Received: 21 May 2023 | Revised: 4 December 2023 | Accepted: 12 December 2023

DOI: 10.1111/jvs.13228

Journal of Vegetation Science

RESEARCH ARTICLE

Environmental filtering is the primary driver of community

assembly in forest–grassland mosaics: A case study based on
CSR strategies

László Erdős1,2 | Khanh Vu Ho3,4 | Ákos Bede-­Fazekas1,5 | György Kröel-­Dulay1 |

Csaba Tölgyesi6 | Zoltán Bátori6,7 | Péter Török2,8,9
1
Institute of Ecology and Botany, HUN-­REN Centre for Ecological Research, Vácrátót, Hungary
2
HUN-­REN-­UD Functional and Restoration Ecology Research Group, Debrecen, Hungary
3
Doctoral School of Environmental Sciences, University of Szeged, Szeged, Hungary
4
Faculty of Natural Resources-­Environment, Kien Giang University, Kien Giang, Vietnam
5
Department of Environmental and Landscape Geography, Institute of Geography and Earth Sciences, ELTE Eötvös Loránd University, Budapest, Hungary
6
MTA-­SZTE ‘Momentum’ Applied Ecology Research Group, University of Szeged, Szeged, Hungary
7
Department of Ecology, University of Szeged, Szeged, Hungary
8
Department of Ecology, University of Debrecen, Debrecen, Hungary
9
Polish Academy of Sciences, Botanical Garden – Center for Biological Diversity Conservation in Powsin, Warszawa, Poland

Correspondence
Khanh Vu Ho, Doctoral School of Abstract
Environmental Sciences, University of
Aims: Ecological strategies can provide information about plant community assembly
Szeged, Szeged, Hungary; Faculty of
Natural Resources-­Environment, Kien and its main drivers. Our aim was to reveal the dominant strategies of the vegetation
Giang University, Kien Giang, Vietnam.
types of forest–grassland mosaics and to deduce the assembly processes responsible
Email: [email protected] and ho.
[email protected] for their species composition.
Location: Hungary.
Funding information
The National Research, Development Methods: We investigated eight vegetation types of Hungarian forest–steppes. The
and Innovation Office, Hungary (grant
trade-­off between three key traits related to leaf size and economics was used to
number FK 134384 for LE, K 124796 and
FK 142428 for ZB, and K 137573 and KKP calculate Grime's competitive–stress tolerance–ruderal (CSR) value for each species,
144068 for PT); University of Szeged
based on which the mean value for each vegetation type was determined. Detrended
Open Access Fund (Grant number: 6571);
The New National Excellence Program of correspondence analysis (DCA) ordination was used to reveal the compositional dif-
the Ministry for Culture and Innovation
ferences among the vegetation types under study. To analyze how ecological strate-
from the source of the National Research,
Development and Innovation Fund, gies correlate with the compositional gradient, we used linear regression between
Hungary (ÚNKP-­22-­5-­SZTE-­538 for ZB)
plot ordination scores (the first DCA scores) and each strategy (C, S, and R). Linear
Co-­ordinating Editor: Marcos Carlucci mixed-­effect models were used to evaluate the differences between the vegetation
types regarding each strategy (C, S, and R).
Results: Each vegetation type was dominated by the stress-­tolerator strategy, indicat-
ing the prominent role of environmental filtering in community assembly. However,
ecological strategies differed significantly among the communities. The importance of

László Erdős and Khanh Vu Ho contributed equally to this work and are joint first authors.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in
any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
© 2024 The Authors. Journal of Vegetation Science published by John Wiley & Sons Ltd on behalf of International Association for Vegetation Science.

J Veg Sci. 2024;35:e13228.  wileyonlinelibrary.com/journal/jvs | 1 of 10

https://doi.org/10.1111/jvs.13228
 | Journal of Vegetation Science

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 of 10 ERDŐS et al.

the stress-­tolerator strategy decreased toward the less harsh end of the gradient (i.e.,
from grasslands to forests), while the competitor strategy showed a reverse pattern.
The ruderal strategy was weakly correlated with the gradient, although its proportion
increased toward the harsh end of the gradient.
Conclusions: With ongoing climate change, an increasing importance of environmen-
tal filtering is expected in the assembly of the vegetation types in the studied for-
est–grassland mosaics. We suggest that CSR strategies offer a useful tool for studying
plant-­community assembly rules along environmental gradients.

KEYWORDS
competitor, CSR, environmental filter, forest–steppe, leaf trait, ruderal, stress-­tolerator

1 | I NTRO D U C TI O N As emphasized by Grime and Pierce (2012), there are no living

beings that would be exclusively C-­, S-­, or R-­selected. For example,
Plant community assembly has long been at the focus of ecological a species that shows a high degree of C-­selection also has to cope
research and remains an issue under intensive scientific discussion with some level of S-­ and R-­selection. This means that in reality, in-
(Götzenberger et al., 2012; Dias et al., 2020). Assembly rules determine dividuals have to pass all three components of the CSR-­filter, even
how species of a regional species pool are selected and fit together to though each component may represent a greater or smaller obstacle.
form local communities (Menezes et al., 2020). In addition to its theo- The filter proposed by Grime and Pierce (2012) selects individu-
retical importance, knowledge of assembly processes has outstanding als possessing traits that are directly beneficial to competitive ability,
practical implications as well: it can help to predict plant communities' stress tolerance, and the survival of the population by completing
responses to environmental changes and to restore (near-­)natural com- the individual life cycle between two destructive events. These
munities (e.g., Temperton et al., 2004; Münkemüller et al., 2020). traits (i.e., traits that are directly connected to the CSR strategies)
The assembly of plant communities is usually represented as a display a plant's general strategy. Consequently, by studying the
series of various filters that define which traits (and therefore, which traits and strategies of the species composing a plant community,
species possessing these traits) can enter the realized local plant com- we can gather information about the primary drivers that determine
munity (e.g., Keddy, 1992). The most widespread model includes three community composition.
filters acting in concert: the dispersal filter determines which species In the forest–steppe zone, which is a transitional zone between
arrive at the site, whereas the environmental filter and the biotic fil- closed-­c anopy forests and steppes (Erdős et al., 2018a), it is as-
ter select species that can tolerate the local abiotic factors and the sumed that forest and grassland patches, in most cases, repre-
biotic interactions from the co-­existing species respectively (e.g., sent alternative stable states, that is, they appear under the very
Götzenberger et al., 2012; Hulvey & Aigner, 2014; Halassy et al., 2016). same primary environmental conditions and are stable in time
Grime and Pierce (2012) proposed a different scheme, based (Petraitis, 2013; Erdős et al., 2023). However, secondary differ-
on three basic ecological phenomena that shape vegetation: com- ences emerge among the patches, evoked by the vegetation itself,
petition, stress, and disturbance. According to Grime (1974, 1977) further stabilizing the pattern of the individual patches. For exam-
and Grime and Pierce (2012), competition means that co-­occurring ple, trees and shrubs increase the humus and moisture content of
individuals strive to capture the same units of resource, stress is un- the upper soil and mitigate daily temperature extremes, changes
derstood as environmental constraints that limit production, while that favor the continuous existence of the forest and hinder the
disturbance is the partial or complete destruction of biomass. Grime establishment of grassland species (Erdős et al., 2023). As a result,
and Pierce (2012) argue that every plant species faces an evolu- significant environmental differences have been revealed among
tionary trade-­off among (1) developing strong competitive ability the various plant communities, which can be arranged along an
(competitors, C), (2) withstanding environmental stress (stress tol- environmental harshness gradient (Erdős et al., 2018b, 2020; Ho
erators, S), and (3) enduring regular biomass destruction (ruderals, et al., 2024). Generally, daytime temperatures in the growing sea-
R). According to this view, plants have to pass a filter that favors son are low in large forest patches and increase through smaller-­
competitors, stress tolerators, or ruderals in productive, harsh, or sized forest patches and edges to grasslands, which are the
disturbed environments respectively. Grime and Pierce (2012) also hottest habitats, especially in summer. Nighttime temperatures
stressed that there is no hierarchy among the different components show a reverse trend, and thus, large forests have the lowest and
of the filter; that is why their model includes a single filter instead of grasslands the highest daily temperature fluctuations. Both the
a series of filters. moisture content of the upper soil layer and daytime air humidity
 |

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ERDŐS et al. 3 of 10
Journal of Vegetation Science

are the highest in large forest patches and become progressively 10.3–10.5°C and an average annual precipitation of 520–550 mm
lower toward smaller-­sized forest patches and edges, and are the (Tölgyesi et al., 2016). Grasslands have humus-­poor sandy soil with
lowest in grasslands. low water retention capacity (Várallyay, 1993); however, the humus
In this study, our aim was to reveal the dominant strategies of the content is slightly higher in forest patches (Erdős et al., 2014).
vegetation types of sandy forest–steppes, and infer the assembly The area belongs to the westernmost outposts of the forest–
processes responsible for their species composition. We hypoth- steppe zone. Although the overwhelming majority of forest–steppe
esized that in the more productive environment of forest patches, ecosystems of the region have been converted to agricultural fields
community composition would be driven by competitive exclusion, or tree plantations (Molnár et al., 2012), a few sandy forest–steppe
while under the harsher conditions of open grasslands, stress (i.e., areas have survived and are currently under legal protection. Within
environmental filtering) would be the most important driver — ex- the study sites, the vegetation is characterized by a mosaic-­like pat-
pressed also in the composition of traits. tern of forest and grassland patches (Figure 1b). The variously sized
forest patches of poplar–juniper stands (Junipero-­Populetum albae)
exhibit a cover of 50%–80% and are dominated by Populus alba with
2 | M ATE R I A L S A N D M E TH O DS heights ranging from 10 m to 15 m. In the shrub layer, typical spe-
cies include Berberis vulgaris, Crataegus monogyna, Juniperus com-
2.1 | Study area munis, Ligustrum vulgare, and Prunus spinosa, and their cover ranges
from 5% to 80%. The herb layer hosts species such as Carex lipa-
The study was carried out in the Kiskunság Sand Ridge in central rocarpos, Cynoglossum officinale, Euphorbia cyparissias, and Teucrium
Hungary, a lowland area in the heart of the Carpathian Basin be- chamaedrys. Forest patches range in size from a few dozen square
tween the rivers Danube and Tisza. We selected thirteen legally pro- meters to as large as 1 ha.
tected sites with near-­natural vegetation (Figure 1a; Appendix S1). There are three types of grasslands in the study sites. The
The subcontinental climate with sub-­Mediterranean influences is closed perennial grassland (Astragalo austriacae-­
Festucetum rupi-
characteristic of this region with an average annual temperature of colae), which has a relatively high cover (typically more than 80%),

F I G U R E 1 (a) The location of Hungary

in Europe (brown) and thirteen study sites
(red dots) in the Kiskunság Sand Ridge
(gray) in central Hungary. (b) A mosaic of
forests and grasslands in the study region.
 | Journal of Vegetation Science

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 of 10 ERDŐS et al.

is dominated by Festuca rupicola, Stipa borysthenica, and Stipa cap- and specific leaf area (SLA, mm2/mg), based on the observation that
illata. Other typical species are, among others, Achillea pannonica, species with the C strategy invest resources in increasing LA; spe-
Calamagrostis epigejos, Euphorbia cyparissias, and Poa angustifolia. cies with the S strategy invest in retaining LDMC; and species with
The most widespread grassland type is the open perennial grass- the R strategy invest primarily in the ability to increase SLA) (Pierce
land (Festucetum vaginatae), which is dominated by Festuca vaginata, et al., 2017). It is widely accepted that these traits strongly represent
Stipa borysthenica, and Stipa capillata. The overall total vascular veg- the leaf economics and plant size spectra (sensu the global spectrum
etation cover varies between 40% and 70%. Other typical species of plant form and function; Díaz et al., 2016). Trait data were ex-
include Alkanna tinctoria, Centaurea arenaria, Koeleria glauca, and tracted from Hungarian databases (Lhotsky et al., 2016a, 2016b; E-­
Syrenia cana. Vojtkó et al., 2020; Gyalus et al., 2022; McIntosh-­Buday et al., 2022).
The open annual grassland vegetation (Secali sylvestris-­Brometum Data were retrieved from PADAPT, the Pannonian Database of Plant
tectorum) has a vascular vegetation cover of 20%–50% and is domi- Traits (Sonkoly et al., 2023). However, eight of the 289 taxa (2.77%)
nated by Bromus tectorum and Secale sylvestre. Other common spe- did not have SLA data in this database. Therefore, we used the LEDA
cies include Bromus squarrosus, Poa bulbosa, Silene conica, and Viola database published by Kleyer et al. (2008) to provide missing values
arvensis. for these species.
Plant species nomenclature is in accordance with Király (2009), We calculated the strategy (separate C, S, and R values) for each
while plant association names are based on Borhidi et al. (2012). species based on the trade-­off among the three above leaf traits by
the “StrateFy” tool, which regresses trait values against the princi-
pal component analysis (PCA) axes extracted from global leaf traits
2.2 | Field sampling (Pierce et al., 2017). This method enables determining CSR values for
a wide range of vascular plant species globally (Pierce et al., 2017).
We distinguished and analyzed eight vegetation types at each site: Based on the C, S, and R components of the strategy for each spe-
large forest patches (>0.5 ha), medium forest patches (0.2–0.4 ha), cies, we calculated both the unweighted and weighted mean C, S,
small forest patches (<0.1 ha), north-­facing forest edges, south-­ and R values for each plot, using presence/absence data for the
facing forest edges, closed perennial grasslands, open perennial former situation and square-­root-­transformed cover scores for the
grasslands, and open annual grasslands. Based on the earlier stud- latter one.
ies of Erdős et al. (2018b, 2020), we used 5 m × 5 m plots for sam-
pling forest interiors and grasslands, while 2 m × 12.5 m plots were
employed to study forest edges, preventing their extension into the 2.4 | Data analysis
interiors of the adjacent vegetation types. In this study, we defined
an edge as the area outside of the outermost tree trunks but still To analyze the differences in species composition among the veg-
under the canopy. Edge plots were established along the relatively etation types, we applied detrended correspondence analysis (DCA),
straight peripheral zones of forest patches that were larger than 0.2 which was performed on square-­root-­transformed cover scores.
hectares in size. Detrending was performed using the default number of rescaling
A total of 494 plots were used: 60 plots in large forest patches, cycles (4) and segments (26). The first DCA axis was used for the
64 plots in medium forest patches, 60 plots in open annual grass- quantification of vegetation types along the compositional gradient,
lands, 50 plots in closed perennial grasslands, and an equal count which provides a continuous number interpretable as “compositional
of 65 plots in each of the other vegetation types (Appendix S1). The similarity to open grassland” or “compositional dissimilarity to a large
number of replicates was determined by the lack or rarity of cer- forest.” The analysis was conducted using the vegan package in R
tain vegetation types at some study sites. We visually estimated the version 4.1.2 (R Core Team, 2021; Oksanen et al., 2022).
cover of all vascular plant species in all vegetation layers (canopy, For all vegetation types, the strategy values (C, S and R) per plot
shrub, and herb) in each plot in spring (April–May) and summer (July– were visualized using a ternary graph, which was created using the
August). The highest cover value for each species was then applied ggtern package in R (Hamilton & Ferry, 2018).
for subsequent data analyses. Seven taxa unidentified at the species To reveal how ecological strategies (C, S, and R) correlate with
level (Acer sp., Epipactis sp., Fraxinus sp., Hieracium sp., Lathyrus sp., the gradient, we used linear regression between plot ordination
Prunus sp., and Silene sp., none of which was found in more than scores (the first DCA scores) and each strategy (C, S, and R). The
three of the 494 plots), were excluded from the analyses involving models were visually checked using the diagnostic plots.
strategy. To evaluate the differences between the vegetation types re-
garding each strategy (C, S, and R), we used linear mixed-­effect
models. In our modeling approach, the site was treated as a random
2.3 | Leaf traits and ecological strategies factor, while the vegetation type was considered a fixed factor.
We applied the glmmTMB package in R to construct the models,
To determine the strategy of the species, three leaf traits were using the Gaussian family distribution (Brooks et al., 2017). The
used: leaf area (LA, mm2), leaf dry matter content (LDMC, mg/g), models were visually checked using the performance package in
 |

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ERDŐS et al. 5 of 10
Journal of Vegetation Science

R (Lüdecke et al., 2021). In order to find a significant proportion follows: component C (χ2 = 1637, p < 0.001), component S (χ2 = 242,
of variability, an analysis of variance was performed on the linear p < 0.001), and component R (χ2 = 123, p < 0.001). Based on the
mixed-­effect models. Afterwards, we used the emmeans package pairwise comparisons (Appendix S3), the highest component C was
to perform pairwise comparisons among the levels of the fixed found in large and medium forest patches, and it gradually decreased
factor and adjusted the p-­values using the Bonferroni method toward the end of the vegetation gradient (Figure 4d). Component
(Lenth, 2022). S depicted a gradually increasing trend from large forest patches
to open annual grasslands (Figure 4e). Open perennial and annual
grasslands exhibited the highest component R, whereas north-­facing
3 | R E S U LT S edges displayed the lowest component R (Figure 4f). Again, patterns
were similar with weighted values, although some differences did
The DCA ordination revealed a compositional gradient along emerge (Appendix S4). For example, component S had a more equal
the first DCA axis with the following order of vegetation types: presence along the gradient for weighted than for unweighted val-
large forest patches — medium forest patches — small forest ues, and the relationship was more hump-­backed. In addition, there
patches — north-­facing edges — south-­facing edges — closed peren- was a more pronounced change of component R along the gradient
nial grasslands — open perennial grasslands — open annual grass- for weighted than for unweighted values.
lands (Figure 2). It demonstrated that the first DCA scores can be
used in the subsequent analyses and interpreted as a compositional
gradient. 4 | DISCUSSION
The ternary plot showing unweighted values revealed that the av-
erage CSR values were located along the RS-­axis with a smaller con- In this work, we studied the vegetation types of forest–grassland
tribution from component C (Figure 3a). The contributions of mean mosaics in eastern Central Europe. In the framework of the CSR
components S and R were 50%–60% and 20%–30% respectively. theory (Grime & Pierce, 2012), it is possible to infer community as-
The study found a smaller contribution from mean component C sembly processes from the strategies of plant communities. For ex-
(<25%). When taking a closer look (Figure 3b), a separation between ample, if a plant community is dominated by the competitor strategy,
values across vegetation types was found, forming different groups: this suggests that the competition filter is the most notable obstacle
large and medium forest patches belonged to one group, small for- for individuals to enter the realized local plant community. However,
est patches, north-­ and south-­facing edges, and closed grasslands as already noted in the Introduction, no species can be regarded as
formed another group, and open perennial and open annual grass- exclusively C-­, S-­, or R-­selected, which also applies to communities.
lands formed the third group. Generally, the ternary plot based on Thus, a given plant community that is dominated by the competi-
weighted values showed a rather similar pattern (Appendix S2). tor strategy also has a certain level of environmental stress and dis-
The scores of the sample plots on the primary ordination axis turbance. Moreover, there may be additional drivers shaping plant
were negatively associated with the C strategy (Figure 4a) but pos- communities, such as dispersal, which is not assessed by the CSR
itively associated with the S strategy (Figure 4b). A weak but posi- approach used in the present work.
tive relation was observed between the R strategy and DCA1 scores The eight studied vegetation types formed a compositional gra-
(Figure 4c). The vegetation type affected each type of strategy as dient (Figure 2). Generally, each vegetation type was dominated

F I G U R E 2 DCA ordination scattergram of the 494 plots. Large symbols indicate the centroids for each vegetation type. Ordination
ellipses were drawn based on standard deviation of point scores, where the directions of the major axes of the ellipses were defined by the
weighted correlation. CG, closed perennial grasslands; LF, large forest patches; MF, medium forest patches; NE, north-­facing forest edges;
OA, open annual grasslands; OP, open perennial grasslands; SE, south-­facing forest edges; SF, small forest patches.
 | Journal of Vegetation Science

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 of 10 ERDŐS et al.

F I G U R E 3 Ternary plots showing unweighted mean values of CSR strategies for the eight vegetation types. The red triangle in plot
(a) shows the boundaries of plot (b). Larger symbols indicate the mean value for each vegetation type. Vegetation type abbreviations are
according to the caption of Figure 2.

F I G U R E 4 Relationships between plot scores on the primary DCA ordination axis and each unweighted strategy component: (a) C; (b)
S, and (c) R; p-­value and adjusted R-­squared were calculated using linear regression; Slope: the slope value of the regression line; the blue
line is the regression line, and the gray area around the line represents the 95% confidence interval. Box plots demonstrate the variability
of each strategy component: (d) C, (e) S, and (f) R in the eight communities. Those that do not share a letter are significantly different at the
significance level of α = 0.05. Vegetation type abbreviations are according to the caption of Figure 2.

by the stress-­tolerator strategy (Figure 3a), which reflects the rela- high interannual variations (from <350 mm in some years to >800 mm
tively harsh environmental conditions prevailing in the study region. in others) (Tölgyesi et al., 2016). The very low water retention capac-
According to Grime and Pierce (2012), stress-­tolerator plant species ity of the sandy soils in the region, alongside their low humus content
have an advantage over other species in unproductive and variable (Várallyay, 1993) further increase the environmental stress.
environments. In the Kiskunság Sand Ridge, most species encounter Although each vegetation type examined in the present
a harsh environment, as the amount of precipitation is low and shows work was dominated by the stress-­tolerator strategy, ecological
 |

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ERDŐS et al. 7 of 10
Journal of Vegetation Science

strategies differed markedly among the studied vegetation types smaller forest patches and forest edges have a limited ability to buf-
(Figures 3b and 4). This is similar to the findings of Rosenfield fer environmental stress (Erdős et al., 2023), that is, they are harsher
et al. (2019), who revealed distinct plant strategies along a rela- than the larger forest patches.
tively short gradient in South America. In our study region, com- Focusing on the three grassland types, competition seems to
petition proved to be a relatively important force in community be more important and stress tolerance slightly less important in
assembly in large and medium forest patches, while its impor- closed grasslands than in open ones. Similarly, comparing differ-
tance progressively diminished along the gradient toward the ent grasslands of the Tibetan Plateau, Yu et al. (2022) showed that
grasslands. Competition seemed to play the most subordinate the importance of competition increased and the importance of
role in the open grassland vegetation. The stress-­tolerator strat- stress tolerance decreased with increasing productivity. Kelemen
egy showed a reverse trend. Our environmental measurements et al. (2013) compared various lowland grassland communities and
(Erdős et al., 2018b; Ho et al., 2024) suggest that the forests of the found that competition was more important in stable and productive
study region are more productive and less harsh than grasslands, environments, whereas stress-­tolerator species gained advantage in
with forest edges typically providing intermediate environments. alkaline and highly variable environments. The fact that stress toler-
Trees and shrubs reduce environmental stress by providing rela- ance is more important in open grasslands than in closed ones may
tively cool and humid circumstances under the canopy during the reflect the harsher conditions in the former types (Ho et al., 2024).
growing season, including the hot and dry months of late sum- In closed grasslands, there is probably intensive competition both
mer. Also, the canopy reduces daily temperature variation and for light (above ground) and water and nutrients (below ground). In
mitigates extremes. In addition, forests have increased soil mois- contrast, the widely spaced individuals in open grasslands probably
ture and improved soil humus content compared to grasslands. experience less competition. Open annual grasslands, in particular,
Thus, our results are consistent with the predictions of Grime contain individual plants scattered on a relatively open surface, with
and Pierce (2012) and Adler et al. (2013), who argued that along small roots and limited leaf surfaces, suggesting weak competition.
productivity gradients, a shift in the importance of abiotic vs bi- When using unweighted values, the R strategy was only weakly
otic factors can be expected, with abiotic constraints becoming related to DCA1 scores (Figure 4c) and played the most important
more important toward the harsh end (in our case, grasslands) and role in the open grassland types (Figure 4f). With weighted values,
competition becoming more important toward the more produc- however, the R strategy was much more strongly related to DCA1
tive end of the gradient (in our case, forests). Our results also fit scores, and this strategy proved much more important in open an-
the findings of Dayrell et al. (2018), who reported that in south- nual grasslands than in open perennial grasslands (Appendix S4).
east Brazil competition dominates community assembly in for- The R strategy is typical in early successional stages, on open, re-
est patches, whereas environmental stress is more important in cently exposed surfaces (e.g., Caccianiga et al., 2006), which fits our
grasslands. Similarly, Negreiros et al. (2014) claim that grasslands, findings. The open perennial grassland (OP) has considerable open
especially those in highly unproductive environments, tend to be sand surfaces between the dominant tussock grasses, where small
dominated by the stress-­tolerator strategy. annual plants are typical, similarly to other vegetation types with an
When using weighted instead of unweighted values important contribution of component R (Li & Shipley, 2017; Pierce
(Appendix S4), the importance of stress was further emphasized, as et al., 2017). Open annual grasslands (OA) usually emerge as a result
component S was high throughout the full gradient, and differences of disturbance, most often wind erosion or trampling by grazers and
among the habitats were only moderate. Since filters have an influ- browsers (Fekete, 1992; Borhidi et al., 2012), although identifying
ence on species' abundances in a given community, this reinforces the exact disturbance agents needs further research.
our findings with unweighted values and makes our results more ro- To sum up, our hypothesis that community assembly would be
bust. The somewhat hump-­backed shape of the curve suggests that dominated by competitive exclusion in the forest patches and stress
stress loses some importance toward the end points of the gradient, (i.e., environmental filtering) would dominate in grasslands was sup-
probably because competition becomes more important in large and ported only partly. While it is true that the importance of compe-
medium forest patches, while disturbance increases in importance in tition was larger in woody habitats (i.e., forests and edges) than in
the open annual grasslands. grasslands, environmental filtering was the most important factor in
We suggest that competition for light is an important force that each of the vegetation types (Figure 5). Disturbance played the most
shapes forest communities in the study region, while competition for important role in the open perennial and the open annual grassland
water and nutrients may be more limited in this vegetation type. The communities.
herb layer is sparse and individuals are usually widely spaced, sug- A significant drying tendency has been observed in Hungary
gesting low levels of competition, especially because the upper soil during the last few decades (Jaagus et al., 2022), and the trend is
layer is relatively moist (Erdős et al., 2018b). Woody species reach projected to continue during the 21st century (Sábitz et al., 2014).
much deeper soil layers, resulting in reduced competition between As a result, we expect that vegetation types become even more
them and herbs. stressed by aridity. This may result in increasing importance of en-
We found obvious differences in strategies among the differ- vironmental filtering, which already dominates the assembly of the
ently sized forest patches and the forest edges, suggesting that plant communities in the Kiskunság Sand Ridge.
 | Journal of Vegetation Science

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 of 10 ERDŐS et al.

Török). The New National Excellence Program of the Ministry for

Culture and Innovation from the source of the National Research,
Development and Innovation Fund, Hungary (ÚNKP-­22-­5-­SZTE-­538
to Zoltán Bátori). University of Szeged Open Access Fund (Grant
number: 6571).

DATA AVA I L A B I L I T Y S TAT E M E N T

Data necessary to reproduce the analyses are deposited at Zenodo:
https://​doi.​org/​10.​5281/​zenodo.​10255319 (Erdős et al., 2023).

ORCID
László Erdős https://orcid.org/0000-0002-6750-0961
Khanh Vu Ho https://orcid.org/0000-0002-9117-7789
Ákos Bede-­Fazekas https://orcid.org/0000-0002-2905-338X
F I G U R E 5 Assembly in the eight studied vegetation types of the György Kröel-­Dulay https://orcid.org/0000-0002-0695-1232
forest–grassland mosaics in the Kiskunság Sand Ridge. The three Csaba Tölgyesi https://orcid.org/0000-0002-0770-2107
components of the filter proposed by Grime and Pierce (2012) are Zoltán Bátori https://orcid.org/0000-0001-9915-5309
shown as separate filters to increase clarity, but their order should
Péter Török https://orcid.org/0000-0002-4428-3327
not be interpreted as a hierarchy. Darker colors indicate dominating
role of a filter. C, competition; D, disturbance; S, stress. Vegetation
REFERENCES
type abbreviations are according to the caption of Figure 2.
Adler, P.B., Fajardo, A., Kleinhesselink, A.R. & Kraft, N.J.B. (2013) Trait-­
based tests of coexistence mechanisms. Ecology Letters, 16, 1294–
1306. Available from: https://​doi.​org/​10.​1111/​ele.​12157​
Similar to Catorci et al. (2011) and Han et al. (2021), we found Borhidi, A., Kevey, B. & Lendvai, G. (2012) Plant communities of Hungary.
that the CSR strategy types offer a useful tool for the study of plant-­ Budapest: Academic Press.
Brooks, M.E., Kristensen, K., van Benthem, K.J., Magnusson, A., Berg,
community assembly rules along gradients. However, a considerable
C.W., Nielsen, A. et al. (2017) glmmTMB balances speed and flex-
limitation of our study was the limited length of the gradient. Studies ibility among packages for zero-­inflated generalized linear mixed
encompassing more productive (and at the same time, less harsh) modeling. The R Journal, 9, 378–400. Available from: https://​doi.​
vegetation types (such as Lhotsky et al., 2016a, 2016b), or studies org/​10.​3929/​ethz-­​b-­​00024​0 890
Caccianiga, M., Luzzaro, A., Pierce, S., Ceriani, R.M. & Cerabolini, B.
that contain a longer section of the disturbance gradient (e.g., Pierce
(2006) The functional basis of a primary succession resolved by
et al., 2007) could provide additional information on assembly rules. CSR classification. Oikos, 112, 10–20. Available from: https://​doi.​
org/​10.​1111/j.​0 030-­​1299.​2006.​14107.​x
AU T H O R C O N T R I B U T I O N S Catorci, A., Vitanzi, A. & Tardella, F.M. (2011) Variations in CSR strate-
László Erdős designed the study, László Erdős, Khanh Vu Ho, Ákos gies along stress gradients in the herb layer of submediterranean
forests (Central Italy). Plant Ecology and Evolution, 144, 299–306.
Bede-­Fazekas, György Kröel-­Dulay, Csaba Tölgyesi, Zoltán Bátori,
Available from: https://​doi.​org/​10.​5091/​plece​vo.​2011.​621
and Péter Török conducted the field works, Khanh Vu Ho and Ákos Dayrell, R.L.C., Arruda, A.J., Pierce, S., Negreiros, D., Meyer, P.B.,
Bede-­Fazekas analyzed the data, László Erdős, Khanh Vu Ho, Ákos Lambers, H. et al. (2018) Ontogenetic shifts in plant ecological
Bede-­Fazekas, György Kröel-­Dulay, Csaba Tölgyesi, Zoltán Bátori, strategies. Functional Ecology, 32, 2730–2741. Available from:
https://​doi.​org/​10.​1111/​1365-­​2435.​13221​
and Péter Török interpreted the results, László Erdős and Khanh Vu
Dias, A.T.C., Rosado, B.H., De Bello, F., Pistón, N. & De Mattos, E.A.
Ho wrote the first draft of this manuscript, and László Erdős, Khanh (2020) Alternative plant designs: consequences for community as-
Vu Ho, Ákos Bede-­Fazekas, György Kröel-­Dulay, Csaba Tölgyesi, sembly and ecosystem functioning. Annals of Botany, 125, 391–398.
Zoltán Bátori, and Péter Török edited the manuscript. Available from: https://​doi.​org/​10.​1093/​aob/​mcz180
Díaz, S., Kattge, J., Cornelissen, J.H., Wright, I.J., Lavorel, S., Dray,
S. et al. (2016) The global spectrum of plant form and function.
AC K N O​W L E D
​ G E ​M E N T S Nature, 529, 167–171. Available from: https://​d oi.​org/​10.​1038/​
The authors thank Dolly Tolnay for her help with the trait data- natur​e16489
bases. We would also like to acknowledge the financial support from Erdős, L., Ambarlı, D., Anenkhonov, O.A., Bátori, Z., Cserhalmi, D., Kiss,
M. et al. (2018a) The edge of two worlds: a new review and syn-
the János Bolyai Research Scholarship of the Hungarian Academy
thesis on Eurasian forest-­steppes. Applied Vegetation Science, 21,
of Sciences to Zoltán Bátori and László Erdős, and the Stipendium 345–362. Available from: https://​doi.​org/​10.​1111/​avsc.​12382​
Hungaricum Scholarship for Khanh Vu Ho. Erdős, L., Ho, K.V., Bátori, Z., Kröel-­Dulay, G., Ónodi, G., Tölgyesi, C. et al.
(2023) Taxonomic, functional, and phylogenetic diversity peaks do
not coincide along a compositional gradient in forest-­grassland mo-
F U N D I N G I N FO R M AT I O N
saics. Journal of Ecology, 111, 182–197. Available from: https://​doi.​
The National Research, Development and Innovation Office, org/​10.​1111/​1365-­​2745.​14025​
Hungary (grant number FK 134384 to László Erdős; K 124796 and Erdős, L., Kröel-­D ulay, G., Bátori, Z., Kovács, B., Németh, C., Kiss, P.J.
FK 142428 to Zoltán Bátori; and K 137573 and KKP 144068 to Péter et al. (2018b) Habitat heterogeneity as a key to high conservation
 |

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ERDŐS et al. 9 of 10
Journal of Vegetation Science
value in forest-­g rassland mosaics. Biological Conservation, 226, Vegetation Science, 24, 1195–1203. Available from: https://​doi.​org/​
72–80. Available from: https://​d oi.​o rg/​10.​1016/j.​b iocon.​2018.​ 10.​1111/​jvs.​12027​
07.​029 Király, G. (2009) Új magyar füvészkönyv. Jósvafő: Aggtelek National Park.
Erdős, L., Tölgyesi, C., Horzse, M., Tolnay, D., Hurton, A., Schulcz, N. et al. Kleyer, M., Bekker, R.M., Knevel, I.C., Bakker, J.P., Thompson, K.,
(2014) Habitat complexity of the Pannonian forest-­steppe zone Sonnenschein, M. et al. (2008) The LEDA Traitbase: a database
and its nature conservation implications. Ecological Complexity, 17, of life-­history traits of the northwest European flora. Journal of
107–118. Available from: https://​doi.​org/​10.​1016/j.​ecocom.​2013.​ Ecology, 96, 1266–1274. Available from: https://​doi.​org/​10.​1111/j.​
11.​0 04 1365-­​2745.​2008.​01430.​x
Erdős, L., Török, P., Szitár, K., Bátori, Z., Tölgyesi, C., Kiss, P.J. et al. (2020) Lenth, R.V. (2022) Emmeans: estimated marginal means, aka least-­squares
Beyond the forest-­grassland dichotomy: the gradient-­like organiza- means. Version 1.7.5. Available from: https://​CRAN.​R-­​proje​c t.​org/​
tion of habitats in forest-­steppes. Frontiers in Plant Science, 11, 236. packa​ge=​emmeans.
Available from: https://​doi.​org/​10.​3389/​fpls.​2020.​0 0236​ Lhotsky, B., Csecserits, A., Kovács, B. & Botta-­Dukát, Z. (2016a) New
E-­Vojtkó, A., Balogh, N., Deák, B., Kelemen, A., Kis, S., Kiss, R. et al. plant trait records of the Hungarian flora. Acta Botanica Hungarica,
(2020) Leaf trait records of vascular plant species in the Pannonian 58, 397–400. Available from: https://​doi.​org/​10.​1556/​abot.​58.​
flora with special focus on endemics and rarities. Folia Geobotanica, 2016.​3-­​4.​8
55, 73–79. Available from: https://​doi.​org/​10.​1007/​s1222​4-­​020-­​ Lhotsky, B., Kovács, B., Ónodi, G., Csecserits, A., Rédei, T., Lengyel, A.
09363​-­​7 et al. (2016b) Changes in assembly rules along a stress gradient
Fekete, G. (1992) The holistic view of succession reconsidered. Coenoses, from open dry grasslands to wetlands. Journal of Ecology, 104, 507–
7, 21–29. 517. Available from: https://​doi.​org/​10.​1111/​1365-­​2745.​12532​
Götzenberger, L., de Bello, F., Bråthen, K.A., Davison, J., Dubuis, A., Li, Y. & Shipley, B. (2017) An experimental test of CSR theory using a
Guisan, A. et al. (2012) Ecological assembly rules in plant commu- globally calibrated ordination method. PLoS One, 12, e0175404.
nities—approaches, patterns and prospects. Biological Reviews, 87, Available from: https://​doi.​org/​10.​1371/​journ​al.​pone.​0175404
111–127. Available from: https://​doi.​org/​10.​1111/j.​1469-­​185X.​ Lüdecke, D., Ben-­Shachar, M.S., Patil, I., Waggoner, P. & Makowski, D.
2011.​0 0187.​x (2021) Performance: an R package for assessment, comparison
Grime, J.P. (1974) Vegetation classification by reference to strategies. and testing of statistical models. Journal of Open Source Software,
Nature, 250, 26–31. Available from: https://​doi.​org/​10.​1038/​ 6, 3139.
250026a0 McIntosh-­Buday, A., Sonkoly, J., Takács, A., Balogh, N., Kovacsics-­Vári,
Grime, J.P. (1977) Evidence for the existence of three primary strategies G., Teleki, B. et al. (2022) New data of plant leaf traits from Central
in plants and its relevance to ecological and evolutionary theory. Europe. Data in Brief, 42, 108286. Available from: https://​doi.​org/​
The American Naturalist, 111, 1169–1194. 10.​1016/j.​dib.​2022.​108286
Grime, J.P. & Pierce, S. (2012) The evolutionary strategies that shape eco- Menezes, B.S., Martins, F.R., Dantas Carvalho, E.C., Souza, B.C., Silveira,
systems. Chichester: John Wiley & Sons. A.P., Loiola, M.I.B. et al. (2020) Assembly rules in a resource gra-
Gyalus, A., Barabás, S., Berki, B., Botta-­Dukát, Z., Kabai, M., Lhotsky, B. dient: competition and abiotic filtering determine the structuring
et al. (2022) Plant trait records of the Hungarian and Serbian flora of plant communities in stressful environments. PLoS One, 15,
and methodological description of some hardly measurable plant e0230097. Available from: https://​doi.​org/​10.​1371/​journ​al.​pone.​
species. Acta Botanica Hungarica, 64, 451–454. Available from: 0230097
https://​doi.​org/​10.​1556/​034.​6 4.​2022.​3-­​4.​14 Molnár, Z., Biró, M., Bartha, S. & Fekete, G. (2012) Past trends, present
Halassy, M., Singh, A.N., Szabó, R., Szili-­Kovács, T., Szitár, K. & Török, K. state and future prospects of Hungarian forest-­steppes. In: Werger,
(2016) The application of a filter-­based assembly model to develop M.J.A. & van Staalduinen, M.A. (Eds.) Eurasian Steppes. Berlin:
best practices for Pannonian sand grassland restoration. Journal of Springer, pp. 209–252.
Applied Ecology, 53, 765–773. Available from: https://​doi.​org/​10.​ Münkemüller, T., Gallien, L., Pollock, L.J., Barros, C., Carboni, M.,
1111/​1365-­​2664.​12618​ Chalmandrier, L. et al. (2020) Dos and don'ts when inferring assem-
Hamilton, N.E. & Ferry, M. (2018) Ggtern: ternary diagrams using gg- bly rules from diversity patterns. Global Ecology and Biogeography,
plot2. Journal of Statistical Software, 87, 1–17. 29, 1212–1229. Available from: https://​doi.​org/​10.​1111/​geb.​
Han, X., Huang, J. & Zang, R. (2021) Shifts in ecological strategy spec- 13098​
tra of typical forest vegetation types across four climatic zones. Negreiros, D., Stradic, S.L., Fernandes, G.W. & Renno, H.C. (2014) CSR
Scientific Reports, 11, 14127. Available from: https://​doi.​org/​10.​ analysis of plant functional types in highly diverse tropical grass-
1038/​s 4159​8-­​021-­​93722​-­​7 lands of harsh environments. Plant Ecology, 215, 379–388. Available
Ho, K.V., Süle, G., Kovács, B. & Erdős, L. (2024) Strong differences in from: https://​doi.​org/​10.​1007/​s1125​8-­​014-­​0302-­​6
microclimate among the habitats of a forest-­steppe ecosystem. Oksanen, J., Simpson, G.L., Blanchet, F.G., Kindt, R., Legendre, P., Minchin,
Időjárás, 128. Available from: https://​doi.​org/​10.​28974/​​idoja​ras.​ P.R. et al. (2022) Vegan: community ecology package. Version 2.6-2.
2024.​1 Available from: https://​CRAN.​R-­​proje​c t.​org/​packa​ge=​vegan​.
Hulvey, K.B. & Aigner, P.A. (2014) Using filter-­based community assem- Petraitis, P. (2013) Multiple stable states in natural ecosystems. Oxford:
bly models to improve restoration outcomes. Journal of Applied Oxford University Press.
Ecology, 51, 997–1005. Available from: https://​doi.​org/​10.​1111/​ Pierce, S., Luzzaro, A., Caccianiga, M., Ceriani, R.M. & Cerabolini, B.
1365-­​2664.​12275​ (2007) Disturbance is the principal α-­scale filter determining niche
Jaagus, J., Aasa, A., Aniskevich, S., Boincean, B., Bojariu, R., Briede, A. differentiation, coexistence and biodiversity in an alpine commu-
et al. (2022) Long-­term changes in drought indices in eastern and nity. Journal of Ecology, 95, 698–706. Available from: https://​doi.​
Central Europe. International Journal of Climatology, 42, 225–249. org/​10.​1111/j.​1365-­​2745.​2007.​01242.​x
Available from: https://​doi.​org/​10.​1002/​joc.​7241 Pierce, S., Negreiros, D., Cerabolini, B.E.L., Kattge, J., Díaz, S., Kleyer, M.
Keddy, P.A. (1992) Assembly and response rules: two goals for predic- et al. (2017) A global method for calculating plant CSR ecological
tive community ecology. Journal of Vegetation Science, 3, 157–164. strategies applied across biomes world-­wide. Functional Ecology,
Available from: https://​doi.​org/​10. ​2307/​3235676 31, 444–457. Available from: https://​doi.​org/​10.​1111/​1365-­​2435.​
Kelemen, A., Török, P., Valkó, O., Miglécz, T. & Tóthmérész, B. (2013) 12722​
Mechanisms shaping plant biomass and species richness: plant R Core Team. (2021) R: a language and environment for statistical comput-
strategies and litter effect in alkali and loess grasslands. Journal of ing. Available from: https://​cran.​r-­​proje​c t.​org/​bin/​windo​ws/​base/​
 | Journal of Vegetation Science

16541103, 2024, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jvs.13228 by CENTRE FOR ECOLOGICAL RESEARCH HAS LIBRARY, Wiley Online Library on [20/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 of 10 ERDŐS et al.

Rosenfield, M.F., Müller, S.C. & Overbeck, G.E. (2019) Short gradient, Appendix S1. The 13 study sites used in this study with coordinates
but distinct plant strategies: the CSR scheme applied to subtropical
and elevation, and the distribution of relevés across plant
forests. Journal of Vegetation Science, 30, 984–993. Available from:
https://​doi.​org/​10.​1111/​jvs.​12787​ communities and sites.
Sábitz, J., Pongrácz, R. & Bartholy, J. (2014) Estimated changes of Appendix S2. Ternary plots of weighted mean values of CSR
drought tendency in the Carpathian Basin. Hungarian Geographical strategies with square-­root-­transformed cover scores for the eight
Bulletin, 63, 365–378. Available from: https://​doi.​org/​10.​15201/​​
vegetation types.
hunge​obull.​63.4.​1
Sonkoly, J., Tóth, E., Balogh, N., Balogh, L., Bartha, D., Bata, K. et al. (2023) Appendix S3. Statistical results of the pairwise comparisons of
PADAPT 1.0 – the Pannonian dataset of plant traits. Scientific Data, unweighted components C, S, and R.
10, 742. Available from: https://​doi.​org/​10.​1038/​s 4159​7-­​023-­​ Appendix S4. Relationships between plot scores on the primary DCA
02619​-­​9
ordination axis and each weighted strategy component using square
Temperton, V.M., Hobbs, R.J., Nuttle, T. & Halle, S. (2004) Assembly rules
and restoration ecology: bridging the gap between theory and practice. root transformation of cover scores: (a) C; (b) S, and (c) R; p-­value and
Washington, DC: Island Press. adjusted R-­squared were calculated using linear regression; Slope:
Tölgyesi, C., Zalatnai, M., Erdős, L., Batori, Z., Hupp, N.R. & Körmöczi, L. the slope value of the regression line; blue line is the regression line,
(2016) Unexpected ecotone dynamics of a sand dune vegetation
and the gray area around the line represents the 95% confidence
complex following water table decline. Journal of Plant Ecology, 9,
interval. Box plots demonstrate the variability of each strategy
40–50. Available from: https://​doi.​org/​10.​1093/​jpe/​r tv032
Várallyay, G. (1993) Soils in the region between the Rivers Danube and component: (d) C, (e) S, and (f) R in the eight communities.
Tisza (Hungary). In: Szujkó-­L acza, J. & Kováts, D. (Eds.) The flora of
the KiskunságNational Park. Hungarian Natural History Museum:
Budapest, pp. 21–42.
Yu, J., Hou, G., Zhou, T., Shi, P., Zong, N. & Sun, J. (2022) Variation of How to cite this article: Erdős, L., Ho, K.V., Bede-­Fazekas, Á.,
plant CSR strategies across a precipitation gradient in the alpine
Kröel-­Dulay, G., Tölgyesi, C., Bátori, Z. et al. (2024)
grasslands on the northern Tibet plateau. Science of the Total
Environment, 838, 156512. Available from: https://​doi.​org/​10.​ Environmental filtering is the primary driver of community
1016/j.​scito​tenv.​2022.​156512 assembly in forest–grassland mosaics: A case study based on
CSR strategies. Journal of Vegetation Science, 35, e13228.
Available from: https://doi.org/10.1111/jvs.13228
S U P P O R T I N G I N FO R M AT I O N
Additional supporting information can be found online in the
Supporting Information section at the end of this article.