Drivers of tree community assembly during tropical forest post-fire succession in anthropogenic savannas

Highlights

• We surveyed transects in an area of recent forest post-fire recovery in savanna.

• We analyzed spatio-temporal shifts in trait and phylogenetic community composition.

• Fire and drought entail environmental filtering in young edge communities.

• Temporal changes are faster in edge communities than in forest interior.

• We provide insights on the processes driving tropical forest post-fire recovery.

Abstract

In the context of global change, tropical forests are increasingly affected by fires. Understanding the ecological processes driving forest recovery in fire-modified landscapes is a critical issue.

We analyzed spatial and temporal (8 years) changes in functional and phylogenetic composition of tree communities during forest post-fire recovery in anthropogenic savannas. We used null models to infer the main assembly processes driving forest succession along three 90-m transects running from an advancing savanna-forest edge to forest interior in New Caledonia. We also evaluated if successional changes differed between large and small trees, or depended on the demography of remnant savanna trees.

We found coordinated shifts from drought- and fire-resistance towards shade-tolerance strategies, involving leaf, stem, and architectural traits along transects. Our results indicate stronger environmental filtering and faster temporal changes in composition of young edge communities. In forest interior, our results suggest slower compositional changes, with an important role of light limitation in community assembly. These non-random patterns depended on both the decline of savanna trees and compositional changes among forest species. We also found contrasting community patterns depending on tree size, supporting a stronger influence of environmental filtering on small trees.

Our work emphasized the dominance of deterministic assembly processes driving tropical forest post-fire succession. Our study suggests that fire and drought drive environmental filtering during early succession at the forest edge, entailing constraints on multiple functional dimensions. As succession progresses, light-limitation becomes a stronger driver of community assembly, and community composition becomes more stable in time. Our study provides insights for a better understanding of the processes guiding tropical forest succession in the particular context of post-fire forest recovery.

Introduction

Tropical forest recovery maintains biodiversity and ecosystem services in human-modified landscapes (Arroyo-Rodríguez et al., 2017; Boukili and Chazdon, 2017; Chai et al., 2016), and studying ecological successions allows to better understand, predict, and assist forest recovery (Buzzard et al., 2015; Cheesman et al., 2018). Specific successional pathways are often poorly known and vary with environmental context (Arroyo-Rodríguez et al., 2017; Boukili and Chazdon, 2017; Lohbeck et al., 2013). Ongoing climate change and deforestation increase the influence of droughts and wildfires on tropical forests (Allen et al., 2010; Brando et al., 2014). Understanding how these factors affect the processes driving tropical tree assembly during forest recovery remains challenging (Boukili and Chazdon, 2017; Buzzard et al., 2015).

The assembly of tropical tree community depends on a range of ecological mechanisms spanning from deterministic, niche-based processes to neutral dynamics (Hubbell, 2001; Kraft et al., 2015). Niche theories predict that community assembly rely on species niche differences regarding biotic (i.e., species interactions) and abiotic (i.e., environmental filtering) constraints, while neutral theory assumes that community assembly depends only on stochastic immigration-drift dynamics (Adler et al., 2007). If environmental filtering drives community assembly, species establishment and survival should depend on how their ecological strategies match with local environmental conditions, and species occupying similar environments are expected to have similar, convergent strategies (Boukili and Chazdon, 2017; Kraft et al., 2015). Biotic interactions such as competition may entail either limiting similarity, i.e., divergence in species strategies (Abrams, 1983), or competitive dominance, i.e., the selection of species with similar competitive strategies and the exclusion of other species (Mayfield and Levine, 2010). Alternatively, without the influence of deterministic drivers, tree community assembly would rely on the sole influence of stochastic processes as predicted by the neutral theory, so that the composition of communities should be random with respect to species strategies (Swenson et al., 2012).

The successional pathway of tropical forest generally begins with the establishment of fast-growing, light-demanding pioneer species with acquisitive strategies (Chazdon, 2008; Lohbeck et al., 2013; Swenson et al., 2012). Nonetheless, the influence of strong abiotic constraints (e.g., dryer conditions) at the beginning of the succession can entail environmental filtering, favoring species with conservative and resistance abilities (Buzzard et al., 2015; Fang et al., 2019; Letcher et al., 2012; Lohbeck et al., 2013). As succession progresses, the number and size of individuals increase and canopy closes, which causes an increase in the influence of the light limitation on species recruitment (Buzzard et al., 2015; Chazdon, 2008; Pacala et al., 1996). As more and more species establish during succession, biotic interactions are expected to play an increasing role in community assembly, entailing either limiting similarity (Letcher et al., 2012; Lohbeck et al., 2014) or competitive dominance (Buzzard et al., 2015). If species strategies are sufficiently conserved across phylogenies (i.e., phylogenetically related species tend to have more similar strategies than less related species), niche-based assembly processes should likewise entail patterns of convergence or divergence in phylogenetic composition of communities (Cavender-Bares et al., 2009; Mayfield and Levine, 2010; Swenson et al., 2012; Swenson, 2013; Webb et al., 2002). Hence, investigating spatial and temporal changes in community phylogenetic structure can provide additional information about deterministic processes driving community assembly (Mouquet et al., 2012; Swenson et al., 2012; Swenson, 2013). Nonetheless, environmental filtering and competition can act together in shaping community composition, which can blur their respective signatures (Bernard-Verdier et al., 2012; Götzenberger et al., 2016). Processes driving forest succession can also act differently depending on tree size. Notably, some studies suggested that small trees may undergo stronger environmental filtering than older and larger canopy trees (Baldeck et al., 2013; Fang et al., 2019; Lasky et al., 2015). In addition, rapid shifts in constraining factors during forest succession should entail rapid changes in community composition, while compositional changes should be slower in late-successionnal communities where the environment is relatively constant (Swenson et al., 2012).

Forest-savanna landscapes are characterized by highly contrasting environments with sharp spatial transitions (Charles-Dominique et al., 2018; Gignoux et al., 2016; Ibanez et al., 2013a). In anthropogenic savannas, frequent anthropogenic fires maintains the coexistence of a flammable grass layer with an open-canopy tree layer (Hoffmann et al., 2009; Staver et al., 2011). In such environments, the establishment and survival of forest trees is strongly limited by fire (Cardoso et al., 2016; Hoffmann et al., 2009, 2012a, 2012b), while light is generally not a limiting factor (Charles-Dominique et al., 2018; Geiger et al., 2011). However, prolonged fire-free periods (i.e., 10–15 years) allow forest saplings to establish in savanna (Geiger et al., 2011; Hoffmann et al., 2012a). The influence of drought, which is amplified by low tree cover, can also limit the establishment of forest species in savanna (Hoffmann et al., 2012b; Ibanez et al., 2013a). In addition, soil nutrient availability is generally lower in savannas than in forest areas (Silva et al., 2013). On the other hand, in forest interior, closed canopy provides wetter conditions that prevent fire progression and light availability becomes more limiting for species recruitment (Hoffmann et al., 2012b; Ibanez et al., 2013a). The boundary between these alternative ecosystem states is generally constituted by a distinct forest edge (Geiger et al., 2011; Ibanez et al., 2013a). Such environmental contrasts entail striking differences in strategies between savanna and forest species (Charles-Dominique et al., 2018; Hoffmann et al., 2012a; Maracahipes et al., 2018; Ratnam et al., 2011). Notably, the recruitment of light-demanding savanna species is limited by shading under a closed canopy, which can drive a progressive decline of remnant populations of savanna trees in regenerating forests (Charles-Dominique et al., 2018; Geiger et al., 2011). However, few studies have explicitly addressed how the environmental context associated with anthropogenic savannas influences the pathway of tropical forest succession according to species strategies (Cardoso et al., 2021).

Ecological strategies can be characterized based on species functional traits (Violle et al., 2007). Bark has a major functional role in protecting stem against external damage from fire, as well as various factors as drought and herbivory (Pausas, 2015; Rosell, 2016). Notably, thicker bark is associated with higher survival rate in fire-prone areas (Charles-Dominique et al., 2017; Hoffmann et al., 2012a; Silva and Batalha, 2010). The leaf economic spectrum is a key functional dimension opposing acquisitive, fast-growing strategies based on low-cost, thin leaves with high photosynthetic rates, to conservative strategies with thick and long-lived leaves, more adapted to environments with low water and nutrient availability (Baraloto et al., 2010; Moles, 2018; Wright et al., 2004). In fire-prone landscapes, thicker and denser leaves also reduce leaf flammability (Pausas et al., 2017). Leaf area determines species transpiration surface and is negatively related with drought resistance (Wright et al., 2017). Furthermore, higher leaf area increases species capacity to intercept light, conferring greater competitive ability in light-limiting conditions (Díaz et al., 2016; Moles, 2018; Pierce et al., 2013). Another key functional trait of trees is wood density. Light wood is associated with high hydraulic efficiency, fast growth, but high mortality rate, while dense wood is associated with high drought resistance and mechanical support, slow growth, and low mortality rate (Chave et al., 2009). Finally, plant architecture determines light capture strategy, through horizontal and/or vertical space exploration, and should be related to competitive ability and shade tolerance (Ford, 2014; Küppers, 1989; Poorter et al., 2006). Specifically, vertical growth allows species to rapidly reach the canopy and have greater access to light, while lateral growth optimizes light capture under low light availability by both avoiding self-shading and increasing shading for adjacent competitors (Charles-Dominique et al., 2018; Gignoux et al., 2016; Millet et al., 1999). In savannas, trees have more space and light is less limiting, so that savanna species are generally smaller and invest less in vertical exploration than forest species (Charles-Dominique et al., 2018; Gignoux et al., 2016). However, investing in vertical growth allows trees to quickly produce leaves and buds higher than flame height, and thus escape from fire injuries (Gignoux et al., 2016). Yet, although the influence of species architecture on their ability to compete and survive has been suggested since three decades (Givnish, 1988; Küppers, 1989), little is known about its role regarding community-level assembly processes (Charles-Dominique et al., 2018; Ford, 2014; Subedi et al., 2019; Verbeeck et al., 2019).

In this study, we investigated the patterns of spatial (i.e., edge to interior) and temporal (i.e., 8 years interval) variations in the functional and phylogenetic composition of tree communities along a savanna-to-forest transition in New Caledonia. We focused on key functional traits representing drought and fire resistance, resource-use and light capture strategies to characterize the successional pathways leading to forest expansion in savanna. In light of these patterns, we inferred the processes that are likely to drive forest recovery and succession in anthropogenic savannas. Our general hypothesis was that the prevailing drivers of community assembly shift from environmental filtering related to fire and drought at the forest edge, where post-fire succession starts, to increasing light limitation in forest interior communities. Using a null model approach, we addressed several expectations in line with this hypothesis. We expected spatial trait patterns to reflect a shift from fire- and drought-resistance, as well as light-demanding strategies in young edge communities, to shade-tolerance strategies in forest interior. If constraints generated by environmental filtering are relaxed during succession, we should observe an increase in functional and phylogenetic diversity from the edge to forest interior. Then, if succession progressed during the studied time interval, we expected temporal change in community mean trait values to reflect a successional pathway similar to the one revealed by spatial patterns. Finally, we expected temporal change in taxonomic, functional and phylogenetic composition to be faster in younger edge communities than in forest interior, where recruitment limitation should be stronger and environmental conditions should be more stable. Because different tree size classes may undergo different constraints during succession, we evaluated if the observed spatial and temporal changes in community composition differed between small and large trees. As areas of recent forest recolonization included remnant savanna trees (Melaleuca quinquenervia, Myrtaceae), we also assessed whether the signature of changing assembly processes relied on the traits and demography of savanna species or if they reflect congruent compositional shifts among forest species.