Phytoplankton functional group dynamics ... - Wiley Online Library

Ecology, 95(12), 2014, pp. 3335–3343 Ó 2014 by the Ecological Society of America

Phytoplankton functional group dynamics explain species abundance distribution in a directionally changing environment CHENG-HAN TSAI,1 TAKESHI MIKI,1,2 CHUN-WEI CHANG,1,2 KANAKO ISHIKAWA,3 SATOSHI ICHISE,3 MICHIO KUMAGAI,3,4 1,5,6 AND CHIH-HAO HSIEH 1 Institute of Oceanography, National Taiwan University, Taipei 10617 Taiwan Research Center for Environmental Changes, Academia Sinica, Taipei 10617 Taiwan 3 Lake Biwa Environmental Research Institute, Otsu 520 0022 Japan 4 Lake Biwa Sigma Research Center, Ritsumeikan University, Kusatsu 525 0058 Japan 5 Institute of Ecology and Evolutionary Biology, National Taiwan University, Taipei 10617 Taiwan 2

Abstract. The mechanism underlying species abundance distribution (SAD), particularly the characteristics of ‘‘excess of rare species,’’ remains controversial. The current equilibrium theory cannot explain the transient dynamics of SAD, which is essential for predicting biodiversity response to environmental changes. Using a unique 32-yr-long phytoplankton community data set from a pelagic site of Lake Biwa, Japan, we show that the dynamics of functional groups driven by environmental variation explain the excess of rare species over time. First, most of the rare species belong to the littoral group supplied through dispersal, whereas the common species belong to the pelagic group. Second, the littoral group was negatively influenced by environmental changes (i.e., lake warming, water-level manipulation, and partial re-oligotrophication), mechanistically explaining truncation of the excess of rare species in the SAD associated with biodiversity loss in Lake Biwa. Our findings imply the significance of an ecological trait-based approach for SAD theory and managing biodiversity in a changing environment. Key words: functional response group; Lake Biwa; lake environmental changes; phytoplankton biodiversity; reservoir management; species abundance distribution.

INTRODUCTION The commonness and rarity of species represent a fundamental feature of community structure, which may emerge from resource partitioning among species (MacArthur 1957, Sugihara 1989, Sugihara et al. 2003, Connolly et al. 2005, McGill et al. 2007). Species abundance distribution (SAD) is a concise representation of the commonness and rarity of species and has been extensively studied as a means of understanding mechanisms structuring communities (Preston 1948, May 1975, Magurran 2004). In an effort to elucidate the patterns and mechanisms underlying SAD, recent studies have focused on the SAD of species-rich communities, for example, tropical trees (Volkov et al. 2003), reef corals and fishes (Dornelas et al. 2006), and soil microbes (Dumbrell et al. 2010). In addition, numerous theoretical models have been proposed to explain SAD through niche partitioning (MacArthur 1957, Sugihara 1980), demographic stochasticity (Volkov et al. 2003), statistical effects (May 1975, McGill 2003, Sizling et al. 2009), and reconciliation of these factors (Engen and Lande 1996, Chisholm and Pacala 2010, Vergnon et al. 2012). More recently, some studies Manuscript received 16 October 2013; revised 5 May 2014; accepted 21 May 2014. Corresponding Editor: P. R. Leavitt. 6 Corresponding author. E-mail: [email protected]

argued that SAD is not ecologically informative because species interactions indirectly influence SAD through regulating total species richness and abundance (Supp et al. 2012, Locey and White 2013). However, the majority of hypotheses have focused on static patterns, and mechanisms of temporal evolution of SAD have remained largely unknown. Indeed, recent reviews note that exploring temporal variation of SAD provides an additional empirical basis for testing competing SAD hypotheses, as well as for formulating new hypotheses concerned with community assembly rules (Magurran 2007, McGill et al. 2007). More importantly, understanding temporal variation of SAD has practical implications for the detection and management of biodiversity loss in response to environmental changes such as habitat degradation, pollution, and warming (Gray 1981, Magurran et al. 2010). A number of challenges must be addressed to better understand the dynamics of SAD under environmental change. Firstly, the recent resurgence of interest in the ‘‘excess of rare species’’ (i.e., many rare species and a few dominant species) of SAD has stimulated debate on the underlying mechanisms of rarity and SAD (Magurran and Henderson 2003, Volkov et al. 2003, McGill et al. 2007, Sizling et al. 2009, Chisholm and Pacala 2010). Nonetheless, the extent to which these mechanisms can explain temporal changes in rarity and SAD remains unclear. Moreover, the response of excess of rare species

3335

3336

CHENG-HAN TSAI ET AL.

and SAD to environmental change is less predictable; for example, the proportion of rare species in a community may be influenced by factors such as dispersal (Palardy and Witman 2011), productivity (Gray 1981, Chase 2010), and predation (Schoener and Spiller 1996, Magurran and Henderson 2012). Secondly, the uncertainty of predicting rarity also underscores the importance of characterizing species into functional response groups, since species from different functional groups can respond differently to the same driver (Gray 1989, Hooper et al. 2005). However, there is no current consensus as to how and to what extent the dynamics of functional group responses to environmental change can explain the temporal evolution of rarity and SAD. Thirdly, even though functional groups may play key roles in structuring rarity and SAD, targeting the essential functional traits (or niche dimensions) for coexisting species is empirically difficult (McGill et al. 2007, Litchman and Klausmeier 2008). A potential traitbased approach for explaining SAD is to decompose a community into transient vs. resident species groups (Magurran and Henderson 2003, Magurran 2007). This idea originates from the common observation that ecological communities are often influenced by recruitment processes such as species dispersal and migration. In applying this idea, previous studies have demonstrated that species commonness and rarity could be explained by representing rare and common species with transient and resident species/groups (Magurran and Henderson 2003, Ulrich and Zalewski 2006, Dolan et al. 2009, Coyle et al. 2013), respectively. For instance, Magurran and Henderson (2003) have argued that the dynamics of transient vs. resident species may represent two distinct (i.e., stochastic vs. deterministic) community assembly processes in shaping SAD. This implies different sensitivities of functional groups to external fluctuations. Finally, in order to elucidate the environmental drivers of rarity and SAD through time (Magurran 2007), the aforementioned questions emphasize the need for field data collected from long-term, methodologically consistent monitoring of ecological communities and their abiotic environments. Here, we analyzed a unique long-term (32 years of semimonthly sampling; 1978–2009) phytoplankton community and environmental data set collected from the pelagic zone of the largest lake in Japan, Lake Biwa. One of the advantages of this time series was that the sampling procedure and taxonomic resolution of the phytoplankton data set was very consistent over the entire study period. Such a long-term, highly resolved data set is suitable for the analysis of rarity, SAD, and species diversity through time. Additionally, over the course of the past few decades, Lake Biwa has experienced dramatic environmental changes, such as eutrophication, re-oligotrophication, water-level manipulation, and lake warming (Nakanishi and Sekino 1996, Hsieh et al. 2010, Hsieh et al. 2011), providing us with an

Ecology, Vol. 95, No. 12

opportunity to investigate long-term environmental effects. Phytoplankton communities play a major role in maintaining biodiversity and ecosystem functions in aquatic ecosystems (Reynolds 2006, Cardinale 2011). Thus, understanding their community structure and dynamics in a context of environmental change is relevant for predicting and managing the consequences of biodiversity loss of aquatic producers, particularly for those human-disturbed watersheds such as Lake Biwa. Here, we investigated SAD through categorization of trait-based functional groups, and our aim was to demonstrate that functional differences between two groups (i.e., littoral group representing transient species vs. pelagic group representing resident species) could explain the commonness and rarity of species in the phytoplankton community. More importantly, we focused on identifying environmental factors which determine the appearance/disappearance of the littoral group through time, and elucidating how this mechanism in turn drives the temporal dynamics of rarity, SAD, and species diversity. MATERIALS

AND

METHODS

Phytoplankton community and environmental data The phytoplankton community data set (a total of 190 species) includes the semimonthly time series from January 1978 to December 2009 collected by the Lake Biwa Environmental Research Institute in a pelagic site of Lake Biwa (Appendix A: Figs. A1–2). This is a unique data set with defined and consistently applied sampling protocols and species identification and enumeration methods (see Appendix A; Hsieh et al. 2010). Importantly, enumeration and identification of phytoplankton species were performed by a single expert (S. Ichise) over the entire 32 years; therefore, the possibility of sampling bias that may exclusively affect temporal variation of the SAD is precluded. The environmental data set includes meteorological and limnological measurements from different institutions spanning the same time period (Appendix A). Specific details for sampling locations and protocols regarding the data sets are presented in Appendix A. Time series and functional data analyses We investigated the temporal variation of the phytoplankton SAD in Lake Biwa. In this study, we used the number of individuals as the index of abundance for each phytoplankton species to analyze SAD (Appendix A). We used the cumulative sum of semimonthly data of species within a given year to show the interannual variation of the phytoplankton SAD, as well as to remove the strong seasonality of the phytoplankton community (Appendix B). We constructed the temporal SAD diagram following the classic Preston approach (Preston 1948, Magurran 2004). We found dramatic temporal variation of SAD; in particular, the left-hand tail was cut after 1986 (see Results).

December 2014

SPECIES RARITY IN CHANGING ENVIRONMENTS

To examine whether the truncated left-hand tail of the SAD was caused by sampling artifact (e.g., unveiling effect), we first analyzed the correlation between total abundance and species richness time series at an interannual scale. We followed the rationale that sampling artifact could be ruled out in the absence of a positive correlation of time series between total abundance (proxy for sampling intensity) and species richness. In addition, we considered that unveiling effect could be ruled out if a truncated-left hand tail of SAD could not be reproduced by a random sampling of phytoplankton community matrices through time. Therefore, we constructed a null model for the temporal SAD using a community resampling procedure (Ulrich and Gotelli 2010). Specifically, we randomly drew phytoplankton individuals from the metacommunity (in which the relative species abundance is calculated by the cumulated species abundances for all sampling years) to generate a null model of community matrix. Resampling procedure and sampling probability were constrained by the total abundance of phytoplankton in the given year and the relative species abundance of metacommunity (here, defined as the cumulated species abundances for all sampling years). See Appendix C for a detailed explanation of the null model. We carried out the simulation 100 times and compared the null model results to the observations for temporal SAD and total species richness time series. Because the metacommunity for generating a null model of SAD comes from cumulative species abundances over all years, the richness of the SAD generated from the null model for each year is on average higher than the richness of SAD of the observation. We standardized the number of species in octaves by their mean and standard deviation in a given year (i.e., standardized richness) for generating the diagram illustrating temporal SAD, so that interannual changes in the left-hand tail of SAD derived from empirical data vs. null model could be compared. Note that this null model is used to evaluate pure sampling artifact and does not constrain total richness of a given year. In addition to considering sampling (unveiling) effects, one may ask whether the observed SAD for a particular year can simply result from simultaneous statistical constraints of total species richness and total abundance (i.e., statistical artifact) without invoking any ecology (Locey and White 2013). To address this issue, we performed two additional null model analyses (Appendix C). First, we cumulated the SAD at a decadal scale (i.e., 10-yr window) and used these decadal SADs to generate null communities following the aforementioned resampling procedure (Ulrich and Gotelli 2010). This simulation strategy produces similar total species richness and constrained total abundance as the observed data. Second, we performed the feasible set analyses for all possible combinations of SAD, given the constraints of species richness and total abundance

3337

of the community (Locey and White 2013). Specifically, we conducted 1000 instances of homogeneous sampling for all combinations of SAD, given the species richness and total abundance for a given year. Here, we followed the procedure of Locey and White (2013), but used improved algorithms with high sampling efficiency (Appendix C). We also followed the procedure of Locey and White (2013) to compare the null SADs with the observed SAD. Then, to explore possible causes for the temporal variation of SAD, we investigated the dynamics of functional groups of phytoplankton. We focused on the hypothesis that functional group dynamics, particularly changes in ratio of transient and core species, drive temporal variation of SAD (Appendix D). Accordingly, we categorized all recorded phytoplankton species into littoral and pelagic functional groups based on Reynolds functional groups (Appendix D: Table D1; Reynolds et al. 2002, Reynolds 2006). We used littoral species richness to total species richness ratio (hereafter, littoral ratio) time series at an inter-year scale as an index of temporal variation of SAD, because we found that the littoral species were low in abundance (i.e., rare species) and contributed mostly to the left-hand tail of SAD (see Results). The littoral ratio would be more informative and concise than other indices because it directly indicated the proportion of the littoral functional group (transient group) in a pelagic phytoplankton community. Finally, we linked the temporal variation of SAD with meteorological and limnological variables, using timeseries correlation and linear-model analyses of the littoral ratio responses to environmental variables. The environmental variables included surface water temperature, buoyancy frequency, wind mixing index, total phosphorous, water level, precipitation, river inflow, Secchi depth, zooplankton abundance, and Arctic Oscillation index (Appendix E: Fig. E1). Both interyear and intra-year variation of environmental variables were investigated to account for different effects of temporal variability at multiple time scales. River inflow and zooplankton abundance cannot be resolved at an intra-year scale and were analyzed only at an inter-year scale (Appendix E). To consider potential interactive effects of environmental variables on the littoral ratio, we included pairwise interaction terms of environmental variables in multivariate linear models (Appendix F). Higher-order terms were not considered due to limitation of sample size. All of the time series were normalized by mean and standard deviation prior to correlation and linear-model analyses. Aikake’s information criteria corrected for small sample size (AICc) were used for model selection and inferences of multiple models (Appendix F). In order to account for serial dependence in time-series data, the stationary bootstrap with accelerated bias corrections was used to compute confidence limits and to perform hypothesis tests in correlation and linear-model analyses (Efron and

3338



group dynamics have significantly contributed to temporal variation of total species richness. Moreover, since the littoral species represent most of the rare species in the community, the dynamics of the littoral group can in turn drive rarity and SAD through time. Dynamics of littoral group resulted in truncated rarity of SAD

FIG. 1. Species abundance distribution of the phytoplankton community cumulated from 1978 to 2009. In total, 190 species were recorded. White and black bars indicate the number of littoral (N ¼ 40) and pelagic (N ¼ 150) species in a given numerical abundance class (Preston’s octave; log2transformed numerical abundance), respectively.

Tibshirani 1993, Politis and Romano 1994). All analyses were performed in the R 2.12.2 statistical software (R Development Core Team 2011) using boot, AICcmodavg packages, and codes modified from references (Politis and White 2004, Ulrich and Gotelli 2010). RESULTS Littoral group consisted of rare species

We observed truncated rare species (i.e., the left-hand tail is cut) of the SAD in the late 1980s at an interannual time scale (Fig. 3a, c), which was also coupled with the temporal variation of the littoral group. The observed truncation is robust even when SAD is examined on a seasonal basis at an interannual time scale (Appendix B: Fig. B1). One may argue that such a truncation of SAD could occur with changes in sampling effort; that is, fewer singletons with a greater intensity of sampling effort (commonly known as the unveiling effect; Preston 1948, Nee et al. 1991, McGill et al. 2007). In order to confirm that the observed truncation of rare species of SAD was not due to the unveiling effect, we performed two tests. First, we assessed whether or not a random sampling of individuals of species can produce a truncation of SAD. The null model built from an average of 100 resamplings of the community (see Materials and methods) showed no truncation of SAD through time (Fig. 3b, d). That is, the simulated sampling effect always produces rare species (i.e., the left-hand tail) in SAD (Fig. 3b), which does not agree with our empirical observation of truncation in late 1980s (Fig. 3a). Second, we wanted to know whether species richness increases with the total phytoplankton abundance (a proxy for sampling effort) through time. We found that the time series of species richness was not positively correlated, but negatively correlated with the

Through examining the community structure from SAD over 32 years with respect to the functional grouping of phytoplankton, we found that most of the littoral species were rare species and contributed mostly to the left-hand tail of the SAD (Fig. 1). The finding that littoral species (littoral species in Appendix D: Table D1; N ¼ 40) were also rare taxa is insensitive to the definition of rare species (Table D2; see Appendix D for sensitivity tests). In contrast, the pelagic species (pelagic species in Appendix D: Table D1; N ¼ 150) were considerably more abundant than the littoral species and constituted most of the right-hand part of the SAD (Fig. 1). Temporal variation of the littoral group We found that the littoral group was not temporally persistent in its presence. We used the littoral ratio to represent the littoral group dynamics through the past three decades (Fig. 2). Many of the littoral species in Lake Biwa disappeared in late 1980s, and the time-series analysis showed that the littoral ratio was positively correlated with the total species richness (r ¼ 0.64, P , 0.01) at an interannual scale (Fig. 2). Thus, the littoral

FIG. 2. Interannual variation of the littoral ratio (littoral species richness to total species richness ratio) and species richness in the phytoplankton community. The solid and dashed lines indicate the littoral ratio and species richness in phytoplankton community, respectively. Positive correlation between the two time series is significant (r ¼ 0.64, P , 0.01) after accounting for the serial dependence.

December 2014


3339

FIG. 3. Interannual and decadal variation of the species abundance distribution (SAD). Interannual-scale SAD of the phytoplankton community is shown (a) from the field data, and (b) from the average of 100 resamplings of the field data. The horizontal axis indicates the interannual scale, and the vertical axis indicates the abundance classes using the Preston octave (log2transformed numerical abundance). Colors indicate the standardized species richness in octaves for a given year. A truncated lefthand tail of SAD can be found after 1985 in panel a relative to the persistent rare species in panel (b). Decadal-scale SAD of the phytoplankton community is shown (c) from the field data, and (d) from the resampling of the field data. Panels of (c) and (d) from top to bottom represent the time series of SAD on a decadal scale. Log-transformed total abundance (log(A)) of the phytoplankton community and the corresponding time interval are denoted for each panel.

total abundance at an interannual scale (r ¼ 0.61, P , 0.01). In contrast, under the random resampling null model, the time series of species richness is positively correlated with the total abundance (r ¼ 0.88, P , 0.01), in a manner consistent with the unveiling effect. We conclude that the truncated left-hand tail of SAD is largely driven by littoral species dynamics and not because of the unveiling effect. Furthermore, we evaluated whether the observed SADs could simply result from simultaneous statistical constraints of species richness and total abundance (Locey and White 2013).

When constraining both species richness and total abundance, we found that the null models cannot reproduce the observed SAD (Appendix C); therefore, the observed SAD cannot be explained by statistical artifact. Environmental drivers for the temporal variation of littoral group and SAD We found that the dynamics of the littoral group played an important role in affecting the temporal evolution of SAD, particularly the left-hand tail. A

3340


critical question is which environmental factors determine littoral group dynamics in Lake Biwa. We focused on those mechanisms that were most likely related to the history of environmental change in Lake Biwa since 1978. Among these mechanisms were (1) changes in precipitation, water level, and river inflow, (2) increased water-column temperature and the abrupt change of the Arctic Oscillation index in late 1980s, (3) re-oligotriphication, and (4) zooplankton predation. We tested four hypotheses addressing each of the four mechanisms potentially contributing to temporal variation of the littoral ratio in phytoplankton community. Hypothesis 1: immigration.—Since littoral species typically inhabit streams or the littoral zone, we considered the role of immigration as a key process influencing the composition of the pelagic phytoplankton community. We hypothesized that the cause of truncation of the left-hand tail of SAD (or reduced littoral ratio) is the reduced availability and/or immigration rate of littoral species. Thus, we analyzed waterlevel fluctuations in order to examine the availability of the littoral zone, since many of the littoral species in our study are either attached to littoral macrophytes or live in shallow habitats, which are both sensitive to waterlevel variations. That is, when the temporal variability of water level is greater, the habitat (transition zone in the coastal area of the watershed) for the littoral species is likely larger, and, thus, the reservoir of littoral species also becomes larger (e.g., Van Geest et al. 2007, Zohary and Ostrovsky 2011). In addition, we analyzed precipitation, river inflow, and wind strength as factors likely determining the immigration rate of littoral species through circulation (Akitomo et al. 2009, Palardy and Witman 2011). In Lake Biwa, we found that the proportion of littoral species in the phytoplankton community is positively correlated with intra-year variation of water level (r ¼ 0.393, P , 0.05; Table 1). In accordance, results of multivariate linear model analyses showed that intra-year variation of water level best explains the littoral ratio (Appendix F: Table F1). However, littoral ratio is not significantly related to precipitation, river inflow, or wind mixing (Table 1). Hypothesis 2: water warming.—We tested the hypothesis that water warming drives local extinction of littoral species through (1) differential physiological tolerance among functional groups (Huertas et al. 2011) and (2) changes in interspecific competition caused by altered water-column mixing regimes (Daufresne et al. 2009, Winder et al. 2009). We found a negative correlation between the littoral ratio and surface water temperature at an inter-year scale (r ¼0.298, P , 0.01; Table 1). In keeping with previous demonstrations of the influence of the Arctic Oscillation on inter-year variation of water temperature in Lake Biwa (Hsieh et al. 2010), we also found a negative correlation between the littoral ratio and the Arctic Oscillation index (r ¼ 0.295, P , 0.05; Table 1). The results of multivariate linear modeling indicate that both surface water temperature and the


Arctic Oscillation index significantly explained the littoral ratio; however, the Arctic Oscillation index alone has the lower AICc (Appendix F: Table F1). Hypothesis 3: productivity.—Higher productivity can sustain more species in stochastic environments (e.g., Chase 2010). As such, sufficient nutrients help enhance the survival of the rare dispersal-mediated species. We hypothesized that the truncated left-hand tail of SAD arises in Lake Biwa when rare species (littoral species) are competitively excluded under low-nutrient conditions. To test this hypothesis, we used total phosphorous and Secchi depth as proxies for nutrient availability and lake trophic condition. A marginally significant positive correlation between the littoral ratio and total phosphorous (r ¼ 0.312, P ¼ 0.075; Table 1) was found, partially supporting this hypothesis. Linear-model analyses showed that intra-year variation of Secchi depth and inter-year variation of total phosphorous likely explained the littoral ratio (Appendix F: Table F1). Hypothesis 4: predation.—It has been shown that predators can promote or suppress the persistence of rare species (e.g., Caswell 1978, Schoener and Spiller 1996). We tested the hypothesis that the truncation of the left-hand tail of SAD (reduced littoral ratio) was due to changes in predation pressure. Here, we used total zooplankton abundance as a proxy for predation pressure on the phytoplankton community (Appendix A), and found that the correlation was not significant (Table 1). DISCUSSION Here, we empirically demonstrate that functional traits can explain the commonness and rarity of species. In this study, common and rare species belong to the pelagic and littoral phytoplankton groups, respectively. Common species are related to the pelagic-adapted functional traits, whereas rare species are related to littoral traits. Accordingly, we propose that this functional separation of SAD is due to the effects of two community structuring factors, (1) competition, where littoral species are less adapted to pelagic environments and therefore tend to be rare when compared to resident pelagic species, and (2) immigration, where littoral species are sensitive to dispersal or environmental stochasticity. Our findings support the hypothesis that functional traits associated with habitat use (i.e., transient vs. resident species) can explain the excess of rare species in SAD (Magurran and Henderson 2003; Fig.1). Moreover, we show that changes in the proportion of functional groups in community (i.e., littoral ratio) explains the temporal truncation of rare species of the SAD (Figs. 2 and 3), but not the unveiling effect (Preston 1948, Nee et al. 1991, McGill et al. 2007), or statistical artifacts constrained by species richness and total abundance (Locey and White 2013). This counterintuitive result demonstrates the strength and importance of using functional-based mechanisms rather than

December 2014


purely statistical mechanisms as an alternative hypothesis for explaining SAD (Magurran and Henderson 2003, McGill 2003, McGill et al. 2007). Distinguishing functional groups from statistical artifacts in SAD patterning is especially striking given the current knowledge gap in understanding the temporal variation of SAD (Magurran 2007). Importantly, the truncation of the left-hand tail of SAD is driven by functional groups in response to environmental changes. Firstly, our analysis suggests that variation in the availability of littoral species from the littoral zone (evaluated by intra-year variation of water level) explains the temporal variation of the littoral ratio (i.e., immigration hypothesis). Further analyses indicated that the intra-year variation of water level was not driven by the variation of precipitation (Appendix E: Table E2). Rather, reduced intra-year variation of water level in Lake Biwa might be attributable to the management policy that aims to stabilize the water level (Appendix A). Further examination of species-specific habitat preference for the littoral group in Lake Biwa indicated that all types of coastal habitat (e.g., macrophyte zone and rocky shore) may be degraded under anthropogenic stresses (Appendix G). Accordingly, we conclude that this anthropogenic factor (i.e., managed water level) most likely contributed to the immigration processes of littoral species in Lake Biwa, and thus the changes in rarity of SAD. Secondly, the climate-driven lake warming was also found as a key driver (i.e., water warming hypothesis), indicating that warming generates unfavorable conditions for littoral species. One potential mechanism may be that warming favors small-sized phytoplankton (Daufresne et al. 2009, Winder et al. 2009). However, in spite of slightly larger cell sizes for littoral vs. pelagic groups (Appendix H), our analysis demonstrates no significant difference in the mean size of littoral and pelagic groups (Appendix H: Fig. H1). Moreover, although the mean size of both littoral and pelagic groups decreased with time, the rate of change of cell size was found to be similar (Appendix H: Fig. H2). The correlations of littoral ratio against buoyancy frequency (directly related to sinking velocity of phytoplankton) and cubic-transformed wind strength (proxy for water mixing intensity) are not significant (Table 1). Accordingly, changes in community size structure alone cannot explain the truncated left-hand tail of SAD. Rather, we suggest that physiological constraints, as opposed to size-dependent mechanisms, have driven the dynamics of littoral species in Lake Biwa, and thus the changes in rarity of SAD. Finally, the productivity hypothesis cannot be ruled out. Anthropogenic re-oligotrophication (decreased nutrient supply) of the lake partly explains the reduced proportion of littoral species. Note that the aforementioned hypotheses may not be independent in determining the temporal variation of littoral ratio. In fact,

3341

TABLE 1. Results of correlation analyses for the proportion of littoral species in the phytoplankton community vs. environmental variables.

Environmental variables

Correlation coefficient

P

Inter-year variation Arctic Oscillation index Buoyancy frequency Lake surface temperature Total phosphorous Precipitation River inflow Wind mixing Water level Secchi depth Zooplankton abundance

0.295 0.119 0.298 0.312 0.083 0.012 0.089 0.074 0.192 0.183

,0.05 0.355 ,0.01 0.075 0.338 0.489 0.377 0.395 0.483 0.126

Intra-year variation Buoyancy frequency Lake surface temperature Total phosphorous Precipitation Wind mixing Water level Secchi depth

0.087 0.057 0.281 0.051 0.021 0.393 0.036

0.364 0.379 0.446 0.165 0.427 ,0.05 0.507

Notes: The time series of annual mean value was used to represent inter-year variation; the time series of coefficient of variation within a given year was used to represent intra-year variation. Significant correlations were set at P , 0.05, after accounting for the temporal autocorrelation.

effects of water temperature and productivity may interact to decrease the littoral ratio in Lake Biwa (Appendix F). Overall, our analyses suggest that a reduced proportion of littoral species and, therefore, the truncated lefthand tail of SAD for the phytoplankton community could be mainly driven by natural (i.e., climatic-driven lake warming) and anthropogenic (i.e., managed water level and, partly, re-oligotrophication) environmental variations in Lake Biwa. This implies that the wellknown excess of rare species of SAD can be very dynamic and prone to variable external forcing, which emphasizes the need for considering transient states in addition to existing equilibrium theories of SAD (Harte 2003, Sugihara et al. 2003, Volkov et al. 2003, Magurran 2004, Sizling et al. 2009). Hence, we conclude that the excess of rare species and temporal evolution of SAD can be produced by ecological mechanisms and not as byproducts of sampling or statistical artifacts. This distinction emerges only when the effects of the temporal dimension and functional group responses to environmental change are considered. Our finding that immigration processes drive the changes in excess of rare species of SAD may have implications for recent theories attempting to reconcile dispersal- and niche-assembly perspectives of community structuring (e.g., Chisholm and Pacala 2010, Vergnon et al. 2012). While our findings suggest that the dispersal-assembly process (i.e., immigration) can determine the temporal evolution of the left-hand tail of SAD, our results do not necessarily lead to the

3342


conclusion that dispersal assembly is more important than niche assembly for understanding community structure and SAD (Sugihara et al. 2003, Volkov et al. 2003). In fact, the left-hand tail of SAD can be indicative of the dispersal-driven immigrant species, while the right-hand part of SAD can indicate the niche ordering of resident species (Magurran and Henderson 2003). As such, although our findings may suggest the separated effects of dispersal- and niche-assembly processes on SAD, it is possible that the interplay between species immigration and local species interactions (i.e., the relative strength of dispersal- and niche-assembly processes) synergistically determines the changes in functional groups and the shape of SAD. We propose that the changes in rarity of SAD through time (i.e., the cut of the left-hand tail of the SAD) can represent a biodiversity indicator for the impaired community structure resulting from the loss of environmentally vulnerable functional groups. In our case, rare species belonging to the littoral functional group are largely driven out by environmental changes in the lake ecosystem, and, particularly, disconnected metacommunity in littoral habitats. Thus, the truncated SAD through time could represent a signal for deteriorated habitats or recruitment barriers. In addition, the truncated SAD has practical implications for management. Watershed management often attempts to stabilize (or minimize) the variation of water level (Wantzen et al. 2008). In contrast, our findings imply that mimicking the natural variability of water level aids maintenance of phytoplankton biodiversity, and therefore buffers against negative impacts of water warming. Importantly, we found that interactions between climatic and anthropogenic factors (e.g., temperature vs. trophic status, or water level vs. trophic status; Appendix F) can affect the functional group dynamics of phytoplankton and in turn reorganize the community structure and SAD. As future environmental changes from climatic and anthropogenic mechanisms may interact in complex manners (Savage et al. 2010), the truncated SAD calls for more precautionary mitigation strategies to manage biodiversity loss of aquatic producers and increase ecosystem resilience. ACKNOWLEDGMENTS This study was supported by the National Taiwan University and Ministry of Science and Technology of Taiwan (101-2621B-002-004-MY3). Comments from Akash Sastri, Tzung-Su Ding, Wei-Ting Lin, and George Sugihara improved the manuscript. Algorithms developed by Meng-Tsung Tsai greatly improved the simulation efficiency in null model test. This is a contribution of Bio-Asia FASCICLE. LITERATURE CITED Akitomo, K., K. Tanaka, and M. Kumagai. 2009. Annual cycle of circulations in Lake Biwa, part 2: mechanisms. Limnology 10:119–129. Cardinale, B. J. 2011. Biodiversity improves water quality through niche partitioning. Nature 472:86–113. Caswell, H. 1978. Predator-mediated coexistence: non-equilibrium model. American Naturalist 112:127–154.


Chase, J. M. 2010. Stochastic community assembly causes higher biodiversity in more productive environments. Science 328:1388–1391. Chisholm, R. A., and S. W. Pacala. 2010. Niche and neutral models predict asymptotically equivalent species abundance distributions in high-diversity ecological communities. Proceedings of the National Academy of Sciences USA 107: 15821–15825. Connolly, S. R., T. P. Hughes, D. R. Bellwood, and R. H. Karlson. 2005. Community structure of corals and reef fishes at multiple scales. Science 309:1363–1365. Coyle, J. R., A. H. Hurlbert, and E. P. White. 2013. Opposing mechanisms drive richness patterns of core and transient bird species. American Naturalist 181:E83. Daufresne, M., K. Lengfellner, and U. Sommer. 2009. Global warming benefits the small in aquatic ecosystems. Proceedings of the National Academy of Sciences USA 106:12788– 12793. Dolan, J. R., M. E. Ritchie, A. Tunin-Ley, and M. D. Pizay. 2009. Dynamics of core and occasional species in the marine plankton: tintinnid ciliates in the north-west Mediterranean Sea. Journal of Biogeography 36:887–895. Dornelas, M., S. R. Connolly, and T. P. Hughes. 2006. Coral reef diversity refutes the neutral theory of biodiversity. Nature 440:80–82. Dumbrell, A. J., M. Nelson, T. Helgason, C. Dytham, and A. H. Fitter. 2010. Relative roles of niche and neutral processes in structuring a soil microbial community. ISME Journal 4:337–345. Efron, B., and R. Tibshirani. 1993. An introduction to the bootstrap. Chapman and Hall, New York, New York, USA. Engen, S., and R. Lande. 1996. Population dynamic models generating the lognormal species abundance distribution. Mathematical Biosciences 132:169–183. Gray, J. S. 1981. Detecting pollution induced changes in communities using the log-normal distribution of individuals among species. Marine Pollution Bulletin 12:173–176. Gray, J. S. 1989. Effects of environmental stress on species rich assemblages. Biological Journal of the Linnean Society 37: 19–32. Harte, J. 2003. Tail of death and resurrection. Nature 424: 1006–1007. Hooper, D. U., et al. 2005. Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecological Monographs 75:3–35. Hsieh, C. H., et al. 2010. Phytoplankton community reorganization driven by eutrophication and warming in Lake Biwa. Aquatic Sciences 72:467–483. Hsieh, C. H., Y. Sakai, S. Ban, K. Ishikawa, T. Ishikawa, S. Ichise, N. Yamamura, and M. Kumagai. 2011. Eutrophication and warming effects on long-term variation of zooplankton in Lake Biwa. Biogeosciences 8:1383–1399. Huertas, I., M. Rouco, V. Lopez-Rodas, and E. Costas. 2011. Warming will affect phytoplankton differently: evidence through a mechanistic approach. Proceedings of the Royal Society B 278:3534–3543. Litchman, E., and C. A. Klausmeier. 2008. Trait-based community ecology of phytoplankton. Annual Review of Ecology, Evolution, and Systematics 39:615–639. Locey, K. J., and E. P. White. 2013. How species richness and total abundance constrain the distribution of abundance. Ecology Letters 16:1177–1185. MacArthur, R. H. 1957. On the relative abundance of bird species. Proceedings of the National Academy of Sciences USA 43:293–295. Magurran, A. E. 2004. Measuring biological diversity. Blackwell Publishing, Oxford, UK. Magurran, A. E. 2007. Species abundance distributions over time. Ecology Letters 10:347–354. Magurran, A. E., et al. 2010. Long-term datasets in biodiversity research and monitoring: assessing change in ecological

December 2014


communities through time. Trends in Ecology and Evolution 25:574–582. Magurran, A. E., and P. A. Henderson. 2003. Explaining the excess of rare species in natural species abundance distributions. Nature 422:714–716. Magurran, A. E., and P. A. Henderson. 2012. How selection structures species abundance distributions. Proceedings of the Royal Society B 279:3722–3726. May, R. M. 1975. Patterns of species abundance and diversity. Pages 81–120 in M. L. Cody and J. M. Diamond, editors. Ecology and evolution of communities. Belknap/Harvard University Press, Cambridge, Massachusetts, USA. McGill, B. J. 2003. Does Mother Nature really prefer rare species or are log-left-skewed SADs a sampling artefact? Ecology Letters 6:766–773. McGill, B. J., et al. 2007. Species abundance distributions: moving beyond single prediction theories to integration within an ecological framework. Ecology Letters 10:995– 1015. Nakanishi, M., and T. Sekino. 1996. Recent drastic changes in Lake Biwa bio-communities, with special attention to exploitation of the littoral zone. GeoJournal 40:63–67. Nee, S., P. H. Harvey, and R. M. May. 1991. Lifting the veil on abundance patterns. Proceedings of the Royal Society B 243: 161–163. Palardy, J. E., and J. D. Witman. 2011. Water flow drives biodiversity by mediating rarity in marine benthic communities. Ecology Letters 14:63–68. Politis, D. N., and J. P. Romano. 1994. The stationary bootstrap. Journal of the American Statistical Association 89:1303–1313. Politis, D. N., and H. White. 2004. Automatic block-length selection for the dependent bootstrap. Econometric Reviews 23:53–70. Preston, F. W. 1948. The commonness and rarity of species. Ecology 29:254–283. R Development Core Team. 2011. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. www.r-project.org Reynolds, C. S. 2006. The ecology of phytoplankton. Cambridge University Press, Cambridge, UK. Reynolds, C. S., V. Huszar, C. Kruk, L. Naselli-Flores, and S. Melo. 2002. Towards a functional classification of the freshwater phytoplankton. Journal of Plankton Research 24:417–428. Savage, C., P. R. Leavitt, and R. Elmgren. 2010. Effects of land use, urbanization, and climate variability on coastal eutrophication in the Baltic Sea. Limnology and Oceanography 55:1033–1046.

3343

Schoener, T. W., and D. A. Spiller. 1996. Devastation of prey diversity by experimentally introduced predators in the field. Nature 381:691–694. Sizling, A. L., D. Storch, E. Sizlingova, J. Reif, and K. J. Gaston. 2009. Species abundance distribution results from a spatial analogy of central limit theorem. Proceedings of the National Academy of Sciences USA 106:6691–6695. Sugihara, G. 1980. Minimal community structure: an explanation of species abundance patterns. American Naturalist 116: 770–787. Sugihara, G. 1989. How do species divide resources? American Naturalist 133:458–463. Sugihara, G., L. F. Bersier, T. R. E. Southwood, S. L. Pimm, and R. M. May. 2003. Predicted correspondence between species abundances and dendrograms of niche similarities. Proceedings of the National Academy of Sciences USA 100: 5246–5251. Supp, S. R., X. Xiao, S. K. M. Ernest, and E. P. White. 2012. An experimental test of the response of macroecological patterns to altered species interactions. Ecology 93:2505– 2511. Ulrich, W., and N. J. Gotelli. 2010. Null model analysis of species associations using abundance data. Ecology 91:3384– 3397. Ulrich, W., and M. Zalewski. 2006. Abundance and cooccurrence patterns of core and satellite species of ground beetles on small lake islands. Oikos 114:338–348. Van Geest, G. J., H. Coops, M. Scheffer, and E. H. van Nes. 2007. Long transients near the ghost of a stable state in eutrophic shallow lakes with fluctuating water levels. Ecosystems 10:36–46. Vergnon, R., E. H. van Nes, and M. Scheffer. 2012. Emergent neutrality leads to multimodal species abundance distributions. Nature Communications 3:1–6. Volkov, I., J. R. Banavar, S. P. Hubbell, and A. Maritan. 2003. Neutral theory and relative species abundance in ecology. Nature 424:1035–1037. Wantzen, K. M., K. O. Rothhaupt, M. Mo¨rtl, M. Cantonati, G. La´szlo´, and P. Fischer. 2008. Ecological effects of waterlevel fluctuations in lakes: an urgent issue. Hydrobiologia 613:1–4. Winder, M., J. E. Reuter, and S. G. Schladow. 2009. Lake warming favours small-sized planktonic diatom species. Proceedings of the Royal Society B 276:427–435. Zohary, T., and I. Ostrovsky. 2011. Ecological impacts of excessive water level fluctuations in stratified freshwater lakes. Inland Waters 1:47–59.

SUPPLEMENTAL MATERIAL Ecological Archives Appendices A–H are available online: http://dx.doi.org/10.1890/13-1946.1.sm