Previous theoretical work has highlighted the potential for natural enemies to mediate the coexistence of species with similar life-histories via density-dependent effects on survivorship. For plant pathogens to play this role, they must differ in their ability to infect or induce disease in different host plant species. In tropical forests characterized by high diversity, these effects must extend to phylogenetically closely related species pairs. Mortality at the seed and seedling stage strongly influences the abundance and distribution of tropical tree species, but the host preferences and spatial distributions of fungi are rarely determined.

We examined how host species identity, relatedness, and seed viability influence the composition of fungal communities associated with seeds of four co-occurring pioneer trees (Cecropia insignis, C. longipes, C. peltata, and Jacaranda copaia). Seeds were buried in mesh bags in five common gardens in the understory of a lowland tropical forest in Panama and retrieved at intervals from 1-30 months. A subset of the seeds in each bag was used to determine germination success. One half of each remaining seed was tested for viability; the other half was used to culture and identify seed-infecting fungi.

Seeds were infected by fungi after burial. Although fungal communities differed in viable vs. dead seeds, and across burial locations, community composition primarily varied as a function of plant species identity (30.7% of variation in community composition vs. 4.5% for viability and location together), even for congeneric Cecropia species. Phylogenetic reconstruction showed that relatedness of fungi mostly reflected differences between Jacaranda (Bignoniaceae) and Cecropia (Urticaceae).

Although the proportion of germinable seeds decreased gradually over time for all species, intraspecific variation in survival was high at the same location (e.g., ranging from 0-100% for C. peltata) suggesting variable exposure or susceptibility to seed pathogens.

Synthesis: Our study provides evidence under field conditions that congeneric tree species with similar life-history differ markedly in seed-associated fungal communities when exposed to the same soilborne fungi. This is a critical first step supporting pathogen mediated coexistence of closely related tree species.

Seeds of all species were collected from at least five different maternal sources in 2012-2014 in accordance with seasonal variation in fruiting phenology. Mature infructescences of Cecropia spp. were collected directly from trees or from freshly fallen material beneath their crowns. Jacaranda copaia trees can attain 35 m in height when mature, precluding collection of ripe fruits directly from the crown. For this species we used recently fallen seeds estimated to have been on the soil surface for a short time (from minutes to a few hours). Seeds of J. copaia were separated from dehiscent woody seed pods and the wings were cut from each seed. Seeds of Cecropia (technically fruits; Lobova et al. 2003) were removed from whole infructescences. Immediately after collection seeds were cleaned with tap water and 0.7% sodium hypochlorite for 2 min to remove small pieces of pulp. Clean seeds of all species were air-dried at room temperature (∼22^oC) for several days under low red:far red irradiance.

We established five common garden plots (9 x 15 m each) in lowland forest at Barro Colorado Island (BCI). Common gardens were at least 350 m apart (average distance 800 m), and were located under closed-canopy forest to limit seed germination (see Table S1 for a summary of common garden characteristics; see also Zalamea et al. 2015, Ruzi et al. 2017, and Sarmiento et al. 2017). They were placed in multiple forest- and soil- types that varied slope and aspect (Baillie et al. 2006). No adults of the study species occurred within 20 m of the garden edges. Dry seeds from all maternal sources of each species were thoroughly mixed, pooled, and placed into seed bags. Each seed bag contained 45 seeds of one species mixed with 10 g of sterile forest soil (previously autoclaved at 121^oC for 2 h). Each set of soil and seeds was enclosed in a nylon mesh bag (pore size = 0.2 mm) and covered with aluminium mesh (pore size = 2 mm) to exclude seed predators but not microorganisms. Twenty-eight bags per plant species were distributed among four replicate subplots per garden, where they were buried 2 cm below the soil surface and 40 cm apart. A total of 140 seed bags and a total of 6300 seeds per species were included in the burial experiment. Seeds of each species were buried less than a month after seed collection, matching the natural phenology of fruit production and seed dispersal. We evaluated initial infection rates and confirmed the viability status of the seed lot before burial by processing 400 fresh seeds (i.e., seeds that were not buried) per species (200 seeds for germination; 200 seeds for fungal culture). 

Seed processing and retrieval after burial:
Four seed bags for each species in each garden were collected at each of seven time points: 1, 3, 6, 12, 18, 24, and 30 months after burial. Seeds were retrieved by rinsing the contents of each bag with tap water over a sieve. Seeds were partitioned for tests of (i) germination (10 seeds per seed bag) and (ii) seed viability and identification of seed-associated fungi (10 seeds per seed bag). In general, seeds were recovered intact, precluding lethal germination as a source of seed mortality. In 2% of cases, we recovered fewer than 20 seeds per bag, suggesting that small invertebrate seed predators penetrated or chewed the nylon mesh. In these cases, the recovered seeds were partitioned evenly among germination and culturing approaches. We did not find 28 (i.e., 5%) of bags that were buried, likely reflecting displacement by small mammals or rain events. To record the time series of seed germination for each species, 10 seeds from each seed bag were placed in a Petri dish lined with paper towel, moistened with sterile distilled water, and sealed with two layers of Parafilm® (Zalamea et al. 2015; 2018). Seeds were incubated in a shade house on BCI under ca. 30% sunlight, high red:far-red irradiance (ca. 1.4), and ambient temperature (23-30°C). Germination (i.e., radicle protrusion) was scored over six weeks.