Introduction

The relationship between ecosystem productivity and biodiversity has been examined well over 1,000 times since it was proposed (Pianka 1966; Odum 1969), with multiple conceptual frameworks and empirical tests. Yet evidence of its direction is inconsistent among (e.g., (Mittelbach et al.2001) and even within (e.g., Adler et al. 2011) studies: sometimes it takes on the expected hump shape, other times a positive relationship, while in some cases a negative or non-existent relationship (Mittelbachet al. 2001; Gillman & Wright 2006; Pärtelet al. 2007). These inconsistencies have led to criticism and intense debate (e.g., Fridley et al.2012; Tredennick et al. 2016) behind the mechanisms underpinning these relationships (Abrams 1995) and methodological approaches (Whittaker 2010). Focusing on pinpointing the mechanisms that underlie diversity-productivity relationships could address how different biodiversity patterns emerge, identify significance of the patterns, and allow for predictive models. A parallel goal in population ecology is understanding mechanisms that yield variability in the performance of individual species along gradients of environmental stress (i.e. productivity): the ‘stress-gradient hypothesis’ (Malkinson & Tielbörger 2010). Although the stress-gradient hypothesis tends to be tested at geographic scales and for individual species, such as across a species’ range (Ziffer-Berger et al. 2014), similar processes likely occur within landscapes. Despite the similar goals of these two subfields, to our knowledge, methods from population ecology have not been applied to the study of mechanisms underlying diversity-productivity relationships in community ecology.
To understand how methods of testing the stress-gradient hypothesis apply to diversity-productivity relationships, it is useful to see the net biodiversity of communities as collections of populations across environments. If a population can persist in an environment, it adds to the richness of the community. ‘Persistence’ is defined as a population’s ability to maintain itself via survival and reproduction, i.e., per capita population growth rates (λ ) ≥ 1. In studies of the stress-gradient hypothesis, fitness is commonly measured on individuals that have been transplanted across the stress gradient. When averaged across individuals, absolute fitness is equivalent to per capita population growth (Brady et al. 2019). By employing methods that measure persistence, we can then identify (i) where along the productivity gradient different species are expected to persist and whether occupancy mirrors persistence, (ii) if different species exhibit similar or dissimilar relationships with productivity, and consequently, (iii) what biodiversity patterns emerge as we aggregate species together in a given environment. For example, biodiversity-productivity relationships might emerge if species differ in how sensitive they are to conditions across the gradient, if the magnitude of species interactions differs between environments, or if species are increasingly lost in poor conditions due to an increased effect of stochasticity within smaller population sizes (Lande 1993; Shoemakeret al. 2020). In order to make these inferences, persistence of all species under consideration must be tested together across the extent of the same productivity gradient. As we will describe, adopting this approach can help answer two outstanding questions about why populations and communities are affected by environmental gradients and address the implicit assumption that patterns in species occurrence and biodiversity across gradients mirror persistence.
The first outstanding question is to what extent are species interactions (e.g. competition, facilitation) versus environmental filtering driving biodiversity-productivity relationships. Productivity gradients may affect biodiversity patterns directly, due to abiotic suitability (i.e., species’ fundamental niches), or indirectly, by modifying the strength and direction of species interactions (i.e., species’ realized niches) (Chase & Leibold 2009; Germain et al. 2018; Carscadden et al. 2020). Theory suggests that as conditions improve, communities crowd causing competition for limiting resources (Harpole & Tilman 2007; Craine & Dybzinski 2013), thus reducing opportunities for niche partitioning and coexistence (Harpole & Tilman 2007). Yet the intensity of biotic interactions and consequences for diversity across productivity gradients may not be so straightforward: competition may vary linearly or nonlinearly or even transition to facilitative interactions (a key prediction of the stress-gradient hypothesis) with abiotic stress (Bertness & Callaway 1994; He et al. 2013; Louthan et al. 2021). To disentangle direct effects of productivity on persistence from indirect effects through biotic interactions, one could measure λ on populations of different species transplanted across a productivity gradient, removing neighbors (i.e., isolating the direct component) or leaving biotic interactions intact. Comparing these λ elucidates how biotic interactions change in magnitude and direction across the productivity gradient. Although intuitive and straightforward, we emphasize that applying this kind of manipulative approach has only been proposed to test the stress-gradient hypothesis in the last half decade (Louthan et al.2015; Jones & Gilbert 2016), and thus, empirical applications are still uncommon, especially when scaling from single species to a community context. Studies taking this approach would mutually benefit both fields.
The second outstanding question is to what degree does species richness reflect local conditions vs. conditions at larger scales via linkages with dispersal. More specifically, a common goal in the stress-gradient literature is to ask if range limits (i.e., the spatial extent of a species’ distribution) reflect niche limits, as opposed to limited dispersal (Hargreaveset al. 2014). By transplanting individuals into locations where absent, thereby simulating dispersal, one can ask if species can maintain populations, and thus if dispersal limitation decreases diversity. This method can be easily extended to test dispersal’s other effect: the maintenance of “sink” populations due to dispersal excess that, in the absence of dispersal, would fail to persist (Howe et al.1991; Pulliam 2000). Sinks may be common if there is high heterogeneity in local conditions in space or time relative to the scale of dispersal (Jorgensen & Fath 2008; Zelnik et al. 2021). By measuring growth rates of transplanted individuals along with species natural distributions, we can determine how dispersal interacts with the biotic and abiotic environment to result in biodiversity patterns (Leibold & Chase 2017). Compared to tests of the stress-gradient hypothesis, dispersal is less often invoked as a mechanism affecting biodiversity-productivity relationships (but see Pärtel & Zobel 2007) despite being key to theories of biodiversity maintenance (i.e., metacommunity theory) (Leibold et al.2004; Leibold & Chase 2017).
In this study, we bridge from population to community ecology, testing the interplay and strengths of local and regional processes that may underlie the stress-gradient and biodiversity-productivity relationships. To do so, we combined occupancy surveys, neighbor manipulations, and fitness assays on transplanted individuals of four annual plant species in a highly heterogeneous natural grassland that spans harsh to productive serpentine habitat. We used these data to test: (1) if species-level occupancy patterns align with persistence under natural conditions, (2) the role of environmental filtering versus biotic interactions in generating stress-gradient and diversity-productivity relationships, and (3) the influence of dispersal on diversity patterns. We predicted a poor alignment between occupancy and persistence due to both dispersal of species beyond their niches and transient populations caused by high spatiotemporal environmental variability. Consistent with the stress-gradient hypothesis, we predicted that competition would be most intense in high productivity environments, weakening or even transitioning into facilitation in less productive environments. Finally, paralleling metacommunity theory, we hypothesized that dispersal would play a key role in structuring diversity at the community level, while exploring if dispersal varies across the productivity gradient and if dispersal excess vs. limitation dominates. By integrating the stress-gradient hypothesis for populations with diversity-productivity and metacommunity theory for communities, we quantified both local and regional level processes structuring biodiversity, thereby testing how well species occupancy mirrors persistence and the mechanisms that may lead to mismatches.