ArticlePDF Available

Long‐term empirical evidence, early warning signals and multiple drivers of regime shifts in a lake ecosystem

Wiley
Journal of Ecology
Authors:

Abstract and Figures

Catastrophic regime shifts in various ecosystems are increasing with the intensification of anthropogenic pressures. Understanding and predicting critical transitions are thus a key challenge in ecology. Previous studies have mainly focused on single environmental drivers (e.g. eutrophication) and early warning signals (EWSs) prior to population collapse. However, how multiple environmental stressors interact to shape ecological behaviour and whether EWSs were detectable prior to the recovery process in lake ecosystems are largely unknown. We present long‐term empirical evidence of the critical transition and hysteresis with the combined pressures of climate warming, eutrophication and trophic cascade effects by fish stocking in a subtropical Chinese lake in the Yangtze floodplain. The catastrophic regime shifts are cross‐validated by 64‐year multi‐trophic level monitoring data and paleo‐diatom records. We show that EWSs are detectable in both the collapse and recovery trajectories and that including body size information in composite EWSs requires shorter time‐series data and can improve the predictive ability of regime shifts. Although full recovery has not yet been observed, EWSs prior to recovery provide us with the opportunity to take measures for a clear‐water regime. Climate warming and top‐down cascade effects have a negative influence on water clarity by altering lower trophic level abundance and body size, which, in turn, have a negative effect on macrophyte abundance. Furthermore, we identify a shift in the dominant driving forces from bottom‐up to top‐down after regime shifts, decoupling the relationships between nutrients and biological components and thus decreasing the efficiency of nutrient reduction. Synthesis. This study provides new insights into ecological hysteresis under multiple external stressors and improves our understanding of trait‐based early warning signals in both the collapse and recovery processes in natural freshwater ecosystems. For management practice, our work suggests that slowing down climate warming and weakening the fish predation pressure on food webs are necessary to increase the effectiveness of nutrient reduction in the restoration of lakes.
Infographic representing the characteristics of the clear‐water state (left, before the mid‐1980s) and the turbid‐water state (right, after the mid‐1980s) in the Donghu Lake ecosystem. Compared with the vegetated clear‐water state, the turbid state is characterized by a loss of submerged macrophytes, a high water nutrient concentration, a high fish stocking density, a low abundance of crustaceans, a high density of phytoplankton and a smaller body size of zooplankton and phytoplankton. A schematic diagram summarizes how fish stocking, climate warming and eutrophication influence the feedbacks between water clarity and submerged macrophytes. Red arrows represent positive effects, black arrows represent negative effects and the grey dashed arrow represents a nonsignificant path. The arrow width is proportional to the strength of the relationship. The numbers above the arrows indicate the path coefficients. Model fit summary: χ² = 10.83, df = 7, p = 0.146. A three‐dimensional conceptual model in the upper graph shows ecosystem behaviour under different pressure scenarios of eutrophication, fish stocking and climate warming. The nutrient‐limited bottom‐up effect was predominant in the forward clear‐water regime, whereas the top‐down cascade effect was predominant in the backward turbid‐water regime. With the increase in the fish stocking density and temperature, the hysteresis of the pressure response increased (indicated by the lower recovery threshold). In addition, the possible synergic effects of fish stocking and warming will further strengthen hysteresis, and lake recovery efforts (e.g. nutrient reduction) will be less effective as a result
… 
This content is subject to copyright. Terms and conditions apply.
3182  
|
wileyonlinelibrary.com/journal/jec Journal of Ecology. 2021;109:3182–3194.© 2020 British Ecological Society
Received: 3 June 2020 
|
Accepted: 26 Oc tober 2020
DOI: 10.1111/1365-2745.13544
RECONCILING RESILIENCE ACROSS
ECOLOGICAL SYSTEMS, SPECIES AND SUBDISCIPLINES
Research Article
Long-term empirical evidence, early warning signals and
multiple drivers of regime shifts in a lake ecosystem
Haojie Su1,2 | Rong Wang3| Yuhao Feng2| Yanling Li4| Yun Li1,3| Jun Chen1|
Chi Xu5| Shaopeng Wang2| Jingyun Fang2| Ping Xie1,4
1Donghu E xperimental Station of Lake
Ecosystems, State Key Laborator y of
Freshwater Ecolog y and Biotechnology,
Institute of Hydrobiolog y, Chinese Academy
of Sciences, Wuhan , China
2Depar tment of Ecology, College of Ur ban
and Environment al Sciences, Peking
University, Beijing, China
3Nanjing Institute of Geography &
Limnology, Chinese Academy of Sciences,
Nanjing, China
4Institute for Ecologica l Research and
Pollution Control of Plateau Lakes, School of
Ecology and Environment al Scie nce, Yunnan
University, Kunming, China
5School of Life Sciences, Nanjing University,
Nanjing, China
Correspondence
Ping Xie
Email: xieping@ihb.ac.cn
Funding information
The Strategic Priority Research Progr am of
the Chinese Aca demy of Sciences, Grant /
Award Number: XDB31000000; Th e Major
Science a nd Technology Program for Water
Pollution Control and Treatment, Grant/
Award Number: 2017ZX07203-00 4; The
Nationa l Key Research and Developm ent
Program of China, Grant /Award Number:
2017YFA0605201
Handling Editor: Guillaume de Lafontaine
Abstract
1. Catastrophic regime shifts in various ecosystems are increasing with the intensi-
fication of anthropogenic pressures. Understanding and predicting critical transi-
tions are thus a key challenge in ecology. Previous studies have mainly focused on
single environmental drivers (e.g. eutrophication) and early warning signals (EWSs)
prior to population collapse. However, how multiple environmental stressors in-
teract to shape ecological behaviour and whether EWSs were detectable prior to
the recovery process in lake ecosystems are largely unknown.
2. We present long-term empirical evidence of the critical transition and hysteresis
with the combined pressures of climate warming, eutrophication and trophic cas-
cade effects by fish stocking in a subtropical Chinese lake in the Yangtze flood-
plain. The catastrophic regime shifts are cross-validated by 64-year multi-trophic
level monitoring data and paleo-diatom records.
3. We show that EWSs are detectable in both the collapse and recovery trajecto-
ries and that including body size information in composite EWSs requires shorter
time-series data and can improve the predictive ability of regime shifts. Although
full recovery has not yet been observed, EWSs prior to recovery provide us with
the opportunity to take measures for a clear-water regime.
4. Climate warming and top-down cascade effects have a negative influence on
water clarity by altering lower trophic level abundance and body size, which, in
turn, have a negative effect on macrophyte abundance. Furthermore, we identify
a shift in the dominant driving forces from bottom-up to top-down after regime
shifts, decoupling the relationships between nutrients and biological components
and thus decreasing the efficiency of nutrient reduction.
5. Synthesis. This study provides new insights into ecological hysteresis under mul-
tiple external stressors and improves our understanding of trait-based early
warning signals in both the collapse and recovery processes in natural freshwa-
ter ecosystems. For management practice, our work suggests that slowing down
climate warming and weakening the fish predation pressure on food webs are
necessary to increase the effectiveness of nutrient reduction in the restoration
of lakes.
  
|
 3183
Journal of Ecolog
y
SU et al.
1 | INTRODUCTION
A burgeoning literature shows that ecosystems (freshwater, marine
and terrestrial) respond nonlinearly to environmental drivers, as an
abrupt transition occurs when a driver exceeds a threshold (Beisner
et al., 2003; Scheffer et al., 2001). Ecosystems often exhibit hys-
teresis even when external drivers return to previous conditions,
leading to not only serious degradation of ecological functions and
services but also difficulty in restoration practice (Hilt et al., 2017).
Intensive efforts have been made to understand regime shifts using
mathematical models (Biggs et al., 2009; Holling, 1973; May, 1977;
Scheffer et al., 2009), the reconstruction of paleo-ecological com-
munities (Bruel et al., 2018; Kong et al., 2017; Mcgowan et al., 2005;
Wang et al., 2012) and population-level and whole-lake experiments
(Carpenter et al., 2011; Faassen et al., 2015). However, there is little
direct empirical evidence to confirm system-level nonlinear changes
in natural ecosystems (Capon et al., 2015), as monitoring programmes
are often limited by short time-series lengths and low sampling res-
olutions. Although regime shifts can also be tested indirectly using
methods of space-for-time substitutions in field observations (Su,
Wu, et al., 2019; van Nes & Scheffer, 2005; Ward et al., 2018) or
remote sensing archives (Staver et al., 2011; Xu et al., 2016), spa-
tial data often have different temporal contexts and cannot provide
site-specific mechanisms of change.
Ecological resilience is the capacity of an ecosystem to maintain
its structure, functions and processes in the face of perturbations
(Holling, 1973). Theory suggests that early warning signals (EWSs)
ca n be us e d to mo n ito r subt le ch ange s in the sp atia l and te mpo r al be-
haviour of ecological systems as they approach a threshold (Carpenter
& Brock, 20 06; Clements & Ozgul, 2018; Dakos et al., 2012; Guttal
& Jayaprakash, 2008; Kefi et al., 2014; Scheffer et al., 20 09). These
statistical metrics are based on the phenomenon of critical slowing
down (CSD), which is characterized by a reduction in the recover y
rate after a small disturbance as an ecosystem approaches a cata-
strophic transition (Dakos et al., 2008). EWSs indicate a loss of re-
silience, in the sense that an ecosystem becomes more vulnerable
and can more easily tip to an alternative state. Increases in variance
and autocorrelation in systems prior to a transition have been shown
both theoretically and experimentally (Carpenter & Brock, 2006;
Spears et al., 2017), providing an ideal method for depicting the dy-
namics of resilience. Composite EWSs have also been proposed to
increase overall predictive ability by combining multiple indicators
(Drake & Griffen, 2010). In addition, including trait information (e.g.
body size) in composite EWSs provides more robust predictions than
traditional abundance-based time-series indicators alone (Arkilanian
et al., 2020; Baruah et al., 2020; Clements et al., 2017; Clement s &
Ozgul, 2016). However, most studies on ecological resilience and/
or EWSs focused on the deterioration of environments, whereas
ecological resilience in the recovery process received much less at-
tention (but see Clements et al., 2019).
Climate warming, eutrophication and fish stocking are three
major anthropogenic stressors undermining the functioning of
aquatic ecosystems (Daskalov et al., 2007; Moss et al., 2003; Smith &
Schindler, 2009). For instance, unprecedented rates of climate warm-
ing can induce phenological mismatches within food webs (Durant
et al., 2007; Schweiger et al., 2008; Winder & Schindler, 2004),
alter the forging behaviour of consumers (Woodward et al., 2010)
and reduce the body size of aquatic individuals (Forster et al., 2012;
Yvon-Durocher et al., 2011), which may alter the feedbacks that
maintain a st able regime and the resilience of ecosystems to distur-
ban ce (Wernberg et al., 2010). Fur thermore, warmin g an d fish stock-
ing may interact in a synergistic way through ecological networks
(Christensen et al., 2006; Jackson et al., 2016), which enlarge the
strength of top-down cascade effects. Thus, clarifying how multi-
ple stressors interact and shape the structure and functioning of
ecosystems is critical for understanding the dynamics of ecological
behaviour.
Plankton community abundance and body size are important
ecosystem properties and can be used to explore the mechanisms
through which water clarity functionally declines. Body size is an
important species trait that links the structure and stability of food
webs (Brose et al., 2019; Spanbauer et al., 2016). This metric is re-
lated to energy requirement s, gape size and life history, and it plays
a vital role in determining the strength of trophic cascades (Delong
et al., 2015). A previous study showed that bottom-up effects from
nutrients primarily structure plankton abundance, whereas commu-
nity composition and body size distributions are mainly determined
by top-down cascade ef fects from fish (Lemmens et al., 2018).
Water clarity is functionally correlated with not only phytoplank-
ton abundance but also body size. Under the same biomass con-
ditions, it is generally accepted that systems with smaller body
sizes of suspended particles have lower water clarity (Bhargava &
Mariam, 1991). As ecosystems in nature are often affected by mul-
tiple stressors simultaneously (Crain et al., 2010), changes in the rel-
ative importance of bottom-up and top-down effec ts may decouple
the relationships between nutrients and biological processes, which
cause the hysteretic response of an ecosystem to nutrient reduction
in the back ward recovery trajectory.
In the present work, using monitoring data spanning 6 4 years
(1956–2019) and microfossil records reconstructed from dated
se dim e nt co r es in th e subt r opi c al sha llo w La ke Do n ghu (F igu re S1),
we provide empirical quantitative evidence of abrupt shifts and
hysteresis at the ecosystem level. We then assess whether trait-
based composite EWSs were reliably detectable in advance of
both the collapse and recovery processes and evaluate how short
the time-series data are required to be before a collapse/recovery
KEYWORDS
body size, catastrophic transitions, climate warming, ecological resilience, eutrophication,
hysteresis, paleolimnology, trophic cascade
3184 
|
  
Journal of Ecology
SU et al.
is present in composite EWSs. Finally, we unravel the underlying
mechanisms of how climate warming, fish stocking and eutrophi-
cati on intera c t in dr iving nonlin ear reg ime shi fts. We hi gh light that
the transition of the dominant driving forces from bottom-up to
top-down caused hysteresis that prevented the recovery efforts
of nutrient reduction.
2 | MATERIALS AND METHODS
2.1 | Study site
Lake Donghu (30°33′02″N, 114°21′40″E), a subtropical shallow
lake in the middle and lower Yangtze Plain, is a city-central lake
located in Wuhan, Hubei Province (Figure S1). The mean depth of
Lake Donghu is 2.21 m, and the lake surface area is 32 km2. The
ever-growing human population and rapid development of indus-
try, agriculture and animal husbandry in the drainage basin have led
to severe eutrophication since the 1970s. The total nitrogen (TN)
conc entratio n inc rea sed from 0.76 mg /L in th e 195 0 s to 2. 35 mg /L
in the 1990s and decreased to 1.31 mg/L in the 2010s; the total
phosphorus (TP) concentration increased from 0.068 mg/L in the
1950s to 0.176 mg/L in the 1990s and decreased to 0.107 mg/L in
the 2010s. From 2002, municipal engineering was carried out to
separate rain and sewage water around the lake. Consequently,
the water nutrient concentration has shown a decreasing trend in
recent years (Figure S2a,b).
Beginning in 1972, fish stocking activities were implemented
to increase the fish yield of the lake (Figure S2c), including in-
creasing the stocking densities of bighead carp and silver carp,
improving the proportion of large fingerlings, reconstructing
fish screens and controlling predatory fish. These juvenile fishes
subsist on natural food resources in the water column without
artificial feeding. The fish yield increased from 98.25 kg/ha be-
fore the 1970s to 1,194.23 kg/ha in 2003 and then decreased to
757.6 kg/ha after the 2010s. The mean proportion of planktivo-
rous fish (i.e. silver carp and bighead carp) was 84.71% from 1973
to 1978 and increased to 98% after the 1990s. The fish yield of
grass carp Ctenopharyngodon idellus was 8.9% from 1973 to 1976
and decreased to less than 1% after that time. Before the 1970s,
submerged macrophyte vegetation was abundant in the lake. An
investigation in 1962–1963 showed that the vegetation coverage
was 83%, with a mean biomass of 1,614 g/m2. Submerged mac-
rophyte vegetation decreased to 5.8 g/m2 in 1975, and the once
dominant species, Potamogeton maackianus, disappeared in that
year and since then has never recovered (Liu, 1991).
2.2 | Long-term field observations
Long-term monitoring data (1956–2019) were obtained from the
Donghu Experimental Station, which is under the framework of
the Chinese Ecosystem Research Net work (CERN). The monitoring
programme collects ecosystem-level abiotic (e.g. TP, TN, water clar-
ity and temperature) and biotic variables (e.g. phytoplankton, zoo-
plankton and fish). Surveys were conducted monthly at two sampling
stations in the Guozheng Lake area representing coastal (Station I)
and pelagic zones (Station II; Figure S1). As water temperature is
highly correlated with air temperature (Figure S3), the monthly mean
air temperature was used as a substitute for the water temperature
in further analyses. Water samples from each site were collected
0.5 m below the water surface and 0.5 m above the lake bottom
using a 5 L Schindler sampler and then mixed together for subse-
quent analyses. The water clarity was characterized as the Secchi
depth, which was measured with a Secchi disk. The water samples
were taken to the laboratory to analyse gross primary productiv-
ity, TN and TP concentrations in accordance with standard methods
(Huang et al., 1999). Fish yield data were obtained annually from the
Fishery Management Committee of Lake Donghu.
Mixed water samples (1 L) were preserved with 4% formal-
dehyde and 1% Lugol's solution and concentrated to 50 ml after
sedimentation for at least 48 hr. Phytoplankton and rotifers were
counted in the concentrated samples under 400× magnification.
The biomass of each taxon was calculated using the approximating
geometric forms. Mixed water samples (10 L) were sieved through
64-μm plankton nets and preserved with 5% formalin for further
analysis of crustacean zooplankton. To maintain the consistency
of the data, phytoplankton and zooplankton were sampled and
identified with the same method in recent decades to reduce ob-
servation bias. We considered functional groups (e.g. cladocerans,
copepods and rotifers for zooplankton) in our analysis, which is
rarely influenced by species identification. The phytoplankton were
categorized into six major groups with distinct sizes and functions,
including Cyanophyta, Chlorophyta, Bacillariophyta, Pyrrophyta,
Euglenophyta and Chrysophyta. Furthermore, to ensure the accu-
racy of the data, we resampled, reanalysed or recounted the samples
if abnormal values were found. The body sizes of the cladocerans
and phy toplankton were calculated as the biomass divided by the
density, with mg per individual (mg/ind.) as the unit.
2.3 | Paleo-limnological records
On 8 July 2010, a 50-cm sediment core was collected in Lake Donghu
(30°33′02″N, 114°21′40″E) at a water depth of 2 m using a Kajak
gravity corer with a 58 mm diameter. The sediment core was sliced
at 0.5 cm intervals for dating and diatom analyses. The chronologies
of the core were obtained by measuring 210Pb and 137Cs radionuclide
activities in contiguous samples at the State Key Laboratory of Lake
Science and Environment of the Chinese Academy of Sciences (CAS),
Nanjing, China. A total of 98 sediment samples were prepared for
diatom analysis using the standard method (Battarbee et al., 2001).
Diatom species were identified using oil immersion at 1,000 magni-
fication under an Olympus microscope (BX51). Diatom concentra-
tions were estimated using DVB microspheres, which are expressed
as relative percent abundances.
  
|
 3185
Journal of Ecolog
y
SU et al.
2.4 | Statistical analyses
Change-point analysis was used to test the changes in the mean
value of water clarity. The sequential F-statistic of water clarity was
calculated at a confidence level of p < 0.01 with a moving window
of 5% of the time series using the strucchange package in r 4.0.0
(Zeileis et al., 2001). The change point occurred at the maximum
of the F-statistic, which was determined by the ‘breakpoints’ func-
tion. The same method was also applied to nine other state variables
to test the reliability of the estimated breakpoint date, that is, the
abundance of cladocerans, copepods, rotifers, Cyanophyta and total
phytoplankton, the PC1 of phytoplankton community composition,
the proportion of Cyanophyta to total phytoplankton abundance,
and the body size of cladocerans and phytoplankton. Furthermore,
diatom assemblage records over time retrieved from the sediment
core were analysed using principal component analysis (PCA) and
chronological clustering to confirm the identification of breakpoints.
Changes in the mean principal component (PC) of diatoms were also
assessed by change-point analysis. Chronological clustering was
performed using the ‘coniss’ method on a bray distance matrix using
the rioja package in r (Juggins, 2015). The output of the dendrogram
indicated the main temporary regimes.
To demonstrate hysteresis in the ecosystems, we plotted the re-
sponse state variables (i.e. water clarity, total phytoplankton abun-
dance, cladoceran abundance, the body size of phytoplankton and
the body size of cladocerans) against the main environmental stress
over time. PCA was used to extract the main time trends from the
multiple environmental stresses. The first PC axis scores are the-
oretical variables that contain the most information regarding the
original total variance (Jolliffe, 2002). Here, the main environmental
stress was calculated as the first PC of TN, TP, temperature and fish
(which explained 53.5% of the total variance).
Composite EWSs were first proposed by Drake and Griffen
(2010) and have since been commonly used (Arkilanian et al., 2020;
Clements et al., 2017, 2019; Clements & Ozgul, 2016). Following
Clements and Ozgul (2016), we also consider body size decline prior
to a population approaching a collapse as an EWS. In doing so, the
normalized value of body size was multiplied by −1 so that it can
be included in the composite EWS (i.e. increasing when approaching
a tipping point). To derive the composite EWS, we first normalized
each leading indicator (AR1, CV and body size) by subtracting the
long-run average of the respective indicator and dividing by the
long-run standard deviation (Dr ake & Grif fen, 2010). Thus, each sta-
tistic at time t (
wt
) was calculated as follows:
where
w1:t
is the mean of a statistic from times 1 to t and
SD (
w
1:
t
)
is
the standard deviation over the same period. The composite EWS was
then calculated by summing the values for the leading indicators to be
included at each time point (e.g. Figure S4 for AR1 + CV + Size.phy-
toplankton). An EWS was considered to be present if the composite
metric at any time exceeds its running mean value by more than 2σ
(more than 2 when the composite metric was normalized).
We present the composite EWS in two different ways. The first
is an analysis using the entire time series. This method exhibit s the
detailed dynamics of the composite EWS and provides general in-
formation on how predictable the tipping point was. Second, we also
calculated the minimum length of required time-series data to gener-
ate an EWS, as the length of data used in the analysis may affect the
predictive ability. We did so by analysing the whole time series prior
to the tipping point and then iteratively −1 year to the analysis until
all composite EWSs vanished. This approach provides information on
whether our results are sensitive to the window size and whether in-
cluding trait-based EWSs are more predictive than traditional leading
indicators. In this study, we analysed the EWSs prior to the collapse
and recovery independently to avoid the effects of state shift itself on
the resilien ce ind icators . The diatom data use d in the col la pse path are
the sum abundance of macrophyte-attached species (i.e. Eunotia sp.,
Epithemia sorex and Gyrosigma accuminatum), which are the dominant
species in a clear-water regime. The diatom data used in the recovery
path are the sum abundance of pelagic species. In addition, we used
the traditional single early warning indicator SD’ to test whether fl ic k-
ering was observed before a bifurcation (Figure S5). Water clarity and
the PC1 of phytoplankton community composition were used as state
variables. Significance was determined by comparing the original non-
parametric Kendall tau correlation coefficient to the trends obtained
from the surrogate data (Dakos et al., 2012). The ‘surrogates_ews’
func tion in the early warnings package was used to perform this anal-
ysis, and the default setting was used. That is, the size of the rolling
window was 50%, and the Gaussian detrending method was used
(bandwidth was 10%) prior to analysis.
Theil–Sen regression was used to robustly explore the effects of
fish stocking and temperature on water clarity, plankton abundance
and body size. Theil–Sen estimation is a robust method that deter-
mines the slope of the regression line via the median of the slopes.
The mblm package in r was used to perform this analysis. Bottom-up
and top-down control are known to be critical determinants of eco-
system structure and functions in aquatic habitats. The correlation
coefficient between the time series of nutrients (or fish) and pop-
ulation abundance has probably been commonly used to assess
the type and strength of trophic control within ecosystems (Boyce
et al., 2015; Carpenter & Kitchell, 1996; Frank et al., 2005; Ripple
et al., 2016). Strong positive correlations indicate bottom-up control,
and strong negative correlations between adjacent trophic levels in-
dicate top-down control, as predators suppress the abundance of
their prey. We tested whether the dominant forces changed before
and after regime shifts using the Pearson correlation coefficient.
Structural equation modelling (SEM) was used to explore the
pathways by which fish stocking, temperature and nutrients affected
wat er clarity through the foo d we b an d to test a posit ive fe ed ba ck be-
tween water clarity and macrophyte abundance (represented by the
macrophy te-attached diatom abundance). We first constructed a full
model that included all possible pathways; then, nonsignificant path-
ways were eliminated to optimize the model. SEM, which involved
w
t=
wtw1:t
SD
(
w
1:
t
),
3186 
|
  
Journal of Ecology
SU et al.
sets of multiple regression analyses, allowed a rigorous estimation of
the causal relationship network (Grace, 2006). The standardized path
coefficients between two variables represented the relative strength
of a relationship. For the category of nutrients that encompassed
more than one parameter (i.e. TN and TP), the first PC (PC1), which
explained 78.1% of the total variance, was introduced as a new vari-
able into the SEM. We used a χ2 tes t and p values to evaluate the fit of
the models (e.g. models were considered to have a good fit when the
p values ranged from 0.1 to 1). As missing values are not allowed for
many analyses (e.g. composite EWS analysis and SEM), we used the
linear interpolation method in the analysis. We conducted the SEM
analyses using AMOS 21.0 (Amos Development Corporation).
3 | RESULTS
3.1 | Abrupt transition of multiple state variables
The lake experienced a regime shift in the mid-1980s from a clear-
water state to a phytoplankton-dominated turbid state. The eco-
system exhibited hysteresis, as the values of environmental drivers
dropped below the values at which bi-stability was observed in the
forward collapse. We provide direct evidence of regime shifts for mul-
tiple ‘state variables’ in the mid-1980s in Lake Donghu. Specifically,
we selected 10 state variables that comprehensively represented the
biotic and abiotic conditions of the ecosystem to robustly assess the
exis ten ce of reg im e shi f ts (Table 1; Fi gur e 1) . Th e spe cie s com posit ion
of phytoplankton has shown profound changes in recent decades
(Fi gu re S6). Wate r cl ar it y, as well as the parameters of the ab undance,
species composition and body size of phytoplankton and zooplank-
ton, all generally showed abrupt changes in the middle of the 1980s
(Table 1; Figure 1), except for rotifer (1982) and copepod abundances
(1996). All state variables presented in Table 1 showed significant dif-
ferences (p < 0.001) in their mean values before and after regime
shifts, for example, water clarity showed an abrupt change in 1985,
with the mean water clarity decreasing from 1.77 to 0.86 m after the
shift (Table 1). In addition, Lake Donghu has experienced increas-
ing environmental stresses in recent decades. The annual mean air
temperature increased significantly (R2 = 0.31, p < 0.001; Figure S2;
Table 1), with the mean temperature increasing from 16.70 ± 0.41
to 17.30 ± 0.49 after the regime shift. The fish stocking density
also showed a significant increase (R2 = 0 .69, p < 0.001; Figure S2),
with the mean fish yield increasing from 263.09 ± 210.10 to
903 .67 ± 180.49 kg/ha after the regime shift. The TN in the water
column also increased significantly (p = 0.015) after the regime shift,
but for TP, the change was not statistically significant.
3.2 | Hysteresis in the recovery path
By plotting these same state variables—namely, water clarity, log10
total phytoplankton density, cladoceran density, the body size of
phytoplankton and body size of cladocerans—as a function of envi-
ronmental stress, we reveal a folded bifurcation, which indicates a
TABLE 1 Overview of the 10 state variables and four stressors. The breakpoint and mean value (mean ± SD) of the clear-water regime
and turbid-water regime, the direction of the shif t and the significance of the difference after regime shifts are listed. Here, the breakpoint
is the last year of the clear-water regime, and 1985 was used as the breakpoint to compare the mean values of the stressors before and after
the critical transition
Parameter Breakpoint Clear-water regime Turbid-water regime
Direction of
shift p
State variables Water clarity, m 1985 1.77 ± 0.32 0.86 ± 0.13 Decrease <0.001
Cladoceran, ind./L 1986 2 9.46 ± 12.03 6.07 ± 4.88 Decrease <0.001
Copepods, ind./L 1996 61.24 ± 35. 52 14.6 0 ± 14 .8 3 Decrease <0.001
Rotifer, ind./L 1982 842.17 ± 7,49 9.13 2,406.23 ± 945.06 Increase <0.001
Log10 total phytoplankton,
ind./L
1986 5.88 ± 0.40 7. 74 ± 0.34 Increase <0.001
Log10 Cyanophyta, ind./L 1986 5.32 ± 0.56 7. 52 ± 0.30 Increase <0.001
PC1 of the phytoplankton
assemblages
1985 0.51 ± 0.33 0.34 ± 0 .19 Increase <0.001
Cyanophy ta, % 1985 30.13 ± 1 7.9 5 72.24 ± 14.10 Increase <0.001
Body size of cladoceran, mg/
ind.
1986 0.06 ± 0.02 0.02 ± 0.01 Decrease <0.001
Log10 body size of
phytoplankton, mg/ind.
1986 −4.60 ± 0.28 −7.0 8 ± 0.37 Decrease <0.001
Stressors Total nitrogen, mg/L 1.32 ± 0.53 1.90 ± 0.91 Increase 0.015
Total phosphorus, mg/L 0.12 ± 0.08 0.15 ± 0.07 Increase 0.245
Temperature, 16.70 ± 0.41 17. 3 0 ± 0.49 Increase <0.001
Fish, kg/ha 263.09 ± 210.10 91 7. 5 4 ± 185.82 Increase <0.001
  
|
 3187
Journal of Ecolog
y
SU et al.
symptom of hysteresis (Figure 1, right column). The ecosystem was
alternatively in either a clear or a turbid state when the stress was in
the range of values between −0.21 and 0.08. The temporal trajec-
tory followed by the ecosystem state suggests a different forward
and backward path characterizing hysteresis in the lake's recovery
(Figure 1, right column).
3.3 | Evidence of regime shifts from paleo-
diatom records
Our paleo-diatom records retrieved from sediment cores also de-
tected the presence of a regime shift (Figure 2), in agreement with
the patterns found in the long-term ecological monitoring data. The
chronological clustering of diatom assemblages identified two tem-
poral groups, with a division in 1984. These findings were consistent
with a visual inspection of the PC1 (accounting for 83.5% of the total
variance) of the sediment core. Specifically, the diatom community
was dominated by planktonic Aulacoseira granulata before 1984 (49
16 cm), which had a mean abunda nce of 60.6%. After 1984 (16–0 cm),
A. granulata declined rapidly, and Stephanodicus minutulus became the
dominant species. The periphytic taxa, such as Eunotia sp., Epithemia
sorex and Gyrosigma accuminatum, which are often attached to sub-
me rge d mac r oph y tes , sho wed a decr e asi ng tr end af ter th e mi d-198 0s.
3.4 | EWS analysis
The maximum values of normalized composite EWSs were greater
than the 2σ threshold, indicating that EWSs were present for dif-
ferent organisms (phytoplankton, cladocerans and paleo-diatom)
in both the collapse and recovery trajectories (Figure 3a,c). In ad-
dition, composite metrics that include body size information gen-
erally have smaller value of minimum leng th of required data and
higher proportion of years with detectable EWS (Table S1). That
is, including trait-based EWSs have stronger predictive ability than
abundance-based metrics alone in both the collapse and recover y
trajectories (Figure 3b,d). EWSs were detectable up to 10 years be-
fore a collapse when macrophyte-attached diatom abundance data
were used (AR1 + CV), whereas detectable E WSs prior to recovery
would take at least 13 years when trait-based cladoce ran dat a were
used (AR1 + CV + Size). The normalized SD score increased in both
the collapse (p = 0.015 and 0.095 for water clarity and phytoplank-
ton PC1, respectively) and recovery (p = 0.025 and 0.005 for water
clarity and phy toplankton PC1, respectively), which showed an in-
creasing variation approaching the tipping point (Figure S5).
3.5 | Multiple factors in driving the ecosystem to a
turbid state
Theil–Sen regressions showed that both the fish yield and tempera-
ture had negative relationships with water clarity (both p < 0.001;
Figure 4a,b) and cladoceran abundance (both p < 0.001; Figure 4e,f)
and positive relationships with the abundance of total phytoplank-
ton (both p < 0.001; Figure 4c,d). Furthermore, the fish yield and
temperature exhibited negative relationships with phytoplankton
FIGURE 1 Abrupt shifts and hysteresis, respectively, in
Lake Donghu in state variables (a, b) water clarity, (c, d) log10
phytoplankton abundance, (e, f) cladoceran abundance, (g, h) log10
phytoplankton body size and (i, j) cladoceran body size. A sequential
F-statistic with a 5% sliding window was used to detect regime
shifts for the entire period. The blue horizontal lines in the left
column represent the mean value before and after regime shifts.
The blue curves in the right column were fitted using the quadratic
parabola equation. The red dashed lines indicate the border of the
two alternative stable states. Data points are coloured by year and
clearly show the hysteretic trajectory along environmental stress.
Here, environmental stress was calculated as the first principal
component of temperature, water total nitrogen, total phosphorus
and the fish yield
(ind./L)(ind./L)(mg/ind.)
(mg/ind.)
(a)
(c)
(e)
(g)
(i) (j)
(h)
(f)
(d)
(b)
3188 
|
  
Journal of Ecology
SU et al.
body size (both p < 0.001; Figure 4g,h) and cladoceran body size
(both p < 0.001; Figure 4i,j).
Int er es ting ly, despite th e negative ef fe cts of fish an d warming
on crustacean zooplankton, cladocerans showed an increasing
trend before regime shifts (Figure 1e). Furthermore, we found
that TN and TP had positive relationships w ith gross primar y pro-
ductivity (r = 0.727 and 0.676, p < 0.001 and p = 0.001), roti-
fer abundance (r = 0.653 and 0.562, p < 0.0 01 and p = 0.005),
FIGURE 2 Evidence for regime
shifts in Lake Donghu from paleo-
limnological diatom records. Only taxa
with a mean abundance above 2% are
shown. The first principal components
of diatom community assemblages
account for 83.5% of the total variance.
The grey horizontal line represents the
critical transition year, that is, 1984.
Chronological clustering analysis with the
Bray–Curtis distance was used
FIGURE 3 Performance of the composite early warning signal (EWS) prior to the (a, b) collapse and (c, d) recovery processes. (a, c) show
the mean trends of the normalized composite EWS using whole time-series data and (b, d) show how short the EWS is present prior to the
tipping point. The horizontal black dashed line indicates a 2σ threshold where an EWS was considered to be present. The solid dot s in (b, d)
indicate that EWSs are present in that year and the colours represent the different composite EWSs (see legends in the top left of Figure 3a).
Grey shaded areas indicate the shortest length of required time-series data for the appearance of the EWSs. EWSs were detectable with as
little as 10 years (collapse) and 13 years (recovery). PP, phytoplankton; Cla, cladoceran
(a)
(c) (d)
(b)
  
|
 3189
Journal of Ecolog
y
SU et al.
copepod abundance (r = 0.675 and 0.638, both p < 0.001) and
cladoceran abundance (r = 0.567 and 0.373, p = 0.005 and
p = 0.079) in the clear-water state while these relationships
weakened or disappeared after regime shifts (Table 2). These
results suggest that the bottom-up effect was dominant in the
clear-water stage, whereas top-down forces were dominant in
the turbid-water stage. Changes in the relative impor tance of
environmental drivers may influence ecosystem responses and
FIGURE 4 Effects of the fish yield
and temperature, respectively, on (a, b)
water clarity, (c, d) log10 phytoplank ton
abundance, (e, f) cladoceran abundance,
(g, h) log10 phytoplank ton size and (i, j)
cladoceran size. Slopes of the regression
line were determined by the Theil–Sen
slope estimator. Body size was calculated
as the ratio of biomass to abundance, with
mg per individual as the unit
(ind./L)
(ind./L)
(mg/ind.)
(mg/ind.)
(kg/ha)
p < 0.001
p < 0.001
p < 0.001
p < 0.001
p < 0.001 p < 0.001
p < 0.001
p < 0.001
p < 0.001
p < 0.001
(a)
(c)
(e)
(g)
(i) (j)
(h)
(f)
(d)
(b)
3190 
|
  
Journal of Ecology
SU et al.
behaviours, consequently leading to hysteretic recovery in re-
sponse to nutrient reduction.
The SEM results suggest that fish stocking, warming and eutro-
phication had cascade effects on the food webs in various ways.
Our model fit the data well (χ2 = 10.83, df = 7, p = 0.146) and
explained 49%, 79%, 72% and 81% of the variation in cladoceran
abundance, phytoplankton abundance, macrophyte abundance
and water clarity, respectively (Figure 5). Fish had a direct negative
influence on macrophyte abundance (−0.35). In addition, fish and
temperature directly and indirectly had negative effects on water
Correlated variables
Clear-water regime Turbid-water regime
n r p n r p
TN-gross primar y
production
20 0.727 <0.001 35 0. 61 <0.001
TN-rotifer 23 0.653 <0.001 49 0.157 0.281
TN-copepods 23 0.675 <0.001 47 0.292 0.046
TN-cladoceran 23 0.567 0.005 49 0.123 0.401
TP-gross primary
production
20 0. 676 0.001 35 0 .414 0.013
TP-rotifer 23 0.562 0.005 49 0.16 4 0.260
TP-copepods 23 0.638 <0.001 47 0.223 0 .132
TP-cladoceran 23 0.373 0.079 49 0.067 0.649
TABLE 2 Pearson correlations
between lake water nutrients (TN and TP)
and primary production and zooplankton
abundance (rotifer, copepods and
cladoceran) in the clear-water regime
and turbid-water regime. The number
of observation years (n), correlation
coefficients (r) and significance values (p)
are indicated
FIGURE 5 Infographic representing the characteristics of the clear-water state (left, before the mid-1980s) and the turbid-water state
(right, after the mid-1980s) in the Donghu Lake ecosystem. Compared with the vegetated clear-water state, the turbid state is characterized
by a loss of submerged macrophytes, a high water nutrient concentration, a high fish stocking density, a low abundance of crustaceans,
a high density of phytoplankton and a smaller body size of zooplankton and phytoplankton. A schematic diagram summarizes how fish
stocking, climate warming and eutrophication influence the feedbacks between water clarity and submerged macrophytes. Red arrows
represent positive effects, black arrows represent negative effects and the grey dashed arrow represents a nonsignificant path. The arrow
width is proportional to the strength of the relationship. The numbers above the arrows indicate the path coefficients. Model fit summary:
χ2 = 10.83, df = 7, p = 0.146. A three-dimensional conceptual model in the upper graph shows ecosystem behaviour under different pressure
scenarios of eutrophication, fish stocking and climate warming. The nutrient-limited bottom-up ef fect was predominant in the forward clear-
water regime, whereas the top-down cascade effect was predominant in the backward turbid-water regime. With the increase in the fish
stocking density and temperature, the hysteresis of the pressure response increased (indicated by the lower recovery threshold). In addition,
the possible synergic effects of fish stocking and warming will further strengthen hysteresis, and lake recovery efforts (e.g. nutrient
reduction) will be less effective as a result
  
|
 3191
Journal of Ecolog
y
SU et al.
clarity, which, in turn, exerted a negative feedback effect on mac-
rophyte abundance.
4 | DISCUSSION
In recent decades, the concepts of regime shift and alternative
stable states in multiple ecosystem types, such as lakes, oceans,
coral reefs and arid lands (Scheffer et al., 2001), have gained
much attention. However, a previous study reviewed 135 papers
on f re shwa te r ecosy st em s a nd showed that t here is litt le em pi ri-
cal evidence in natural ecosystems to confirm pressure-induced
nonlinear changes (Capon et al., 2015). It is generally accepted
that regime shifts should occur at the system-wide level, with
changes being detectable across multiple physical and biological
components, including reconstructed palaeontological commu-
nities (Dong et al., 2020; Kong et al., 2017; Zhang et al., 2018).
Although the methods for reconstructing microfossil records are
restricted by low resolution and the few taxa, paleo-limnological
proxies can provide long-term reference data to track ecosys-
tem dynamics . In th e present s tudy, we p rovided cros s-v alidated
multidisciplinary evidence to verify critical transitions in mul-
tiple trophic levels and water environments and the existence
of alternative stable states within a range of environmental
conditions.
4.1 | EWS detection prior to the collapse and
recovery processes
Ecosystems worldwide are facing biodiversity loss and ecosys-
tem function degradation under multiple anthropogenic pres-
sures. Thus, predicting ecosystem collapse (forward shift) and
recovery (backward shift) is a critical challenge because it is re-
lated to management costs and conservation efficiency. Many
efforts have been made to predict a regime shift; however,
whether EWSs can be detectable in a real-world recovery pro-
cess is largely overlooked. Although Clements et al. (2019) ob-
served EWSs of recovery with marine fish catch data, tests in
natural freshwater ecosystems have not yet been conducted. In
this study, a composite EWS is detectable prior to regime shifts
in both the collapse and recovery processes. Although the full
recovery (shift back to a clear state) has not yet been obser ved,
EWS s indic ate a lo s s of resi l ien ce of th e tu rbid st ate , which po i nts
to the system being on a recovery trajector y. We also found that
including body size information in composite E WSs can be more
robust for signalling abrupt changes in ecosystem structure and
functions (Clements et al., 2017; Clements & Ozgul, 2016). As
the upper bound of the diatom series data is 2010, the compos-
ite EWS of diatoms (AR1 + CV) has a lower predictive ability of
recover y than that of phy toplankt on and clad ocer ans (F ig ur e 3d),
indicating that an EWS metric closer to the tipping point has
higher pre dictive abilit y. Our resu lt s su ggested that EWSs c an be
de tec t abl e even wh en ful l reco ve r y was not obs er ve d , w hic h p ro-
vide s u s wit h t he op por tun ity to take meas ur es to re store a cl ear-
wate r eco sys tem . In man ag eme nt pr actice , in add ition to nutr ien t
reduction, more positive measures are needed (e.g. a reduction
in the fish stocking density and the planting of submerged mac-
rophytes) to promote the recovery transition to a macrophyte-
dominated clear-water state.
4.2 | Mechanisms underlying the critical
transition and hysteresis
This study included data on multiple stressors and trophic levels,
which allowed us to explore internal trophic interactions and re-
veal the mechanisms underlying critical transitions in the whole
ecosystem. Multiple anthropogenic stressors potentially exert posi-
tive feedbacks that provoke an unpredictable ‘ecological surprise’
in lake ecosystems (Birk et al., 2020; Davis et al., 2010; Jackson
et al., 2016). For instance, the synergistic effects of herbivor y
and periphyton shading trigger macrophyte loss and regime shifts
(Hidding et al., 2016). Our results showed that climate warming and
the trophic cascade by fish stocking had direct and indirect negative
influences on water clarity (Figure 5), jointly driving the ecosystem
to a turbid state and decoupling the relationships between nutrient s
and biological components. Body size miniaturization of plankton
and the flourishing of phytoplankton abundance are probably key
to understanding the decline in water clarity, as small organisms
floating in the water column have a high specific surface area and
can more efficiently scatter and reflect light. In turn, the decline in
water clarity has a negative effect on macrophyte abundance, form-
ing a positive feedback loop in which ‘the more turbid the water, the
fewer macrophytes there are’ (Su, Chen, et al., 2019; van der Heide
et al., 2011; this study, Figure 5).
Understanding the mechanisms of hysteresis is critical for
management practices. It has been assumed that the recovery
to a macrophyte-dominated ecosystem can be hampered by low
wate r clar ity, gra zin g by fis h o r bir ds, wa te r-le ve l fluc tuati ons an d
a lack of propagules or seeds in the sediment (Hilt et al., 2006).
In the present study, we propose a three-dimensional conceptual
model to iconically depict the mechanisms of critical transition
and hysteresis (Figure 5). In the forward bottom-up controlled
system, according to the theory of alternative lake equilibria, eu-
trophication facilitates primary producers and thereby reduces
water clarity. In the backward trajector y, fish-driven top-down
forces control ecosystem behaviour, forming an ecological hys-
teresis of recovery even when the nutrient levels drop below the
values at which the ecosystem collapses. As respiration is more
sensitive to warming than photosynthesis (Allen et al., 2005;
Yvon-Durocher et al., 2010), warmer temperatures will
strengthen the top-down forces due to the increased metabolic
rate and resource demand (Kratina et al., 2012). Furthermore, it
is suggested that both fish stocking and warming will increase
water nutrient availability, for example, by direct excretion of
3192 
|
  
Journal of Ecology
SU et al.
fish, elevated nutrient release from sediment, enhanced miner-
alization rates and increased evaporation (Domis et al., 2013;
Gudasz et al., 2010), further weakening the efforts of nutrient
reduction and enhancing hysteresis. Moreover, food webs may
show evolutionary responses to top-down effects (e.g. altering
genotype frequencies in the cladoceran population), making the
recovery process slow or even impossible (Kuparinen & Merilä,
2007; Stokes & Law, 20 00).
5 | CONCLUSIONS
We provide convincing empirical evidence for pressure-induced
nonlinear ecological changes in a subtropical shallow lake, cross-
validated by long-term monitoring and paleo-diatom data. The
Donghu Lake ecosystem shifts from a macrophyte-dominated clear-
water state to a phytoplankton-dominated turbid-water state under
the combined pressures of climate warming, eutrophication and
trophic cascade by fish stocking. We found that EWSs were detect-
able in both the forward collapse and backward recovery trajecto-
ries and that including body size information in composite EWSs
was more robust for assessing ecological resilience approaching a
transition. In a multi-driver-controlled ecosystem, dominant driv-
ers changing from bottom-up to top-down forces after the critical
transition lowered the effectiveness of nutrient reduction in the
turbid regime. Therefore, in practice, in addition to nutrient reduc-
tion, more positive management measures, such as reducing the fish
stocking density, slowing climate warming and carr ying out macro-
phyte planting, are important and necessary for the anticipatory re-
covery to a clear-water state.
ACKNOWLEDGEMENTS
We would like to thank all members who have ever participated
in the monitoring programme of the Donghu Lake ecosystem.
This work was supported by the Strategic Priority Research
Program of the Chinese Academy of Sciences (XDB3100 0000),
the National Key Research and Development Program of China
( 2 01 7 Y F A0 6 05 20 1 ) an d th e Ma jo r Sc i e n c e an d Te ch no l o g y P r o gr am
for Water Pollution Control and Treatment (2017ZX07203-00 4).
We wo uld like to thank Soni a Ke fi fo r pro vidin g use ful feed ba ck on
earlier versions of the manuscript. The authors declare no conflict
of interest.
AUTHORS' CONTRIBUTIONS
P.X. and H.S. design ed the re searc h; H.S., Y.L ., Ya. L. and J.C. contrib-
uted to the sampling and data collection; H.S. and Y.F. devised the
figure structure and performed the data analyses; H.S. wrote the
manuscript and P.X., R.W., C.X., S.W. and J.F. substantially contrib-
uted to revisions.
DATA AVAIL ABIL IT Y S TAT EME NT
Long-term monitoring data can be obtained from Donghu Experimental
Station of Lake Ecosystems (http://dhl.cern.ac.cn/meta/metaData).
Data were also available from the Dr yad Digital Repository https://doi.
org/10.5061/dryad.7m0cf xpsf (Su et al., 2020).
ORCID
Haojie Su https://orcid.org/0000-0003-4780-1094
Chi Xu https://orcid.org/0000-0002-1841-9032
Shaopeng Wang https://orcid.org/0000-0002-9430-8879
REFERENCES
Allen, A., Gillooly, J., & Brown, J. (20 05). Linking the global carbon cycle
to individual metabolism. Functional Ecology, 19(2), 202–213. https://
doi.org/10.1111/j.1365-2435.2005.00952.x
Arkilanian, A. A., Clements, C. F., Ozgul, A., & Baruah, G . (2020). Effec t
of time series length and resolution on abundance- and trait-based
early warning signals of population declines. Ecology, 101(7), e0304 0.
htt ps://doi.org/10.10 02 /ecy.30 40
Baruah, G., Clements, C. F., & Ozgul, A. (2020). Eco-evolutionary processes
underlying early warning signals of population declines. Journal of
Animal Ecology, 89(2), 436–448. https://doi.org/10.1111/1365-2656.
13097
Battarbee, R., Jones, V., & Flower, R. (2001). Tracking environmental
change using lake sediments: Volume 3: Terrestrial, algal, and siliceous
indicators (Vol. 3): Springer.
Beisner, B. E., Haydon, D. T., & Cuddington, K. (2003). Alternative sta-
ble states in ecology. Frontiers in Ecology and the Environment, 1(7),
376–382. https://doi.org/10.2307/3868190
Bhargava, D., & Mariam, D. W. (1991). Effects of suspended particle size
and concentration on reflectance measurements. Photogrammetric
Engineering and Remote Sensing, 57(5), 519–529.
Biggs, R., Carpenter, S. R., & Brock, W. A. (2009). Turning back from
the brink: Detecting an impending regime shif t in time to avert it.
Proceedi ngs of the National Academy of Sciences of the United States
of America, 106 (3), 826–831. https://doi.org/10.1073/pnas.08117
29106
Birk, S., Chapman, D., Carvalho, L ., Spears, B. M., Andersen, H. E.,
Argillier, C., Auer, S., Baat trup-Pedersen, A., Banin, L., Beklioğlu,
M., Bondar-Kunze, E., Borja, A., Branco, P., Bucak, T., Buijse, A. D.,
Cardoso, A. C., Couture, R .-M., Cremona, F., de Zwart, D., … Hering,
D. (2020). Impacts of multiple stressors on freshwater biota across
spatial scales and ecosystems. Nature Ecol ogy & Evolution, 4(8), 1060–
1068 . ht tps://doi.org/10.1038/s4155 9- 020-1216- 4
Boyce, D. G., Frank, K. T., Worm, B., & Leggett, W. C. (2015). Spatial
patterns and predictors of trophic control in marine ecosystems.
Ecology Letters, 18(10 ), 1001–1011. https://doi.org/10 .1111/el e.
12481
Brose, U., Archambault, P., Barnes, A. D., Bersier, L.-F., Boy, T., Canning-
Clode, J., Conti, E., Dias, M., Digel, C., Dissanayake, A., Flores, A.
A. V., Fussmann, K., Gauzens, B., Gray, C ., Häussler, J., Hirt, M. R.,
Jacob, U., Jochum, M., Kéfi, S., … Iles, A. C. (2019). Predator traits
determine food-web architecture across ecosystems. Nature Ecology
& Evolution, 3(6), 919–927. https://doi.org/10.10 38/s4155 9-019-
089 9-x
Bruel, R., Marchet to, A., Bernard, A., L ami, A., Sabatier, P., Frossard, V.,
& Perga, M. E. (2018). Seeking alternative stable states in a deep
lake. Freshwater Biology, 63(6), 553–56 8. ht tp s://doi.o rg /10.1111/
fwb.130 93
Capon, S. J., Lynch, A. J. J., Bond, N., Chessman, B. C., Davis, J., Davidson,
N., Finlayson, M., Gell, P. A., Hohnberg, D., Humphrey, C., Kingsford,
R. T., Nielsen, D., Thomson, J. R., Ward, K., & Nally, R. M . (2015).
Regime shifts, thresholds and multiple stable states in freshwa-
ter ecosystems; a critical appraisal of the evidence. Science of the
Total Environment, 534, 122–130. https://doi.org/10.1016/j.scito
tenv.2015.02.045
  
|
 3193
Journal of Ecolog
y
SU et al.
Carp enter, S. R., & Brock, W. A. (20 06). Rising variance: A leading indica-
tor of ecological transition. Ecology Letters, 9(3), 311–318. http s://do i.
org /10.1111/j .1461- 024 8. 20 05 .0 0877.x
Carpenter, S. R ., Cole, J. J., Pace, M. L., Batt, R., Brock, W. A., Cline, T.,
Coloso, J., Hodgson, J. R., Kitchell, J. F., Seekell, D. A., Smith, L., &
Weidel, B. (2011). Early warnings of regime shifts: A whole-eco-
system experiment. Science, 332 (6033), 1079–1082. ht tps://doi.
org/10.1126/scien ce.1203672
Carpenter, S. R., & Kitchell, J. F. (1996). The trophic cascade in lakes.
Cambridge University Press.
Christensen, M. R., Graham, M. D., Vinebrooke, R. D., Findlay, D. L., Paterson,
M. J., & Turner, M. A. (2006). Multiple anthropogenic stressors cause
ecological surprises in boreal lakes. Global Change Biology, 12(12),
2316–2322. https://doi.org/10.1111/j.1365-2486.2006.01257.x
Clements, C. F., Blanchard, J. L., Nash, K. L., Hindell, M. A., & Ozgul, A .
(2017). Body size shifts and early warning signals precede the his-
toric collapse of whale stocks. Nature Ecology & Evolution, 1(7), 188.
htt ps://doi.org/10.10 38/s4155 9-017-0188
Clements, C. F., McCarthy, M. A., & Blanchard, J. L. (2019). Early warning
signals of recover y in complex sys tems. Nature Communications, 10 .
https://doi.org/10.1038/s4146 7-019-09684 -y
Clements, C. F., & Ozgul, A . (2016). Including trait-based early warning
signals helps predict population collapse. Nature Communications, 7.
htt ps://doi.org/10.10 38/ncomm s10984
Clements, C. F., & Ozgul, A. (2018). Indicators of transitions in biological
systems. Ecology Letters, 21(6), 905–919. ht tp s://doi.o rg /10.1111/
ele.1294 8
Crain, C. M., Kroeker, K., & Halpern, B. S. (2010). Interactive and cumulative ef-
fects of multiple human stressors in marine systems. Ecology Letters, 11(12),
1304–1315. https://doi.org/10.1111/j.1461-0248.2008.01253.x
Dakos, V., Carpenter, S. R., Brock, W. A ., Ellison, A. M., Gut tal, V., Ives,
A. R., Kéfi, S., Livina, V., Seekell, D. A., van Nes, E. H., & Scheffer, M.
(2012). Methods for detecting early warnings of critical transitions
in time series illustrated using simulated ecological data. PLoS ONE,
7(7). https://doi.org/10.1371/journ al.pone.0041010
Dakos, V., Scheffer, M., van Nes, E. H., Brovkin, V., Petoukhov, V., & Held,
H. (2008). Slowing down as an early warning signal for abrupt climate
change. Proceedings of the National Academy of Sciences of the United
States of America, 105(38), 14308–14312. https://doi.org/10.1073/
pnas.08024 30105
Daskalov, G. M., Grishin, A. N., Rodionov, S., & Mihneva, V. (2007). Trophic
ca sca des tr igg e red by ove rfis hin g rev eal po ssi ble me cha nis ms of eco-
system regime shifts. Proceedi ngs of the National Academy of Sciences
of the United States of America, 104(25), 10518–10523. https://doi.
org /10.1073/pnas.07011 0010 4
Davis, J., Sim, L., & Chambers, J. (2010). Multiple stressors and regime shifts
in shallow aquatic ecosystems in antipodean landscapes. Freshwater
Biology, 55, 5–18. https://doi.org/10.1111/j.1365-2427.2009.02376.x
De senerpont domis, L. N., Elser, J. J., Gsell, A. S., Huszar, V. L. M., Ibelings,
B. W., Jeppesen, E., Kosten, S., Mooij, W. M., Roland, F., Sommer, U.,
Van donk, E., Winder, M., & Lürling, M. (2013). Plankton dynamics
under different climatic conditions in space and time. Freshwater
Biology, 58(3), 463–482. https://doi.org/10.1111/fwb.12053
DeLong, J. P., Gilbert, B., Shurin, J. B., Savage, V. M., Barton, B. T.,
Clements, C. F., Dell, A. I., Greig, H. S., Harley, C. D. G., Kratina,
P., McCann, K. S., Tunney, T. D., Vasseur, D. A., & O’Connor, M. I.
(2015). The body size dependence of trophic cascades. The American
Naturalist, 185(3), 354–366. https://doi.org/10.1086/679735
Dong, X., Kat tel, G., & Jeppesen, E. (2020). Subfossil cladocerans as
quantit ative indicator s of past ecological conditions in Yangtze River
Basin lakes, China. The Science of the Total Environment, 728, 138794.
https://doi.org/10.1016/j.scito tenv.2020.138794
Drake, J. M., & Griffen, B. D. (2010). Early warning signals of extinction in
deteriorating environments. Nature, 467(7 314 ), 456 459. http s: //do i.
org /10.103 8/nat ur e09389
Durant , J. M., Hjermann, D. Ø., Ottersen, G., & Stenseth, N. C. (2007).
Climate and the match or mismatch between predator requirements
and resource availability. Climate Research, 33(3), 271–283. https://
doi.org/10.3354/cr033271
Faassen, E. J., Veraart, A. J., Van Nes, E. H., Dakos, V., Lürling, M., & Scheffer,
M. (2015). Hysteresis in an experimental phytoplankton population.
Oikos, 124(12), 1617–1623. https://doi.org/10.1111/oik.02006
Forster, J., Hirst, A. G., & Atkinson, D. (2012). Warming-induced reductions
in body size are greater in aquatic than terrestrial species. Proceedings
of the National Academy of Sciences of the United States of America,
109(47), 19310–19314. https://doi.org/10.1073/pnas.12104 60109
Frank, K. T., Petrie, B., Choi, J. S., & Leggett , W. C. (2005). Trophic c as-
cades in a formerly cod-dominated ecosystem. Science, 308(5728),
1621–1623. https://doi.org/10.1126/scien ce.1113075
Grace, J. B. (2006). Struc tural equation modeling an d natural systems.
Cambridge University Press.
Gudasz, C., Bastviken, D., Steger, K., Premke, K., Sobek, S., & Tranvik, L. J. (2010).
Temperature-controlled organic carbon mineralization in lake sediments.
Nature, 466(7305), 478–481. https://doi.org/10.1038/natur e09186
Guttal, V., & Jayaprakash, C. (2008). Changing skewness: An early warn-
ing signal of regime shifts in ecosystems. Ecology Letters, 11(5), 450–
460. https://doi.org/10.1111/j.1461-0248.2008.01160.x
Hidding, B., Bakker, E. S., Hootsmans, M. J., & Hilt, S. (2016). Synergy be-
tween shading and herbivory triggers macrophyte loss and regime
shifts in aquatic systems. Oikos, 125(10), 1489–1495. https://doi.org/
10.1111/oik.03104
Hilt, S., Brothers, S., Jeppesen, E., Veraart, A. J., & Kosten, S. (2017).
Translating regime shifts in shallow lakes into changes in ecosystem
functions and services. BioScience, 67, 928–936. https://doi.org/
10.1093/biosc i/bix10 6
Hilt, S., Gross, E. M., Hupfer, M., Morscheid, H., Mählmann, J., Melzer, A.,
Poltz, J., Sandrock, S., Scharf, E.-M., Schneider, S., & van de Weyer,
K. (2006). Restoration of submerged vegetation in shallow eutrophic
lakes – A guideline and state of the art in Germany. Limnologica, 36(3),
155–171. https://doi.org/10.1016/j.limno.2006.06.001
Holling, C. S. (1973). Resilience and stabilit y of ecological systems.
Annual R eview of Ecology Evoluti on and Systematics , 4(4) , 1–23 . htt ps: //
doi.org/10.1146/annur ev.es.04.110173.000245
Huang, X., Chen, W., & C ai, Q. (1999). Sur vey, observation and analysis of
lake ecolog y. Standard methods for observation and analysis in Chinese
Ecosystem Re search Network, Series V. Standards Press of China.
Jackson, M. C., Loewen, C. J., Vinebrooke, R. D., & Chimimba, C. T.
(2016). Net effects of multiple stressors in freshwater ecosystems:
A meta-analysis. Global Change Biology, 22(1), 180–189. https://doi.
org /10.1111/gcb.13 028
Jolliffe, I. (20 02). Principal component analysis. Springer.
Juggins, S. (2015). rioja: Analysis of Quaternary science data. R package
version (0.9-5). The Comprehensive R Archive Network.
Kéfi, S., Guttal, V., Brock, W. A., Carpenter, S. R., Ellison, A. M., Livina, V. N.,
Seekell, D. A., Scheffer, M., van Nes, E. H., & Dakos, V. (2014). Early warn-
ing signals of ecological transitions: Methods for spatial patterns. PLo S
ONE, 9(3), e92097. https://doi.org/10.1371/journ al.pone.0092097
Kong, X., He, Q., Yang, B., He, W., Xu, F., Janssen, A . B. G ., Kuiper, J. J.,
van Gerven, L. P. A., Qin, N., Jiang, Y., Liu, W., Yang, C., Bai, Z., Zhang,
M., Kong, F., Janse, J. H., & Mooij, W. M. (2017). Hydrological regula-
tion drives regime shifts: Evidence from paleolimnology and ecosys-
tem modeling of a large shallow Chinese lake. Global Change Biology,
23(2), 737–754. https://doi.org/10.1111/gcb.13416
Kratina, P., Greig, H. S., Thompson, P. L., Carvalho-Pereira, T. S. A., &
Shurin, J. B. (2012). Warming modifies trophic cascades and eutro-
phication in experimental freshwater communities. Ecology, 93(6),
1421–1430. https://doi.org/10.1890/11-1595.1
Kuparinen, A., & Merilä, J. (2007). Detecting and managing fisheries-
induced evolution. Trends in Ecology & Evolution, 22(12), 652–659.
https://doi.org/10.1016/j.tree.2007.08.011
3194 
|
  
Journal of Ecology
SU et al.
Lemmens, P., Declerck, S. A., Tuytens, K., Vanderstukken, M., & De Meester, L.
(2018). Bottom-up effects on biomass versus top-down effects on iden-
tity: A multiple-lake fish community manipulation experiment. Ecosystems,
21(1), 166–177. https://doi.org/10.1007/s1002 1-017-0144-x
Liu, J. K. (1991). Ecological studies of Lake Donghu. Science Press.
May, R. M. (1977). Thresholds and breakpoint s in ecosystems with a
multiplicity of stable states. Nature, 269 (5628), 471–477. https://doi.
org /10.1038/269471a0
McGowan, S., Leavitt, P. R., Hall, R. I., A nderson, N. J., Jeppesen, E., &
Odgaard, B. V. (20 05). Controls of algal abundance and community
composition during ecosystem state change. Ecolog y, 86(8), 2200–
2211. https://doi.org/10.1890/04-1029
Moss, B., Mckee, D., Atkinson, D., Collings, S. E., Eaton, J. W.,
Gill, A. B., Har vey, I., Hatton, K ., Heyes, T., & Wilson, D.
(2003). How important is climate? Effect s of warming, nutri-
ent addition and fish on phytoplankton in shallow lake micro-
cosms. Journal of Applied Ecology, 40(5), 782–792. https://doi.
org/10.1046/j.1365-2664.2003.0 0839.x
Ripple, W. J., Estes, J. A., Schmitz, O. J., Constant, V., Kaylor, M. J., Lenz,
A., Motley, J. L., Self, K. E., Taylor, D. S., & Wolf, C. (2016). What is
a trophic cascade? Trends in Ecology & Evolution, 31(11), 842–849.
https://doi.org/10.1016/j.tree.2016.08.010
Scheffer, M., Bascompte, J., Brock , W. A ., Brovkin, V., Carpenter, S. R.,
Dakos, V., Held, H., van Nes, E. H., Rietkerk, M., & Sugihara, G. (2009).
Early-warning signals for critical transitions. Nature, 461(7260), 53–
59. http s://doi.org /10.1038/natur e0 8227
Scheffer, M., Carpenter, S., Foley, J. A., Folke, C., & Walker, B. (20 01).
Catastrophic shift s in ecosystems. Nature, 413, 591–596. http s://doi.
org /10.103 8/3509800 0
Schweiger, O., Set tele, J., Kudrna, O., Klotz, S., & Kühn, I. (2008).
Climate change can cause spatial mismatch of trophically interact-
ing species. Ecology, 89(12), 3472–3 479. ht tps://doi .or g/10.189 0/
07-174 8 .1
Smith, V. H., & Schindler, D. W. (2009). Eutrophication science: Where
do we go from here? Trends in Ecology & Evolution, 24(4), 201–207.
https://doi.org/10.1016/j.tree.2008.11.009
Spanbauer, T. L., Allen, C. R ., Angeler, D. G., Eason, T., Frit z, S. C.,
Garmestani, A. S., Nash, K. L ., Stone, J. R., Stow, C. A ., & Sundstrom,
S. M. (2016). Body size dis tributions signal a regime shift in a lake
ecosystem. Pro ceedings of the Royal Society B: Biologica l Sciences,
283(1833). https://doi.org/10.1098/rspb.2016.0249
Spears, B. M., Futter, M. N., Jeppesen, E., Huser, B. J., Ives, S., Davidson,
T. A., Adrian, R., Angeler, D. G., Burthe, S. J., Carvalho, L., Daunt , F.,
Gsell, A. S., Hessen, D. O., Janssen, A. B. G., Mackay, E. B., May, L.,
Moorhouse, H., Olsen, S., Søndergaard, M., … Thackeray, S. J. (2017).
Ecological resilience in lakes and the conjunction fallacy. Nature
Ecology & Evolution, 1(11), 1616–1624. https://doi.org/10.1038/
s4155 9-017-0333-1
Staver, A. C., Archibald, S., & Levin, S. A. (2011). The global extent and de-
terminants of savanna and forest as alternative biome states. Science,
334(6053), 230–232. https://doi.org/10.1126/scien ce.1210465
Stokes, K., & L aw, R. (200 0). Fishing as an evolutionar y force. Mari ne
Ecology Progress, 208, 307–309. Retrieved from https://www.jstor.
org/stabl e/24863829
Su, H., Chen, J., Wu, Y., Chen, J., Guo, X., Yan, Z., Tian, D. I., Fang, J., &
Xie, P. (2019). Morphological traits of submerged macrophytes re-
veal specific positive feedbacks to water clarity in freshwater eco-
systems. Science of the Total Environment, 684, 578–586. https://doi.
org/10.1016/j.scito tenv.2019.05.267
Su, H., Wang, R., Feng, Y., Li, Y., Li, Y., Chen, J., Xu, C., Wang, S., Fang,
J., & Xie, P. (2020). Dat a from: Long-ter m empirical evidence, early
warning signals and multiple drivers of regime shifts in a lake
ecosystem. Dryad Digital Repository, ht tps://do i.o rg /10.50 61/dr y-
ad.7m0cf xpsf
Su, H., Wu, Y., Xia, W., Yang, L., Chen, J., Han, W., Fang, J., & Xie, P.
(2019). Stoichiometric mechanisms of regime shifts in freshwater
ecosystem. Water Research, 149, 302–310. https://doi.org/10.1016/
j.watres.2018.11.024
van der Heide, T., van Nes, E. H., van Katwijk, M. M., Olff, H., & Smolders,
A. J. (2011). Positive feedbacks in seagrass ecosystems–evidence
from large-scale empirical data. PLoS ONE, 6(1), e16504. https://doi.
org/10.1371/journ al.pone.0016504
van Nes, E. H., & Scheffer, M. (2005). Implications of spatial heterogene-
ity for catastrophic regime shifts in ecosystems. Ecology, 86(7), 1797–
1807. https://doi.org/10.1890/04-0550
Wang, R., Dearing, J. A., Langdon, P. G., Zhang, E., Yang, X., Dakos, V., &
Scheffer, M. (2012). Flickering gives early warning signals of a criti-
cal transition to a eutrophic lake state. Nature, 492(7429), 419–422.
https://doi.org/10.1038/natur e11655
Ward, D. F., Wotherspoon, S., Melbourne-Thomas, J., Haapkylä, J., &
Johnson, C. R. (2018). Detecting ecological regime shifts from transect
data. Ecological Monographs, 88(4), 694–715. https://doi.org/10.1002/
ecm.1312
Wernberg, T., Thomsen, M. S., Tuya, F., Kendrick, G. A., Staehr, P. A., &
Toohey, B. D. (2010). Decreasing resilience of kelp beds along a lat-
itudinal temperature gradient: Potential implications for a warmer
future. Ecology Letters, 13(6), 685–694. https://doi.org/10.1111/
j.1461-0248.2010.01466.x
Winder, M., & Schindler, D. E. (2004). Climate change uncouples trophic in-
teractions in an aquatic ecosystem. Ecology, 85(8), 2100–2106. https://
doi.org/10.1890/04-0151
Woodward, G., Perkins, D. M., & Brown, L. E. (2010). Climate change and
freshwater ecosystems: Impacts across multiple levels of organization.
Philosophical Transactions of the Royal Society B: Biological Sciences, 365,
2093–2106. ht tps://doi.org/10.1098/rs tb. 2010.0055
Xu, C., Hantson, S., Holmgren, M., Nes, E. H., Staal, A., & Scheffer, M. (2016).
Remotely sensed canopy height reveals three pantropical ecosystem
states. Ecology, 97(9), 2518–2521. https://doi.org/10.1002/ecy.1470
Yvon-Durocher, G., Jones, J. I., Trimmer, M., Woodward, G., & Montoya,
J. M. (2010). Warming alters the metabolic balance of ecosystems.
Philosophical Transac tions of the Royal Society B: Biological Sciences,
365(1549), 2117–2126. https://doi.org/10.1098/rstb.2010.0038
Yvon-Durocher, G., Montoya, J. M., Trimmer, M., & Woodward, G. (2011).
Warming alters the size spectrum and shifts the distribution of bio-
mass in freshwater ecosystems. Global Change Biology, 17( 4), 16 81–
1694. https://doi.org/10.1111/j.1365-2486.2010.02321.x
Zeileis, A., Leisch, F., Hornik, K., & Kleiber, C. (2001). strucchange: An R
package for testing for structural change in linear regression models.
Journal of Statistical Software, 7(2), 1–38. https://doi.org/10.18637/
js s .v 0 0 7. i0 2
Zhang, K., Dong, X., Yang, X., Kattel, G., Zhao, Y., & Wang, R. (2018).
Ecological shif t and resilience in China's lake systems during the last
two centuries. Global and Planetary Change, 165, 147–159. https://
doi.org/10.1016/j.glopl acha.2018.03.013
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Supporting Information section.
How to cite this article: Su H, Wang R, Feng Y, et al. Long-term
empirical evidence, early warning signals and multiple drivers
of regime shifts in a lake ecosystem. J Ecol. 2021;109:3182
3194. https://doi.o rg /10.1111/1365-2745.1354 4
... Various restoration measures can promote a regime shift in shallow lakes from an algae-dominated turbid-water state to a submerged macrophyte-dominated clear-water state (Shi et al., 2023). Although early warning signals can be detected before regime collapse and recovery, there is a time-lag effect in this regime transition (Su et al., 2020). Thus, it is difficult to restore the water quality of lakes that are degraded by human activity. ...
... Sedimentary records show that water quality in many lakes has deteriorated in Wuhan City since the mid-twentieth century, suggested by the loss of submerged macrophytes (Qiu et al. 2001), gradual increases in cyanobacterial pigments (Zeng et al. 2023) and eutrophic diatom and chironomid species (Cao et al. 2020;Su et al. 2020;Chen et al. 2022b). Since continuous monitoring data of water quality in most lakes began in 2005, and hence limited knowledge is available on decadal changes in lake water quality. ...
Article
Full-text available
A series of restoration measures such as municipal wastewater treatment and aquaculture closures have been implemented in Wuhan City during recent years. In order to explore the impact of restoration measures and climate change on lake water quality, long-term (2005–2021) water quality data of 47 lakes were explored to reveal spatiotemporal changes in lake water quality. Percentages of polluted lakes were calculated according to six water-quality parameters, including total phosphorus (TP), ammonia nitrogen (NH3-N), chemical oxygen demand (COD), biological oxygen demand (BOD), chemical oxygen demand using potassium permanganate as oxidant (CODMn) and petroleum contamination (PET), at interannual and monthly timescales. At the interannual timescale, percentages of COD, BOD, CODMn and PET pollution decreased significantly, suggestive of water quality improvement during recent years. At the monthly timescale, low percentages of NH3-N and BOD pollution in March 2020 probably resulted from the sharp reduction in human activities during the COVID-19 lockdown. At the monthly timescale, temperature was positively correlated with percentage of CODMn pollution, but negatively correlated with percentage of NH3-N pollution; precipitation was negatively correlated with percentage of BOD pollution. The similarity of water-quality parameters generally decreased with an increase in geographical distance between each pair of lakes. However, the regression coefficients between the similarity of lake water quality and the geographical distance decreased with time, probably resulting from enhanced similarity of water quality parameters among all lakes with rapid urbanization. Our results highlight the importance of active restoration measures for sustainable management of lakes in Wuhan City, as well as in similar developing regions.
... Benefiting from long-term and high-quality continuous empirical data, paleolimnology provides a powerful method to reveal the real-world effects of biodiversity on ecological resilience. 41 Lake sediment subfossils including diatoms, chironomids, and cladocerans can provide valuable information on the dynamics of community structure and species interactions with environmental change ( Figure 1C), and hence valuable insight into the transition process of ecosystems. 42,43 In this study, we conducted a data synthesis of 53 lakes worldwide based on integrated paleoecological subfossil records over the past hundreds of years. ...
Article
Full-text available
Ecological resilience is critical for ecosystems to persist in the face of perturbations without shifting to a different state. Global biodiversity loss in multiple ecosystems is considered to be associated with decreasing ecological resilience and increasing the risk of ecosystem collapse. However, how temporal changes in biodiversity affect ecological resilience in natural ecosystems remains poorly elucidated. By analyzing subfossil records of diatoms, chironomids, and cladocerans from 53 lake sediment cores across the globe, we found that species richness showed an increasing trend with time, while beta diversity and ecological resilience presented a decreasing temporal trend when ecosystems are approaching the abrupt shift. Asynchronous fluctuations among species and temporal stability at species level are suggested to be the mechanisms that contribute to the maintenance of temporal community stability. We found species richness and beta diversity have positive effects on species asynchrony but negative effects on species stability. However, we found that species asynchrony and species stability had negative relationships with ecological resilience, with species richness and beta diversity overall having no positive effects on ecological resilience. We highlighted that biodiversity effects on ecological resilience are not only the role of species richness, but also the species assemblage and network complexity of species-species interactions. Our study indicates that increased species diversity and community heterogeneity may be not beneficial for the ecosystem to recover from disturbances at a lake ecosystem scale, which has great implications for the assessment of ecological resilience and predicting ecosystem collapse in future global environmental change scenarios.
... One of the most common biological features, which has been documented in many shallow lakes, is the transition from macrophytes to algae (clear to turbid state; Huang et al. 2021). Other important biological features of lake shift are changes in fish community structure (Bao et al. 2021;Su et al. 2021). Such changes can have cascading effects in lakes, most implying increased predation on larger zooplankton, which in turn means less grazing on phytoplankton and in consequence higher algal biomass per unit of available phosphorus (Jeppesen et al. 2010;Meerhoff et al. 2012). ...
Article
Full-text available
Climate change exerts a profound impact on lakes, eliciting responses that range from gradual to abrupt transitions. When reaching critical tipping points, the established lake dynamics stand to undergo substantial modifications, setting off a chain reaction that reverberates through the entire ecosystem. This lake shift ripples into related ecosystem services and even influences the well‐being of human communities. Despite the importance of lake shifts, we lack a systematic overview of their occurrence, mainly due to the lack of systematic data at the global scale. We reviewed the literature focusing on climate‐related lake shifts and assessed how satellite Earth Observation (EO) has contributed to the research topic, and what we can unlock from this novel data. Our results show that EO data are used in only 9% of studies on lake shifts, although this fraction has increased since 2012. EO data is most commonly used to assess shifts in surface extent, ice coverage, or phytoplankton phenology. These variables are directly observable and the spatio‐temporal resolution of EO satellites is of great advantage. But lake shifts can also be identified indirectly from EO data, as in the example of the vertical mixing of lake water, which can be described on the basis of surface patterns. In all possible applications, we expect increasing use of EO satellites in the future, including the development of early warning systems that promise to provide timely alerts regarding impending lake shifts, thus serving as a vanguard against abrupt alterations that could ripple through interconnected ecosystem services.
Article
Full-text available
Global ocean warming and extreme climate events pose a severe threat to marine biodiversity by inducing species redistribution and ecosystem reorganization. It is important to quantify the impacts of marine heatwaves (MHWs) on marine cetacean habitats to avoid rapid ecosystem shifts. Here we utilized detected breakpoints and early warning indicators derived from sightings data spanning from 2009 to 2021 to assess the distribution change of Indo-Pacific humpback dolphin (Sousa chinensis). We found that (1) during the 2014 MHWs, the encounter rates (number of on-effort sightings per 100 km) of humpback dolphins in Hong Kong waters significantly decreased, with the breakpoint occurring during the autumn of 2014; (2) Since 2014, Hong Kong waters have experienced more prolonged and frequent MHWs, with a significant reduction of core habitat by 26%; according to Granger causality analysis, changes in sea surface temperature drove shifts in dolphin distribution; (3) Our analysis revealed a co-existence of rapid annual increases in MHWs and high habitat usage, with the marine park located in Southwest Lantau being particularly at risk. This study on dolphin distribution shifts and their relationships with marine heatwaves in Southern China made a contribution to our understanding of the action of marine cetaceans' response to climate change. Additionally, it highlights the importance of considering MHWs in dolphin conservation efforts.
Article
Full-text available
Ecology Letters (2010) 13: 685–694 Successful mitigation of negative effects of global warming will depend on understanding the link between physiological and ecological responses of key species. We show that while metabolic adjustment may assist Australasian kelp beds to persist and maintain abundance in warmer waters, it also reduces the physiological responsiveness of kelps to perturbation, and suppresses canopy recovery from disturbances by reducing the ecological performance of kelp recruits. This provides a warning not to rely solely on inventories of distribution and abundance to evaluate ecosystem function. The erosion of resilience is mediated by a shift in adult-juvenile interactions from competitive under cool to facilitative under warm conditions, supporting the prediction that positive interactions may become increasingly important in a warmer future. Kelp beds may remain intact but with a lower threshold for where additional impacts (e.g., extreme storms or reduced water quality) will lead to persistent loss of habitat and ecological function.
Article
Full-text available
The idea that alternative stable states may exist in communities has been a recurring theme in ecology since the late 1960s, and is now experiencing a resurgence of interest. Since the first papers on the subject appeared, two perspectives have developed to describe how communities shift from one stable state to another. One assumes a constant environment with shifts in variables such as population density, and the other anticipates changes to underlying parameters or environmental “drivers”. We review the theory behind alternative stable states and examine to what extent these perspectives are the same, and in what ways they differ. We discuss the concepts of resilience and hysteresis, and the role of stochasticity within the two formulations. In spite of differences in the two perspectives, the same type of experimental evidence is required to demonstrate the existence of alternative stable states.
Article
Full-text available
Climate and land-use change drive a suite of stressors that shape ecosystems and interact to yield complex ecological responses, i.e. additive, antagonistic and synergistic effects. Currently we know little about the spatial scale relevant for the outcome of such interactions and about effect sizes. This knowledge gap needs to be filled to underpin future land management decisions or climate mitigation interventions, for protecting and restoring freshwater ecosystems. The study combines data across scales from 33 mesocosm experiments with those from 14 river basins and 22 cross-basin studies in Europe producing 174 combinations of paired-stressor effects on a biological response variable. Generalised linear models showed that only one of the two stressors had a significant effect in 39% of the analysed cases, 28% of the paired-stressor combinations resulted in additive and 33% in interactive (antagonistic, synergistic, opposing or reversal) effects. For lakes the frequency of additive and interactive effects was similar for all spatial scales addressed, while for rivers this frequency increased with scale. Nutrient enrichment was the overriding stressor for lakes, generally exceeding those of secondary stressors. For rivers, the effects of nutrient enrichment were dependent on the specific stressor combination and biological response variable. These results vindicate the traditional focus of lake restoration and management on nutrient stress, while highlighting that river management requires more bespoke management solutions.
Article
Full-text available
Catastrophic regime shifts in shallow lakes are hard to predict due to a lack of clear understanding of the associate mechanisms. Theory of alternative stable states suggests that eutrophication has profound negative effects on the structure, function and stability of freshwater ecosystems. However, it is still unclear how eutrophication destabilizes ecosystems stoichiometrically before a tipping point is reached. The stoichiometric homeostasis (H), which links fine-scale process to broad-scale patterns, is a key parameter in ecological stoichiometry. Based on investigation of 97 shallow lakes on the Yangtze Plain, China, we measured nitrogen (N) and phosphorus (P) concentrations of the aboveground tissues of common submerged macrophyte species and their corresponding sediments. We found submerged macrophytes showed significant stoichiometric homeostasis for P (H P) but not for N (H N). Furthermore, H P was positively correlated with dominance and stability at the species level, and community production and stability at the community level. Identifying where macrophyte community collapse is a fundamental way to quantify their resilience. Threshold detection showed that macrophyte community dominated by high-H P species had a higher value of tipping point (0.08 vs. 0.06 mg P L À1 in lake water), indicating their strong resilience to eutrophication. In addition, macrophytes with high H P were predominant in relative oligotrophic sediments and have higher ability in stabilizing the water environment compared to those low-H P ones. Our results suggested that ecosystem dominated by homeostatic macrophyte communities was more productive, stable and resilient to eutrophication. Eutrophication-induced stoichiometric imbalance may destabilize the ecosystem by altering the community structure from high-to low-H P species. Efforts should be focused on maintaining and restoration of high homeo-static communities to make ecosystem more resilient, which can significantly improve our understanding of the critical transition mechanisms.
Article
Full-text available
Natural populations are increasingly threatened with collapse at the hands of anthropogenic effects. Predicting population collapse with the help of generic early warning signals (EWS) may provide a prospective tool for identifying species or populations at highest risk. However, pattern‐to‐process methods such as EWS have a multitude of challenges to overcome to be useful, including the low signal‐to‐noise ratio of ecological systems and the need for high quality time series data. The inclusion of trait dynamics with EWS has been proposed as a more robust tool to predict population collapse. However, the length and resolution of available time series are highly variable from one system to another, especially when generation time is considered. As yet, it remains unknown how this variability with regards to generation time will alter the efficacy of EWS. Here we take both a simulation‐ and experimental‐based approach to assess the impacts of relative time series length and resolution on the forecasting ability of EWS. We show that EWS’ performance decreases with decreasing time‐series length. However, there was no evident decrease in EWS performance as resolution decreased. Our simulations suggest a relative time series length between 10 and five generations as a minimum requirement for accurate forecasting by abundance‐based EWS. However, when trait information is included alongside abundance‐based EWS, we find positive signals at lengths one‐half of what was required without them. We suggest that, in systems where specific traits are known to affect demography, trait data should be monitored and included alongside abundance data to improve forecasting reliability.
Article
Full-text available
Environmental change can impact the stability of ecological systems and cause rapid declines in populations. Abundance‐based early warning signals have been shown to precede such declines, but detection prior to wild population collapses has had limited success, leading to the development of warning signals based on shifts in distribution of fitness‐related traits such as body size. The dynamics of population abundances and traits in response to external environmental perturbations are controlled by a range of underlying factors such as reproductive rate, genetic variation and plasticity. However, it remains unknown how such ecological and evolutionary factors affect the stability landscape of populations and the detectability of abundance and trait‐based early warning signals. Here, we apply a trait‐based demographic approach and investigate both trait and population dynamics in response to gradual and increasing changes in the environment. We explore a range of ecological and evolutionary constraints under which stability of a population may be affected. We show both analytically and with simulations that strength of abundance‐ and trait‐based warning signals are affected by ecological and evolutionary factors. Finally, we show that combining trait‐ and abundance‐based information improves our ability to predict population declines. Our study suggests that the inclusion of trait dynamic information alongside generic warning signals should provide more accurate forecasts of the future state of biological systems.
Article
Cladocerans (Crustacea: Branchiopoda) are an essential component of the food webs of shallow lakes. However, their potential significance as sentinels of environmental change in subtropical regions of South-East Asia, is not well elucidated. In this study, we examined the distribution pattern of cladoceran subfossils in the surface sediments of 64 shallow lakes located in the middle and lower reaches of the Yangtze River Basin (MLYB). Using constrained ordination and multivariate regression trees (MRT), we explored the relationships between the distribution of cladocerans and contemporary environmental variables. The results indicated that macrophyte abundance is the most important regulator, followed by chlorophyll a and total phosphorus that all were important drivers of cladoceran community dynamics. Despite intensive aquaculture in the MLYB lakes over the past decades, fish predation did not have a significant influence on the cladoceran community, likely because the fish predation pressure was generally high. The MRT analysis showed distinct variations in the composition of cladoceran communities and was able to identify the key indicator species of various stressors, including Bosmina coregoni, small Alona (A. intermedia, A. rectangula) and Leydigia acanthocercoides, Leydigia leydigi and Chydorus sphaericus. Boosted Regression Tree (BRT), a novel numerical model, was employed to establish the quantitative relationship between cladoceran composition and environmental variables. The model was used to reconstruct historical macrophyte coverage over the past 60 years in a typical Yangtze lake, Liangzi Lake, based on sedimentary cladoceran records. The BRT-inferred macrophyte coverage was validated well by the macrofossil records in the same lake. Using subfossil cladocerans as indicator of the past environmental change in MLYB lakes is potentially useful given their key ecological role. However, caution should be taken due to species-specific differences in preservation in the sediment and confounding environmental processes that to some extent may bias the conclusions.