Abstract Background Lowland dipterocarp forests in Southeast Asia represent one of the world’s most significant biodiversity hotspots, yet they have experienced extensive fragmentation by rapid land-use change. Rubroshorea leprosula , a near-threatened and economically important dipterocarp, occurs widely across Peninsular Malaysia, Sumatra and Borneo, yet its evolutionary history and adaptive potential remain poorly understood. Understanding how and why genetic variation is structured across this geographic range is essential for predicting the resilience of tropical tree species, particularly where geographical isolation and heterogeneous environments may drive localized adaptation. Results Whole-genome resequencing of 194 individuals from 37 natural populations revealed moderate genome-wide diversity, with notable regional contrasts. Populations from Peninsular Malaysia and Sumatra exhibited higher heterozygosity and shared ancestral variation, while Bornean populations showed signals of recent demographic expansion. Population structure analyses identified two major genetic clusters corresponding to Western (Peninsular Malaysia + Sumatra) and Eastern (Borneo) lineages, with a small number of admixed individuals in northern Peninsular Malaysia suggesting historical gene flow. Demographic modelling supports divergence during the Mid- to Late-Pleistocene followed by low, asymmetric post-divergence gene flow, consistent with secondary contact during post-glacial forest reconnection across Sundaland. Integration of outlier detection ( F ST and pcadapt) with environmental association analyses (RDA and LFMM2) identified candidate loci associated with stress response, signalling and metabolic pathways, suggesting that environmental heterogeneity across the species’ range has contributed to adaptive genomic differentiation. Conclusions This study reveals strong west-east genetic divergence in R. leprosula and identifies climate-associated loci linked to adaptation across temperature, elevation and precipitation gradients. These genome-wide patterns clarify evolutionary processes and provide a foundation for climate-informed conservation management of threatened dipterocarp forests.

While habitat loss and hunting remain the major drivers of biodiversity declines, sublethal disturbances, such as human presence, recreation, and noise also impact wildlife. In response, wildlife often adjust their spatiotemporal behaviors. This study assesses how terrestrial wildlife responds to sublethal disturbances associated with sustainable logging activities. Using camera traps, we conducted a 2-year continuous survey of two logging compartments within an active, sustainably logged forest reserve in central Sabah, Malaysian Borneo. The survey encompassed periods before, during, and after logging activities, and we obtained daily felling data. Using generalized linear mixed models, we assessed the impact of logging on the spatiotemporal activity of 10 terrestrial mammal and galliform species, considering spatial factors like logging intensity and distance from logging activity, and temporal factors such as days since logging. Four species had a significant and strong response to logging intensity, with varied directions of effect. Only for one species, the response to logging intensity was mediated by time since logging. Though only descriptive, we found no evidence of diel activity shifts, and, for all other species detected during our survey, general patterns in the number of records across each logging period were mixed. Our results highlight the variable and species-specific responses to logging activity. At a local scale, many species exhibit resilience in their spatiotemporal activity patterns, but for affected species, the effects of logging appear to linger over time. The overall limited response could be due to the relatively low disturbance nature of reduced-impact logging.

Information on tropical Asian vertebrates has traditionally been sparse, particularly when it comes to cryptic species inhabiting the dense forests of the region. Vertebrate populations are declining globally due to land-use change and hunting, the latter frequently referred as "defaunation." This is especially true in tropical Asia where there is extensive land-use change and high human densities. Robust monitoring requires that large volumes of vertebrate population data be made available for use by the scientific and applied communities. Camera traps have emerged as an effective, non-invasive, widespread, and common approach to surveying vertebrates in their natural habitats. However, camera-derived datasets remain scattered across a wide array of sources, including published scientific literature, gray literature, and unpublished works, making it challenging for researchers to harness the full potential of cameras for ecology, conservation, and management. In response, we collated and standardized observations from 239 camera trap studies conducted in tropical Asia. There were 278,260 independent records of 371 distinct species, comprising 232 mammals, 132 birds, and seven reptiles. The total trapping effort accumulated in this data paper consisted of 876,606 trap nights, distributed among Indonesia, Singapore, Malaysia, Bhutan, Thailand, Myanmar, Cambodia, Laos, Vietnam, Nepal, and far eastern India. The relatively standardized deployment methods in the region provide a consistent, reliable, and rich count data set relative to other large-scale pressence-only data sets, such as the Global Biodiversity Information Facility (GBIF) or citizen science repositories (e.g., iNaturalist), and is thus most similar to eBird. To facilitate the use of these data, we also provide mammalian species trait information and 13 environmental covariates calculated at three spatial scales around the camera survey centroids (within 10-, 20-, and 30-km buffers). We will update the dataset to include broader coverage of temperate Asia and add newer surveys and covariates as they become available. This dataset unlocks immense opportunities for single-species ecological or conservation studies as well as applied ecology, community ecology, and macroecology investigations. The data are fully available to the public for utilization and research. Please cite this data paper when utilizing the data.

Pangolins are the most trafficked mammals in the world and are severely threatened by poaching the loss, degradation, and fragmentation of habitat. In Malaysian Borneo, conservation initiatives for the Sunda pangolin (Manis javanica) are hindered by a paucity of data on their distribution and population size. Using MaxEnt niche modelling and consolidated species location data, we projected the distribution of Sunda pangolins in Sabah. Additionally, we assessed the accessibility of their forest habitats to humans to understand potential threats. Our model indicated that, as of 2015, approximately half of Sabah’s land area (39,530km²) is suitable for pangolins, with 43% in protected forests, 38% in production forests, and 19% outside of these areas. Alarmingly, our data suggest that nearly all (91%) of these suitable habitats are relatively easily accessible to poachers. Our findings provide a state-level baseline understanding of Sunda pangolin distribution and assess potential threats in Sabah. These can inform short- and long-term conservation management plans for pangolin to safeguard this critically endangered species.

AbstractGlobal change research has largely focused on the effects of drought on forest dynamics while the importance of excessive rainfall that can cause waterlogged soils has largely been assessed in riparian zones or seasonally flooded sites. However, increased rainfall may also cause decreased growth and survival of tree species in lowland aseasonal tropical forests due to increased risk from potentially more extensive and frequent waterlogged soils. We used a Bayesian modelling approach on a tree dynamics dataset from 2004 to 2017 to test the concomitant effects of rainfall excess and deficit and dry period length on tree growth and survival across a network of experimentally planted trees in a primary aseasonal forest in Malaysia. Growth declined in 48% of the species and survival decreased in 92% of the species during periods of high rainfall while as little as 4% of species had decreased growth or survival with drought and long dry periods. Climate change is projected to cause more frequent and severe rainfall deficit and excess, and our results suggest increased rainfall may have stronger negative effects on aseasonal tropical forests than that of severe drought.

Tropical secondary forests are widespread in the anthropogenically modified landscapes. Because tropical secondary forests were known to show a steady rate of above-ground biomass (AGB) recovery, they were expected to play a crucial role in carbon sequestration. However, in a preliminary survey of logged-over forests in Borneo, some patches covered with ferns and vines do not seem to recover steadily despite the fact that several decades have passed since the last logging. The presence and abundance of fern/vine thickets may affect the AGB recovery, but few studies have investigated the effects of the thickets. We established a total of seventeen 20-m radius circular plots in logged-over forests, Sabah, Malaysia, with a varying degree of fern/vine coverage, and tested the hypothesis that the greater fern/vine coverage would retard the AGB recovery. The net AGB accumulation rate from 2014 to 2019 was lower in the forests with a higher fern/vine coverage. Our lower bound of the accumulation rate was much lower than the previously reported rates elsewhere. The number of newly recruited and small-diameter trees was lower, and the mortality of remnant trees was higher with increasing fern/vine coverage. The growth rate of the extant pioneer trees, which are expected to significantly contribute to the initial build-up phase of the secondary succession, was inhibited when ferns and vines covered the canopy. Our study suggests that the secondary succession is arrested and the recoverability of forest stands is lower than formerly predicted when they are covered by thick ferns and vines.

Understanding the physical processes associated with wind-fire interaction is of significant interest due to the recent occurrences of massive wildfires and broad applications in the scientific and engineering fields. This research aims to study the characteristics of wind-blown flames in a controlled flow environment. To this end, the aerodynamic and thermal structures of the near-wall boundary layer above the fuel-rich surface were investigated experimentally. The flow characteristics were quantified in a small-scale wind tunnel with a 0.48 m2 cross-sectional area using a high-temperature Kanomax Anemomaster. Propane was used above gaseous burners with a heat release rate of approximately 18 kW. The results show that the flame reaction considerably altered the velocity profiles. In the boundary layer without flame reaction, the velocity increases monotonically with the ground height, z. However, in the boundary layer with a diffusion flame, a local acceleration of mean wind velocity known as the fire-induced wind was observed to increase with z to a maximum near the flame zone. Our findings also show an increase in velocity variances and turbulence intensity due to the fire-induced winds. This observation indicates that shear-generated turbulence likely plays a fundamental role in the physics of fire spread. We also observed a satisfactory agreement between the previously conducted numerical simulation and the current experimental data regarding the intensity of fire-induced winds at their respective normalised locations behind the fire line. This study can be further investigated to predict the resultant wind loads to buildings or structures during bushfire events, which will have significant implications for building protection against bushfire attacks.

Patterns of co-occurrence among species can help reveal the structure and assembly of ecological communities. However, studies have been limited by measuring co-occurrence in either space or time but not both simultaneously. This is especially problematic in systems such as masting forests where resources are highly variable, meaning that spatial use and co-occurrence patterns can change on fine spatiotemporal scales. We develop an analytical framework for assessing species co-occurrence at fine spatial and temporal scales simultaneously and apply these models to a camera trapping dataset from Borneo. We sought to determine how substantial variation in food availability across space and time affects co-occurrence among terrestrial vertebrates. We detect many significant, mostly positive, co-occurrence patterns among species, but almost entirely in unlogged forest and during dipterocarp mast years. The most strongly co-occurring pair of species, bearded pig (Sus barbatus) and sambar (Rusa unicolor), only positively co-occur in areas and years when fruit is locally abundant. Species occurrences in logged forest and non-mast years are mostly random with respect to other species. This suggests that frugivore–granivore species positively co-occur when resources are plentiful (i.e., large trees are present and fruiting), likely because they use the same resources; these patterns disappear when food availability is lower. Our approach demonstrates the utility of measuring co-occurrence in space and time together and highlights the importance of resource abundance for driving the co-occurrence structure of communities. Furthermore, our method could be broadly applied to other systems to assess fine-scale spatiotemporal patterns across a range of taxa.

The role of conspecific density dependence (CDD) in the maintenance of species richness is a central focus of tropical forest ecology. However, tests of CDD often ignore the integrated effects of CDD over multiple life stages and their long-term impacts on population demography. We combined a 10-year time series of seed production, seedling recruitment and sapling and tree demography of three dominant Southeast Asian tree species that adopt a mast-fruiting phenology. We used these data to construct individual-based models that examine the effects of CDD on population growth rates (λ) across life-history stages. Recruitment was driven by positive CDD for all species, supporting the predator satiation hypothesis, while negative CDD affected seedling and sapling growth of two species, significantly reducing λ. This negative CDD on juvenile growth overshadowed the positive CDD of recruitment, suggesting the cumulative effects of CDD during seedling and sapling development has greater importance than the positive CDD during infrequent masting events. Overall, CDD varied among positive, neutral and negative effects across life-history stages for all species, suggesting that assessments of CDD on transitions between just two stages (e.g. seeds seedlings or juveniles mature trees) probably misrepresent the importance of CDD on population growth and stability.

Forest degradation has been most frequently defined as an anthropogenic reduction in biomass compared with reference biomass in extant forests. However, so-defined "degraded forests" may widely vary in terms of recoverability. A prolonged loss of recoverability, commonly described as a loss of resilience, poses a true threat to global environments. In Bornean logged-over forests, dense thickets of ferns and vines have been observed to cause arrested secondary succession, and their area may indicate the extent of slow biomass recovery. Therefore, we aimed to discriminate the fern thickets and vine-laden forests from those logged-over forests without dense ferns and vines, as well as mapping their distributions, with the aid of Landsat-8 satellite imagery and machine learning modeling. During the process, we tested whether the gray-level co-occurrence matrix (GLCM) textures of Landsat data and Sentinel-1 C-band SAR data were helpful for this classification. Our study sites were Deramakot and Tangkulap Forest Reserves-commercial production forests in Sabah, Malaysian Borneo. First, we flew drones and obtained aerial images that were used as ground truth for the supervised classification. Subsequently, a machine-learning model with a gradient-boosting decision tree was iteratively tested in order to derive the best model for the classification of the vegetation. Finally, the best model was extrapolated to the entire forest reserve and used to map three classes of vegetation (fern thickets, vine-laden forests, and logged-over forests without ferns and vines) and two non-vegetation classes (bare soil and open water). The overall classification accuracy of the best model was 86.6%; however, by combining the fern and vine classes into the same category, the accuracy was improved to 91.5%. The GLCM texture variables were especially effective at separating fern/vine vegetation from the non-degraded forest, but the SAR data showed a limited effect. Our final vegetation map showed that 30.7% of the reserves were occupied by ferns or vines, which may lead to arrested succession. Considering that our study site was once certified as a well-managed forest, the area of degraded forests with a high risk of loss of resilience is expected to be much broader in other Bornean production forests.

Data used in Williams, Ong, Brodie, & Luskin (2021) Fungi and insects compensate for lost vertebrate seed predation in an experimentally defaunated tropical forest. Nature Communications.We conducted a nested exclosure experiment in Sabah, Malaysian Borneo. We established 10 replicate experimental blocks, spaced 75 m apart along a transect. Within each experimental block, we established six treatments per species. For treatment 1 (control), seeds were placed outside a large vertebrate exclosure, accessible to all seed predators. For treatments 2-6, seeds were placed inside the exclosure, excluding large vertebrate seed predators. For treatments 3-6, seeds were protected by small, closed-top wire mesh rodent exclosures. For treatments 4 and 6, seeds were sprayed with insecticides. For treatments 5 and 6, seeds were sprayed with fungicide. Thus, treatment 4 was only treated with insecticide, treatment 5 was only treated with fungicide, and treatment 6 was treated with both insecticide and fungicide. Both insecticides and fungicide were applied twice per week, and we sprayed an equivalent volume of water on other treatments to reduce bias associated with repeatedly visiting sites. In each treatment in each block, we placed 10 seeds of each of our five tree species within a 30 cm diameter circle. We monitored seeds for 11 weeks, at which point all seeds were either established or assumed to be dead. Over the monitoring period, we assessed how many seeds died before they could germinate, how many germinated but died before they could establish, and how many successfully established. Seeds germinated when the radicle emerged and were considered to have established when the cotyledons unfurled. For seeds that died, we attributed mortality to either vertebrate predation or non-vertebrate mortality. Mortality was attributed to vertebrates either based on chewed seeds and tooth marks, or if seeds disappeared. From these data, we determined seed fate at two stages: germination (stage 1) and seedling establishment (stage 2).

Overhunting reduces important plant-animal interactions such as vertebrate seed dispersal and seed predation, thereby altering plant regeneration and even above-ground biomass. It remains unclear, however, if non-hunted species can compensate for lost vertebrates in defaunated ecosystems. We use a nested exclusion experiment to isolate the effects of different seed enemies in a Bornean rainforest. In four of five tree species, vertebrates kill many seeds (13–66%). Nonetheless, when large mammals are excluded, seed mortality from insects and fungi fully compensates for the lost vertebrate predation, such that defaunation has no effect on seedling establishment. The switch from seed predation by generalist vertebrates to specialist insects and fungi in defaunated systems may alter Janzen–Connell effects and density-dependence in plants. Previous work using simulation models to explore how lost seed dispersal will affect tree species composition and carbon storage may require reevaluation in the context of functional redundancy within complex species interactions networks.

Aims Non-structural carbohydrates (NSCs) are plant storage compounds used for metabolism,transport,osmoregulation and regrowth following the loss of plant tissue.Even in conditions suitable for optimal growth,plants continue to store NSCs.This storage may be due to passive accumulation from sink-inhibited growth or active reserves that come at the expense of growth.The former pathway implies that NSCs may be a by-product of sink limitation,while the latter suggests a functional role of NSCs for use during poor conditions.Methods Using 13C pulse labelling,we traced the source of soluble sugars in stem and root organs during drought and everwet conditions for seedlings of two tropical tree species that differ in drought tolerance to estimate the relative allocation of NSCs stored prior to drought versus NSCs assimilated during drought.We monitored growth,stomatal conductance,stem water potential and NSC storage to assess a broad carbon response to drought.Important Findings We found that the drought-sensitive species had reduced growth,conserved NSC concentrations in leaf,stem and root organs and had a larger proportion of soluble sugars in stem and root organs that originated from pre-drought storage relative to seedlings in control conditions.In contrast,the drought-tolerant species maintained growth and stem and root NSC concentrations but had reduced leaf NSCs concentrations with a larger proportion of stem and root soluble sugars originated from freshly assimilated photosynthates relative to control seedlings.These results suggest the drought-sensitive species passively accumulated NSCs during water deficit due to growth inhibition,while the drought-tolerant species actively responded to water deficit by allocating NSCs to stem and root organs.These strategies seem correlated with baseline maximum growth rates,which supports previous research suggesting a trade-off between growth and drought tolerance while providing new evidence for the importance of plasticity in NSC allocation during drought.

Cladding damage of tall buildings often occurs during hurricane events, and the damage is not only costly but also impacts the economy, and more importantly, threatens lives. One solution to reduce the cladding pressure is the application of modified double-skin facade (DSF) which is already of great interest to architects due to improving building energy efficiency. This paper investigates the effects on the cladding pressures of a building fitted with a DSF with vertical openings in the external skin mounted in front of the windward face of the building. The effects of the vertical openings on the building's surface pressure were investigated by conducting pressure measurements in the wind tunnel. The associated flow mechanisms were revealed by using particle image velocimetry technique. The DSF without opening increases both the mean suctions and fluctuating pressures on both side and leeward faces, and hence this common façade configuration leads to an undesirable effect on building claddings under strong winds. In contrast, the DSF with the opening(s) can effectively reduce wind pressures on the side and leeward faces. Therefore, creating vertical openings on the external skin of DSF is a practical approach to enhance the wind-resistance performance of building claddings.

Liana cutting is a management practice currently applied to encourage seedling regeneration and tree growth in some logged tropical forests. However, there is limited empirical evidence of its effects on forest demographic rates in Southeast Asia. We used 22 four-hectare plots in the Sabah Biodiversity Experiment (a reduced impact logging site) enrichment line planted with 16 dipterocarp species to assess the effects of complete liana cutting on tree growth and survival. We compared plots where lianas were only cut along planting lines (standard enrichment line planting) with those with one (2014) or two rounds (2011 and 2014) of complete liana cutting. We found increased seedling growth following the first complete liana cut in 2011 relative to the enrichment line planting, consistent with previous studies. The response after 3years to the cutting in 2014 depended on whether lianas had been previously cut or not: in twice-cut plots, seedling growth was not significantly different from the standard enrichment planting controls, whereas growth in plots with only one complete cut in 2014 was significantly slower. Seedling survival decreased through time for both once- and twice-cut liana treatments but remained stable in controls. Sapling growth after the 2014 liana cutting showed a similar pattern to seedling growth, while tree growth following the 2014 liana cutting was significantly lower than controls regardless of whether lianas were cut twice (2011 and 2014) or once (2014). Differences in response between the two rounds of liana cutting were likely due to changes in precipitation2011 was followed by consistent rainfall while 2014 was followed by two severe droughts within 2years. Synthesis and applications. Our results generally support the widely reported positive effects of liana cutting on tree growth and survival. However, reduced growth and survival after the 2015/2016 El Nino suggests that drought may temporarily undermine the benefits of liana cutting in logged tropical forests. Managers of similar areas in SE Asia should consider halting liana cutting during El Nino events. In other tropical areas, seedling survival should be monitored to assess to what extent results from SE Asia are transferable. Our results generally support the widely reported positive effects of liana cutting on tree growth and survival. However, reduced growth and survival after the 2015/2016 El Nino suggests that drought may temporarily undermine the benefits of liana cutting in logged tropical forests. Managers of similar areas in SE Asia should consider halting liana cutting during El Nino events. In other tropical areas, seedling survival should be monitored to assess to what extent results from SE Asia are transferable.

The Forest Stewardship Council (FSC) has initiated a new sustainability mechanism, the ecosystem-services certification. In this system, management entities who wish to be certified for the maintenance of ecosystem services (carbon, biodiversity, watershed, soil and recreational services) must verify that their activities have no net negative impacts on selected ecosystem service(s). Developing a robust and cost-effective measurement method is a key challenge for establishing a credible certification system. Using a single method to evaluate a bundle of ecosystem services will be more efficient in terms of transaction costs than using multiple methods. We tested the efficiency of a single method, “biodiversity observation for land and ecosystem health (BOLEH)”, to simultaneously evaluate biodiversity and carbon density on a landscape scale in FSC-certified tropical production forests in Sabah, Malaysia. In this method, forest intactness based on the tree-generic compositional similarity with that of a pristine forest was used as an index of biodiversity. We repeated BOLEH in 2009 and 2014 in these forests. Our analysis could detect significant spatiotemporal changes in both carbon and forest intactness during these five years, which reflected past logging intensities and current management regimes in these forests. Enhancement of these ecosystem services occurred in the forest where sustainable management with reduced-impact logging had long been implemented. In this paper, we describe the procedure of the BOLEH method, and results of the pilot test in these forests.

Precipitation patterns are changing across the globe causing more severe and frequent drought for many forest ecosystems. Although research has focused on the resistance of tree populations and communities to these novel precipitation regimes, resilience of forests is also contingent on recovery following drought, which remains poorly understood, especially in aseasonal tropical forests. We used rainfall exclusion shelters to manipulate the interannual frequency of drought for diverse seedling communities in a tropical forest and assessed resistance, recovery and resilience of seedling growth and mortality relative to everwet conditions. We found seedlings exposed to recurrent periods of drought altered their growth rates throughout the year relative to seedlings in everwet conditions. During drought periods, seedlings grew slower than seedlings in everwet conditions (i.e., resistance phase) while compensating with faster growth after drought (i.e., recovery phase). However, the response to frequent drought was species dependent as some species grew significantly slower with frequent drought relative to everwet conditions while others grew faster with frequent drought due to overcompensating growth during the recovery phase. In contrast, mortality was unrelated to rainfall conditions and instead correlated with differences in light. Intra-annual plasticity of growth and increased annual growth of some species led to an overall maintenance of growth rates of tropical seedling communities in response to more frequent drought. These results suggest these communities can potentially adapt to predicted climate change scenarios and that plasticity in the growth of species, and not solely changes in mortality rates among species, may contribute to shifts in community composition under drought.

The original version of this Article contained an error in the third sentence of the abstract and incorrectly read “Here, using long-term plot monitoring records of up to half a century, we find that intact forests in Borneo gained 0.43 Mg C ha−1 year−1 (95% CI 0.14–0.72, mean period 1988–2010) above-ground live biomass”, rather than the correct “Here, using long-term plot monitoring records of up to half a century, we find that intact forests in Borneo gained 0.43 Mg C ha−1 year−1 (95% CI 0.14–0.72, mean period 1988–2010) in above-ground live biomass carbon”. This has now been corrected in both the PDF and HTML versions of the Article.

Aim Financial incentives to manage forests sustainably, such as certification or carbon storage payments, are assumed to have co-benefits for biodiversity conservation. This claim remains little studied for rain forest mammals, which are particularly threatened, but challenging to survey.Location Sabah, Malaysia, Borneo.Methods We used photographic data from three commercial forest reserves to show how community occupancy modelling can be used to quantify mammalian diversity conservation co-benefits of forest certification. These reserves had different management histories, and one was certified by the Forest Stewardship Council.Results Many threatened species occupied larger areas in the certified reserve. Species richness, estimated per 200x200-m grid cell throughout all reserves, was higher in the certified site, particularly for threatened species. The certified reserve held the highest aboveground biomass. Within reserves, aboveground biomass was not strongly correlated with patterns of mammal richness (Spearman's rho from 0.03 to 0.32); discrepancies were strongest along reserve borders.Main conclusions Our approach provides a flexible and standardized tool to assess biodiversity and identify winners of sustainable forestry. Inferring patterns of species richness from camera-trapping carries potential for the objective designation of high conservation value forest. Correlating species richness with aboveground biomass further allows evaluating the biodiversity co-benefits of carbon protection. These advantages make the present approach an ideal tool to overcome the difficulties to rigorously quantify biodiversity co-benefits of forest certification and carbon storage payments.

Occasional periods of drought are typical of most tropical forests, but climate change is increasing drought frequency and intensity in many areas across the globe, threatening the structure and function of these ecosystems. The effects of intermittent drought on tropical tree communities remain poorly understood and the potential impacts of intensified drought under future climatic conditions are even less well known. The response of forests to altered precipitation will be determined by the tolerances of different species to reduced water availability and the interactions among plants that alleviate or exacerbate the effects of drought. Here, we report the response of experimental monocultures and mixtures of tropical trees to simulated drought, which reveals a fundamental shift in the nature of interactions among species. Weaker competition for water in diverse communities allowed seedlings to maintain growth under drought while more intense competition among conspecifics inhibited growth under the same conditions. These results show that reduced competition for water among species in mixtures mediates community resistance to drought. The delayed onset of competition for water among species in more diverse neighbourhoods during drought has potential implications for the coexistence of species in tropical forests and the resilience of these systems to climate change.