Discussion

In this study, we applied integrated modelling procedures (SEM) to productivity and waterbirds functional diversity data collected from lakes and reservoirs in the arid and semi-arid northern China covering over 5 million km2 with the aim to explore macroecological pattern involving geographic gradients in inland aquatic systems. Both lake productivity and waterbirds FD displayed strong geographical variations across northern China (Figure 3). We found that the geographic position exerted effects on lake productivity and waterbirds FD in a similar way, i.e. through its influences on climatic conditions, which was defined by seven bioclimatic variables in this study. This causality from geographic position to climatic conditions was significant and consistent in all three SEMs. Specifically, our analyses showed an unambiguous decreasing elevational gradient for both lake productivity and waterbirds FD (Herzog, Kessler, & Bach, 2005; Rahbek, 1995). An increasing gradient with location coordinates was also obvious; and the effects of latitude (through its effects on climate) on both lake productivity and waterbirds FD were relatively weaker than those of longitude. Moreover, the “latitudinal gradient”, which predicts species diversity decreases when moving away from the equator towards northern latitudes (Jetz, Thomas, Joy, Hartmann, & Mooers, 2012), was not supported. Instead, we found evidence of a reverse trend in our analyses (Figure 4a). The opposite latitudinal gradient in this study should be treated with cautions as the study focused on the temperate zone that have a rather narrow latitudinal range and strong negative effects of elevation might obscure the real latitudinal pattern. Nevertheless, our analysis results supported a number of hypotheses underlying the geographic gradients by Field et al. (2009): 1) Climate; 2) Productivity; and 3) Size of ecosystem. More importantly, our results revealed a hierarchical structure of drivers that regulate the observed biodiversity and productivity patterns (Macneil et al., 2009).

Drivers of geographic and elevational patterns of lake productivity

The lake productivity model had relatively high performance in explaining the observed geographic and elevational gradients (R2 for separated and integrated models was 0.93 and 0.96, respectively, Figure 5 and 7). Both linear regression analysis (Figure 4d, 4e) and structural equation modeling (Figure 5 and 7) showed that lake productivity was positively related with water TN and TP concentration. Moreover, the results from the integrated SEM indicated that the effect of nutrient, which was defined by TN and TP concentrations in water, was much stronger than that of climate. The close relationship between nutrient, lake chlorophyll a concentration, and lake productivity (Jeppesen et al., 2005; Smith, 1979) was expected because nitrogen and phosphorus are important limiting nutrients in freshwater ecosystems (Elser et al., 2007; Schindler, 1974). Lake ecosystem productivity depends on the supply of nutrients, especially phosphorus (Wetzel, 2001). According to previous studies, phosphorus loading alone could explain 79-95% of the variances in lake chlorophyll a concentration (Schindler, 1978).
Climatic conditions, measured mainly by temperature and precipitation variables including mean temperature of the wettest quarter, mean temperature of the warmest quarter, annual total precipitation, precipitation of driest month, and precipitation of the warmest quarter also had positive effects on lake productivity, except for mean diurnal range which showed the opposite trend (Figure 5 and 7). Climatic variation has been found to influence the magnitude of chlorophylla concentration (O’Reilly, Alin, Plisnier, Cohen, & Mckee, 2003). Lake productivity increased with air temperature, which is the function of solar energy input. (Warner & Lesht, 2015) reported that air temperature and precipitation were identified as important predictors, which had positive effects on chlorophyll a . Our results are consistent with those studies, although the mechanisms are not clear. One possibility is that higher air temperature reduces ice cover, which facilitates wind-induced mixing and nutrient resuspension (Nicholls, 1998; Schwab, Eadie, Assel, & Roebber, 2009). And increases in the form of rain would cause increased run-off, which could bring more nutrients to the lakes from non-monitored or diffuse non-point sources. However, these hypotheses were not fully explored in our study, and further researches are needed to understand the mechanisms on how climate influences lake productivity. In addition, in our study, it was mean diurnal range, mean temperature of the wettest quarter, mean temperature of the warmest quarter showed significant effects on lake productivity. For which, it may due to the sampling time (i.e. we monitored lake chl-α in summer).

Drivers of geographic and elevational patterns of waterbirds functional diversity

The waterbirds FD SEMs achieved sufficient performance in explaining the spatial variations in waterbirds FD (R2 for both separated and integrated models were 0.52 and 0.56 respectively, Figure 6 and 7). Geographic position, defined by latitude, longitude and altitude, had dominating effect on waterbirds FD through its influence on climate. Climate is typically a strong descriptor of broad-scale richness patterns (Hawkins et al., 2003), and the theory that climate’s control of energy drives the global richness gradient has generated an extensive literature quantifying the relationship between species richness and climatic variables (Whittaker et al., 2007). Our results demonstrated that waterbirds FD increased with temperature, giving empirical affirmation to the species-energy hypothesis in that species diversity increases with environmental temperature (Allen, Brown, & Gillooly, 2002). Temperature is one of major determinants of latitudinal and altitudinal gradients in animal diversity (Allen et al., 2002; Rohde, 1992), which may be explained by energy hypothesis, although the underlying mechanism remains unknown (Hawkins et al., 2003). The model results also showed that waterbirds FD increased significantly with precipitation. Furthermore, the modelled path coefficients indicated that precipitation was more important than that of temperature in our system. Precipitation is one of the resource-based estimates of available energy, especially in arid and semi-arid ecosystems (Brown & Davidson, 1977), where biodiversity patterns are strongly related to precipitation amount (Waide et al., 1999). This is particularly true for waterbirds, whose distribution is generally determined by rainfall (Wen et al., 2016) through changing habitat availability, like water depth, habitat area and habitat diversity (Canepuccia, Isacch, Gagliardini, Escalante, & Iribarne, 2007). As most of the lakes included in our study are located in the arid and semi-arid region of China, precipitation is critical to maintain enough habitat area for waterbirds.
Lake morphology, which was defined solely by lake area in this study, had positive effect on waterbirds FD. This pattern resembles the common species-area relationship observed in many ecosystems (Arrhenius, 1921; Guadagnin, Maltchik, & Fonseca, 2009; Keil, Storch, & Jetz, 2015; MacArthur & Wilson, 1967; Nogues-Bravo & Araujo, 2006; Rosenzweig, 1995). According to the theory of island biogeography (MacArthur & Wilson, 1967), large and more diverse ecosystems are likely to harbor more species due to higher immigration rates and lower extinction rates. Indeed, lakes in arid and semi-arid zones can be regarded as aquatic islands in a terrestrial world, offering an explanation for the positive species-area relation in our analysis. This finding is consistent with Suter (Suter, 1994), who reported species-area relationship for waterfowl assemblages on the 20 major Swiss lakes north of the Alps. Likewise, Froneman et al. (2001) described a positive species-area relationship in waterbirds communities in farm ponds, South Africa, and Guadagnin et al. (2009) also described that the number of waterbirds species in wetlands from the Atlantic coastal zone of Brazil showed a positive relationship with their size.

Relationship between waterbirds species richness and lake productivity

Many studies revealed that productivity affects diversity (Carpenter et al., 1987; Dodson et al., 2000; Mittelbach et al., 2001), especially for plants (Chase & Leibold, 2002). Nonetheless, no general consensus concerning the form of the pattern has emerged based on theoretical considerations or empirical findings (Waide et al., 1999). And positive, negative, and hump-shaped patterns were common at most spatial scales and no one pattern predominated (Mittelbach et al., 2001). For avian species, particularly waterbirds, there are only a few studies presented the relationship between diversity and productivity (Hawkins et al., 2003; Hurlbert & Allen, 2004). Results of our integrated SEM gave evidence to support the causality from lake productivity to waterbirds FD albeit the relationship was weak in comparison to other factors. As the majority of waterbirds forage on the riparian zone of lakes (mean trait value of foraging at ground was greater than 55% for all lakes, Appendix A1) and have plants as their major diets (Appendix A2), this weak causality is expected.

Conclusions and caveats

A major contribution of this study is that our findings reveal the key environmental drivers of large-scale patterns in lake productivity and functional diversity using advanced statistical techniques (i.e. SEM). This approach showed that the observed geographical and altitudinal gradients in lake productivity and waterbirds FD (Figure 3) can be partly explicated by the gradients in climatic conditions, which is in term significantly related to the geographic position of the lakes on the earth surface. As the relationship between productivity and species diversity in arid and semi-arid ecosystems has not been addressed, our study could contribute to the mechanistic explanations underlying the observed broad – scale biodiversity gradients in arid and semi-arid region. However, site-specific factors, such as lake morphology (for waterbirds) and nutrients (for productivity) impose their effects independently (Figure 5-7), and their effects could be more important than climatic variables (e.g. for lake productivity). These results, although supporting some primary macroecological biodiversity theories such as species–energy and species–resource hypothesis, could not lead to a mechanism that unifies these theories (Mcgill, 2010), exemplifying one key limitation of statistical analyses: statistical relationships do not necessarily reveal the underlying mechanisms regulating waterbirds biodiversity (Stomp et al., 2011). For example, the SEM indicates that altitude has a strong negative effect (indirectly through climate and lake productivity, Figure 7) on waterbirds FD. However, these causal paths could be driven by other environmental variables that co-vary with altitude but were not measured in our study. For instance, seasonal variation in environmental conditions increases at higher elevation, which could reduce species diversity by excluding sensitive species with a narrow tolerance range (Currie et al., 2004). Nevertheless, through articulating the dominant processes, our results could contribute to future studies seeking mechanistic explanations underlying the observed macroecological phenomena.