Pre?dispersal seed predation could help explain premature fruit drop in a tropical forest

Seed production and seed survival are critical elements in the life cycle of a plant, giving rise to next generation mediating population- community-level dynamics (Clark et al., 2007; Green 2014; Maron & Crone, 2006; Turnbull 2000). Mortality events that happen early can be bottleneck for recruitment individuals population, with sometimes disproportionate effects on community composition (Fenner Thompson, 2005; James 2011; Rother 2013; Roughgarden 1988; A number processes contribute success or failure developing seed, including nutrient availability (Bertamini Nedunchezhian, Martinez-Alcantara 2012), microclimatic conditions (Einhorn Arrington, 2018), weather (Najeeb 2017; Reichardt 2020) interactions between plant other organisms. For species relying biotic pollination, visits from pollinators crucial reproductive success, often manifesting as direct positive correlation pollinator visitation rate set (Karron Steffan-Dewenter 2001). Following successful might target range enemies, pre-dispersal insect predators 2005), deleterious fitness. commonly observed phenomenon plants is some fruits prematurely abscised, is, they drop mother prior completing their development. Plant which regularly abscise large proportion form taxonomically ecologically diverse group (Stephenson, 1981) it currently unknown whether shared ancestry contributes variation rates premature fruit across species. The ecological agricultural research literature reports several causes drop. example, changes resource availability, through mechanisms such herbivory 1980), leaf shading 2018) drought (Nussbaumer 2020; Pérez-Pérez 2008; 2020), trigger due competition resources among (Bawa Webb, 1984; Goubitz 2002; Stephenson, 1981). Developmental genetic abnormalities (Bradbury, 1929; Forino 1987; Kraus, 1915), pollen quality (Goubitz 2002) damage abiotic means, example frost (Rodrigo, 2000; Rodrigo Tagliasacchi 2006), an individual unlikely reach maturity thereby minimise cost parent Damage by natural enemies also lead drop, seed/fruit predation pathogen attack (Akinsanmi 2016; Boucher Sork, 1979; Planes 2014). Regardless exact mechanism causing resulting mortality could—if reduces viable seeds produced plant—have important population (Turnbull its wider consequences, remains largely unexplored context tropical forest (but see Bawa Jones Comita, 2010). Of particular interest potential enemy-triggered which, if widespread showing patterns density dependence (e.g. locally abundant experiencing higher levels drop), could coexistence these highly systems (Chesson, Although role maintaining high local diversity has received substantial attention communities since Janzen's (1970) Connell's (1971) seminal papers, bulk conducted date focused attacking young seedlings after dispersal Comita Holl Lulow, 1997; Levi 2019). While been highlighted potentially source facilitator forests (Gillett, 1962; Gripenberg, 2018; Janzen, 1970), enemy period dispersal—as developing—is only rarely considered this Taking advantage 31-year dataset rain well-studied woody Barro Colorado Island, Panama, we relate 201 (trees, shrubs, lianas) (a) phylogeny, (b) traits attributes hypothesised associated (c) previously documented one (insect predators) known settings (Follett, Tzanakakis 1997). Through approach, assess likely contributors explore consequences biodiversity maintenance forests. Island (BCI; 9°9?N, 79°51?W) 16 km2 island situated Gatun Lake Panama Canal, supports semi-deciduous 35-m tall canopy. Smithsonian Tropical Research Institute custodian Nature Monument behalf Republic Panama. BCI 50-ha plot was first established within CTFS-ForestGEO network 1982 (Anderson-Teixeira 2015). Since 1987, weekly censuses have part project coordinated S. J. Wright 1999, 2003, Calderon, 1995). traps (n = 200) 1987 continuously monitored then. These located at 13.5 m intervals alternating sides 4–10 2.7 km pre-existing trails plot. Another 50 were added naturally occurring tree fall gaps 2002 2004 since. final 200 2 each original 2011 censused 23 months. Each trap consists 0.8 PVC frame 1 mm mesh covering 0.5 m2 floor. In censuses, all encountered identified species, counted categorised mature immature. long-term study, very collected immediately shortly flowering termed ‘aborted’. not included study. Fruits labelled ‘immature’ dropped weeks months later. distinction immature based examination endocarps: If endocarp filled material solid (rather than hollow liquid soft substances), then mature. Our species-specific estimates abscission relative frequencies fruits. did conduct systematic germination trials full cannot rule out possibility scored would able germinate. We still confident vast majority cases, inside indeed inviable (given timing happens many species) typically translates studied community. used information shrubs recorded during 1988–2018 (31 calendar years). All analyses data preparation r v 4.0.5 (R Core Team, 2021). Counts multiplied average seed-to-fruit ratios (S. Wright, unpubl. data) seeds. counts 1,615 censuses) summed traps. This done separately year dataset. 90% once year, few December–January period, approach will generate overall fruiting events. Not every single but total there 5295 unique × combinations Species fewer 10 omitted ensure representative sample size, yielding 3,286 observations To contributed tested signal susceptibility (see Section 1) abscised larger Additionally, group—insect predators—we test attacked, associations obtained archived (Gripenberg 2019a, 2019b; Information how estimated, sources provided Table 1. relationship variable interest, constructed generalised linear mixed models (GLMMs) binomial distributions using lme4 package (Bates 2020). combined into two-column response variable. explanatory attribute (listed presence predators, random effects. most variables, available subset independently (i.e. continuous covariates separate models). Two variables (local abundance conspecifics mass) log-transformed standardised (mean 0, SD aid model fitting interpretation. dharma (Hartig, scaled residuals simulation fit models. plotted verify specified correctly. Durbin–Watson statistic temporal autocorrelation additionally AR(1) covariance structure glmmtmb (Brooks 2021) effect results. assessed pairwise correlations Spearman's rank coefficient, multicollinearity. Results visualised ggplot2 (Wickham, 2016) coefficient 95% confidence intervals. similarities community, estimated phylogenetic (the tendency related more similar drawn tree) abscised. phylogeny 2019b), 184 our dataset, mean proportions analyses. Both Pagel's lambda (Pagel, 1997, 1999) Blomberg's K (Blomberg 2003) ‘phylosig’ phytools (Revell, 2012). ‘contMap’ function visualise trait mapped onto tree. Premature common. 1,222,863 years 632,835 (52%) account (species-specific) numbers per fruit, corresponds 3,159,062 (39%) 8,062,517 being either diaspores No 29 these, 20 (69%) had 100 period. remaining 172 57 50% when averaged study (Figure 1). Correlations relatively weak S1). 867 combinations, 47 (5%) showed significant (Durbin–Watson <1 p < 0.05) (3%) negative >3 residuals. Fitting no Other investment mechanical defences, investigated demonstrated 0.05 2; relationships rate, overlap production, line predictions. saw clear four attributes: abundance, mass, time fruiting) height. detected two crop size defences. 28% BCI. attacked least predator prematurely. Estimates low (K 0.002, 0.10; ? 0.33, 0.005; Figure 3). stages determinants populations later (Green 2014), earliest stages—when attached plant—has (Gripenberg, analysis reveals site responsible deaths up 39% initiated. Rates signal, suggesting explain small amount found five six predicted abscission. key aim enemies—insect predators—show guild. suggests insects may play triggering association make prone agreement previous studies forests, interspecific explained defensive (Cárdenas Coley, 1983, Schuldt mass addition providing food resource, exposed longer time, take develop (Moles Westoby, 2003). Locally colonised by, support host-specific (Hanski, 2001; Pacala Crawley, 1992). taller Canopy trees crops apparent understorey (Castagneyrol 1968). greater height increase ‘apparency’ (Feeny, 1976). Leaf cause (Petzold 2009) abundance) apparency probably herbivores, possibly contributing (Floater Zalucki, Hughes, times tended contrasts prediction result satiation generalist reducing fruits, however, predators. case host specific, know guild: 2019a), specific. reared samples representing 478 (80%) 20% restricted ranges. fact same S2). produce long year) predictions, lower temporally unstable populations. defences yield statistical support, although detect direction. Plants invest approximately biomass structures do producing embryo endosperm tissues Reducing investing both come offspring. selection traits, given lost via selective agents plants. observation before maturation, should ecologists interested long-standing question ecology what allows so coexist (Wright, 2002). prerequisite stable 2000): must growth rare suggest fitness costs mortality) common specialist exhibited further dependence, envisaged Janzen–Connell hypothesis (Connell, 1971; 1970). current evidence circumstantial, conspecific 2018). major implications coexistence, caution needs taken interpreting results presented here. First, uses correlative caused triggers unrelated relationships. As height—which enemies—also correlates light availability. (Nevertheless, limitation driver drop.) Second, while die triggered those survived absence enemies. It suggested flowers realistically maturation 1981), them fail cross-fertilised. trees, Ghazoul Satake proposed ‘sacrificial sibling hypothesis’ rather aborting inbred stage, retain surplus low-quality act sinks diluting (Ghazoul Satake, 2009). logistically challenging, manipulative experiments measure output helpful interpretation Third, argued methodological overestimated displacement canopy dispersers inflated However, reason why just brought removed above them, spread ha plot, believe quantified reality. summary, overlooked favour better-known post-dispersal shaping structure. hinted driving stabilising dependence. results, encourage consider investigating communities. supported PhD studentship E.E.J QMEE CDT, funded NERC grant NE/P012345/1. S.G. Royal Society University Fellow. collection Environmental Sciences Program Institution 2008 thereafter. Data funding Academy Finland (S.G.'s postdoctoral project; 138299) Fellowship). thank anonymous reviewers feedback version manuscript Pablo Medrano-Vizcaíno Spanish translation abstract. authors declare conflict interest. S.G., E.E.J. S.J.W. conceived idea developed following discussions J.M.B. T.O.; curation (including data); O.C. rain; methodology, formal analyses, visualisations wrote initial draft. subsequent revisions. peer review history article https://publons.com/publon/10.1111/1365-2745.13867. yearly Dryad Digital Repository https://doi.org/10.5061/dryad.4mw6m909j (Calderón 2022). public repositories https://doi.org/10.5061/dryad.230j5ch 2019b) 2019a). code relating repository GitHub https://github.com/ee-jackson/premature-fruit-drop Zenodo https://doi.org/10.5281/zenodo.5767681 (Jackson Supinfo Please note: publisher content functionality any supporting supplied authors. Any queries (other missing content) directed corresponding author article.