Abstract: | A major challenge in predicting species’ distributional responses to climate change involves resolving interactions between abiotic and biotic factors in structuring ecological communities. This challenge reflects the classical conceptualization of species’ regional distributions as simultaneously constrained by climatic conditions, while by necessity emerging from local biotic interactions. A ubiquitous pattern in nature illustrates this dichotomy: potentially competing species covary positively at large scales but negatively at local scales. Recent theory poses a resolution to this conundrum by predicting roles of both abiotic and biotic factors in covariation of species at both scales, but empirical tests have lagged such developments. We conducted a 15-y warming and herbivore-exclusion experiment to investigate drivers of opposing patterns of covariation between two codominant arctic shrub species at large and local scales. Climatic conditions and biotic exploitation mediated both positive covariation between these species at the landscape scale and negative covariation between them locally. Furthermore, covariation between the two species conferred resilience in ecosystem carbon uptake. This study thus lends empirical support to developing theoretical solutions to a long-standing ecological puzzle, while highlighting its relevance to understanding community compositional responses to climate change.A readily observable phenomenon in nature is the tendency for the distributions of potentially competing species to covary positively at large spatial scales but negatively at small scales (1, 2). This scale dependence in patterns of species covariation is a defining phenomenon in ecology (3), and a classic illustration of it derives from MacArthur’s observations of Dendroica sp. warblers in mixed forests of the northeastern United States (1) and related theoretical work (4, 5). However, while opposing patterns of species covariation at large and local scales are ubiquitous, assigning causality to interacting drivers of such patterns in natural systems is challenging. Originally, theory explained this phenomenon as a product of distinct types of drivers of species abundance and distribution at large versus local scales. According to this framework, regional factors, such as climate, determine species’ distributions over large scales, while biotic interactions such as exploitation and interference determine presence, absence, and relative abundances of species at local scales (5–10). Hence, species with similar resource demands should, and often do, overlap spatially (covary positively) at broad scales as their distributions track abiotic niche requirements such as favorable climatic conditions (11). Meanwhile, the same species should, and often do, covary negatively at smaller spatial scales, where local biotic interactions such as competition, interference, niche complementarity, or exploitation by consumers or pathogens promote exclusion or segregation (5, 12–14). More recent theoretical developments have, however, highlighted the potential for roles of both types of drivers in patterns at both scales (7, 15, 16). Understanding whether, and how, climate and biotic interactions simultaneously influence species’ covariation at large and local scales has been repeatedly identified as a key challenge in improving predictions of species’ distributional and biodiversity responses to climate change (15, 17, 18).In contrast to progress in theory, field experimental tests of such potential interactions between biotic and abiotic factors in opposing patterns of species covariation at large and local scales have been lacking (14), in part because of the challenges inherent in conducting sufficiently controlled field experiments over suitably long time scales (19, 20). Consequently, novel empirical support for the role of, for example, biotic interactions in large scale patterns of species covariation has been strictly observational (21). Application of more robust empirical tests of predictions deriving from recent theory on this topic may also improve understanding of the consequences of patterns of species covariation at opposing spatial scales for important aspects of ecosystem function (22), including carbon exchange (23–26). Here, we present results of a 15-y warming and herbivore-exclusion experiment conducted at a remote arctic field site aimed at investigating influences of both drivers on patterns of covariation between two dominant shrub species at local and large spatial scales. The experimental design targets temperature as the abiotic limiting factor and herbivory (and associated ancillary effects) as the biotic limiting factor (Methods).The two focal shrub species in this study, dwarf birch (Betula nana) and gray willow (Salix glauca), hereafter “birch” and “willow,” respectively, are the most abundant plant species at our study site in low-arctic Greenland (27), and their functional role in ecosystem CO2 exchange far exceeds that of any other vascular plant species at the site (28, 29). Furthermore, the two species are codominant across much of the Arctic () (30, 31), but some experimental evidence indicates that Betula has the capacity to outcompete Salix at local scales in the Arctic due to its greater developmental plasticity and ability to invest rapidly in stem growth (32). Hence, although annual sampling throughout the duration of our experiment has assessed aboveground dynamics of all components of the plant community (Methods), our focus here is on patterns of covariation between birch and willow. Although birch is generally more common than willow across the study site (SI Appendix), the two species share similar distributions across the site, occur mainly on low to mid elevation slopes and plateaus, and predictably avoid arid steep slopes and stagnant mesic or saturated lowlands and fens (). Each of the two species readily forms monospecific “shrub islands” at the local scale ( and SI Appendix, Fig. S3).Open in a separate window(A) Circum-Arctic distributions of the two focal shrub species, dwarf birch (B. nana) and gray willow (S. glauca). Shaded polygons were derived from published range maps (30, 65). Point locations were derived from occurrence records (66–68) and the GBIF data portal (www.gbif.org). (B and C) Landscape and local scale views of patterns of covariation between the two species at the study site near Kangerlussuaq, Greenland. (B) South-facing hillside and lowland plains at the study site illustrating cooccurrence of dwarf birch (B. nana) and gray willow (S. glauca) at the landscape scale. (C) Monospecific shrub islands of each species are evident at smaller plot scales at the study site. In both photographs, birch appears dark or olive green, while willow appears lighter green. Image credit: E.P. |