5.1. Water availability and soil nutrients
Climate changes are strongly related to changes in rainfall patterns worldwide, involving reductions in the total amount of annual precipitation and/or increase in the dry season duration (Grossiord et al. 2016). Furthermore, at local scales, soil water availability is highly variable along topographic gradients, with wetter valleys and drier ridges (Fig. 3, Gibbons & Newbery 2002). Both changes in soil moisture and reduced precipitation strongly drive changes in physiological traits (Pezzola et al. 2017). For instance, drought-induced water stress decreases photosynthesis rate, which leads to slow and reduced plant growth (McDowell et al. 2008). In addition, low water availability decreases the physiological mechanisms (secondary metabolites) related to plant defense and favors the reproduction of insects (Mattson and Haack 1987, McDowell et al. 2011), increasing woody species vulnerability to death (Das et al. 2016). Water availability also plays a vital role in forest recovery, increasing the growth rate of plant tissues during post-drought periods (Poorter et al. 2016, Álvarez-Yépiz et al. 2018). Due to its strong spatial and temporal variations, water availability is the most important driver of resistance and recovery rates.
Soil nutrients also play an essential role in plant communities distributed across topographic gradients (Fig. 3, Guan et al. 2015). Soil nutrients are key factors of photosynthesis, which is an essential mechanism of plant survival and growth (Fatichi et al. 2014). Nutrient stoichiometry can change across topographic gradients due to the unidirectional fluxes from ridges to valleys and of the consequent loss of nutrients in the ridges (Werner and Homeier 2015). Therefore, nutrient limitation can lead to high nutrient competition in ridges (Werner and Homeier 2015). Nitrogen (N), phosphorus (P) and carbon (C) are the main elements required to plant growth and survival, and change in their ratios can have strong consequences for tropical forest communities (Zhang et al., 2012, 2017). For instance, P of tropical forest soil is positively correlated with growth rate and negatively related with mortality rate (Soong et al. 2020). Tropical plants in valleys may require large amounts of leaf N to deal with the intense shading of forest understory (Torres-Leite et al. 2019). Droughts can alter N and P cycles in ecosystems, either directly through changes in N mineralization and P sorption (Mariotte et al. 2017) or indirectly through changes in plant nutrient uptake and growth (Mariotte et al. 2020). Changes in litter N, P and C content alter decomposition rates because decomposers require nutrients from either litter or soil for their functioning (Gartner and Cardon 2004). Deficiency in plant-available nutrients in ridge is enhanced through a positive feedback driven by poor litter decomposability (Werner and Homeier 2015). Furthermore, more frequent droughts slow down litter decomposition and reduce plant-available nutrients (da Silva et al. 2020).
Microclimate variation across topographic gradients
Vegetation structure (e.g., height, stem density, and canopy biomass) and local topography (e.g., elevation, slope and aspect) strongly influence the microclimate (Hardwick et al. 2015). Understory and canopy species are under different microclimatic conditions due to the vertical structuration in tropical forests. For instance, canopy cover of larger trees regulate the irradiance and light intensity entering the understory layer (Nepstad et al. 2002, Wright et al. 2010). Thus, the light availability is more variable and limiting for understory species than for large trees (Wright et al. 2010). The high drought-induced mortality of large trees can change the microclimate of the understory (Redmond et al. 2018, Zellweger et al. 2020). Small forest fragments have lower abundance of large trees due to high edge effects (Dantas de Paula et al. 2011) and lack suitable microclimatic sites for the persistence of species during drought periods (Hardwick et al. 2015, Laurance et al. 2018). During a water stress period, the low leaf area of large deciduous trees lead to high light availability in the understory (Smith et al. 2019). In turn, the increase in light availability induces high rates of photosynthesis in the understory, thus affecting species adapted to shaded conditions (Guan et al. 2015).
Other microclimate variables, such as air temperature and humidity, are controlled by light availability that reaches the understory stratum. For example, in the valleys with dense forest canopy, air humidity is higher and temperature is lower than ridges with open forest canopy (Fig. 3, Jucker, Hardwick, et al. 2018). The high humidity and low temperature in valley can protect plant species during drought periods. However, the opening of forest canopies during droughts due to leaf loss can affect species adapted to low temperatures and high air humidity (Smith et al. 2019). The high transpiration rates are strongly related to high air moisture in the understory layer (Hardwick et al. 2015). In addition, ridges can show different microclimate conditions depending on the topographic aspect. For example, west facing ridges are warmer due to higher exposure to afternoon sun than east facing ridges that are exposed to morning sun (Stephenson 1990). Therefore, microclimate variables, controlled by topography and vegetation structure, are strongly related to plant growth and mortality, especially for understory species.
Biotic factors
Drought resilience of dominant and low abundance species
Plant tropical communities are complex systems due to high diversity and large differences in species abundances. The most basic classification of species in a plant community is based on abundance patterns and it separates species into dominants, subordinates and transients (Whittaker 1965, Grime 1998). In a plant community, few species are classified as dominant species, which are the most abundant, and account for a higher proportion of the overall biomass of the community. On the other hand, low abundance species (subordinates and transients) represent lower amount of biomass, but are the main determinants of plant diversity (Whittaker 1965, Grime 1998, Mariotte 2014). Dominant and low abundance species differ in their functional traits and play different roles in the ecosystem (Mariotte 2014). In general, dominant species are competitively superior (Mariotte 2014) and respond to environmental filtering, while low abundance species respond to niche differentiation (Maire et al. 2012). Low abundance species can promote the diversity of climbing plants (Garbin et al. 2012) and affect ecosystem functioning (Mariotte et al. 2015). Dominant species play an important role in structuring the species distributions (Wei et al. 2020) due to its homogenous distribution pattern (Mariotte 2014). For example, in a topographic gradient, they can occur in valleys and ridges (Hollunder et al. 2014). On the other hand, as low abundance species have a more aggregated spatial distribution (Mariotte 2014, Garbin et al. 2016), it is expected that they occur in specific habitats and not across a whole topographic gradient (Hollunder et al. 2014). The response to drought depends on species and their spatial distribution and thus, taking into account both habitats (valleys vs. ridges) and species groups (dominant vs. low abundance species) can improve our understanding of the processes that drive the response of tropical forests to severe droughts.
Low abundance species are less numerous, making populations of these species more vulnerable to local extinction induced by climate change (Greenwood et al. 2017). Furthermore, populations of low abundance species that occur with few individuals and with an aggregated spatial pattern (Mariotte 2014) can be even more reduced due to habitat loss. On the other hand, dominant species are expected to be more resistant to local extinction due to their higher number of individuals (Greenwood et al. 2017). Therefore, locally, and perhaps even globally, species diversity can decrease under the future climate scenarios due to the role of low abundance species as main determinants of plant diversity. Dominant species play important roles on ecosystem functioning because of their large amount of biomass (Grime 1998) and thus, they can act as a biotic filter in the establishment and survival of low abundance species (Khalil et al. 2019). The mortality of dominant species during droughts can also affect the performance of low abundance species. Therefore, understanding how dominant and low abundance species respond to drought events will highlight their roles in promoting resilience of tropical plant communities.
Use of traits to unravel drought resilience mechanisms
A functional trait is any feature which impacts fitness indirectly via effects on growth, reproduction and survival at individual and species level (Diaz and Cabido 2001, Violle et al. 2007). Traits are classified based on responses to the environment and/or common effects on ecosystem processes (Lavorel and Garnier 2002). There are traits related to responses to environmental variables, such as resources and disturbances (i.e., response traits), and traits that determine the effects of plants on ecosystem functions (i.e., effect traits), such as biogeochemical cycling and disturbance resistance (Lavorel and Garnier 2002). Furthermore, species may exhibit intraspecific variation in functional traits, which can represent an important strategy to resist and recover from drought impacts (Hof et al. 2011). For instance, small forest fragments sustain smaller populations with low phenotypic variability in traits, making them more sensitive to climate change effects (Hof et al. 2011). Trait-based approaches can help to understand the physiological mechanisms of drought tolerance in different species, as well as how forests will respond to future climate scenarios. Thus, ecologists have to include key traits directly and mechanistically relevant to plant survival and growth. For instance, carbon sequestration, water use efficiency and photosynthesis rate are mechanisms related to plant growth and mortality that can be assessed by measuring plant traits.
Traits can be measured at different levels, from tissue (e.g. stomatal and vein density), organ (e.g., specific leaf area [SLA], leaf dry matter content [LDMC], woody density [WD]) to whole-plant level (e.g., height, root to shoot ratio) (Pérez-Harguindeguy et al. 2013). Plant ecologists have been using trait-based approaches in order to understand mechanisms that explain resilience mechanisms. On one hand, at the community level, functional diversity can enhance resource use and increase the potential for facilitative interactions, which decrease the negative impacts of droughts (Gazol and Camarero 2016). On the other hand, at smaller levels, individual and species are able to change their traits to tolerate droughts. For example, trees can increase chlorophyll concentration in their leaves during El Niño periods (Nunes et al. 2019).
Trait combinations have the potential to explain the growth-mortality and resistance-recovery trade-offs. Traits that are linked to resource capture (e.g., leaf area), photosynthetic capacity (e.g.,specific leaf area and leaf nitrogen content) and nutrient and water uptake (e.g., root length and diameter) generally have positive relationships with growth and negative relationships with mortality (Pérez-Harguindeguy et al. 2013). In turn, traits that are related to structural safety (e.g., wood density) or longevity (e.g.,leaf dry-matter content and leaf thickness) have negative relationships with growth, but are often positively related to survival (Poorter and Bongers 2006). However, the growth–trait relationships can exhibit inconsistences among species group, biomes, and spatial scales. For instance, although WD is widely used to predict drought impacts, it has been suggested that this trait does not provide a mechanistic understanding of drought-induced mortality (O’Brien et al. 2017). Recently, it was found that low wood density is a strong predictor of mortality for angiosperms, but not for gymnosperms (Anderegg et al. 2016). Furthermore, high wood density can be associated to high mortality rate caused by hydraulic failure (Hoffmann et al. 2011). Therefore, the use of specific key traits directly related to growth and mortality is essential to investigate physiological mechanisms of resilience.
The importance of tree size
Tree size is one of the most used traits to understand drought effects on plants (Meakem et al., 2017; Prado-Junior et al., 2017; Shenkin et al., 2018). Tree size is an effect trait, due to its key role in controlling light availability and heterogeneity, and in influencing the structure of understory species communities (Sercu et al. 2017). Tree size is also a response trait related to light competition due to the greater access to light that larger trees can have when compared to understory plants (Laughlin 2014). Plant size can be negatively related to stand density across topographic gradients, from valleys with low stem density and bigger trees (in height and diameter) to ridges with high stem density and smaller trees (Werner and Homeier 2015). This trait can be helpful to understand which trees are more vulnerable to mortality, but its use in drought studies is still limited in tropical forest.
Most studies showed that larger trees are more sensitive to drought effects due to their higher demands for water comparing to smaller trees, which are expected to be more resistant (Phillips et al. 2010, Moser et al. 2014). Nevertheless, small trees can face more water limitation due to their shallower roots (Gibbons and Newbery 2002) and thus they also show higher mortality rates (Rocha et al. 2020) and a more accentuated growth-mortality trade-off pattern than larger trees (Zhu et al. 2017). Trees with a single stem may be more vulnerable to drought effects than shrubs with multiple stems, which can leave one or more viable as an insurance to survive (Tanentzap et al. 2012). In addition, older and taller trees can show high resistance to precipitation variations (Giardina et al. 2018). These findings suggest that small and big trees are more vulnerable to droughts and trees with intermediate sizes (e.g. understory individuals) are more resistant to drought events. Indeed, tree mortality may increase, decrease, or show a U-shaped curve with increasing tree size (Zhu et al. 2017). Although tree size is a common trait used in drought studies, we do not know clearly if there is a size-dependency in mortality.
The importance of hydraulic traits
Traits can be classified according to their responses to different factors (Li et al. 2015): leaf economic traits (e.g., SLA and WD) and hydraulic traits (e.g., stomata density and water-use efficiency). Leaf economic traits are related to light use and carbon sequestration while hydraulic traits are related to gas exchange and water transport capacity (Li et al. 2015). Variation in hydraulic traits is independent of variation in economic traits in tropical forests (Powell et al. 2017), but hydraulic traits are strongly influenced by plant size and water availability (Liu et al. 2019). Plants can reduce water loss by shedding their leaves or closing their stomata to maintain plant water potential and avoid or reduce the risk of xylem cavitation (Vitória et al. 2019). The sensitivity of trees to drought is best predicted by hydraulic traits (Anderegg et al. 2016, Li et al. 2020) and forests with higher diversity in hydraulic traits are more resistant to drought impacts (Anderegg et al. 2018). Hydraulic failure was the main mechanism explaining tree mortality during the 2015/2016 drought, while wood density, SLA, tree size and foliar nutrients were poorly correlated with drought vulnerability in a seasonally dry tropical forest in Costa Rica (Powers et al. 2020). Hydraulic failure occurs when the water transport to the canopy is decreased, resulting in desiccation and death of plant tissues (Hoffmann et al. 2011). In addition, tropical rainforests showed high water use efficiency to deal with the negative impacts of 2015/2016 drought (Nunes et al. 2019). Therefore, the use of hydraulic traits and leaf economic traits, as well as their relationships, can help to understand the cause of mortality and reduced growth.
Deciduousness is a phenological but also a hydraulic trait related to drought avoidance strategy for trees from dry tropical trees (Wolfe et al. 2016), but it is rarely used in drought studies. Deciduous species lose their leaves during drought periods to reduce transpiration and maintain water balance (Wolfe et al. 2016). When the water is available again, these species show high stomatal conductance and photosynthesis, both important mechanisms to grow and increase canopy cover (Lambers et al. 2008). The duration of the deciduousness strongly affect carbon sequestration by tropical deciduous and it has the potential to indicate the magnitude of the drought (Singh and Kushwaha 2016). Deciduousness can explain why the recovery is faster in ridges and in dry forests. However, most of the studies investigating drought response mechanism focused on leaf economic traits, and not on hydraulic traits. Repeated droughts can act as filter selecting tolerant species (Aguirre-Gutiérrez et al. 2020) and increasing the abundance of drought-tolerant deciduous species (Fauset et al. 2012). Focusing on traits related to water use and balance is a way to fill most of the remaining mechanistic gaps regarding drought effects (Martinez-Vilalta et al. 2019).
Conclusion
This literature review summarizes the role of abiotic and biotic factors mediating drought resilience in tropical forests. Our synthesis highlights that the regional climate conditions shape the forest types, and the topography controls biotic and abiotic factors at a local scale in different forest types. Both dry tropical forests and ridges are more sensitive to droughts than moist tropical forest and valleys but the mechanisms explaining these patterns remain unknown. Field studies are essential to identify local and regional differences in drought resilience and to predict the future of tropical forests. This literature review also highlights the main gaps in drought resilience research, which are: 1) to identify mechanisms explaining both the growth-mortality and resistance-recovery trade-offs; 2) to understand how different functional groups (dominant vs. subordinate species, shade-species vs. sun-species, trees vs. shrubs) deal with droughts; 3) to describe the physiological mechanisms explaining the forest resilience of different habitats (valley versus ridge), forest types (moist versus dry forests) and successional stages (secondary versus primary forests); 4) to understand how droughts change the microclimate in different habitats.