031 (forest) and a = 0.9247 and b = ‐5.497 (páramo). We opted
for this specific empirical relation rather than published general relations, since the latter
were demonstrated to systematically underestimate soil dry bulk density when applied to
soil types and/or areas other than those calibrated (De Vos et al. 1995).

For the estimation of total Al and Fe concentrations (Alt and Fet) in ground, dried and heated
(900 °C) mineral samples (fine earth fraction) and concentrations of major and trace
elements in ground, dried and heated (500 °C) organic samples (litter and ectorganic
horizons) a Perkin Elmer Optima 3000 XL ICP‐OES was used. Samples were destructed in a
hot HF/H2SO4 mixture (Jackson et al. 1986). Fe(II) content of the mineral sample was
estimated (without heating) by extraction with a cold HF/H2SO4/H3BO3 solution, followed by
colorimetric detection after complexation with 1.10‐orthophenantroline (Begheijn 1979). In
the same sample Fe(II) was determined again, after reduction of all Fe(III) to Fe(II) using
hydroquinone to get the Fe(III) content (Begheijn 1979). This total Fe that was
colorimetrically determined was scaled to the total Fe measured by ICP‐OES in the hot
HF/H2SO4 mixture and Fe(III) concentrations were adjusted accordingly. Loss on ignition
(LOI) was determined at 950 °C.
93
Acid ammonium oxalate (AAO) extractable Al, Si and Fe (Alo, Sio and Feo), sodium
pyrophosphate extractable Al, Fe and C (Alp, Fep and Cp) and KCl extractable Al (Alk) were
determined in fieldmoist samples (fine earth fraction) as described by Van Reeuwijk et al.
(2002). Dithionite‐citrate extractable Fe (Fed) was determined in ground and dried samples
as described by Van Reeuwijk et al. (2002). Allophane contents were calculated according to
the formula proposed by Parfitt & Wilson (1985) and modified by Mizota & Van Reeuwijk
(1989):

%Sio Equation 5.3
%Allophane = 100
(23.4 − 5.1x)

with x = (Alo‐Alp)/Sio representing the atomic ratio of allophane. As recommended by Mizota
& Van Reeuwijk (1989) we applied a maximum (Alo‐Alp)/Sio atomic ratio of 2.5, assuming
that an excess in Al should be allocated to hydroxy‐Al in the interlayered minerals.
Ferrihydrite content was estimated using Feo, according to Parfitt & Childs (1988) and Childs
et al. (1991):

%Ferrihydrite = 1.7%Feo Equation 5.4

We expressed both allophane and ferrihydrite contents on a SOM free basis to allow
comparison with the clay content which is on a SOM free basis by default. The crystalline
clay content was calculated by subtracting allophane content and ferrihydrite content from
the clay content, implying that the crystalline clay content includes some strongly bound
SOM.

The fraction < 2μm was obtained with the sieve‐pipette method, after SOM removal with
H2O2 (Gee & Or 2002). To avoid irreversible aggregation upon drying (Kubota 1972; Shoji et
al. 1993), we used fieldmoist samples. We applied either NaOH or HCl for pH adjustment to
achieve maximum chemical dispersion, according to Shoji et al. (1993). Amorphous
constituents were removed from the fraction < 2μm by AAO extraction (Van Reeuwijk et al.
2002). Oriented clay samples were then prepared for X‐ray diffraction (XRD) analysis using a
filter‐membrane technique after the following treatments: a) Mg‐saturation of air dried
sample and b) ethylene‐glycol solvated Mg‐saturated sample, c) K‐saturation of air dried
sample and d) K‐ saturated sample heated to 300 °C and/or e) 550 °C, according to Brown et
al. (1972). The samples were scanned with an X‐ray diffractometer from 2° to 35° 2θ with Cu
Kα radiation and a curved crystal monochromator. It proved to be impossible to completely
remove SOM from the samples and therefore clay orientation was not ideal, making
interpretation and quantification difficult. Nevertheless, results within and between soil
profiles were consistent. Clay minerals were visually identified by their diagnostic peaks.

Soil water retention characteristics were determined by placing undisturbed water
saturated soil ring samples on a hanging water column and draining them successively to
94
matric potentials between pF 1 – 2.2 (Richards 1948). For the water content at pF 2.7 a
kaolin table was used (Stakman et al. 1969). For the larger suctions of pF 3.0 and pF 4.2
ceramic pressure membrane apparatus were used (Stakman et al. 1969). The samples that
were used on the ceramic plates comprised soil that was crumbled into aggregates. An
additional point at very low‐pressure head was obtained with the method of very low
evaporation equilibrium as described by Stakman et al. (1969). Water retention curves were
converted to a cumulative pore size distribution according to the Laplace‐Jurin law.


Spearman correlations (bivariate) between variables were calculated using the SPSS
Correlate procedure and considered significant when p < 0.01. Differences in means
between 2 groups (being either topsoil versus subsoil or forest versus páramo) were
considered significant when p < 0.05, as determined with the non‐parametric Mann
Whitney’s U test for 2 independent samples (SPSS Inc. 2001). We treated the mineral soil
samples of forest patch profile G5a as páramo, because this site was only recently colonised
by forest (Bakker et al. 2008; Jansen et al. 2008). We regarded the mineral topsoil sample of
G11 as a subsoil sample, because of its consistent behaviour as such which is probably due
to truncation of the topsoil.

Results

Soil properties are summarised in Table 5.1, while Figure 5.1a and b show physical soil
properties. A statistical comparison between the means of forest and páramo soils and
between topsoils and subsoils with regard to several soil properties is presented in Table
5.2. Correlations between organic carbon content and several soil properties are given in
Table 5.3, while correlations between those soil properties were mentioned in the text only.
Finally, organic carbon stocks are presented in Table 5.4.

Figure 5.1 a) Water retention curves and b) cumulative pore size distribution.
× = forest site G11 at 20 cm, ■ = páramo site G7 at 10 cm and 30 cm (●), and ○ = páramo site G9 at 20 cm.
95
Table 5.1 Selected soil properties (bulk soil samples) ‐ forest sites
96

Site Horizona Depthb bulk density pHCaCl2 pHKCl C C/N Cp Clay Allophane Ferrihydrite Alp/Alo Alt Alo Alp Alk Fet Fed Feo Fep
[‐] [‐] [cm] [gcm‐3] [‐] [‐] [%] [‐] [%] [%] [%] [%] [‐] [%] [%] [%] [cmolckg‐1] [%] [%] [%] [%]
F ‐5 3.13 3.06 45 36 1.2 7.6 0.68
G11
Ah1 10 0.67 4.22 4.34 7.8 16 3.5 20 8.0 1.3 0.31 7.8 2.5 0.78 2.7 3.1 0.88 0.66 0.42
L ‐75 4.60 4.68 53 84 0.01 0.01
F ‐65 0.11 2.53 2.27 54 41 0.06 0.81 0.03
G1 F ‐45 0.12 2.29 1.82 53 43 0.13 3.5 0.06 0.01
H ‐25 0.10 2.27 1.87 52 41 0.28 7.1 0.10 0.03
Ah 7 0.43 3.18 3.35 21 27 10.8 4 0.00 1.3 1.00 5.0 0.91 0.94 16.1 1.5 0.52 0.44 0.67
Lv ‐98 4.46 4.37 53 105 0.01 0.00 0.01
F1am ‐96 3.82 3.71 53 47 0.07 0.63 0.03
F2am ‐81 0.07 2.63 2.33 53 52 0.07 0.63 0.03
F2am ‐58 0.13 2.32 1.94 53 42 0.17 5.2 0.08 0.04
G2
H ‐38 0.12 2.21 1.80 53 45 0.09 2.6 0.03 0.01
H ‐18 0.14 2.32 2.05 49 35 0.70 11.3 0.22 0.05
Ah 7 0.38 3.76 3.87 20 24 9.6 5 0.00 1.1 1.00 5.5 1.4 1.4 9.9 1.4 0.39 0.38 0.38
Bw 37 0.69 4.12 4.19 8.2 23 4.4 11 1.6 0.20 0.87 7.8 1.3 1.1 3.3 2.2 0.25 0.10 0.18
L ‐35 3.92 3.89 50 74 0.10 0.01
F ‐20 0.10 2.49 2.30 50 24 0.49 7.1 0.20
F ‐10 2.44 2.29 50 24 0.62 13.9 0.27 0.08
G5a
Ah 10 0.45 3.67 3.77 18 18 8.5 8 0.00 2.0 1.00 5.6 1.3 1.5 13.5 2.4 0.99 0.77 0.77
Ah 25 0.45 4.05 4.10 13 17 5.5 15 5.0 2.1 0.58 6.9 2.2 1.3 5.4 2.7 1.1 0.90 0.61
Bw 50 0.58 4.19 4.36 8.9 18 3.9 17 4.7 1.8 0.45 7.8 1.8 0.83 1.8 3.2 1.5 0.86 0.66
a
= Horizon nomenclature of ectorganic horizons according to Green et al. (1993)
b
= depth of sample, the mineral soil surface was set at 0 cm
Table 5.1 Continued from previous page – páramo sites

Site Horizona Depthb bulk density pHCaCl2 pHKCl C C/N Cp Clay Allophane Ferrihydrite Alp/Alo Alt Alo Alp Alk Fed Fet Feo Fep
[‐] [‐] [cm] [gcm‐3] [‐] [‐] [%] [‐] [%] [%] [%] [%] [‐] [%] [%] [%] [cmolckg‐1] [%] [%] [%] [%]
L ‐1 3.65 3.61 42 91 0.06 0.04
Ah 15 0.34 3.95 4.12 20 16 6.6 12 0.00 1.9 1.00 5.2 1.4 1.5 7.1 0.79 2.2 0.65 0.60
G5b
Ah 35 0.41 4.17 4.34 15 16 5.4 12 2.6 1.8 0.78 6.6 1.9 1.4 3.3 0.96 2.7 0.72 0.63
Bw1 55 0.54 4.44 4.58 9 15 3.6 16 6.8 1.6 0.38 8.0 2.3 0.88 1.0 1.1 3.5 0.78 0.57
L ‐1 3.76 3.81 42 86 0.04 0.02
Ah 15 0.33 3.90 4.02 22 17 6.5 10 0.00 2.2 1.00 4.9 1.4 1.4 8.6 0.82 2.1 0.71 0.54
G6
Ah 30 0.36 4.05 4.15 19 17 6.9 8 0.00 2.1 1.00 5.3 1.6 1.7 6.5 0.91 2.2 0.77 0.64
AB 55 0.43 4.40 4.46 12 16 4.5 18 7.2 2.4 0.42 7.5 2.6 1.1 1.8 1.3 3.2 1.1 0.62
L ‐1 3.82 3.80 42 92 0.05 0.02
Ah1 10 0.35 3.92 4.12 20 16 6.7 12 0.00 2.3 1.00 5.5 1.5 1.5 7.3 0.85 2.3 0.82 0.60
G7
Ah1 30 4.10 4.27 16 16 5.8 12 1.9 2.2 0.85 6.3 1.9 1.6 4.4 0.99 2.6 0.87 0.62
Ah2 47 0.42 4.30 4.45 13 16 4.8 17 6.5 2.5 0.48 7.4 2.5 1.2 2.3 1.2 3.1 1.1 0.69
L ‐2 3.49 3.47 41 92 0.13 0.08
G10 Ah 19 0.32 4.03 4.19 20 16 6.8 0.60 2.4 0.96 5.6 1.6 1.5 7.2 0.89 2.3 0.88 0.56
Ah 49 0.47 4.43 4.53 11 16 4.4 16 6.5 2.2 0.43 7.4 2.5 1.1 1.8 1.1 3.0 1.0 0.55
L ‐1 3.32 3.31 41 72 0.12 0.06
G9 Ah 10 0.33 4.05 4.21 17 16 5.9 9 0.75 1.8 0.94 5.8 1.3 1.2 5.9 0.70 2.1 0.69 0.46
Ah 25 0.44 4.15 4.33 13 15 5.2 11 2.1 1.5 0.77 6.7 1.7 1.3 4.3 0.81 2.4 0.65 0.37
a
= Horizon nomenclature of ectorganic horizons according to Green et al. (1993)
b
= depth of sample, the mineral soil surface was set at 0 cm
97
General soil characteristics

We refer to the upper 20 cm of the mineral soil profile as ‘topsoil’, below this depth we use
the term ‘subsoil’. The pHCaCl2 in the mineral soil was always lower than 4.5 and significantly
lower in the topsoil (pHCaCl2 = 3.8) than in the subsoil (pHCaCl2 = 4.2). Soil pHKCl was
significantly higher than pHCaCl2 in all mineral soil samples. Soil pHCaCl2 was highly negatively
correlated with organic C content (r = ‐0.85, p < 0.001) and exchangeable, i.e. KCl
extractable, Al (r = ‐0.99, p < 0.001). Dry bulk density was significantly lower in the topsoils
(0.37 gcm‐3) than in the subsoils (0.49 gcm‐3) and met the diagnostic criterion for Andosols
according to the WRB classification (< 0.9 gcm‐3). The water retention curve is shown in
Figure 5.1a and the pore size distribution derived from it is presented in Figure 5.1b.
Porosity was ~ 75 %, of which ~ 75 % may be attributed to micro (< 0.4 μm) and mesopores
(0.4 – 6.0 μm) and of which ~ 70 % is smaller than 3 μm.

Table 5.2 Mann Whitney’s test for differences between means (mineral soil samples) of forest and páramo or
topsoil and subsoil, significant differences (p < 0.05) indicated with italic font. For every variable the number of
samples (n), mean and standard deviation (s.d.) are given.

Forest Páramo Topsoil Subsoil

a b
Variable Unit n mean s.d. n mean s.d. p n mean s.d. n mean s.d. p
C [%] 4 14 7 16 15 4 0.705 8 20 2 12 12 3 <0.001
C/N [‐] 4 22 5 16 16 1 0.089 8 19 4 12 17 2 0.123
Cp [%] 4 7.1 3.7 16 5.7 1.3 0.850 8 7.7 1.8 12 4.8 1.0 <0.001
Cp/C [‐] 4 0.50 0.03 16 0.38 0.05 0.003 8 0.39 0.09 12 0.41 0.05 0.247
Alt [%] 4 6.52 1.48 16 6.4 1.0 0.925 8 5.38 0.32 12 7.12 0.79 0.001
Alo [%] 4 1.53 0.69 16 1.84 0.46 0.257 8 1.34 0.20 12 2.06 0.44 0.001
Alp [%] 4 1.06 0.27 16 1.32 0.26 0.089 8 1.38 0.21 12 1.19 0.29 0.143
Alk [%] 4 0.07 0.06 16 0.05 0.03 0.345 8 0.08 0.03 12 0.03 0.01 <0.001
(Alo – Alk) [%] 4 1.45 0.73 16 1.79 0.48 0.257 8 1.26 0.22 12 2.04 0.45 0.001
(Alo – Alp) [%] 4 0.48 0.84 16 0.55 0.60 0.699 8 0.02 0.03 12 0.88 0.61 0.001
(Alp – Alk) [%] 4 0.99 0.26 16 1.27 0.24 0.059 8 1.29 0.23 12 1.16 0.28 0.217
Alp/Alo [‐] 4 0.79 0.33 16 0.75 0.25 0.699 8 0.99 0.02 12 0.61 0.23 0.001
Alp/C [‐] 4 0.04 0.02 16 0.04 0.01 0.925 8 0.03 0.01 12 0.04 0.01 <0.001
Alp/Cp [‐] 4 0.08 0.03 16 0.10 0.01 0.186 8 0.08 0.02 12 0.11 0.01 0.002
(Alp+Fep)/Cp [‐] 4 0.09 0.04 16 0.13 0.01 0.047 8 0.10 0.03 12 0.13 0.01 <0.001
Fet [%] 4 2.03 0.77 16 2.63 0.46 0.108 8 2.03 0.37 12 2.83 0.43 0.002
Fed [%] 4 0.51 0.27 16 1.00 0.21 0.011 8 0.74 0.20 12 1.01 0.31 0.014
Feo [%] 4 0.40 0.23 16 0.83 0.14 0.005 8 0.67 0.17 12 0.79 0.27 0.105
Fep [%] 4 0.41 0.20 16 0.59 0.09 0.131 8 0.57 0.12 12 0.55 0.15 0.817
(Fed – Feo) [%] 4 0.11 0.09 16 0.17 0.15 0.450 8 0.08 0.07 12 0.22 0.14 0.007
Fep/Feo [‐] 4 0.90 0.18 16 0.73 0.13 0.072 8 0.84 0.15 12 0.71 0.14 0.082
Clay [%] 4 10.1 7.3 15 12.8 3.3 0.317 7 8.6 3.2 12 14.4 3.4 0.007
Cryst. clay [%] 4 6.7 3.9 16 7.4 2.4 0.925 8 5.9 3.6 12 8.2 1.3 0.123
Allophane [%] 4 2.4 3.8 16 2.8 2.8 0.699 8 0.2 0.3 12 4.4 2.7 0.001
Ferrihydrite [%] 4 1.0 0.5 16 2.0 0.3 0.002 8 1.9 0.5 12 1.8 0.6 1.000
Bulk density [gcm‐3] 4 0.54 0.16 15 0.41 0.08 0.134 8 0.37 0.05 11 0.50 0.11 0.006
pHCaCl2 [‐] 4 3.8 0.5 16 4.1 0.2 0.344 8 3.8 0.3 12 4.2 0.1 <0.001
pHKCl [‐] 4 3.9 0.4 16 4.3 0.2 0.185 8 4.0 0.3 12 4.3 0.1 0.002
a
= For explanation of subscripts see method section
b
= Ratios are atomic ratios, except for C/N which is a mass ratio
98
Soil organic matter

We define SOM as the total of all biologically derived organic matter residing in the soil
matrix (referred to as mineral soil) and directly on the soil surface (referred to as ectorganic
horizons), in line with Baldock & Skjemstad (2000). The molecular composition of SOM in
the soil profiles studied was described by Nierop et al. (2007).

C/N ratios of forest litter (C/N = 88, mass ratio) and páramo litter (C/N = 87) did not differ
significantly. The C/N ratio of forest litter was significantly higher than the C/N ratio of the
ectorganic material of the forest floor (C/N = 39). Páramo litter (pHCaCl2 = 3.6) was more
acidic than forest litter (pHCaCl2 = 4.3), but acidity of forest litter increased rapidly with depth
to pHCaCl2 values as low as 2.2 in the ectorganic horizons. Forest litter contained significantly
more K, Ca, Mg, Cu, Sr and P and significantly less Fe, Cr, Ti, Ni and Li than páramo litter.
Ectorganic material of the forest floor contained significantly more Ca, Mg, Al, Cu, Ti, Sr, P
and Li and less Mn than páramo litter. Forest litter contained significantly more Ca and Mn
and significantly less Na, Fe, Cu, Cr, Ti, Ni and Li than ectorganic material from the forest
floor.

Organic C content in the mineral soil was 15 ± 5 % and significantly higher in topsoils (20 ± 2
%) than in subsoils (12 ± 4 %). The C/N ratio in the mineral soils was on average 18 ± 3. C/N
ratios did not differ significantly between forest and páramo soils and between topsoils and
subsoils. The Cp/C ratio of the mineral soil averaged 0.40 ± 0.07 and was significantly higher
in forest soils (0.50 ± 0.03) than in páramo soils (0.38 ± 0.05) while topsoils and subsoils
showed similar Cp/C ratios.

Mineralogy of the clay fraction

Total clay content was on average 12 ± 4 mass % and significantly lower in the topsoils (9 ± 3
%) than in the subsoils (14 ± 3 %). Allophane content ranged between 0 % and 8 % (mean 3
%) and was significantly lower in the topsoils (0.16 ± 0.3 %) than in the subsoils (4.4 ± 2.6),
and did not differ between forest and páramo. Ferrihydrite content was 1.8 ± 0.6 % and did
not differ between topsoils and subsoils, but was significantly lower in forest soils (1.0 ± 0.5
%) than in páramo soils (2.1 ± 0.3 %). The correlation between total clay content and the
sum of allophane and ferrihydrite was high and positive (r = +0.89, p < 0.001), indicating that
good dispersion was achieved with the sieve‐pipette method. Allophane was present at soil
pHCaCl2 below 4.5. Allophane content increased with soil pHCaCl2 (r = +0.91 and p < 0.001).

Crystalline clay content averaged at 7 ± 3 % and did not differ significantly between topsoil
and subsoil or between forest and páramo. On average 65 ± 14 % of the clay fraction
consisted of crystalline components. The crystalline clay fraction contained predominantly
Al‐hydroxy interlayered minerals and clay sized primary minerals (albite, amphibole,
cristobalite and quartz) and only trace amounts (if any) of kaolinite, illite and halloysite.
99
Aluminium and iron species

Total amorphous aluminium (Alo‐Alk) amounted to 1.7 ± 0.5 % and 30 – 100 % of it was in
organic amorphous form (Alp‐Alk). Total amorphous aluminium content was significantly
lower in topsoils (1.3 ± 0.2 %) than in subsoils (2.0 ± 0.4 %). Alp/Alo was significantly higher in
topsoils (1.0 ± 0.03) than in subsoils (0.6 ± 0.2). Alp/Alo showed negative correlations with
clay content (r = ‐0.85, p < 0.001), allophane content (r = ‐0.99, p < 0.001) and soil pHCaCl2 (r =
‐0.92, p < 0.001). Exchangeable Al was on average 6 ± 4 cmolckg‐1 and was significantly
higher in topsoils (9 ± 4 cmolckg‐1) than in subsoils (3 ± 2 cmolckg‐1).

SOM versus other soil constituents

Organic C content in the mineral soil was positively correlated with Cp content (r = +0.91 and
p < 0.001). Organic C content increased with increasing amorphous organic Al content (as
Alp‐Alk, r = +0.60 and p = 0.005), exchangeable Al (r = +0.87 and p < 0.001) and Alp/Alo (r =
+0.88 and p < 0.001). Organic C content decreased with increasing clay content (r = ‐0.71
and p = 0.001), allophane content (r = ‐0.86 and p < 0.001), total and inorganic amorphous
Al (as Alo‐Alk, r = ‐0.64 and p = 0.002 and Alo‐Alp, r = ‐0.87 and p < 0.001) and crystalline Fe
oxides (as Fed‐Feo, r = ‐0.60, p = 0.005). Organic C content was not correlated with crystalline
clay content. These correlations were similar using Cp content instead of C content.

Table 5.3 Spearman’s correlation coefficient (r) between organic carbon
content (%) and several variables (mineral soil profiles), only significant
(p < 0.01) correlations are given, n indicates the number of samples.
Organic carbon [%]
Variablea r p n
Alt ‐0.97 <0.001 20
Alo ‐0.63 0.003 20
Alp +0.64 0.002 20
Alk +0.87 <0.001 20
Alp/Alo +0.88 <0.001 20
(Alo – Alk) ‐0.64 0.002 20
(Alo – Alp) ‐0.87 <0.001 20
(Alp – Alk) +0.60 0.005 20
Fet ‐0.76 <0.001 20
Fed ‐ ‐ 20
Feo ‐ ‐ 20
Fep ‐ ‐ 20
Fep/Feo ‐ ‐ 20
(Fed – Feo) ‐0.60 0.005 20
Clay ‐0.71 0.001 19
Crystalline clay ‐ ‐ 20
Allophane ‐0.86 <0.001 20
Ferrihydrite ‐ ‐ 20
Sample depth ‐0.65 0.002 20
Bulk density ‐0.80 <0.001 19
pHCaCl2 ‐0.85 <0.001 20
pHKCl ‐0.77 <0.001 20
a
= For explanation of subscripts see method section
100

(Alp+Fep)/Cp atomic ratios were on average 0.12 ± 0.02 in the mineral soil. Topsoils had
significantly lower (Alp+Fep)/Cp atomic ratios (0.10 ± 0.03) than subsoils (0.13 ± 0.01). C/Alp
ratios ranged between 17 and 50 (mean 27), while Cp/Alp ratios ranged between 8 and 26
(mean 11) and both were significantly higher in topsoils than in subsoils.

Organic carbon stocks

Organic carbon stocks in the ectorganic and mineral horizons are presented in Table 5.4,
together with global estimates for Andosols as derived from Batjes (1996, humic Andosols)
and Eswaran et al. (1993).

Organic carbon stocks in the ectorganic layer ranged between 5 and 60 kgCm‐2 representing
20 – 65 % of the total organic carbon stock up to 30 cm. Organic carbon stocks in the current
soils (excluding the overprinted zone) were on average 29 ± 11 kgCm‐2. Organic carbon
stocks in current topsoils (up to 15 cm) were on average 11 ± 2 kgCm‐2 at the forest sites and
12 ± 5 kgCm‐2 at the páramo sites and differences between these means were not
significant.

Organic carbon stocks averaged 22 ± 5 kgCm‐2 up to 30 cm, 53 ± 10 kgCm‐2 up to 100 cm and
87 ± 12 kgCm‐2 up to 200 cm, representing 165 ± 36 %, 180 ± 34 % and 290 ± 41 %
respectively relative to global averages (Table 3). It should be noted that while most soil
profiles were thicker than 100 cm, none of them reached a maximum depth of 200 cm
(Table 5.4).

‐2
Table 5.4 Organic carbon stocks (kgm ) for ectorganic horizons and mineral soil profiles, the latter up to the
depth of the current soil profile (excluding overprinted zone), 15 cm, 30 cm, 100 cm and 200 cm depth. Italic
font indicates that organic carbon stock calculations were based on bulk samples rather than high resolution
sampling (other profiles).

Batjes Eswaran et
G11 G1 G2 G5a G5b G6 G7 G10 G9 all sites
(1996) al. (1993)
a
TCDectorganic 5 37 60 21 ‐ ‐ ‐ ‐ ‐ 31 ± 23 ‐ ‐
b
TCDcurrent ‐ 11 ‐ 32 37 40 32 ‐ 21 29 ± 11 ‐ ‐
TCD15 13 11 10 8 11 11 11 21 10 12 ± 4 ‐ ‐
TCD30 19 22 32 17 21 20 20 28 19 22 ± 5 13 ± 8 ‐
c
TCD≤ 100 30 49 58 55 56 60 61 61 46 53 ± 10 29 ± 18 31
d
TCD< 200 ‐ ‐ ‐ 78 103 95 85 73 ‐ 87 ± 12 30 ± 13 ‐
a
= G11 (10 cm thick), G1 (75 cm thick), G2 (100 cm thick), G5a (35 cm thick)
b
= Depth of current soil up to overprinted zone for G11, G2 and G10 not known due to lack of high resolution
sampling. G1 (11 cm), G5a (57 cm), G5b (65 cm), G6 (65 cm), G7 (55cm), G9 (35 cm).
c
= For sites G11 (55 cm), G1 (71 cm), G2 (100 cm) and G9 (82 cm) this already represents the maximum depth.
d
= Up to maximum depth of soil profile, i.e. G5a (140 cm), G5b (192 cm), G6 (160 cm), G7 (147 cm) and G10 (125 cm)

101
Discussion

Vegetation and soil formation

In the study area, besides leaching, both forest and páramo vegetation contributed to soil
acidification, as evidenced by the consistently lower soil pH in mineral topsoils than in
subsoils. Nevertheless, topsoil pH at the forest sites, including the forest patch that was only
recently forested, was even lower than topsoil pH at the páramo sites, in accordance with
the trends reported by Ugolini et al. (1988) and Dahlgren et al. (1991). This may be
explained by the higher amount of bases (e.g. Ca, Mg) retained in the forest floor than in
páramo litter, thus rendering less bases available for neutralisation of the acidity. Moreover,
it is likely that the ectorganic horizons of the forest produce more organic acids contributing
to soil acidification. Because of the already low pH (~5.2) of the (rhyo‐)dacitic parent
material (Tonneijck et al. 2008a), the soils of both forest and páramo sites clearly developed
into the non‐allophanic type, with their high Alp/Alo ratios in the topsoils, high levels of
exchangeable Al, generally low allophane and ferrihydrite contents and relatively high
contents of crystalline clay in the clay fraction.

Due to vegetation‐enhanced soil acidification, the soils studied were likely transformed from
allophanic to non‐allophanic ones over time. We conclude that the soils initially must have
been allophanic, because contrary to global trends (Parfitt & Kimble 1989), allophane is
currently present in the subsoils at a soil pH well below 4.5. This clearly indicates that
subsoil allophane is a relic pertaining to the initial stages of soil formation, when the pH was
still higher and SOM contents were lower. Moreover, clay contents increased with depth,
while classical weathering would have led to the opposite trend, pointing to the breakdown
of pre‐existing allophane in the topsoils. Other possible explanations for the increasing clay
content with depth seem improbable. First, surface removal of fine particles is highly
unlikely, because this process would not preferentially remove amorphous minerals
whereas in our case the increase in clay content with depth was related only to an increase
in amorphous clay content. Second, eluviation in topsoils and subsequent illuviation in
subsoils is unlikely, because dispersion of the clay fraction of volcanic ash soils is notoriously
difficult and therefore transport of clay minerals is generally reported to be negligible
(Dahlgren et al. 1991; Aran et al. 2001). With regard to the topsoils, it is likely that
environmental conditions at the onset of soil formation also favoured allophane formation,
provided that soil pH did not immediately drop to levels below 4.5.

Although both non‐allophanic in nature, a striking difference between soils under forest and
páramo vegetation is the accumulation of SOM both in and on top of the mineral soil in the
forest, versus the accumulation of SOM only in the mineral soil in the páramo. This feature
even leads to the classification of the forest soils as Histosol rather than Andosol. Because
both aboveground and belowground litter input is reported to be rather similar in Andean
upper montane forest and páramo (Hofstede & Rossenaar 1995; Röderstein et al. 2005), this
102
must largely be related to slower litter decomposition and slower incorporation by soil
fauna into the soil in the forest. Apparently, forest litter is less palatable than páramo litter.
C/N ratios are similar, but tannins are more abundant in (Clusia) forest litter than in páramo
litter at our sites (Nierop et al. 2007), and these may indeed hamper decomposition (Kraus
et al. 2003). With regard to vertical transport of organic matter into the soil, the dominant
soil faunal species in the forest soils do not redistribute material vertically, while the
dominant soil faunal species in the páramo indeed caused bioturbation and concurrent
transport of SOM (Tonneijck & Jongmans 2008).

SOM stabilisation in the mineral soil

SOM stabilisation in soils depends on the one hand on the composition of SOM itself and on
the other hand on its protection against decomposition by various soil constituents and by
environmental conditions (Baldock & Skjemstad 2000). In volcanic ash soils, SOM is
generally suggested to be protected against decomposition in organo‐mineral and/or
organo‐metallic complexes (Shoji et al. 1993; Baldock & Skjemstad 2000; Dahlgren et al.
2004). In the following sections we will first discuss the molecular composition of SOM in
the soils studied (summarised from data presented by Nierop et al. 2007) and then focus on
the different mechanisms of SOM protection by soil constituents.

Despite differences at the molecular level between litter of forest and páramo in our study
area, the molecular composition of SOM eventually accumulating in the mineral soil profile
was remarkably similar and dominated by aliphatic compounds of probably primary (plant‐
derived) origin. The scarcity of lignin in the mineral soil suggests that decomposition is
performed predominantly by fungi rather than bacteria, since fungi are the only
decomposers capable of breaking down lignin (Kögel‐Knabner 2002). Additionally, the acidic
soil pH favours fungal rather than bacterial decomposition (Bumpus 1993). Unpublished
micromorphological data indicates that fungal hyphae and spores occur frequently in the
ectorganic horizons of the forest soils, but that these are scarce in the mineral soils of both
forest and páramo sites (provided by J. Van Mourik 2008). Apparently, microbial
transformations predominantly take place in ectorganic horizons, while in the mineral soil
matrix it is the soil fauna that may further transform SOM (Tonneijck & Jongmans 2008).
Because aliphatic compounds are only present in limited amounts in plants, their relative
enrichment in the soil indicates extensive transformation of other primary OM compounds.
Extensive transformation of primary OM has also been reported in allophanic Andosols,
although in these soils, besides lignin, aliphatic components were also scarce while
secondary (microbial) polysaccharides predominantly accumulated (Nierop et al. 2005;
Buurman et al. 2007). Accumulation of polysaccharides is mainly attributed to stabilisation
by allophane (Baldock & Skjemstad 2000; Nierop et al. 2005; Buurman et al. 2007), which
explains the much lower abundance of polysaccharides in our non‐allophanic Andosols with
lower allophane contents (Nierop et al. 2007).

103
Assuming all organic matter in organo‐metallic complexes to be part of the Cp fraction, at
most 40 % of the accumulated carbon in the soils studied can be explained by formation of
such complexes, since Cp/C was 0.4 on average. In fact, the actual percentage of carbon in
organo‐metallic complexes was most likely even less than 40 % as Cp encompasses more
than organo‐metallic complexes (Mizota & Van Reeuwijk 1989). The aliphatic compounds
that predominate in the soils studied are not typically stabilised in organo‐mineral/metallic
complexes due to their generally low amount of reactive groups and long chain lengths. This
explains the high C/Alp ratios compared to the Cp/Alp ratios. Therefore, SOM stabilised in
these organo‐metallic complexes is probably from secondary origin, as was also suggested
by Buurman et al. (2007) for non‐allophanic Andosols from Costa Rica.

Organo‐mineral complexes did not significantly contribute to total organic carbon
stabilisation, as evidenced by the negative correlations between C content, clay content and
allophane content and the absence of a correlation between C content and crystalline clay
content. Podwojewski et al. (2002) and Powers & Schlesinger (2002) reported similar trends
in other non‐allophanic volcanic ash soils. This clearly sets non‐allophanic Andosols aside
from many other soil types including allophanic Andosols, because clay minerals in general
and in particular those with a high specific surface area such as allophane are found to
enhance SOM accumulation (Torn et al. 1997; Baldock & Skjemstad 2000). Many models of
global soil carbon dynamics incorporate clay content as a controlling factor of SOM
stabilisation (Smith et al. 1997). We conclude that this assumption may not be adequate in
the case of non‐allophanic Andosols, which may lead to serious underestimation of organic
carbon stocks in these soils.

Instead, our results imply that a striking amount of organic carbon in the volcanic ash soils
studied is stabilised by other processes than direct protection in organo‐metallic or organo‐
mineral complexes. The strong positive correlation between C content and exchangeable Al
indicates that the latter contributed to organic carbon stabilisation, albeit indirectly.
Exchangeable Al exceeded 2 cmolckg‐1 in all topsoils and the greater part of the subsoils and
may thus be regarded as ‘toxic’ (Shoji et al. 1993; Dahlgren et al. 2004). Moreover, high
levels of exchangeable Al result in soil acidification through hydrolysis. Although these
‘toxic’ levels of exchangeable Al did not prevent transformation of most primary organic
compounds as discussed before, it may explain the relative accumulation of aliphatic
compounds. In allophanic Andosols lacking toxic levels of Al in most of the soil profile,
aliphatic compounds were indeed degraded (Nierop et al. 2005; Buurman et al. 2007).

In addition to Al toxicity, physical protection of SOM by soil constituents may play an
important role, through the arrangement of pores and soil particles determining SOM
accessibility and water and oxygen availability to decomposer organisms (Baldock &
Skjemstad 2000). Decomposition is inhibited at higher proportions of small pores, because
of the exclusion of decomposer organisms and because of the associated high water
retention capacity facilitating anaerobic conditions. Upon weathering, porosity in Andosols
104
generally increases markedly, which is accompanied by a shift from macropores (> 6.0 μm)
to micropores (< 0.4 μm) (Shoji et al. 1993). In the soils studied, porosity in representative
páramo and forest soil horizons was indeed very high with an exceptionally high proportion
of micro‐ and mesopores. Approximately 70 % of the pores was smaller than 3 μm, which
was reported as the lower limit for bacteria entry (Baldock & Skjemstad 2000 and
references therein). The actual proportion of micro and mesopores must be even larger,
because of aggregation during determination of the water retention curve. In view of the
low contents of allophane in non‐allophanic Andosols, this high porosity is due to the
accumulation of SOM itself, which apparently leads to a positive feedback. Buytaert et al.
(2005) and Poulenard et al. (2003) indeed reported that organic colloids caused a high fine
porosity in non‐allophanic Andosols in Ecuadorian páramo. Soil fauna played an important
role in enhancing the fine porosity in the páramo soils studied, since soil faunal activity
resulted in a massive microstructure without visible voids (Tonneijck & Jongmans 2008).

Organic carbon stocks

Organic carbon stocks in volcanic ash soils under upper montane forest and páramo sites
are very high, roughly doubling global averages given for Andosols (Eswaran et al. 1993;
Batjes 1996). The mineral soil profile under forest vegetation may to some extent be a relic
from a period with páramo vegetation, because of natural UFL fluctuations which is a typical
process in these tropical Andean ecosystems. Based on micromorphological observations
we conclude that at least the current topsoil of G1 must have been formed under forest
vegetation (Tonneijck & Jongmans 2008). To deduce the effect of vegetation type on carbon
accumulation, we therefore compared carbon stocks contained in the mineral topsoils (up
to 15 cm) and concluded that organic carbon stocks under forest and páramo vegetation
were similar (Table 3). Podwojewski et al. (2002) and Poulenard et al. (2003) already
reported very high organic carbon stocks in other non‐allophanic páramo soils throughout
Ecuador and we now demonstrate that similar organic carbon stocks may accumulate in
mineral soils under forest vegetation.

Sites G1 and G2 represent undisturbed forest, and the carbon stocks in their ectorganic
horizons exceeded the already huge organic carbon stocks in their mineral soil profiles. The
organic carbon stocks contained in ectorganic horizons are very sensitive to land‐use
change, e.g. deforestation or reforestation. This is illustrated by the much lower carbon
stock at site G11, which may be explained by its proximity to agricultural fields causing
disturbance. Contrary, site G5a contains a considerable organic carbon stock in its
ectorganic horizon in view of its recent colonisation by forest.

105
Implications

The capacity of a soil to stabilise organic carbon is finite and depends on the saturation of
the particular stabilisation mechanism involved (Baldock & Skjemstad 2000). Accumulation
of SOM in ectorganic horizons as a result of its inherent recalcitrance will eventually
approach a steady state where input and output are balanced. Stabilisation of secondary
OM in organo‐metallic complexes is probably already at a maximum in our soils, as indicated
by the high (Alp+Fep)/Cp ratios (Dahlgren et al. 2004). However, primary aliphatic compounds
may continue to accumulate as long as the pH remains strongly acidic and the porosity
remains high and dominated by small pore sizes. Mean residence times of SOM in volcanic
ash soils are generally reported to cover very long time scales (Shoji et al. 1993; Torn et al.
1997; Dahlgren et al. 2004). In the soils studied, SOM in ectorganic layers (forest) is at least
stabilised at the century time scale, while SOM in the mineral soil (forest and páramo) is
stabilised for several millennia (Tonneijck et al. 2006; Tonneijck & Jongmans 2008).

Most types of land‐use change involving upper montane forest and páramo ecosystems are
expected to cause a net flow of carbon from the soil to the atmosphere. Deforestation and
subsequent conversion to agricultural land would at least cause the immediate and
substantial loss of carbon from the ectorganic horizons. Liming to alleviate Al toxicity would
evidently promote the loss of carbon from the soil to the atmosphere by reducing soil
acidification, which would not only increase decomposition of aliphatic compounds but also
of OM in organo‐metallic complexes as demonstrated by Takahashi et al. (2006). Exposing
bare soil, tilling and draining volcanic ash soils for agricultural use would alter physical soil
properties because of irreversible changes upon drying, eventually leading to soil erosion
and transforming the soils from sinks to sources of carbon (Shoji et al. 1993; Poulenard et al.
2001; Dahlgren et al. 2004; Buytaert et al. 2006). Burning and grazing of the páramo will
reduce OM input into the soil, will increase soil erosion and will prevent the UFL from
moving upwards (Hofstede 1995a; Laegaard 1992; Poulenard et al. 2001; Podwojewksi et al.
2002; Buytaert et al. 2005). Contrary, reforestation below the natural UFL with native tree
species and protection of the páramo from grazing and burning is expected to enhance the
carbon sink function of non‐allophanic Andosols.

106
Conclusions

Organic carbon stocks contained in volcanic ash soils under natural upper montane forest
and páramo in northern Ecuador were exceptionally high, roughly doubling global averages
for this soil type. The organic carbon stabilisation mechanisms involved are 1) direct
stabilisation of probably secondary OM in organo‐metallic (Al‐humus) complexes; 2) indirect
protection of primary SOM (notably aliphatic compounds) through low soil pH and toxic
levels of Al and 3) physical protection of OM as a result of a high porosity and great
proportion of micropores. Moreover, in the case of the forest soils, inherent recalcitrance of
OM itself was responsible for the vast accumulation of OM in ectorganic horizons. By
enhancing soil acidification, both páramo and forest vegetation contributed to the
transformation of initially allophanic to non‐allophanic volcanic ash soils, thus increasing
SOM accumulation in the soils and inducing positive feedbacks. SOM in ectorganic layers is
at least stabilised at the century time scale, while SOM in the mineral soil is stabilised for
several millennia. Most types of land‐use change will result in substantial carbon loss from
these ecosystems to the atmosphere. This stresses once more the urgent need to protect
the Tropical Andes ‘hotspot’ from destructive land‐use change, not only for the sake of
preserving its outstanding biodiversity but also for its function as carbon sink.

Acknowledgements

assistance. Furthermore, we acknowledge the fellow members of the RUFLE program:
Marcela Moscol Olivera, Henry Hooghiemstra and Antoine Cleef for their valuable
contributions to this research. Jan Van Mourik is acknowledged for the micromorphological
observations. Finally, we wish to thank Bert De Leeuw, Leo Hoitinga, Piet Wartenbergh and
Ton Van Wijk for their assistance in the laboratory. This research was funded by WOTRO
(WAN 75‐405) and the University of Amsterdam and generously sponsored by Fjällraven in
the form of clothing and gear.

107
6 Modelling vertical organic matter distribution in a
volcanic ash soil in the Ecuadorian páramo:
impact of soil volume change
To be submitted in revised form to the European Journal of Soil Science by
Tonneijck FH, Velthuis M, Bouten W, Van Loon E, Sevink J & Verstraten JM.

Abstract

Volcanic ash soils contain very large stocks of soil organic matter (SOM) per unit area.
Consequently, they constitute potential sources or sinks for the greenhouse gas CO2 that are
of global importance and thus merit special attention. Because a large fraction of SOM is
generally stored in the subsoil, a thorough understanding of the processes affecting the
vertical distribution of SOM is needed to properly predict impacts of climate change and/or
land use change on soil organic carbon stocks. Soil faunal bioturbation (‘bioturbation’) is
often cited as the dominant process influencing the vertical distribution of SOM. An
additional but often overlooked process that affects the vertical distribution of SOM is soil
volume change. Soil volume change may influence depth‐dependent processes such as
bioturbation. We previously investigated the role of bioturbation in volcanic ash soils under
grassland vegetation in the northern Ecuadorian Andes and hypothesised that bioturbation
has shifted upwards in time in response to soil thickening. In the current investigation, we
determined the extent of soil volume change with a geochemical mass balance involving
immobile element concentrations. Furthermore, we tested our hypothesis with a one‐
dimensional partial differential equation model. Our results show that soil volume change is
a major soil forming process in the volcanic ash soil studied. SOM accumulation resulted in
more than a tripling of initial soil thickness within 4800 years, from ~20 cm to 77 cm. Our
modelling exercise confirmed that an upward shifting bioturbation zone in response to soil
thickening is indeed plausible. We conclude that soil volume change cannot be ignored
when modelling soil organic carbon dynamics in SOM‐rich soils such as volcanic ash soils.

Keywords
soil organic carbon, model, vertical distribution, bioturbation, volcanic ash soil, volume
change
109
Introduction

Because soil organic matter (SOM) constitutes the largest pool of terrestrial organic carbon,
soils are potential sources or sinks of the greenhouse gas CO2 that are of global importance
(Eswaran et al. 1993; Batjes 1996; Swift 2001; Lal 2004). This is reflected in global models for
carbon cycling by incorporating SOM dynamics. Most models of SOM dynamics focus on the
topsoil (< 25 cm depth) and lack spatial dimensions (Elzein & Balesdent 1995; Smith et al.
1997; Jenkinson et al. 2008). However, a large fraction of SOM is stored in the subsoil
(Batjes 1996), where its stability against decomposition is supposedly higher. Thus, a
thorough understanding of the processes affecting the vertical distribution of SOM is
needed to properly predict impacts of climate change and/or land use change on soil
organic carbon stocks (Elzein & Balesdent 1995; Jobbágy & Jackson 2000; Jenkinson et al.
2008). The vertical distribution of SOM in volcanic ash soils merits special attention, because
volcanic ash soils (mainly Andosols) contain very large stocks of SOM per unit area (Eswaran
et al. 1993; Batjes 1996; Lal 2004; Tonneijck et al. 2008b). Consequently, the impact of
climate change and land use change on SOM in volcanic ash soils may be particularly strong
(Veldkamp 1994; Rhoades et al. 2000).

The vertical distribution of SOM in soils is determined by 1) input ‐ above and belowground ‐
of litter, 2) decomposition and 3) vertical transport. In the case of volcanic ash soils tephra
deposition also influences the vertical distribution of SOM. With respect to the vertical
transport of SOM, soil faunal bioturbation (further referred to as ‘bioturbation’) is often
cited as a major process (Anderson 1988; Lavelle 1988; Rasse et al. 2006). Elzein &
Balesdent (1995) demonstrated in a modelling study that bioturbation is the dominant
mode of vertical transport, relative to leaching, in a range of soil types. Leaching of SOM is
rather insignificant in volcanic ash soils because of large metal‐to‐SOM ratios (Aran et al.
2001; Dahlgren et al. 2004). Generally, bioturbation is modelled with a diffusion equation of
which the diffusion constant is assumed to either be uniform or to decrease with depth
(O’Brien & Stout 1978; Elzein & Balesdent 1995; Bruun et al. 2007). However, in many soils
the actual bioturbation pattern may be much more complex because of the presence of
different groups of soil faunal species operating at different depths (Anderson 1988; Lavelle
1988; Lee & Foster 1991; Tonneijck & Jongmans 2008).

An additional process that may affect the vertical distribution of SOM is soil volume change.
Soil volume changes during soil formation because of e.g. chemical element loss upon
weathering, SOM incorporation and biological activity (Brimhall & Dietrich 1987; Chadwick
et al. 1990; Salvador‐Blanes et al. 2007). These soil forming processes all affect the dry bulk
density of the soil by changing the mean solid soil particle density and/or the porosity (Lee
& Foster 1991; Blanco‐Canqui et al. 2006; Rühlmann et al. 2006) and thus may cause the soil
to collapse or to thicken depending on their relative importances (Salvador‐Blanes et al.
2007). Volume change influences soil forming processes that are related to soil depth, such
as weathering, decomposition, bioturbation and rooting. While this effect is taken into
110
account in soil genesis models (Chadwick et al. 1990; Salvador‐Blanes et al. 2007), to our
knowledge it was not previously taken into account in models dealing specifically with soil
organic carbon dynamics. Still, the effect of soil volume change on the vertical distribution
of SOM may be substantial, particularly in the case of SOM‐rich soils. Soil volume change
can be calculated with a geochemical mass balance approach involving immobile element
concentrations (e.g. Ti, Al, Zr) and soil dry bulk density, provided that parent material
characteristics are reasonably well known (Chadwick et al. 1990). In soil carbon models, soil
volume change can effectively be taken into account by predicting soil dry bulk density from
SOM content, since SOM content explains most of the variation in soil dry bulk density in
many soil types (e.g. De Vos et al. 2005; Heuscher et al. 2005).

We previously investigated the effect of tephra deposition and bioturbation on the vertical
distribution of SOM in volcanic ash soils under tropical montane grassland (páramo)
vegetation in the northern Ecuadorian Andes. These soil profiles consist of a thick current
soil in a tephra deposit superimposed on a SOM‐rich paleosol in a preceding tephra deposit
(Tonneijck et al. 2008a). We observed a progressive downward increase in SOM from the
relatively SOM‐poor subsoil of the current soil into the SOM‐rich top of the paleosol, further
referred to as the ‘overprinted zone’. Bioturbation by two different groups of soil faunal
species, operating predominantly in either the topsoil or the subsoil, was found to be
responsible for this vertical distribution of SOM (Tonneijck & Jongmans 2008). The great
abundance of bioturbation features throughout the overprinted zone seemingly conflicts
with the gradual increase in SOM content with depth in that zone. In view of the magnitude
of bioturbation in the overprinted zone, one would have expected SOM in the entire
overprinted zone to be homogenised. Therefore, we hypothesised that bioturbation must
have shifted upwards over time in response to soil thickening, rendering deeper positions in
the soil out of reach of bioturbation (Tonneijck & Jongmans 2008).

The aims of this research were 1) to determine the extent of soil volume change and 2) to
test our hypothesis of the effect of soil thickening on bioturbation, i.e. upward shifting, for a
volcanic ash soil under páramo vegetation in northern Ecuador. To reach our aims we
calculated soil volume change using a geochemical mass balance approach and we
conducted a ‘proof of concept’ by testing our hypothesis with a one‐dimensional partial
differential equation model that was constructed specifically for this purpose.

Description of study area and sites

The study area is located in the nature protection area of Guandera Biological Station in
northern Ecuador, near the border with Colombia (Figure 1.1). For the current study, we
selected a site (G7, N 0˚35’48’’/ W 77˚41’25’’) at 3860 m a.s.l. that has been covered by
páramo vegetation for the last millennia (Bakker et al. 2008). Dominant species in the
páramo are bunch‐grass Calamagrostis effusa KUNTH (STEUD.) and stem rosette Espeletia
pycnophylla CUATREC. The páramo belongs to the globally leading Tropical Andes
111
biodiversity ‘hotspot’ (Myers et al. 2000). Mean annual precipitation is around 1900 mm
and mean annual temperature ranges from 12 °C at 3000 m a.s.l. to 4 °C at 4000 m a.s.l.
Both precipitation and temperature show little seasonal variation. The soil climate is
isomesic and perudic.

The soil profile was formed in three tephra deposits of Holocene age, with similar mineral
assemblage (Tonneijck et al. 2008a). The timing of the last tephra deposit is estimated at
2779 cal BC (Tonneijck et al. 2008a), thus resulting in ~4800 years of soil formation. Related
to this tephra stratigraphy the soil contained a current soil, a paleosol and a second heavily
truncated or immature paleosol, each at least 40 cm thick. The main focus of this research
was on the current soil profile. Aboveground litter was mostly present as hanging dead
material in the tussock grasses or on the stem rosettes rather than in a litter layer. Between
the current soil and the first paleosol an overprinted zone with transitional organic carbon
contents, grain size distribution and chemical element ratios was present. In the soil horizon
designations we used ‘1/2’ as a prefix to indicate this overprinted zone. A thin placic horizon
was present at the boundary between the first and second paleosol, which is ascribed to a
strong textural contrast hindering drainage of the otherwise well drained soils. The soil
classified as an Andosol according to the World Reference Base (FAO 2006).


First we describe sampling and laboratory procedures. Then the method to determine the
extent of soil volume change with a geochemical mass balance approach involving immobile
elements is outlined in a separate section. In the subsequent section the construction of the
dynamic model to test our hypothesis of an upward shifting bioturbation zone is discussed.

Sampling and laboratory procedures

The datasets used for geochemical mass balance calculations and modelling were presented
by Tonneijck et al. (2008a) and Tonneijck & Jongmans (2008). We briefly discuss the
sampling and laboratory procedures here. We took bulk soil samples for physical and
chemical analyses and ring samples for soil dry bulk density estimation and undisturbed box
samples (with dimensions of 7 cm height x 5 cm width x 4 cm depth) for the preparation of
thin sections in the same soil pit at regular depths but respecting horizon boundaries.
Additionally, we took undisturbed vertical soil samples (monoliths) using one or two metal
gutters of 75 cm height x 5 cm width x 4 cm depth. For additional estimation of organic
carbon contents, samples were cored from the monoliths by means of a corer of 0.75 cm
diameter. All samples were stored at 2 ˚C under field moist conditions.

Total carbon was measured with a VarioEL (Elementar) CNS auto‐analyser. Total carbon
equals organic carbon, since carbonates were absent. We assumed that SOM content equals
twice the amount of total organic carbon content according to Van Reeuwijk (2002) and we
112
use both terms interchangeably. Soil dry bulk density was determined by weighing oven‐
dried volumetric samples (105 ˚C, until constant weight > 24 h). Contents of the immobile
elements Ti and Al were estimated with a Perkin Elmer Optima 3000 XL ICP‐OES in ground,
dried and heated (900 ˚C) bulk samples (n = 4 for current soil and top of overprinted zone
and n = 13 for the reference material) after destruction in a hot HF/H2SO4 mixture.

Geochemical mass balance calculation

For the calculation of soil volume change, we used the strain equation similar to Brimhall &
Dietrich (1987) and Chadwick et al. (1990):

b
ρ s ,r C i, r Equation 6.1
ε i (x , t end ) = −1
b
ρ s ,w (x , t end )C i,w (x , t end )

Where ε(x,tend) is strain (cmcm‐1), or change in thickness at tend = 4800 (y) over initial
thickness at t0 = 0 (y) due to soil formation, at depth x (cm) with subscript i referring to
strain determination by use of an immobile strain index element. Depth x is defined relative
to the soil surface (xtop), with positive values below the soil surface, which is common in soil
science (FAO 1991). Depth x is time‐dependent due to soil volume change, which we do not
explicitly mention in the equations for reasons of readability. Furthermore, bρs,r and bρs,w
represent the soil dry bulk density (gcm‐3) of the reference material (subscript r) and
weathered material (subscript w) respectively and Ci,r and Ci,w represent the mass
percentage (%) of immobile element i in the reference material and weathered material
respectively. Note that bρs,r and Ci,r are not depth‐dependent. We used the relatively
unweathered second paleosol as reference material, which is justified because the tephra
deposit it was formed in likely originated from the same volcanic source as the last tephra
deposit as evidenced by their similar Ti/Al ratios and similar mineral assemblage (Tonneijck
et al. 2008a). To account for natural variability in this reference material, minimum and
maximum strain values were obtained by using the standard deviations of bρs,r (n = 8) and Ci,r
(n = 13).

We used linear regressions to interpolate strain at a given soil depth x (Figure 6.1a and b):

εi (x , tend ) = ax + b Equation 6.2

Parameters ‘a’ and ‘b’ are presented in the captions of Figure 6.1a and b. The r2 of these
linear regressions exceeded 0.92 in all cases. In the overprinted zone, mixing of the
relatively SOM‐poor current soil with the SOM‐rich paleosol caused somewhat higher strain
values. Therefore, applying linear regressions represents a conservative estimate of strain in
the subsoil.

113
Figure 6.1 Strain versus sample depth (cm) computed with a) Ti as immobile index element or b) Al as
immobile index element.
● = minimum strain, with εTi,w = ‐0.0366 x + 3.6625 and εAl,w = ‐0.0355 x + 3.7852
○ = maximum strain, with εTi,w = ‐0.0535 x + 5.8227 and εAl,w = ‐0.0428 x + 4.6948
× = average strain, with εTi,w = ‐0.0447 x + 4.7069 and εAl,w = ‐0.0393 x + 4.2301

To calculate the initial thickness at t0 of the entire current soil, i.e. before volume change,
we applied the following equation, using equation 6.2:

L(x , t end ) 1 Equation 6.3
L(x , t 0 ) = = →
1 + ε(x , t end ) 1 + ax + b
x current x current x
⎡ ln(ax + b + 1) ⎤ current
∫ ∫
1
L current (t 0 ) = L(x , t 0 )dx = dx = ⎢ ⎥⎦ →
1 + ax + b ⎣ a x top
x top x top
1
L current (t 0 ) = (ln(ax current + b + 1) − ln(b + 1))
a

With L(x,t0) representing the initial thickness (cm) of a layer and L(x,tend) representing the
thickness of this layer after soil formation which we arbitrarily set at 1 cm. The total
observed thickness of the entire current soil at tend is 77 cm and referred to as Lcurrent(tend),
being the thickness from the mineral soil surface (xtop) up to the middle of the overprinted
zone (xcurrent), see Figure 6.3. L(x,t0) was integrated from xtop to xcurrent to obtain the initial
thickness of the entire current soil i.e. Lcurrent(t0), see Figure 6.3. We computed minimum and
maximum Lcurrent(t0) using minimum and maximum strain values.

Correct application of this technique depends on immobility of at least one element.
Therefore, we used Ti and Al as probable immobile elements to allow for an evaluation of
internal consistency. Furthermore, immobility of an element can be checked against
another supposedly immobile element and vice versa, by calculating its transport (Chadwick
et al. 1990) with:

114
Equation 6.4
b
ρ s ,w (x , t end )C j, w (x , t end )
τ j (x , t end ) = (ε i (x , t end ) + 1)
bρ C
s ,r j, r

Where τj (x,tend) is the mass fraction of mobile element j added or subtracted to the system.
To account for natural variability in the reference material, transport was calculated using
minimum and maximum strain values.

Dynamic modelling

General model structure

The soil profile is described by a vector where each element represents a soil layer, with an
initial thickness of 5 cm. For pragmatic reasons we assume that layers have a unit area of 1
cm2. During soil formation each layer is subject to volume change as explained in the
introduction and therefore layer thickness is variable in time. The upper boundary of the
system is the soil surface (xtop) and the lower boundary is chosen just above the placic
horizon (xbase).

The model uses time steps of five years. We start our model run (t0 = 0 y) at the moment of
the last tephra deposition, i.e. before formation of the current soil and overprinted zone.
The timing of the last tephra deposit is estimated at 2779 cal BC (Tonneijck et al. 2008a),
thus resulting in ~4800 years (tend) of soil formation. It should be noted that due to mixing of
the top of the paleosol with the subsoil of the current soil, it has become impossible to
obtain an accurate timing of the event of tephra deposition with radiocarbon dating and this
timing should thus be regarded as a ‘best guess’.

Organic carbon mass and mineral mass are the state variables. We use mass based units
rather than volume based units, because volume changes during simulation. The fluxes are
determined by diffusion and decay equations, which are discussed in the following sections.

Variable layer thickness

Geochemical mass balance calculations (see Results section) showed that soil volume
increase due to SOM incorporation is by far the dominant process in the soil profile studied.
Therefore, we used an empirical relation between organic carbon content and soil dry bulk
density as based on data of páramo soils from the same study area (Tonneijck et al. 2008a,
r2 = 0.98) to recalculate soil dry bulk density of a layer at depth x and year t:

Equation 6.5
b
ρ (x , t) = cedfOC (x ,t)
s ,w

With c = 0.9247 gcm‐3 and d = ‐5.497 and fOC(x,t) = fraction of organic carbon gg‐1.
115
We opted for this specific empirical relation rather than published general relations, since
the latter were demonstrated to systematically underestimate soil dry bulk density when
applied to soil types and/or areas other than those calibrated (De Vos et al. 1995). Because
of the exponential nature of the relation between organic carbon and soil dry bulk density
we assume that this relation reflects the combined effect of all soil forming processes and
not only of SOM incorporation.

Soil dry bulk density is subsequently used to recalculate the thickness of a layer at depth x
and year t according to the following equation:

m (x , t) Equation 6.6
L(x , t) = b b
ρs ,w (x , t)U

With L(x,t) = the thickness of a layer; mb(x,t) = the sum of mineral and SOM mass (g), of that
layer and U = unit area (cm2). Integrating L(x,t) from xtop to xbase gives Ltot(t) for the entire soil
profile. In order to keep the layers from growing uncontrollably, layer splitting is introduced.
Layers that grow beyond a thickness threshold of 10 cm are divided into two equally thick
layers, while taking into account the slope of the organic fraction with depth to reduce the
effect of numerical errors.

Input

Organic carbon enters the soil as aboveground (shoots) and belowground (roots) litter. The
annual rate of aboveground litter production was assumed to be constant during the model
period and calibrated, since long term data is not available. We assumed that the
aboveground to belowground litter ratio equaled the aboveground to belowground biomass
ratio (Hofstede & Rossenaar 1995). Rooting is typically superficial in the páramo with ~80 %
of root biomass located in the first 10 cm of soil (Hofstede & Rossenaar 1995), which was
also observed in the soil profile studied (Tonneijck & Jongmans 2008). In accordance we
assumed root litter input to decrease exponentially with depth x, in line with Elzein &
Balesdent (1995), according to the following equation:

−g ⋅ x Equation 6.7
e
Ioc _ r (x ) = R ⋅ Ioc _ s
g

With Ioc_r(x) = organic carbon input by root litter (gy‐1); R = shoot litter to root litter ratio; Ioc_s
= organic carbon input by shoot litter (gy‐1); and g = a distribution parameter which is chosen
such that 80 % of the total root litter input is contained in the top 10 cm of the soil. As
plants grow on the soil surface, rooting depth should be defined relative to the soil surface.

116
Organic carbon stocks are exceptionally high in the páramo soils of the study area, because
SOM is protected against decomposition by various mechanisms including the formation of
organo‐metallic complexes (Tonneijck et al. 2008b), which is a typical feature of volcanic ash
soils (Shoji et al. 1993; Dahlgren et al. 2004). Analysis of the molecular composition of SOM
in the soil profile studied showed that its composition was dominated by aliphatic
compounds of primary (plant‐derived) origin (Nierop et al. 2007). Because aliphatic
compounds are only present in limited amounts in plants, their relative enrichment in the
soil indicates severe transformation of other primary OM compounds. Therefore, we
assume that decomposition of easily decomposable plant material is so fast that it does not
affect the organic carbon‐depth profile at the time scale of our model and only resistant
plant material (RPM) enters the mineral soil. We assumed that 10 % of the litter consists of
RPM, being a rough estimate based on the molecular composition of SOM and aboveground
litter in the soil profile studied (Nierop et al. 2007).

Decay

RPM is decomposed according to a one compartment model, where the decay rate is
assumed to be first order and therefore directly proportional to the organic carbon mass of
a layer:

∂mOC (x , t) Equation 6.8
= −k(x).mOC (x , t)
∂t

With mOC(x,t) = mass of organic carbon (g) and k(x) = a decay constant (y‐1). We assume that
this decay constant k(x) is proportional to bioturbation (equation 6.10 and 6.11), because
we previously found that microbial decomposition predominantly takes place in the litter
layer rather than in the mineral soil, while soil fauna may further transform SOM in the
mineral soil (Tonneijck et al. 2008b).

Organic carbon transport

Vertical transport of dissolved OM by leaching can be assumed insignificant in volcanic ash
soils, because of the large metal‐to‐SOM ratios that strongly limit its mobility (Aran et al.
2001; Dahlgren et al. 2004). Moreover, we did not find evidence of SOM illuviation (coatings
or radiocarbon age inversions) in the soil studied. Tonneijck & Jongmans (2008)
demonstrated previously that bioturbation is the main process responsible for SOM
transport in the soils studied. Change in mass of organic carbon for a layer due to transport
by bioturbation is modelled with the diffusion equation, similar to Elzein & Balesdent
(1995):

Equation 6.9
∂mOC (x , t) ∂ 2 f (x , t)mb (x , t)
= D(x) OC
∂t ∂x 2

With D(x) = the diffusion coefficient (cm2y‐1) of that layer, see also equation 6.10.

117
Bioturbation patterns
Micromorphological data (Tonneijck & Jongmans 2008) showed that bioturbation was
caused by two groups of soil faunal species producing clearly distinct pedofeatures.
Bioturbation features of the first group were concentrated in the topsoil and decreased in
abundance with depth, while contrary bioturbation features of the second group increased
with depth and were most abundant in the subsoil. Because fresh bioturbation features of
both species groups were present up to the soil surface we concluded that both species
groups are currently active (Tonneijck & Jongmans 2008). To represent this observed
bioturbation pattern in the model we used a function based on two Gaussian‐shaped
kernels to describe the diffusion coefficient D with depth x:
for D(x) ≥ h Equation 6.10
D(x) = DA .e −(x + DB .e −(x − x B )
2 2
).w A .w B

D(x) = 0 for D(x) < h

With DA = the height of the first Gaussian kernel (cm2y‐1); wA = a parameter (cm‐1) influencing
the width of the first kernel; DB = the height of the second Gaussian kernel (cm2y‐1); wB = a
parameter (cm‐1) influencing the width of the second Gaussian kernel, xB = the depth of DB
(cm) and h = a D(x) beyond which the value of the kernels is set to zero. To test our
hypothesis of an upward shifting bioturbation zone in response to soil thickening, equation
6.10 is defined relative to the soil surface. We assume that the páramo vegetation
immediately colonised the last tephra deposit and also that the soil faunal community
immediately colonised this tephra deposit according to this bioturbation pattern.
As explained previously, the decay constant k from equation 6.8 is proportional to
bioturbation as represented by equation 6.10 through:
Equation 6.11
k(x) = qD(x)

With q = a conversion parameter (cm‐2).

Initial settings
The last tephra deposit buried a pre‐existing tephra deposit in which a paleosol was present.
The presence of this paleosol influences the vertical distribution of organic carbon in the
current soil due to bioturbation, and must therefore be taken into account in our model.
Because of the specific vertical pattern of bioturbation and the organic carbon gradient, the
direction of organic carbon transport from the paleosol is upward throughout the modelling
period and consequently no organic carbon enters the paleosol. The estimate of the initial
organic carbon‐depth profile of the paleosol at t0 thus should compensate for loss of organic
carbon due to decay and for soil volume change due to mixing. This was achieved by linearly
extrapolating the currently measured organic carbon‐depth profile of the paleosol up to a
depth x, which was calibrated. The initial thickness of the entire current soil before soil
118
volume change, i.e. Lcurrent(t0) see Figure 6.3, covering this paleosol was calculated using a
geochemical mass balance approach (equation 6.3) and we also calibrated this parameter.
Comparison of the calibrated initial thickness to the calculated initial thickness represents
an independent check of the extent of soil volume change.

Model calibration
To explore the effect of the vertical bioturbation pattern, we first calibrated the model by
manually adjusting parameters related to the first Gaussian kernel (referred to as
‘bioturbation A’) while excluding the second Gaussian kernel (referred to as ‘bioturbation
B’). Next, we added bioturbation B to the optimal scenario for bioturbation A. Because of
soil thickening, the effect of bioturbation B could not be tested without including
bioturbation A in the model. Finally, we adjusted the parameters ‘initial thickness of the
entire current soil’, ‘aboveground organic carbon input’ and ‘decay’ (through conversion
parameter q of equation 6.11) within realistic ranges to get an optimal fit to the observed
total soil thickness (up to the placic horizon) at the end of the modelling period. We used
the root mean squared error between model predicted and observed organic carbon
fraction i.e. foc(x,tend) as the error criterion to be minimised.

Results
Soil volume change
Soil properties are presented in Table 6.1. Immobility of Al and Ti were checked by
evaluating plots of transport of these elements calculated with strain based on Ti or Al
respectively as immobile element index against strain, see Figure 6.2a and b. Immobility is
established when a portion of the transport line falls within the horizontal band of
‘transport’ of the immobile index element as caused by natural variation of that element in
the reference material (Chadwick et al. 1990).

Table 6.1. Soil properties of bulk samples from soil profile G7 and from 3BCb horizons (reference material) of
several soil profiles (including G7) in the same study area.
Horizona Depth Organic C Dry bulk density Soil pHCaCl2 TiO2 Al2O3

[‐] [cm] [mass %] [gcm‐3] [‐] [mass %] [mass %]
Ah1 10 20.1 0.35 3.92 0.33 10.4
Ah1 30 16.3 0.40 4.10 0.37 11.9
Ah2 47 12.7 0.42 4.30 0.43 13.9
1/2Ah2 63 13.5 0.41 4.34 0.49 14.7
1/2Ahb 90 19.6 0.34 4.16 0.57 14.0
2Ahb 110 15.6 0.39 4.24 0.64 15.7
2Bwb 140 5.3 0.71 4.56 0.71 19.9
2Bsb 148 3.8 n.db 4.88 0.57 16.1
3BCb 154 2.7 n.db 4.84 0.67 19.7
3BCb 170 0.38 0.93 5.36 0.70 19.9
3BCb all ‐ 0.81 ± 1.11 (n=15) 0.95 ±0.04 (n=8) 5.11 ± 0.32 (n=15) 0.66 ± 0.10 (n=13) 19.31 ± 0.85 (n=13)
a
= prefix 1/2 refers to the overprinted zone, see description of study area and sites
b
= not determined
119
Measurements show that soil volume increased throughout the current soil profile, as
indicated by positive strain values of up to 4.7 in the topsoils and up to 2.1 in the subsoils.
Strain decreased linearly with depth, although the subsoil samples that were located in the
overprinted zone had a relatively high strain value because of mixing with the SOM‐rich
paleosol. The initial thickness of the entire current soil was estimated at 17 – 26 cm (average
21 cm) with Ti as immobile index element and at 20 – 24 cm (average 22 cm) with Al as
immobile index element. Thus, the current soil (now 77 cm thick, up to the middle of the
overprinted zone) increased 300 – 440 % in thickness relative to its initial thickness.

Figure 6.2 Transport τ (fraction, gains are positive and losses are negative) of a) Ti and b) Al computed with
strain ε based on Al and Ti as immobile element index respectively, against strain based on that immobile
index element, for 4 samples taken from the current soil and top of the overprinted zone. Immobility is
established when a portion of the transport line falls within the horizontal band of ‘transport’ of the immobile
index element as caused by natural variation of that element in the reference material (Chadwick et al. 1990).

Modelling results

The model output at several moments in time and the model fit to the data (at tend = 4800 y)
are shown in Figure 6.3. The parameters used to obtain this fit are presented in Table 6.2.
The model output at t0 shows the initial paleosol settings. Organic carbon content was
relatively high at the top of the paleosol.

Bioturbation A regulated organic carbon distribution in the current soil and the original top
of the paleosol thus shaping the top of the overprinted zone (Figure 6.3, tend), while
bioturbation B shaped the organic carbon distribution at the bottom of the overprinted
zone (Figure 6.3, tend). Bioturbation A had to be slow and had to operate over a relatively
large depth interval (> 60 cm), while bioturbation B had to be relatively fast and had to
cover a small depth interval (12 cm) in order to represent the observations.

120
Figure 6.3 Modelled (—) organic carbon‐depth profiles with bioturbation A and B at several moments in time
(t in y). The current situation is represented by the figure for tend = 4800 y and includes measurements (○) and
a modelled organic carbon‐depth profile without bioturbation B (‐ ‐ ‐). The initial thickness of the entire
current soil, i.e. Lcurrent(t0), is indicated in figure t0 and its total thickness after soil formation, i.e. Lcurrent(tend), is
indicated in figure tend.

Figure 6.4 Modelled organic carbon‐depth profiles with bioturbation A and B and measured (○) organic
2 ‐1 2 ‐1
carbon‐depth profiles as obtained with a) widthA = ~98 cm and DA = 0.03 cm y (— —) or 0.3 cm y (‐ ‐ ‐), b)
2 ‐1
DA = 0.1 cm y and widthA = 180 cm (‐ ‐ ‐) or 60 cm (— —) and c) an initial thickness of the entire current soil
of 30 cm (‐ ‐ ‐) or 10 cm (— —). Other parameter settings in these figures are equivalent to Table 6.2.
121
The effect of DA and wA describing the Gaussian kernel of bioturbation A (equation 6.10) is
shown in figure 6.4a and b respectively. Increasing DA at fixed wA results in homogenisation
of the organic carbon‐depth profile of the current soil. With a relatively low DA, a large wA
(i.e. small width) limits transport to a smaller distance resulting in organic carbon
accumulation in the topsoil rather than the subsoil while contrary a small wA (i.e. large
width) results in a more even distribution of organic carbon throughout the soil. The
sensitivity of the organic carbon‐depth profile to the initial thickness of the entire current
soil is shown in Figure 6.4c.

Table 6.2 Model fit to the data for several parameters as compared to data reported in the literature. These
ranges were used as limits for model calibration.

Parameter Symbol Unit Fit Literature data Reference
Initial thickness of Geochemical mass balance
Lcurrent(t0) [cm] 20 10 – 30
the entire current soil calculation, this paper
Aboveground input of Hofstede & Rossenaar (1995)
Ioc_s [gy‐1] 53 10 – 1000
organic carbon (shoot litter) Tonneijck (1999)
Shoot litter: root litter R [‐] Fixed 1:0.35 Hofstede & Rossenaar (1995)
Personal communication Nierop
Resistant plant material RPM [%] Fixed 10
(2008)
Van Dam et al. (kp 1997)
Turnover timeA (1/k) [y] 10000 4500 – 170000
Torn et al.(1997)
Van Dam et al. (kp 1997)
Turnover timeB (1/k) [y] 5000 4500 – 170000
Torn et al.(1997)
Diffusion coefficient 0.42 – 3.97 Elzein & Balesdent (1995)
DA [cm2y‐1] 0.1
(maximum value), bioturbation A 1 – 15 Van Dam et al. (1997)
Representing the depth of the
Kernel width, bioturbation A widthA [cm] ~98 0 – 200
whole soil profile
Diffusion coefficient
DB [cm2y‐1] 0.2 –
(maximum value), bioturbation B
Kernel width, bioturbation B widthB [cm] 12 0 – 200
Depth of DB x B [cm] 60 60 – 80 Tonneijck & Jongmans 2008

Discussion

Extent of volume change

We conclude that soil volume change is a major soil forming process in the volcanic ash soil
studied. The current soil more than tripled in thickness over time, as based on three
independent methods; namely geochemical mass balance calculations 1) with Al and 2) with
Ti as immobile index element and 3) modelling of organic carbon dynamics. The narrower
range in total initial soil thickness calculated with Al than calculated with Ti is due to the
higher measurement accuracy of the former. Using a thicker total initial current soil in the
122
model resulted in dilution of the organic carbon content in the model output (Figure 6.4c),
which could not be corrected for by increasing organic carbon input (Ioc_s) because the soil
would then become too thick in total due to soil volume increase.

Figure 6.2 demonstrates that Al and Ti were indeed immobile and thus confirms the
reliability of the calculations involving Ti or Al as immobile index element. In all cases a
portion of the transport line falls within the horizontal band representing ‘transport’ due to
natural variation of the immobile index element in the reference material (Figure 6.2a and
b, Chadwick et al. 1990). Although immobile, this does not mean that Al is inert (Chadwick
et al. 1990), because when released during weathering it is stabilised in situ e.g. in Al‐humus
complexes (Tonneijck et al. 2008b).

Soil thickening is clearly the result of SOM incorporation in the mineral soil, mainly by
bioturbation. Barois et al. (1998) also mentioned soil thickening in combination with SOM
accumulation in the case of Mexican Andosols, but did not quantify it. Other common soil
forming processes did not significantly affect soil volume in the soil studied. First,
bioturbation itself did not necessarily lead to an increased porosity (Tonneijck & Jongmans
2008) and therefore could not explain the great extent of soil thickening. Second, soil
collapse due to leaching of elements apparently did not counterbalance soil thickening,
even though weathering of volcanic ash soils under perudic conditions is rapid. This may be
explained by complexation of elements with SOM (e.g. Al, Fe) and by biocycling of elements
(e.g. Si, K, P), preventing leaching. Moreover, element leaching does not necessarily lead to
soil volume change (Chadwick et al. 1990). In view of the vast accumulation of SOM that is
characteristic for volcanic ash soils (Shoji et al. 1993; Dahlgren et al. 2004), we expect that
soil thickening is an important process for this soil type in general and probably also for
other SOM‐rich soils (e.g. Chernozems).

Evaluation of parameters
Although we used realistic ranges to calibrate the parameters of our model (see Table 6.2),
there is no guarantee that there are no other combinations of parameters that would
produce equal or better results. Nonetheless, we conclude that the parameter set used to
obtain our model fit is a plausible one, as discussed in this section.
Diffusion constants were low compared to diffusion constants estimated by Elzein &
Balesdent (1995) and Van Dam et al. (1997). However, when applying 3 times larger
diffusion constants in the model this already resulted in near homogenisation of organic
carbon content with depth (Figure 6.4a). Diffusion rates increase with increasing biological
activity (Elzein & Balesdent 1995). In the field we indeed rarely encountered soil fauna when
excavating soil pits. This observation would indeed support the low diffusion rates in the
model. The abundance of bioturbation features observed using micromorphological
techniques (Tonneijck & Jongmans 2008) does not necessarily contradict low diffusion rates,
since it is the cumulative result of bioturbation over several millennia. Bioturbation rates
123
may indeed be low due to the unfavourable soil environment, i.e. perudic conditions and
acidic soil pH. Because bioturbation B only covers a small vertical distance at every timestep,
its final effect on the organic carbon‐depth profile is minimal (Figure 6.3, tend). Endogeic soil
faunal species, like the ones represented by bioturbation B in the model (Tonneijck &
Jongmans 2008), create extensive horizontal burrowing networks indeed resulting in SOM
transport over short vertical distances only (Anderson 1988; Lee & Foster 1991).
In our model structure, decay was correlated to bioturbation (equation 6.11), because we
found that microbial transformations predominantly take place in the litter layer, while in
the mineral soil matrix it is the soil fauna that further transforms SOM (Tonneijck et al.
2008b). Therefore, turnover times were also low, which is in line with turnover rates
generally reported for the stable pool of organic carbon in volcanic ash soils (Torn et al.
1997; Van Dam et al. 1997). Because the parameters ‘aboveground litter input’ and ‘decay’
are obviously correlated, which is a general problem when modelling organic carbon
dynamics, ‘aboveground litter input’ was at the low end of the range reported in the
literature (Hofstede & Rossenaar 1995; Tonneijck 1999). However, it is possible that the
proportion of RPM was overestimated, which would then automatically lead to higher
estimates for aboveground litter input.

Upward shifting
The good fit between data and model confirms (sensu Oreskes et al. 1994) that our
hypothesis of an upward shifting bioturbation zone in response to soil thickening is indeed
able to clarify the observed vertical distribution of SOM. Although this does not necessarily
imply that our hypothesis is true, at least we show that it is plausible. It is well known that
soil fauna is related to depth, because of preferred environmental conditions such as oxygen
supply, water content and temperature (Lavelle 1988; Lee & Foster 1991), explaining
upward shifting in response to soil thickening. In this hypothesis, organic carbon content is
apparently not a limiting factor for the soil fauna represented by bioturbation B, since
upward shifting implies movement towards lower organic carbon contents. Alternatively,
correlating the bioturbation pattern to organic carbon content evidently results in
homogenised organic carbon‐depth profiles.
It is also plausible that roots have shifted upwards in response to soil thickening. This may
explain the apparent paradox of the abundance of root‐derived organic molecules (e.g.
suberin) throughout the current soil profile (Nierop et al. 2007), while rooting was clearly
superficial according to field observations and micromorphological observations (Tonneijck
& Jongmans 2008).
Because soil volume change may influence depth‐dependent soil forming processes, as
clearly illustrated by our modelling exercise, it should be taken into account when modelling
soil organic carbon dynamics, especially in the case of SOM‐rich soils. This pertains to
volcanic ash soils, but probably also to Chernozems and Phaeozems which are also
characterised by bioturbation and carbon accumulation.
124
Implications

The studied soil profile contains paleoecological proxies (e.g. pollen and plant species
specific organic molecules or ‘biomarkers’) that are used to reconstruct the vegetation
history in the study area (Jansen et al. 2008a; Moscol Olivera & Hooghiemstra 2008). Since
the use of paleoecological proxies contained in soils depends on their preservation in
chronostratigraphic order, the degree and type of bioturbation is of key importance. Based
on the model structure and parameter set presented in this investigation, we conclude that
bioturbation may have been slow. In combination with upward shifting bioturbation, this
may explain the linear increase of age with depth observed in soil profiles of the study area,
including the soil profile investigated in this chapter (Tonneijck & Jongmans 2008).
Therefore, in line with our previous investigations (Tonneijck et al. 2008a; Tonneijck &
Jongmans 2008), the results from the present study suggest that as long as paleoecological
proxies were transported through the soil profile similarly to bulk SOM, they will probably
be distributed in a (crude) chronostratigraphic order. However, a major consequence is also
that any sample taken inevitably would produce a mixed signal on a high resolution
temporal scale.

Recommendations

The modelling exercise presented in this chapter served its purpose of testing our
hypothesis. Nevertheless, the model of organic carbon dynamics may be improved by e.g.
accounting for labile organic carbon, by further adjusting the initial paleosol settings and
possibly by using more parsimonious and asymmetrically shaped functions to even better
describe the soil faunal distribution and mixing. Model calibration should also be done in an
automated procedure to gain insight in the model structure, the parameter space,
correlation between parameters and sensitivity of parameters. Extrapolation of short term
field data/experiments to long term modelling periods is uncertain. Nonetheless,
measurement of several parameters such as leaf and root litter production and SOM decay
rates at various soil depths in this and other Andean páramo ecosystems would enhance the
parameter ranges used in model calibration. A study of the vertical migrations of soil fauna
currently present in the soil studied would further underpin the conclusions of this research.
In addition to modelling organic carbon‐depth profiles, modelling of radiocarbon depth‐
profiles may be useful to further constrain the model. Our model should be applied to other
areas and other (SOM‐rich) soil types as well to contribute to a greater understanding
organic carbon dynamics. Furthermore, when applied backwards, our model may be used in
combination with the VERHIB model (Jansen et al. 2008b) that was made to reconstruct the
past vegetation by unravelling the ‘bulk’ biomarker signal contained in the soil using
biomarker signals of individual plant species.

125
Conclusions

Our results show that soil volume change is a major soil forming process in the volcanic ash
soil studied. SOM accumulation resulted in more than a tripling of initial soil thickness
within 4800 years, from ~20 to 77 cm. Furthermore, our modelling exercise confirmed that
our hypothesis of an upward shifting bioturbation zone in response to soil thickening is
indeed plausible. Because of its effect on depth‐ dependent processes, e.g. decomposition,
bioturbation and rooting, soil volume change cannot be ignored when investigating soil
organic carbon dynamics at the millennia time scale, particularly in the case of SOM‐rich
soils that are of major importance to the global carbon cycle.

Acknowledgements

assistance. Furthermore, we acknowledge the fellow members of the RUFLE program: Boris
Jansen, Marcela Moscol Olivera, Klaas Nierop, Henry Hooghiemstra and Antoine Cleef for
their valuable contributions to this research. Finally, we wish to thank Leo Hoitinga and Ton
Van Wijk for their assistance in the chemical laboratory. This research was funded by
WOTRO (WAN 75‐405) and the University of Amsterdam and generously sponsored by
Fjällraven in the form of clothing and gear.

126

127
Synthesis

The main objective of this PhD research has been to gain insight into the processes affecting
the vertical distribution and stabilisation of SOM in volcanic ash soils in tropical montane
forest and páramo ecosystems in the northern Ecuadorian Andes. The purpose of this final
chapter is to present the main conclusions of this research and place them in a broader
framework, to discuss its scientific and practical implications and to identify topics that
warrant further investigation.

Main conclusions

Vertical distribution of SOM: the tephra stratigraphic framework

Tephra deposition clearly formed the framework for SOM accumulation in the soils studied
(Chapter 3). Three superimposed tephra deposits of Holocene age were identified at every
study site. No evidence for a finer lithological stratification within these tephra deposits was
found. Related to this tephra stratigraphy all soils contained a multisequum consisting of a
SOM‐rich current soil, a SOM‐rich paleosol and a SOM‐poor second heavily truncated or
immature paleosol, each at least 40 cm thick. Between the current soil and the first paleosol
an overprinted zone with gradually increasing SOM contents and transitional grain size
distribution and chemical element ratios was present, indicating that bioturbation most
likely played an important role in the youngest pedogenetic phase. Unravelling the tephra
stratigraphy clearly improved understanding of the vertical distribution of SOM and vice
versa and directed the research towards the study of pedogenetic processes and
bioturbation in particular.

In general, overprinting makes accurate timing of the event of tephra deposition by
radiocarbon dating impossible. Without overprinting, the exact depth of the top of a
paleosol can be identified by an abrupt increase in organic carbon content and the position
for dating events of tephra deposition would then be just above and just below the
maximum organic carbon content. However, with overprinting, the position of the
maximum organic carbon content depends on e.g. the depth of a bioturbation zone
extending into the preceding tephra deposit. In the latter case, the maximum organic
carbon content would represent an overestimation of the timing of last tephra deposition.
Optically Stimulated Luminescence (OSL) dating does not appear to offer a solution in the
short term, because of instability of the quartz luminescence signal, causing gross
underestimation of timing of tephra deposition. This is currently under further investigation
(personal communication J. Wallinga 2008).
129
In general, we conclude that failure to recognize overprinted zones may seriously hamper
determination of the tephra chronostratigraphy and understanding of subsequent soil
forming processes. Overprinting is common in soils developed in distal facies with complex
tephra chronology, and particularly in such soils it is not straightforward to unravel the
tephra stratigraphy. Still, we conclude that this is possible provided that techniques
traditionally used to investigate tephra stratigraphy (mineral assemblages, chemical
element ratios, grain size distributions) are combined with organic carbon contents at high
vertical resolution (i.e. sampling intervals of 1 cm).

Vertical distribution of SOM: complex bioturbation patterns

In the study area, the soil faunal community in the forest is dominated by epigeic species,
whereas endogeic species prevail in the páramo (Chapter 4). Epigeic species typically inhabit
the ectorganic layers and do not actively redistribute material, while endogeic species ingest
both mineral soil particles and SOM and create extensive networks of subhorizontal
burrows in the mineral soil transporting SOM over short vertical distances (Anderson 1988;
Lavelle 1988; Lee & Foster 1991). One of the main findings of the RUFLE project was that all
currently forested study sites were previously under páramo vegetation at some point
during the Holocene as a result of UFL dynamics, while the páramo sites were never under
forest throughout the Holocene (Bakker et al. 2008; De Boer 2008; Jansen et al. 2008a). A
change from páramo to forest vegetation results in a change from endogeic to epigeic
species. Because the latter species do not redistribute material vertically, SOM in the forest
starts to accumulate predominantly on top of the mineral soil profile in ectorganic layers
rather than in the mineral soil. Thus, in the area studied, the vertical distribution of SOM in
mineral soils under forest vegetation is to some extent a relic from a period with páramo
vegetation, although fine tree roots and some illuviation of SOM leached from the
ectorganic layers may currently influence the mineral topsoil.

In the páramo, bioturbation was largely responsible for the vertical transport of SOM within
and between the two upper tephra deposits. Illuviation of SOM did not play a significant
role and root input was mainly superficial. The influence of bioturbation on the vertical
distribution of SOM was complex because of interaction between different endogeic species
operating preferentially in either the topsoil or the subsoil. In the current soil, a decrease in
bioturbation by small endogeic species with depth combined with dominant OM supply
from the soil surface and topsoil resulted in a decrease in SOM content and a linear increase
of age with depth. Bioturbation by larger endogeic species was massive throughout the
overprinted zone, which seemed to conflict with the gradual increase in SOM content and
radiocarbon age with depth in that zone. In view of the magnitude of bioturbation in the
overprinted zone, one would have expected bulk organic carbon and radiocarbon in this
entire zone to have been homogenised. This may be explained by predominant mixing over
short vertical distances, because mixing of material with small differences in carbon and
radiocarbon content over short distances does not cause complete homogenisation even if
130
the bioturbation zone covers larger distances. An additional explanation is an upward shift
of the soil fauna, rendering deeper positions in the soil out of reach of bioturbation over
time. It was hypothesised that such an upward shift could have occurred in response to soil
thickening due to SOM accumulation. This hypothesis was tested in a modelling exercise
(Chapter 6). The good fit between data and model confirmed (sensu Oreskes et al. 1991)
that our hypothesis of an upward shifting bioturbation zone in response to soil thickening is
indeed able to clarify the observed vertical distribution of SOM.

Because of the high degree of endemism of soil fauna in the tropical montane Andean
region (Feijoo et al. 2004), the results of this investigation cannot simply be extrapolated to
other areas. Nevertheless, complex bioturbation patterns with depth seem to be the rule
rather than an exception also in other volcanic ash soils (Barois et al. 1998, Van der
Hammen & Beglinger 1989) and other soil types (Lavelle 1988; Lee & Foster 1991; Pitkanen
& Nuutinen 1997; Jimenez & Decaëns 2000; Shuster et al. 2001).

Vertical distribution of SOM: soil thickening

Soil volume change is a major soil forming process in the volcanic ash soils studied (Chapter
6). In the literature on volcanic ash soils this has not yet been properly acknowledged. SOM
accumulation resulted in more than a tripling of initial soil thickness within 4800 years, from
~20 to 77 cm. Although weathering of volcanic ash soils under perudic conditions is known
to be rapid, soil collapse due to leaching of elements did not at all counterbalance soil
thickening in our case, which may be explained by complexation of elements with SOM and
by biocycling.

Soil volume change clearly affects depth‐related processes, e.g. decomposition, bioturbation
and rooting, as demonstrated in a modelling exercise (Chapter 6). Therefore, soil volume
change cannot be ignored when investigating soil organic carbon dynamics, particularly in
the case of SOM‐rich soils that are of major importance for the global carbon cycle. This
clearly pertains to volcanic ash soils (Andosols), but probably also to Chernozems and
Phaeozems which are also characterised by bioturbation and carbon accumulation.

Soil organic carbon stabilisation: a combination of mechanisms

In the non‐allophanic volcanic ash soils studied, the organic carbon stabilisation mechanisms
involved were: 1) direct stabilisation of OM in organo‐metallic (Al‐humus) complexes; 2)
indirect protection of OM (notably aliphatic compounds) through low soil pH and toxic
levels of Al; and 3) physical protection of OM as a result of a high porosity and great
proportion of micropores (Chapter 5). Moreover, in the case of the forest soils, inherent
recalcitrance of OM itself contributed to the substantial accumulation of OM in ectorganic
horizons. By enhancing soil acidification, both páramo and forest vegetation contributed to
the formation of non‐allophanic characteristics, thus increasing SOM accumulation in the
131
soils and inducing positive feedbacks. In the forest and páramo soils of the study area,
decomposition of SOM is performed predominantly by fungi rather than bacteria. Microbial
SOM transformations predominantly take place in the ectorganic horizons, while in the
mineral soil matrix it is the soil fauna that further transforms SOM.
In volcanic ash soils throughout the world Al‐humus complexes are found to be of overriding
importance in determining organic carbon stabilisation, rather than amorphous organo‐
mineral complexes e.g. with allophane (Percival et al. 2000; Podwojewski et al. 2002;
Powers & Schlesinger 2002; Poulenard et al. 2003; Matus et al. 2006; Egli et al.; 2007; Dümig
et al. 2008). In most of these soils organic carbon content is found to be also related to
acidic pH and humid conditions and when measured with high (toxic) levels of exchangeable
Al, like in the soils studied. Yet, compelling evidence exists that high contents of amorphous
constituents like allophane result in organic carbon stabilisation in other volcanic ash soils
(Torn et al. 1997; Powers & Schlesinger 2002; De Koning et al. 2003; Basile‐Doelsch et al.
2005; Huygens et al. 2005).
Powers & Schlesinger (2002) reported a non‐linear relationship between Alp/Alo ratios and
organic carbon stock in Costa Rican volcanic ash soils, which may be the key to this apparent
paradox. At Alp/Alo < 0.5, i.e. in allophanic volcanic ash soils, organic carbon stocks increased
with decreasing Alp/Alo implying an increase with amorphous constituents like allophane.
Contrarily, at Alp/Alo > 0.5, i.e. in non‐allophanic volcanic ash soils, organic carbon stock
increased with increasing Alp/Alo implying an increase with Al‐humus complexes. The
difference between allophanic and non‐allophanic volcanic ash soils also seems to hold at
molecular level (Nierop et al. 2005; Buurman et al. 2007; Nierop et al. 2007). SOM in
allophanic volcanic ash soils is dominated by organic molecules that are normally easily
decomposed and that are to a large extent of microbial origin (polysaccharides, including
chitin and proteins), while SOM in non‐allophanic volcanic ash soils is predominantly
aliphatic (lipids and aliphatic macromolecules) and probably plant‐derived. Both types do
have in common that the microporosity is very high which contributes to organic carbon
stabilisation, in the case of allophanic soils because of the microstructure of allophane (Shoji
et al. 1993) and in the case of non‐allophanic soils mainly because of the high SOM contents
(Chapter 5; Poulenard et al. 2003; Buytaert et al. 2005).
Although this ‘binary composition’ of volcanic ash soils is well known (Mizota & Van
Reeuwijk 1989; Shoji et al. 1993; Dahlgren et al. 2004), few researchers systematically apply
this concept and e.g. do not specifically report the type of volcanic ash soil investigated or
even Alp/Alo ratios. Moreover, there appears to be confusion regarding the terminology. For
example, the ‘allophanic’ soils of Rodríguez‐Rodríguez et al. (2006) still had an average
Alp/Alo > 0.5. Many non‐allophanic soils have a high Alp/Alo in the topsoil combined with a
lower Alp/Alo in the subsoil due to lower SOM supply and higher pH there. Thus, only
topsoils should be considered when designating the type of volcanic ash soil. To truly
increase understanding of the mechanisms of organic carbon stabilisation in volcanic ash
soils, we argue that research should clearly distinguish between the two types.
132
Organic carbon stocks: forest and páramo soils as carbon sinks

Organic carbon stocks are very large in the natural forest and páramo soils studied (Chapter
5), amounting to 22 ± 5 kgCm‐2 up to 30 cm, 53 ± 10 kgCm‐2 up to 100 cm and 87 ± 12 kgCm‐2
up to 200 cm, thus roughly double global averages reported for volcanic ash soils (13 ± 8
kgCm‐2 up to 30 cm, 29 ± 18 kgCm‐2 up to 100 cm and 30 ± 13 kgCm‐2 up to 200 cm). As
explained previously in this synthesis, the mineral soil profile under forest vegetation may to
some extent be a relic from a period with páramo vegetation, because of natural UFL
fluctuations which is a typical process in these tropical Andean ecosystems. However, when
comparing mineral forest topsoils that were indeed only influenced by forest vegetation, to
mineral páramo topsoils organic carbon stocks were still similar. Podwojewski et al. (2002)
and Poulenard et al. (2003) already reported very high organic carbon stocks in other non‐
allophanic páramo soils throughout Ecuador and we now demonstrated that similar organic
carbon stocks may accumulate in mineral soils under forest vegetation. In the forest, an
additional 31 ± 23 kgCm‐2 was stored in thick ectorganic horizons. The latter organic carbon
stocks are particularly sensitive to land‐use change, e.g. deforestation or reforestation.

SOM in ectorganic layers (forest) is at least stabilised at the century time scale as indicated
by radiocarbon dating, while SOM in the mineral soil (forest and páramo) is stabilised for
several millennia. Accumulation of SOM in ectorganic horizons as a result of its inherent
recalcitrance will eventually approach a steady state where input and output are balanced.
Stabilisation of SOM in organo‐metallic complexes is probably already at a maximum in our
soils, but SOM may still continue to accumulate as long as the pH remains strongly acidic
and the porosity remains high and dominated by small pore sizes.

Implications

Implications for global carbon cycling

The tropical Andean ecosystems should urgently be protected from destructive land‐use
change, not only for the sake of preserving their outstanding biodiversity but also for their
function as exceptionally large, long term organic carbon sink. In view of the increased
population and corresponding increased pressure on land, this presents a real challenge.
Most types of land‐use change involving natural forest and páramo ecosystems are
expected to cause a net flow of carbon from the soil to the atmosphere for a long period of
time. Deforestation and subsequent conversion to agricultural land would at least cause the
immediate and substantial loss of carbon from the ectorganic horizons. Liming to alleviate Al
toxicity would evidently promote the loss of carbon from the soil to the atmosphere by
reducing soil acidification, which would not only increase decomposition of plant‐derived
compounds but also of OM in organo‐metallic complexes as demonstrated by Takahashi et
al. (2006). Exposing bare soil, tilling and draining volcanic ash soils for agricultural use would
alter physical soil properties because of irreversible changes upon drying, eventually leading
133
to soil erosion and transforming the soils from sinks to sources of carbon (Shoji et al. 1993;
Poulenard et al. 2001; Dahlgren et al. 2004; Buytaert et al. 2006). Burning and grazing of the
páramo will reduce OM input into the soil, will increase soil erosion and will prevent the UFL
from moving upwards (Laegaard 1992; Hofstede 1995a; Poulenard et al. 2001; Podwojewksi
et al. 2002; Buytaert et al. 2005).

Implications for modelling of organic carbon dynamics

Many models of global soil organic carbon dynamics incorporate clay content as a
controlling factor determining SOM stabilisation (Smith et al. 1997). This assumption is not
adequate in the case of many volcanic ash soils (this research, Percival et al. 2000; Powers &
Schlesinger 2002; De Koning et al. 2003; Matus et al. 2006) and may thus lead to serious
underestimation of organic carbon stocks, which seems ironical in view of the huge organic
carbon accumulation in precisely this soil type (Chapter 5). Alternatively, Alp/Alo ratios may
be of more value for predictions of organic carbon contents for both allophanic and non‐
allophanic volcanic ash soils.

Most models of global soil organic carbon dynamics focus on the topsoil (< 25 cm depth)
and lack spatial dimensions (Elzein & Balesdent 1995; Smith et al. 1997; Jenkinson et al.
2008). Because of the vast amount of organic carbon transported to and stored in the
subsoil, we greatly encourage recent initiatives to include vertical transport in such models
(Jenkinson et al. 2008). However, in the case of organic carbon rich soils ‐ volcanic ash soils,
but possibly also Chernozems and Phaeozems ‐ we argue that organic carbon models should
incorporate soil volume change due to SOM accumulation and its effect on depth‐
dependent processes such as bioturbation, decomposition and rooting (Chapter 6).
Furthermore, these models should also account for the complexity of bioturbation patterns
and consequently for the decomposition patterns in the soil rather than assuming uniform
or decreasing bioturbation and decomposition with depth.

Implications for the interpretation of paleorecords

The use of paleoecological proxies (pollen and biomarkers) contained in terrestrial soils for
the reconstruction of UFL dynamics depends amongst others on their preservation in
chronostratigraphic order. Therefore, in the case of volcanic ash soils, paleoecological
analysis should be accompanied by a thorough investigation of both tephra stratigraphy and
bioturbation. Multiple deposition of thin tephra layers clearly offers a different framework
for SOM accumulation than does incidental deposition of thick tephra layers, while
bioturbation may disorder the chronostratigraphy through vertical mixing.

Because the soils studied were formed in few relatively thick tephra deposits rather than
several thinner ones, tephra sedimentation rate could be excluded as a driver of
chronostratigraphic ordering. With regard to bioturbation, vertical mixing of SOM is halted
134
as soon as forest vegetation colonises a site (Chapter 4), so that the chronostratigraphic
order is intact from that moment onward. In contrast, vertical mixing of SOM takes place
throughout the soil profile at sites with páramo vegetation. Nevertheless, due to short
vertical distance mixing over intervals < 5 cm (i.e. ~500 years) in combination with an
upward shifting zone of bioturbation, a (crude) chronostratigraphic order of SOM is still
maintained also in páramo soils, as evidenced by an increase in radiocarbon age with depth.
However, a major consequence with respect to paleoecological investigations is also that
any sample taken in these soils inevitably produces a mixed signal at a high resolution
temporal scale. When applied backwards, the model presented in Chapter 6 may be used in
combination with the VERHIB model (Jansen et al. 2008b) that was made to reconstruct the
past vegetation by unravelling the ‘bulk’ biomarker signal contained in the soil from
biomarker signals of individual plant species.

The lack of seasonality in precipitation in the study area probably limited the distance of
vertical movement of the soil fauna at short time scales. In regions with a more pronounced
seasonality, vertical movement of the soil fauna in response to soil moisture oscillations
(Lavelle 1988; Jimenez & Decaëns 2000) may result in mixing of organic carbon and
radiocarbon contents over larger vertical distances than the 5 cm we estimated.

As long as paleoecological proxies are distributed through the soil profile by similar
processes as SOM, they will also be in (crude) chronostratigraphic order. Many researchers
(Havinga 1963; Walch et al. 1970; Keatinge 1983; Davidson et al. 1999) demonstrated that
bioturbation is a key process affecting incorporation and redistribution of pollen in most
soils, which is also likely to be the case for biomarkers. Other processes that theoretically
may influence the distribution of paleoecological proxies include erosion/deposition,
leaching and subsequent illuviation and root input in the case of biomarkers. However,
these other processes were found to be of minor importance at the study sites. Sites with
erosional/depositional features could be identified by anomalous grain size and
geochemical data (Chapter 3). While some researchers suggest that pollen may leach
through the soil (Dimbleby 1957; Walch et al. 1970; Kelso 1994) others dispute this
(Keatinge 1983; Davidson et al. 1999; Van Mourik 1999; Tyler et al. 2001). Moreover, it was
not observed in Colombian volcanic ash soils (Bakker & Salomons 1989) and the great
proportion of micropores is likely to prevent substantial leaching of pollen in the soils
studied. With regard to the biomarkers (notably n‐alkanes and n‐alcohols), these were not
affected by leaching, having been specifically selected for their limited leaching potential
(Jansen et al. 2006; Jansen et al. 2007). Finally, the biomarker record was not disturbed by
roots, because root biomarkers had a sufficiently different composition from leaf
biomarkers to allow for their distinction in the studied soils (Jansen et al. 2006; Jansen et al.
2007). Havinga (1971) suggested that selective degradation of pollen from some plant
species relative to others plays a role, but this was not observed in Colombian volcanic ash
soils (Bakker & Salomons 1989) nor were biomarkers selectively degraded (Jansen et al.
2008b).
135
Humic acids (HA) proved to be the most reliable fraction of SOM for radiocarbon dating of
volcanic ash soils (Chapter 2), because this fraction is not rejuvenated by fresh roots or by
illuviation of younger SOM (except just underneath a thick forest floor). Using humic acids
for radiocarbon dating offers the possibility of dating samples derived from small vertical
sample intervals. Contrary, obtaining charcoal fragments for dating often requires large
sample volumes and consequently large vertical sample intervals. Thus, for linking pollen
and biomarkers to a high resolution time scale, dating of HA may in many cases be more
suitable than dating of charcoal fragments.

Further research

The research presented in this thesis has greatly increased knowledge on the processes
affecting the vertical distribution and stabilisation of SOM in volcanic ash soils in natural
forest and páramo ecosystems and its scientific and practical implications were discussed.
Nonetheless, new questions arose, leading to the identification of topics warranting further
research, as for example the identification of soil faunal species currently present in the soil
and a study of their vertical migrations at short time scales to further underpin the
conclusions of this research. Another topic that merits further attention is the vertical
distribution and stabilisation of SOM under different types of land use to enable
quantification and modelling of the impact of land‐use change on soil organic carbon
dynamics. Finally, the effect of soil volume change on depth‐dependent processes should be
studied in a range of volcanic ash soils and other SOM‐rich soils.
136
References

Adriano DC. 2001. Trace elements in terrestrial environments. Biogeochemistry,
bioavailability and risks of metals. Springer‐Verlag, New York, Berlin, Heidelberg:
867 pp.
Aerts‐Bijma AT, van der Plicht J & Meijer HAJ. 2001. Automatic AMS sample combustion and
CO2 collection. Radiocarbon 43(2): 293‐298.
Anderson JM. 1988. Invertebrate‐mediated transport processes in soils. Agriculture,
Ecosystems & Environment 24(1‐3): 5‐19.
Aran D, Gury M & Jeanroy E. 2001. Organo‐metallic complexes in an Andosol: a comparative
study with a Cambisol and Podzol. Geoderma 99(1‐2): 65‐79.
Bader MY, Rietkerk M & Bregt A. 2007. Vegetation Structure and Temperature Regimes of
Tropical Alpine Treelines. Arctic, Antarctic, and Alpine Research 39(3): 353‐364.
Bakker J, Moscol Olivera M & Hooghiemstra H. 2008. Holocene environmental change at the
upper forest line in northern Ecuador. The Holocene 18(6): 877 ‐ 893.
Bakker JGM & Salomons JB. 1989. A palaeoecological record of a volcanic soil sequence in
the Nevado del Ruiz area, Colombia. Review of Palaeobotany and Palynology
60(1‐2): 149‐163.
Baldock JA & Skjemstad JO. 2000. Role of the soil matrix and minerals in protecting natural
organic materials against biological attack. Organic Geochemistry 31(7‐8): 697‐710.
Barois I, Dubroeucq D, Rojas P & Lavelle P. 1998. Andosol‐forming process linked with soil
fauna under the perennial grass Mulhembergia macroura. Geoderma 86(3‐4): 241‐
260.
Barragan R, Geist D, Hall M, Larson P & Kurz M. 1998. Subduction controls on the
compositions of lavas from the Ecuadorian Andes. Earth and Planetary Science
Letters 154(1‐4): 153‐166.
Basile‐Doelsch I, Amundson R, Stone WEE, Borschneck D, Bottero JY, Moustier S, Masin F &
Colin F. 2005. Mineralogical control of organic carbon dynamics in a volcanic ash
soil on La Réunion. European Journal of Soil Science 56(6): 689‐703.
Batjes NH. 1996. Total carbon and nitrogen in the soils of the world. European Journal of Soil
Science 47(2): 151‐163.
Begheijn LT. 1979. Determination of iron(II) in rock, soil and clay. Analyst 104: 1055‐1061.
Birks HJB & Birks HH. 1980. Quaternary Palaeoecology. Edward Arnold, London: 289 pp.
Blanco‐Canqui H, Lal R, Post WM, Izaurralde RC & Shipitalo MJ. 2006. Organic Carbon
Influences on Soil Particle Density and Rheological Properties. Soil Science Society
of America Journal 70(4): 1407‐1414.
137
Blott SJ & Pye K. 2001. Gradistat: A grain size distribution and statistics package for the
analysis of unconsolidated sediments. Earth Surface Processes and Landforms
26(11): 1237‐1248.
Bradley RS. 1999. Paleoclimatology: Reconstructing Climates of the Quaternary. 2nd ed.
Academic Press, New York: 610 p.
Brimhall GH & Dietrich WE. 1987. Constitutive mass balance relations between chemical
composition, volume, density, porosity, and strain in metasomatic hydrochemical
systems: Results on weathering and pedogenesis. Geochimica et Cosmochimica
Acta 51(3): 567‐587.
Brown G, Edwards B, Ormerod EC & Weir AH. 1972. A simple diffractometer heating stage.
Clay Minerals 9: 407‐413.
Bruun S, Christensen BT, Thomsen IK, Jensen ES & Jensen LS. 2007. Modeling vertical
movement of organic matter in a soil incubated for 41 years with 14C labeled straw.
Soil Biology and Biochemistry 39(1): 368‐371.
Bumpus JA. 1993. White rot fungi and their potential use in soil bioremediation. In: Soil
Biochemistry, editors Bollag JM & Stotzky G, Marcel Dekker, New York: 65‐100.
Buurman P, Garcia Rodeja E, Martinez Cortizas A & Van Doesburg JDJ. 2004. Stratification of
parent material in European volcanic and related soils studied by laser‐diffraction
grain‐sizing and chemical analysis. Catena 56(1‐3): 127‐144.
Buurman P, Peterse F & Almendros Martin G. 2007. Soil organic matter chemistry in
allophanic soils: a pyrolysis‐GC/MS study of a Costa Rican Andosol catena.
European Journal of Soil Science 58(6): 1330‐1347.
Buytaert W, Célleri R, De Bièvre B, Cisneros F, Wyseure G, Deckers J & Hofstede R. 2006.
Human impact on the hydrology of the Andean páramos. Earth‐Science Reviews
79(1‐2): 53‐72.
Buytaert W, Wyseure G, De Bièvre B & Deckers J. 2005. The effect of land‐use changes on
the hydrological behaviour of Histic Andosols in south Ecuador. Hydrological
Processes 19(20): 3985‐3997.
Chadwick OA, Brimhall GH & Hendricks DM. 1990. From a black to a gray box ‐ a mass
balance interpretation of pedogenesis. Geomorphology 3(3‐4): 369‐390.
Childs CW, Matsue N & Yoshinaga N. 1991. Ferrihydrite in volcanic ash soils of Japan. Soil
Science and Plant Nutrition 37(2): 299‐311.
Dahlgren RA, Saigusa M & Ugolini F. 2004. The nature, properties and management of
volcanic soils. Advances in Agronomy 82: 113‐182.
Dahlgren RA, Ugolini FC, Shoji S, Ito T & Sletten RS. 1991. Soil‐Forming Processes in Alic
Melanudands under Japanese Pampas Grass and Oak. Soil Science Society of
America Journal 55(4): 1049‐1056.
Dalsgaard K & Odgaard BV. 2001. Dating sequences of buried horizons of podzols developed
in wind‐blown sand at Ulfborg, Western Jutland. Quaternary International 78(1): 53‐60.
Davidson DA, Carter S, Boag B, Long D, Tipping R & Tyler A. 1999. Analysis of pollen in soils:
processes of incorporation and redistribution of pollen in five soil profile types. Soil
Biology and Biochemistry 31(5): 643‐653.
138
De Boer E. 2008. Reconstruction of the natural upper forest line of northern Ecuador: A
comparative study of pollen analysis and biomarker analysis. MSc thesis,
Universiteit van Amsterdam, Amsterdam: 56 pp.
De Koning GHJ, Veldkamp E & López‐Ulloa M. 2003. Quantification of carbon sequestration
in soils following pasture to forest conversion in northwestern Ecuador. Global
Biogeochemical Cycles 17(4): 1098‐1107.
De Vos B, Van Meirvenne M, Quataert P, Deckers J & Muys B. 2005. Predictive Quality of
Pedotransfer Functions for Estimating Bulk Density of Forest Soils. Soil Science
Society of America Journal 69(2): 500‐510.
Deer WA, Howie RA & Zussman J. 1993. An introduction to the rock‐forming minerals.
Longman, Harlow: 696 pp.
Di Pasquale G, Marziano M, Impagliazzo S, Lubritto C, de Natale A & Bader MY. 2008. The
Holocene treeline in the Northern Andes (Ecuador): first evidence from soil
charcoal. Palaeogeography, Palaeoclimatology, Palaeoecology 259(1): 17‐34.
Dimbleby GW. 1957. Pollen analysis of terrestrial soils. New Phytologist 56(1): 12‐28.
Droux A & Delaloye M. 1996. Petrography and geochemistry of Plio‐Quaternary calc‐alkaline
volcanoes of Southwestern Colombia. Journal of South American Earth Sciences
9(1‐2): 27‐41.
Dumanski J. 2004. Carbon Sequestration, Soil Conservation, and the Kyoto Protocol:
Summary of Implications. Climatic Change 65(3): 255‐261.
Dümig A, Schad P, Kohok M, Beyerlein P, Schwimmer W & Kögel‐Knabner I. 2008. A mosaic
of nonallophanic Andosols, Umbrisols and Cambisols on rhyodacite in the southern
Brazilian highlands. Geoderma 145(1‐2): 158‐173.
Egli M, Alioth L, Mirabella A, Raimondi S, Nater M & Verel R. 2007. Effect of climate and
vegetation on soil organic carbon, humus fractions, allophanes, imogolite,
kaolinite, and oxyhydroxides in volcanic soils of Etna (Sicily). Soil Science 172(9):
673‐691.
Ellenberg H. 1979. Man's influence on tropical mountain ecosystems in South America.
Journal of Ecology 67(2): 401‐416.
Elzein A & Balesdent J. 1995. Mechanistic simulation of vertical distribution of carbon
concentrations and residence times in soils. Soil Science Society of America Journal
59: 1328‐1335.
Erdtman G. 1943. An introduction to pollen analysis. The Chronica Botanica Company,
Waltham Mass: 239 pp.
Eswaran H, Van Den Berg E & Reich P. 1993. Organic Carbon in Soils of the World. Soil
Science Society of America Journal 57(1): 192‐194.
FAO (Food and Agriculture Organisation of the United Nations). 1990. Guidelines for soil
description. FAO/ISRIC, Rome: 70 pp.
FAO (Food and Agriculture Organisation of the United Nations). 2006. Global Forest
Resources Assessment 2005. Progress towards sustainable forest management.
FAO, Rome. FAO Forestry Paper 147: 320 pp.
139
FAO (Food and Agriculture Organisation of the United Nations). 2006. World reference base
for soil resources 2006. A framework for international classification, correlation
and communication. FAO/ISRIC/IUSS, Rome. World Soil Resources Reports 103: 128
pp.
Farley KA. 2007. Grasslands to Tree Plantations: Forest Transition in the Andes of Ecuador.
Annals of the Association of American Geographers 97(4): 755‐771.
Feijoo MA, Quintero H & Fragoso CE. 2004. Patrón de distribución y listado de especies de
las lombrices de tierra (Annelida, Oligochaeta) en Colombia. Acta Zoologica
Mexicana 20(2): 197‐220.
Fieldes M & Perrott KW. 1966. The nature of allophane in soils. Part 3. New Zealand Journal
of Soil Science 9: 623‐629.
Gee GW & Or D. 2002. Particle‐size analysis. In Methods of soil analysis, Part 4: Physical
Methods, editors Dane JH & Topp GC. Soil Science Society of America Inc.,
Madison, Wisconsin: 255‐293.
Gonzalez‐Perez JA, Gonzalez‐Vila FJ, Almendros G & Knicker H. 2004. The effect of fire on
soil organic matter‐a review. Environment International 30(6): 855‐870.
Graphpad Software Inc. 1999. Graphpad Prism Users Guide Version 3. The fast, organized
way to analyze and graph scientific data. San Diego: 108 pp.
Green RN, Trowbridge RL & Klinka K. 1993. Towards a taxonomic classification of humus
forms. In: Forest science monograph 29, editors Hyde WF, Winget KW & Smith B.
Society of American Foresters, Washington D.C.: 1‐50.
Hall M & Mothes P. 1994. Tefroestratigrafia holocenica de los volcanes principales del valle
interandino, Ecuador. Estudios de Geografia 6.
Havinga AJ. 1963. A palynological investigation of soil profiles in cover sand. Mededelingen
Landbouwhogeschool Wageningen 63: 11‐93.
Havinga AJ. 1971 An experimental investigation into the decay of pollen and spores in
various soil types. In: Sporopollenin, editors Brooks J, Grant PR, Muir M, Van Gijzel
P & Shaw G. Academic Press, London: 446–478.
Hedberg O. 1964. Features of Afro‐Alpine plant ecology. Acta Phytogeographica Suecica 49:
1‐144.
Hetier, JM, Guillet B, Brousse, R, Delibrais, G & Maury RC. 1983. 14C dating of buried soils in
the volcanic Chaine des Puys (France). Bulletin Volcanologique 46(2): 193‐201.
Heuscher SA, Brandt CC & Jardine PM. 2005. Using Soil Physical and Chemical Properties to
Estimate Bulk Density. Soil Science Society of America Journal 69(1): 51‐56.
Hofstede RGM & Rossenaar AJGA. 1995. Biomass of grazed, burned and undistubed páramo
grasslands, Colombia. II. Root mass and aboveground:belowground ratio. Arctic
Alpine Research 27(1): 13‐18.
Hofstede RGM, Mondragon MX & Rocha CM. 1995. Biomass of grazed, burned and
undisturbed paramo grasslands, Colombia. I. Above ground vegetation. Arctic and
Alpine Research 27(1): 1‐12.
140
Hofstede RGM. 1995a. Effects of Livestock Farming and Recommendations for Management
and Conservation of Páramo Grasslands (Colombia). Land Degradation and
Rehabilitation 6(3): 133‐147.
Hofstede RGM. 1995b. The effects of grazing and burning on soil and plant nutrient
concentrations in Colombian páramo grasslands. Plant and Soil 173(1): 111‐132.
Hörmann PK & Pichler H. 1982. Geochemistry, petrology and origin of the Cenozoic volcanic
rocks of the Northern Andes in Ecuador. Journal of Volcanology and Geothermal
Research 12(3‐4): 259‐282.
Huygens D, Boeckx P, Van Cleemput O, Oyarzún C & Godoy R. 2005. Aggregate and soil
organic carbon dynamics in South Chilean Andisols. Biogeosciences 2(2): 159‐174.
Instituto Geofisico. 2006 Centros Volcánicos con Actividad Plio‐Cuaternaria.
http://www.igepn.edu.ec/, Instituto Geofisico, Escuela Politecnica Nacional, Quito,
Ecuador.
IPCC (Intergovernmental Panel on Climate Change). 2007. Climate Change 2007: Synthesis
Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report
of the Intergovernmental Panel on Climate Change. (Core Writing Team, editors
Pachauri RK & Reisinger A ). IPCC, Geneva: 104 pp.
Jackson ML, Lim CH & Zelazny LW. 1986. Oxides, hydroxides and aluminosilicates. In:
Methods of Soil Analysis, Part 1 – Physical and mineralogical methods, editor Klute
A. Soil Science Society of America Inc., Madison, Wisconsin: 101‐150.
Jansen B, Haussmann NS, Tonneijck FH, Verstraten JM & De Voogt P. 2008a. Characteristic
straight‐chain lipid ratios as a quick method to assess past forest ‐ páramo
transitions in the Ecuadorian Andes. Palaeogeography, Palaeoclimatology,
Palaeoecology 262(3‐4): 129‐139.
Jansen B, Nierop KGJ, Hageman JA, Cleef AM & Verstraten JM. 2006. The straight‐chain lipid
biomarker composition of plant species responsible for the dominant biomass
production along two altitudinal transects in the Ecuadorian Andes. Organic
Geochemistry 37(11): 1514‐1536.
Jansen B, Nierop KGJ, Tonneijck FH, Van der Wielen FWM & Verstraten JM. 2007. Can
isoprenoids in leaves and roots of plants serve as biomarkers for past vegetation
changes? A case study from the Ecuadorian Andes. Plant and Soil 291(1): 181‐198.
Jansen B, Van Loon E, Hooghiemstra H & Verstraten JM. 2008b. Analysis of plant biomarkers
greatly improves reconstruction of palaeo‐environments. Geology (submitted).
Jenkinson DS, Poulton PR & Bryant C. 2008. The turnover of organic carbon in subsoils. Part
1. Natural and bomb radiocarbon in soil profiles from the Rothamsted long‐term
field experiments. European Journal of Soil Science 59(2): 391‐399.
Jimenez JJ & Decaëns T. 2000. Vertical distribution of earthworms in grassland soils of the
Colombian Llanos. Biology and Fertility of Soils 32(6): 463‐473.
Jobbágy EG & Jackson RB. 2000. The vertical distribution of soil organic carbon and its
relation to climate and vetetation. Ecological Applications 10(2): 423‐436.
Keatinge TH. 1983. Development of pollen assemblage zones in soil profiles in Southeastern
England. Boreas 12(1): 1‐12.
141
Keese J, Mastin T & Yun D. 2007. Identifying and Assessing Tropical Montane Forests on the
Eastern Flank of the Ecuadorian Andes. Journal of Latin American Geography 6(1):
63‐83.
Kelso GK. 1994. Pollen Percolation Rates in Euroamerican‐Era Cultural Deposits in the
Northeastern United States. Journal of Archaeological Science 21(4): 481‐488.
Kögel‐Knabner I. 2002. The macromolecular organic composition of plant and microbial
residues as inputs to soil organic matter. Soil Biology and Biochemistry 34(2): 139‐
162.
Kovda IW, Lynn W, Williams D & Chichagova O. 2001. Radiocarbon age of Vertisols and its
interpretation using data on gilgai complex in the north Caucasus. Radiocarbon
43(2): 603‐609.
Kraus TEC, Dahlgren RA & Zasoski RJ. 2003. Tannins in nutrient dynamics of forest
ecosystems ‐ a review. Plant and Soil 256(1): 41‐66.
Kristiansen S, Dalsgaard K, Holst MK, Aaby B & Heinemeier J. 2003. Dating of prehistoric
burial mounds by 14C analysis of SOM fractions. Radiocarbon 45(1): 101‐112.
Krumbein WC & Pettijohn FJ. 1938. Manual of Sedimentary Petrography. Appleton Century
Crofts, New York: 549 pp.
Krumbein WC. 1934. Size frequency distributions of sediments. Journal of Sedimentary
Petrology 4: 65‐77.
Kubota T. 1972. Aggregate‐formation of allophanic soils: Effect of drying on the dispersion
of the soils. Soil Science and Plant Nutrition 18: 79‐87.
Laegaard S. 1992. Influence of fire in the grass páramo vegetation of Ecuador. In: Páramo,
an Andean ecosystem under human influence, editors Balslev H & Luteyn JL.
Academic Press, London: 151‐170.
Lal R. 2004. Soil carbon sequestration to mitigate climate change. Geoderma 123(1‐2): 1‐22.
Lavelle P. 1988. Earthworm activities and the soil system. Biology and Fertility of Soils 6(3):
237‐251.
Lee KE & Foster RC. 1991. Soil fauna and soil structure. Australian Journal of Soil Research
29(6): 745‐775.
Lee KE. 1959. The earthworm fauna of New Zealand. Government Printer, Wellington.
Bulletin of the New Zealand Department of Scientific and Industrial Research 130:
486 pp.
Leuschner C, Moser G, Bertsch C, Röderstein M & Hertel D. 2007. Large altitudinal increase
in tree root/shoot ratio in tropical mountain forests of Ecuador. Basic and Applied
Ecology 8(3): 219‐230.
López‐Sandoval MF. 2004. Agricultural and settlement frontiers in the Tropical Andes: The
páramo belt of Northern Ecuador, 1960‐1990. Institut für Geographie an der
Universität Regensburg, Regensburg, Regensburger Geographische Schriften 37:
180 pp.
Luteyn JL. 1992. Páramos: Why study them? In: Páramo: An Andean ecosystem under
human influence, editors Balslev H & Luteyn JL. Academic Press, London: 1‐14.
142
Manning MR, Lowe DC, Melhuish WH, Sparks RJ, Wallace G, Brenninkmeijer CAM & McGill
RC. 1990. The use of Radiocarbon measurements in atmospheric studies.
Radiocarbon 32(1): 37‐58.
Matlab 2004. Version 7.1 (release 14), copyright 1984–2004 The Mathworks Inc.
Matus F, Amigo X & Kristiansen SM. 2006. Aluminium stabilization controls organic carbon
levels in Chilean volcanic soils. Geoderma 132(1‐2): 158‐168.
Meijer HAJ, Pertuisot MF & van der Plicht J. 2006. High‐accuracy 14C measurements for
atmospheric CO2 samples by AMS. Radiocarbon 48(3): 355–372.
Meijer HAJ, van der Plicht J, Gislefoss JS & Nydal R. 1995. Comparing long‐term atmospheric
14
C and 3H records near Groningen, the Netherlands with Fruholmen, Norway and
Izana, Canary Island 14C stations. Radiocarbon 37 (1): 39‐50.
Mena Vasconez P, Medina G & Hofstede RGM. 2001. Los páramos del Ecuador.
Particularidades, problemas, perspectivas. Quito: 299 pp.
Miedema R, Pape T & Van de Waal GJ. 1974. A method to impregnate wet soil samples,
producing high‐quality thin sections. Netherlands Journal of Agricultural Science
22: 37‐39.
Mizota C & Van Reeuwijk LP. 1989. Clay mineralogy and chemistry of soils formed in volcanic
material in diverse climatic regions. ISRIC, Wageningen, Soil Monographs 2: 185
pp.
Mook WG & Streurman HJ. 1983. Physical and chemical aspects of radiocarbon dating. Pact
8(II.1): 31‐55.
Mook WG & van de Plassche O. 1986. Radiocarbon dating. In: Sea‐level Research: A Manual
for the Collection and Evaluation of Data, editor Van de Plassche O. Geo Books,
Norwich: 525‐560.
Moore PD, Webb JA & Collinson ME. 1991. Pollen analysis. 2nd ed. Blackwell Scientific
Publications, Oxford: 207 pp.
Morris A. 1997. Afforestation Projects in Highland Ecuador: Patterns of Success and Failure.
Mountain Research and Development 17(1): 31‐42.
Moscol Olivera MC & Cleef AM. 2008a. A phytosociological study of the páramo along two
altitudinal transects in El Carchi province, northern Ecuador. Phytocoenologia (in
press).
Moscol Olivera MC & Cleef AM. 2008b. Vegetation composition and altitudinal distribution
of Andean rain forests in El Angel and Guandera reserves, northern Ecuador.
Phytocoenologia (in press).
Moscol Olivera MC & Hooghiemstra H. 2008. A palaeoecological study of upper forest line
changes in Guandera site (Northern Ecuador) during the last 3000 years. Review of
Palaeobotany and Palynology (in preparation).
Murphy CP. 1986. Thin section preparation of soils and sediments. AB Academic Publishers,
Berkhamsted: 149 pp.
Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB & Kent J. 2000. Biodiversity
hotspots for conservation priorities. Nature 403(6772): 853‐858.
143
Nanzyo M. 2002. Unique properties of volcanic ash soils. Global environmental research
6(2): 99‐112.
Neira D, Van Den Berg B & De la Torre F. 2006. El Mecanismo de Desarrollo Limpio en
Ecuador: Un diagnóstico rápido de los retos y oportunidades en el Mercado de
Carbono. Banco Interamericano de Desarrollo, Quito: 99 pp.
Nesse WD. 2004. Introduction to optical mineralogy. Oxford University Press Inc., New York,
348 pp.
Nierop KGJ & Buurman P. 1999. Insoluble organic matter fractions in incipient podzol B
horizons: preservation of aliphatic biopolymers from roots. Humic Substances in
the Environment 1(2): 29‐37.
Nierop KGJ, Buurman P & de Leeuw JW. 1999. Effect of vegetation on chemical composition
of H horizons in incipient podzols as characterized by 13C NMR and pyrolysis‐
GC/MS. Geoderma 90(1‐2): 111‐129.
Nierop KGJ, Tonneijck FH, Jansen B & Verstraten JM. 2007. Organic Matter in Volcanic Ash
Soils under Forest and Páramo along an Ecuadorian Altitudinal Transect. Soil
Science Society of America Journal 71(4): 1119‐1127.
Nierop KGJ, Van Bergen PF, Buurman P & Van Lagen B. 2005. NaOH and Na4P2O7 extractable
organic matter in two allophanic volcanic ash soils of the Azores Islands ‐ a
pyrolysis GC/MS study. Geoderma 127(1‐2): 36‐51.
Nydal R & Lövseth K. 1983. Tracing bomb 14C in the atmosphere 1962‐1980. Journal of
Geophysical Research 88(C6): 3621‐3642.
O'Brien BJ & Stout JD. 1978. Movement and turnover of soil organic matter as indicated by
carbon isotope measurements. Soil Biology and Biochemistry 10(4): 309‐317.
Oreskes N, Shrader‐Frechette K & Belitz K. 1994. Verification, Validation, and Confirmation
of Numerical Models in the Earth Sciences. Science 263(5147): 641‐646.
Orlova LA & Panychev VA. 1993. The reliability of radiocarbon dating buried soils.
Radiocarbon 35(3): 369‐377.
Parfitt RL & Childs CW. 1988. Estimation of forms of Fe and Al ‐ a review, and analysis of
contrasting soils by dissolution and Mossbauer methods. Australian Journal of Soil
Research 26 (1): 121‐144.
Parfitt RL & Kimble JM. 1989. Conditions for Formation of Allophane in Soils. Soil Science
Society of America Journal 53(3): 971‐977.
Parfitt RL & Wilson AD. 1985. Estimation of allophane and halloysite in three sequences of
volcanic soils, New Zealand. In: Volcanic soils, Catena Supplement 7, editors Caldas
Fernandez E & Yaalon DH. Catena Verlag, Braunschweig: 1‐8.
Pawluk S. 1987. Faunal micromorphological features in moder humus of some Western
Canadian soils. Geoderma 40 (1‐2): 3‐16.
Percival HJ, Parfitt RLS & Neal A. 2000. Factors Controlling Soil Carbon Levels in New Zealand
Grasslands: Is Clay Content Important? Soil Science Society of America Journal
64(5): 1623‐1630.
144
Pessenda LCR, Gouveia, SEM & Aravena R. 2001. Radiocarbon dating of total soil organic
matter and its comparison with 14C ages of fossil charcoal. Radiocarbon
43(2): 595‐601.
Pitkanen J & Nuutinen V. 1997. Distribution and abundance of burrows formed by
Lumbricus terrestris L. and Aporrectodea caliginosa Sav. in the soil profile. Soil
Biology and Biochemistry 29(3‐4): 463‐467.
Podwojewski P, Poulenard J & Herbillon AJ. 2002. Overgrazing effects on vegetation cover
and properties of volcanic ash soil in the páramo of Llangahua and La Esperanza
(Tungurahua, Ecuador). Soil Use and Management 18(1): 45‐55.
Poulenard J, Podwojewski P & Herbillon AJ. 2003. Characteristics of non‐allophanic Andisols
with hydric properties from the Ecuadorian páramos. Geoderma 117(3‐4): 267‐
281.
Poulenard J, Podwojewski P, Janeau JL & Collinet J. 2001. Runoff and soil erosion under
rainfall simulation of Andisols from the Ecuadorian Páramo: effect of tillage and
burning. Catena 45(3): 185‐207.
Powers JS & Schlesinger WH. 2002. Relationships among soil carbon distributions and
biophysical factors at nested spatial scales in rain forests of northeastern Costa
Rica. Geoderma 109(3‐4): 165‐190.
Rahman S, Takaki H, Tamai M & Nagatomo, Y. 1996. Distribution of zinc, manganese,
copper, cobalt and nickel in andosol profiles. Soil Science and Plant Nutrition
42(4): 881‐891.
Ramsay PM & Oxley ERB. 1996. Fire temperatures and postfire plant community dynamics
in Ecuadorian grass páramo. Plant Ecology 124(2): 129‐144.
Rasse DP, Mulder J, Moni C & Chenu C. 2006. Carbon turnover kinetics with depth in a
French loamy soil. Soil Science Society of America Journal 70(6): 2097‐2105.
Reimer PJ, Baillie, MGL, Bard E, Bayliss A, Beck JW, Bertrand CJH, Blackwell PG, Buck, CE,
Burr GS, Cutler KB, Damon PE, Edwards RL, Fairbanks RG, Friedrich M, Guilderson
TP, Hogg AG, Hughen KA, Kromer B, McCormac G, Manning S, Ramsey CB, Reimer
RW, Remmele S, Southon JR, Stuiver M, Talamo S, Taylor FW, van der Plicht J &
Weyhenmeyer CE. 2004. IntCal04 Terrestrial Radiocarbon Age Calibration, 0–26 Cal
Kyr BP. Radiocarbon 46(3): 1029‐1058.
Rhoades CC, Eckert GE & Coleman D. 2000. Soil carbon differences among forest,
agriculture, and secondary vegetation in lower montane Ecuador. Ecological
Applications 10(2): 497‐505.
Richards LA. 1948. Porous plate apperatus for measuring moisture retention and
transmission by soil. Soil Science 66: 105‐110.
Röderstein M, Hertel D & Leuschner C. 2005. Above‐ and below‐ground litter production in
three tropical montane forests in southern Ecuador. Journal of Tropical Ecology
21(5): 483‐492.
Rodríguez Rodríguez A, Arbelo CD, Guerra JA, Mora JL, Notario JS & Armas CM. 2006.
Organic carbon stocks and soil erodibility in Canary Islands Andosols. Catena 66(3):
228‐235.
145
Rühlmann J, Körschens M & Graefe J. 2006. A new approach to calculate the particle density
of soils considering properties of the soil organic matter and the mineral matrix.
Geoderma 130(3‐4): 272‐283.
Rumpel C, Kögel‐Knabner I & Bruhn F. 2002. Vertical distribution, age, and chemical
composition of organic carbon in two forest soils of different pedogenesis. Organic
Geochemistry 33 (10): 1131‐1142.
Salomons JB. 1986. Paleoecology of volcanic soils in the Colombian Central Cordillera
(Parque Nacional Natural de los Nevados). Universiteit van Amsterdam. Cramer J,
Berlin, Stuttgart, Amsterdam. Dissertationes Botanicae 95: 212 pp.
Salvador‐Blanes S, Minasny B & McBratney AB. 2007. Modelling long‐term in situ soil profile
evolution: application to the genesis of soil profiles containing stone layers.
European Journal of Soil Science 58(6): 1535‐1548.
Sarmiento L. 1986. Ecological features of climate in high tropical mountains. In: High altitude
tropical biogeography, editors Vuilleumier F & Monasterio M. Oxford University
Press, New York, Oxford: 11‐45.
Sauer W. 1971. Geologie von Ecuador. Gebrüder Borntreger, Berlin: 316 pp.
Scharpenseel HW & Becker‐Heidmann P. 1992. Twenty‐five years of radiocarbon dating
soils: paradigm of erring and learning. Radiocarbon 34(3): 541‐549.
Schnitzer M. 1982. Organic matter characterization. In: Methods of soil analysis. Part 2:
Chemical and microbiological properties, editor Page A. American Society of
Agronomy, Soil Science Society of America, Madison, Wisconsin: 581‐594.
Shoji SM, Nanzyo M & Dahlgren RA. 1993. Volcanic ash soils: genesis, properties and
utilization. Elsevier, Amsterdam, London, New York, Tokyo: 288 pp.
Shore JS, Bartley DD & Harkness DD. 1995. Problems encountered with the 14C dating of
peat. Quaternary Science Reviews 14(4): 373‐383.
Shuster WD, Subler S & McCoy EL. 2001. Deep‐burrowing earthworm additions changed the
distribution of soil organic carbon in a chisel‐tilled soil. Soil Biology and
Biochemistry 33(7‐8): 983‐996.
Smith P, Smith JU, Powlson DS, McGill WB, Arah JRM, Chertov OG, Coleman K, Franko U,
Frolking S, Jenkinson DS, Jensen LS, Kelly RH, Klein‐Gunnewiek H, Komarov AS, Li C,
Molina JAE, Mueller T, Parton WJ, Thornley JHM & Whitmore AP. 1997. A
comparison of the performance of nine soil organic matter models using datasets
from seven long‐term experiments. Geoderma 81(1‐2): 153‐225.
Soethe N, Lehmann J & Engels C. 2006. The Vertical Pattern of Rooting and Nutrient Uptake
at Different Altitudes of a South Ecuadorian Montane Forest. Plant and Soil
286(1): 287‐299.
Sollins P, Homann P & Caldwell BA. 1996. Stabilization and destabilization of soil organic
matter: mechanisms and controls. Geoderma 74(1‐2): 65‐105.
Sommer M, Kaczorek D, Kuzyakov Y & Breuer J. 2006. Silicon pools and fluxes in soils and
landscapes ‐ a review. Journal of Plant Nutrition and Soil Science 169(3): 310‐329.
SPSS Inc. 2001. SPSS Base 11.0 User's Guide. Chigago, Illinois.
146
Stakman WP, Valk GA & Van der Harst GG. 1969. Determination of soil moisture retention
curves. 1. Sand box apparatus. 2. Pressure membrane apparatus. Institute for Land
and Water Management Research, Wageningen.
Stern CR. 2004. Active Andean volcanism: its geologic and tectonic setting. Revista geologica
de Chile 31(2): 161‐206.
Stevenson FJ. 1994. Humus chemistry. Genesis, Composition, Reactions. 2nd ed. John Wiley,
New York: 496 pp.
Stoops G. 2003. Guidelines for analysis and description of soil and regolith thin sections. Soil
Science Society of America Inc., Madison, Wisconsin: 184 pp.
Swift MJ, Heal OW & Anderson JM. 1979. Decomposition in terrestrial ecosystems. Studies in
Ecology. Blackwell Scientific Publications, Oxford: 371 pp.
Swift R. 2001. Sequestration of carbon by soil. Soil Science 166 (11): 858‐871.
Takahashi T & Shoji S. 2002. Distribution and classification of volcanic ash soils. Global
Environmental Research 6(2): 83‐97.
Takahashi T, Ikeda Y, Fujita K & Nanzyo M. 2006. Effect of liming on organically complexed
aluminum of nonallophanic Andosols from northeastern Japan. Geoderma 130(1‐
2): 26‐34.
Telford RJE, Heegaard E & Birks HJB. 2004. All age‐depth models are wrong: but how badly?
Quaternary Science Reviews 23(1‐2): 1‐5.
Theng BKG. 1980. Soils with variable charge. Society of Soil Science, Lower Hutt: 448 pp.
Tonneijck FH & Jongmans AG. 2008. The influence of bioturbation on the vertical
distribution of soil organic matter in volcanic ash soils: a case study in northern
Ecuador. European Journal of Soil Science 59(6): 1063‐1075.
Tonneijck FH, Hageman JA, Sevink J & Verstraten JM. 2008a. Tephra stratification of volcanic
soils in Northern Ecuador. Geoderma 144(1‐2): 231‐247.
Tonneijck FH, Jansen B, Nierop KGJ, Verstraten JM, Sevink L & De Lange L. 2008b. Carbon
stocks and stabilisation mechanisms in volcanic ash soils under natural upper
montane forest and grassland in northern Ecuador. Global Biogeochemical Cycles
(submitted).
Tonneijck FH, Van der Plicht J, Jansen B, Verstraten JM & Hooghiemstra H. 2006.
Radiocarbon dating of soil organic matter fractions in Andosols in Northern
Ecuador. Radiocarbon 48(3): 337‐353.
Tonneijck FH. 1999. Carbon fixation in the soil – volcanic soils in the tropical environments
of Ecuador. Modelconcept and field study. Student report, Universiteit van
Amsterdam, Amsterdam, Quito: 85 pp.
Torn MSS, Trumbore E, Chadwick OA, Vitousek PM & Hendricks DM. 1997. Mineral control
of soil organic carbon storage and turnover. Nature 389(6647): 170‐173.
Tyler AN, Carter S, Davidson DA, Long DJ & Tipping R. 2001. The extent and significance of
bioturbation on 137Cs distributions in upland soils. Catena 43(2): 81‐99.
Ugolini FC, Dahlgren R, Shoji S & Ito T. 1988. An Example of Andosolization and
Podzolization as Revealed by Soil Solution Studies, Southern Hakkoda,
Northeastern Japan. Soil Science 145(2): 111‐125.
147
Van Dam D, Van Breemen N & Veldkamp E. 1997. Soil organic carbon dynamics: variability
with depth in forested and deforested soils under pasture in Costa Rica.
Biogeochemistry 39(3): 343‐375.
Van der Hammen T & Beglinger E. 1989. Soil fauna in the Parque los Nevados transect. In:
Studies on Tropical Andean Ecosystems, editors Van der Hammen T, Diaz‐
Piedrahita S & Julio Alvarez V. Gebrüder Borntraeger, Berlin: 443‐454.
Van der Plicht J, Wijma S, Aerts AT, Pertuisot MH & Meijer HAJ. 2000. Status report: The
Groningen AMS facility. Nuclear Instruments and Methods in Physics Research
Section B: Beam Interactions with Materials and Atoms 172(1‐4): 58‐65.
Van der Plicht J. 2005. WinCal25. Groningen, the Netherlands, Center for Isotope Research,
University of Groningen.
Van Mourik JM, Wartenbergh PE, Mook WG & Streurman HJ. 1995. Radiocarbon dating of
palaeosols in aeolian sands. Mededelingen Rijks Geologische Dienst 52: 425‐440.
Van Mourik JM. 1999. The use of micromorphology in soil pollen analysis. The interpretation
of the pollen content of slope deposits in Galicia, Spain. Catena 35(2‐4): 239‐257.
Van Reeuwijk LP. 2002. Procedures for soil analysis. ISRIC, Wageningen, Technical paper, 6th
edition: 120 pp.
Veldkamp E. 1994. Soil organic carbon dynamics in pastures established after deforestation
in the humid tropics of Costa Rica. PhD thesis, Wageningen Universiteit en
Researchcentrum, Wageningen: 117 pp.
Veneklaas EJ. 1990. Nutrient Fluxes in Bulk Precipitation and Throughfall in Two Montane
Tropical Rain Forests, Colombia. Journal of Ecology 78(4): 974‐992.
Wada K. 2002. Allophane and imogolite. In: Soil mineralogy with environmental applications,
editor Dick WE. Soil Science Society of America Inc., Madison, Wisconsin: 1051‐
1087.
Walch KM, Rowley JR & Norton NJ. 1970. Displacement of pollen grains by earthworms.
Pollen et spores 12: 39‐44.
Wang YR, Amundson R & Trumbore S. 1996. Radiocarbon dating of soil organic matter.
Quaternary Research 45(3): 282‐288.
Ward JH. 1963. Hierarchical groupings to optimize an objective function. Journal of the
American Statistical Association 58: 234–244
Wilcke W, Yasin S, Abramowski U, Valarezo C & Zech W. 2002. Nutrient storage and
turnover in organic layers under tropical montane rain forest in Ecuador. European
Journal of Soil Science 53(1): 15‐27.
Wille M, Hooghiemstra H, Hofstede RGM, Fehse J & Sevink J. 2002 Upper forest line
reconstruction in a deforested area in Northern Ecuador based on pollen and
vegetation analysis. Journal of Tropical Ecology 18(3): 409‐440.
Wunder S. 1996. Deforestation and the Uses of Wood in the Ecuadorian Andes. Mountain
Research and Development 16(4): 367‐381.
148
149
Summary

Volcanic ash soils in Andean ecosystems
Unravelling organic matter distribution and stabilisation
The Ecuadorian Andes harbours fragile tropical montane forest and grassland (páramo)
ecosystems of exceptional biodiversity that are threatened by land‐use change, e.g.
deforestation, grazing and burning. Land‐use change may turn volcanic ash soils – the
dominant soil type in this region – from sinks into sources of CO2 to the atmosphere, which
affects the global carbon cycle. To estimate potential carbon gains or losses from soils upon
land‐use change more detailed knowledge on the processes affecting the vertical
distribution and stabilisation of soil organic matter (SOM) is urgently needed. Meanwhile,
ecologically sustainable reforestation and restoration of degraded areas in the Ecuadorian
Andes is severely hindered, because the natural position of the upper forest line (UFL) in the
absence of human interference is still subject to scientific debate. To enhance
understanding of UFL dynamics, analysis of paleorecords contained in volcanic ash soils has
great potential. However, for a correct interpretation of these paleorecords knowledge on
the processes affecting the vertical distribution of SOM, including paleoecological proxies, is
required.

In volcanic ash soils, boundary conditions for the vertical distribution of SOM are set by
geogenetic processes, specifically the amount and timing of tephra depositions. However,
the actual vertical distribution of SOM depends on subsequent pedogenetic processes: e.g.
aboveground and belowground litter input, organic matter decomposition, bioturbation and
leaching. SOM accumulation in volcanic ash soils is generally related to the formation of
organo‐mineral and/or organo‐metallic complexes, each protecting SOM against
decomposition through mechanisms that are still subject to debate. The main objective of
this PhD research has therefore been to gain insight into the processes affecting the vertical
distribution and stabilisation of SOM in volcanic ash soils in tropical montane forest and
páramo ecosystems in the northern Ecuadorian Andes, by studying the influence of
geogenetic and pedogenetic processes. This research thus contributes to sustainable
management of volcanic ash soils as carbon sinks and supports the reconstruction of the
natural position of the UFL, which is needed for ecologically sustainable reforestation and
restoration of degraded areas.

A general description of the physiography of the research area was provided in Chapter 1.
To increase understanding of the processes affecting the vertical distribution of SOM in
volcanic ash soils, and more specifically its chronological ordering (‘chronostratigraphy’),
radiocarbon dating of SOM is required, which was the topic of Chapter 2. Interpretation of
151
SOM radiocarbon ages is not straightforward, because they are influenced by pedogenetic
processes. Dating of fractions of SOM obtained by alkali–acid extraction is promising,
although which (non‐mobile) fraction (humic acid or humin) renders the most accurate
radiocarbon dates is still subject to debate. To determine which fraction should be used for
radiocarbon dating of volcanic ash soils and to evaluate if the chronostratigraphy of SOM is
suitable for paleoecological research, we measured radiocarbon ages of both fractions and
related calibrated ages to soil depth. We compared the time frames covered by these soils
with those of peat sequences nearby to provide independent evidence. Humic acid (HA) was
significantly older than humin, except for the mineral soil samples just beneath a forest floor
(ectorganic horizons) where the opposite was true. In peat sections, radiocarbon ages of HA
and humin were equally accurate. In the soils, calibrated ages increased significantly with
increasing depth. Age inversions and homogenisation were not observed at the applied
sampling distances. We conclude that in volcanic ash soils lacking a thick ectorganic horizon,
dating of HA renders the most accurate results, since humin was contaminated by roots.
Contrarily, in mineral soil samples just beneath a forest floor, humin ages were more
accurate, because HA was then contaminated by younger HA illuviated from the ectorganic
horizons that hardly contain Al or Fe. Overall, the chronostratigraphy of SOM in the volcanic
ash soils studied appears to be suitable for paleoecological research.

In Chapter 3 the influence of geogenetic processes on the vertical distribution of SOM was
investigated by unravelling the tephra stratigraphy of the volcanic ash soils studied. We
combined proxies traditionally used in stratigraphic research (mineral assemblages, grain
size distribution, and chemical element ratios) with soil organic carbon contents and
radiocarbon dating both at a high vertical resolution. Our results show that these soil
profiles were each formed in three distinct tephra deposits. Although the deposits
contained a similar assemblage of minerals, we were able to differentiate these deposits
because of their characteristic organic carbon distribution, grain size distribution and typical
SrO to Na2O, CaO or crystalline Al2O3 ratios. Unravelling the tephra stratigraphy improved
understanding of the vertical distribution of soil organic carbon, including paleoecological
proxies, in the studied soils. Besides, we concluded that bioturbation must have played an
important role in current pedogenesis, resulting in overprinting (merging, mixing) of the
paleosol. Surprisingly, in spite of bioturbation, a linear age depth relationship exists, leading
to the hypothesis that the zone of bioturbation shifted upwards during soil formation.
Therefore, we tentatively concluded that paleoecological proxies are stratified in our soils,
albeit probably more crudely than in undisturbed peat bogs or lake sediments. To further
increase understanding of the vertical distribution of paleoecological proxies in soils,
detailed research on bioturbation was recommended, which was subsequently done in
Chapter 4.

In Chapter 4 the impact of bioturbation on the vertical distribution of SOM was studied. Soil
faunal bioturbation (‘bioturbation’) is often cited as a major process influencing the vertical
distribution of SOM. The influence of bioturbation on vertical SOM transport is complex
152
because it is the result of interaction between different groups of soil faunal species that
redistribute SOM through the soil profile in distinct ways. We performed a semi‐quantitative
micromorphological analysis of soil faunal pedofeatures and related their occurrence to the
vertical distribution of SOM and high resolution radiocarbon dating. The páramo soil data
suggest that bioturbation was largely responsible for the vertical distribution of SOM, while
illuviation and root input were of minor importance. Bioturbation was caused by endogeic
species, which typically mix the soil only over short vertical distances. Short vertical distance
mixing was apparently enhanced by upward shifting of bioturbation as a result of soil
thickening due to SOM accumulation. A change from páramo to forest vegetation was
accompanied by a change from endogeic to epigeic species. These latter species do not
redistribute material vertically, which eventually resulted in the formation of thick
ectorganic horizons in the forest.

In Chapter 5 soil organic carbon stocks were quantified and mechanisms of soil organic
carbon stabilisation were examined. To clarify the role of chemical and physical carbon
stabilisation mechanisms in the volcanic ash soils studied, we applied selective extraction
techniques, performed X‐ray diffraction analysis of the clay fraction and estimated pore size
distribution. Our results show that organic carbon stocks under both vegetation types
roughly doubled global averages for volcanic ash soils. The carbon stabilisation mechanisms
involved are 1) direct stabilisation of SOM in organo‐metallic (Al‐humus) complexes; 2)
indirect protection of SOM through low soil pH and toxic levels of Al and 3) physical
protection of SOM due to a very high microporosity. Moreover, in the case of the forest
soils, inherent recalcitrance of SOM itself was responsible for substantial accumulation in
ectorganic horizons. Both vegetation types contributed to soil acidification, thus increasing
SOM accumulation and inducing positive feedbacks. Most types of land‐use change
involving natural Andean forest and páramo will result in immediate and substantial carbon
loss to the atmosphere.

In Chapter 6 the processes affecting the vertical distribution of SOM were further explored
in a modelling exercise, integrating knowledge obtained in the previous chapters and with a
special focus on the impact of soil volume change due to SOM incorporation. The impact of
soil volume change on the vertical distribution of SOM is often overlooked, although it may
influence depth‐dependent processes such as bioturbation. We first determined the extent
of soil volume change with a geochemical mass balance involving immobile element
concentrations. Then, we tested our hypothesis of an upward shifting bioturbation zone in
response to soil thickening with a one‐dimensional partial differential equation model. Our
results show that soil volume change is a major soil forming process in the volcanic ash soil
studied. SOM accumulation resulted in more than a tripling of initial soil thickness within
4800 years, from ~20 cm to 77 cm. Our modelling exercise confirmed that an upward
shifting bioturbation zone in response to soil thickening is indeed plausible. We conclude
that soil volume change cannot be ignored when modelling soil organic carbon dynamics in
SOM‐rich soils such as volcanic ash soils.
153

Finally, in the Synthesis the conclusions of these chapters were placed in a broader
framework and combined to define implications of this research for 1) sustainable
management of volcanic ash soils as carbon sinks and 2) paleoecological research on the
natural position of the UFL, which is needed for ecologically sustainable reforestation and
restoration of degraded areas. With respect to the first topic, the tropical Andean
ecosystems should urgently be protected from destructive land‐use change, not only for the
sake of preserving their outstanding biodiversity but also for their function as exceptionally
large long term organic carbon sink. Most types of land‐use change affecting natural forest
and páramo ecosystems are expected to cause a net flow of carbon from the soil to the
atmosphere for a long period of time. In the case of volcanic ash soils, models of soil organic
carbon dynamics may be improved by incorporating Alp/Alo ratios instead of clay content, by
taking into account vertical transport of SOM by bioturbation due to different groups of soil
faunal species and by accounting for soil volume change. With respect to the second topic,
vertical mixing of SOM is halted as soon as forest vegetation colonises a site, so that the
chronostratigraphic order is intact from that moment onwards. In contrast, vertical mixing
of SOM takes place throughout the soil profile at sites with páramo vegetation.
Nevertheless, due to short vertical distance mixing over intervals < 5 cm (i.e. ~500 years) in
combination with an upward shifting zone of bioturbation, a (crude) chronostratigraphic
order of SOM is maintained in the soils studied. As long as paleoecological proxies are
distributed through the soil profile by similar processes as SOM, they will also be in (crude)
chronostratigraphic order. However, a major consequence is also that any sample taken in
these soils inevitably produces a mixed signal at a high resolution temporal scale.
154
Samenvatting

Vulkanische asbodems in andiene ecosystemen
Ontrafelen van de distributie en stabilisatie van organische stof
De Ecuadoriaanse Andes herbergt fragiele tropisch alpiene bos‐ en grasland (páramo)
ecosystemen die worden gekenmerkt door een enorme biodiversiteit. Deze ecosystemen
worden bedreigd door veranderingen in landgebruik zoals ontbossing, begrazing en
branden. Verandering in landgebruik zou vulkanische asbodems – het dominante
bodemtype in deze regio – kunnen veranderen van netto opslagplaatsen (sinks) naar netto
bronnen (sources) van CO2 richting de atmosfeer en dat heeft effect op de mondiale
koolstofcyclus. Om de toe‐ of afname in bodemkoolstof na verandering in landgebruik te
kunnen inschatten, is er op korte termijn meer kennis nodig over de processen die de
verticale distributie en stabilisatie van organische stof in bodems (SOM) beïnvloeden.
Intussen wordt ecologisch duurzame herbebossing en herstel van gedegradeerde gebieden
in de Ecuadoriaanse Andes belemmerd omdat de positie van de bosgrens (UFL) in
afwezigheid van menselijk ingrijpen nog steeds onderhevig is aan wetenschappelijke
discussie. Analyse van paleo‐archieven (paleorecords) die opgeslagen zijn in vulkanische
asbodems lijkt een veelbelovende manier om het begrip van de UFL dynamiek te vergroten.
Voor een correcte interpretatie van deze paleorecords is echter kennis van de processen die
de verticale distributie van SOM beïnvloeden noodzakelijk.
In vulkanische asbodems is de verticale distributie van SOM allereerst afhankelijk van
geogenetische processen, met name van de hoeveelheid en het moment van tefra‐afzetting.
Daarnaast hangt de uiteindelijke verticale distributie van SOM af van de daaropvolgende
pedogenetische processen zoals aanvoer van boven‐ en ondergrondse biomassa, afbraak
van SOM, bioturbatie en uitspoeling. Accumulatie van SOM in vulkanische asbodems wordt
meestal gerelateerd aan de vorming van complexen van organische stof met metalen (Al) of
mineralen (allofaan), die SOM beschermen tegen afbraak door mechanismen die nog
onderwerp van wetenschappelijke discussie zijn. Het hoofddoel van dit promotieonderzoek
was daarom: het verkrijgen van inzicht in de processen die de verticale distributie en
stabilisatie van SOM beïnvloeden in vulkanische asbodems in bos‐ en páramo ecosystemen
in het noorden van de Ecuadoriaanse Andes door het onderzoeken van geogenetische en
pedogenetische processen. Dit promotieonderzoek draagt daarmee bij aan duurzaam
beheer van vulkanische asbodems als koolstof sink en ondersteunt onderzoek naar de
natuurlijke positie van de UFL, nodig om herbebossing en herstel van gedegradeerde
gebieden op een ecologisch duurzame manier mogelijk te maken.

Een algemene beschrijving van de fysiografie van het onderzoeksgebied is opgenomen in
Hoofdstuk 1. Hoofdstuk 2 behandelt de datering van SOM met de koolstof‐14 methode. Dit
155
verschaft kennis over de processen die de verticale distributie van SOM in vulkanische
asbodems beïnvloeden en in het bijzonder de chronologische ordening daarvan
(‘chronostratigrafie’). Interpretatie van 14C dateringen van SOM is complex, omdat deze
beïnvloed worden door pedogenetische processen. Dateren van specifieke SOM fracties
zoals verkregen na loog‐zuur extractie is veelbelovend, hoewel nog onduidelijk is welke
immobiele fractie (huminezuren of humine) de meest nauwkeurige 14C datering oplevert.
Om te bepalen welke fractie gebruikt zou moeten worden voor 14C datering aan vulkanische
asbodems en om te bekijken of de chronostratigrafie van SOM geschikt is voor
paleoecologisch onderzoek hebben we beide fracties gedateerd en de gecalibreerde
leeftijden gerelateerd aan bodemdiepte. Ter controle hebben we de tijdspanne verkregen in
bodems vergeleken met die in venen in de nabije omgeving. Huminezuren waren significant
ouder dan humine, behalve voor minerale bodemmonsters net onder de bosvloer
(ectorganische horizonten) waar het omgekeerde gold. In venen waren de 14C leeftijden van
huminezuren en humine even nauwkeurig. De gecalibreerde leeftijd nam significant toe met
de diepte in zowel de venen als de bodems. Leeftijdinversies en homogenisatie werden niet
waargenomen met de toegepaste bemonsteringsintervallen. We concluderen dat in
vulkanische asbodems zonder dikke ectorganische horizonten datering aan huminezuren
het meest nauwkeurig is, omdat humine vervuild is door jonge wortels. In minerale
bodemmonsters net onder dikke ectorganische horizonten geldt het omgekeerde en is
datering van humine het meest nauwkeurig, omdat huminezuren in dat geval vervuild zijn
door inspoeling van jonge huminezuren afkomstig uit de dikke ectorganische horizonten die
nauwelijks Al of Fe bevatten. In het algemeen kunnen we concluderen dat de
chronostratigrafie van SOM in de bestudeerde vulkanische asbodems geschikt is voor
paleoecologisch onderzoek.
In hoofdstuk 3 wordt de invloed van geogenetische processen op de verticale distributie van
SOM onderzocht door de tefrastratigrafie van de bestudeerde vulkanische asbodems te
ontrafelen. Hiertoe hebben we proxies die traditioneel gebruikt worden in stratigrafisch
onderzoek (zoals mineralencombinaties, korrelgrootteverdelingen en verhoudingen tussen
chemische elementen) gecombineerd met SOM gehalten en 14C dateringen met een hoge
verticale resolutie. Onze resultaten toonden aan dat deze bodemprofielen elk gevormd zijn
in drie verschillende tefra‐afzettingen. Ondanks dat deze afzettingen eenzelfde
mineralencombinatie vertonen waren we toch in staat ze van elkaar te onderscheiden,
dankzij de karakteristieke organische koolstofdistributie, korrelgrootteverdeling en typische
verhoudingen van SrO tegen Na2O, CaO en kristallijne Al2O3. Het ontrafelen van de
tefrastratigrafie vergrootte het begrip van de verticale distributie van SOM, inclusief
paleoecologische proxies, in de bestudeerde bodems. Daarnaast concludeerden we dat
bioturbatie een belangrijke rol gespeeld moet hebben in de huidige bodemvorming,
resulterend in het ‘overprinten’ (samenvoegen, mengen) van de paleosol. Ondanks
bioturbatie is er verrassend genoeg een lineaire relatie tussen leeftijd en diepte. Op basis
daarvan hebben we de hypothese geformuleerd dat de bioturbatiezone langzaam omhoog
geschoven moet zijn tijdens bodemvorming. We concluderen daarom voorzichtig dat de
paleoecologische proxies chronologisch gestratificeerd zijn in deze bodems, hoewel
156
waarschijnlijk op een wat grovere schaal dan in ongestoorde venen of meerafzettingen.
Onze aanbeveling om onderzoek naar bioturbatie te doen om meer inzicht te verkrijgen in
de verticale distributie van paleoecologische proxies in vulkanische asbodems is vervolgens
uitgewerkt in Hoofdstuk 4.
In Hoofdstuk 4 wordt de impact van bioturbatie op de verticale distributie van SOM
onderzocht. Bioturbatie door bodemfauna (kortweg ‘bioturbatie’) wordt vaak genoemd als
belangrijk bodemvormend proces dat van invloed is op de verticale distributie van SOM. De
invloed van bioturbatie op verticaal transport van SOM is complex omdat dit het resultaat is
van interactie tussen verschillende typen bodemfauna. Deze typen transporteren SOM op
verschillende manieren door het bodemprofiel. We hebben een semi‐kwantitatieve
micromorfologische analyse uitgevoerd met betrekking tot door de bodemfauna
geproduceerde verschijnselen en het voorkomen daarvan gerelateerd aan de verticale
distributie van SOM en 14C dateringen, beide met hoge verticale resolutie. De data
verkregen uit páramobodems suggereerden dat met name bioturbatie verantwoordelijk is
voor de verticale distributie van SOM, terwijl inspoeling en beworteling een ondergeschikte
rol spelen. De waargenomen bioturbatie blijkt te worden veroorzaakt door endogeïsche
soorten, die gekenmerkt worden door menging van de bodem over korte verticale
afstanden. Menging over korte verticale afstanden wordt klaarblijkelijk versterkt door het
omhoogschuiven van de bioturbatiezone als gevolg van het aangroeien van de bodem door
SOM accumulatie. Een verandering van páramo‐ naar bosvegetatie wordt vergezeld van een
verandering van endogeïsche soorten naar epigeïsche soorten. Epigeïsche soorten
veroorzaken geen verticaal transport van materiaal, wat in het bos uiteindelijk resulteert in
accumulatie van organische stof in dikke ectorganische horizonten bovenop de minerale
bodem.
In Hoofdstuk 5 zijn de organische koolstofvoorraden (stocks) in de bodem gekwantificeerd
en zijn mechanismen van koolstofstabilisatie geanalyseerd. Om de rol van chemische en
fysische koolstofstabilisatiemechanismen te verklaren hebben we selectieve
extractietechnieken toegepast, X‐ray diffractie‐analyse aan de kleifractie uitgevoerd en de
poriëngrootteverdeling onderzocht. Onze resultaten toonden aan dat stocks van organische
koolstof in de bodem onder beide vegetatietypen grofweg twee keer zo groot zijn als het
mondiale gemiddelde voor vulkanische asbodems. De betrokken koolstof‐
stabilisatiemechanismen zijn: 1) directe stabilisatie van SOM door complexvorming met
metalen (Al‐humus); 2) indirecte bescherming van SOM door een zure bodem pH en
toxische hoeveelheden Al en 3) fysische bescherming van SOM door een zeer hoge
microporositeit. Daarnaast speelt de slechte afbreekbaarheid van SOM zelf een rol bij de
accumulatie in ectorganische horizonten onder bos. Beide vegetatietypen blijken bij te
dragen aan bodemverzuring en op die manier SOM accumulatie te versterken, resulterend
in positieve terugkoppelingen. De meeste veranderingen in landgebruik in natuurlijke
andiene bos‐ en páramo ecosystemen zullen onmiddellijk leiden tot verlies van koolstof uit
de bodem naar de atmosfeer.
157
In Hoofdstuk 6 zijn de processen die de verticale distributie van SOM beïnvloeden nader
onderzocht met een model. Hierbij hebben we de kennis verkregen in de voorgaande
hoofdstukken geïntegreerd en ons in het bijzonder gericht op de impact van veranderingen
in bodemvolume door SOM accumulatie. De impact van veranderingen in bodemvolume op
de verticale distributie van SOM wordt vaak genegeerd, hoewel het diepte‐afhankelijke
processen zoals bioturbatie kan beïnvloeden. Eerst hebben we de mate van bodemvolume‐
verandering vastgesteld met een geochemische massabalans, waarbij we gebruik maakten
van concentraties immobiele elementen. Daarna hebben we onze hypothese van een
omhoog schuivende bioturbatiezone ten gevolge van bodemaangroei getoetst met een
ééndimensionaal partiële differentiaalvergelijking model. Onze resultaten demonstreerden
dat bodemvolumeverandering een zeer belangrijk proces is in de bestudeerde vulkanische
asbodems. Accumulatie van SOM heeft meer dan een verdriedubbeling van de initiële
bodemdikte, van ~20 cm naar 77 cm, veroorzaakt binnen 4800 jaar. Het model bevestigde
dat een omhoogschuivende bioturbatiezone ten gevolge van bodemaangroei inderdaad
plausibel is. We concluderen daarom dat veranderingen in bodemvolume niet genegeerd
mogen worden in modellen die de dynamiek van organische koolstof in SOM‐rijke bodems
zoals vulkanische asbodems beschrijven.
Tot slot zijn de conclusies van deze voorgaande hoofdstukken in een breder kader geplaatst
en gecombineerd om implicaties te formuleren voor: 1) het ecologisch duurzaam beheren
van vulkanische asbodems als sinks voor organische koolstof en 2) paleoeocologisch
onderzoek naar de natuurlijke ligging van de bosgrens, waarover kennis vereist is om
ecologisch duurzame herbebossing en herstel van gedegradeerde gebieden mogelijk te
maken. Met betrekking tot het eerste thema concluderen we dat de andiene ecosystemen
dringend beschermd moeten worden tegen destructief landgebruik, niet alleen om de
overweldigende biodiversiteit te beschermen maar ook vanwege hun functie als zeer grote,
lange termijn koolstof sink. De meeste veranderingen in landgebruik met betrekking tot
natuurlijke bos‐ en páramo ecosystemen zullen gedurende langere tijd een netto verlies van
koolstof uit de bodem veroorzaken. In het geval van vulkanische asbodems kunnen
modellen over de dynamiek van organische koolstof in de bodem verbeterd worden door
Alp/Alo verhoudingen te gebruiken in plaats van kleigehalte, door rekening te houden met
verticaal transport van SOM door bioturbatie veroorzaakt door verschillende typen
bodemfauna en door rekening te houden met veranderingen in bodemvolume. Met
betrekking tot het tweede thema merken we op dat de verticale menging van SOM stopt
zodra bosvegetatie een gebied koloniseert, zodat de chronostratigrafische ordening vanaf
dat moment intact blijft. In gebieden met páramovegetatie speelt verticale menging van
SOM een rol door het hele profiel heen. Desondanks blijft een grove chronostratigrafische
ordening van SOM in stand in de bestudeerde páramobodems doordat menging slechts over
korte verticale afstanden (< 5 cm of ongeveer 500 jaar) plaatsvindt en de bioturbatiezone
omhoog schuift tijdens bodemvorming. Mits paleoecologische proxies op dezelfde manier
door de bodem gedistribueerd worden als SOM zullen deze ook grofweg
chronostratigrafisch geordend zijn. Een belangrijke consequentie is echter ook dat elk
monster onvermijdelijk een gemengd signaal bevat op een tijdschaal met fijnere resolutie.
158
Resumen

Los suelos de ceniza volcánica en ecosistemas andinos
Desentrañando la distribución y estabilización de la materia orgánica
Los Andes Ecuatorianos alojan frágiles ecosistemas tropicales montañosos de bosque y
pastizales (páramo) de excepcional biodiversidad, los cuales están amenazados por cambios
del uso de la tierra, por ejemplo la deforestación, el pastoreo y la quema. Estos cambios
podrían hacer que los suelos de ceniza volcánica – el tipo de suelo más abundante en esta
región – pasen a ser ya no un depósito de dióxido de carbono sino una fuente de éste hacia
la atmósfera. De esta forma afecta al ciclo global del carbono. Para determinar los
incrementos o las pérdidas de carbono orgánico en los suelos debido a los cambios del uso
de la tierra, urge tener un conocimiento más detallado de los procesos que influyen en la
distribución vertical y la estabilización de la materia orgánica del suelo (SOM). Mientras
tanto la reforestación y recuperación de áreas degradadas de forma ecológicamente
sustentable se ven severamente limitadas, porque la posición altitudinal del límite superior
de bosque (UFL) en ausencia de interferencia humana es aún objeto de debate. Para
mejorar el entendimiento de la dinámica del UFL, es prometedora la investigación del
registro fósil contenido en los suelos de ceniza volcánica. Sin embargo, para una
interpretación correcta de esos registros fósiles es necesario un conocimiento de los
procesos que influyen en la distribución vertical de la materia orgánica del suelo (SOM).
En los suelos de ceniza volcánica, las condiciones básicas para la distribución vertical de la
SOM están determinadas por los procesos geogenéticos, específicamente la cantidad y el
momento de deposición de la tefra. Sin embargo, la distribución vertical de la SOM actual
depende de los siguientes procesos pedogenéticos: la entrada de la SOM por hojarasca y
raíces, la descomposición de la SOM, la bioturbación y la lixiviación. Acumulación de la SOM
en los suelos de ceniza volcánica en general está relacionada con la formación de complejos
órgano‐minerales y/o órgano‐metálicos, que protegen la SOM contra la descomposición por
mecanismos que son aún objeto de debate. El objetivo principal de esta tesis doctoral ha
sido por lo tanto: el obtener más conocimiento acerca de los procesos que influyen en la
distribución vertical y la estabilización de la SOM en los suelos de ceniza volcánica en los
ecosistemas de bosque y de páramo en el norte de los Andes Ecuatorianos, mediante el
estudio de la influencia de procesos geogenéticos y pedogenéticos. De esta forma esta tesis
contribuye al manejo sustentable de los suelos de ceniza volcánica como depósitos de
carbono orgánico y ayuda a la reconstrucción de la posición altitudinal natural del UFL, que
es necesaria para lograr la reforestación y la recuperación de áreas degradadas de forma
ecológicamente sustentable.

159
En el Capítulo 1 se presentó una descripción general de la fisiografía del área de estudio.
Para mejorar el entendimiento de los procesos que influyen en la distribución vertical de la
SOM en los suelos de ceniza volcánica, y en particular su ordenamiento cronológico, es
necesario datar la SOM por radiocarbono (14C) y este tema se presentó en el Capitulo 2. La
interpretación de dataciones 14C de la SOM no es simple, porque éstas son influenciadas por
procesos pedogéneticos. Datar fracciones de SOM por extracción álcali‐ácido es
prometedor, pero qué fracción inmóvil (ácido húmico o humina) dará las dataciones 14C más
precisas es aún objeto de debate. Para determinar qué fracción debe ser usada para datar
por 14C los suelos de ceniza volcánica y para evaluar si la cronoestratigrafía de la SOM es
apropiada para la investigación paleoecológica, datamos ambas fracciones y relacionamos
las edades calibradas con profundidades de suelo. Comparamos los intervalos de tiempo
cubiertos por los suelos estudiados con aquellos de secciones de turberas cercanas para
obtener evidencia independiente. La edad de ácido húmico (HA) fue significativamente
mayor que la de la humina, excepto para las muestras de suelo mineral justo debajo de un
suelo de bosque (horizontes ectorgánicos), donde se presenta la situación opuesta. En
secciones de turbera las edades 14C de HA y humina fueron igual de precisas. En los suelos,
las edades calibradas aumentan de manera significativa con la profundidad. No se ha
observado inversiones de edad u homogenización en los intervalos de muestreo aplicados.
En conclusión, datar el HA en suelos de ceniza volcánica sin horizonte orgánico grueso da
resultados más precisos, ya que la humina fue contaminada por raíces. Contrariamente a
esto, en muestras de suelo mineral tomadas justo por debajo de horizontes ectorgánicos,
las edades de la humina fueron más precisas, porque en este caso el HA fue contaminado
por HA más joven iluviado desde horizontes ectorgánicos que apenas contienen metales (Al
y Fe). En general, la cronoestratigrafía de la SOM resulta apropiada para investigaciones
paleoecológicas.
En el Capitulo 3 fue investigada la influencia de los procesos geogenéticos en la distribución
vertical de la SOM por dilucidación de la estratigrafía de las tefras de los suelos ceniza
volcánica estudiados. Combinamos los proxies usados tradicionalmente en estudios
estratigráficos (asociaciones de minerales, distribuciónes granulométricas y proporciónes de
elementos químicos) con contenidos de SOM y dataciones 14C, ambos con una alta
resolución vertical. Nuestros resultados indican que cada uno de estos suelos se ha formado
en tres distintos depósitos de tefra. Aunque estos depósitos contenían una asociación
similar de minerales, fue posible diferenciar entre depósitos debido a su característica
distribución vertical de carbono orgánico, su granulometría, y la proporción típica de SrO a
Na2O, CaO o Al2O3 cristalino. El dilucidar la estratigrafía de las tefras mejoró la comprensión
de la distribución vertical del SOM, incluidos los proxies paleoecológicos, en los suelos
estudiados. Además, concluimos que probablemente la bioturbación juega un papel
importante en la pedogénesis actual, resultando en superposición (fusión, mezcla) del
paleosuelo. Es sorprendente ver que, a pesar de la bioturbación, existe una relación lineal
entre edad y profundidad, lo que conduce a la hipótesis de que la zona de la bioturbación se
movió hacia arriba durante la formación del suelo. Por lo tanto, concluimos
160
provisionalmente que los proxies paleoecológicos se presentan estratificados en nuestros
suelos, aunque en cierto modo más toscamente que en las turberas o los sedimentos
lacustres. Para mejorar el entendimiento de los procesos que influyen en la distribución
vertical de los proxies paleoecológicos de los suelos, recomendamos investigar la
bioturbación con más detalle, lo que a continuación hicimos en el Capitulo 4.
En el Capitulo 4 fue investigado el impacto de la bioturbación por la fauna de suelos
(bioturbación) en la distribución vertical de la SOM. Muchas veces se ha referido a la
bioturbación como el principal proceso que influye en la distribución vertical de la SOM.
Esta influencia es muy compleja porque es el resultado de la interacción entre diferentes
grupos de especies de fauna del suelo que redistribuyen la SOM de maneras distintas a
través del perfil del suelo. Realizamos un análisis micromorfológico semicuantitativo de los
rasgos edafológicos de la fauna del suelo y relacionamos su presencia con la distribución
vertical de la SOM y las dataciones 14C de alta resolución. Los datos de suelo de páramo
sugieren que la bioturbación es ampliamente responsable de la distribución vertical de la
SOM, mientras que la lixiviación y entrada de raíces fueron de menor importancia. La
bioturbación fue causada por especies endogeicas, que mezclan el suelo típicamente sólo
por cortas distancias verticales. La mezcla vertical de corta distancia fue aparentemente
realzada por el desplazamiento hacia arriba de la zona de la bioturbación como resultado
del engrosamiento del suelo debido a la acumulación de la SOM. Un cambio de vegetación
de páramo a bosque fue acompañado por un cambio de especies endogeicas a especies
epigeicas. Estas últimas no redistribuyen material verticalmente, lo que origina
eventualmente la formación de gruesos horizontes ectorgánicos en el bosque.
En el Capitulo 5 los stocks de carbono orgánico del suelo fueron cuantificados y los
mecanismos de estabilización del SOM fueron examinados. Para aclarar el papel químico y
físico de los mecanismos de estabilización en los suelos de ceniza volcánica estudiados,
aplicamos técnicas de extracción selectiva, llevamos a cabo análisis de difracción con rayos X
de la fracción arcillosa y estimamos la distribución del tamaño de poros. Nuestros resultados
muestran que los stocks de carbono orgánico fueron excepcionalmente elevados bajo
ambos tipos de vegetación, duplicando aproximadamente el promedio global en reportado
para los suelos de ceniza volcánica. Los mecanismos de estabilización de SOM involucrados
son 1) estabilización directa de la SOM en complejos órgano‐metálicos (Al‐humus); 2)
protección indirecta de la SOM a través de un pH bajo del suelo y de niveles tóxicos de Al y
3) la protección física de la SOM causada por una microporosidad muy alta. Además, en el
caso de suelos de bosque, la resistencia inherente de la SOM fue responsable de una
acumulación sustancial en los horizontes ectorgánicos. Ambos tipos de vegetación
contribuyeron a la acidificación del suelo, incrementando así la acumulación de SOM e
induciendo a un retorno positivo. La mayoría de los cambios del uso de la tierra que
involucran a los ecosistemas de bosque y páramo andino resultarán en una pérdida
considerable e inmediata de carbono orgánico del suelo hacia la atmósfera.
161
En el Capitulo 6 los procesos que influyen en la distribución vertical de la SOM fueron
estudiados en un modelo preliminar, integrando el conocimiento obtenido en los capítulos
previos y con un énfasis especial en el impacto del cambio del volumen de suelo debido a la
incorporación de SOM. El impacto del cambio del volumen de suelo es frecuentemente
desestimado, aunque podría influenciar los procesos que dependen de la profundidad de
suelo como la bioturbación. Primero, determinamos la magnitud del cambio en volumen
con un balance de masa geoquímica que involucra a los elementos químicos inmóviles. En
seguida, probamos nuestra hipótesis de un desplazamiento hacia arriba de la zona de la
bioturbación como resultado del engrosamiento del suelo en un modelo uni‐dimensional
basado en ecuaciones diferenciales en derivadas parciales. Nuestros resultados indican que
el cambio en volumen de suelo es un proceso pedogenético principal en los suelos
estudiados. La acumulación de SOM resultó en más de una triplicación del grueso inicial de
suelo en menos de 4800 años, desde ~20 cm hacia 77 cm. Nuestro modelo confirmó que un
desplazamiento hacia arriba de la zona de la bioturbación es ciertamente plausible como
resultado del engrosamiento del suelo. Concluímos que no se pueden negar los cambios del
volumen de suelo en los modelos de la dinámica del carbono orgánico que involucra a los
suelos ricos en SOM como los suelos de ceniza volcánica.
Finalmente, en la Síntesis las conclusiones de los capítulos previos fueron puestas en un
marco más amplio y combinadas para definir implicaciones para 1) el manejo sustentable de
los suelos de ceniza volcánica como depósitos de carbono orgánico y 2) el análisis
paleoecológico sobre la posición natural del UFL que es necesario para la reforestación y
recuperación de áreas degradadas de forma ecológicamente sustentable. Acerca del primer
tema, los ecosistemas andinos deben ser protegidos urgentemente de los cambios
destructivos del uso de la tierra, no solamente con el propósito de preservar su gran
biodiversidad, sino también por su función como depósito enorme de carbono. La mayoría
de los cambios del uso de la tierra que involucran a los ecosistemas andinos resultarán en
una perdida considerable e inmediatamente de carbono orgánico del suelo hacia la
atmósfera por un periodo largo. En el caso de los suelos de ceniza volcánica, los modelos de
la dinámica del carbono orgánico pueden ser mejorados incorporando los ratios Alp/Alo en
lugar del contenido de arcilla, tomando en cuenta el transporte vertical de la SOM por la
bioturbación causada por diferentes grupos de fauna de suelo y explicando el cambio del
volumen de suelo. Acerca del segundo tema, la mezcla vertical de la SOM se detiene tan
pronto como un bosque coloniza un sitio y el ordenamiento cronológico sigue quedando
intacto desde ese momento. Por el contrario, en el páramo la bioturbación afecta todo el
suelo. Sin embargo, cierto orden cronoestratigráfico de la SOM sigue siendo conservado en
estos suelos también, debido a la mezcla por cortas distancias verticales sobre intervalos < 5
cm (o ~500 años) combinada con un desplazamiento hacia arriba de la zona de la
bioturbación. Sí los proxies paleoecológicos son distribuidos en los suelos por los mismos
procesos que la SOM, también tendrán un cierto orden cronoestratigráfico. Sin embargo,
una consecuencia importante es también que cada muestra tomada de estos suelos
produce inevitablemente una señal mezclada a una escala temporal de alta resolución.
162
Appendix I List of RUFLE publications

This list contains publications as a result of the RUFLE project up to 2008, other than those
presented in this PhD thesis.
Bakker J, Moscol Olivera M & Hooghiemstra H. 2008. Holocene environmental change at the
upper forest line in northern Ecuador. The Holocene 18(6): 877 ‐ 893.
Jansen B & Nierop KGJ. 2008. Methyl ketones in high altitude Ecuadorian Andosols confirm
excellent conservation of plant‐specific n‐alkane patterns. Organic Geochemistry
(in press).
Jansen B, Nierop KGJ, Hageman JA, Cleef AM & Verstraten JM. 2006a. The straight chain
lipid biomarker composition of plant species responsible for the dominant biomass
production along two altitudinal transects in the Ecuadorian Andes. Organic
Geochemistry. 37: 1514‐1536.
Jansen B, Nierop KGJ, Kotte MC,de Voogt P & Verstraten JM. 2006b. The applicability of
accelerated solvent extraction (ASE) to extract lipid biomarkers from soils. Applied
Geochemistry. 21: 1006‐1015.
Jansen B, Haussmann NS, Tonneijck FH, de Voogt P & Verstraten JM. 2008a. Characteristic
straight‐chain lipid ratios as a quick method to assess past forest – páramo
transisitons in the Ecuadorian Andes. Palaeogeography, Palaeoclimatology,
Palaeoecology. 262: 129‐139.
Jansen B, Nierop KGJ, Tonneijck FH, van der Wielen FW &Verstraten JM. 2007. Can
isoprenoids in leaves and roots of plants along altitudinal gradients in the
Ecuadorian Andes serve as biomarkers? Plant and Soil. 291: 181‐198.
Jansen B, Van Loon E, Hooghiemstra H & Verstraten JM. 2008b. Analysis of plant biomarkers
greatly improves reconstruction of palaeo‐environments. Geology (submitted).
Moscol Olivera MC & Cleef AM. 2008a. A phytosociological study of the páramo along two
altitudinal transects in El Carchi province, northern Ecuador. Phytocoenologia (in
press).
Moscol Olivera MC & Cleef AM. 2008b. Vegetation composition and altitudinal distribution
of Andean rain forests in El Angel and Guandera reserves, northern Ecuador.
Phytocoenologia (in press).
Nierop KGJ & Jansen B. 2008. Extensive transformation of organic matter vs. excellent lipid
preservation at the upper, superhumid Guandera páramo. Chemical Geology
(submitted).
Nierop KGJ, Tonneijck FH, Jansen B & Verstraten JM. 2007. Organic matter in volcanic ash
soils under forest and páramo along an Ecuadorian altitudinal transect. Soil Science
Society of America Journal. 71: 1119‐1127.

163
Profile number G11 Elevation (m asl) 3330 (n=1) Topography Hilly
Date of description 21 May 2004 Inclination (˚) 8 Landform Mountain
Author Femke Tonneijck MSc. Exposition (˚) 260 Landelement Slope
GPS coordinates 0˚ 35’ 12’’ N / 77˚ 42’37’’ W Soil T 10 cm (˚C) 11.2 Position Upper slope
Location Northern Ecuadorian Andes, province of Cajas, canton Soil T 30 cm (˚C) Slope Concave
Huaca, parroquia Mariscal Sucre, near biological station Land use Grazing and logging Vegetation Forest patch
Guandera, forest patch along dirt road to Mariscal Sucre, WRB 2006 soil
Folic, Hemic, Aluandic, Fulvic Andosol (Dystric, Placic, Thixotropic)
outcrop classification
1
Horizon Depth Colour Description Diagnostics
L
F ‐10 – 0 2.5 YR 2/2 No mottling observed, clay loam, no coarse fragments observed, massive structure, very friable, non sticky, Folic, Hemic
common very fine to fine vughs, few medium vughs, high porosity, common very fine to fine roots, very few
medium roots, clear, smooth boundary to
2
Ah1 0 – 40 5 YR 2/2 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, non sticky, no Aluandic,
1/2 voids observed, few very fine to fine roots, very few medium roots, clear, smooth boundary to Fulvic
2Ahb 40 – 55 5 YR 1.7/1 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, non sticky, no Silandic,
voids observed, very few very fine to medium roots, clear, smooth boundary to Melanic
2Bhsb 55 – 65 5 YR 2/1 Many fine to medium prominent mottles, 2.5 YR 4/8, with clear boundaries, very few fine to medium Placic
prominent mottles, 10 YR 6/8, with clear boundaries, clay loam, no coarse fragments observed, massive
structure, broken compacted layer with nodular iron, friable, non sticky, no voids observed, very few very
fine to fine roots, clear, irregular boundary, with tonguing (5 YR 3/2) into the BC (25 x 3 cm, 3 times within 1
meter)
2BCb 65 – 125 Mixed colour Mixed colour consists of 45% 10 YR 7/6, 45% 10 YR 6/8, 10% 10 YR 1.7/1, no mottling observed, coarse sand, Vitric
10 YR 6/6 from 75 ‐ 95 cm: common fine to medium subangular gravel, slightly weathered; from 95 ‐ 135 cm: few fine
Appendix II Soil profile descriptions
to medium subangular gravel, slightly weathered, single grain, loose, non sticky, common very fine to fine
interstital voids, high porosity, no roots observed, clear, smooth boundary to
2C 125 – Mixed colour Mixed colour consists of 45% 2.5 Y 8/2, 45% 2.5 Y 8/6, 10% 2.5 Y 2/1, no mottling observed, coarse sand, few Vitric
180+ 10 YR 7/3 fine gravel, subangular, fresh, single grain, loose, non sticky, common very fine to fine interstital voids, high
porosity, no roots observed

1
Overprinted zone not identified exactly, because of lack of monolith sample, but the sample at 30 cm probably belonged to the overprinted zone, see Chapter 3
2
Topsoil was truncated, see Chapter 5

Profile number G1 Elevation (m asl) 3501 ± 20 (n = 9) Topography Undulating
GPS coordinates 0˚ 35’ 27’’ N/ 77˚ 42’ 1’’ W Soil T 10 cm (˚C) 9.8 Position Upper slope
Location Northern Ecuadorian Andes, province of Carchi, canton Soil T 30 cm (˚C) 9.8 Slope Convex
Huaca, parroquia Mariscal Sucre, biological station Land use Nature conservation Vegetation Tropical montane forest
Guandera, near signpost T9 on the forest trail. WRB 2006 soil
Folic, Hemic, Endo‐aluandic, Endofulvic Histosol
classification
3
L ‐76 – ‐75 Undecomposed litter
F ‐75 – ‐35 10 R 2/2 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky,
common very fine to medium vughs, high porosity, common very fine to fine roots, few medium and very
few coarse roots, clear smooth boundary to
Folic, Hemic
H ‐35 – 0 2.5 YR 2/2 No mottling observed, clay loam, no coarse fragments observed, massive structure, very friable, non sticky,
common very fine to medium vughs, high porosity, common very fine to fine roots, few medium to coarse
roots, clear smooth boundary to
Ah 0 – 12 7.5 YR 2/2 No mottling observed, silty clay, no coarse fragments observed, massive structure, friable, non sticky, no
voids observed, very few very fine to medium roots, clear smooth boundary to
Bw 12 – 40 7.5 YR 2/3 Very vague and diffuse mottling probably from root origin, clay loam, no coarse fragments observed,
Aluandic,
massive structure, friable, non sticky, no voids observed, very few very fine to medium roots, at 15 cm on
Fulvic
1/2 the other side of the pit a lense of silty material occurs (colour 2.5 Y 7/4, 10 cm thick and 40 long), gradual
smooth boundary to

2Ahb 40 – 70 7.5 YR 1.7/1 No mottling observed, silty clay, no coarse fragments observed, massive structure, friable, non sticky, no Silandic,
voids observed, very few very fine to medium roots, irregular abrupt boundary to Melanic

2Bsb 70 – 72 2.5 YR 4/8 No mottling observed, coarse sand, no coarse fragments observed, single grain structure, cemented, Placic
extremely firm, non sticky, voids not determined, no roots observed, irregular abrupt boundary to
3BCb 72 – 85+ 10 YR 2/2 No mottling observed, coarse sand, abundant fine and few medium subangular, slightly weathered gravel Vitric
(2.5 Y 6/8), single grain structure, loose, non sticky, voids not determined, no roots observed

3
Overprinted zone from 15 – 39 cm, see Chapter 3

Profile number G2 Elevation (m asl) 3520 ± 14 (n=9) Topography Undulating
GPS coordinates 0˚ 35’ 27’’ N / 77˚ 42’ 1’’ W Soil T 10 cm (˚C) 9.4 Position Upper slope
Guandera, near signpost T10 of the forest trail. WRB 2006 soil
Folic, Hemic, BathyAluandic, Bathyfulvic Histosol
classification
4
Lv ‐99 ‐ ‐98 Undecomposed litter
F1 ‐98 ‐ ‐93 10 R 2/2 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky,
many fine to medium vughs, high porosity, few very fine and medium roots, common fine roots and very
few coarse roots, clear smooth boundary to
F2 ‐93 ‐ ‐48 10 R 2/3 No mottling observed, clay loam, no coarse fragments observed, massive structure, very friable, slightly
sticky, many fine and common medium vughs, high porosity, few very fine, common fine, very few medium Folic, Hemic
to coarse roots, gradual smooth boundary to
H ‐48 – 0 2.5 YR 2/3 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, slightly sticky, no
voids observed, few very fine, common fine, very few medium to coarse roots (coarse roots concentrated at
boundary with next horizon), abrupt smooth boundary to
Ah 0 – 20 5 YR 2/2 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, slightly sticky, no
voids observed, very few medium roots, abrupt smooth boundary to
Aluandic,
Bw 20 – 56 7.5 YR 3/3 Very few distinct coarse dull yellowish brown (10 YR 5/3) mottles with clear boundaries, clay loam, no coarse
Fulvic
fragments observed, massive structure, friable, slightly sticky, no voids observed, very few medium roots,
abrupt smooth boundary to
2Ahb 56 – 101 5 YR 1.7/1 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, slightly sticky, no Aluandic,
voids observed, very few medium roots (water springs along root channels), abrupt smooth boundary to Melanic
3BCb 101 – 10 YR 3/4 Obtained by augering, no further data. Vitric
112+

4
Overprinted zone not identified, because of lack of monolith sample, see Chapter 3

Profile number G4 Elevation (m asl) 3616 ± 12 (n=12) Topography Hilly
GPS coordinates 0˚ 35’ 53’’ N / 77˚ 41’ 41’’ W Soil T 10 cm (˚C) 8.6 Position Middle slope
Location Northern Ecuadorian Andes, province of Cajas, canton Soil T 30 cm (˚C) 8.7 Slope Complex
Guandera, along forest trail near upper forest line. WRB 2006 soil Folic, Hemic, Aluandic, Fulvic, Thaptaluandic, Andosol (Thixotropic,
5 6
classification Colluvic , Dystric)
7
L ‐11 ‐ ‐10 2.5 YR 2/2
F ‐10 – 0 2.5 YR 2/1 No mottling observed, common very fine to fine vughs, few medium vughs, high porosity, common very fine
Folic
to fine roots, few medium and very few coarse roots, clear smooth boundary to
Ah 0 – 25 2.5 YR 2/2 No mottling observed, clay loam, no coarse fragments observed, massive structure, very friable, non sticky,
few very fine, few fine to medium vughs, medium porosity, few very fine to fine roots, very few medium to
coarse roots, clear smooth boundary to Aluandic,
AB 25 – 60 50% 2.5YR No mottling observed, clay loam, in the upper 10 cm very few fine subangular strongly weathered gravel (10 Fulvic
1.7/1, 50% YR 4/6), massive structure, friable, non sticky, very few very fine to fine vughs, low porosity, very few very
2.5 YR 2/1 fine to fine roots, clear wavy boundary to
2Ahb 60 – 135 10 YR 4/6 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, slightly sticky,
very few very fine to fine vughs, low porosity, very few very fine, fine and coarse roots, at 90 cm water
Aluandic
springs at root channels, on the border to the next horizon a lens with colour 10 YR 3/2 occurs, clear smooth
boundary to
3BCb 135 – 2.5 YR 2/2 No mottling observed, coarse sand, fine gravel, single grain, loose, non sticky, voids not determined, no
Vitric
145+ roots observed
5
Fulvic or Melanic qualifier not used because melanic index not available
6
Whether sedimentation was indeed human‐induced is not certain, but soil properties clearly deviated from general trends, see Chapter 3
7
Overprinted zone not identified, because of lack of monolith sample, see Chapter 3

Profile number G5a Elevation (m asl) 3697 ± 9 (n=9) Topography Hilly

Location Northern Ecuadorian Andes, province of Carchi, canton Soil T 30 cm (˚C) 9.1 Slope Complex
Huaca, parroquia Mariscal Sucre, biological station Land use Nature conservation Vegetation Forest patch
Guandera, forest island above forest line WRB 2006 soil Folic, Hemic, Aluandic, Endomelanic Cambisol (Hyperhumic, Thixotropic,
classification Dystric)
8
L ‐36 ‐ ‐35 Undecomposed litter
F ‐35 – 0 2.5 YR 2/4 No mottling observed, common very fine to fine vughs, few medium and very few coarse vughs, high Folic, Hemic
porosity, common very fine to fine roots, few medium roots and very few coarse roots (coarse roots
concentrated on boundary with F), clear smooth boundary to
Ah 0 – 40 5 YR 1.7/1 No mottling, observed silty clay, no coarse fragments observed, massive structure, friable, non sticky, very Aluandic,
few very fine to fine vughs, low porosity, very few very fine to coarse roots, gradual smooth boundary to Melanic

Bw 40 – 70 5 YR 2/2 No mottling observed, silty clay, very few fine subangular gravel, 2.5 Y 8/3, strongly weathered, massive Silandic
structure, friable, non sticky, very few very fine to fine vughs, low porosity, very few very fine to medium
1/2 roots, gradual smooth boundary to
2Ah1b 70 – 90 5 YR 1.7/1 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, non sticky, very
few very fine to fine vughs, low porosity, very few very fine to fine roots, clear smooth boundary to
2Ah2b 90 – 105 5 YR 1.7/1 No mottling observed, clay loam, very few medium and few fine subangular gravel, 5 YR 5/8 (1%), 2.5 Y 7/6
Silandic,
(1%), 2.5 Y 8/2 (7%), strongly weathered (sandy texture) , massive structure, friable, non sticky, very few
Melanic
very fine to fine vughs, low porosity, very few very fine to fine roots, gradual smooth boundary to
2Ah3b 105 – 140 5 YR 1.7/1 No mottling observed, silty clay, no coarse fragments observed, massive structure, friable, slightly sticky,
very few very fine to fine vughs, low porosity, very few very fine to fine roots, abrupt smooth boundary to
2Bsb 140 – 42 5 YR 7/8 No mottling observed, coarse sand, no coarse fragments observed, single grain ‐ cemented (continuous Placic
compacted iron layer), firm, non sticky, very few very fine to fine interstital voids, low porosity, no roots
observed, abrupt smooth boundary (although within one meter one pocket of 15 cm width and 10 cm depth
is observed) to
3BCb 142 – 155+ Mixed colour Mixed colour consists of 45 % 10 YR 6/8 (bright yellowish brown), 45 % 2.5 Y 8/6 (yellow), 10 % 2.5 Y 2/1 Vitric
10 YR 6/8 (black), no mottling observed, coarse sand, few fine gravel subangular, slightly weathered, single grain,
loose, non sticky, very few very fine to fine interstital voids, high porosity, no roots observed

8
Profile number G5b Elevation (m asl) 3694 ± 13 (n=9) Topography Hilly
Location Northern Ecuadorian Andes, province of Carchi, canton Soil T 30 cm (˚C) 9.9 Slope Complex
Huaca, parroquia Mariscal Sucre, biological station Land use Nature conservation, Vegetation Páramo
Guandera, páramo next to forest island of site G5A sporadic burning
WRB 2006 soil
Aluandic, Melanic, Umbric Andosol (Thixotropic)
classification
9
L Undecomposed litter
Ah 0 – 45 5 YR 1.7/1 No mottling observed, clay loam, no coarse fragments observed, massive structure, very friable, slightly Aluandic,
sticky, common very fine to fine, few medium vughs, medium porosity, common very fine, few fine, very few Melanic,
medium to coarse roots, clear smooth boundary to Umbric
Bw1 45‐ 70 7.5 YR 2/1 No mottling observed, clay loam, no coarse fragments observed, massive structure, very friable, slightly
Silandic
sticky, few very fine to fine vughs, low porosity, few very fine, very few fine roots, clear smooth boundary to
Bw2 70 – 115 7.5 YR 2/1 No mottling observed, clay loam, from 70 ‐ 85 cm: common fine, few medium, very few coarse subangular Silandic
gravel, 60% 10 YR 6/8, 39% 10 YR 7/4, 1% 5 YR 6/8, strongly weathered; from 85 ‐ 100 cm: few fine
1/2 subangular gravel, 10 YR 7/4, strongly weathered; from 100 ‐ 115 cm: few fine, common medium, very few
coarse subangular gravel, 60% 10 YR 6/8, 39% 10 YR 7/4, 1% 5 YR 6/8, strongly weathered, massive
structure, friable, non sticky, few very fine to fine vughs, few very fine, very few fine roots, clear smooth
boundary to
2Ahb 115 – 175 5 YR 1.7/1 No mottling observed, clay loam, from 140 ‐150 cm: common fine subangular gravel, 99% 2.5 Y 8/2, 1% 2.5 Silandic,
1
YR 4/8 , massive structure, very friable, slightly sticky, few very fine to fine vughs, very few very fine to fine Melanic
roots, clear smooth boundary to
2Bwb 175 – 192 7.5 YR 2/2 No mottling observed, clay loam, few fine subangular gravel (silty), 10 YR 7/6, strongly weathered, massive Silandic
structure, friable, non sticky, voids not determinded, roots not determined, abrupt smooth boundary to
2Bsb 192 – 194 2.5 YR 4/8 No mottling observed, coarse sand, no coarse fragments observed, single grain ‐ cemented (continuous Placic
weakly cemented iron layer), firm, non sticky, voids not determinded, no roots observed, abrupt smooth
boundary to
3BCb 194 – Mixed colour Mixed colour consists of 45% 2.5 Y 8/3 (pale yellow), 45% 2.5 Y 7/8 (yellow), 10% 2.5 Y 2/1 (black), no Vitric
200+ 2.5 Y 7/4 mottling observed, coarse sand, few fine and very few medium to coarse subangular gravel (silty), slightly
weathered, single grain, loose, non sticky, common very fine to fine interstital voids, no roots observed
9
Overprinted zone from 65 ‐ 116 cm, see Chapter 3

Location Northern Ecuadorian Andes, province of Carchi, canton Soil T 30 cm (˚C) 9.4 Slope Straight
Guandera, between site G5A/B and site G7 sporadic burning
WRB 2006 soil
classification
10
Ah 0 – 40 5 YR 1.7/1 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky, Aluandic,
(Black) few very fine to fine vughs, very few medium vughs. Medium porosity, common very fine, few fine, very few Melanic,
medium to coarse roots, gradual, smooth boundary to Umbric

AB 40 – 80 5 YR 2/1 Mottling (2.5 Y 8/4) not further described, silty clay, no coarse fragments observed, massive structure, Silandic
(Black) friable, non sticky, very few very fine to medium vughs, few very fine to fine, very few medium roots,
1/2 gradual, smooth boundary to
2Ahb 80 – 140 5 YR 1.7/1 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, slightly Aluandic,
(Black) sticky, very few very fine to medium vughs, very few very fine to medium roots, gradual, smooth boundary Melanic
to
2Bwb 140 – 160 7.5 YR 3/2 Very few, very fine, distinct mottles, 2.5 Y 7/8 and 2.5 YR 4/8, with sharp boundaries, clay loam, no coarse Silandic
(Brownish fragments observed, massive structure, friable, non sticky, very few very fine to fine vughs, very few very
black) fine to fine roots, clear, smooth boundary to
3BCb 160 – mixed colour Mixed colour consists of 45% 2.5 Y 8/3 (pale yellow), 45% 2.5 Y 7/6 (bright yellowish brown), 10% 2.5 Y 1.7/1 Vitric
180+ 2.5 Y 7/4 (black), very few, very fine distinct mottles, 2.5 Y 7/8, with sharp boundaries and common medium to coarse
(light yellow) prominent elongated mottles with clear boundaries and a horizontal orientation, medium sand, no coarse
fragments observed, single grain structure, loose, non sticky, common very fine to fine interstital voids, high
porosity , no roots observed
NB colour between 160 – 165 cm: mixed colour 10 YR 5/4 (dull yellowish brown)
10
Overprinted zone from 65 ‐ 108 cm, see Chapter 3

Guandera, 20 m below ‘El Itto’ sign of the army at the sporadic burning
crest. WRB 2006 soil Aluandic, Melanic, Umbric, Thaptosilandic, Thaptomelanic, Andosol
classification (Thixotropic)
11
Ah1 0 – 40 5 YR 1.7/1 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky, Aluandic,
(black) few very fine to fine, very few medium vughs, common very fine, few fine, very few medium roots, clear, Melanic,
smooth boundary to, Umbric
Ah2 40 – 70 7.5 YR 2/1 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky, Silandic
(black) very few very fine to fine vughs, few very fine, very few fine roots, clear, smooth boundary to,
1/2
2Ahb 70 – 130 7.5 YR 1.7/1 Very few medium (30 cm x 3 cm) distinct mottles, 5 YR 3/4, with clear boundaries (possibly old roots), silty Silandic,
(black) clay, no coarse fragments observed, massive structure, friable, slightly sticky, very few very fine to fine Melanic
vughs, few very fine, very few fine roots, gradual, smooth boundary to,
1
2Bwb 130 – 147 Top: 10 YR Very few medium distinct mottles (2.5 YR 3/4) with sharp boundaries , clay loam, many very fine to fine Silandic
2/1 (black), subangular gravel (2.5 Y 8/2 and 2.5 Y 7/6) strongly weathered, massive structure, very friable, non sticky,
Bottom: 10 very few very fine to fine vughs, very few very fine to fine roots, abrupt, smooth boundary to,
YR 3/2
(brownish
black)
2Bsb 147 – 150 2.5 YR 4/8 No mottling observed, coarse sand, no coarse fragments observed, single grain structure – cemented: Placic
(reddish continuous weakly cemented iron layer, very firm, non sticky, no voids observed, no roots observed, abrupt,
brown) smooth boundary to,
3BCb 150 – Mixed colour: Mixed colour consists of of 45% 2.5 Y 7/8, 45% 2.5 Y 8/4 and 10% 10 YR 1.7/1, no mottling observed, coarse Vitric
170+ 2.5 Y 7/4 sand, coarse fragments not determined, single grain structure, loose, non sticky, interstital voids, no roots
(light yellow) observed

11
Overprinted zone from 55 – 98 cm

Guandera, just below the watershed divide sporadic burning
WRB 2006 soil
Aluandic, Fulvic, Umbric Andosol (Placic, Thixotropic)
classification
12
Ah 0 – 35 7.5 YR 2/2 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky, Aluandic,
(brownish voids n.d., common fine to very fine, very few medium to coarse roots, clear, smooth boundary to Fulvic,
black) Umbric
Bw1 35 – 55 7.5 YR 3/4 No mottling observed, silty clay, few coarse subangular gravel (2.5 Y 2/3) silty strongly weathered, massive
(dark brown) structure, friable, non sticky, voids n.d., common very fine to fine roots, very few medium to coarse roots, Silandic
clear, smooth boundary to,
1/2
Bw2 55 – 70 7.5 YR 2/2 No mottling observed, silty clay, no coarse fragments observed, massive structure, very friable, non sticky, Silandic
(brownish voids n.d., few very fine to fine roots, very few medium, clear, smooth boundary to,
black)
2Ahb 70 – 80 7.5 YR 1.7/1 No mottling observed, clay loam, no coarse fragments observed, massive structure, friable, slightly sticky, Aluandic
(black) voids n.d., very few fine and medium roots, few fine roots, abrupt, smooth boundary to,
2Bsb 80 – 83 2.5 YR 4/8 No mottling observed, coarse sand, no coarse fragments observed, single grain – cemented: continuous Placic
(reddish weakly cemented iron layer, very firm, non sticky, voids n.d., no roots observed, abrupt, smooth boundary
brown) to,
3BCb 83 – 100+ 10 YR 7/8 No mottling observed, coarse sand, few fine to medium gravel, very few coars gravel, subangular, slightly Vitric
(yellow weathered, single grain, loose, non sticky, voids n.d., no roots observed.
orange)

12

Guandera, site facing the bog, just below an artificial sporadic burning
canal WRB 2006 soil
classification
13
Ah 0 – 124 5 YR 1.7/1 No mottling observed, silty clay, from 114 ‐ 124 cm: very few fine to medium subangular gravel, strongly Aluandic,
(Black) weathered, massive structure, very friable, non sticky, voids n.d., from 0 ‐ 40 cm: common very fine to fine Melanic,
roots, very few medium to coarse roots, from 41 ‐ 124 cm: few very fine to fine roots, very few medium Umbric
roots, clear, smooth boundary to,
2 Silandic,
Melanic
2Bs 124 – 127 2.5 YR 5/8 No mottling observed, coarse sand, no coarse fragments observed, single grain – cemented: continuous, Placic
(Bright weakly cemented iron layer, very firm, non sticky, voids n.d., no roots observed, clear, smooth boundary to,
reddish
brown)
3BC 127 – 2.5 Y 8/6 No mottling observed, coarse sand, few fine to medium and very few coarse subangular gravel, slightly Vitric
140+ (yellow) weathered, single grain, very friable, non sticky, voids n.d., no roots observed,

13
2Ahb from 80 – 124 cm, see Chapter 3

Bulk density pHH2O pHCaCl2 pHKCl EC C/N C/S C Cp Cfulvic acids Chumic acids Chumine
[gcm‐3] [‐] [μScm‐1] [‐] [mass %]
Ectorganic horizons Minimum 0.073 2.78 2.21 1.80 85.0 24.1 148 44.7
(forest) Maximum 0.14 4.48 3.82 3.71 269 51.5 288 53.6
Mean 0.11 3.25 2.59 2.31 190 39.2 206 51.3
Std. Deviation 0.021 0.54 0.48 0.59 61.6 8.72 40.2 2.70
n 8 10 11 11 9 11 11 11
Minimum 0.32 3.66 3.18 3.35 17.2 15.9 123 17.0 5.92 2.73 6.66 3.82
Maximum 0.45 4.57 4.05 4.21 87.1 27.3 189 21.9 10.8 4.51 11.4 8.88
Current topsoils Mean 0.37 4.10 3.81 3.96 41.6 18.9 145 19.7 7.66 3.83 8.34 6.66
Std. Deviation 0.051 0.36 0.28 0.29 24.9 4.27 19.6 1.63 1.75 0.78 2.08 2.19
n 8 8 8 8 8 8 8 8 8 4 4 4
Minimum 0.36 4.04 4.05 4.10 5.90 14.8 98.9 7.82 3.54 2.40 2.96 2.24
Maximum 0.69 5.15 4.44 4.58 30.8 23.3 134 19.2 6.92 4.08 5.95 5.98
Current subsoils Mean 0.50 4.58 4.22 4.34 14.0 16.8 123 12.2 4.83 3.41 4.66 3.78
Std. Deviation 0.11 0.31 0.14 0.15 7.08 2.22 11.9 3.49 0.99 0.59 1.22 1.40
n 11 12 12 12 12 12 12 12 12 6 6 6
Minimum 0.34 4.15 4.01 4.14 4.50 14.5 107 7.90 3.38 2.36 2.95 2.01
Maximum 0.67 5.26 4.41 4.50 18.7 30.0 167 20.3 10.7 3.51 12.0 3.96
Overprinted zone Mean 0.49 4.75 4.25 4.37 9.53 19.8 136 12.3 5.93 2.87 5.84 2.52
Appendix III Soil properties
Std. Deviation 0.12 0.37 0.13 0.12 3.95 4.25 18.9 4.22 2.18 0.42 3.16 0.76
n 12 12 12 12 12 12 12 12 12 7 7 7
Minimum 0.37 4.08 3.85 3.90 3.90 16.9 87.0 5.30 2.65 2.35 1.64 1.28
Maximum 0.71 5.24 4.64 4.66 17.0 34.3 207 20.3 14.1 4.04 7.52 3.19
Paleosol Mean 0.49 4.69 4.30 4.38 9.16 21.7 132 12.2 6.65 3.11 5.21 2.06
Std. Deviation 0.094 0.38 0.19 0.18 3.92 4.20 32.3 3.73 2.51 0.54 2.08 0.63
n 16 16 16 16 16 16 16 16 16 9 9 9
Minimum 4.99 4.88 9.50 21.0 74.8 3.78 1.97
Maximum 5.25 4.91 13.0 24.3 108 5.83 3.78
Placic horizon Mean 5.12 4.90 5.09 11.3 22.4 94.1 4.63 2.73 3.20 1.68 0.91
Std. Deviation 0.18 0.021 2.47 1.72 17.2 1.06 0.94
n 0 2 2 1 2 3 3 3 3 1 1 1
Minimum 0.89 4.75 4.45 4.60 4.00 17.1 21.8 0.085 0.01
Maximum 0.99 5.53 5.61 5.31 23.0 32.8 91.1 4.01 2.53
Unweathered ash Mean 0.94 5.27 5.15 5.00 10.7 24.0 48.4 0.88 0.65 2.16 1.11 0.68
Std. Deviation 0.037 0.23 0.31 0.21 6.10 4.80 22.0 1.23 0.83
n 7 12 12 12 12 12 12 12 9 1 1 1

LOI Gravel Sand Silt Clay Alt Ald Alo Alp Alk Allophane Alp/Alo (Alo ‐ Alp)/Sio pHNaF
950˚C OM‐free
[mass %] [mass %] [‐]
Minimum 0.064 0.070 0.006 5.99
Maximum 1.20 0.31 0.125 6.29
Ectorganic horizons
Mean 0.35 0.15 0.049 6.07
(forest)
Std. Deviation 0.36 0.094 0.040 0.15
n 11 6 11 4
Minimum 34.3 0.00 13.6 69.8 3.93 4.85 1.01 0.91 0.94 0.053 0.00 0.94 0.00 7.49
Maximum 42.8 0.00 21.0 82.4 12.5 5.80 1.50 1.57 1.54 0.144 0.75 1.00 1.01 10.7
Current topsoils Mean 38.7 0.00 17.6 73.8 8.63 5.38 1.35 1.34 1.38 0.085 0.17 0.99 0.24 9.81
Std. Deviation 2.77 0.00 2.78 5.35 3.22 0.32 0.16 0.20 0.21 0.032 0.31 0.02 0.44 1.38
n 8 7 7 7 7 8 8 8 8 8 8 8 8 5
Minimum 17.0 0.00 18.1 54.8 8.25 5.33 1.28 1.27 0.78 0.009 0.00 0.31 0.00 11.0
Maximum 38.1 0.00 29.4 72.2 20.0 8.03 2.06 2.63 1.70 0.058 7.99 1.00 2.74 11.4
Current subsoils Mean 26.0 0.00 22.1 63.5 14.4 7.12 1.71 2.06 1.19 0.029 4.41 0.61 1.90 11.2
Std. Deviation 6.35 0.00 3.35 5.23 3.40 0.79 0.20 0.44 0.29 0.015 2.66 0.23 0.87 0.13
n 12 12 12 12 12 12 12 12 12 12 12 12 12 8
Minimum 18.6 0.00 22.4 36.2 12.0 6.56 1.50 1.99 0.79 0.010 3.92 0.30 1.75 11.0
Maximum 40.1 0.00 51.7 61.6 22.4 9.04 3.45 3.67 2.67 0.048 10.4 0.78 2.76 11.5
Overprinted zone Mean 26.2 0.00 31.0 51.8 17.2 7.78 2.26 2.78 1.39 0.027 7.00 0.49 2.43 11.3
Std. Deviation 7.46 0.00 8.75 7.61 3.10 0.65 0.60 0.57 0.58 0.013 1.88 0.13 0.33 0.14
n 12 12 12 12 12 12 12 12 12 12 12 12 12 10
Minimum 13.8 0.00 24.9 29.8 9.61 5.90 1.96 2.26 0.60 0.005 0.00 0.19 0.00 11.3
Maximum 37.5 1.94 55.2 58.3 25.0 10.5 3.36 4.45 2.72 0.118 14.7 1.00 2.89 11.5
Paleosol Mean 26.4 0.21 40.6 39.9 19.5 8.69 2.67 3.59 1.43 0.030 10.4 0.41 2.46 11.4
Std. Deviation 6.09 0.59 8.06 8.32 4.36 1.16 0.38 0.62 0.47 0.026 3.67 0.19 0.71 0.070
n 16 16 16 16 16 16 16 16 16 16 16 16 16 11
Minimum 11.6 12.1 60.6 18.1 7.64 7.59 1.35 1.98 0.50 0.003 6.03 0.18 2.13
Maximum 16.7 14.7 73.6 24.9 14.6 8.53 2.30 3.05 0.70 0.010 10.3 0.29 2.65
Placic horizon Mean 14.0 13.4 68.7 21.1 10.2 8.19 1.92 2.60 0.59 0.007 8.47 0.23 2.33 10.6
Std. Deviation 2.55 1.86 7.10 3.43 3.84 0.52 0.51 0.55 0.10 0.005 2.21 0.05 0.28
n 3 2 3 3 3 3 3 3 3 2 3 3 3 1
Minimum 1.17 0.98 66.3 8.73 0.41 9.22 0.27 0.47 0.05 0.000 1.80 0.10 1.44 9.92
Maximum 11.7 62.8 90.1 24.8 8.91 10.8 1.50 3.03 0.60 0.008 10.3 0.22 2.74 10.8
Unweathered ash Mean 3.70 18.7 81.5 15.2 3.28 10.2 0.60 1.18 0.18 0.003 3.96 0.14 2.22 10.5
Std. Deviation 3.25 20.6 6.61 4.64 2.39 0.47 0.38 0.80 0.17 0.002 2.61 0.04 0.41 0.31
n 12 12 12 12 12 12 12 12 12 12 12 12 12 6

Fet Fed Feo Fep Ferrihydrite Alo + ½ Feo P retention Melanic
OM free Index
[mass %] [%] [‐]
Minimum 0.026 0.013
Maximum 0.68 0.081
Ectorganic horizons
Mean 0.16 0.038
(forest)
Std. Deviation 0.19 0.026
n 11 6
Minimum 1.38 0.39 0.38 0.38 1.08 1.13 100 1.63
Maximum 2.37 0.99 0.88 0.77 2.44 2.00 100 2.09
Current topsoils Mean 2.03 0.74 0.67 0.57 1.87 1.68 100 1.81
Std. Deviation 0.37 0.20 0.17 0.12 0.48 0.26 0.000 0.19
n 8 8 8 8 8 8 4 8
Minimum 2.16 0.25 0.097 0.18 0.20 1.32 100 1.64
Maximum 3.47 1.47 1.12 0.69 2.55 3.17 100 1.77
Current subsoils Mean 2.83 1.01 0.79 0.55 1.81 2.46 100 1.69
Std. Deviation 0.43 0.31 0.27 0.15 0.63 0.55 0.000 0.059
n 12 12 12 12 12 12 6 4
Minimum 1.87 0.26 0.19 0.11 0.44 2.11 97.4 1.50
Maximum 3.94 1.91 1.34 0.87 3.64 4.32 100 1.58
Overprinted zone Mean 3.04 1.24 0.94 0.54 2.18 3.25 99.6 1.53
Std. Deviation 0.68 0.58 0.44 0.25 1.12 0.74 0.97 0.041
n 12 12 12 12 12 12 7 3
Minimum 1.37 0.14 0.11 0.081 0.26 2.51 98.5 1.53
Maximum 4.02 2.39 2.24 1.47 5.20 5.03 100 1.65
Paleosol Mean 3.06 1.30 1.03 0.62 2.32 4.11 99.8 1.59
Std. Deviation 0.82 0.62 0.52 0.33 1.18 0.77 0.49 0.055
n 16 16 16 16 16 16 9 5
Minimum 10.3 9.54 5.92 0.84 10.9 5.73
Maximum 18.9 17.1 11.5 1.61 21.3 7.72
Placic horizon Mean 13.4 12.3 8.41 1.34 15.8 6.80
Std. Deviation 4.72 4.13 2.82 0.44 5.26 1.01
n 3 3 3 3 3 3
Minimum 2.83 0.24 0.15 0.008 0.25 0.57 31.1
Maximum 5.10 1.63 1.31 0.52 2.42 3.68 48.2
Unweathered ash Mean 3.56 0.64 0.41 0.089 0.73 1.38 39.9
Std. Deviation 0.57 0.51 0.33 0.15 0.61 0.95 8.59
n 12 11 12 12 12 12 3

Exchangeable Exchangeable Cations Base
pH KCl Acidity Al3+ Fe2+ K+ Na+ NH4+ Ca2+ Mg2+ Mn2+ CECeffective Saturation
[‐] [mmolc kg‐1] [%]
Minimum 2.41 28.5 9.67 0.280 2.94 0.56 3.22 12.9 4.52 0.17 283 7.8
Maximum 3.75 389 194 0.814 19.4 8.43 64.6 572 77.1 10.8 700 96
Ectorganic horizons
Mean 2.80 259 80.7 0.330 7.76 2.55 26.9 123 29.7 1.55 422 35
(forest)
Std. Deviation 0.44 126 60.9 0.161 5.43 2.08 20.3 164 24.8 3.36 108 30
n 11 11 11 11 11 11 11 11 11 11 11 11
Minimum 3.76 55.5 0.77 0.060 0.99 0.82 0.26 0.82 0.51 0.01 68.3 2.3
Maximum 4.32 177 16.31 0.401 3.72 1.85 5.44 6.38 3.32 0.13 184 19
Current topsoils Mean 4.18 98.0 4.56 0.182 2.43 1.40 1.52 3.52 1.70 0.04 107 10
Std. Deviation 0.18 43.2 4.89 0.118 0.87 0.38 1.63 2.06 1.00 0.04 40.9 5.8
n 8 8 8 8 8 8 8 8 8 8 8 8
Minimum 4.21 12.2 0.33 0.056 0.20 0.56 0.11 0.66 0.20 0.01 15.9 3.6
Maximum 4.61 73.8 4.43 0.172 2.61 4.33 1.48 4.06 2.43 0.10 81.0 31
Current subsoils Mean 4.42 35.0 1.08 0.095 1.05 1.22 0.36 2.21 0.87 0.03 40.3 15
Std. Deviation 0.10 18.8 1.14 0.046 0.75 1.03 0.38 1.02 0.66 0.04 19.2 7.5
n 12 12 12 12 12 12 12 12 12 12 12 12
Minimum 4.30 13.7 0.33 0.060 0.17 0.26 0.11 0.95 0.22 0.01 17.2 4
Maximum 4.50 53.7 2.08 0.140 0.64 1.10 0.69 5.62 1.83 0.10 59.5 20
Overprinted zone Mean 4.42 31.3 0.74 0.096 0.39 0.72 0.25 2.20 0.76 0.02 35.4 13
Std. Deviation 0.07 14.6 0.52 0.040 0.17 0.25 0.17 1.29 0.51 0.04 15.1 5.0
n 12 12 12 12 12 12 12 12 12 12 12 12
Minimum 4.20 7.98 0.33 0.055 0.092 0.22 0.11 1.03 0.11 0.01 11.3 4.0
Maximum 4.60 140.5 1.87 0.140 1.02 5.51 0.62 9.17 5.78 0.11 146 35
Paleosol Mean 4.42 35.0 0.52 0.096 0.42 1.23 0.17 2.85 1.11 0.02 40.6 17
Std. Deviation 0.09 30.7 0.38 0.039 0.27 1.29 0.13 1.93 1.42 0.03 31.1 9.6
n 16 16 16 16 16 16 16 16 16 16 16 16
Minimum 4.48 1.56 0.37 0.140 0.060 0.17 0.12 1.93 0.85 0.01 8.56 19
Maximum 4.75 14.6 0.45 0.246 2.50 1.71 0.15 2.32 1.03 0.01 18.1 82
Placic horizon Mean 4.62 8.07 0.41 0.193 1.28 0.94 0.14 2.12 0.94 0.01 13.3 51
Std. Deviation 0.19 9.21 0.05 0.075 1.72 1.09 0.02 0.27 0.12 0.00 6.75 44
n 2 2 2 2 2 2 2 2 2 2 2 2
Minimum 4.39 1.13 0.33 0.065 0.11 0.12 0.11 0.34 0.06 0.01 1.85 19
Maximum 4.92 9.63 1.20 0.183 3.67 9.00 0.40 1.30 0.53 0.09 14.4 98
Unweathered ash Mean 4.72 2.89 0.51 0.137 0.65 1.26 0.17 0.74 0.24 0.02 5.69 48
Std. Deviation 0.14 2.26 0.24 0.027 1.05 2.47 0.08 0.28 0.19 0.03 3.80 24
n 12 12 12 11 12 12 12 12 12 12 12 12
Dankwoord

Ecuador 2004, het RUFLE team trekt kriskras door de Andes, op zoek naar nevelbos en
páramo. We staan aan het begin van een groot project, en de sfeer is enthousiast en
verwachtingsvol. Het veldwerk is behoorlijk zwaar: het klimaat koud en nat, de bodems
glibberig en de hellingen steil. Maar de spectaculaire vergezichten en Latijns Amerikaanse
sfeer maken alles goed. Vandaag, Amsterdam 2008. Mijn promotieonderzoek is afgerond en
ik ben zowel blij dat het gelukt is als melancholisch omdat deze memorabele fase voorbij is.
Ik heb enorm genoten van elke stap.

Veel mensen hebben bijgedragen aan deze onvergetelijke ervaring en ik wil deze laatste
pagina’s gebruiken om iedereen te bedanken. Allereerst mijn promotor Koos Verstraten.
Zijn vertrouwen in mijn kunnen was zeer waardevol en dat maakt hem wat mij betreft een
echte ‘people manager’. Zijn gretigheid met betrekking tot datasets en de eindeloze
mogelijkheden die hij ziet om er papers van te maken zijn erg motiverend. Dan mijn co‐
promotor Jan Sevink. Zijn creatieve geest en overzicht hebben mijn onderzoek echt
geïnspireerd. Zijn veldkennis is onvergelijkbaar. Ik dank zowel Koos als Jan voor het snelle
becommentariëren van mijn papers, dat stimuleerde de voortgang enorm. Graag bedank ik
de overige leden van de promotiecommissie: Willem Bouten, Peter Buurman, Toine
Jongmans, Henry Hooghiemstra en Hans van der Plicht voor hun cruciale rol bij mijn
promotie, en voor het daadwerkelijk lezen van het gehele manuscript.

Dan natuurlijk de overige RUFLE teamleden. Boris Jansen was een geweldig klankbord en ik
heb veel plezier beleefd aan onze soms verhitte discussies. Marcela Moscol tiene una gran
perseverancia, ella ha pasado por momentos difíciles en Holanda sin su familia pero a pesar
de eso trabajó muy duro. Eramos un buen equipo! Klaas Nierop was altijd enthousiast over
het RUFLE project en heeft zijn specialistische kennis gul gedeeld. Henry Hooghiemstra´s
kritische blik ten opzichte van de bijdrage die bodemonderzoek zou kunnen leveren aan de
interpretatie van paleorecords heeft me gestimuleerd om het nut ervan aan te tonen.
Antoine Cleef was altijd erg betrokken en positief, zowel wat betreft mijn onderzoek als mijn
familie. De strijd der Mastodonten in het veld was mooi om mee te maken en ik zal de
queste naar de ‘vierkante fles’ niet licht vergeten.

Expreso mi gratitud al Ministerio del Ambiente del Ecuador, y en particular al Dr. Galo
Rosales, al Ing. Rómulo Aguirre y al Dr. Carlos Molina por brindarnos su apoyo en la
obtención de la autorización de investigación para conducir nuestro trabajo en el Norte del
Ecuador. Agradezco al guardaparques Miguel Cavascango por su ayuda entusiasta y a Luz
del Hostal Los Frailejones por sus excelentes habilidades culinarias. Mi reconocimiento
179
especial a Chris James y Chulpi, a Martha Muñoz y su familia, a don José Cando y a los
muchos voluntarios extranjeros de Jatun Sacha por los buenos momentos que compartimos
en el campo. Al mismo tiempo reconozco el apoyo de Susan Poats, Carla Gavilanes, David
Suárez y Paul Arellano de Randi Randi. Quisiera extender mis agradecimientos a Ecopar por
facilitar nuestra investigación brindándonos las instalaciones de sus oficinas. Fue siempre un
gran placer el trabajar con los Ecopares: gracias entre otros a Alfredo López, Nikolay Aguirre,
Luis Ordoñez, Erica Narváez, María Agusta Almeida, Luis González, Kaia Ambrose, Kevin
Cueva, Pool Segarra, Andrea Arias, Jhonny López y a los estudiantes colombianos. Mi
agradecimiento va también para Miriam y Vicenta Moreno y José Mazo por su estupenda
hospitalidad y amistad. Agradezco a Hugo Navarrete de la Pontificia Universidad Católica del
Ecuador por su colaboración. Deseo además expresar mi agradecimiento a Robert Hofstede
por transmitirme su pasión por el páramo e iniciarme al modo de vida latinoamericano.
Agradezco a María Victoria Arbeláez y a Elisa por tantas reuniones agradables. Del mismo
modo no quiero dejar de reconocer a Minard Hall y Patricia Mothes (Escuela Politécnica
Nacional), quienes compartieron conmigo su amplio conocimiento de las tefras
ecuatorianas. Agradezco asimismo a los siguientes colegas investigadores en Guandera:
Maaike Bader, Mario Marziano y Stefania Nonce por su entusiasmo.

Die rare RUFLE monsters waren erg bewerkelijk in het laboratorium, en ik wil Leo Hoitinga,
Leen de Lange, Ton van Wijk, Piet Wartenbergh, Bert de Leeuw, Niko de Wilde‐de‐Ligny,
Frans Backer en Wijnanda Koot bedanken voor hun eindeloze geduld en flexibiliteit. Zonder
al het laboratoriumwerk was dit proefschrift er niet geweest. Mijn dank gaat ook uit naar de
verschillende co‐auteurs en anonieme reviewers die bijgedragen hebben aan de papers in
dit proefschrift. Marijke Duyvendak en Stephen Vandenbijlaard van de Bibliotheek van de
Universiteit van Amsterdam wil ik hartelijk bedanken voor het snelle aanleveren van soms
onmogelijk vindbare referenties. Dankzij Charles Hogervorst heeft het bodemkundige
taalgebruik niet compleet gezegevierd over correct Engels. Marcela Moscol, Antoine Cleef,
Jan Peter Lesschen en Irina Venegas hebben meegewerkt aan de Spaanse samenvatting.
Marjo Oligschläger heeft gezorgd dat het binnenwerk er prachtig uitziet en Jan van Arkel
heeft de omslag mooi vorm gegeven. Dit onderzoek werd gesubsidieerd door WOTRO en de
Universiteit van Amsterdam, waar ik dankbaar voor ben. Het RUFLE veldwerk werd
bovendien gesponsord door Fjällraven dankzij directeur Jurg Noorman, in de vorm van
outdoor kleding en materiaal. Nu weet ik dat slecht weer echt niet bestaat als je de juiste
kleding draagt. Het veldwerk was vele malen onaangenamer geweest zonder Fjällraven.

Verder ben ik mijn verschillende kamergenoten op de UvA dankbaar voor de vele gedeelde
uurtjes. In het bijzonder natuurlijk opperkamergenoten Maartje (paranimf) en Sanneke.
Onze balspelletjes op de kamer waren hilarisch en boden de nodige afleiding. Ik heb veel
goede herinneringen aan onze gezamenlijke UvA tijd. Verder wil ik ook kamergenoten
Ingrid, Laurens en Casper bedanken voor de gezelligheid en leuke gesprekken. En natuurlijk
jaargenoten Jan Peter en Klaske voor de vele gezellige lunchwandelingen en koffiepauzes.
Het was leuk om samen te werken met Geert Jan, Gilles en Deanna in de ICG
180
promovendicommissie. Joris, Marieke, Julian, Mirjam, Mirella, Marjolein en Jerry wil ik
bedanken voor de gezellige koffiepauzes, wetenschappelijke discussies, gezamenlijke
cursussen en/of leuke veldwerken. Mirjam Vriend (†) was zeer behulpzaam en ik zal haar
missen op onze jaarlijkse studiereünie. Vele andere collega’s hebben bijgedragen aan een
prettige werksfeer, onder andere Erik, Jan, Emiel, Harry, Albert, Annemiek, Lucy, John, Joost,
Pim, Fred, Joke en Mijke. Samenwerking met studenten Ray, Niels, Richard, Natalie, Johan,
Erik en Misha was leuk en leerzaam.

Annewieke (paranimf) zit in hetzelfde schuitje maar dan op een compleet ander onderwerp
en het was heel verfrissend om van gedachten te wisselen. Bedankt voor de
scherpzinnigheid en betrokkenheid. Elisabeth en Miriam bedankt voor de gezellige avondjes
met kaasfondue. De studiedames Marieke, Maartje, Inka, Riekje en Gertruud: wat is het
bijzonder dat we elkaar nog steeds zo vaak zien, bedankt voor de vriendschap. Ellen wil ik
graag bedanken voor de heldere analyse van mijn psyche van tijd tot tijd. Kasper en Roos
voor de gedeelde passie voor goede koffie en lekker eten. Matthijs, El bandalo, herinnert
me gelukkig af en toe aan de noodzaak tot feesten. Mede fysisch geograaf Hiddo wil ik
bedanken voor de leuke discussies.

Tot slot wil ik natuurlijk mijn familie en schoonfamilie bedanken hoewel dat bijna
onmogelijk is. Wat fijn dat ik uit zo’n warm nest kom. Mijn ouders wil ik bedanken voor hun
onvoorwaardelijke vertrouwen, dat vormt een geweldige en onvervangbare basis. Ze
hebben hun avontuurlijke en eigenwijze dochter altijd gesteund met raad en daad en dat
was vast niet altijd even gemakkelijk. Mijn zus Sanne wil ik bedanken voor de leuke en
inspirerende gesprekken en Erwin voor zijn analytische blik. Het was heel bijzonder om mijn
familie in Ecuador te mogen rondleiden. Schoonouders Joke en Henk bedankt voor de
oprechte interesse en nieuwsgierigheid. Katja, Mel, Gosse, Hanno en Perle bedankt voor de
vrolijke noot. Opa Tonneijck, Oma Kim en Oma de Rooy wil ik bedanken voor het meeleven
op afstand. Mijn lieve zoon Karel heeft bijgedragen door meestal heel goed te slapen ’s
nachts en natuurlijk door zijn relativerende vrolijkheid en onbezorgdheid. De verschillende
oppassen en de crèchedames wil ik bedanken voor de goede zorgen. En dan natuurlijk
Kasper, mijn partner in crime en altijd in voor avontuur, die mij vaak beter begrijpt dan ik
mezelf begrijp. Dit boekje heb ik aan hem opgedragen, mijn dank is groot.

181
182
Cover
The cover contains a fragment of a Tigua painting (private collection) by
Juan Luis Cuyo, showing the intricate links between nature and humanity
in the Andes. Naturally we have tried to contact Juan Luis Cuyo
concerning the copyright of his painting, but unfortunately we did not
succeed. Since this thesis is not for commercial purposes we feel we have
made fair use of it and we hope that he is pleased that we brought it to
the attention of the scientific community.

La portada contiene un fragmento de una pintura de Tigua (colección
privada), obra de Juan Luis Cuyo, en la que se muestra los intrincados
lazos entre la naturaleza y el ser humano en Los Andes. Naturalmente
intentamos contactar a Juan Luis Cuyo con respecto a los derechos de
autor de su obra, pero desafortunadamente no tuvimos éxito. Como esta
tesis no tiene fines comerciales pensamos haber hecho un uso justo de
ella y esperamos que él esté satisfecho de que la hayamos puesto en
conocimiento de la comunidad científica.

Photographs accompanying chapters, all made in Guandera
Introduction Close up of cloudforest p 6
1 View of upper forest line p 10
2 Irregular upper forest line p 18
3 Camping on top of the world p 38
4 Forest patch G5a (middle of photo) p 66
5 Cloudforest p 88
6 Site G6: the RUFLE team in action p 108
Synthesis Agricultural lands, cloudforest in the p 128

back
Summary Paramo, forest patch, cloudforest and p 150
agricultural lands
Dankwoord Boris Jansen, Marcela Moscol Olivera p 178
and Femke Tonneijck

FSC
Products carrying the Forest Stewardship Council (FSC) label are

independently certified to assure consumers that they come from forests
that are managed to meet the social, economic and ecological needs of
present and future generations. This paper is totally chlorine free.

Thesis

Uploaded by

Copyright:

Available Formats

You might also like

Thesis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis

Uploaded by

Copyright:

Available Formats

UvA-DARE (Digital Academic Repository)

Citation for published version (APA):

UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

Download date: 30 Oct 2017

Graag nodig ik u uit voor het bijwonen van

VOLCANIC ASH SOILS IN ANDEAN ECOSYSTEMS

UNRAVELLING ORGANIC MATTER

Vrijdag 6 februari 2009 10.00 in de Agnietenkapel

Low forest 3600 .

Soil type (FAO) Altitude Current vegetation Coordinatesa Cordillera Site

Site Horizona Depth b.d.b Ctotal pHCaCl2 Rootsc

Site Horizon Sample Sand fraction Silt fraction

3BC / G11 F170

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Forest Páramo Topsoil Subsoil

Horizona Depth Organic C Dry bulk density Soil pHCaCl2 TiO2 Al2O3

Shoot litter: root litter R [‐] Fixed 1:0.35 Hofstede & Rossenaar (1995)

Kernel width, bioturbation B widthB [cm] 12 0 – 200

Depth of DB x B [cm] 60 60 – 80 Tonneijck & Jongmans 2008

Profile number G5a Elevation (m asl) 3697 ± 9 (n=9) Topography Hilly

Synthesis Agricultural lands, cloudforest in the p 128

Products carrying the Forest Stewardship Council (FSC) label are

You might also like