1. Ecologists seek a general scheme to classify the diversity of plant responses to environmental factors into a few strategies (e.g. competitor -C, stress-tolerant -S, ruderal-R), while plant physiologists seek a mechanistic scheme to explain such different responses (e.g. tolerance, escape, avoidance). So far, few attempts have been made to combine both perspectives into plant eco-physiological strategies. Moreover, the relative contribution of different strategies to maintain both community structure and ecosystem functioning during drought has rarely been assessed. This limits our capacity to predict how extreme events caused by climate change will affect plant communities.

2. Here, we present an integrated framework to identify plant eco-physiological strategies and to estimate their contribution to community originality (diversity of trait-combinations), dominance (species relative frequency), and ecosystem functioning (productivity and evapotranspiration).

3. We applied this framework in a tropical montane grassland and found three co-occurring eco-physiological strategies in this community (S-tolerance/avoidance, CS-escape/tolerance and CR-escape/avoidance). While CS-species contributed more to dominance and functionality, CR- and S-species contributed more to originality. Therefore, all three strategies were important to support the grassland form and function.

Synthesis: Plants exhibit different strategies, as well as different contributions to community and ecosystem attributes. We developed an integrated approach to both identify strategies and estimate their relative contribution. Thereby, as droughts intensify, we can better predict which plants are more likely to be lost and how their loss will impact the communities and ecosystem where they occur. This knowledge is necessary for specifying conservation priorities and for developing more efficient conservation practices.

We collected branchlets (for woody species) or whole individuals (for herbaceous species) from 10 individuals per species (N = 51 plant species in total), then re-cut the samples under water and rehydrated for at least 2 h. After rehydration, we sampled five mature leaves (excluding the petiole) per individual and determined their water saturated fresh mass (FM, g) using a precision balance (0.0001 g). Subsequently, we scanned (300 dpi resolution) the leaves and used the software Image J version 1.48 to obtain the leaf area (LA, mm², Pérez-Harguindeguy et al., 2013). Then, we oven-dried the leaves at 50 °C for 72 h, to determine the leaf dry mass (DM, g). LA, fresh and dry mass were used to obtain: specific leaf area – SLA, mm² mg⁻¹ [SLA = LA/DM], leaf dry matter content, % - LDMC [LDMC = (DM/FM)/10], and leaf succulence – SUC, g H₂O dm^-2 [(FM – DM)/LA]. To determine the stem specific density (SSD, mg mm^–3), we cut 5 cm stem sections and removed the bark with a knife. Stem sections were then submersed in water for at least 30 min. Next, we obtained the stem fresh volume by following the Archimedes principle of water displacement (Rosado & de Mattos, 2010). Finally, the stems were oven-dried for five days at 60 °C, and then SSD was calculates as: Stem dry mass/ stem fresh volume. We measured foliar water uptake (FWU, %) on five individuals per species following methods from Limm et al. (2009). To obtain stomatal and venation traits, we used one mature and healthy leaf of five individuals per species. Each leaf was subjected to the diaphanization technique (Strittmater, 1973), stained with safranin and mounted on glass slides. The, we used a digital camera coupled in a light Microscope (Olympus Cx40, Spectra Services, Ontario, USA) to take digital photographs of both abaxial and adaxial leaf surfaces up to 1 mm² of area sampled per leaf surface. Next, we analysed the digital images using the software Image Pro Plus version 4.5, to obtain: stomatal density (SD, n° mm^-2), by counting the number of stomata per unit of leaf area; venation density (VD, mm mm^-2), by summing up the length of all minor veins and dividing by the total area sampled, and the average fraction of the leaf surface allocated to stomatal pores (Fsp, %), by multiplying SD by the average anatomical maximum stomatal pore area (which was measured on 10 stomata per digital leaf image, using the software Image J). We measured leaf water potentials at predawn (Ψ_pd, MPa) and midday (Ψ_md, MPa) and stomatal conductance at midday (g_s) at each month from Jun 2016 to Aug 2017. Measurements were obtained from two individuals per species (2-3 leaves per individuals) using a pressure chamber (model 1505D-EXP, PMS, Albany, OR, USA) and a leaf porometer (model SC1, Decagon Devices, Pullman, WA, USA), respectively. Midday measurements were conducted from 12 to 14h, and predawn measurements from 4 to 6h. From those data, we calculated the maximum midday stomatal conductance (g_{s max}) and the iso-/anisohydric behaviour (Δslope; Meinzer et al., 2016). We also obtained pressure–volume curves for five individuals per species during the dry season (Jun-Aug 2017) using the bench drying technique (Turner, 1988). Hydraulic traits, except FWU, could not be directly obtained for the whole set of species, as their measurement is too time-consuming. So, for those traits we selected a subset of 12 species. The missing values for the hydraulic traits were then imputed using a Bayesian hierarchical probabilistic matrix factorization (BHPMF) (Schrodt et al., 2015). We collected seeds from five individuals per species, monthly from 2016 to 2017. From each individual we selected five seeds, and measured their size (SS, cm) by taking digital photographs with a stereomicroscope and then measuring the largest dimension of the seed using the software Image J. Resprouting ability (RA, %) was experimentally obtained following Moreira et al. (2012) protocol. We assessed species and strategies functionality based on their relative contribution to productivity (gross primary productivity - GPP, g C m^-2 d^-1) and water fluxes (transpiration - ET, kg H₂O m^-2 d^-1). GPP and ET were modelled for a subset of 12 species, using a stomatal optimization model from Eller et al. (2020). We assessed dominance, by measuring species relative frequency in the study area during the summer (peak standing biomass) and using the quadrant method (300 m² of area sampled). Finally, we assessed species originality as the Euclidean distance of a target species to the centre of the multidimensional trait space defined by the PCA results (Leitão et al., 2016; Violle et al., 2017).

Traits definition/meaning:

Leaf area (LA, mm²) : One-sided area of an individual leaf ;

Leaf dry matter content (LDMC, %): Leaf dry mass divided by its water-saturated fresh mass;

Leaf succulence (SUC, g H₂O dm^-2): Leaf capacity to store water, measured as (saturated leaf mass - leaf dry mass)/leaf area;

Specific leaf area (SLA, mm² mg ⁻¹ ): Represents the construction cost of a unit leaf area: one-sided leaf area/leaf dry mass ;

Stem specific density (SSD, mg mm^–3): Describes the carbon investment per unit volume of stem: stem dry mass/ stem fresh volume;

Foliar water uptake (FWU, %): The movement of water coalesced on the leaf surface into the leaf ;

Stomatal density (SD, n° mm^-2): Number of stomata per unit of leaf area;

Vein density (VD, mm mm^-2): Length of minor veins per unit of leaf area;

Average fraction of the leaf surface allocated to stomatal pores (Fsp, %): The stomatal density multiplied by the average anatomical maximum stomatal pore area;

Maximum midday stomatal conductance (g_{s max,} mmol m^-2 s^-1): A measure of the water loss from plant leaves controlled by stomatal aperture;

Iso-/anisohydric behaviour (Δslope, MPa MPa^-1): The slope of a linear regression fitted to a plot of log (pre-dawn leaf water potential Ψ_pd- midday leaf water potential Ψ_md) vs. Ψ_pd;

Leaf water potential at the turgor loss point (Ψ_tlp, MPa): The negative water potential at which leaf cells lose turgor;

Seed size (SS, cm): The longest dimension of the seed;

Resprouting ability (RA, %): Species ability to regenerate after the destruction of its above-ground biomass by using reserves stored in basal or below-ground tissues;

Competitiveness (C, %): species percentage of competitor strategy in the CSR triangle;

Stress-tolerance (S, %): species percentage of stress-tolerant strategy in the CSR triangle;

Ruderalism (R, %): species percentage of ruderality strategy in the CSR triangle;

Originality (range 0 - redundant to 1 - original): the Euclidean distance of a target species to the centre of the multidimensional trait space;

Gross primary productivity  (C m^-2 d^-1): carbon biomass produced per area per day;

Transpiration (kg H2O m^-2 d^-1): kilograms of water transpired per unit of leaf area per day;

Predawn leaf water potential (MPa): leaf water potential measured before the sunrise;

Frequency (%): number of sampling units where a certain species occurred, divided by total number of sampling units (N = 300)

Categories of dominance (categorical): rare if  frequency ≤ 5%, subordinate if 5%<  frequency < 50%, and dominant if  frequency  ≥ 50%.