Discussion

Multiple conceptual pillars of community ecology, from spatial and temporal species accumulation curves to the diversity-productivity relationship, are built from patterns of species richness (Ugland et al. 2003; van Ruijven & Berendse 2005; Ulrich 2006). Patterns of diversity across landscapes are commonly evoked to infer assembly mechanisms (Scheiner 2003) and are used to inform conservation and restoration practices (Fleishman et al.2006). Here, we find that for both populations and the aggregate community, species occurrence is strongly mismatched from species persistence across a productivity gradient, emphasizing that occupancy patterns are a poor predictor of where conditions are suitable for maintaining biodiversity. Critically, by comparing occurrence patterns to persistence of transplanted seeds under conditions with and without biotic interactions, we disentangled the impacts of environmental filtering, interactions between neighbors, and dispersal on community assembly and diversity patterns (Vellend 2010; HilleRisLambers et al. 2012). Below, we discuss how our results bridge from patterns to processes, advancing our understanding of the mechanisms structuring the diversity-productivity relationship by integrating niche, range limit, metacommunity, and stress-gradient theories.
Species’ distributions across the productivity gradient largely supported literature on the diversity-productivity relationship. In our study, occupancy patterns of the aggregate community show a linear increase with productivity, rather than the also commonly observed unimodal, hump-shaped relationship (Mittelbachet al. 2001; Pärtel & Zobel 2007; Fraser et al. 2015). Given that our experiment is representative of a smaller area on the harsher side of the global gradient, we may expect to see the linear slope descend into a hump-shaped curve if surveyed at a larger spatial scale (Venail et al. 2010; Germain et al. 2019). Critically, our observed diversity-productivity relationship is driven primarily by sink populations, yielding a mismatch between species’ distributions, realized niches, and fundamental niches (Pulliam 2000; Chase & Leibold 2009; Carscadden et al. 2020).
While environmental filtering structured species’ fundamental niches at the species level (Fig. 1; blue lines), it exhibited a relatively weak effect on community diversity (Fig. 3; purple line), despite being evoked as a seminal process structuring populations and communities in heterogeneous landscapes (Laliberté et al.2014; Hallett et al. 2019). Both occupancy and persistence without neighbors was highest in productive ends of the gradient for three quarters of species, implying that species distributions commonly matched their fundamental niches. However, community occupancy patterns showed a weaker link to persistence without biotic interactions (Fig. 3), with persistence increasing only slightly with productivity. This suggests other mechanisms were dominant in driving community diversity, thus countering previous literature showing environmental filtering as the predominant mechanism shaping diversity patterns (Laliberté et al.2014; Le Bagousse-Pinguet et al. 2017).
Species interactions promoted strong mismatches between persistence and occupancy for both the species-level and community. At the species-level, occupancy was consistently highest in productive environments while persistence with neighbors tended to be maximized in harsher environments (Fig. 1 B-C, F-G). Unexpectedly, we found that distributions matched species’ fundamental niche more closely than the realized niche. One explanation is that seeds deposited in a neighborhood that had a suitable community of neighbors in the maternal generation may be unlikely to experience a similarly suitable community of neighbors in the current generation (Germain et al.2022). Neighborhood stochasticity would be especially pronounced in annual systems with high turnover and with high spatiotemporal variability, like California grasslands. More generally, our results comparing species persistence with and without neighbors corroborates range limit theory: in harsh environments abiotic conditions determine persistence, whereas in productive species interactions set boundaries on persistence (Grinnell 1917; Louthan et al. 2015; Allbee et al. 2023). Yet, forBromus and Plantago, positive interactions among species promoted persistence in harsh environments, thus potentially extending species’ ranges (Stephanet al. 2021). Scaling to the community, these differences in species interactions across productivity lend strong support to the stress-gradient hypothesis. Biotic interactions shift from positive in harsh areas to competitive in more productive environments, leading to strong differences in persistence under environmental filtering alone versus when considering environment by species interaction effects (Fig. 3).
The gaps between patterns of species distributions versus persistence with neighbors points to the dominant role of dispersal in driving population and community dynamics. We found collective effects between dispersal, species interactions, and environment, emphasizing that processes structuring communities are interactive rather than sequential (Kraft et al.2015; Cadotte & Tucker 2017). At the species level, mismatches between occupancy and persistence with neighbors (Fig. 1) were driven by an excess of dispersal, yielding sink populations, especially in productive areas. Given this mismatch, our results support niche theories that allow realized niches to differ in distribution from, or even exceed, the fundamental niche (Pulliam 2000; Louthan et al. 2015; Carscadden et al. 2020). Further, the strong role of excess dispersal underscores the operational value of distinctly defining species’ realized niche vs distributions (Soberón & Nakamura 2009). At the community scale, we find that in harsh areas, species tend to be aligned to their environment (Fig. 4). However, in more productive environments, sink populations dominate—especially when species interactions further restrict persistence. These results likely depend on the scale of environmental heterogeneity; for example, Pinto et. al. (2010) found that sinks were found in fine scales but not at coarse scales. Considering seedbanks as a means of dispersal through time could elucidate further links between dispersal, biotic interactions and environmental filtering along gradients (Wisnoski & Shoemaker 2022). This study reveals the importance of considering a metacommunity perspective, which emphasizes dispersal as a critical driver of diversity patterns, when making inferences from observational data. We identified a need for a more nuanced metacommunity theory (Leibold et al. 2004; Leibold & Chase 2017; Thompson et al. 2020) including the interplay between environmental conditions and propagule pressure (Allbee et al. 2023).
Overall, our results emphasize the benefits of linking species to community patterns and contradicts the assumption that persistence patterns can be approximated via occupancy. While many methods attempt to statistically link species occurrence patterns to underlying mechanisms of dispersal, biotic interactions, and environmental filtering, (Kraftet al. 2011; Leibold et al. 2022) inferring patterns from processes is inherently fraught (Tucker et al.2016; Blanchet et al. 2020). Rather, by empirically measuring occupancy and persistence under abiotic and biotic environments where we manually dispersed seeds, we directly quantified the mechanisms structuring populations and communities, and in doing so integrate several theories in ecology. We suggest that future work in annual plant systems compare multi-year dynamics to disentangle spatial and temporal assembly mechanisms, especially given the hypothesized role of environmental fluctuations in maintaining coexistence (Hallett et al.2019; Van Dyke et al. 2022). Extending these methods across larger gradients could inform species distribution modeling under future climate scenarios (Buckleyet al. 2010). Similarly, we envision these methods informing long-term success in restoration practices and the potential need for active management of competitors or regular seed addition (Shackelford et al. 2021; Aoyama et al. 2022). Our results provide a cautionary tale of inferring mechanisms that structure populations and communities, or in forecasting future species distributions or diversity, based solely on current occupancy patterns.