Description of Experiment and Data Collection This experiment was done in the tropical dry forests at the Parque Nacional Palo Verde, located in the Área de Conservación Arenal-Tempisque (10.358N, 85.358W), and in Parque Nacional Santa Rosa, located in the Área de Conservación Guanacaste (10.848N, 85.628W), in northwestern Costa Rica. The study consisted of twenty-one 20 x 50 m plots distributed across the two areas. Seven plots at Parque Nacional Palo Verde and seven plots in Sector Santa Rosa were placed in a forest type consisting of a mixture of drought-deciduous and evergreen species (referred to as Palo Verde and Santa Rosa). In Sector Santa Rosa, an additional seven plots were placed in an evergreen forest once dominated by the live oak Quercus oleoides with low abundance of co-occurring drought-deciduous and evergreen species (referred to as Santa Rosa Oak). Rainfall was measured separately for Palo Verde and Santa Rosa to get accurate local conditions. For each site, we used monthly rainfall to calculate the longest period of consecutive months with rainfall below 50 mm per month between each seedling census to determine annual dry season length for that census interval. This period typically began in December and ended in May but using rainfall quantity and not fixed dates allowed the dry season to vary among years. From 2015 to 2025 the average dry season length was 5 months but ranged from 3 to 7 months. We also calculated the total rainfall of the dry period as dry season severity which ranged from 1 mm to 113 mm over the 10 years. The experiment started in November 2016 following the ENSO-induced drought, all seedlings were measured for height, identified to species and tagged in a 1 x 50 m transect bisecting each forest plot. Sampling of seedlings was recorded for each 1 x 1 m subplot along the transect (50 subplots in total) to monitor the spatial distribution of seedlings along the plot longitude. Seedling sampling was carried out annually at the end of the rainy season (November to December) from 2016 to 2025. In total, we recorded 1532 seedlings of 91 species from 70 genera, equating to approximately 1765 recruiting seedlings per hectare per year. Species traits of the tree community were compiled from previous studies in the area (Powers et al., 2020; Powers & Tiffin, 2010; Vargas G. et al., 2021). To supplement gaps in local measurements, we accessed the TRY database (Request No. 30374) to assess trait measurements of the local forest species from regional studies (Kattge et al., 2020) for seed mass and specific-leaf area. Seed mass coverage was 55 species (42 without the TRY data). Specific-leaf area coverage was 65 species (59 without the TRY data). In addition, we used the BIOMASS package (Réjou-Méchain et al., 2017) to obtain average wood density values for 8 species from Central American populations and 6 species from global measurements. Wood density coverage was 82 species (68 without the BIOMASS data). Turgor loss point coverage was 46 species and was not supplemented. Description of each column in SeedlingTotalData.txt pid – factor defining plot and census number (e.g. 1 1 is plot one in census one) id – name of seedling (genus_species epithet) date – data of measurement dir – offset direction from plot center plot – plot number splot – subplot number from 1 to 50 meters spp.no – tag number of the seedling in the plot genus – genus of individual spp – specific epithet of individual hgt – height of seedling in cm muerto – survival (1) and mortality (0) of individual DBH – diameter at breast height once individual reaches 1.3 m day – date as a numerical number census – cenusu number with 1 being 2016 and and 10 being 2025 time – number of days from the first measurement of the individual global_day – number of days from the first measurement of the plot recruitSap – observation of 1 if the seedling reaches 1.3 m height site – three sites SRO is Santa Rosa Oak, PV is Palo Verde and SR is Santa Rosa disp_type – seed dispersal type (a is animal, w is wind, and d is gravity/autochory) leaf_type – leaf phynology (deciduous, semideciduous and evergreen) N_type – legume or nonlegume seed – seed weight in grams sla – specific leaf area (cm2 g-1) wood – wood density (g cm-3) dryl – dry season length (months) drys – dry season severity (mm of rain) wets – wet season severity (mm of rain) TLP – turgor loss point (MPa) age – estimated age of the forest since disturbance Description of each column in TLP.txt genus – genus of individual spp – specific epithet of individual TLP – turgor loss point (MPa) Description of each column in FullTrait.txt SpeciesID – name of seedling (genus_species epithet) family – family of the species genus – genus of individual spp – specific epithet of individual disp_type – seed dispersal type (a is animal, w is wind, and d is gravity/autochory) leaf_type – leaf phynology (deciduous, semideciduous and evergreen) N_type – legume or nonlegume seed – seed weight in grams sla – specific leaf area (cm2 g-1) wood – wood density (g cm-3) lwc – leaf water content (%) d15N – isotope delta 15 N ratios d13C – isotope delta 13 C ratios leafP – leaf P in mg per gram leafN – leaf N in mg per gram leafC – leaf C in mg per gram Description of RCode SaplingRecruitmentRandomSlopes.R - Code for analysis of recruitment to sapling (Figure 4) SeedlingMortalityRandomSlopes.R - Code to generate seedling survival (Figure 2 and 3) SeedlingRecruitmentRandomSlopes.R - Code to generate seedling recruitment (Figure 1) SeedlingTraitStrapCWM.R - Code to generate community-weighted trait means (Figure 5)