Spatial synchrony occurs when properties measured in populations (e.g., species abundance), communities (species richness) or ecosystems (e.g., primary productivity) vary synchronously. Several studies have demonstrated synchrony of temporal dynamics in local population densities (Knops and Koenig 1998; Shanker and Sukumar 1999; Cattadori et al 1999; Liebhold et al 2004a). Other studies also revealed synchrony as an ubiquitous pattern, occurring in different regions (temperate and tropical), ecosystem types (e.g., aquatic and terrestrial) and on different groups of organisms (Moran 1953a, b; Ranta et al 1995, 2008; Bj??rnstad et al 1999; Ims and Andreassen 2000; Koenig 2001; Tedesco et al 2004; Lansac-T??ha et al 2008; Stange et al 2011).
Thus, spatial synchrony has important implications for population persistence. Empirical and theoretical studies indicate that synchrony enhance is associated with the increasing of regional extinction probability (Paradis et al. 1999). The more synchronous a metapopulation is, the lower is the expectation of the persistence time (Liebhold et al. 2004a). In this context, a process that causes extinction of a local population would lead all populations that fluctuate synchronously to the same risk of extinction. Besides, some level of synchronization can also allow local populations to function as source and restore extinct populations (Matter and Roland 2010).
Asynchronous population dynamics require a greater number of monitoring locations (Burrows et al. 2002). In this context, another important implication for spatial synchronization studies is the possibility to reduce the effort of sampling, since synchronous dynamic populations can be monitored in few environments (Stoddard et al. 1998; Anneville et al. 2004; Rhodes and Jonz??n 2011). Assuming regionalized or synchronous dynamic, the data from local “sentinels” can then be extrapolated to the whole area of interest (Anneville et al. 2004).
In general, the ubiquity of synchronous patterns has motivated ecologists in the search of underlying mechanisms. For example, geographical distance between local populations are generally negatively correlated with spatial synchrony, in other words, synchrony values decrease with increasing distances between pairs of local populations (Ranta et al. 1999; Koenig 2002). This synchrony decline can be explained in both terms of lower dispersal rates between environments separated by greater distances (Ranta et al. 1995; Lande et al. 1999; Paradis et al. 1999), as well as the decrease of environmental similarity versus distance (Ranta et al. 1999; Koenig 2002).
Trophic interactions between species can also synchronize local population dynamics (Buonaccorsi et al. 2001; Liebhold et al. 2004a). For example, on Lake Inari islands (Finland), mustelids predation was the primary cause of synchronous dynamics of small rodents populations (Heikkila et al. 2012). A study involving rodent populations Ondatra zibethicus L. and the predator Neovison vison Schreber in Canada also obtained similar results (Estay et al. 2011). However, those studies emphasized that different processes (dispersal, predation and fluctuations in precipitation) can explain the patterns of synchrony.
Environmental factors also exhibit a synchronizing effect on local populations dynamics (Liebhold et al. 2004b). In many regions, spatially disjunct populations may exhibit synchronic fluctuations determined by environmental variations (Hudson and Cattadori 1999; Koenig 1999, 2002). This pattern is known as Moran effect (Moran 1953a). Evidence for the influence of environmental factors synchronizing the population dynamics are frequent. For example, hydrological fluctuations were probably decisive in synchrony patterns of fish populations from three drainage basins in Africa (Tedesco et al. 2004). Variations in precipitation were also the main synchronizing mechanism of bird populations in England (Cattadori et al. 2000). However, the identification of the main cause of synchronization is often difficult because all the mechanisms can produce almost identical signatures for synchrony between populations (Liebhold et al. 2004a).
Climatic and hydrological factors have been recognized as important in influencing patterns of synchrony of planktonic populations (George et al. 2000; Downing et al. 2008; Rusak et al. 2008; Xu et al. 2012). However, the patterns of synchrony of planktonic populations can be strongly influenced by local processes, including variations in limnological features (e.g. Lansac-T??ha et al. 2008). There are also examples of studies in a unique environment with high levels of synchrony and a decrease of such synchrony with increasing distance between the sampling sites (Seebens et al. 2013).
In this context, the aim of this study was to quantify the level of synchrony between local zooplankton populations in a tropical reservoir (Lajes, RJ) using data collected in a long-term ecological study. Considering the closeness of the sampling points (i.e., all performed in a single system), we expected to find high levels of synchrony. These high levels would be explained by the high rates of dispersal (from upstream region toward the dam area) as well as the Moran effect (i.e., influence of density independent factors that act in a similar way in the different local populations dynamics). Furthermore, synchrony decay with increasing distance between sites is predicted according to the dispersion hypothesis, mainly due to spatial heterogeneity in the reservoir (Soares et al. 2008). The presence of synchronous patterns can reveal the effects of density independent processes, in a regional scale, on population dynamics (e.g., Moran effect). Moreover, the lack of synchrony can suggest local factors as the most important regulators of population dynamics (Kratz et al. 1987).
The sampling were performed in Lajes reservoir (Rio de Janeiro; Figure 1) a tropical reservoir located into Serra do Mar formation, surrounded by fragments of Atlantic Forest. It was built between 1905 and 1908, the water retention time is about 300.67 days, the maximum area is 47.8 km2, volume of 450 x 106 m3 and average depth of 15 m (maximum = 40 m). The Lajes reservoir has a warm monomitic default, remaining stratified mostly year. According to total phosphorus and orthophosphate levels these reservoir is classified as oligo-mesotrophic (Branco et al. 2009).
Sampling and Analysis
Samples were taken monthly from August 2001 to December 2009 at six sites distributed along the main axis of Lajes reservoir. Between July 2003 and October 2004, we were not able to perform the samples. Therefore, temporal series data was about 85 months. Zooplankton samples were collected at each location by filtering 20 liters of sub-surface water with a plankton net (68 ??m) and preserved in 4% formalin. Zooplankton samples were analyzed under a binocular microscope using a Sedgewick-Rafter counting chamber. For further analyses only taxa that occurred in at least a quarter of total samples were used (23 taxa distributed among protozoans, rotifers, cladocerans and copepods), to avoid inconsistencies arising from the large number of zero. Abiotic parameters like water temperature, dissolved oxygen, electric conductivity, pH, water transparency (Secchi disc), were measured in situ using a multi-parametric probe (YSI 6920). Water samples were also collected for the determination of nitrate, ammonium, nitrite, total phosphorus (Total-P), orthophosphate (Ortho-P) and chlorophyll-a concentrations (Chlor-a) and biochemical oxygen demand (BOD) (Table S1) according to standard methods (APHA et al. 2005).
In order to assess spatial synchrony a matrix of six columns was built (referring to the six sampling points), each matrix containing the density data of taxa and large groups of zooplankton. With the assistance of mSynch function (modified to Spearman correlation) the average correlation and confidence interval “bootstrap” (Bj??rnstad et al. 1999) was calculated. A correlation matrix was estimated for each of the taxa (S1, … ,S23) and also for large groups (SProtozoa, SRotifera, SCladocera e SCopepoda), as well as for each environmental variable (ETemp, … ,EBOD), latter with Pearson correlation (Buonaccorsi et al. 2001).
A matrix of environmental distance was generated (D), i.e. Euclidean distance between points, considering the average values of standardized environmental variables. A geographical distance matrix (G) was also calculated, using Euclidean distance between points and based on geographic coordinates (decimal degrees).
The Mantel tests were performed to evaluate whether the levels of population synchrony for large groups (SProtozoa, SRotifera, SCladocera e SCopepoda) were correlated with geographical distance (G) or environmental synchrony (ETemp, … ,EBOD). Mantel test has received several recent criticisms (e.g. Harmon and Glor 2010; Legendre and Fortin 2010; Guillot and Rousset 2013). However, it is still often used in ecological and evolutionary studies (Diniz-Filho et al. 2013), and, more importantly, is one of the few appropriate tests when the hypothesis under consideration can be formulated in terms of distances (Legendre and Fortin 2010).
Multiple regression analyses with distance matrices (MRM, Lichstein 2006; Zapala and Schork 2006; Haynes et al 2013) were used to model the levels of population synchrony (S1, … ,S23) and large groups (SProtozoa, SRotifera, SCladocera e SCopepoda) as a function of geographical (G) and environmental distances (D), even as a function of geographic distance (G) and average environmental synchrony (EM).
All analyses were performed with R software (R Core Team 2013). Mantel tests and MRM were carried out using the mantel and MRM functions available on ecodist package (Goslee and Urban 2007) and mSynch function (Spearman) on ncf package (Bj??rnstad 2012) was used to estimate average levels of synchrony and confidence intervals.
It was found 186 zooplankton taxa and the highest density was recorded in July 2007 at sample site 1 (14.19 million ind.m-3) and the lowest value was measured in May 2006 at the same site and in January 2007 at point 6 (100 ind.m-3). The density fluctuation of all zooplankton groups showed a similar dynamics over the temporal series (Figure 2).
The following taxa are those occurred in at least a quarter of the samples: Centropyxis aculeata, Difflugia spp. (Protozoa), Ascomorpha ecaudis, A. saltans, Collotheca spp., Conochilus coenobasis, C. unicornis, Hexarthra spp., Keratella americana, K. cochlearis, Polyarthra sp., Ptygura sp., Synchaeta sp. (Rotifera), Bosmina hagmanni, Ceriodaphnia silvestrii, Daphnia gessneri, Diaphanosoma birgei (Cladocera), larvae, juveniles and adults of Notodiaptomus cearenses, Mesocyplos sp. and Thermocyclops sp. (Copepoda).
The synchrony of environmental data showed particularly high values for temperature, dissolved oxygen, nitrite, nitrate and pH, while the lowest values were recorded for Total-P, Ortho-P and chlorophyll-a (Figure 3). For zooplankton groups, Rotifera showed the highest values followed by Copepoda, Cladocera and Protozoa (Figure 4). At taxon scale, Cyclopoida nauplii showed the highest values of synchrony followed by Collotheca, Keratella, Hexarthra, Calanoida nauplii and Bosmina (Figure 5). Nauplii, beyond the high level of synchrony, showed a high density and frequency of occurrence. Additionally, we observed positive relationship between some nauplii feature as the mean synchrony with density and with frequency of occurrence (?? = 0.52 and ?? = 0.73, respectively). It is also important to note that the values of environmental synchrony were generally higher than levels of population synchrony.
For Protozoa group the population synchrony matrix was significant and negatively correlated with geographic distances between points. The fluctuation of protozoa density tended to be more synchronized at nearby sample points (Table 2, Figure 6). The synchrony matrices estimated with water transparency and electric conductivity were significant and positively correlated with synchrony of different groups of zooplankton. Similar results were recorded for environmental synchrony matrices estimated with the variables nitrate (except for rotifers) and chlorophyll-a (except for cladocerans). Synchrony matrices calculated with ammonium ion and BOD variables were correlated with protozoa synchrony matrix. The matrix of water temperature synchrony was significantly associated with synchrony matrices of protozoa, cladocerans and copepods. On the other hand, the synchrony of the zooplankton groups were independent of those matrices estimated with dissolved oxygen, pH, nitrite, Ortho-P and Total-P.
According to the results of multiple regression analysis (MRM), the matrix of population synchrony was significantly correlated with geographic distance only for protozoans and few other taxa (Table 3). Actually, most taxa present environmental distance as more important predictor for variability in population synchrony. Consequently, the population synchrony declined more consistently with environmental distances than with geographic distances (Figure 6). Another interesting result was the environmental synchrony decay with the increasing of geographical distance between the points (Figure 7).
Likewise, the multiple regression models using the average environmental synchrony and geographical distance showed greater environmental effect on population synchrony of groups and zooplankton taxa (Table 4). Hence, the explanatory power of multiple models using environmental synchrony was generally higher than the model that used environmental distance (Tables 3 and 4).
Synchrony is a significant and recurring result for different taxonomic resolutions at the present time series study. The synchrony of zooplankton populations have been detected in different types of ecosystems, both temperate (Rusak et al. 1999, 2002) and tropical (Caliman et al. 2010). The results obtained in this study and for several other groups of organisms (from insects to mammals: Moran 1953a, b; Ranta et al 1995; Koenig and Knops 1998; Cattadori and Hudson 1999; Shanker and Sukumar 1999; Koenig 2001; Lansac-T??ha et al 2008; Stange et al 2011) and different spatial scales (Koenig and Knops 2013; Seebens et al 2013) strongly suggest that, like environmental synchrony (Koenig 2002) the population synchrony is an ubiquitous phenomenon.
Different processes can generally cause population synchrony (Moran 1953a; Paradis et al. 2000; Koenig 2002). One possible cause is the dispersion that can lead to synchronization in environments with populations spatially close enough to allow continuous exchange of individuals. Passive transport by water currents induced for wind can drive the horizontal distribution of lacustrine zooplankton (Seebens et al. 2013). In reservoirs, the flow of water from upstream stream ecosystems may be the main factor determining the horizontal distribution of zooplankton (Marzolf 1990). Hence, the transport phenomena can have great importance as synchronizing agent in reservoirs (Lansac-T??ha et al. 2008). For Protozoa, in part, this scenario seems plausible when the explanatory variables used were spatial and environmental distances. The importance of the spatial component can be explained considering this a pseudoplanktonic group. Thus, the dynamics of these organisms is associated with fluctuations in the water flow that transport individuals from the substrate into the water column (Lansac-T??ha et al 2008; Velho et al 2013).
The Moran effect (Moran 1953a) is another possible cause for the population synchrony, that is, driven by environmental factors synchrony. This study results indicate that the similarity of limnological dynamics (i.e. environmental synchrony matrix) and/or similarity between sites considering the average of the limnological variables (i.e. environmental distance matrix) were important determinants of population synchrony. On the other hand, most of population synchrony matrices showed no significant relationship with distance. In addition to this, the average level of environmental synchrony showed a decay with increasing distance. Therefore, this combination of results (i.e., the greater importance of environmental similarity between different regions of the reservoir than that of geographic distance), is a strong evidence that the spatial synchrony of zooplankton populations was strongly influenced by environmental synchrony. In this context, the inference about the importance of the Moran effect can be maintained even though environmental synchrony has declined with increasing geographic distance (Ranta et al 1999; Koenig 2002; Lansac-T??ha et al 2008; Fox et al 2011). This result is consistent with a growing number of studies that point the Moran effect as an important factor driving population synchrony (Hudson and Cattadori 1999; Lundberg et al 2000; Lima-Ribeiro et al 2007; Koenig and Knops 2013).
Different limnological variables (related to density dependent and independent processes) can influence demographic rates of zooplankton populations. Although it is difficult to establish the relative importance of different variables measured in determining patterns of synchrony, the results of the simple Mantel test suggest that the main variables (i.e., those with average correlation higher than 0.6) were nitrate, transparency, electrical conductivity, temperature and chlorophyll-a. To a greater or lesser degree, these variables are correlated with meteorological conditions. For example, precipitation may increase rates of runoff and the concentration of ions. Assuming this process, it is also expected the water transparency decrease followed by increase of nitrate levels. Variables strongly influenced by weather conditions were also major determinant in synchrony of aquatic organisms density reported in many studies (Magnuson et al 1990; Grenouillet et al 2001; Anneville et al 2004; Rusak et al 2008).
Some studies have shown that variables with more direct association to the biological component of the system have less predictive power for the spatial population synchrony (Kratz et al 1997; Baines et al 2000; Rusak et al 2008; Caliman et al 2010). However, the results of this study showed that the matrix of environmental synchrony based on chlorophyll-a were significantly correlated with synchrony matrices for protozoans, rotifers and copepods in Lajes reservoir. Taking into account that Lajes is an environment with low phytoplankton biomass (Soares et al. 2008), the dynamics of this variable can directly affect the dynamics of zooplankton.
While previous synchrony studies were developed in discrete systems (e.g., different lakes), a unique environment as such reservoir can also be considered an excellent system to understand ecological synchrony due to its spatial heterogeneity. The present work demonstrates the existence of spatial synchrony at zooplankton community and environmental variables. For this particular situation, the levels of synchronization within the monitoring area indicate that temporal variation patterns are similar regardless of the sampled site. However, the values of synchrony were not so high to justify a reduction in the number of monitoring points (Rhodes and Jonz??n 2011).
The importance of regionalized environmental dynamics in synchronizing the population dynamics has often been suggested (Haynes et al 2013), including multiple studies in aquatic ecosystems (Cottenie et al 2003; Kent et al 2007; Seebens et al 2013). However, surprisingly few studies, in particular with planktonic populations (e.g., George et al. 2000) demonstrated an association between population and environmental synchrony matrices. An auspicious direction for future work would be to share the data (used in previous studies on population synchrony) with the objective of quantifying the relative importance of the main synchronizing agents using a standardized analytical protocol (as the one used by Haynes et al. (2013), and in the present study). This is an open invitation to interested.