Introduction
Environment plays a key role in shaping community assembly (Kraft et al., 2015). Across an environmental gradient, species performance relates to their functional traits that mediate the ability to withstand different environmental conditions (Adler et al., 2013; Paine et al., 2011; Poorter et al., 2008; Sonnier et al., 2010). Trait-environment linkages influence compositional change across space and also shape community responses to anthropogenic change (Bernard-Verdier et al., 2012; Cornwell and Ackerly, 2009; Harrison and LaForgia, 2019; Méndez-Toribio et al., 2020; Poorter et al., 2019; Venail et al., 2015). In human-modified landscapes, remnant habitat exist as fragments embedded within a matrix of other land-uses (Barlow et al., 2007; Melo et al., 2013). Fragments experience edge effects that alter the microclimate (De Frenne et al., 2019, 2015; Zellweger et al., 2020) and forest-climate feedbacks that can enhance local climatic stress (Arroyo-Rodríguez et al., 2017; Laurance, 2004). Both processes can decouple the how different traits mediate community composition across macro-scale environmental gradients in fragments compared to contiguous forest (Fernandes Neto et al., 2019; Krishnadas et al., 2018b; Lebrija-Trejos et al., 2010; Poorter et al., 2019; Zirbel and Brudvig, 2020).
Trait-environment linkages can vary between fragments and contiguous forest because of changes in the relative importance of different stressors. For trees, consider wood density and specific leaf area (SLA). Denser wood and lower SLA correlate with greater shade-tolerance (Poorter et al., 2008; Wright et al., 2010). In contiguous forest, tree species with denser wood or lower SLA increase in wetter sites where shade limits species performance. However, greater light availability in fragments could increase abundances of species with lighter wood and/or higher SLA in wetter sites, which would decouple the macro-scale linkages between traits and water availability. Alternatively, because denser wood and lower SLA also correlate with dry-tolerance, they may be favored where climatic water deficit increases across the contiguous forest (Muscarella et al., 2016; O’Brien et al., 2017; Sterck et al., 2006). Warmer, drier local conditions in fragments (Asbjornsen et al., 2004) may favor dry-tolerant traits even in wetter sites. However, if dry microclimates are more deleterious in drier climates, fragments may show tighter the macro-scale linkage between water deficit and lower SLA and/or higher wood density. Instead of individual traits, environmental filters may select for multi-trait phenotypes (Umaña and Swenson, 2018), or assembly may be a response to a composite suite of environmental variables (Shen et al., 2019).
Environmental gradients also shape community assembly via constraints on trait combinations, reflected in trait covariance (Dwyer and Laughlin, 2017). Environmental stress can curtail viable trait combinations or select for specific phenotypes among locally co-occurring species, resulting in stronger covariance at more stressful sites (Dwyer and Laughlin, 2017; Umaña and Swenson, 2018). The sign of covariance also matters to recognize the phenotypes being favored. Positive covariance indicates selection for more extreme trait combinations (e.g., large seeded and dense wooded or large seeded and tall) as may happen if multiple stressors impose coordination across a suite of traits. Negative covariance may imply opposing selective pressures on traits, e.g., higher light availability supports smaller-seeded species but concomitant effects of dry conditions select for species shorter stature or thicker leaves (Muscarella et al., 2016; Tyree, 2003).
In fragments, limited arrival of larger-seeded species and warmer, drier microenvironments can constrain viable combinations of seed size and SLA, height or wood density. Trait covariance could therefore be tighter in fragments compared to contiguous forests, especially where macroclimatic stress also increases. Alternatively, greater heterogeneity of microenvironments within fragments could support more phenotypes than contiguous forest, resulting in a weaker influence of the macroenvironment on trait covariance. For instance, heterogeneous light levels within fragments may allow a wider range of viable SLA or wood density values for the same seed size, whereas more uniformly shaded conditions in contiguous forests restrict trait combinations. Finally, trait covariance could reflect the diversity of the trait space shaped by environmental gradients or local biotic interactions. An increase in the functional diversity of individual traits, i.e., wider occupancy of available niche space, could increase the number of possible phenotypes arising from trait combinations.
Here, I examined how trait-environment linkages and constraints on trait combinations varied between fragments and contiguous forests across macroenvironmental gradients in a human-modified wet tropical landscape in peninsular India. Specifically, I asked:
Do fragmented and contiguous forests differ in trait-mediated species distributions across macroenvironmental gradients?
In contiguous forest, I expected that traits associated with shade-tolerance (denser wood and lower SLA) would increase in wetter sites where there is strong competition for light. Greater light availability in fragments would favor smaller-seeded, lighter-wooded species on average, but larger seeded and denser wooded species would increase in fragments located in warmer, drier conditions. Drier climate in fragments would favor shorter species (Méndez-Toribio et al., 2020). Across contiguous forest however, uniformly shaded conditions may buffer the effects of drier climates allowing similar height profiles in relation to climatic dryness (Arroyo-Rodríguez et al., 2017; Davis et al., 2019; De Frenne et al., 2019).
Do contiguous forests and fragments differ in whether the environment selects for individual traits or multi-trait phenotypes (i.e., trait syndromes) and does this influence composition along environmental gradients?
If selection along environmental gradients operates via individual traits, trait-environment correlations will be stronger for individual traits than multi-trait phenotypes in both contiguous forest and fragments (Butterfield and Suding, 2013; Krishnadas et al., 2018b). Alternatively, successful phenotypes may be determined by a combination of traits. Species with trait values associated with resource-conservative strategies would be expected to increase in more stressful conditions such as warmer, drier climates. If small-scale abiotic conditions in fragments modulate or override the effects of broad-scale environmental gradients, then landscape-level trait-environment linkages seen in contiguous forests will change for fragments.
Does the environment constrain trait combinations in fragments vs. contiguous forest? Or, does trait covariance correlate with the overall trait space occupied by constituent species?
In contiguous forest, I expected stronger trait covariance as sites became drier and warmer. I expected fragments to show stronger correlations between trait covariance and climate gradients as forest loss would reduce the buffering of climatic stress and allow less leeway for different phenotypes (Zellweger et al., 2020), i.e., stronger selection for specific phenotypes leading to tighter constraints on trait combinations. I also expected trait covariance to decrease with higher species richness, functional diversity, or variance of individual traits.