Aridity and forest age mediate landscape scale patterns of tropical forest resistance to cyclonic storms

1 INTRODUCTION

Climate change is expected to increase the intensity of cyclonic storms, hereafter hurricanes, due to rising ocean heat energy (Seneviratne et al., 2023). Large areas of tropical forests experience hurricane disturbances, which strongly influences forest function and structure (Lugo, 2008). The high biological and functional diversity of these ecosystems poses challenges for predicting their response to increased hurricane intensity (Lin et al., 2020; McLaren et al., 2019; Uriarte et al., 2019). It is expected that forest resistance to hurricane disturbance will vary based on factors, such as forest age, water availability, structure, topography, land-use history and species composition (Feng et al., 2020; Uriarte et al., 2009). This complexity highlights the need for a unifying framework to study forest resistance to hurricanes across broad environmental gradients.

Classic ecological theory suggests that, as ecosystems age, their attributes will favour slower energy exchange with greater ecosystem homeostasis (Odum, 1969). This provides a framework that links ecological succession with ecosystem stability in the face of perturbations (Poorter et al., 2023). In the context of hurricane disturbances, this can be defined as an increase in forest resistance with forest age. However, for this to hold true, the functional properties of forests conferring resistance to hurricane disturbances should covary along with ecosystem development, which can be context dependent (Vitousek & Reiners, 1975; Zak, 2014). It is crucial, then, to evaluate whether the factors influencing tropical forest function along gradients of succession mediate resistance to hurricanes.

During hurricane disturbances, tree mortality primarily occurs through uprooting and stem breakage (Lugo, 2008), especially among tall tree species with low wood density (WD; Curran et al., 2008; Helmer et al., 2023b; Ibanez et al., 2024; Ogle et al., 2006; Taylor et al., 2023; Uriarte et al., 2019; Zimmerman et al., 1994). Community-level patterns of plant traits, such as WD or tree height, vary across gradients of water availability and forest age (Bruelheide et al., 2018). These traits link to species' growing strategies, where conservative traits such as high WD and short stature are found in ‘slow-safe’ species and the opposite in ‘fast-risky’ species (Díaz et al., 2016; Reich, 2014). For instance, in arid regions, plant species tend to show shorter stature, deep roots, high WD and xylem resistant to drought stress (Olson et al., 2018; Tumber-Dávila et al., 2022; Vargas G et al., 2022), representing a physiological strategy associated with higher survival during drought stress (Anderegg et al., 2016). Conversely, low WD species typically dominate young stands at the wet end of the aridity spectrum while high WD species dominate young stands at the dry end (Poorter et al., 2019; Figure 1a).

FIGURE 1

Open in figure viewerPowerPoint

Hurricanes act as a major disturbance agent in tropical forests. (a) The relation between CWM-WD and forest stand age for three forest types associated with climatic conditions for a given life zone in Puerto Rico, Vieques and Culebra islands. (b) The hypothesized prediction that hurricanes will reinitiate the functional trajectory, but the resulting functional composition will depend on the available species pool given the effect of environmental drivers on forest age (panel a). In (b), points represent CWM-WD through time, the grey-shaded area represents the 95% confidence interval of the CWM-WD, the straight blue line is the mean trendline of CWM-WD against time, the vertical dashed line represents a hurricane event and the three dashed lines represent hypothesized functional trajectories. (c) Basal area mortality rates (MR), measured as the percentage basal area loss during the census interval before and after Hurricane Maria for 180 Forest Inventory Analysis plots in Puerto Rico, Vieques and Culebra.

The known variation in plant traits along aridity gradients suggests that many of the traits associated with drought tolerance, such as dense wood or deep roots, also enhance forest resistance to hurricanes. However, whether this assumption holds true across broad geographic scales remains an open question. Recent remote sensing analyses suggest that large reductions in canopy greenness and increases in non-photosynthetic vegetation may be associated with tall forest canopies, hurricane exposure, old stand age or wetter forests (Feng et al., 2020; Hall et al., 2020; Leitold et al., 2022; Van Beusekom et al., 2018). However, only one study to date has performed a ground validation of hurricane disturbances using a small sample size (n < 30) in which canopy height was the most important forest characteristic explaining biomass loss during hurricanes (Hall et al., 2020). Conversely, a global data synthesis of 74 forest plots confirmed the importance of WD in mediating tree damage caused by hurricane disturbances (Ibanez et al., 2024), yet the plots were located only in three types of forests (tropical montane, lowland tropical and lowland subtropical rainforests). This evidence underscores the need to investigate the contribution of forest age and community-level functional traits in determining the severity of hurricane disturbances across climatic gradients.

Climate and forest type may also be an important factor mediating how hurricanes affect plant community composition. In wet forests, hurricanes cause an increase in the number of forest gaps (Lugo, 2008), potentially favouring fast-growing drought-vulnerable plant species (Alonso-Rodríguez et al., 2022; Smith-Martin et al., 2022). However, in tropical dry forests, biomass recovery after hurricanes mostly occurs through the re-sprouting of snapped trees, and large re-organizations of the plant community are rarely seen (Curran et al., 2008; Van Bloem et al., 2007). These lines of evidence suggest that the functional composition of plant communities can follow many alternative routes depending on environmental factors and the characteristics of surviving tree species (Figure 1b). Identifying how environmental gradients affect the rate at which community-level trait values change over time (e.g. trait velocities) can provide valuable information on whether plant communities become more resistant to future disturbances (Trugman et al., 2020). To our knowledge, no study has extensively explored community-level changes as a function of hurricane disturbances across broad gradients of forest stand age and aridity (Lin et al., 2020).

In this work, we evaluate key factors that influence the patterns of hurricane disturbance on forest ecosystems across broad environmental gradients. Specifically, we tested whether aridity, forest structure, stand age and functional composition explained the variability in basal area mortality rates, changes in canopy height and cover after Hurricanes Irma and Maria in the archipelago of Puerto Rico. We leveraged forest basal area mortality rates on 180 out of 338 permanent plots from U.S. Forest Inventory and Analysis (FIA) data with censuses before and after the hurricanes. We supplemented FIA data with functional traits for 410 tree species, along with remotely sensed measurements of canopy height, canopy cover and stand age. Given that wet and dry forests exhibit opposite trends in their wood density trajectories along succession (Lohbeck et al., 2013; Poorter et al., 2023), we hypothesized that tropical forest hurricane resistance patterns will converge in late successional stages along aridity gradients. Specifically, older wet forests will increase resistance due to the increasing abundance of high wood density species, while older dry forests will decrease resistance owing to the increasing abundance of low wood density species (Figure 1a). We expected these patterns to remain even after accounting for the differences in topographic slope and proximity to the eye of the storm (Feng et al., 2020). We further hypothesized that aridity would limit the increase in community-level trait values indicative of vulnerability to drought (i.e. low WD) following hurricane disturbance. Specifically, we asked: (1) How do climate and stand age mediate forest diversity, structure and community-level trait values? (2) How do forest structure, basal area mortality, community-level traits and trait velocities change after Hurricanes Irma and Maria? (3) Do climate and stand age determine the trajectory of trait velocities after the hurricanes? (4) Are hurricane-related changes in forest basal area mortality directly and indirectly mediated by community-level traits?