Long-term droughts may drive drier tropical forests towards increased functional, taxonomic and phylogenetic homogeneity

Tropical ecosystems adapted to high water availability may be highly impacted by climatic changes that increase soil and atmospheric moisture deficits. Many tropical regions are experiencing significant changes in climatic conditions, which may induce strong shifts in taxonomic, functional and phylogenetic diversity of forest communities. However, it remains unclear if and to what extent tropical forests are shifting in these facets of diversity along climatic gradients in response to climate change. Here, we show that changes in climate affected all three facets of diversity in West Africa in recent decades. Taxonomic and functional diversity increased in wetter forests but tended to decrease in forests with drier climate. Phylogenetic diversity showed a large decrease along a wet-dry climatic gradient. Notably, we find that all three facets of diversity tended to be higher in wetter forests. Drier forests showed functional, taxonomic and phylogenetic homogenization. Understanding how different facets of diversity respond to a changing environment across climatic gradients is essential for effective long-term conservation of tropical forest ecosystems.


Supplementary discussion on forest community dynamics
To characterise the dynamics at the community level we calculated the changes in the species' basal area, temporal diversity metrics, such as the community turnover, species appearances and disappearances and mean rank shifts, and the variance ratio community stability metric (1,2) . All community dynamics analyses were carried out using the "codyn" package in R platform (v3.4.1; http://cran.r-project.org) (3) .
There were large changes in species basal area along the climatic gradient and across time (Supplementary whereas Strombosia pustulata and Antiaris toxicaria showed the strongest total basal area decreases (-3.39 and -3.09 m 2 respectively). The species with larger increases and decreases in basal area did not appear clustered in specific locations of the phylogenetic tree that contained all species present in the studied plots (Supplementary Figure 4a). However, at the plot level (phylogenetic tree per plot), the groups/clades changing the most in basal area became more apparent but a strong variation between census plots was still observed (Supplementary The overall community dynamics with species turnover and species appearances and disappearances, mean rank shifts and variance ratio did not differ significantly (P-val >0.05) along the climatic gradient. Moreover, most sampling plots (12 of them) have increased their basal area with an annual rate of between 0.005 (BBR-14) and 0.36 m 2 (KDE-02) and only eight showed annual rate decreases of between -0.03 (DRA-04) to -0.21 m 2 (FUR-07).
Such changes in basal area however where not significantly related to changes in functional (R 2 =-0.01, P-val=0.94), taxonomic (R 2 =0.28, P= 0.21) or phylogenetic diversity (R 2 =-0.17, P= 0.46). The ratio may decrease as to prevent water loss. Stronger changes in more water-limited forests and thus decreases in this trait with increases in the abundance of deciduous species. (4,5) Kp Water transport capacity; Index of hydraulic efficiency and possible trade off with hydraulic safety.

Supplementary
Reduction as a result of acclimation to drier environments. Possibly with stronger changes and being higher in communities with usually wet conditions. (6) VLF Related to water transport capacity of stem. Larger values represent higher possible water conductivity at the partial cost of lower mechanical support. May represent hydraulic efficiency.
Expected reduction to decrease cavitation given lack of water resources under drought. (7,8) VD Ensure sufficient water supply from the roots to the leaves.
Deciduous species may show wider vessel diameter than evergreen as they avoid dry season cavitation risk. (6,9,10) Larger vessels diameter is associated with species with rapid water transport to support high photosynthetic rates. Wider vessels may be more susceptible to implosion and have increased risk to embolism and cavitation.
ρV Water transport capacity. Fewer but larger vessels (lower density) may facilitate water transport.
Expected increases in dry environments in association with decrease of vessel size, to maintain water flow and lower cavitation risk (8) AreaL Relevant as a main light capture mechanism.
Higher leaf area could result in more leaf transpiration and thus water loss under a drying climate.
Under a drying climate it may increase in deciduous species and expect decreases in evergreens as to limit water loss by transpiration and for increasing cooling. (11,12) SLA Important for photosynthetic capacity, light capture, water loss, net assimilation rate, leaf life span.
May increase if acquisitive species, e.g. deciduous species, become more abundant with a drying climate. (13)(14)(15) NL Essential for metabolic reactions involved in light capture, photosynthetic capacity and growth. Restricted availabilities limit plant carbon acquisition and growth Drought effects may be compensated if nitrogen fixing species (mainly Fabaceae) become more abundant. May be more dependent on soil conditions than on climate. (16)(17)(18) PL Needed nutrient for metabolic reactions that include light capture, related to photosynthetic capacity and growth. Lack of P may limit carbon acquisition and growth Decreases under a drying climate and possible not strong effect under short term droughts or in wet forests. May be more dependent on soil conditions than on climate. (16)(17)(18) ThicknessL Trade-off between decreasing water transpiration at the expense of higher construction investment. May decrease under a drying climate as a result of increasing in deciduous species which may tend to have thinner leaves.
It is expected that thicker leaves become more common under larger water deficits for evergreen species but may decreases for acquisitive deciduous species. (11,12) Amax Maximum CO2 assimilation. Index of leaf photosynthetic capacity.
Higher for species with fast resources turnover, e.g. deciduous vs evergreens. Increase with abundance of such species. (6,19,20) Asat Saturated photosynthetic rate. Index of leaf photosynthetic capacity.
Declines with higher temperatures and lower precipitation. However, Asat is also dependent on CO2 fertilization and N and P levels. (6,19,20) Heightmax Proxy of species position in the vertical light gradient in the forest canopy, with taller species accessing higher light levels than shorter species also given their usually wider crowns.
Taller species that can access more light resources may increase if they can also avoid cavitation risks and have a fast energy turnover as is the case for deciduous species. Otherwise, shorter species with slow growing patterns and with low vessel cavitation risks, e.g. given periods of drought, may become more dominant. (21,22) WD Relevant for mechanical strengths, stem vulnerability to xylem cavitation.
Expected to be higher in areas with lower water resources, and thus increase with a drying climate. (15,(22)(23)(24) Phenology -Deciduous/ Evergreens Deciduous species have low investment in leaf construction, rapid leaf turnover and high photosynthetic capacity. Reduction of water transpiration and avoidance of xylem cavitation are important for their success -drought avoiders.
With a drying climate increasing are expected as such species may be better adapted to long and intense periods of drought in comparison to evergreen species, which may tend to decrease in abundance (25)(26)(27) Evergreens have high investment in leaf construction, slow leaf turnover, lower photosynthetic capacity -drought resistant

Guilds
Certain guilds have been shown to be better adapted to droughts, e.g. NPLD than others, e.g. SB With a drying climate, guilds as NPLD and Pioneers may become more abundant specially if the LA:SA ratio decreases which may negatively affect the abundance of other guilds as Shade Bearers. (27) Nitrogen Fixers Higher productivity given N uptake. Higher leaf nitrogen content and photosynthetic capacity than non-nitrogen fixers. Likely with high rates of photosynthesis over wet periods and accumulation of carbon for foliage production after drought.
Expected increase as such species may have access to limiting resources important for photosynthesis such as nitrogen, which may confer them advantages in a drying climate (25,28) LA

Supplementary Table 4. Specification of models fitted in the R statistical environment for each one of the Diversity Metrics (FDis: functional diversity, Simpson: taxonomic diversity and MPD: phylogenetic diversity).
The diversity metric was fitted as a response to climatic and soil drivers. For all three diversity metrics the soil PC axes were fitted as quadratic terms. For the MPD models the time between censuses was used as an extra covariate in order to account for its possible role in determining changes in MPD across the different vegetation plots (given the weack but significant correlation between changes in MPD and the years between censuses). Amax: CO2-saturated assimilation rates; Asat: light-saturated photosynthetic rates; Heightmax: maximum adult size; WD: wood density.

Supplementary Table 6. Model selection table based on Leave One Out crossvalidation information criterion (LOOIC).
The best model, the one with the lowest LOOIC and highest ELPD, for each diversity metric is highlighted in grey. The best models are used in subsequent analysis. For more specific information on the terms included in each model see Table S2. 4.5 3.24 LOOIC: Leave one out information criterion; ELPD LOO: Leave one out expected log predicted density; ELPD diff: difference in expected log predicted density; P LOO: Effective number of parameters; SE LOOIC: standard error of LOOIC; SE ELPD LOO: Standard Error of ELPD; SE P LOO: Standard error of P. Table S7. Linear regression results for the second best models, based on the leave one out cross-validation information criterion (LOOIC), explaining the functional (FDis), taxonomic (Simpson) and phylogenetic (MPD) diversity changes as a function of climatic and soil drivers. Several different models were fitted (see Supplementary Table 4