Ecological Engineering: A B B C A D B

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Ecological Engineering 136 (2019) 30–37

Contents lists available at ScienceDirect

Ecological Engineering
journal homepage: www.elsevier.com/locate/ecoleng

Assessing the context and ecological effects of river restoration – A meta- T


analysis

Weiwei Lua,b, Raquel Arias Fontb,c, Shuiping Chenga, , Jieqiong Wangd, Johannes Kollmannb
a
College of Environmental Science and Engineering, Key Laboratory of Yangtze River Water Environment, Ministry of Education, Tongji University, 1239 Siping Road,
Shanghai 200092, China
b
Chair of Restoration Ecology, Technical University of Munich, Emil-Ramann-Straße 6, 85354 Freising, Germany
c
School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, B15 2TT Birmingham, UK
d
College of Architecture and Urban Planning, Tongji University, 1239 Siping Road, Shanghai 200092, China

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

Keywords: A large number of qualitative and semi-quantitative analyses have been performed to summarize the effects of
Biodiversity river restoration, while systematic and quantitative assessments of the literature on river restoration are still
Degradation type rare. To fill this gap, we conducted a meta-analysis of 55 peer-reviewed publications describing assessments of
Ecosystem services river restoration with indicators of biodiversity or ecosystem services (ES). The core of the study was to identify:
Organism group
(i) the overall ecological effects of river restoration; (ii) the restoration effects on different organism groups and
Restoration age
types of ES, including the relationship between biodiversity and ES; and (iii) the influence of the restoration
Restoration techniques
context, that is degradation types, restoration techniques and restoration age. The meta-analysis revealed that
river restoration significantly enhanced ecological conditions. However, restoration effects varied among or-
ganism groups and ES, with plant abundance and regulating services showing higher responses to restoration.
Moreover, a trend of a positive correlation was found between biodiversity and ES. Restoration activities showed
most significant effects in the context of hydromorphological degradation or land-use changes, and channel
reconfiguration and riparian buffer establishment had a particularly favourable outcomes. Only regulating
services showed a positive correlation with restoration age; thus, long-time monitoring is necessary to fully
understand restoration effects. Overall, our meta-analysis provides evidence-based advice for future projects and
assessment of river restoration.

1. Introduction et al., 2015).


Evaluation of restoration effects is essential for any scientific and
Rivers and floodplains are among the ecosystems that have suffered practical progress in river restoration (Kondolf, 1995; Palmer et al.,
most from human interventions including channelization, pollution and 2005). The studies that monitored and evaluated the effects of river
destruction of riparian vegetation (Antón et al., 2011; Pedersen et al., restoration mainly focused on changes in biodiversity, primarily on
2007). Many rivers have changed from dynamic and diverse ecosystems invertebrates, fish and macrophytes (Antón et al., 2011; Li et al., 2014;
to static and homogeneous ones, resulting in enormous losses of bio- Reeves et al., 2016; Tullos et al., 2009), while few paid attention to
diversity and ecological functions (Allan and Castillo, 2007; Tockner changes of in ES (McMillan et al., 2014; Pander et al., 2015). Combined
et al., 2009). More recently, river restoration projects have been im- studies evaluating changes in biodiversity and ES are even more scarce
plemented at an increasing rate in many regions for recreating habitats, (Dolph et al., 2015; Wu et al., 2013) and, to our knowledge, no study
and for improving biodiversity and ecosystem services (ES) (Bernhardt exists in the peer-reviewed literature on the relationship between bio-
et al., 2005; Nilsson et al., 2016). However, in contrast to the rapid diversity and ES during river restoration. Furthermore, contrasting
growth in demand and practice of river restoration, little is known evidence has been published on restoration effects on biodiversity of
about the actual effects of restoration and the influence of the re- river ecosystems. For example, some studies showed that restoration
storation context, i.e. the degradation type, the restoration age and the improved the conditions of certain groups of organisms (Besacier-
specific techniques applied. This is due to a lack of adequate monitoring Monbertrand et al., 2014; de Jong and Cowx, 2016), while others found
and assessment (Bernhardt et al., 2005; Jähnig et al., 2011; Hering almost no biotic effects of river restoration, for example on benthic


Corresponding author.
E-mail address: shpcheng@tongji.edu.cn (S. Cheng).

https://doi.org/10.1016/j.ecoleng.2019.06.004
Received 20 September 2018; Received in revised form 28 May 2019; Accepted 9 June 2019
Available online 15 June 2019
0925-8574/ © 2019 Elsevier B.V. All rights reserved.
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

invertebrates (Friberg et al., 2014; Haase et al., 2013; Lepori et al., resulted in a list of 460 references to be included in the meta-analysis.
2005), macrophytes (Lorenz et al., 2012; Pedersen et al., 2007) or fish For this subsample, five criteria were used to decide whether or not
(Antón et al., 2011; Lorenz et al., 2013; Schmutz et al., 2014). De- the study could be included: (1) the research systems should be a river
gradation types, restoration techniques (de Jong and Cowx, 2016) and or stream ecosystem (excluding wetland, floodplain and marine eco-
restoration age (Hering et al., 2015; Paillex et al., 2017) most likely system); (2) changes in variables relating to biodiversity or/and ES
impact the effects of river restoration, but have rarely been studied. should be reported in a quantitative way; (3) information should be
Thus, a comprehensive assessment is needed of the restoration outcome presented that compared the restored and unrestored conditions or
(“ecological effects”) on both biodiversity and ES, based on all scientific control conditions; (4) the work should be a field study; and (5) the
studies available. restoration should have taken place at least 1 year earlier. Finally,
Existing reviews of the effects of river restoration are mostly based combined with eight papers traced from two other review papers (i.e.
on qualitative (Hering et al., 2013; Pander and Geist, 2013; Olden et al., Pan et al., 2016; Rubin et al., 2017) and two papers (i.e. Vehanen et al.,
2014; Rubin et al., 2017) or semi-quantitative analyses (Palmer et al., 2010; Louhi et al., 2016) from other papers’ reference lists, 55 studies
2014), but cannot infer general conclusions from inconsistent results of formed the basis of the meta-analysis (see Supplementary Materials).
multiple studies. A meta-analysis performed as an integrative and
quantitative approach, with more reliable results, would overcome this 2.2. Data extraction
challenge (Chung et al., 2006).
Some meta-analyses of restoration effects on ecosystem conditions Data of mean value, standard deviation and sample size of different
(Rey Benayas et al., 2009), or on fish, macroinvertebrates and macro- measured metrics in restored and unrestored areas, and additional in-
phytes (Miller et al., 2010; Whiteway et al., 2010; Kail et al., 2015) formation (i.e. river name, degradation types, restoration age, restora-
have already been carried out. However, despite the previous con- tion techniques) were extracted from the text, the figures and/or tables
tributions made by practitioners and scientists, practical knowledge is in the papers selected. Data were extracted from figures using
lacking on restoration effectiveness based on degradation type and re- PlotDigitizer (http://plotdigitizer.sourceforge.net/). Biodiversity data
storation technique, and on the influence of the period since restora- were differentiated according to abundance, density and species
tion. Our meta-analysis aims to fill this knowledge gap. number of different organism groups: fish, invertebrates, periphyton,
We quantified the ecological effects of river restoration based on the macrophytes and riparian vascular plants. Only a few studies explicitly
results of studies published in the peer-reviewed literature in recent referred to the concept of ES. However, based on the definition of dif-
thirty decades. Changes in biodiversity are measured with the in- ferent ES classified by the MEA (2005), and the concrete metrics
dicators of abundance, density and species numbers of five organism measured in each study, we adopted the identification and classification
groups: fish, invertebrates, periphyton, macrophytes and riparian vas- of reported variables as measures of ES (Table A1). For example, as
cular plants. ES indicators were classified according to the Millennium indicator for regulating services Chemical Oxygen Demand (COD) was
Ecosystem Assessment (MEA), with a focus on regulating and sup- classified as a contribution to ecosystem regulation. Biodiversity data
porting services (MEA, 2005). Regulating services were considered as were differentiated in abundance, density and species number of dif-
those functions providing societal benefits by the modification or sta- ferent organism groups, i.e. fish, invertebrate, periphyton, macrophyte
bilization of ecological processes. This could be the regulation of cya- and riparian vascular plant. Other information including river name,
nobacterial growth by water purification capacity of riparian zones or project region, river length, restoration age, cause of degradation
stabilization of local climate by the microclimate regulation of rivers. (channelization, damming, deforestation, fishing and land-use) and
As for supporting services, we classified the publications as maintaining restoration measures (boulder addition, gravel introduction, wood ad-
the certain compartments of the ecosystems functioning, i.e. nutrient dition, channel reconfiguration and riparian buffer establishment) were
cycling or primary productivity linked to basal food sources and oxygen also collected. Studies without well-defined organism group or or-
concentrations. ganism group with too small sample size (i.e. phytoplankton), active
The meta-analysis had three objectives, (1) to identify the overall restoration measures or measures without well-defined clear definitions
ecological effect of river restoration, i.e. to test whether or not re- were classified as “others”.
storation has a positive effect on ecosystem conditions; (2) to in-
vestigate the effects of river restoration on different biodiversity com-
2.3. Quantifying ecological effect and publication bias
ponents and ES, including potential correlations between biodiversity
and ES; and (3) to disentangle the influences of the restoration context.
Effect size that reveals the amount of the treatment effect or the
strength of the interaction within two variables (Borenstein et al., 2009)
2. Materials and methods
is an important currency in meta-analyses, and can be used to quantify
the ecological effect of river restoration. The effect size of response ratio
2.1. Literature search
R (Hedges et al., 1999) can be calculated without replications, which
are most commonly used to express the effect magnitude of an ex-
We made a systematic search of the peer-reviewed literature to
perimental treatment. Variables that are expected to have lower values
identify quantitative evidence of the ecological effects of river re-
after restoration (e.g., nutrient concentrations) were reciprocally
storation. In the ISI Web of Knowledge database, all relevant studies
transformed to ensure that all results representing a good ecological
were identified on 27 April 2017, using the following keywords:
state pointed in the same (positive) direction.
(stream* OR river*) AND (restor* OR rehabilit* OR revitali*) AND
The response ratio R (representing the ln response ratio) and its
(service* OR function* OR biodiversity*). The initial search was refined
variance were computed as:
to the subject areas “Environmental Sciences”, “Ecology”, “Water
− −
Resources”, “Biodiversity Conservation”, “Engineering Environment”, R = ln[(Xa + 1)/(Xb + 1)] (1)
“Fisheries”, “Rehabilitation”, “Environmental Studies”, “Plant − −
Sciences”, “Multidisciplinary Sciences”, “Biology” and “Zoology”, using where Xa and Xb were the means of the values under restored and
the “Refine Results” option in Web of Knowledge. Based on this search unrestored conditions. The response ratio is dimensionless with va-
string, 3131 references were retrieved. However, despite the initially lues > 0 indicating river improvement and < 0 denoting river de-
filtered papers might contain the search keywords and belong to the gradation. To avoid the occurrence of 0 in the natural logarithm of the
subject areas, the actual content may miss the aim of our analysis. Thus, equation, “1” was added to both the numerator and denominator of the
we scrutinised the title and abstract of each of these publications, which equation (Moreno-Mateos et al., 2012), which demonstrated has little

31
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

effect on conclusions (Scheiner and Gurevitch, 2001). Calculations were number) responded to river restoration in different ways, and the or-
done in R project version 3.4.3 using the metafor (Viechtbauer, 2010) ganism groups also showed different magnitudes in their responses to
and compute.es packages (Del Re, 2013), while the actual classification restoration (Fig. 2).
of the response ratio for detailed analysis can be seen in Appendix A. Overall, river restoration had a favourable effect on biodiversity
Supplementary material. (Fig. 2), and the effects of river restoration on abundance, species
To assess the potential publication bias, a funnel plot (Fig. A1) and number and density were all positive (p < 0.05), with the estimated
Egger’s regression test were performed (Egger et al., 1997). Further response ratio and 95% confidence intervals being 0.50 [0.35, 0.65],
information assessing publication bias can be found in the supple- 0.18 [0.04, 0.32] and 0.20 [0.01, 0.39], respectively. The estimated
mentary material. response ratio of abundance differed from that of species number and
density, and was higher for abundance compared to species number
2.4. Statistical analyses (p < 0.001) and density (p < 0.001). This suggested that river re-
storation had greater effects on species abundance than on species
The meta-analysis implies certain dependencies that need to be numbers and densities.
addressed: location dependence (i.e. studies performed in the same River restoration caused a significant increase in abundance and
river), within-research-group dependence (studies performed by the species numbers of fish, invertebrates and macrophytes. Periphyton and
same research group or authors) and study dependence (different riparian vascular plants also showed positive effects for abundance and
measurements extracted from the same study) (Gurevitch and Hedges, density. In other organism groups there were few or no effects of river
1999). Dependency was assessed using a three-level meta-analysis restoration, for example in the density metric of invertebrates.
(Cheung, 2014). A mixed effect model followed as: Significant differences in restoration effects between organism groups
existed in abundance and species number components (p < 0.05). The
yij = β1 x ij + u (2) ij + u (3) j + eij (2) estimated response ratio of abundance of invertebrates (0.25) was sig-
nificantly lower compared to that of macrophytes (0.66) and periph-
where yij represents the ith effect size in the jth cluster, β1 is the vector
yton (1.11). Similarly, there was a weaker response ratio of species
with the regression coefficients including the intercept, x ij is a vector
number for invertebrates (0.17) compared to macrophytes (0.42) and
with the predictors, u (2) ij and u (3) j are the heterogeneity at levels 2 and
fish (0.43).
3, respectively, andeij is the known sampling variance in the ith effect
Even with a positive effect on the overall ES, river restoration af-
size in the jth cluster. For testing the effect of discrete predictors (i.e.
fected the general ES and particular ES in different ways (Table 1). The
organism group, functions), the intercept was constrained to 0. Statis-
estimated response ratio to regulating services was positive and its 95%
tical analyses were developed in R project version 3.4.3 (https://www.
confidence intervals did not include 0, indicating that river restoration
r-project.org/) using the metaSEM package (Cheung, 2015). To obtain
can improve the regulating services. However, there was no significant
correlation coefficients between restoration effect and time, Spearman
effect on supporting services. Hence, it suggested that river restoration
rank tests were calculated in R project version 3.4.3 (https://www.r-
improved the regulating services more strongly than that supporting
project.org/).
ones. Similarly, only habitat structure regulation (p = 0.002) showed
significant effects of river restoration. Thus, although river restoration
3. Results had a different effect on ecosystem functions or services, the overall ES
were improved by restoration.
3.1. Overall ecological effect analysis The meta-analysis also tested whether or not the changes in ES were
positively related to the change in biodiversity due to river restoration.
The 55 studies yielded 435 results on river restoration effect from 19 With 31 pairs of studies containing both biodiversity and ES variables, a
countries, with Germany, the USA and Switzerland accounting for the complicated relationship between biodiversity components and ES
largest percentages (23%, 17% and 8%, respectively). The data point emerged (Fig. A2), and the mean RR of biodiversity showed a positive
distribution in the different categories of organism and ecosystem ser- trend with that of ES, but not a linear correlation (p > 0.05, r = 0.24).
vices and the concrete metrics are shown in Table A2 in Supplementary
Materials. Considering all studies together, the estimated overall re- 3.3. Effect of river restoration context
sponse ratio and its 95% confidence intervals were +0.31 and [0.19,
0.42], respectively, situated on the right of the no-effect zero line The meta-analysis explored five types of river degradation causes,
(Fig. 1). This means that the overall ecological effect of river restoration and it showed that restoration activities have most significant effects in
was positive (p < 0.01), despite the response ratios varying sig- hydromorphological degradation context (p = 0.02; Table 2). More-
nificantly among the studies, and one-fourth showing no increase over, restoration activities could also show positive effects in the con-
(R ≤ 0) after restoration. This positive effect was also seen in biodi- text of degradation due to land-use changes (p = 0.05), like intensified
versity and ES as a whole, with the estimated response ratio of biodi- agriculture or road construction, while not for the contexts of channe-
versity being higher than that of ES, suggesting that it is more difficult lization, damming, deforestation and fishing.
to recover ES than biodiversity. The estimated response ratios and its 95% confidence intervals of
The five groups of organisms revealed different effects of river re- restoration techniques of channel reconfiguration and riparian buffer
storation: periphyton showed no significant response to river restora- establishment were all positive. They did not include the zero line for
tion, while fish, invertebrates, macrophytes and riparian vascular plant the overall effects, the ones of biodiversity and ES, indicating that these
groups responded positively to restoration. We identified a significant two restoration techniques affected restoration most strongly (Table 3).
difference in restoration effect between macrophytes and invertebrates, In addition, ES was also positively affected by the restoration technique
with the estimated response ratio of macrophytes (0.49) being higher of wood addition, with significant improvements.
than that of invertebrates (0.20). Similarly, we found that river re- Biodiversity components and general ES types responded differently
storation had a positive effect on regulating services but not on sup- to restoration age (Fig. 3). No correlation was found between restora-
porting services. tion age and the biodiversity components, suggesting that changes in
biodiversity were unrelated to restoration age. Regulating services
3.2. Magnitude of effect of river restoration on biodiversity and ES (p = 0.02) had a positive relationship with restoration age, while this
was not the case for supporting services (p = 0.89), nor for fish
The three biodiversity components (abundance, density and species (p = 0.47), invertebrates (p = 0.99), macrophytes (p = 0.93),

32
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

Fig. 1. Response ratios of biodiversity, ecosystem services (ES) and overall ecological effect of river restoration. In the forest plot the zero line indicates no effect,
with positive effect falling to the right of the zero line.

Fig. 2. Response ratios of four biodiversity components among different organism groups to river restoration.

periphyton (p = NA), phytoplankton (p = NA) and riparian vascular 4. Discussion


plants (p = 0.19). Overall, the results suggest that the restoration effect
had no apparent relation with restoration age. 4.1. Complex effects of river restoration on biodiversity and ES

Overall, there was a positive outcome of river restoration for bio-


diversity and ES. The strongest effects were found in biodiversity stu-
dies that considered species numbers, their abundance and densities.
The increase in abundance was double that of species numbers and

33
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

Table 1 2005; Rosi-Marshall et al., 2006; Li et al., 2014). It also suggests that
Results of the mixed effect multivariate model with ecosystem services (ES) as a one should be cautious when choosing invertebrates for evaluating
covariance used to estimate the response ratio (RR) of ES including the lower restoration due to marginal or contrasting results in different studies
and upper lower confidence intervals (CIlower and CIupper, respectively). (Pedersen et al., 2007; Sarriquet et al., 2007). Moreover, invertebrate
Ecosystem services RR CIlower CIupper p indices may report well on near-natural conditions, while they do not
reflect minor improvements of degraded states (Hahn, 1982; Valentina
Total Ecosystem services 0.219 0.045 0.394 0.013
et al., 2011), especially when using EPT indexes based on Ephemer-
General ES Regulating 0.391 0.037 0.744 0.030
Supporting 0.261 −0.124 0.645 ns
iptera, Plecoptera and Trichoptera. This approach might produce results
Particular ES Connectivity regulation 0.165 −0.299 0.629 ns unrelated to the restored reach, but rather reflect patterns within tri-
Water flow regulation 0.089 −0.348 0.526 ns butaries or the entire catchment (Brosse et al., 2003; Feld and Hering,
Habitat structure regulation 0.642 0.241 1.043 0.002 2007; Giling et al., 2016). Despite invertebrates being directly affected
Water quality 0.118 −0.355 0.591 ns
by the degree of changes in aquatic substrate and the establishment of
Nutrient cycling 0.203 −0.248 0.655 ns
Food/raw materials 0.527 −0.022 1.076 (*) new microhabitats (Hering et al., 2015), the few correlations found
Primary productivity −0.165 −0.841 0.512 ns between habitat heterogeneity and macroinvertebrates further support
this argument (Palmer et al., 2010).
(*) p < 0.10; ns, p ≥ 0.1. The recreation of microhabitats is highly affected by macrophytes,
which also play a significant role in stabilizing river morphology (Jones
Table 2 et al., 1994; Sand-Jensen, 1997). Most effects on macrophytes were
Results of the mixed effect multivariate model with degradation type as a probably related to restoration techniques, for example channel re-
covariance used to estimate the response of ecological effect to different kinds
configuration and river widening. Such techniques can reduce flow
of degradation (response ratio, RR) with the lower and upper lower confidence
velocity and create suitable habitats for pioneer plants (Saldi-Caromile
intervals (CIlower and CIupper, respectively).
et al., 2004; Erwin et al., 2017), which favours other macrophytes and
Degradation types RR CIlower CIupper p riparian vascular plants at later stages of succession. In addition,
Channelization 0.09 −0.08 0.26 ns
macrophytes and helophytes have positive effects on nutrient retention
Damming −0.27 −0.74 0.20 ns and cycling (Ribot et al., 2017).
Deforestation 0.23 −0.34 0.81 ns According to the MEA, nutrient retention and recycling are directly
Hydro-morphological degradation 0.06 0.01 0.11 0.014 related to the supporting services. Thus, the positive restoration effect
Land-use change 0.35 −0.01 0.71 (*)
on macrophytes should improve nutrient dynamics and the resulting
(*) p < 0.10; ns, p ≥ 0.1. ES. However, our meta-analysis produced contrasting results, which
underline the complex relationship between biodiversity and ES.
densities, which suggests that restoration more efficiently increases the Restoration techniques like channel reconfiguration, which can reduce
number of individuals of existing species than it promotes the arrival of flow velocity, might actually decrease the nutrient cycling rate
new biodiversity (Miller et al., 2010; Stoll et al., 2014). (Hasselquist et al., 2015), thus explaining this contrasting result. Fur-
Differences in dispersal potential might partially explain variation thermore, greater input of nutrients will result from connected chan-
among organism groups, with river restoration causing a significant nels, at least under high discharge (Inglett et al., 2008). However,
increase in abundance and species richness of fish, invertebrates and channel reconfiguration may have positive effects on other ecological
macrophytes, but not in the other groups investigated. This is corre- functions related to river-bed dynamics and habitat regulation
sponding with the results showing that river restoration only benefits (Bukaveckas, 2007; Ock et al., 2015).
some groups, i.e. fish, invertebrates and macrophytes (Besacier- Regulating services, which are related to river-bed configuration
Monbertrand et al., 2014; Neale and Moffett, 2016; Reeves et al., 2016). and habitat structure establishment, might benefit significantly from
In addition, the biodiversity response varied greatly among and within channel reconfiguration and wood introduction, as these can create
catchments, mostly due to human intervention. For example, dams habitats and improve habitat heterogeneity (Acuña et al., 2013; Testa
jeopardize species capacity to reach restored river sections (Downs et al., 2011). Moreover, with microhabitat creation and improved ha-
et al., 2011). Hence, dispersal limitation such as habitat fragmentation bitat heterogeneity, macrophytes show positive restoration effects
and lack of a regional species pool might prevent new species from (Januschke et al., 2009; Lorenz et al., 2012). As Pander et al. (2015)
establishing (Brederveld et al., 2011; Sundermann et al., 2011). The demonstrated, improvement in habitat quality for aquatic species mo-
presence of diffuse (e.g. agricultural land use) or point pollution tivates river restoration and often has positive effects on the ecosystem,
(wastewater treatment plants) within the river catchment greatly which can also be used to explain the significant restoration effect on
modified species establishment (Ribot et al., 2017; Lorenz et al., 2013). regulating services.
Species sensitivity to pollution or eutrophication might explain the
limited response of invertebrates to river restoration (Lepori et al.,

Table 3
Results of the mixed effect multivariate model with restoration technique as a covariance used to estimate the response of biodiversity and ecosystem services to
different restoration techniques (response ratio, RR) including the lower and upper lower confidence intervals (CIlower and CIupper, respectively).
Restoration techniques Overall Biodiversity Ecosystem services

RR CIlower CIupper p RR CIlower CIupper p RR CIlower CIupper p

Boulder addition 0.06 −0.15 0.28 ns 0.08 −0.15 0.32 ns −0.19 −0.74 0.36 ns
Channel reconfiguration 0.35 0.22 0.50 6.818e–07 0.36 0.21 0.52 6.985e–06 0.40 0.01 0.80 0.045
Gravel introduction 0.18 −0.08 0.44 ns 0.24 −0.05 0.52 ns −0.16 −0.79 0.48 ns
Riparian buffer 0.74 0.37 1.12 1.153e–04 0.55 0.08 1.03 0.022 0.89 0.20 1.57 0.011
Wood addition 0.26 −0.15 0.68 ns 0.04 −0.41 0.50 ns 1.15 0.30 2.01 0.008
Others 0.04 −0.28 0.36 ns 0.11 −0.23 0.44 – – – –

ns, p ≥ 0.1.

34
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

Fig. 3. Correlation between river restoration age and the response ratio for biodiversity components and ecosystem services (ES).

4.2. Restoration context controlling restoration effects provisioning. This could also explain the restoration debts reported by
many invertebrate studies (Sand-Jensen et al., 1999; Muotka et al.,
We found that river hydromorphological degradation (e.g. adverse 2002).
hydromorphology and drought) and land-use change (intensified agri- Another controversial aspect is that the disturbance caused by the
culture or road construction) were the degradation types that seemed to restoration techniques, which might produce negative outcomes on
be most effectively mitigated by restoration. While channelization and ecosystem biodiversity or ES (Hasselquist et al., 2015). The small
damming have been intensely studied when prioritizing restoration sample size in the advanced age classes within our meta-analysis might
areas (Kupilas et al., 2017; Renöfält et al., 2013; McManamay et al., have prevented us to detect long-term effects of restoration. Further-
2013), the cost–benefit ratio of the respective restoration techniques is more, the influence of climate change and extreme events like floods
not sufficiently understood. In addition, other studies suggest that the and drought periods might have compromised the observed trajectories
widely used addition of wood, boulder or gravel (Mueller et al., 2014; in river restoration (Bond and Lake, 2005).
Pander et al., 2015) might not be so effective in river restoration. In-
stead, more expensive and complex techniques such channel re-
configuration and riparian buffer establishment could bring greater 4.3. Summary and practical recommendations
success. Thus, the pivotal role of river hydromorphology on river eco-
logical state, which has been highlighted in recent European inventories The results of our meta-analysis suggest that river restoration sig-
(Hering et al., 2015), should be taken into consideration when prior- nificantly enhances ecological conditions in rivers, while restoration
itizing restoration efforts. effects vary among organism groups and ES, with plant abundance and
Overall, the results suggest that effects of restoration age are hidden regulating services being restored most effectively. Moreover, a slightly
by other factors, and only regulating services present a significant po- positive correlation was found between biodiversity and ES, and only
sitive relation with it. In several river restoration projects an “unhelpful regulating services showed a positive trend with restoration age.
resilience” of the degraded system was found, when structural im- For river restoration practitioners, firstly, we would suggest them to
provements, for example restored sand bars or recreated backwaters, monitor the restoration projects with long time series to reveal the true
disappeared after a few years (Bauer et al., 2018), or negative factors effects of river restoration (Woolsey et al., 2005; Kollmann et al., 2016)
persisted despite the structural improvement. Moreover, the different and the clear relationships between biodiversity and ES. Secondly, re-
organism groups revealed no apparent correlation with restoration age. storation techniques like channel reconfiguration and riparian buffer
The variation in the response of different organism groups concurred establishment, rather than the addition of wood, boulder or gravel, may
with the contradictory results in the peer-reviewed literature. be more effective in the river restoration based on our results; and more
While the opening of new habitats due to restoration techniques is attention should be paid to the contexts of hydromorphological de-
expected to favour riparian vascular plants and macrophytes in the gradation and land-use change when prioritizing restoration areas.
aquatic zones at the first stage of restoration, no clear relationship was For river restoration effect evaluation researchers, when integrating
found between restoration age and macrophytes (Lorenz et al., 2012). the effects (e.g. meta-analysis), it would be much better to get more
Expectations that consumers such as invertebrates and fish will benefit indices related with biodiversity and ecosystem services. In addition,
from the new availability of substrate and microhabitats (Matthaei researchers should have strict standards no matter for data collections
et al., 1999; Muotka et al., 2002) have been rarely supported. Haase or identification of the success of the restoration projects.
et al. (2013) detected only a slight positive effect on fish, or even a
decrease in fish species number despite increasing abundance with re-
Acknowledgments
storation age (Schmutz et al., 2014).
Species-specific trajectories also contributed to the unclear re-
This study was supported by the Chinese National Natural Science
lationship between restoration age and species recovery. As demon-
Foundation (51578395) and the Fundamental Research Funds for the
strated by Moreno-Mateos et al. (2012), macroinvertebrates required
Central Universities. The authors are grateful to Leonardo Teixeira for
5–10 years to recover, while it took on average 30 years for plant as-
commenting on the statistical analyses. Weiwei Lu also thanks the
semblages to develop towards the reference state. For example, the
China Scholarship Council for supporting her study at the Technical
recovery of aquatic plants can take up to several decades for full de-
University of Munich, Germany. The constructive comments of expert
velopment, which affects macroinvertebrates due to habitat
reviewers on earlier versions of this manuscript are appreciated.

35
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

Declaration of Competing Interest Hedges, L.V., Gurevitch, J., Curtis, P.S., 1999. The meta-analysis of response ratios in
experimental ecology. Ecology 80, 1150–1156.
Hering, D., Borja, A., Carvalho, L., Feld, C.K., 2013. Assessment and recovery of European
None. water bodies: key messages from the WISER project. Hydrobiologia 704, 1–9.
Hering, D., Aroviita, J., Baattrup-Pedersen, A., Brabec, K., Buijse, T., Ecke, F., Friberg, N.,
Appendix A. Supplementary data Gielczewski, M., Januschke, K., Köhler, J., Kupilas, B., Lorenz, A.W., Muhar, S.,
Paillex, A., Poppe, M., Schmidt, T., Schmutz, S., Vermaat, J., 2015. Contrasting the
roles of section length and instream habitat enhancement for river restoration suc-
Supplementary data to this article can be found online at https:// cess: a field study of 20 European restoration projects. J. Appl. Ecol. 52, 1518–1527.
doi.org/10.1016/j.ecoleng.2019.06.004. Inglett, P.W., Inglett, K.S., Reddy, K.R., 2008. Biogeochemical processes and implications
for nutrient cycling. Report in report titled “Summary and synthesis of the available
literature on the effects of nutrients on spring organisms and systems” submitted to
References FDEP Springs Initiative.
Jähnig, S.C., Lorenz, A.W., Hering, D., Antons, C., Sundermann, A., Jedicke, E., Haase, P.,
2011. River restoration success: a question of perception. Ecol. Appl. 21, 2007–2015.
Acuña, V., Díez, J.R., Flores, L., Meleason, M., Elosegi, A., 2013. Does it make economic
Januschke, K., Sundermann, A., Antons, C., Haase, P., Lorenz, A., Hering, D., Aue, D.,
sense to restore rivers for their ecosystem services? J. Appl. Ecol. 50, 988–997.
2009. Untersuchung und Auswertung von ausgewählten Renaturierungsbeispielen
Allan, J.D., Castillo, M.M., 2007. Stream Ecology: Structure and Function of Running
repräsentativer Fließgewässertypen der Flusseinzugsgebiete Deutschlands. Deutscher
Waters. Chapman & Hall, NY.
Rat für Landespflege 82, 23–39.
Antón, A., Elosegi, A., García-Arberas, L., Díez, J., Rallo, A., 2011. Restoration of dead
Jones, C.G., Lawton, J.H., Shachak, M., 1994. Organisms as ecosystem engineers. Oikos
wood in Basque stream channels: effects on brown trout population. Ecol. Freshwater
69, 373–386.
Fish 20, 461–471.
Kail, J., Brabec, K., Poppe, M., Januschke, K., 2015. The effect of river restoration on fish,
Bauer, M., Harzer, R., Strobl, K., Kollmann, J., 2018. Resilience of riparian vegetation
macroinvertebrates and aquatic macrophytes: a meta-analysis. Ecol. Indic. 58,
after restoration measures on River Inn. River Res. Appl. 18, 1–10.
311–321.
Bernhardt, E.S., Palmer, M.A., Allan, J.D., Alexander, G., Barnas, K., Brooks, S., Carr, J.,
Kollmann, J., Meyer, S.T., Bateman, R., Conradi, T., Gossner, M.M., Souza Mendonça, M.,
Clayton, S., Dahm, C., Follstad-Shah, J., Galat, D., Gloss, S., Goodwin, P., Hart, D.,
Fernandes, G.W., Hermann, J.M., Koch, C., Müller, S.C., Oki, Y., Overbeck, G.E.,
Hassett, B., Jenkinson, R., Katz, S., Kondolf, G.M., Lake, P.S., Lave, R., Meyer, J.L.,
Paterno, G.B., Rosenfield, M.F., Toma, T.S.P., Weisser, W.W., 2016. Integrating
O’Donnell, T.K., Pagano, L., Powell, B., Sudduth, E., 2005. Synthesizing US river
ecosystem functions into restoration ecology-recent advances and future directions.
restoration efforts. Science 308, 636–637.
Restor. Ecol. 24, 722–730.
Besacier-Monbertrand, A.L., Paillex, A., Castella, E., 2014. Short-term impacts of lateral
Kondolf, G.M., 1995. Five elements for effective evaluation of stream restoration. Restor.
hydrological connectivity restoration on aquatic macroinvertebrates. River Res. Appl.
Ecol. 3, 133–136.
30, 557–570.
Kupilas, B., Hering, D., Lorenz, A.W., Knuth, C., Gücker, B., 2017. Hydromorphological
Bond, N.R., Lake, P.S., 2005. Ecological restoration and large-scale ecological dis-
restoration stimulates river ecosystem metabolism. Biogeosciences 14, 1989.
turbance: the effects of drought on the response by fish to a habitat restoration ex-
Lepori, F., Palm, D., Brännäs, E., Malmqvist, B., 2005. Does restoration of structural
periment. Restor. Ecol. 13, 39–48.
heterogeneity in streams enhance fish and macroinvertebrate diversity? Ecol. Appl.
Borenstein, M., Hedges, L.V., Higgins, J.P., Rothstein, H.R., 2009. Introduction to Meta-
15, 2060–2071.
analysis. Wiley, Chichester, UK.
Li, K., He, C., Zhuang, J., Zhang, Z., Xiang, H., Wang, Z., Yang, H., Sheng, L., 2014. Long-
Brederveld, R.J., Jähnig, S.C., Lorenz, A.W., Brunzel, S., Soons, M.B., 2011. Dispersal as a
term changes in the water quality and macroinvertebrate communities of a sub-
limiting factor in the colonization of restored mountain streams by plants and mac-
tropical river in South China. Water 7, 63–80.
roinvertebrates. J. Appl. Ecol. 48, 1241–1250.
Lorenz, A.W., Korte, T., Sundermann, A., Januschke, K., Haase, P., 2012. Macrophytes
Brosse, S., Arbuckle, C.J., Townsend, C.R., 2003. Habitat scale and biodiversity: influence
respond to reach-scale river restorations. J. Appl. Ecol. 49, 202–212.
of catchment, stream reach and bedform scales on local invertebrate diversity.
Lorenz, A.W., Stoll, S., Sundermann, A., Haase, P., 2013. Do adult and YOY fish benefit
Biodivers. Conserv. 12, 2057–2075.
from river restoration measures? Ecol. Eng. 61, 174–181.
Bukaveckas, P.A., 2007. Effects of channel restoration on water velocity, transient sto-
Louhi, P., Vehanen, T., Huusko, A., Mäki-Petäys, A., Muotka, T., 2016. Long-term mon-
rage, and nutrient uptake in a channelized stream. Environ. Sci. Technol. 41,
itoring reveals the success of salmonid habitat restoration. Can. J. Fish. Aquat. Sci.
1570–1576.
73, 1733–1741.
Cheung, M.W.L., 2014. Modeling dependent effect sizes with three-level meta-analyses: a
Matthaei, C.H., Peacock, K.A., Townsend, C.R., 1999. Scour and fill patterns in a New
structural equation modeling approach. Psychol. Methods 19, 211–229.
Zealand stream and potential implications for invertebrate refugia. Freshwater Biol.
Cheung, M.W.L., 2015. metaSEM: an R package for meta-analysis using structural
42, 41–57.
equation modeling. Front. Psychol. 5, 1521.
McManamay, R.A., Orth, D.J., Dolloff, C.A., 2013. Macroinvertebrate community re-
Chung, K.C., Burns, P.B., Kim, H.M., 2006. A practical guide to meta-analysis. J. Hand
sponses to gravel addition in a southeastern regulated river. Southeast. Nat. 12,
Surg. 31, 1671–1678.
599–618.
de Jong, M.V.Z., Cowx, I.G., 2016. Long-term response of salmonid populations to habitat
McMillan, S.K., Tuttle, A.K., Jennings, G.D., Gardner, A., 2014. Influence of restoration
restoration in a boreal forest stream. Ecol. Eng. 91, 148–157.
age and riparian vegetation on reach-scale nutrient retention in restored urban
Del Re, A.C., 2013. compute. es: Compute effect sizes. R package version 0.2, 2.
streams. J. Am. Water Resour. As. 50, 626–638.
Dolph, C.L., Eggert, S.L., Magner, J., Ferrington Jr., L.C., Vondracek, B., 2015. Reach-
MEA, 2005. Ecosystem and Human Well-being: Biodiversity Synthesis. World Resources
scale stream restoration in agricultural streams of southern Minnesota alters struc-
Institute, Washington, DC.
tural and functional responses of macroinvertebrates. Freshw. Sci. 34, 535–546.
Miller, S.W., Budy, P., Schmidt, J.C., 2010. Quantifying macroinvertebrate responses to
Downs, P.W., Singer, M.S., Orr, B.K., Diggory, Z.E., Church, T.C., 2011. Restoring eco-
in-stream habitat restoration: applications of meta-analysis to river restoration.
logical integrity in highly regulated rivers: the role of baseline data and analytical
Restor. Ecol. 18, 8–19.
references. Environ. Manage. 48, 847–864.
Moreno-Mateos, D., Power, M.E., Comín, F.A., Yockteng, R., 2012. Structural and func-
Egger, M., Smith, G.D., Schneider, M., Minder, C., 1997. Bias in meta-analysis detected by
tional loss in restored wetland ecosystems. PLoS Biol. 10, e1001247.
a simple, graphical test. BMJ 315, 629–634.
Mueller, M., Pander, J., Geist, J., 2014. The ecological value of stream restoration mea-
Erwin, S.O., Jacobson, R.B., Elliott, C.M., 2017. Quantifying habitat benefits of channel
sures: an evaluation on ecosystem and target species scales. Ecol. Eng. 62, 129–139.
reconfigurations on a highly regulated river system, Lower Missouri River, USA. Ecol.
Muotka, T., Paavola, R., Haapala, A., Novikmec, M., Laasonen, P., 2002. Long-term re-
Eng. 103, 59–75.
covery of stream habitat structure and benthic invertebrate communities from in-
Feld, C.K., Hering, D., 2007. Community structure or function: effects of environmental
stream restoration. Biol. Conserv. 105, 243–253.
stress on benthic macroinvertebrates at different spatial scales. Freshwater Biol. 52,
Neale, M.W., Moffett, E.R., 2016. Re-engineering buried urban streams: daylighting re-
1380–1399.
sults in rapid changes in stream invertebrate communities. Ecol. Eng. 87, 175–184.
Friberg, N., Baattrup-Pedersen, A., Kristensen, E.A., Kronvang, B., Larsen, S.E., Pedersen,
Nilsson, C., Aradottir, A.L., Hagen, D., Halldórsson, G., Høegh, K., Mitchell, R.J., Raulund-
M.L., Skriver, J., Thodsen, H., Wiberg-Larsen, P., 2014. The River Gelså restoration
Rasmussen, K., Svavarsdóttir, K., Tolvanen, A., Wilson, S.D., 2016. Evaluating the
revisited: habitat specific assemblages and persistence of the macroinvertebrate
process of ecological restoration. Ecol. Soc. 21, 41.
community over an 11-year period. Ecol. Eng. 66, 150–157.
Ock, G., Gaeuman, D., McSloy, J., Kondolf, G.M., 2015. Ecological functions of restored
Giling, D.P., Mac Nally, R., Thompson, R.M., 2016. How sensitive are invertebrates to
gravel bars, the Trinity River, California. Ecol. Eng. 83, 49–60.
riparian-zone replanting in stream ecosystems? Mar. Freshwater Res. 67, 1500–1511.
Olden, J.D., Konrad, C.P., Melis, T.S., Kennard, M.J., Freeman, M.C., Mims, M.C., Bray,
Gurevitch, J., Hedges, L.V., 1999. Statistical issues in ecological meta-analyses. Ecology
E.N., Gido, K.B., Hemphill, N.P., Lytle, D.A., McMullen, L.E., Pyron, M., Robinson,
80, 1142–1149.
C.T., Schmidt, J.C., Williams, J.G., 2014. Are large-scale flow experiments informing
Haase, P., Hering, D., Jähnig, S.C., Lorenz, A.W., Sundermann, A., 2013. The impact of
the science and management of freshwater ecosystems? Front. Ecol. Environ. 12,
hydromorphological restoration on river ecological status: a comparison of fish,
176–185.
benthic invertebrates, and macrophytes. Hydrobiologia 704, 475–488.
Paillex, A., Schuwirth, N., Lorenz, A.W., Januschke, K., Peter, A., Reichert, P., 2017.
Hahn, S.S., 1982. Stream channelization: effect on stream fauna. In: Greeson, P.E. (Ed.),
Integrating and extending ecological river assessment: concept and test with two
Biota and Biological Principles of the Aquatic Environment. USGS: Circular, pp.
restoration projects. Ecol. Indic. 72, 131–141.
43–49.
Palmer, M.A., Bernhardt, E.S., Allan, J.D., Lake, P.S., Alexander, G., Brooks, S., Carr, J.,
Hasselquist, E.M., Nilsson, C., Hjältén, J., Jørgensen, D., Lind, L., Polvi, L.E., 2015. Time
Clayton, S., Dahm, C.N., Follstad Shah, J., Galat, D.L., Loss, S.G., Goodwin, P., Hart,
for recovery of riparian plants in restored northern Swedish streams: a chronose-
D.D., Hassett, B., Jenkinson, R., Kondolf, G.M., Lave, R., Meyer, J.L., O’Donnell, T.K.,
quence study. Ecol. Appl. 25, 1373–1389.
Pagano, L., Sudduth, E., 2005. Standards for ecologically successful river restoration.

36
W. Lu, et al. Ecological Engineering 136 (2019) 30–37

J. Appl. Ecol. 42, 208–217. Sand-Jensen, K., Riis, T., Markager, S., Vincent, W.F., 1999. Slow growth and decom-
Palmer, M.A., Menninger, H.L., Bernhardt, E., 2010. River restoration, habitat hetero- position of mosses in Arctic lakes. Can. J. Fish. Aquat. Sci. 56, 388–393.
geneity and biodiversity: a failure of theory or practice? Freshwater Biol. 55, Sarriquet, P.E., Bordenave, P., Marmonier, P., 2007. Effects of bottom sediment restora-
205–222. tion on interstitial habitat characteristics and benthic macroinvertebrate assemblages
Palmer, M.A., Hondula, K.L., Koch, B.J., 2014. Ecological restoration of streams and in a headwater stream. River Res. Appl. 23, 815–828.
rivers: shifting strategies and shifting goals. Annu. Rev. Ecol. Evol. S. 45, 247–269. Scheiner, S.M., Gurevitch, J., 2001. Design and Analysis of Ecological Experiments.
Pan, B., Yuan, J., Zhang, X., Wang, Z., Chen, J., Lu, J., Yang, W., Chen, J., Li, Z., Zhao, N., Oxford University Press, Oxford.
Xu, M., 2016. A review of ecological restoration techniques in fluvial rivers. Int. J. Schmutz, S., Kremser, H., Melcher, A., Jungwirth, M., Muhar, S., Waidbacher, H., Zauner,
Sediment Res. 31, 110–119. G., 2014. Ecological effects of rehabilitation measures at the Austrian Danube: a
Pander, J., Geist, J., 2013. Ecological indicators for stream restoration success. Ecol. meta-analysis of fish assemblages. Hydrobiologia 729, 49–60.
Indic. 30, 106–118. Stoll, S., Kail, J., Lorenz, A.W., Sundermann, A., Haase, P., 2014. The importance of the
Pander, J., Mueller, M., Geist, J., 2015. A comparison of four stream substratum re- regional species pool, ecological species traits and local habitat conditions for the
storation techniques concerning interstitial conditions and downstream effects. River colonization of restored river reaches by fish. PLoS One 9, e84741.
Res. Appl. 31, 239–255. Sundermann, A., Stoll, S., Haase, P., 2011. River restoration success depends on the
Pedersen, M.L., Friberg, N., Skriver, J., Baattrup-Pedersen, A., Larsen, S.E., 2007. species pool of the immediate surroundings. Ecol. Appl. 21, 1962–1971.
Restoration of Skjern River and its valley short-term effects on river habitats, mac- Testa, S.I.I.I., Douglas Shields Jr., F., Cooper, C.M., 2011. Macroinvertebrate esponse to
rophytes and macroinvertebrates. Ecol. Eng. 30, 145–156. stream restoration by large wood addition. Ecohydrology 4, 631–643.
Reeves, D.B., Tate, W.B., Jelks, H.L., Jordan, F., 2016. Response of imperiled Okaloosa Tockner, K., Uehlinger, U., Robinson, C.T., 2009. Rivers of Europe. Academic Press,
Darters to stream restoration. N. Am. J. Fish. Manage. 36, 1375–1385. Heidelberg.
Renöfält, B.M., Lejon, A.G., Jonsson, M., Nilsson, C., 2013. Long-term taxon-specific re- Tullos, D.D., Penrose, D.L., Jennings, G.D., Cope, W.G., 2009. Analysis of functional traits
sponses of macroinvertebrates to dam removal in a mid-sized Swedish stream. River in reconfigured channels: implications for the bioassessment and disturbance of river
Res. Appl. 29, 1082–1089. restoration. J. N. Am. Benthol. Soc. 28, 80–92.
Rey Benayas, J.M., Newton, A.C., Diaz, A., Bullock, J.M., 2009. Enhancement of biodi- Valentina, S.S., Momir, P., Branko, M., Vasil, K., Milica, R., Donka, M., 2011. Water
versity and ecosystem services by ecological restoration: a meta-analysis. Science quality assessment based on the macroinvertebrate fauna-the Pcinja River case study.
325, 1121–1124. Water Res. Manage. 1, 63–69.
Ribot, M., Bernal, S., Nikolakopoulou, M., Vaessen, T.N., Cochero, J., Gacia, E., Sorolla, Vehanen, T., Huusko, A., Mäki-Petäys, A., Mykrä, H., Muotka, T., 2010. Effects of habitat
A., Argerich, A., Sabater, F., Isnard, M., Martí, E., 2017. Enhancement of carbon and rehabilitation on brown trout (Salmo trutta) in boreal forest streams. Freshwater Biol.
nitrogen removal by helophytes along subsurface water flowpaths receiving treated 55, 2200–2214.
wastewater. Sci. Total Environ. 599, 1667–1676. Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. J. Stat.
Rosi-Marshall, E.J., Moerke, A.H., Lamberti, G.A., 2006. Ecological responses to trout Softw. 36, 1–48.
habitat rehabilitation in a Northern Michigan stream. Environ. Manage. 38, 99–107. Whiteway, S.L., Biron, P.M., Zimmermann, A., Venter, O., Grant, J.W., 2010. Do in-stream
Rubin, Z., Kondolf, G.M., Rios-Touma, B., 2017. Evaluating stream restoration projects: restoration structures enhance salmonid abundance? A meta-analysis. Can. J. Fish.
what do we learn from monitoring? Water 9, 174. Aquat. Sci. 67, 831–841.
Saldi-Caromile, K., Bates, K., Skidmore, P.A., Barenti, J., Pineo, D., 2004. Stream habitat Woolsey, S., Weber, C., Gonser, T., Hoehn, E., Hostmann, M., Junker, B., Roulier, C.,
restoration guidelines: Final draft. Co-published by the Washington Departments of Schweizer, S., Tiegs, S., Tockner, K., Peter, A., 2005. Handbook for Evaluating
Fish and Wildlife and Ecology and the U.S. WA: Fish and Wildlife Service. Rehabilitation Projects in Rivers and Streams. EAWAG/WSL, Zurich.
Sand-Jensen, K., 1997. Macrophytes as biological engineers in the ecology of Danish Wu, J., Cheng, S., Li, Z., Guo, W., Zhong, F., Yin, D., 2013. Case study on rehabilitation of
streams. In: Sand-Jensen, K., Pedersen, O. (Eds.), Freshwater Biology. Priorities and a polluted urban water body in Yangtze River Basin. Environ. Sci. Pollut. Res. 20,
Development in Danish Research. G.E.C. Gad, Copenhagen, pp. 74–101. 7038–7045.

37

You might also like