Soil & Tillage Research 180 (2018) 250–258

Contents lists available at ScienceDirect

Soil & Tillage Research

journal homepage: www.elsevier.com/locate/still

Soil quality and tree status in a twelve-year-old apple orchard under three T
mulch-based floor management systems
⁎
Wiktor Rafał Żelaznya, , Maria Licznar-Małańczukb
a
Division of Crop Management Systems, Crop Research Institute, Drnovská 507/73, 161 06 Praha 6 – Ruzyně, Czech Republic
b
Department of Horticulture, Wrocław University of Environmental and Life Sciences, Grunwaldzki 24A, 50-363 Wrocław, Poland

A R T I C LE I N FO A B S T R A C T

Keywords: Interest in abandoning herbicide fallow as the standard in-the-row orchard floor management system continues
Living mulches unabated. Despite research efforts, available relevant data remain insufficient to formulate reliable re-
Yield commendations for individual site conditions. A long-term experiment was therefore initiated in a temperate
Compositional data analysis climate area in south-western Poland. ‘Ligol’ and ‘Pinova’ cultivar apple trees were planted in an unirrigated
orchard in 2004, with treatment plot tree-rows mulched and control plots maintained with herbicide fallow. In
2016, black woven polypropylene fabric and Agrostis vulgaris With. and Festuca ovina L. living mulches were
compared with the herbicide fallow in terms of their effect on soil properties, tree nutrient status and yields.
While the living mulches had a positive influence on soil porosity, humus content and pH, there were substantial
yield reductions; arguably due to competition for water. It is therefore important that more intensive methods of
understory vegetation suppression are explored and more competition-resilient rootstocks sought to counteract
yield loss. The polypropylene cover was primarily associated with a decreased leaf K:Ca+Mg ratio. A synthetic
mulch is a viable choice for certain rain-fed orchards.

1. Introduction
The current standard orchard floor management system in tempe-
rate climates comprises vegetation-free herbicide strips in tree rows and
periodically mown grass cover in drive alleys (Merwin, 2003). While
the approach proved both effective and inexpensive (Lipecki and
Berbeć, 1997), the widespread trend of reducing synthetic pesticides in
plant crops has triggered interest in alternative systems (Hogue et al.,
2010; Yao et al., 2005). One of the promising alternatives is replacing
herbicide strips with various mulches to control weeds in fruit tree
rows. In addition to addressing consumer concerns, mulch-based
orchard floor management systems may contribute to soil conservation
by protecting soil from erosion and improving its biological activity and
water regime (Granatstein and Sánchez, 2009; Lisek, 2014; Tahir et al.,
2015).
Living mulch, i.e. mulch with living plants, has the potential to
reduce nutrient leaching and sequester carbon and nitrogen. However,
the plants can compete with fruit trees for water and nutrients, thus
impairing growth and yield (Granatstein and Sánchez, 2009; Tahir
et al., 2015). Hammermeister (2016) recommended the use of living
mulches only at sites with fertile soils, sufficient water supply and
lacking perennial weed species.
Although the competition issue does not arise in mulching with
dead material, this option involves substantial purchase and labor costs,
which are particularly high for covers requiring frequent renewal
(Lisek, 2014; Tahir et al., 2015). While this problem can be reduced
with a durable synthetic mulch which provides satisfactory weed con-
trol, they also have limitations, including low water permeability, in-
creased summer soil temperatures beyond tree-root tolerance and lack
of sustainability (Granatstein and Sánchez, 2009; Lipecki and Berbeć,
1997).
The published literature on orchard mulch application is sparse and
insufficient to guide satisfactory synthesis. Despite the availability of
recent reviews (Hammermeister, 2016; Lisek, 2014), research still lags
continual progress in fruit-growing technology. Simultaneously, ap-
plicability of early experimental results, such as those employing less
than 400 trees planted per hectare and control treatments using cur-
rently dismissed herbicides (e.g. Glenn et al., 1987; Miller, 1983;
Sanchez et al., 2003), becomes questionable.
Another problem has been insufficient number of long-term studies
(Atucha et al., 2011a). The following two exceptions can be noted: (1)
556 apple trees per hectare planted in the Pacific Northwest (Atucha
et al., 2011a,b; Oliveira and Merwin, 2001; Yao et al., 2005, 2009) and
(2) a Midwestern United States experiment (Sanchez et al., 2003).

⁎
Corresponding author.
E-mail addresses: wzelazny@vurv.cz (W.R. Żelazny), maria.licznar-malanczuk@upwr.edu.pl (M. Licznar-Małańczuk).

https://doi.org/10.1016/j.still.2018.03.010
Received 1 August 2017; Received in revised form 12 March 2018; Accepted 14 March 2018
Available online 27 March 2018
0167-1987/ © 2018 Elsevier B.V. All rights reserved.
W.R. Żelazny, M. Licznar-Małańczuk Soil & Tillage Research 180 (2018) 250–258

However, the latter was conducted in a tart cherry orchard and em- solely on the herbicide, PP, Festuca and Agrostis floor management
ployed simazine in the control treatment, thus hindering comparison systems.
with other published experiments, based on apple trees and glyphosate. Each main plot was divided into three subplots with five trees and
Mulches are a major research subject at the Wrocław University of allocated to different rootstocks. The study was eventually limited to
Environmental and Life Sciences (Poland); where the longest running the P 2 rootstock because it was the only one sufficiently vigorous to
experiment has been conducted in rain-fed rather than irrigated apple enable trees to compete with the living mulches. Like all rootstocks in
trees. We present the results which describe the orchard state after 12 the P-series, P 2 has been bred in Poland. It is a high-yielding rootstock
years of continuous mulching. Our paper provides new primary data related to M.9, exhibiting similar vigour, but better adapted to the
investigating the validity of theories on the long-term influence of Polish climate (Mantinger, 1996). In comparison, the excluded root-
mulches on apple orchard soil properties and estimates of living mulch stocks, P 16 and P 22, provide a more dwarfing effect and are deficient
effects on mature trees in the absence of irrigation. in terms of, respectively, cold-hardiness and soil requirements
(Mantinger, 1996; Szczygieł and Czynczyk, 2002).
2. Material and methods The trees were trained into a slender spindle and the orchard was
fertilised with an average dose of 50 kg N ha−1 yr−1 as ammonium ni-
2.1. Site description and experimental design trate or urea. Tree protection followed current recommendations for
commercial growers. Periodically mown sod was maintained in the
The study was set up at the Fruit Experimental Station in Samotwór drive alleys. The orchard was unirrigated.
(51°6′ N, 16°50′ E), managed by the Wrocław University of
Environmental and Life Sciences (Poland). The orchard lies in a tem- 2.2. Data collection
perate climatic zone on Haplic Luvisol with light loam texture.
According to the Agri4cast Resources Portal (Biavetti et al., 2014), for At the end of April 2016, ‘A horizon’ topsoil samples were collected
the duration of the experiment annual mean temperature in the area from alternate pseudo-replications (Fig. 1). From each plot, four core
ranged from 8.1 to 11.0 °C and annual precipitation sum amounted to samples were randomly collected close to where the three internal tree
392–738 mm (Table 1). trunks stood. Herbaceous vegetation including main root biomass was
The orchard was established in spring 2004 with modified split-plot removed from the sampling points, and top 5 cm of exposed soil was
design. The main-plot factor was an orchard floor management system, collected. This provided a total of 64 core samples.
and rootstock supplied the subplot level. The design included two The sample water content, water content after full saturation
blocks (replications) divided into four pseudo-replication sections (hereafter, WCAFS) and bulk density were determined at the Crop
(Fig. 1). These contained alternating planted ‘Ligol’ and ‘Pinova’ apple Research Institute, Czech Republic, where all laboratory analysis was
tree cultivars separated by ‘Idared’ pollinator lines. Planting was from performed. The bulk density values were converted to porosity with
one-year-old whip-quality nursery stock with 2380 trees per hectare uniform 2.583 Mg m−3 particle density assumed because of low soil
(3.5 × 1.2 m). organic content. This value was obtained using the liquid pycnometer
In spring 2004, floor management systems were incorporated in 1 m method after combining the soil from all samples. The volume of non-
wide strips in the tree rows. Treatments included: control herbicide saturable pores (hereafter, VnSP) was calculated as the difference be-
fallow maintained with two or three annual applications of mixed tween porosity and WCAFS. Some cores were damaged in transporta-
glyphosate (4 L ha−1) and 2-methyl-4-chlorophenoxyacetic acid tion and subsequent handling, thus reducing the final number of ob-
(2 L ha−1), black woven polypropylene fabric (AGRO 84F-170 servations to 58.
TKANINA PP, 94 g m−1; hereafter, PP) and Tropaeolum majus L. and A composite loose soil sample was also collected from each plot by
Agrostis vulgaris With. living mulches. Tropaeolum majus L. was replaced combining six single samples from beside the internal tree bases.
by Festuca ovina L. (hereafter, Festuca) in the second year because of Approximately 20 cm exposed soil was collected by soil auger after
poor performance. The living mulches were maintained by mowing removing plant biomass, but slightly shallower sampling was occa-
with a string trimmer. Although the original design also included sionally necessary to avoid subsoil portions. The total number of sam-
Tagetes patula L. and Trifolium repens L. living mulches, these produced ples was 32, with half collected in April 2016 and the remainder at the
retarded tree growth and weed infested covers, so the study focused beginning of August.
Mehlich-3 extractable P, K, Ca and Mg contents; humus, organic C
Table 1 and total N contents; as well as pH in water and 1 mol KCl L−1 were
Weather descriptors characterizing each year of the experiment. Based on the determined after drying samples and sieving them through 2-mm mesh,
data from the Agri4cast Resources Portal (Biavetti et al., 2014), cell 103128 The C:N ratios were calculated from their relative contents, and four
(51°5′ N, 16°40′ E). runs of herbicide and Festuca aggregate stability assessment (Kemper
Year Mean temperature (°C) Precipitation sum (mm) and Koch, 1966) were performed. These provided percentages of stable
aggregates (SAS).
2004 9.3 449
Two internal trees per experimental plot were randomly chosen for
2005 9.1 552
2006 9.6 622
leaf chlorophyll content measurement and leaf sample collection in
2007 10.2 569 August 2016. Three trees were missing in one pseudo-replication and
2008 10.2 470 additional two in the Festuca treatment, so the pseudo-replication and
2009 9.4 738 the treatment were excluded from analysis. Chlorophyll concentration
2010 8.1 702
was determined by CCM-300 device (ADC BioScientific Ltd.,
2011 10.0 524
2012 9.5 522 Hoddesdon, United Kingdom). This device exploits the relationship
2013 9.3 658 between chlorophyll content per unit leaf surface area and fluorescence
2014 11.0 580 response ratio of 735 and 700–710 nm bands. Conversion is then based
2015 10.9 392 on linear regression (Gitelson et al., 1999). Measurements were taken
2016 10.2 617
on two or four large, undamaged leaves found in middle sections of

251
W.R. Żelazny, M. Licznar-Małańczuk Soil & Tillage Research 180 (2018) 250–258

Fig. 1. Experimental design at orchard (left-hand side) and plot

(right-hand side) scale. Division of the orchard into blocks and
sections (pseudo-replications) planted with the ‘Ligol’ and ‘Pinova’
cultivars is marked with frames. Allocation of the orchard floor
management systems to the rows of the trees is indicated at the
bottom. Allocation of the rootstocks to the subplots of the mag-
nified plot is depicted on the right. Trees in the subplots covered
in the study are marked black. The living mulches, omitted root-
stocks and the lines of the ‘Idared’ pollinators are marked gray.
Pseudo-replications from which soil samples were collected in
April 2016 are connected with a dashed line.

extension shoots on each side of the crowns. This provided 192 mea-
surements. A total of 42 leaf composite samples, each consisting of 12
leaves per side of a crown were derived. These were weighted in the
orchard and transported to the laboratory for dry matter content and N,
P, K, Mg and Ca concentrations after wet combustion determination.
Apples were picked each autumn, with those from severely da-
maged trees ignored, and the fruit yield for the even number of twelve
years (2005–2016) aggregated by summation to minimise the effect of
alternate-year fruit bearing. In this way, 32 observations were obtained.
Dead fruit trees were also recorded in 2016.
2.3. Statistical data analysis
A substantial share of the obtained measurements were composi-
tional (Table 2); with values bound between 0% and 100% of total
sample mass or volume. Many statistical tools, especially parametric
linear models, are unsuitable for analysing raw compositional data.
Analysis can be performed on secondary variables, obtained by using
log ratio based transformations (Pawlowsky-Glahn and Egozcue, 2006).
Each set of compositional variables was extended by a filling value
(fv) variable—the difference between the sum of measured values and
the total sample mass or volume. The three > 2-part compositions were
subjected to isometric log ratio (ilr) transformations based on sequen-
tial partitioning (Egozcue and Pawlowsky-Glahn, 2005) shown in
Fig. 2.
A simple or generalised mixed-effect linear model (Bates et al.,
2015) was fitted for each transformed dependent variable. The orchard
floor management system was the only fixed predictor variable in the
models, and each random model part included a row nested in a block
and crossed with a section nested in a cultivar effects. There were ad-
ditional random terms for the following, dependent on the data origin:
(1) a sample batch for loose soil, (2) a plot nested in a sample batch for
SAS assessment and (3) a tree nested in a plot and crossed with the
effect of crown side for chlorophyll measurement. Simple models were

Table 2
Dependent variables according to their origin and scale of measure. For each scale, a transformation that was applied prior to linear modelling or a GLMM
distribution and link are provided.
Data origin Variable scale of measure

2-part composition (Gaussian + logit) > 2-part composition Continuous ratio Continuous log (none) Binary (binomial + logit)
(ilr transformation) (Gaussian + log)

Soil cores [Water content, fv] [WCAFS, VnSP, fv]

Loose soil [Humus content, fv], [SAS, fv] [N, P, K, Ca, Mg, fv] C:N pHH2O, pHKCl
Leaves [Dry matter content, fv] [N, P, K, Ca, Mg, fv] Chlorophyll contenta
Other Fruit 2005–2016 yield Tree mortality

fv = filling value, WCAFS = water content after full saturation, VnSP = volume of non-saturable pores, SAS = soil aggregates stability.
a
A logarithmic transformation and a simple mixed-effects model.

252
W.R. Żelazny, M. Licznar-Małańczuk Soil & Tillage Research 180 (2018) 250–258

Table 3
Characteristics of soil and leaf samples collected from the herbicide treatment.
Variable n Median Range

3 −3
Soil water content [m m ] 14 0.23 0.18–0.26
porosity [m3 m−3] 14 0.42 0.38–0.44
WCAFS [m3 m−3] 14 0.33 0.27–0.38
aggregates stability [%] 32 47 15–53
humus [%] 8 1.05 0.86–1.13
total N [g kg−1] 8 0.8 0.7–0.9
extractable P [mg kg−1] 8 114 96–131
K [mg kg−1] 8 123 107–166
Ca [mg kg−1] 8 995 766–2130
Mg [mg kg−1] 8 88 57–94
K:Mg 8 1.5 1.3–2.3
C:N 8 7.5 7.2–8.7
pHH2O 8 6.4 6.2–7.4
pHKCl 8 6.2 5.8–7.2
Leaf chlorophyll [mg m−2] 64 566 376–699
dry matter [%] 14 60 56–63
N [%] 14 2.28 1.78–2.46
P [%] 14 0.16 0.13–0.20
K [%] 14 1.23 0.90–1.77
Ca [%] 14 1.31 0.72–1.84
Mg [%] 14 0.20 0.14–0.36

WCAFS = water content after full saturation.

results with findings in studies ignoring the compositional character of

the investigated variables, and to relate outcomes to the critical values
published in official fertilisation recommendations for fruit growers.
All computations were programmed in the R language (R Core
Team, 2017).

3. Results

Fig. 2. Isometric log ratio balances used in sequential binary partitioning of
dependent variables related to soil physical properties (a) and the soil and leaf
nutrient contents (b). fv = filling value, WCAFS = water content after full sa-
turation, VnSP = volume of non-saturable pores.
applied to ilr coordinates and pH values. Generalised mixed-effect
linear models (Bolker et al., 2009) with appropriate distributions and
link functions were fitted to the remaining dependent variables
(Table 2). A logit link was adopted for 2-part compositions (Filzmoser
et al., 2009). A simple model with a logarithmic transformation was
required for chlorophyll measurements because of convergence pro-
blems. Model performance was assessed by visual analysis of residual
distributions.
Mean differences and ratios between herbicide and individual
mulch treatments were estimated and expressed as 95% confidence
intervals, and simple simulations facilitated the interpretation of ilr-
based estimates: A set of 50 plausible treatment–herbicide ilr differ-
ences was randomly generated for each non-herbicide treatment. These
were drawn from a normal distribution centered at the difference point
estimate and with standard deviation equal to the estimation standard
error. The generated ilr effects were added to mean reference ilr values,
i.e. values associated with the control treatment. The sums were then
back-transformed into the compositional space and their values sub-
tracted from back-transformed reference values to obtain raw effects.
For each simulated data point, the degree of support by collected data
was determined by calculation of its Mahalanobis distance from the
centre of the generating distribution. In addition to aiding interpreta-
tion, the back-transformed values enabled us to compare experimental
3.1. Control treatment characteristics
Characteristics of the herbicide control treatment are shown in
Table 3. In contrast to the Soil Profile Analytical Database (Panagos,
2006) average composition levels for European Luvisols managed under
crop agriculture or horticulture, our experimental orchard has soil with
low porosity and humus content; with the latter contributing to low
C:N. The soil is rich in extractable P and Mg and it has a correct K:Mg
ratio according to current fertilisation norms (Sobiczewski, 2015). The
pH levels approximate apple tree requirements (Barden and Neilsen,
2003). P and K leaf concentrations are mostly in the optimum range,
with a number of samples having low N levels or Mg deficits
(Sobiczewski, 2015). Cumulated ‘Ligol’ cultivar apple yields were al-
most 300 t ha−1 compared to the 230 t ha−1 for ‘Pinova’ trees (Fig. 3).
These values are equivalent to 25 and 19 t ha−1 average annual crop.
The low yields can be attributed to the quality of the planting stock,
alternate fruit bearing and the 2007 and 2011 spring frosts, which
coincided with tree bloom. Finally, our orchard maintained high tree
survival rate, with only 5% loss throughout the experiment.
3.2. Mulch effects on soil
Statistical analysis confirmed that covers have a noticeable effect on
soil physical properties. The plots maintained with living mulches had
higher soil porosity than the herbicide treatment (Table 4). The points
in the raw effect simulation plot (Fig. 4) are clustered in the north-
eastern quadrant. This indicates increased volume of both non-capillary
(VnSP) and capillary (WCAFS) pores, resulting in an overall porosity

253
W.R. Żelazny, M. Licznar-Małańczuk
Fig. 3. Cumulated yields according to apple tree cultivar (‘Pinova’ marked gray) and floor management system.
Table 4
Point and interval estimates of mean differences for soil properties between mulch treatments and herbicide fallow
based on data obtained from soil core samples.
Contrast n ilrporosity
PP–herbicide 27 −0.01 [−0.18, 0.17]
Festuca–herbicide 29 0.26 [0.09, 0.43]
Agrostis–herbicide 30 0.28 [0.11, 0.44]
Fig. 4. Simulated mean differences in pore volumes between each mulch
treatment and herbicide fallow with a typical soil from the herbicide treatment
(0.41 m3 m−3 porosity, 0.33 m3 m−3 water content after full saturation) used as
the reference. The contour levels pertain to total porosity difference in cubic
metres per cubic metre. The strength of a given mulch effect for each dimension
(along x axis, along y axis or along the contour gradient) in relation to herbicide
fallow is indicated by the distance of the estimate from the (0, 0) point. The
point sizes are proportional to Mahalanobis distance from the estimate most
supported by data. The plausibility of the simulated estimates to represent the
true population mean decreases with increasing point size.
increase of several percentage points (pp). Measured water content
values revealed the possibility of soil-drying under both polypropylene
and living mulch covers. For instance, soil managed with herbicide
fallow and 0.23 m3 m−3 water content (Table 3) was estimated to have
0.02 m3 m−3 [0.00, 0.03] more moisture than the same soil covered
with PP (Fig. 5(a)). Some evidence of improvement in aggregate sta-
bility was obtained for the Festuca treatment, with estimated 1.38-fold
[0.80, 2.40] (n = 64) increase in stable-to-unstable aggregate ratio. In
raw terms, this translates to an effect of several pp (Fig. 5b).
The presence of living mulches contributed substantially to humus
accumulation (Table 5). Dependent on reference content, the estimated
254
Soil & Tillage Research 180 (2018) 250–258
ilrcapillary Logit water content
0.12 [−0.13, 0.37] −0.10 [−0.18, −0.01]
−0.21 [−0.45, 0.03] −0.08 [−0.16, 0.00]
−0.13 [−0.36, 0.11] −0.11 [−0.20, −0.03]
Agrostis raw effect ranged from approximately 0.4 to almost 1.0 pp, and
the Festuca effect was similar (Fig. 5c). Difference in carbon seques-
tration were reflected in C:N ratio. All of the investigated soil covers
were found to mitigate soil acidification, with the PP treatment effect
possibly exceeding half a pH unit, which is a substantial difference in
view of liming needs (Sobiczewski, 2015). The covers also had a posi-
tive influence on total soil nutrient contents (Table 6), and the effect of
the living mulches was particularly strong, with estimates ranging from
0.5 to beyond 1.5 g kg−1 (Fig. 6a). While for the living mulches an
upward soil N:P balance shift occurred with increased total N con-
centration (Fig. 6b), the opposite was observed in the PP treatment,
where P content increased by approximately 20 mg kg−1 and the effect
on N was weak.
3.3. Mulch effects on apple tree status
The collected data provided no convincing evidence on mulching
influence on leaf dry matter content (Table 7). Trees in the mulched
plots had approximately 10% less chlorophyll than those in the herbi-
cide treatment, thus suggesting slight chlorosis. In assessment of the
influence of soil mulching on leaf nutrient profile, although the
ilrnutrients estimates suggested increased overall nutrient contents, this
could not be confirmed in the simulation (Table 8 vs. Fig. 7a). PP was
associated with a K:Ca+Mg balance shift towards a higher proportion
of the bivalent cations, whose concentration increased by approxi-
mately 0.2 pp (Fig. 7b).
Perhaps the most striking finding was the enormous yield reduction
with the Agrostis mulch. This negative effect occurred despite the im-
proved soil conditions and might amount to at least half of the control
treatment yield (Table 7, cf. Fig. 3). The yield loss point estimate in the
Festuca treatment was approximately 24%, and this is also far from
trivial. Inconclusive evidence was obtained for the influence of the
synthetic mulch on yield and for the relationship between the in-
vestigated floor management systems and fruit tree mortality.
3.4. Performance of statistical models
The residual diagnostics revealed violations of homoskedasticity in
W.R. Żelazny, M. Licznar-Małańczuk
Fig. 5. Raw effect sizes of mulches on soil properties associated with 2-part
compositions in relation to herbicide fallow across a range of reference values.
The strength of a given mulch effect on each property is indicated by the dis-
tance from the zero value along the y axis. Error bars depict 95% confidence
intervals.
255
Soil & Tillage Research 180 (2018) 250–258
soil macronutrient contents, and the high complexity of experimental
design negated application of more robust models to eliminate this
problem. Estimates of these variables are therefore considered less
precise than the cited values suggest. Although some model assumption
violations were also detected in leaf nutrient modelling, these were less
severe.
4. Discussion
4.1. Mulches and soil quality
An important factor in fruit cultivation is the fact that orchards must
be maintained for multiple years with consequent delay in investment
returns. Therefore, sustaining a healthy agroecosystem with high soil
quality is more crucial in fruit growing than in crop cultivation.
Organic matter content is a major indicator of soil health because it
determines many other soil properties (Magdoff, 2001). Living mulches
contributed to increased humus content of approximately 0.5 pp over
our 12 year experiment. This is similar to the Festuca rubra cover effects
observed in the Pacific Northwest by Atucha et al. (2011a), who
achieved 0.56–0.72 pp increase after 13 years of mulch maintenance
under assumed 70–90% humus content in soil organic matter. In
agreement with the increased total soil nutrient contents under our
living mulches, their other results highlighted that additional nitrogen
inputs from understory plant residues contributed to sustaining the
total N balance above zero (Atucha et al., 2011b). Tree nutrient status
in under-fertilised orchards can improve with additional soil N. On the
other hand, excess N can impair tree growth and development and
contribute to water contamination (Atucha et al., 2011b). The increased
C:N ratio observed in our study suggests that additional N was fixed in
the soil and posed no threat to agroecosystem health.
Relative increase in extractable phosphorus in soil covered with
black woven polypropylene fabric was not observed by other authors.
For example, Merwin et al. (1995) reported a negative effect associated
with Belton-Sarlon Weed-Mat when discussing two experiments with
synthetic mulches. This discrepancy may be due to differing experiment
lengths. The cited authors based their results on soil samples collected
after only three years of fabric maintenance, compared with the twelve
years in our study, where the cover became overgrown with a dense
layer of weeds and began to resemble another living mulch. Orchard
understory vegetation, in turn, can contribute to soil P accumulation, as
demonstrated by Miller and Glenn (1985) and observed in our living
mulch treatments.
All mulches investigated in the study were associated with increased
pH levels. For PP, the effect amounted to 0.53 units [0.28, 0.77]. This
order of pH increase on soils prone to acidification in a humid climate
can result in reduced liming needs (Sobiczewski, 2015). An effect of
similar magnitude was reported by Glenn et al. (1987) after three years
of Festuca-based sod maintenance. These authors attributed the effect to
improved buffering capacity of soil covered with living mulch. Other
possible explanations include deprotonation of H+ ions (Oliveira et al.,
2016) and the lack of acidification by synthetic herbicides (Lipecki and
Berbeć, 1997). However, the positive effect of permanent plant cover
on pH does not appear universal (e.g. Atucha et al., 2011a), and there
may be strong interaction between local conditions and living mulch
effects on this soil characteristic.
In addition to chemical properties, soil quality is determined by tilth
and structure (Magdoff, 2001). Whereas our living mulches were as-
sociated with improved porosity, Oliveira and Merwin (2001) reported
the opposite effect, and attributed it to heavy mowing equipment,
which compacted the soil during repeated operation.
4.2. Living mulch competitiveness
High soil quality does not ensure improved cultivated tree growth
conditions when limiting factors are present. Competition between fruit
W.R. Żelazny, M. Licznar-Małańczuk
Table 5
Point and interval estimates of mean differences for non-ilr variables between mulch treatments and herbicide fallow based on data obtained from loose soil samples
(n = 16).
Contrast Logit humus content C:Na
PP–herbicide 0.11 [0.00, 0.22] 1.07 [1.00, 1.14]
Festuca–herbicide 0.50 [0.40, 0.61] 1.11 [1.04, 1.19]
Agrostis–herbicide 0.47 [0.36, 0.58] 1.10 [1.03, 1.18]
a
Response ratio.
Table 6
Point and interval estimates of mean differences for ilr coordinates between mulch treatments and herbicide fallow based on data obtained from loose soil samples
(n = 16).
Contrast ilrnutrients ilrions
PP–herbicide 0.12 [0.01, 0.23] −0.02 [−0.30, 0.25]
Festuca–herbicide 0.25 [0.15, 0.36] −0.03 [−0.31, 0.24]
Agrostis–herbicide 0.22 [0.12, 0.33] 0.02 [−0.26, 0.29]
trees and understory vegetation has seriously hindered living mulch
adoption as a floor management system in commercial orchards
(Granatstein and Sánchez, 2009). Apple trees are particularly vulner-
able to competition because of their sparse roots (Merwin, 2003), so
appropriate living mulch management is crucial for satisfactory results
with this system. It must ensure balance between soil conservation,
weed suppression and control of competition pressure from the cover
crop itself (Bond and Grundy, 2001; Granatstein and Sánchez, 2009).
Experimentally tested approaches to living mulch maintenance
currently provide insufficient suppression of competitive pressure from
understory vegetation. This study determined 46% [7, 69] yield loss in
the Agrostis treatment compared to the control. Older apple tree re-
search of Måge (1982) and Neilsen and Hogue (1992) recorded similar
yield reductions. Negative impacts of living mulches are also substantial
in modern orchards: for example, Atucha et al. (2011a) reported 26%
cumulated yield loss after twelve-year Festuca rubra sod presence and
Hogue et al. (2010) recorded between 11% and 24% decline in five
years, dependent on living mulch species. Although according to
Atucha et al. (2011a) fruit trees growing in living mulches perform
poorly only in the initial growth period, after which they adapt to the
understory competition, we found no evidence of adaptation. Yields
obtained from mulched plots continued to diverge compared to the
herbicide treatment even in the latest experiment years (Fig. 3).
Yield declines can be linked to insufficient water uptake by fruit
trees grown in the presence of living mulches. Merwin et al. (1994)
reported reduction in soil water availability in their mown plant covers
experiment and Miller and Glenn (1985) identified water deficit as the
primary factor limiting growth of their fruit trees. Similarly, this study
was performed in a rain-fed orchard and it provides some observations
of negative influence of living mulches on soil water content. However,
the evidence is not representative of the whole vegetation season, be-
cause it is based on a single data collection campaign. A more extensive
assessment was performed in the wet 2010 year, when there was
702 mm precipitation, but this provided no evidence of a strong effect
of mulches on soil water content. In contrast, there was 622 mm in
2006, and the mulched soil in all except a few observations was much
drier than that in the herbicide treatment throughout the vegetation
season (Licznar-Małańczuk, 2012).
256
Soil & Tillage Research 180 (2018) 250–258
pHH2O pHKCl
0.53 [0.28, 0.77] 0.52 [0.27, 0.77]
0.39 [0.14, 0.63] 0.44 [0.19, 0.69]
0.39 [0.14, 0.63] 0.36 [0.11, 0.61]
ilrN:P ilrK:Ca+Mg ilrCa:Mg
−0.11 [−0.32, 0.11] −0.18 [−0.49, 0.13] 0.24 [−0.53, 1.01]
0.20 [−0.01, 0.41] −0.12 [−0.43, 0.19] 0.13 [−0.63, 0.90]
0.17 [−0.04, 0.38] −0.12 [−0.43, 0.19] 0.14 [−0.63, 0.91]
Merwin (2003) considered irrigation an effective method of ad-
dressing living mulch competition. Accordingly, the negative effect
observed by Merwin et al. (1994) contrasted with the results obtained
by Atucha et al. (2011b) in the same area, but with irrigation in place.
Yao et al. (2009) noted that living mulch inhibits fruit tree root de-
velopment and modifies their distribution in a way that restricts their
access to surface moisture. Therefore, irrigation may be necessary even
in areas with frequent precipitation (Merwin, 2003). In addition to ir-
rigation, various strategies for combating competition between weeds,
living mulch and fruit trees have been proposed and experimentally
evaluated. Living mulch can be suppressed by intensive mowing
(Hammermeister, 2016), chemical means (Merwin et al., 1994) or
combining with another orchard floor management system (Merwin,
2003; Schmid et al., 2004). Largely unexplored possibilities exist in
screening cover plant species and fruit-tree rootstocks for suitable traits
(Granatstein and Sánchez, 2009). It may therefore be possible to find a
combination of management strategies that could be recommended for
orchards in the temperate climates; even without irrigation.
4.3. Tree status in the presence of a synthetic mulch
The issue of competition between understory vegetation and fruit
trees has only marginal significance for synthetic mulches. In theory,
this makes it easier to promote appropriate growth conditions for fruit
trees. Måge (1982) recorded that soil water tension, dubbed ‘soil
moisture’, established under a polypropylene foil in a Norwegian
orchard was seven times lower than in soil where foliar herbicides were
applied, and the summation of third and fourth year yields was 30%
higher. Early assessments by Licznar-Małańczuk (2012), and also the
2016 work, established that PP did not have strong positive effect on
soil water balance. This was most likely due to its non-film structure.
Merwin et al. (1995) assessed different synthetic mulch brands and
obtained the best yields with film foils and less with woven mulches.
Our study highlighted that PP was associated with a relative in-
crease in leaf calcium and magnesium at the expense of potassium.
Reduced leaf K levels is a common phenomenon in high-yielding apple
trees, as the nutrient is exported to developing fruit (Neilsen and
Neilsen, 2003). Since potassium is antagonistic to the bivalent cations
W.R. Żelazny, M. Licznar-Małańczuk Soil & Tillage Research 180 (2018) 250–258

Fig. 7. Simulated mean differences in leaf nutrient concentrations between

Fig. 6. Simulated mean differences in soil nutrient concentrations between each
mulch treatment and herbicide fallow with a typical soil from the herbicide
treatment (0.76 g N kg−1, 110 mg P kg−1, 130 mg K kg−1, 1110 mg Ca kg−1 and
80 mg Mg kg−1) used as the reference. The contour levels at the first plot per-
tain to total nutrient concentration difference in grams per kilogram. The
strength of a given mulch effect for each dimension (along x axis, along y axis or
along the contour gradient) in relation to herbicide fallow is indicated by the
distance of the estimate from the (0, 0) point. The point sizes are proportional
to Mahalanobis distance from the estimate most supported by data. The plau-
sibility of the simulated estimates to represent the true population mean de-
creases with increasing point size.
(Fageria, 2001), the observed decrease in K:Ca+Mg ratio can be in-
terpreted as a sign of appropriate tree nutrient status in the presence of
the synthetic mulch.
On the other hand, in Merwin et al. (1995) research, the Belton-
Sarlon Weed-Mat, which most resembles our PP, contributed to yield
reduction by 13%, and Neilsen and Hogue (1992) reported that woven
polypropylene fabric impaired yield by 28% compared with year-round
each mulch treatment and herbicide fallow with a typical nutrient composition
from the herbicide treatment (2.2% N, 0.2% P, 1.3% K, 1.3% Ca and 0.2% Mg)
used as the reference. The contour levels at the first plot pertain to total nutrient
concentration difference in percentage points. The strength of a given mulch
effect for each dimension (along x axis, along y axis or along the contour gra-
dient) in relation to herbicide fallow is indicated by the distance of the estimate
from the (0, 0) point. The point sizes are proportional to Mahalanobis distance
from the estimate most supported by data. The plausibility of the simulated
estimates to represent the true population mean decreases with increasing point
size.
herbicide fallow. In a more recent study at the same site, but with
fertigated trees, similar mulch outperformed herbicide fallow by ap-
proximately 55% (Neilsen et al., 2003). This lack of consistency in re-
ported effects suggests a high degree of dependency of synthetic mulch
performance on local conditions, orchard management and especially
mulch properties. Nevertheless, considering the high potential gains
associated with this floor management system, commercial orchards
should be encouraged to assess its suitability in their operations, and

Table 7
Point and interval estimates of mean differences for apple tree status (non-ilr variables) between mulch treatments and herbicide fallow.
Contrast Logit leaf dry matter content (n = 28) Leaf chlorophyll contenta (n = 128) Fruit 2005–2016 yielda (n = 16) Tree death oddsa (n = 80)

PP–herbicide 0.00 [−0.06, 0.07] 0.92 [0.85, 1.00] 1.01 [0.70, 1.45] 2.2 [0.4, 12.7]
Festuca–herbicide not estimated not estimated 0.76 [0.49, 1.16] 1.6 [0.3, 9.8]
Agrostis–herbicide −0.03 [−0.10, 0.03] 0.89 [0.82, 0.97] 0.54 [0.31, 0.93] 2.9 [0.5, 16.0]

a
Response ratio.

257
W.R. Żelazny, M. Licznar-Małańczuk
Table 8
Point and interval estimates of mean differences regarding apple tree leaf macronutrient contents between mulch treatments and herbicide fallow (n = 28).
Contrast ilrnutrients ilrions
PP–herbicide 0.06 [0.00, 0.12] −0.02 [−0.14, 0.10]
Festuca–herbicide Not estimated Not estimated
Agrostis–herbicide 0.04 [−0.01, 0.10] 0.03 [−0.09, 0.14]
new research should focus on identifying key factors to maximise suc-
cess in synthetic mulch application.
5. Conclusions
The long-term presence of living mulches in a rain-fed apple orchard
improved soil quality, especially its physical properties and humus
content. The understory vegetation, however, contributed to substantial
yield reductions by competing with the fruit trees for water, and future
research on floor management systems in unirrigated orchards should
focus on addressing the competition issue.
Although the collected data did not enable precise estimation of the
black woven polypropylene fabric effect on long-term fruit yield, its use
indicated the possibility of improved growth conditions of the apple
trees, and this orchard floor management system can therefore be re-
commended for use in some orchards.
Acknowledgements
Funding: This project was supported by the Ministry of Agriculture
of the Czech Republic [project no. RO 0416] and by the Polish Ministry
of Science and Higher Education [statutory activities of the Department
of Horticulture of the Wrocław University of Environmental and Life
Sciences].
References
Atucha, A., Merwin, I.A., Brown, M.G., 2011a. Long-term effects of four groundcover
management systems in an apple orchard. HortScience 46, 1176–1183.
Atucha, A., Merwin, I.A., Purohit, C.K., Brown, M.G., 2011b. Nitrogen dynamics and
nutrient budgets in four orchard groundcover management systems. HortScience 46,
1184–1193.
Barden, J.A., Neilsen, G.H., 2003. Selecting the orchard site, site preparation and orchard
planning and establishment. In: Ferree, D.C., Warrington, I.J. (Eds.), Apples: Botany,
Production, and Uses. CABI Publishing, Wallingford, pp. 237–265.
Bates, D., Mächler, M., Bolker, B.M., Walker, S.C., 2015. Fitting linear mixed-effects
models using lme4. J. Stat. Softw. 67, 1–48.
Biavetti, I., Karetsos, S., Ceglar, A., Toreti, A., Panagos, P., 2014. European meteorological
data: contribution to research, development and policy support. In: Hadjimitsis, D.G.,
Kyriacos, T., Michaelides, S., Papadavid, G. (Eds.), Second International Conference
on Remote Sensing and Geoinformation of the Environment (RSCy2014). vol. 9229 of
Proc. of SPIE. pp. 922907-1–922907-9.
Bolker, B.M., Brooks, M.E., Clark, C.J., Geange, S.W., Poulsen, J.R., Stevens, M.H.H.,
White, J.-S.S., 2009. Generalized linear mixed models: a practical guide for ecology
and evolution. Trends Ecol. Evol. 24, 127–135.
Bond, W., Grundy, A.C., 2001. Non-chemical weed management in organic farming
systems. Weed Res. 41, 383–405.
Egozcue, J.J., Pawlowsky-Glahn, V., 2005. Groups of parts and their balances in com-
positional data analysis. Math. Geol. 37, 795–828.
Fageria, V.D., 2001. Nutrient interactions in crop plants. J. Plant Nutr. 24, 1269–1290.
Filzmoser, P., Hron, K., Reimann, C., 2009. Univariate statistical analysis of environ-
mental (compositional) data: problems and possibilities. Sci. Total Environ. 407,
6100–6108.
Gitelson, A.A., Buschmann, C., Lichtenthaler, H.K., 1999. The chlorophyll fluorescence
ratio F735/F700 as an accurate measure of the chlorophyll content in plants. Remote
Sens. Environ. 69, 296–302.
Glenn, D.M., Miller, S.S., Habecker, M.A., 1987. Effect of soil management and calcium
nitrate fertilization on the availability of soil nitrate and cations in an eastern apple
orchard. J. Am. Soc. Hortic. Sci. 112, 436–440.
Granatstein, D., Sánchez, E., 2009. Research knowledge and needs for orchard floor
management in organic tree fruit systems. Int. J. Fruit Sci. 9, 257–281.
Hammermeister, A.M., 2016. Organic weed management in perennial fruits. Sci. Hortic.
208, 28–42.
258
Soil & Tillage Research 180 (2018) 250–258
ilrN:P ilrK:Ca+Mg ilrCa:Mg
−0.20 [−0.65, 0.25] −0.16 [−0.32, 0.00] −0.02 [−0.09, 0.04]
Not estimated Not estimated Not estimated
−0.23 [−0.68, 0.23] −0.02 [−0.18, 0.14] −0.03 [−0.09, 0.03]
Hogue, E.J., Cline, J.A., Neilsen, G., Neilsen, D., 2010. Growth and yield responses to
mulches and cover crops under low potassium conditions in drip-irrigated apple
orchards on coarse soils. HortScience 45, 1866–1871.
Kemper, W.D., Koch, E.J., 1966. Aggregate Stability of Soils from Western United States
and Canada: Measurement Procedure, Correlations with Soil Constituents.
Agricultural Research Service. United States Department of Agriculture, Washington.
Licznar-Małańczuk, M., 2012. Zastosowanie żywych sciólek w rzędach drzew jabłoni jako
alternatywnego sposobu pielęgnacji gleby w porównaniu z ugorem herbicydowym.
Wydawnictwo Uniwersytetu Przyrodniczego we Wrocławiu, Wrocław.
Lipecki, J., Berbeć, S., 1997. Soil management in perennial crops: orchards and hop
gardens. Soil Till. Res. 43, 169–184.
Lisek, J., 2014. Possibilities and limitations of weed management in fruit crops of the
temperate climate zone. J. Plant Prot. Res. 54, 318–326.
Magdoff, F., 2001. Concept, components, and strategies of soil health in agroecosystems.
J. Nematol. 33, 169–172.
Måge, F., 1982. Black plastic mulching, compared to other orchard soil management
methods. Sci. Hortic. 16, 131–136.
Mantinger, H., 1996. Neue schwachwachsende Apfelunterlagen. Obstbau Weinbau 11,
278–281.
Merwin, I.A., 2003. Orchard-floor management systems. In: Ferree, D.C., Warrington, I.J.
(Eds.), Apples: Botany, Production, and Uses. CABI Publishing, Wallingford, pp.
303–318.
Merwin, I.A., Rosenberger, D.A., Engle, C.A., Rist, D.L., Fargione, M., 1995. Comparing
mulches, herbicides, and cultivation as orchard groundcover management systems.
HortTechnology 5, 151–158.
Merwin, I.A., Stiles, W.C., van Es, H.M., 1994. Orchard groundcover management impacts
on soil physical properties. J. Am. Soc. Hortic. Sci. 119, 216–222.
Miller, S.S., 1983. Response of young ‘Topred Delicious’ apple trees to orchard floor
management and fertilization. J. Am. Soc. Hortic. Sci. 108, 638–642.
Miller, S.S., Glenn, D.M., 1985. Influence of various rates of Ca(NO3)2 fertilizer and soil
management on young apple trees. J. Am. Soc. Hortic. Sci. 110, 237–243.
Neilsen, G.H., Hogue, E.J., 1992. Long-term effects of orchard soil management on tree
vigor and extractable soil nutrients. Can. J. Soil Sci. 72, 617–621.
Neilsen, G.H., Hogue, E.J., Forge, T., Neilsen, D., 2003. Mulches and biosolids affect
vigor, yield and leaf nutrition of fertigated high density apple. HortScience 38,
41–45.
Neilsen, G.H., Neilsen, D., 2003. Nutritional requirements of apple. In: Ferree, D.C.,
Warrington, I.J. (Eds.), Apples: Botany, Production, and Uses. CABI Publishing,
Wallingford, pp. 267–302.
Oliveira, B.S., Ambrosini, V.G., Trapp, T., dos Santos, M.A., Sete, P.B., Lovato, P.E., Loss,
A., Comin, J.J., Lourenzi, C.R., da Rosa Couto, R., Toselli, M., Brunetto, G., 2016.
Nutrition, productivity and soil chemical properties in an apple orchard under weed
management. Nutr. Cycl. Agroecosys. 104, 247–258.
Oliveira, M.T., Merwin, I.A., 2001. Soil physical conditions in a New York orchard after
eight years under different groundcover management systems. Plant Soil 234,
233–237.
Panagos, P., 2006. The European Soil Database. GEO: Connex. Int. (July/August), 32–33.
Pawlowsky-Glahn, V., Egozcue, J.J., 2006. Compositional data and their analysis: an
introduction. Geol. Soc. Lond. Spec. Publ. 264, 1–10.
R Core Team, 2017. R: A Language and Environment for Statistical Computing.
Foundation for Statistical Computing, Vienna, Austria.
Sanchez, J.E., Edson, C.E., Bird, G.W., Whalon, M.E., Willson, T.C., Harwood, R.R.,
Kizilkaya, K., Nugent, J.E., Klein, W., Middleton, A., Loudon, T.L., Mutch, D.R.,
Scrimger, J., 2003. Orchard floor and nitrogen management influences soil and water
quality and tart cherry yields. J. Am. Soc. Hortic. Sci. 128, 277–284.
Schmid, A., Weibel, F., Allemannn, P., Santini, D., 2004. Sandwich-System. Erwerbs
Obstbau 46, 87–94.
Sobiczewski, P. (Ed.), 2015. Metodyka integrowanej ochrony jabłoni (materiały dla
doradców). Instytut Ogrodnictwa, Skierniewice.
Szczygieł, A., Czynczyk, A., 2002. Suitability of some semidwarf and dwarf rootstocks to
three apple cultivars in the sub-Carpathian region. J. Fruit Ornam. Plant Res. 10,
85–93.
Tahir, I.I., Svensson, S.-E., Hansson, D., 2015. Floor management systems in an organic
apple orchard affect fruit quality and storage life. HortScience 50, 434–441.
Yao, S., Merwin, I.A., Bird, G.W., Abawi, G.S., Thies, J.E., 2005. Orchard floor manage-
ment practices that maintain vegetative or biomass groundcover stimulate soil mi-
crobial activity and alter soil microbial community composition. Plant Soil 271,
377–389.
Yao, S., Merwin, I.A., Brown, M.G., 2009. Apple root growth, turnover, and distribution
under different orchard groundcover management systems. HortScience 44,
168–175.