Introduction

One of the oldest and most important macroecology patterns is the spatial variability in biodiversity over broad geographical extents (Field et al., 2009; Hawkins, 2001; Stein, Gerstner, & Kreft, 2014), an appropriate knowledge of which is an essential key to guarantee the provision of multiple ecosystem services (Benedetti et al., 2020). The planet Earth shows striking gradients in the diversity of plants and animals, from high biodiversity in the tropics to low biodiversity in polar and high-mountain regions (Gaston, 2000; Rosenzweig, 1995; Whittaker, Nogues-Bravo, & Araujo, 2007; Willig, Kaufman, & Stevens, 2003). Due to the alarming rate of biodiversity loss in the last decades caused by anthropogenic interruption (Prescott et al., 2016; Waide et al., 1999), a modern resurgence of interests in taxonomic diversity patterns is likely to contribute important insights for developing a more general theory of species diversity (Castro‐Insua, Gómez‐Rodríguez, & Baselga, 2016). Studies in the past decades has documented the broad-scale spatial patterns of species diversity, and explored the mechanisms for such patterns, leading to conceptual insight on the biogeographical variation of species diversity (Devictor et al., 2010; Field et al., 2009; Gaston, 2000). For example, diversity is often highest at intermediate levels of ecosystem productivity (Grime, 1973; Mittelbach et al., 2001; Waide et al., 1999), and species diversity generally increases with habitat area (MacArthur & Wilson, 1967; Rosenzweig, 1995). In addition, environmental heterogeneity is considered as a universal driver of species diversity across taxa, biomes and spatial scales (Stein et al., 2014). Importantly, both theoretical considerations and empirical analyses suggest that patterns are likely scale dependent (Field et al., 2009; Mittelbach et al., 2001; Mouchet et al., 2015).
Most macroecology researches have been focused on terrestrial ecosystems (Currie & Paquin, 1987; Qian, Ricklefs, & White, 2005), and relative fewer studies have explored geographic biodiversity gradients and the underling mechanisms for aquatic ecosystems (Astorga, Heino, Luoto, & Muotka, 2011; Barbour & Brown, 1974; Heino, 2002, 2011; Irz, Argillier, & Thierry, 2004; Jacobsen, 2004), especially the aquatic ecosystems in the arid and semi-arid region, which provide important habitat for diverse species and water resource for human living (Williams, 1999). Moreover, the geographical distribution of study sites is strongly biased towards Europe and North America, with particularly poor coverage in Asia (Field et al., 2009; Fu, Wu, Wang, Lei, & Chen, 2004). Study of species diversity in aquatic ecosystems is as essential as in their terrestrial counterparts (Stendera et al., 2012). Declines in biodiversity are far greater in freshwaters than in most terrestrial ecosystems (Sala et al., 2000), and freshwater ecosystems may well be the most endangered ecosystems in the world (Dudgeon et al., 2006; Millennium Ecosystem Assessment, 2005). In addition, the actual rates of freshwater species extinction due to human interruptions are much higher than natural extinction rates (Naiman & Dudgeon, 2011). Therefore, a better understanding of the global freshwater diversity gradients and the major environmental drivers remains a major topic (Heino, 2011); and such studies serve to address some fundamental questions for conserving freshwater taxa (Tisseuil et al., 2013).
In addition, biodiversity assessments are an important component of conservation planning and increasingly used to identify land-use management practices that maximise both evolutionary value and ecosystem function (Chapman, Tobias, Edwards, Davies, & Vamosi, 2018). Key requirements are to maintain community resilience to environmental disturbance and to preserve ecosystem functions and services across time and space (Socolar, Gilroy, Kunin, & Edwards, 2016). Consequently, it is often proposed that we need to look beyond merely conserving species richness towards maintaining the maximum diversity of evolutionary lineages and associated ecological functions (Bregman et al., 2016). The idea that functional diversity or functional complementarity performs better than species richness as predictors of ecosystem functions is supported by a range of empirical studies (Flynn, Mirotchnick, Jain, Palmer, & Naeem, 2011; Fründ, Dormann, Holzschuh, & Tscharntke, 2013; Petchey & Gaston, 2007). Functional diversity (FD) is a biodiversity component that represents the extent of the functional differences among species based on the distinction of their morphological, physiological and ecological traits (Petchey & Gaston, 2006). Species loss may lead to a reduction in FD depending on the intrinsic redundancy of assemblages (Flynn et al., 2009; Petchey, Evans, Fishburn, & Gaston, 2007). A decrease on the FD of local and regional assemblages could have dramatic consequences for ecosystem functioning because the traits of species, not just the number of taxonomic units, ultimately drive biodiversity-ecosystem functioning relationships (Dı́az & Cabido, 2001; Hooper et al., 2005). Worldwide, strong geographic differences exist in the ecological attributes of birds (Kissling, Sekercioglu, & Jetz, 2012). Thus, we ask how these factors combined to affect the diversity of species assemblages under different environmental conditions.
Waterbirds are ubiquitous components of freshwater systems, and their diversity and abundance have long been recognized as suitable bioindicators of environmental change in aquatic systems (Caro & O’Doherty, 1999; Wen, Saintilan, Reid, & Colloff, 2016) and serves multiple significant functional roles in ecosystems (Barbet-Massin & Jetz, 2015) (Figure 1). However, as with other freshwater biota, macroecological studies of environmental drivers of waterbirds diversity are rare (Shah, Domisch, Pauls, Haase, & Jähnig, 2014; Stendera et al., 2012; Zeng et al., 2019). It is unclear whether similar latitudinal and other broad geographical (e.g. altitudinal) gradients apply to waterbirds as well. In a recent review, Heino (2011) found no clear latitudinal gradients at regional scale while species richness typically attains highest levels in mountainous regions. Using river basins as the spatial unit, however, Tisseuil et al. (2013) found that the ‘climate/productivity’ hypothesis (Field et al., 2009) explained large portion of geographic variance in waterbirds richness, which is consistent to land avian species (Storch et al., 2006). Several factors are known to affect waterbirds diversity at a local scale, such as lake productivity, lake size, and habitat heterogeneity (Barbour & Brown, 1974; Cintra, 2015; Xia et al., 2016). Linking these local scale variables with broad-scale geographical variations in an integrative analysis framework could potentially articulate the leading processes underlying the regional and global waterbirds richness patterns.
Lakes are ideal systems for studying the relationship between species richness and productivity (Dodson, Arnott, & Cottingham, 2000). It provides a unique system to examine macro-ecological pattern involving geographic and elevational gradients in inland aquatic systems. In this study, we apply multiple regression analysis and structural equation models (SEM) to lake productivity and waterbirds FD data collected from 35 lakes and reservoirs across the temperate zone of arid and semi-arid China to examine macro-ecological pattern involving geographic and elevational gradients in inland aquatic systems. Our primary goal is to test the causal path of geographic location climate lake productivity waterbirds FD. In particular, we first develop separate models for lake productivity and waterbirds FD to compare the dominant factors controlling the gradients of the two indicators. We then build an integrative SEM with explicit causality from productivity to waterbirds diversity to quantify the effect of lake productivity on waterbirds FD. Lake productivity is often a strong predictor of freshwater biodiversity (Dodson et al., 2000), including aquatic animals (and zooplankton as well) (Chase & Leibold, 2002) and phytoplankton (Stomp, Huisman, Mittelbach, Litchman, & Klausmeier, 2011). But its effect on waterbirds has rarely been tested. The productivity-richness hypothesis suggests a positive effect of primary productivity on species richness by allowing larger populations to persist, thereby reducing extinction risk and supporting a higher diversity of niche specialists (Tittensor et al., 2010; Willig et al., 2003). In this context, a significantly positive path coefficient from lake productivity to waterbirds could provide strong support to the productivity-richness hypothesis. However, a weak or no causal link between the two might suggest that waterbirds, not like other aquatic biota like fish (Dodson, 2008) that is constrained by water, have access to other energy sources; and their distribution is governed by other broad scale factors. In this study, we applied FD metrics to waterbird communities sampled in 35 lakes across the whole temperate arid and semi-arid northern China to test the productivity-richness hypothesis, and more importantly, to fill in the gaps in our understanding of ecological patterns in aquatic ecosystems across a large geographical scale.