Content uploaded by Truman P Young
Author content
All content in this area was uploaded by Truman P Young on Oct 04, 2020
Content may be subject to copyright.
Synchrony matters more than species richness in plant
community stability at a global scale
Enrique Valencia
a,b,1
, Francesco de Bello
b,c,d
, Thomas Galland
b,c
, Peter B. Adler
e
, Jan Lepš
b,f
, Anna E-Vojtkó
b,c
,
Roel van Klink
g
, Carlos P. Carmona
h
,Ji
rí Danihelka
i,j
, Jürgen Dengler
g,k,l
, David J. Eldridge
m
,
Marc Estiarte
n,o
, Ricardo García-González
p
, Eric Garnier
q
, Daniel Gómez‐García
p
, Susan P. Harrison
r
,
TomášHerben
j,s
, Ricardo Ibáñez
t
, Anke Jentsch
u
, Norbert Juergens
v
, Miklós Kertész
w
, Katja Klumpp
x
,
Frédérique Louault
x
, Rob H. Marrs
y
, Romà Ogaya
n,o
, Gábor Ónodi
w
, Robin J. Pakeman
z
, Iker Pardo
aa
,
Meelis Pärtel
h
, Begoña Peco
bb
, Josep Peñuelas
n,o
, Richard F. Pywell
cc
, Marta Rueda
dd,ee
, Wolfgang Schmidt
ff
,
Ute Schmiedel
v
, Martin Schuetz
gg
, Hana Skálová
j
, Petr ˇ
Smilauer
hh
, Marie ˇ
Smilauerová
b
, Christian Smit
ii
,
MingHua Song
jj
, Martin Stock
kk
, James Val
m
, Vigdis Vandvik
ll
, David Ward
mm
, Karsten Wesche
g,nn,oo
,
Susan K. Wiser
pp
, Ben A. Woodcock
cc
, Truman P. Young
qq,rr
, Fei-Hai Yu
ss
, Martin Zobel
h
,
and Lars Götzenberger
b,c
a
Departamento de Biología y Geología, Física y Química Inorgánica, Escuela Superior de Ciencias Experimentales y Tecnología, Universidad Rey Juan Carlos,
28933, Móstoles, Spain;
b
Department of Botany, Faculty of Science, University of South Bohemia, 37005,
Ceské Bud
ejovice, Czech Republic;
c
Institute of
Botany of the Czech Academy of Sciences, 37982, T
rebo
n, Czech Republic;
d
Centro de Investigaciones sobre Desertificación, 46113, Valencia, Spain;
e
Department of Wildland Resources and the Ecology Center, Utah State University, Logan, UT 84322;
f
Biology Research Centre, Institute of Entomology,
Czech Academy of Sciences, 37005,
Ceské Bud
ejovice, Czech Republic;
g
German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, 04103,
Leipzig, Germany;
h
Department of Botany, Institute of Ecology and Earth Sciences, University of Tartu, 51005, Tartu, Estonia;
i
Department of Botany and
Zoology, Faculty of Science, Masaryk University, 61137, Brno, Czech Republic;
j
Institute of Botany of the Czech Academy of Sciences, 25243, Pr
uhonice, Czech
Republic;
k
Vegetation Ecology Group, Institute of Natural Resource Sciences (IUNR), Zurich University of Applied Sciences, 8820, Wädenswil, Switzerland;
l
Plant Ecology Group, Bayreuth Center for Ecology and Environmental Research (BayCEER), University of Bayreuth, 95447, Bayreuth, Germany;
m
Biological,
Earth and Environmental Sciences, University of New South Wales, 2052, Sydney, Australia;
n
Centre for Ecological Research and Forestry Applications
(CREAF), 08193, Cerdanyola del Vallès, Catalonia, Spain;
o
Spanish National Research Center (CSIC), Global Ecology Unit CREAF-CSIC-Autonomous University
of Barcelona, 08193, Bellaterra, Catalonia, Spain;
p
Instituto Pirenaico de Ecología (IPE-CSIC), 22700, Jaca-Zaragoza, Spain;
q
Center in Ecology and
Evolutionary Ecology (CEFE), Université Montpellier, French National Centre for Scientific Research (CNRS), École pratique des Hautes Études (EPHE),
Research Institute for Development (IRD), Université Paul Valéry Montpellier 3, 34293, Montpellier, France;
r
Department of Environmental Science and
Policy, University of California, Davis, CA 95616;
s
Department of Botany, Faculty of Science, Charles University, Praha, Czech Republic;
t
Department of
Environmental Biology, University of Navarra, Pamplona, Spain;
u
Department of Disturbance Ecology, Bayreuth Center of Ecology and Environmental
Research, University of Bayreuth, Bayreuth, Germany;
v
Research Unit Biodiversity, Evolution & Ecology of Plants, Institute of Plant Science and Microbiology,
University of Hamburg, Hamburg, Germany;
w
Institute of Ecology and Botany, Centre for Ecological Research, Hungarian Academy of Sciences, Vácrátót,
Hungary;
x
Université Clermont Auvergne, INRAE, VetAgro Sup, UMR Ecosystème Prairial, Clermont-Ferrand, France;
y
University of Liverpool, Liverpool,
United Kingdom;
z
The James Hutton Institute, Craigiebuckler, Aberdeen, United Kingdom;
aa
Department of Plant Biology and Ecology, University of the
Basque Country, 48940, Leioa, Spain;
bb
Terrestrial Ecology Group (TEG), Department of Ecology, Institute for Biodiversity and Global Change, Autonomous
University of Madrid, 28049, Madrid, Spain;
cc
UK Centre for Ecology & Hydrology, Crowmarsh Gifford, OX10 8BB, Wallingford, Oxfordshire, United
Kingdom;
dd
Department of Conservation Biology, Estación Biológica de Doñana, 41092, Sevilla, Spain;
ee
Department of Plant Biology and Ecology,
Universidad de Sevilla, 41012, Sevilla, Spain;
ff
Department of Silviculture and Forest Ecology of the Temperate Zones, University of Göttingen, 37077,
Göttingen, Germany;
gg
Community Ecology, Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 8903, Birmensdorf, Switzerland;
hh
Department of Ecosystem Biology, Faculty of Science, University of South Bohemia, 37005,
Ceské Bud
ejovice, Czech Republic;
ii
Conservation Ecology
Group, Groningen Institute for Evolutionary Life Sciences, 11103, Groningen, The Netherlands;
jj
Laboratory of Ecosystem Network Observation and
Modelling, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, 100101, Beijing, China;
kk
Wadden Sea National
Park of Schleswig-Holstein, 25832, Tönning, Germany;
ll
Department of Biological Sciences and Bjerknes Centre for Climate Research, University of Bergen,
5020, Bergen, Norway;
mm
Department of Biological Sciences, Kent State University, Kent, OH 44242;
nn
Botany Department, Senckenberg, Natural History
Museum Goerlitz, 02826, Goerlitz, Germany;
oo
International Institute Zittau, Technische Universität Dresden, 02763, Zittau, Germany;
pp
Manaaki
Whenua–Landcare Research, 7640, Lincoln, New Zealand;
qq
Department of Plant Sciences, University of California, Davis, CA 95616;
rr
Mpala Research
Centre, Nanyuki, Kenya; and
ss
Institute of Wetland Ecology & Clone Ecology/Zhejiang Provincial Key Laboratory of Plant Evolutionary Ecology and
Conservation, Taizhou University, 318000, Taizhou, China
Edited by Nils Chr. Stenseth, University of Oslo, Oslo, Norway, and approved August 5, 2020 (received for review November 20, 2019)
The stability of ecological communities is critical for the stable
provisioning of ecosystem services, such as food and forage
production, carbon sequestration, and soil fertility. Greater biodi-
versity is expected to enhance stability across years by decreasing
synchrony among species, but the drivers of stability in nature
remain poorly resolved. Our analysis of time series from 79
datasets across the world showed that stability was associated
more strongly with the degree of synchrony among dominant
species than with species richness. The relatively weak influence of
species richness is consistent with theory predicting that the effect
of richness on stability weakens when synchrony is higher than
expected under random fluctuations, which was the case in most
communities. Land management, nutrient addition, and climate
change treatments had relatively weak and varying effects on
stability, modifying how species richness, synchrony, and stability
interact. Our results demonstrate the prevalence of biotic drivers
on ecosystem stability, with the potential for environmental drivers
to alter the intricate relationship among richness, synchrony, and
stability.
evenness
|
climate change drivers
|
species richness
|
stability
|
synchrony
Understanding the mechanisms that maintain ecosystem sta-
bility (1) is essential for the stable provisioning of multiple
ecosystem functions and services (2, 3). Although research on
Author contributions: F.d.B., J.L., and L.G. designed research; E.V., F.d.B., T.G., and L.G.
performed research; E.V., C.P.C., and L.G. analyzed data; E.V. and T.G. assembled data;
P.B.A. contributed with datasets; J.L., R.v.K., J. Danihelka, J. Dengler, D.J.E., M.E., R.G.-G.,
E.G., D.G.-G., S.P.H., T.H., R.I., A.J., N.J., M.K., K.K., F.L., R.H.M., R.O., G.Ó., R.J.P., I.P., M.P.,
B.P., J.P., R.F.P., M.R., W.S., U.S., M. Schuetz, H.S., P. ˇ
S., M. ˇ
Smilauerová, C.S., M. Song, M.
Stock, J.V., V.V., K.W., S.K.W., B.A.W., T.P.Y., F.-H.Y., and M.Z. contributed with a dataset;
and E.V., F.d.B., T.G., P.B.A., J.L., A.E.-V., R.v.K., C.P.C., J. Danihelka, J. Dengler, D.J.E., M.E.,
R.G.-G., E.G., D.G.-G., S.P.H., T.H., R.I., A.J., N.J., M.K., K.K., F.L., R.H.M., R.O., G.Ó., R.J.P.,
I.P., M.P., B.P., J.P., R.F.P., M.R., W.S., U.S., M. Schuetz, H.S., P.ˇ
S., M. ˇ
Smilauerová, C.S., M.
Song, M. Stock, J.V., V.V., D.W., K.W., S.K.W., B.A.W., T.P.Y., F.-H.Y., M.Z., and L.G. wrote
the paper.
The authors declare no competing interest.
This article is a PNAS Direct Submission.
Published under the PNAS license.
1
To whom correspondence may be addressed. Email: valencia.gomez.e@gmail.com.
This article contains supporting information online at https://www.pnas.org/lookup/suppl/
doi:10.1073/pnas.1920405117/-/DCSupplemental.
First published September 8, 2020.
www.pnas.org/cgi/doi/10.1073/pnas.1920405117 PNAS
|
September 29, 2020
|
vol. 117
|
no. 39
|
24345–24351
ECOLOGYENVIRONMENTAL
SCIENCES
Downloaded at ucd university library on October 2, 2020
community stability has decades of history in ecology (4), with
stability often measured as the inverse coefficient of variation
across years of community abundance or biomass, the main drivers
of stability remain elusive (5). Both abiotic and biotic drivers [e.g.,
climate, land use, and species diversity (6–8)] are expected to
govern community stability. Among biotic drivers, the hypothesis
that increases in species diversity beget stability in communities
and ecosystems (Fig. 1) (2, 9–11) has generated ongoing debate
(12, 13).
The stabilizing effect of biodiversity has been attributed to
various mechanisms (12). Most biodiversity–stability mechanisms
at single trophic levels involve some form of compensatory dy-
namics, which occur when year-to-year temporal fluctuations in
the abundance of some species are offset by fluctuations of other
species (4, 17). Compensatory dynamics are associated with de-
creased synchrony among species, with synchrony defined as the
extent to which species population sizes covary positively over
time. Decreased synchrony, which is predicted to stabilize com-
munities (Fig. 1A), can result from species-specific responses to
environmental fluctuations (18–20) and from temporal changes
in competitive hierarchies (21), as well as stochastic fluctuations.
Importantly, it is expected that species richness can increase
stability (Fig. 1C) by decreasing synchrony (Fig. 1E). This posi-
tive effect of richness on stability can be, in fact, a result of an
increased chance that the community will contain species with
differing responses to abiotic drivers or competition, leading to a
reduction in synchrony (12). However, the effect of richness on
stability should weaken when synchrony is higher than expected
if species were fluctuating randomly and independently (SI Ap-
pendix, Supplementary Text S1 has expanded information) (14).
At the same time, other biotic drivers, together with richness and
synchrony, have the potential to interact and buffer the effects of
ongoing climatic and land-use changes. These additional biotic
drivers include community evenness, which can both increase or
decrease synchrony (1), or the presence of more stable species
that are characterized by more conservative resource strategies
(22). Long-term empirical data from natural communities can
help us reveal the real-world effects of biotic drivers on com-
munity stability (6).
Here, we explore the generality of biodiversity–synchrony–
stability relationships, and their implications in a global change
context, across multiple ecosystems and a wide range of envi-
ronments. We compiled data from 7,788 natural and seminatural
vegetation plots that had annual measurements spanning at least
6 y, sourced from 79 datasets distributed across the world (SI
Appendix, Fig. S1). Most of the datasets include information
about human activities related to global change through the
application of experimental treatments, including fertilization,
herbivore exclusion, grazing, fire, and climate manipulations
(hereafter environmental treatments). Biodiversity, synchrony,
and stability are known to vary in response to climate and land
use, although knowledge of such responses is limited by lack of
comparative data across major habitats and geographic extent
(8, 13, 16). The compiled data allowed us to compare the re-
lationships between species richness, synchrony [using the logVindex
(16)], and stability against theoretical predictions (summarized in
Fig. 1) across vegetation types, climates, and land uses.
Results and Discussion
Interplay between Species Richness, Synchrony, and Stability. Our
results confirmed the general prevalence of negative synchrony–
stability relationships: 71% of the datasets exhibited nega-
tive and significant relationships (R
2
m=0.19; i.e., variance
explained by the fixed effects over all individual plots) (Fig. 1B).
We found similar results for other synchrony indices (SI Ap-
pendix,Fig.S2). These findings support theoretical predic-
tions (Fig. 1A) and previous empirical evidence (2, 6, 11)
that lower levels of synchrony in species fluctuations stabilize
overall community abundance, despite the large range of vegeta-
tion types, environmental treatments, and biogeographic regions
we considered.
Our results highlight a second global pattern consistent with
theory (Fig. 1C): higher species richness was associated with
greater community stability (R
2
m=0.06) (Fig. 1D). However,
this relationship was not nearly as strong: only 29% of the
datasets showed a positive and significant relationship. The high
proportion of nonsignificant species richness–stability relation-
ships was unexpected, as species richness is generally considered
one of the strongest drivers of stability (8–10, 23). Nevertheless,
in observational datasets species richness may covary with other
Fig. 1. Relationships between synchrony and stability (Aand B), richness
and stability (Cand D), and richness and synchrony (Eand F). Richness and
stability were ln transformed. A,C,andEare the schematic representations
of these relationships following theoretical predictions (1, 12, 14, 15). B,D,
and Fdepict these relationships for each dataset (n=79). Red, blue, and gray
lines represent the statistically significant positive, negative, and nonsignif-
icant slopes, respectively. Black lines show each relationship based on all
plots (n=7,788) using a linear mixed effects model with datasets as a ran-
dom factor; these were all statistically significant. The synchrony index was
logV(16).
Significance
The stability of ecological communities under ongoing climate
and land-use change is fundamental to the sustainable man-
agement of natural resources through its effect on critical
ecosystem services. Biodiversity is hypothesized to enhance
stability through compensatory effects (decreased synchrony
between species). However, the relative importance and in-
terplay between different biotic and abiotic drivers of stability
remain controversial. By analyzing long-term data from natural
and seminatural ecosystems across the globe, we found that
the degree of synchrony among dominant species was the
main driver of stability, rather than species richness per se.
These biotic effects overrode environmental drivers, which
influenced the stability of communities by modulating the ef-
fects of richness and synchrony.
24346
|
www.pnas.org/cgi/doi/10.1073/pnas.1920405117 Valencia et al.
Downloaded at ucd university library on October 2, 2020
factors that influence interannual community variability, poten-
tially masking any direct effect of species richness (24).
Species richness was positively and significantly associated
with synchrony across all studies, and the expected negative re-
lationship predicted by theory was found in only 8% of our
datasets (Fig. 1F). Such low frequencies of negative richness–
synchrony relationships contradict both theoretical predictions
(Fig. 1E) and previous studies. For instance, a recent richness-
manipulated experimental study showed a negative relationship
between richness and synchrony (25), although this could be
driven by the low levels of species richness applied in that ex-
periment. We note that in natural or seminatural communities,
such as those analyzed here, richness often exceeds the low levels
commonly applied in experimental studies that manipulate
richness. Our results showed that while the relationship between
synchrony and species richness across datasets depended on the
index of synchrony considered (Fig. 1Fand SI Appendix, Fig. S2;
SI Appendix, Supplementary Texts S1 and S2 have expanded
information), in most cases it was relatively weak. Our results
thus provide only partial support for the hypothesis that more
diverse communities are more stable due to the negative effect of
richness on synchrony (6, 13, 16). Indeed, we expected to observe
a negative relationship between species richness and synchrony,
particularly for those plots and datasets where the relationship
between species richness and stability was strong.
To better understand our results, we explored a random fluc-
tuation scenario, which we approximated using null models that
disrupt synchrony patterns between co-occurring species (Methods
and SI Appendix, Supplementary Text S2). Specifically, we com-
pared the relationships observed among richness, synchrony, and
stability against values expected under random species fluctua-
tions. We also considered potential mathematical constraints on
these relationships (SI Appendix, Supplementary Texts S1 and S2).
This modeling exercise revealed that the observed relationship
between species richness and stability was weaker than expected
under random species fluctuations (observed relationship R
2
m=
0.059; expected relationship R
2
m=0.157). However, the rela-
tionship between synchrony and stability was greater than expec-
ted under the null model (observed relationship R
2
m=0.191;
expected relationship R
2
m=0.021) (SI Appendix, Supplementary
Text S2), particularly for the index of synchrony we focused on in
the text. Note also that for this index, the observed relationship
between richness and synchrony was lower than expected by
chance (observed relationship R
2
m=0.024; expected relationship
R
2
m=0.082) (Methods) and very weak. Most importantly, syn-
chrony between species was higher than expected under the ran-
dom fluctuations scenario, regardless of the index used (based on
paired ttest, P<0.001; t=6.38; mean observed syn-
chrony =−0.02 and mean expected synchrony =−0.08). These
findings show that, in natural ecosystems, synchrony in species
abundances (positive covariances) is more common than random
fluctuations or negative covariances (26), likely because many
species-rich communities contain ecologically similar species, with
similar responses to weather (14, 27). When synchrony is greater
than expected under random fluctuations, the effect of richness on
synchrony and stability will be reduced (SI Appendix, Supple-
mentary Text S1) (1, 14). Our results provide empirical evidence
that, for a wide range of ecosystems, species richness does pro-
mote stability, but this effect is not necessarily caused by a direct,
negative effect of richness on synchrony.
Predictors of Ecosystem Stability. We examined whether synchrony
and stability are mediated by different drivers, an issue that is
gaining momentum in a global change context (6, 7, 16). We
evaluated the effect of climate, vegetation type, environmental
treatments, and biotic attributes (percentage of woody species,
species evenness and richness) on synchrony and community
stability (SI Appendix, Table S1). Overall, the combined effect of
environmental treatments reduced both temporal synchrony and
stability (Fig. 2 Aand B). While the effect size of the combined
treatments was small compared with biotic factors (SI Appendix,
Table S1), this mostly reflects opposing effects of different treat-
ment types (SI Appendix, Supplementary Text S3 has expanded
information).
Using only those datasets with similar treatments and associ-
ated control plots (fertilization, herbivore exclusion, grazing in-
tensification, removal plant species, fire, and manipulative climate
change drivers), we ran separate analyses to disentangle the effect
of the environmental treatments on synchrony and stability. Fer-
tilization and herbivore exclusion significantly decreased syn-
chrony, whereas intensification of grazing significantly increased
synchrony (Fig. 2C). These relationships were partially unexpected
because previous studies have shown that fertilization could pro-
mote synchrony (10) while grazing intensification could decrease it
(13). However, in agreement with our results, Lepšet al. (16)
demonstrated in a local study that while nutrient enrichment in-
creases competition among plant species, it also decreases stability
by increasing differences in productivity between favorable and
unfavorable years. This could override the potential compensatory
dynamics due to synchrony. Moreover, herbivore exclusion or a
reduction in grazing intensity acted to increase community stability
(Fig. 2D). These results suggest that herbivory affects interspecific
competition, promoting the species best adapted to grazing but
reducing the year-to-year stability of the community (16). Overall,
these results show that changes in environmental drivers, associ-
ated with global change scenarios, can disrupt the interplay be-
tween diversity, synchrony, and stability, even reversing the
expected effects of biotic drivers on stability. Thus, the joint
consideration of a wide variety of factors provides insights into the
relationships underlying synchrony and stability, enhancing the
future prediction of community stability in the face of global
changes.
It should be noted that nutrient addition and/or grazing
pressure could promote directional changes in species compo-
sition, with some species increasing over the years and others
decreasing (28). This could cause a decrease in synchrony values
for indices studied here (29), with the indices reflecting not only
year-to-year fluctuations due to compensatory dynamics but also,
these long-term trends. More research is certainly needed in the
future to account for the effect of directional trends on the in-
terplay of biotic and abiotic effects on stability.
We found that forest understory vegetation was more syn-
chronous and less stable than grasslands, shrublands, and savannas
(Fig. 2B), similarly to Blüthgen et al. (13). We suggest that forest
understory vegetation has weaker compensatory effects that lead
to destabilization. Also, this result could be related to the fact
that we excluded from the analyses the tree layer (i.e., the most
stable vegetation layers in these systems). Alternatively, this
vegetation might support a greater proportion of rare species,
which benefit from shared favorable conditions (30) increasing
the synchrony of the community. Finally, communities with a
greater proportion of woody species were more stable. The
longer life span of woody species and their structural storage of
carbon and nutrients should buffer them against environmental
fluctuations and the fluctuations of other species, although we
note that longer measurement timescales may be required to
accurately capture their dynamics.
Finally, we found evidence of a positive evenness–synchrony
association (Fig. 2A) and a negative evenness–stability associa-
tion (Fig. 2B). In other words, low synchrony is more common in
communities with low evenness that are dominated by a few
species. These communities appear to fluctuate less and are
therefore more stable (31, 32). This finding suggests two potential
ecological mechanisms. First, these few species could be the best-
adapted species and tend to perform well across years (i.e., have
comparatively little fluctuations), thus promoting stability. In some
Valencia et al. PNAS
|
September 29, 2020
|
vol. 117
|
no. 39
|
24347
ECOLOGYENVIRONMENTAL
SCIENCES
Downloaded at ucd university library on October 2, 2020
cases, for example, species with slower growth strategies are lo-
cally more abundant and stable in time (22). Second, a small
number of dominant species with different adaptations (different
traits) (16, 33, 34) could lead to decreased synchrony and in-
creased stability at the community level. If synchrony is a common
feature of vegetation [as suggested by our study and in Houlahan
et al. (26)], evenness can have an effect on stability via synchrony
(Fig. 3). Low synchrony among a small number of dominant
species could thus represent an important stabilizing effect in
ecosystems worldwide.
Direct and Indirect Effects of Abiotic and Biotic Attributes on
Community Stability. To clarify the ensemble of directional ef-
fects of abiotic and biotic factors on community stability, we
generated a piecewise structural equation model (Fig. 3). Our
model explained 88% of the variance in community stability and
confirmed that the most important determinant of stability was
the direct negative effect of synchrony. Analogous results were
found when we evaluated either individual habitats or the con-
trol plots among habitats (SI Appendix, Figs. S3 and S4) or when
other synchrony indices were used (SI Appendix, Fig. S5 Aand
B). Further, mean annual temperature showed a direct, negative
effect on stability, as in other studies (6), which was further
reinforced via its indirect effects on evenness, species richness,
and synchrony (Fig. 3). Communities in more variable climates,
such as Mediterranean environments, should show large varia-
tion in productivity from year to year, increasing synchrony be-
tween species and decreasing stability of the whole community.
Again, the positive associations between species richness–
synchrony and evenness–synchrony suggest that the stabilizing
effect of communities originates from lower synchrony among
the dominant species (35) rather than by the number of species
per se (18, 31), emphasizing the role of evenness in the distri-
bution of abundance over time.
Overall, this study demonstrates the consistent cross-system
importance of the interplay among species richness, synchrony,
and environmental parameters in the prediction of community
stability. As expected, low synchrony and high species richness
defined the primary stabilizing pattern of communities (9).
However, contrary to expectation, the stabilizing effects of spe-
cies richness via synchrony were relatively weak. Yet, despite a
prevalence of synchrony between species found in our commu-
nities, richness had a net positive association with stability (direct
effect + indirect effects =0.23) (Fig. 3), implying an important
Fig. 2. Effects of multiple abiotic and biotic drivers on the synchrony values (Aand C) and stability (Band D) of the different communities. We show the
averaged parameter estimates (standardized regression coefficients) of model predictors and the associated 95% CIs. In Aand B, all of the predictors were
evaluated together using general linear mixed effect models (n=7,788). The colors represent the different drivers of vegetation type: grassland is the
reference level (orange), climatic data (blue), biotic attributes (green), number of measurements (gray), and global change treatments (black). The effects of
each environmental treatment on synchrony values and stability (Cand D) were evaluated separately and only for the studies where each driver was
measured (fertilization: n=1,058, ND [number of datasets evaluated] =17; herbivore exclusion: n=2,284, ND =19; grazing intensity: n=1,920, ND =24;
removal plant species: n=518, ND =8; fire: n=974, ND =11; manipulative climate change: n=122, ND =5).
24348
|
www.pnas.org/cgi/doi/10.1073/pnas.1920405117 Valencia et al.
Downloaded at ucd university library on October 2, 2020
effect of richness unrelated with synchrony. Environmental fac-
tors associated with different global change drivers also directly
or indirectly affect stability and have the potential to reverse the
effects of biodiversity and synchrony on stability, although biotic
factors generally had a stronger effect. Our results suggest that
interventions aiming to buffer ecosystems against the effects of
increasing environmental fluctuations should focus on promoting
the maintenance or selection of dominant species with different
adaptations or strategies that will result in low synchrony, rather
than by focusing on increasing species richness per se. Further,
the evaluation of the direct effects of evenness and environ-
mental drivers on stability adds insights on the complex under-
lying biotic and abiotic relationships. To consider these different
drivers of stability in concert is critical for defining the potential
of communities to remain stable in a global change context.
Methods
We used data from 79 plant community datasets where permanent or
semipermanent plots of natural and seminatural vegetation have been
consistently sampled over a period of 6 to 99 y (SI Appendix, Figs. S1 and S6,
Supplementary Text S4, and Table S2). We focused our analyses on vascular
plants as the main primary producers affecting subsequent trophic levels
and ecosystem functioning. These datasets have some differences, such as
the method used to quantify abundance (e.g., aboveground biomass, visual
species cover estimates, and species individual frequencies), plot size (me-
dian =1m
2
; range =0.04 to 400 m
2
), vegetation type (grassland, shrubland,
savanna, forest, and salt marsh), and number of sampling dates (median =
11.5; range =6 to 38). The studies encompassed different localities with
different species pools and different types of vegetation responding to
different types of treatments. The total number of individual plots was 7,788
across the 79 datasets (number of observations ∼190,900).
Climatic Data. We collected climatic information related to temperature and
precipitation for each of the 7,788 plots using WorldClim (https://www.
worldclim.org/) where location coordinates were available. Where these
were not available, weather data were derived from the study centroid.
Among available variables, we retained four: mean annual temperature
(degrees Celsius) and mean annual precipitation (millimeters), related to
annual trends, and mean annual temperature range and coefficient of
variation of precipitation within years as proxies for annual seasonality (6).
These variables were selected from the 19 available WorldClim climatic
variables because they describe relatively independent climatic features and
account for most of the other climatic relationships observed with our data
(climatic variable correlation is in SI Appendix, Table S3).
Biotic Attributes. In each plot, we calculated stability over time as the inverse
of the coefficient of variation (SD/mean) of the year-to-year fluctuations of
total abundance of that community. This has been widely used as a reliable
estimator of temporal invariability (36). SD was based on n−1 degrees of
freedom. We only included datasets using percentage cover as an estimate
of community structure if the summed cover was not constrained.
Although we did not measure ecosystem services directly, multiple studies
highlight the importance of a stable vegetation (primary producers) for a
stable delivery of multiple key ecosystem processes. For example, biomass and
abundance are often considered to be ecosystem functions in their own right
(e.g., forage production and carbon sink), while these can also act as a proxy
or driver of other functions, including litter quantity, soil organic matter,
evapotranspiration, or erosion control. Clearly, the value of stability depends
on its relationship to the provision of specific ecosystem services, and tem-
poral invariability does not necessarily imply a positive effect on the eco-
system service of interest. Our study aims at identifying ecological drivers of
stability at a global scale.
In each plot, we also calculated various indices that characterize the biotic
attributes of the community averaged over all annual observations: average
species richness [average number of species (2, 37)], the average percentage
of woody species per year, and evenness (using the Evar index) (38):
Evar =1−2π arctan
S
s=1ln(xs)−
S
t=1
ln(xt)S2S
,[1]
where Sis total number of species in the community and x
s
is the abundance
Fig. 3. Piecewise structural equation model showing the direct and indirect effects of multiple abiotic and biotic drivers on the stability across the 79 datasets
(Fisher’sCstatistic: C=14.96, P=0.134, n=7,788). Marginal (R
2
m) values showing variance explained by the fixed effects and conditional (R
2
c) values showing
variance explained by the entire model are provided for each response variable. Solid lines represent positive effects, while dashed lines indicate negative
effects. Blue and red lines represent statistically significant effects, and gray lines represent nonsignificant effects. The width of each arrow is proportional to
the standardized path coefficients (more information is SI Appendix, Table S5).
Valencia et al. PNAS
|
September 29, 2020
|
vol. 117
|
no. 39
|
24349
ECOLOGYENVIRONMENTAL
SCIENCES
Downloaded at ucd university library on October 2, 2020
of the sth species. Finally, we calculated synchrony (log-variance ratio index:
logV) (16) as follows:
log V=lnvar S
i=1xi
S
i=1var(xi)
,[2]
where x
i
is the vector of abundances of the ith species over time. The logV
index ranges from −Inf to +ln(S). For this index, positive values indicate a
common response of the species (synchrony, formally positive sum of co-
variances in the variance–covariance matrix), while values close to zero in-
dicate a predominance of random fluctuations, and negative values indicate
negative covariation between species. One theoretical issue of this index is
that its upper limit is a function of species richness and evenness, ques-
tioning its independence from those parameters. Our results, however, were
not affected by this constraint. It is important to note that the observed
index value can vary considerably within its theoretical range; in fact, the
relationship between richness and logVindex is very weak. The chance of
reaching maximum synchrony decreases with the number of species. To
reach maximum synchrony, there must always be perfect synchrony between
all species pairs, no matter how many species are in the community [i.e., with
nspecies, the correlation of n(n−1)/2 pairs must be perfect (i.e. 1) within
each pair]. The values of synchrony that would be close to the maximum one
were not present in real communities (such as those that are the focus of this
manuscript). Thus, the upper limit of logV, which represents the caveat to
the use of this metric, is not invalidating our results.
To ensure that our results were not biased by the choice of this index, we
calculated other commonly used indices, specifically the Gross (11), Gross’
weighted (13), and phi (39) synchrony indices. Following Blüthgen et al. (13),
we weighted the abundance of species to decrease the influence of rare
species that can vary substantially while having a negligible abundance.
Both Gross and Gross’weighted synchrony indices were positively correlated
with logVindex (r=0.75 and 0.86, respectively) (SI Appendix, Table S4) and
gave concordant results. The phi synchrony index was also positively corre-
lated with the logVindex but negatively with species richness (r=0.48 and
0.41, respectively) (SI Appendix, Table S4), an expected output as this index
builds in the decrease in synchrony with increasing species richness expected
when species have independent population dynamics (39). We only present
the results of logVin the text both for clarity and because the models with
this index had the lowest Akaike information criterion (AIC) values and
explained more variance (R
2
m=0.59) (SI Appendix, Table S1) than those
using the alternate indices. Similarly, this index showed a greater difference
between the observed synchrony–stability relationships and the ones gen-
erated by null models (SI Appendix, Supplementary Text S2 has expanded
information).
Previous research has identified the relationship between stability and
synchrony, both in biological (12) and mathematical terms (1). However, it
has also been shown that stability is affected by a number of other factors
(1, 8, 12, 16, 25). Given these multiple influences, the relationship between
synchrony and stability would not necessarily be expected to be consistently
significant or characterized by a strong correlation. We assessed this rela-
tionship for the different indices in comparison with null models that assume
random, independent species fluctuations (SI Appendix, Supplementary Texts
S1 and S2 have expanded information).
We also considered the vegetation type of each plot based on the char-
acterization of the community by the authors of the study (grassland, shrub-
land, savanna, forest,and salt marsh). Savanna was characterized as a grassland
scatteredwith shrubs and/ortrees while maintaining an open canopy.For forest
plots, we restricted our analysis to datasets that measured understory
vegetation.
Analysis. Linear models were used to evaluate the relationships between 1)
synchrony and species richness, 2) species richness and stability, and 3) syn-
chrony and stability. In all cases, richness and stability were ln transformed to
improve their normality. We obtained the slope and the significance for these
relationships individually for each of the 79 datasets as well as for all of the
plots together. We used a null model approach to compare the observed
values of stability and synchrony and observed richness–synchrony and
richness–stability relationships to expected values under a random fluctua-
tion scenario. To do so, we randomized species abundances within a plot
across years, by means of torus randomizations (also referred to as cyclic
shifts). This approach preserves the temporal sequence of values within a
species but changes the starting year. In each individual plot, the sequence
of abundance values of each species was shifted 999 times, using a modifi-
cation of the “cyclic_shift”function in the codyn package for the R statistical
software (40). This procedure kept the total (i.e., summed) species abun-
dance constant for each species but varied (and therefore, disconnected) the
temporal coincidence of species abundances within years. Based on the 999
randomizations, we calculated values of mean expected synchrony and
stability. We used a paired ttest to evaluate the relationship between ob-
served and expected values of synchrony. We then tested the relationship
between observed species richness, 1) observed and expected synchrony,
and 2) observed and expected stability, using linear mixed effects models
with dataset as a random factor. Additionally, we used the same models to
test the relationship between observed synchrony and stability and expected
synchrony and stability.
We performed linear mixed effects models over all individual plots (n=
7,788) to assess the effects of the abiotic and biotic variables on synchrony
(logV). We included climatic data, vegetation type, percentage of woody
species, evenness, species richness, number of years each plot was sampled,
and environmental treatments as predictors in the model; dataset was a
random factor. Environmental treatments constituted a binary variable (0 =
control plots vs. 1 =environmental treatments). The mean and CI of the
parameter estimates of the predictors were used to model their effects on
synchrony values among all of the plots of the 79 studies. Mean annual
precipitation, temperature annual range, richness, and stability were ln
transformed to improve their normality. All predictors were centered on
their mean and standardized by their SD. For vegetation type, the param-
eter estimates were obtained by fixing grasslands as a reference level for the
other habitats. We analyzed the effects of the biotic and abiotic factors and
synchrony values on stability, using the same approaches previously de-
scribed. Although plot size was originally included in our model, this variable
was not significant (χ
2
<0.01; P=0.95) and so, was removed as predictor. To
evaluate the individual effect of each environmental treatment on syn-
chrony values and stability, treatments were grouped into six categories
(fertilization, herbivore exclusion, grazing intensity, removal, fire, and ma-
nipulative climate change drivers), retaining only datasets where these
treatments were applied or assessed.
Finally, we conducted a stepwise selection of a piecewise structural
equation model (41) to test direct and indirect pathways of biotic and abiotic
factors on stability. A piecewise structural equation model is a confirmatory
path analysis using a d-step approach (42, 43). This analysis is a flexible
framework to incorporate different model structures, distributions, and as-
sumptions. This method is based on an acyclic graph that summarizes the
hypothetical relationships between variables to be tested using the Csta-
tistic (44). We built an initial structural equation model containing all pos-
sible biotic and abiotic relationships, independent of the vegetation type
evaluated. Then, we used the AIC to select the minimal and best model (44)
based on the initial structural equation model, using the step AIC procedure
(41). This process selects the most important paths and removes the majority
of nonsignificant paths. Standardized path coefficients were used to mea-
sure the direct and indirect effects of predictors (45). We conducted the
structural equation model analyses across all individual plots (n=7,788), for
nontreatment plots across all habitats (n=4,013), and for plots of each
vegetation type separately (except in salt marsh). In all of the models,
datasets were considered as a random factor.
All analyses were carried out with R (R Core Team) (46) using packages
piecewiseSEM (47), lme4 (48), and modified source code in codyn (40).
Data Availability. The data that support the findings of this study are avail-
able in a txt format at Figshare (49) (https://doi.org/10.6084/m9.figshare.
7886582.v1).
ACKNOWLEDGMENTS. We thank multiple collaborators for the data they
provided (funding associated with particular study sites is listed in SI Appen-
dix, Supplementary Text S5). We also thank the Lawes Agricultural Trust and
Rothamsted Research for data from the Electronic Rothamsted Archive
(e-RA) database. We were supported by US NSF Grants DEB-8114302, DEB-
8811884, DEB-9411972, DEB-0080382, DEB-0620652, DEB-1234162, and DEB-
0618210; the Nutrient Network (https://nutnet.org/) experiment from NSF
Research Coordination Network Grant NSF-DEB-1042132; the New Zealand
National Vegetation Survey Databank; and Institute on the Environment
Grant DG-0001-13. Data (Dataset 56,SI Appendix, Supplementary Text S4)
owned by NERC Database Right/Copyright NERC. Further support was pro-
vided by the Jornada Basin Long-Term Ecological Research project, Cedar
Creek Ecosystem Science Reserve, and the University of Minnesota. The Roth-
amsted Long-term Experiments National Capability is supported by UK Bio-
technology and Biological Sciences Research Council Grant BBS/E/C/000J0300
and the Lawes Agricultural Trust. This research was funded by Czech Science
Foundation Grant GACR16-15012S and Czech Academy of Sciences Grant
RVO 67985939. E.V. was funded by 2017 Program for Attracting and Retain-
ing Talent of Comunidad de Madrid Grant 2017-T2/AMB-5406.
24350
|
www.pnas.org/cgi/doi/10.1073/pnas.1920405117 Valencia et al.
Downloaded at ucd university library on October 2, 2020
1. L. M. Thibaut, S. R. Connolly, Understanding diversity-stability relationships: Towards
a unified model of portfolio effects. Ecol. Lett. 16, 140–150 (2013).
2. D. Tilman, J. A. Downing, Biodiversity and stability in grasslands. Nature 367, 363–365
(1994).
3. F. Isbell et al., Quantifying effects of biodiversity on ecosystem functioning across
times and places. Ecol. Lett. 21, 763–778 (2018).
4. S. J. McNaughton, Stability and diversity of ecological communities. Nature 274,
251–253 (1978).
5. Y. Hautier et al., Plant ecology. Anthropogenic environmental changes affect eco-
system stability via biodiversity. Science 348, 336–340 (2015).
6. Y. Hautier et al., Eutrophication weakens stabilizing effects of diversity in natural
grasslands. Nature 508, 521–525 (2014).
7. F. Isbell et al ., Biodiversity increases the resistance of ecosystem productivity to cli-
mate extremes. Nature 526, 574–577 (2015).
8. L. M. Hallett et al., Biotic mechanisms of community stability shift along a precipita-
tion gradient. Ecology 95, 1693–1700 (2014).
9. C. de Mazancourt et al., Predicting ecosystem stability from community composition
and biodiversity. Ecol. Lett. 16, 617–625 (2013).
10. J. Zhang et al., Effects of grassland management on the community structure,
aboveground biomass and stability of a temperate steppe in Inner Mongolia, China.
J. Arid Land 8, 422–433 (2016).
11. K. Gross et al., Species richness and the temporal stability of biomass production: A
new analysis of recent biodiversity experiments. Am. Nat. 183,1–12 (2014).
12. K. S. McCann, The diversity-stability debate. Nature 405, 228–233 (2000).
13. N. Blüthgen et al., Land use imperils plant and animal community stability through
changes in asynchrony rather than diversity. Nat. Commun. 7,10697 (2016).
14. D. F. Doak et al., The statistical inevitability of stability-diversity relationships in
community ecology. Am. Nat. 151, 264–276 (1998).
15. T. J. Valone, N. A. Barber, An empirical evaluation of the insurance hypothesis in
diversity-stability models. Ecology 89, 522–531 (2008).
16. J. Lepš, M. Májeková, A. Vítová, J. Dole
zal, F. de Bello, Stabilizing effects in temporal
fluctuations: Management, traits, and species richness in high-diversity communities.
Ecology 99, 360–371 (2018).
17. A. Gonzalez, M. Loreau, The causes and consequences of compensatory dynamics in
ecological communities. Annu. Rev. Ecol. Evol. Syst. 40, 393–414 (2009).
18. E. Allan et al., More diverse plant communities have higher functioning over time due
to turnover in complementary dominant species. Proc. Natl. Acad. Sci. U.S.A. 108,
17034–17039 (2011).
19. M. Loreau, C. de Mazancourt, Biodiversity and ecosystem stability: A synthesis of
underlying mechanisms. Ecol. Lett. 16 (suppl. 1), 106–115 (2013).
20. A. R. Ives, K. Gross, J. L. Klug, Stability and variability in competitive communities.
Science 286, 542–544 (1999).
21. D. Tilman, Biodiversity: Population versus ecosystem stability. Ecology 77, 350–363
(1996).
22. M. Májeková, F. de Bello, J. Dole
zal, J. Lepš, Plant functional traits as determinants of
population stability. Ecology 95, 2369–2374 (2014).
23. D. Tilman, P. B. Reich, J. M. H. Knops, Biodiversity and ecosystem stability in a decade-
long grassland experiment. Nature 441, 629–632 (2006).
24. A. T. Tredennick, P. B. Adler, F. R. Adler, The relationship between species richness
and ecosystem variability is shaped by the mechanism of coexistence. Ecol. Lett. 20,
958–968 (2017).
25. D. Craven et al., Multiple facets of biodiversity drive the diversity-stability relation-
ship. Nat. Ecol. Evol. 2, 1579–1587 (2018).
26. J. E. Houlahan et al., Compensatory dynamics are rare in natural ecological commu-
nities. Proc. Natl. Acad. Sci. U.S.A. 104, 3273–3277 (2007).
27. J. Lepš, Variability in population and community biomass in a grassland community
affected by environmental productivity and diversity. Oikos 107,64–71 (2004).
28. J. Lepš, L. Götzenberger, E. Valencia, F. de Bello, Accounting for long‐term directional
trends on year‐to‐year synchrony in species fluctuations. Ecography 42, 1728–1741
(2019).
29. E. Valencia et al., Directional trends in species composition over time can lead to a
widespread overemphasis of year‐to‐year asynchrony. J. Veg. Sci., 10.1111/jvs.12916
(2020).
30. P. Chesson, N. Huntly, The roles of harsh and fluctuating conditions in the dynamics of
ecological communities. Am. Nat. 150, 519–553 (1997).
31. T. Sasaki, W. K. Lauenroth, Dominant species, rather than diversity, regulates tem-
poral stability of plant communities. Oecologia 166, 761–768 (2011).
32. T. J. Valone, J. Balaban-Feld, Impact of exotic invasion on the temporal stability of
natural annual plant communities. Oikos 127,56–62 (2018).
33. F. de Bello et al., Partitioning of functional diversity reveals the scale and extent of
trait convergence and divergence. J. Veg. Sci. 20, 475–486 (2009).
34. N. Pistón et al., Multidimensional ecological analyses demonstrate how interactions
between functional traits shape fitness and life history strategies. J. Ecol. 107,
2317–2328 (2019).
35. S. E. Koerner et al., Change in dominance determines herbivore effects on plant
biodiversity. Nat. Ecol. Evol. 2, 1925–1932 (2018).
36. B. H. McArdle, K. J. Gaston, The temporal variability of densities: Back to basics. Oikos
74, 165–171 (1995).
37. D. Tilman, C. L. Lehman, C. E. Bristow, Diversity-stability relationships: Statistical in-
evitability or ecological consequence? Am. Nat. 151, 277–282 (1998).
38. B. Smith, J. B. Wilson, A consumer’s guide to evenness indices. Oikos 76,70–82 (1996).
39. M. Loreau, C. de Mazancourt, Species synchrony and its drivers: Neutral and non-
neutral community dynamics in fluctuating environments. Am. Nat. 172, E48–E66
(2008).
40. L. M. Hallett et al., Codyn: An r package of community dynamics metrics. Methods
Ecol. Evol. 7, 1146–1151 (2016).
41. J. B. Grace, Structural Equation Modeling and Natural Systems, (Cambridge University
Press, Cambridge, United Kingdom, 2006).
42. B. Shipley, Confirmatory path analysis in a generalized multilevel context. Ecology 90,
363–368 (2009).
43. E. Laliberté, P. Legendre, A distance-based framework for measuring functional di-
versity from multiple traits. Ecology 91, 299–305 (2010).
44. B. Shipley, The AIC model selection method applied to path analytic models com-
pared using a d-separation test. Ecology 94, 560–564 (2013).
45. J. B. Grace, K. A. Bollen, Interpreting the results from multiple regression and struc-
tural equation models. Bull. Ecol. Soc. Am. 86, 283–295 (2005).
46. R Development Core Team, R: A Language and Environment for Statistical Comput-
ing, Version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria, 2018).
https://www.r-project.org/. Accessed 10 December 2018.
47. J. S. Lefcheck, S. E. M. Piecewise, Piecewise structural equation modelling in R for
ecology, evolution, and systematics. Methods Ecol. Evol. 7, 573–579 (2016).
48. D. Bates, M. Mächler, B. Bolker, S. Walker, Fitting linear mixed-effects models using
lme4. J. Stat. Softw. 67,1–48 (2014).
49. E. Valencia et al, Synchrony matters more than species richness in plant community
stability at a global scale. Figshare. https://doi.org/10.6084/m9.figshare.7886582.v1.
Deposited 18 November 2019.
Valencia et al. PNAS
|
September 29, 2020
|
vol. 117
|
no. 39
|
24351
ECOLOGYENVIRONMENTAL
SCIENCES
Downloaded at ucd university library on October 2, 2020
