1 INTRODUCTION
Fire plays a key role in ecosystems across the Earth, influencing species composition, physical structure, and processes (Bond et al., 2005; Krawchuk et al., 2009; Pausas & Keeley, 2009; Archibald et al., 2018; Bowman et al. 2020). Natural disturbances such as fire have long been recognized as regulators of biological diversity (Connell, 1978; Huston, 1994). Understanding spatial and temporal variation in the fire-diversity nexus is especially critical given the profound anthropogenic alterations of fire regimes across the Earth and their cascading impacts on ecosystems and species, including humans (Bowman et al., 2020; Coop et al., 2020). Ecologists have responded by intensifying their efforts to develop generalizations about species diversity and fire that address the challenges of a fierier world in the Anthropocene (e.g., Perry et al., 2011; Burkle et al., 2015; Enright et al., 2015; Pausas & Ribiero, 2017; He et al., 2019; Bowman et al., 2020; Coop et al., 2020; Miller & Safford, 2020)
Whittaker (1970, 1972) partitioned species diversity into three components: alpha diversity (α), beta diversity (β), and gamma diversity (γ). α is species diversity at a point in the landscape (i.e., a site), which itself can be decomposed into the number of species (richness) and the evenness of abundances among species. α is usually measured as the mean or median diversity of multiple local sites distributed across an area of study. β captures differences in species assemblages from one site to another and has been measured with a wide variety of approaches (Anderson et al., 2011). The combination of α and β produces γ, the total species diversity supported in the larger area—landscape diversity. Although myriad hypotheses have been proposed and tested regarding the relationship between fire and these three levels of diversity, generalizations have been elusive, which signals the need for further conceptual work and hypothesis testing for understanding the variety of ways in which fire influences diversity (Parr & Andersen, 2006; Anderson et al., 2014; Burkle et al., 2015; Kelly & Brotons, 2017; He et al., 2019; Miller & Safford, 2020). In this paper, we test key hypotheses on the impact of wildfire on woody plant diversity in a topographically complex mountain range.
The intermediate disturbance hypothesis (IDH) proposes that species richness—usually α, less commonly β and γ—varies predictably with disturbance gradients in a unimodal, hump-shaped fashion, in which intermediate levels of disturbance intensity or frequency maintain high diversity (Connell, 1978; Sousa, 1979). The IDH has been vigorously debated (Fox, 2013; Huston, 2014), with mixed support across a wide range of disturbance types and taxa (Sheil & Burslem, 2013). Nevertheless, research has repeatedly revealed a hump-shaped relationship between plant species richness and fire severity (DeSiervo et al., 2015; He et al., 2019; Richter et al. 2019; Strand et al. 2019; Miller & Safford, 2020), especially in frequent, low severity fire regimes (Miller & Safford, 2020). The assumed underlying mechanisms vary, but most propose that different fire severities environmentally select for different sets of species. Under no fire or low fire severity, for example, competitive and fire-resistant species should thrive, whereas, after a high severity fire, fast growing, rapidly colonizing species should predominate. The IDH proposes that at intermediate fire severity, both sets of species can coexist, resulting in a peak in species richness. Despite support for the IDH for fire, other studies have detected neutral, linear positive, and negative relationships between species richness and fire severity (He et al., 2019; Miller & Safford, 2020).
Martin & Sapsis (1992) coined the term “pyrodiversity” to capture a growing awareness of the ecological importance of variation across landscapes in fire severity, frequency, size, and other attributes (see also Krawchuk & Moritz, 2011; Perry et al., 2011; Bowman et al., 2016, He et al., 2019). They argued that pyrodiversity promotes variation in plant assemblages among sites (i.e,, β) because, as with the underlying assumption of the IDH, different sets of species thrive under different conditions related to fire, a phenomenon observed for decades in fire prone ecosystems (Romme, 1982; Bond et al., 2005; Pausas & Ribiero, 2017). Mixed-severity fire regimes, for example, provide a complex mosaic of post-fire conditions that should support a wider range of plant species across a landscape than would low severity or high severity fires alone. These arguments promoted an emerging management dictum that prescribed burning aimed at fostering biodiversity (β and γ) should create a broad spectrum of fire patch characteristics to provide conditions required for the regeneration and persistence of a diverse range of native biota (Perry et al., 2011; Bowman et al., 2016; Kelly & Brotons, 2017). Adding patches of fire to an otherwise long unburned but fire-prone area will generally enhance the diversity of most taxa (He et al, 2019), but there’s disagreement about the rigor of field studies, the shape of the relationship between fire and biodiversity, support for underlying assumptions, and the strength of the evidence for broadly applying these ideas to land management (Parr and Anderson, 2006; Perry et al., 2011; Bowman et al., 2016; Kelly & Brotons, 2017).
Miller & Safford (2020) argue that the IDH and the pyrodiversity hypotheses largely ignore the interaction of life history traits and the historic fire regime of particular ecosystems. They propose that the historic fire regime acts as a filter, selecting only those species with the capacity to regenerate and persist under those conditions. Such life history traits shaped over evolutionary time to adapt to the prevailing fire regime are unlikely to confer similar success to fire regimes other than the historic one. As an example, fire resistant tree species, with thick insulative bark and regeneration from seed, perform well and often dominate under frequent, surface fire regimes, but are readily killed in ecosystems with infrequent, stand replacing fires (Barton & Poulos, 2018; Coop et al., 2020). This leads to the hypotheses that α richness and β should peak at the historic fire severity of an ecosystem rather than necessarily at intermediate severity as predicted by the IDH. It predicts further that adding patches of fire outside of the historic fire regime will not necessarily promote variation in species assemblages across sites (i.e., β) or total landscape diversity (γ) (Miller & Safford, 2020).
In an effort to further explore the relationships between landscape variation in fire severity and woody plant species diversity, we examined the impact of a 2011 megafire on α, β, and γ in an arid, fire-prone mosaic of shrub, woodland, and forest ecosystems in Chiricahua National Monument in the Sky Islands of Arizona, USA. Before Euro-American settlement (<1880), frequent surface fires predominated in conifer and conifer-oak forests (Swetnam et al., 1989; Kaib et al., 1996; Swetnam & Baisan, 1996; Barton et al., 2001 Swetnam et al., 2001), whereas more arid woodlands and interior chaparral experienced mixed fire regimes with longer fire intervals (Kaib et al., 1996; Baisan & Morino, 2000; Taylor et al., in press). Starting in the late 1800s, reduction of fine fuel by livestock grazing and then active suppression largely excluded fire for more than a century (Leopold, 1924; Marshall, 1957; Swetnam et al., 2001), until the 2011 Horseshoe Two Megafire, which burned ~90,000 ha across the entire mountain range, a size unprecedented in the historical fire record. This fire was sparked by more than a century of mounting fuel loads and an increasingly warmer and drier climate, part of a regional surge in very large fires with a significant high-severity component throughout the Southwest USA (Dennison et al., 2014; Abatzoglou & Williams, 2016; Westerling, 2016; Singleton et al., 2019).
The Horseshoe Two Megafire offered the opportunity to evaluate temporal shifts in woody plant diversity across a range of fire severities, spanning from unburned to high-severity wildfire. To this end, we sampled woody plant diversity in 138 plots before (2002-2003) and after (2017-2018) the 2011 fire in three vegetation types and spanning wide fire severity and topographic gradients. We specifically addressed (1) whether α, β, and γ changed from the pre- to post-fire sample periods, (2) the extent to which these changes were driven by the Horseshoe Two Fire, (3) the direction and shape of the relationship of α, β, and γ to fire severity and fire variability among plots, (4) whether diversity patterns with respect to fire were tied to the underlying historic fire regimes of the three vegetation types, and (5) the role of topography in shaping biodiversity independent of the Horseshoe Two Fire.