How does avian seed dispersal shape the structure of

Accepted Article

Functional Ecology DR AARÓN GONZÁLEZ-CASTRO (Orcid ID : 0000-0002-5311-1017)

Article type

: Research Article

Editor

: Dr Matthias Schleuning

Section

: Plant-Animal Interactions

How does avian seed dispersal shape the structure of early successional tropical forests?

Aarón González-Castro1,2, Suann Yang1,3, & Tomás A. Carlo1* 1

Department of Biology & Ecology Program, The Pennsylvania State University, 208 Mueller

Laboratory, University Park, PA 16802, USA. 2

Present address: Island Ecology and Evolution Research Group (IPNA-CSIC), Calle

Astrofísico Francisco Sanchez 3, E38206 La Laguna, Tenerife, Canary Islands, Spain 3

Present address: Biology Department, State University of New York Geneseo, Geneseo, USA.

*

Corresponding author: Aarón González-Castro

Island Ecology and Evolution Research Group (IPNA-CSIC), Calle Astrofısico Francisco Sanchez 3, E38206 La Laguna, Tenerife, Canary Islands, Spain. E-mail address: [email protected]

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1365-2435.13250 This article is protected by copyright. All rights reserved.

Accepted Article

ABSTRACT 1. Frugivores shape plant communities via seed dispersal of fleshy-fruited plant species. However, the structural characteristics that frugivores impart to plant communities are little understood. Evaluating how frugivores structure plant communities via the nonproportional use of available fruit resources is critical to understand the functioning of ecosystems where fleshy-fruited plant species are dominant, such as tropical forests. 2. We performed a seed-addition field experiment to investigate how frugivorous birds shape the composition and richness of forests during early stages of secondary succession in cleared areas in Puerto Rico. The experiment tested whether the bird-generated seed rain and the subsequent early successional plant communities were proportional representations of the fleshy-fruited species that dominated the surrounding community. Experimental treatments consisted of patches with 1) seed-additions by wild birds attracted to experimental patches with pole perches, 2) manual seed-additions proportional to fruit abundance at the local scale (≤ 50 m from experimental plots), and 3) manual seedadditions proportional to fruit abundance at the landscape scale (entire study site). 3. Birds’ seed-additions differed in composition and abundance to expectations based on fruit availability at local and landscape scales. Treatments with seeds added by birds had the highest species richness in both the seed rain and the emergence stages despite how, on average, the monthly richness in the landscape-scale treatment was double that of birds and the local-scale treatment. This phenomenon was explained by the highest heterogeneity from the bird seed-addition treatment across months, and the lowest seed per-capita emergence rates in landscape treatments. Rather than reflecting relative fruit abundance, birds biased seed rain and per-capita emergence towards a non-random mixture of both small- and large-seeded species, resulting in richer and distinct plant communities.

This article is protected by copyright. All rights reserved.

Accepted Article

4. Because frugivory and seed dispersal patterns depart from random encounters between frugivores and plants in communities, successional forests are characterized by an overrepresentation of proportionally rare plant species, and decreases in the dominance of many common species. Thus, for regenerating tropical forests, frugivory can function as mechanism that promotes persistence of rare plant species and their coexistence with more abundant plants.

Key-words: anti-apostatic, dispersal limitation, diversity maintenance, tropical forest regeneration, frugivory and seed dispersal, island ecology, recruitment limitation

INTRODUCTION Fruit-eating animals that disperse seeds – i.e. frugivores – are thought of as architects of tropical plant communities where most of the woody vegetation depends on them for dispersal and regeneration (Jordano, 2000, Lázaro et al., 2005; Bascompte & Jordano, 2007; García & Martínez, 2012). However, a continued challenge is understanding how much of plant community structure – i.e., the relative species’ abundance and composition – is shaped by frugivores. For example, we do not know how much structural variation of tropical plant communities is actively shaped by frugivore actions. Understanding this would require comparing the early templates of plant regeneration produced by frugivores with those produced by null models of seed dispersal (Schupp & Fuentes, 1995; Gotelli, 2000). Plant regeneration is a complex process affected by multiple factors such as environmental heterogeneity, disturbance history, species traits, and competition as well as other antagonistic interactions (Weiher & Keddy, 1995; Norden et al., 2009; Terborgh, 2012; Chazdon, 2014). Although studies have established the importance of seed dispersal by frugivores for tropical forest regeneration (e.g., Da Silva et al., 1996; Holl, 1998; Carlo & Morales, 2016), we still need an integrated understanding of how fruit choices of birds interact


Accepted Article

with the temporal dynamics of fruit availability in shaping seed rain composition. Further, we need to quantify how the influence of frugivore choices on the seed rain transfers to the composition of early successional plant communities. Addressing these challenges demand linking resource availability patterns to frugivory and the fate of dispersed seeds at community levels (Carlo & Yang, 2011). Advances could be made with experimental approaches designed to probe the functional relationships between fruit resource availability, frugivory, and the resulting structure of regenerating plant communities. Issues of spatial scale quickly arise when attempting to evaluate the link between fruit choices and fruit availability, mainly because of the highly heterogeneous distribution patterns typical of tropical plant species (Condit et al., 2002). For example, fruit abundance may be more important for frugivores at small local scales than at larger regional scales, and vice versa (García & Ortiz-Pulido, 2004). Thus, interpretations of rarity vs. commonness for plant species in a sample of seed rain, or growing in a forest patch, will depend on the spatial scale at which the relative abundance of the fruit resources were evaluated in the local community (Saracco et al., 2004). Further, the spatial heterogeneity of landscapes and resources therein influences the foraging behaviour and movements of frugivores although this aspect still remain little studied (Morales et al., 2013). Therefore, using a multi-scale approach is necessary for understanding the influences of frugivore choices on the structure of regenerating plant communities. Frugivory, seed dispersal, and plant regeneration dynamics are also shaped by plant species traits such as their temporal patterns of fruit production (i.e., phenophase, GonzálezCastro et al., 2012; Yang et al., 2013), and the mass of seeds (Muller-Landau, 2008; Norden et al., 2009). For example, plant species that produce fruits for long time periods may have greater opportunities for interacting with frugivores, and be thus dispersed more frequently than species with shorter fruiting periods (González-Castro et al., 2012). Seed mass also influences the number of seeds that are ingested and dispersed by frugivores and the seed per-capita establishment rates, with more seeds being dispersed from the small-seeded species (Emer et al., 2018), while large seeded species experience higher per-capita establishment (Foster & Janson,


Accepted Article

1985). Thus, plant community structure is influenced in part by how plant species traits affect frugivory, seed dispersal, and emergence. Here we hypothesize that frugivores are functionally important for regenerating tropical plant communities because they modify the community-level seed rain of plant species relative to their availability in the environment, leading to changes in the diversity and composition of recruiting seedlings. For instance, morphological and phenological mismatches prevent or limit interactions between frugivores and certain fruiting species (sensu Olesen et al., 2011). In addition, nutritional qualities of fruit may influence frugivore behavior (Levey and Martínez del Río, 2001) to include a wider array of fruiting species compared to what would be expected from plant relative abundances (Carlo & Morales, 2016). Thus, interactions of frugivores with rare plant species can be surprisingly common, and function as an equalizing mechanism in plant communities where a few species are common while many are rare (Carlo & Morales, 2016; Morán-López et al., 2018). Still, we need a fuller understanding of how these processes combine to create structural attributes in plant communities. To address this gap, we designed a seed-addition experiment to investigate how frugivorous birds – the dominant seed-dispersal agents in tropical ecosystems – shape the composition of a regenerating forest. Our design combined measures of temporal fruit resource availability in the community at two spatial scales (a local and a large scale) and plant functional traits (phenophase length, seed mass) with seed-addition treatments designed to evaluate the structural characteristics of plant communities assembled from frugivore dispersal. We tested the null hypothesis that the seed rain generated by birds mirrors fruit availability – at local or large scales – and its fluctuations over time. Alternatively, birds would generate seed rains that deviate from the null expectation, promoting assemblages that are more diverse and structurally distinct.


Accepted Article

MATERIAL AND METHODS Study site The study was conducted from February 2013 through October 2014 at Finca Montaña, Puerto Rico (18°28’03.00” N; 67°06’46.97” W). Finca Montaña is a research and conservation area administered by the College of Agriculture of the University of Puerto Rico at Mayagüez. It contains 240 ha of managed cattle pastures and 190 ha of second-growth (over 50 years old) subtropical moist karstic forest (Ewel and Whitmore 1973) that are dominated by native species (Martinuzzi et al., 2013). There are six forest fragments: four that are small (1-3 ha), and two that are larger (87 and 100 ha). Annual rainfall is 100-125 cm (Daly et al., 2003). Rainfall can occur in any month, but typically there is a February-May dry season and an August-November wet season. Bird-dispersed plant species dominate the plant community, comprising ~70% of recorded species and including trees, shrubs, lianas, and mistletoes. For information on the abundance and characteristics of plant and bird species present at the study site see Carlo and Morales (2016).

Experimental design To quantify how frugivorous birds shape the composition of early successional karstic forests we conducted a seed-addition experiment in ten 6 x 6 m fenced plots (Fig. 1), each plot containing a block of four seed-addition treatments applied to 1.0 m2 experimental patches. These plots were constructed in active cattle pastures of Finca Montaña. Because of the seasonality of our system (wet and dry seasons), the chronological order of seed-additions could affect the composition of the experimental communities if plant species that arrive first interfere or facilitate the establishment of subsequently added species (i.e., priority effects). As we were focused on characterizing the effect of birds on community structure despite seasonality, and not the priority effects, we decided to start half of the plots in February 2013 at the beginning of the dry season, which corresponds to a period of low fruit abundance and diversity in the plant


Accepted Article

community, and other half in September 2013, which corresponds to the wet season and to the peak of fruit abundance and diversity (Carlo & Morales, 2016). This design equally distributes any priority effects due to seasonality. Plots were located 30-180 m from forest edge (see Figure S1 in Supporting Information). Because our goal was to isolate the functional role of frugivores for the composition of regenerating plant communities, our design specifically controlled for the effects of factors such as the soil seed bank, local variation in soil properties, and competition from pre-existing pasture vegetation. First, to control for these factors, we excavated the area of our four 1.0 m2 experimental patches (Fig. 1) to a depth of 0.3 m after moving the vegetation height in each plot to approximately 15 cm. Additionally, we controlled the existing vegetation around the experimental patches by covering each plot with three layers of heavy-duty polyester fabric (Scotts Landscape Fabric, Marysville, Ohio, USA). This included the sides and bottom of the four 1.0 m2 experimental patches to eliminate the effects of the seed bank and/or any remnant underground vegetation that could have invaded patches from the soil (e.g., rhizomes, tubers, etc.). For additional protection and suppression of weed growth around the 1.0 m2 patches, a layer of pine bark nugget mulch (Timberline Pine Bark Nuggets, Oldcastle Retail Inc, Atlanta, GA, USA) was added on top of the fabric, leaving the 1.0 m2 patches uncovered. At the start of the experiment (February for plots 1-5, September for plots 6-10), patches were filled to surface level with local soil (Oxisol, Eutrustox, Coto series, USGS 2006), collected at a single pasture location to control for the effects of soil heterogeneity on plant emergence. To access this soil, we dug a 1.0-m deep trench with a backhoe. All soil was collected from the B-horizon at a depth 0.3-0.5 m to eliminate the influence of germination from the seed bank. An important feature of our experiment is that our seed-addition treatments are intended to address the complex issues of spatial scale along with temporal fluctuations in fruit availability and seed dispersal activity of birds. We randomly assigned the patches within each plot to one of four treatments: natural seed-addition by birds (hereafter “Bird”), manual seedadditions based on seed abundance of ornithochorous plant species at the scale of about 1-km


Accepted Article

radius of the center of Finca Montaña (hereafter “Landscape”), manual seed-additions based on the seed abundance of ornithochorous species found within a 50-m radius of each plot (hereafter “Local”), and a negative control patch to which no seeds were experimentally added (Fig. 1). The Landscape treatment, encompassing the extent of the study site, represents the scale that will contain most of the expected foraging activity of frugivorous birds on a heterogeneous landscape (Morales et al., 2013). We defined the Local treatment as 50-m radius because dispersal is usually a short-distance process that is more strongly influenced by fruit availability nearby (Saracco et al., 2004; Nathan, 2006). Every Bird patch was equipped with a central, three-meter galvanized steel pipe (2.5 cm in diameter). At the top of each pipe, we installed two perpendicular wooden rods (50 cm long, 0.5 cm in diameter) to serve as perches (Fig. 1). We verified the effectiveness of these perches in attracting birds by using video cameras to record the activity and behaviour of visitors (see Caraballo-Ortiz et al., 2017). To sample the seed rain under the bird perches we attached two plastic seed traps (14 cm x 14 cm area) to the pipe 1.5 m from the ground (Fig. 1). For consistency, a 0.5-m pipe of the same type used for the perch in Bird patches was placed at the center of these other patches (Fig. 1). We used bird netting (Bird-X®, Dalen Inc., Knoxville, TN, USA) to exclude birds from the Local, Landscape, and control treatments.

Simulating seed dispersal across time, sites, and scale Seed availability measurements Treatments simulated seed dispersal as chance encounters between birds and fruits available in the environment across times, locations, and the Local and Landscape spatial scales. Availability of seeds (within fruits) was assessed during the first week of each month by counting the number of ornithochorous fruits that were already ripe, or could ripen during that month (Caraballo-Ortiz et al., 2017). Availability for seed-additions in Local treatments was measured by counting ripe fruit on all individuals of each ornithochorous plant species found


Accepted Article

within a 50-m radius from each plot each month, then multiplying the fruit counts of each species by the mean number of seeds per fruit of the species. For the Landscape treatment, we incorporated the density of mature plants, mean number of ripe fruits per individual and mean number of seeds per fruit. For this we first determined the density of reproductive individuals of ornithochorous plant species on the landscape from six 100 x 100 m plots in open areas, eleven 100 x 2 m transects in forest interior areas, and three transects along vegetated fence lines in open areas (Figure S1). To estimate the seed abundance each month at the landscape scale we counted fruits on 345 permanently tagged plants, representing 63 ornithochorous species. The number of individuals that were tagged per plant species ranged from 1 to a maximum of 10 depending on their availability in the study site. The sample included 36 tree species, 17 shrubs, 8 lianas, and one mistletoe. We calculated seed density (per m2) as the product of the mean number of fruits per plant species per month, the density of reproductive plants in the study area and the mean number of seeds per fruit. Rarefaction curve analysis shows that our sample size at each scale (Local, Landscape) was adequate to capture the diversity of the available fruiting plant species in the study site (Figure S2).

Seed-addition treatments In addition to including spatio-temporal variation in fruit availability into our seed-addition treatments, we also considered that birds could add variable amounts of seeds to different locations. Thus, to make seed-addition densities of Local and Landscape treatments equivalent to the density of seeds dispersed by birds at Bird treatment, we scaled the density of seeds found in seed traps to the patch size (1.0 m2) by multiplying the number of seeds in the trap by five. We empirically determined this scaling value from the ratio of seeds falling on a 1.0-m2 plastic fabric placed below perches, to the number of seeds found in seed traps. This scaling value was the average ratio obtained across ten perches installed specifically to estimate the scaling factor.


Accepted Article

We checked seed traps under bird perches and added seeds to Local and Landscape treatments three times a week. To determine the density to be multiplied by the scaling factor, we first counted and identified seeds in the seed trap of the Bird treatment before dropping the seeds from the height of the trap into their soil patch. Seeds added to Local and Landscape treatments were obtained from fruits collected in the study site and cleaned manually to remove pulp. The species and the number of seeds from each species to be included in each addition event were determined by a bootstrapping procedure in which the probability of a given species being included was proportional to its relative availability at the local and landscape scales each month (see Appendix S1). Plant emergence was measured in each experimental patch by tallying the number of emerged individuals of each species every six months (i.e., at six and 12 months from the start of the plot). We calculated cumulative emergence in terms of density of emerging individuals (per m2) and species richness.

Data analyses Seed-additions. First, we assessed the effects of seed-addition treatments on the resulting plant species richness. For this we calculated the average species richness of all seed-addition events per treatment per plot (a conservative approach that eliminates the repeated measure, per plot), then applied a one-way ANOVA where the averaged species richness of seed-additions per plot (response) was a function of treatments (Bird, Landscape, Local, n = 10 for each) and plot (block). We compared means using a Tukey HSD post-hoc test. Control treatments were excluded from analyses given there were no woody species growing on them; these plots served to verify that bird exclosures worked properly. Next, we compared the temporal heterogeneity of the seed-addition treatments by calculating Jaccard’s index of similarity for the seeds added in consecutive months (e.g., the composition similarity between seeds added in April and May,


Accepted Article

May and June, etc.). We took the average of all inter-month comparison values per treatment per plot (again to eliminate the repeated measures, final n = 10 per treatment). We compared the average Jaccard’s index as a function of treatments (Bird, Landscape, Local) and plot (block variable) using non-parametric ANOVA (Kruskal-Wallis, with non-parametric multiple comparisons using Wilcoxon’s method).

Post-dispersal analyses. We compared several emergence and community composition parameters among seed-addition treatments. The species richness of recruited seedlings was modeled with a Generalized Linear Mixed Model (GLMM) for count data (Poisson errors) as a function of addition treatment (fixed effect) and plot (random effect). The ‘post-dispersal richness loss’ was calculated as the difference between the cumulative richness of seeds added, and the cumulative richness of plants emerging in plots from each treatment (n = 10 per treatment). This variable was modeled with a GLMM for proportion data (binomial errors) as a function of addition treatments (fixed effects) and plot (random effect). Multiple comparisons were made in with the multcomp package (Hothorn et al. 2017) in R for both GLMMs. Lastly, we compared the density (per m2) of emerging plants among treatments by using a nonparametric ANOVA (Kruskal-Wallis test with non-parametric multiple comparisons using Wilcoxon’s method) because of non-normality of data, including plots as blocks. To examine how plant traits such as plant species’ seed mass and phenophase length (i.e., the length of the fruiting season) affected the emergence we used a Generalized Linear Mixed Model (binomial errors, lme4 package in R), where the proportion of emerging seedlings (response) was modeled as a function of treatment, average species’ seed mass, phenophase length (number of months bearing ripe fruit), with plot and plant species as random effects. We included all two-way interactions and deliberately left out three-way effects (sensu Scheiner & Gurevitch 1993). To analyze the structure of the communities of bird-dispersed plants resulting from seed-addition treatments we used Non-Metric Multidimensional Scaling (vegan package in R; Oksanen et al., 2018) with Chao’s distance index. The data matrix consisted of 24 columns of plant species (all


Accepted Article

plant species emerged in at least one experimental patch), and 25 rows of experimental patches where at least one plant emerged (Bird =9, Landscape = 9, Local = 7). All analyses were performed in R version 3.4.3 (R Core Team, 2017).

Results Seed-additions and fruit availability. In total, seeds from 53 plant species were added in our experiment. From these, birds added 4889 seeds from 45 plant species (including 7 morphospecies). Density-equivalent experimental additions resulted in reduced total plant species richness added to the Landscape (37 species) and Local treatments (21 species, Figure S3). This overall higher richness resulting from avian seed dispersal across all experimental plots contrasted with the higher average species richness added to the Landscape treatment each month compared to Bird and Local treatments (Fig. 2A; F 2,29 = 102.7, p < 0.0001). However, there was a significantly lower similarity of species composition (Jaccard’s index) between seed-additions across successive months in Bird treatments, showing a greater month-to-month species’ turnover than in the Landscape and Local treatments (Fig. 3; χ2 = 17.8, df = 2, P < 0.0001). The richness of ornithochorous plant species producing fruits in the study site varied little between months (Fig. 2B, black line). On the other hand, density of available fruits showed a high temporal variability (Fig. 2B, gray line) and had no correlation with the number of species added to any treatment.

Plant emergence after seed-additions. Overall, 26 plant species emerged across all the experimental patches up to one year after the start of each experimental plot (Fig. 4). Experimental addition patches were early-successional communities where many of the emerged species became established as saplings or small trees, with some individuals reaching sexual maturity while the experiment was still running. Across all experimental patches, the Bird addition treatment had the highest cumulative richness with 24 emerging plant species,


Accepted Article

which was 33% more emergence than the Landscape treatment, and more than twice the emergence richness at Local treatments. When examining seed-to-seedling transition, the average richness of emerged plants was the highest for the Bird treatment, intermediate for Landscape, and the lowest for Local (Fig. 5A; χ2 = 12.6, df = 2, P = 0.0018). The Landscape treatment suffered more loss of species richness than the Bird and the Local treatments (Fig. 5B;

χ2 = 24.45, df = 2, P < 0.0001). Plant recruits achieved the highest density in Bird treatments, followed by Local and Landscape treatments (Fig. 5C; χ2 = 8.95, df = 2, P = 0.0114). Seedling emergence was significantly affected by species seed mass and the additiontreatment, with significant two-way interactions (Table 1, Tables S1 & S2). Seed mass had a strong positive effect on emergence, especially for plant species with large seeds added by birds (Figure S4, Table S2). However, birds also dispersed very small-seeded plants with high emergence. In addition, seed mass had a negative effect on emergence, but only in Landscape treatments (thus the interaction). We also found a significant two-way interaction between addition-treatment and species’ fruiting phenophase length for seedling emergence (Table 1, Table S2). Plant species with long phenophases that were frequently dispersed by birds were more likely to establish, compared to other species with long phenophases that were less frequently dispersed by birds, but frequently added in the local and landscape seed-addition treatments because of their high availability (Table S2).

When examining the composition of the emergent plant communities, we find the largest structural differences between the Local and Bird treatments (Fig. 4). The NMDS best solution was two-dimensional with a final stress of 0.17. Communities were not differentiated along axis one, but axis two separated the Bird treatment plots (negative scores on axis two), from most Local treatment plots (positive scores on axis two). The ordination plot also shows groups of plant species more strongly associated with the centroid of each treatment (i.e., species names lying within the 95% confidence ellipses). For example, some species were more


Accepted Article

strongly associated with Bird treatments, either because they were more common or because they were exclusively found in Bird plots. The less diverse Local treatments were characterized by species common in open areas, while Landscape treatments did not show characteristic species associations in the ordination space (Fig. 4).

Discussion Here we experimentally addressed the relationships of frugivory to the dynamics of fruit availability and plant species’ traits in shaping regenerating plant communities on cleared tropical lands. Results support the hypothesis that frugivory and seed dispersal patterns depart from random frugivore-plant encounters in the environment. This leads to early-regenerating plant communities with a characteristic "frugivore signature" in their patterns of species composition and abundance, including an over-representation of many proportionally rare plant species, and decreases in the dominance of many common species.

Seed rain structure Our results show that frugivores impart characteristics to seed rains that can translate into compositional differences of early-regenerating plant communities. Specifically, the biases of frugivores resulted in a more temporally heterogeneous seed rain compared to when dispersal was made proportional to the relative availability of fruiting species in the community (Fig. 3). These frugivore biases resulted in higher species richness and transmitted to distinct dominance patterns in the early successional plant communities (Fig. 4). Thus, revising our understanding of how frugivores shape plant community structure is important given that ecologists have placed more emphasis on post-dispersal processes such as predation, competition, and abiotic stresses in identifying the mechanistic drivers of community structure (Weiher & Keddy, 1995; Chesson, 2000; Terborgh, 2012; Vellend, 2016).


Accepted Article

Our experiment identifies novel properties of frugivore-generated seed rains, revealing the functional importance of frugivores for the structuring of plant communities. First, we found that several rare plant species were present in the bird-generated seed rain at higher frequencies than expected by their abundance. This illustrates an anti-apostatic (sensu Allen et al., 1998) dispersal bias toward many plant species of low relative abundance, producing a more diverse seed rain as compared to what is available in the environment. Second, disproportionally high dispersal rates were also found for some common plant species (Figure S3). Functionally speaking, the general signature of avian frugivory for community-wide seed rains may be characterized by the overrepresentation of not only a few common, but also many rare plant species. This feature may constitute one of the drivers that explain why many tropical plant communities with a high prevalence of ornithochorous species show a hyperdominance of a few species, yet still maintain a high diversity because of many rare species (ter Steege et al., 2013). Phenotypic trait matching (Jordano, 2000; Olesen et al., 2011) and nutrient complementarity effects (Morán-López et al., 2018) are the likely functional mechanistic underpinnings of the frugivore signature on seed rain. Future studies could examine if these patterns also exist in other environments such as a mature tropical forest with higher frugivore diversity. Non-random interactions of frugivores with the plant community can also explain the apparent contradiction of the cumulative species richness of seed-additions being higher for the Bird treatment than the Landscape treatment. Since our Local and Landscape seed-addition treatments were made in accordance to fruit availability, plant species that dominated seedadditions between consecutive months tended to be the same dominant species that have long fruiting phenophases (Caraballo-Ortiz et al., 2017). On the other hand, rare-biased dispersal by birds integrated different rare plant species as they became available, leading to more heterogeneous seed-additions between months (Fig. 3). It is still possible that the seed rain patterns of the Bird treatment match the fruit availabilities of an intermediate spatial scale between our Local and Landscape scales. However, this is unlikely because the anti-apostatic


Accepted Article

biases in the dispersal of rare plant species remain constant given that plant species that are rare at the Local and Landscape scale will also be rare at intermediate spatial scales (Morán-López et al. 2018).

Recruitment limitations following dispersal Curiously, the Landscape treatment had the lowest densities of emerged plants, suggesting that the Landscape treatment was more recruitment-limited than the Bird or the Local treatments (Fig. 5C). Depulping fruits by hand rather than by bird-guts could have reduced seed germination (Traveset et al., 2008). However, the Local treatment also used manually-depulped seeds, yet had similar emergence and density to the frugivore gut-treated seeds of the Bird treatment (Fig. 5C). More likely, the environmental conditions at the experimental patches favored the emergence of ruderal species already common in open habitats and dispersed by birds (Graham & Page, 2012). Emerging communities recorded in the Bird treatment showed a mixture of small- and large-seeded species (Figure S4), contrasting with studies reporting a positive relationship of seedling emergence with seed mass (e.g., Muñoz et al., 2017). This may be explained by the dominant role of small birds (e.g., tyrant flycatchers and mockingbirds, see Carlo and Morales 2016) in dispersing seeds. Thus, our findings might be explained in part by trait matching between small-bodied birds and small-seeded fruits that are common in fragmented tropical landscapes (Emer et al., 2018). In fact, other studies have reported how the general positive relationship between seed mass and seedling emergence can sometimes be offset by higher dispersal rates of small-seeded species (Moles & Westoby, 2004). Still, large bird species may also be important for plant colonization of degraded areas by enhancing long-distance seed dispersal of plant species having large seeds (González-Varo et al., 2014).


Accepted Article

Although long fruiting periods may enhance the chance of plants to interact with frugivore agents (González-Castro et al., 2012) and to encounter favorable environmental conditions to germinate and survive (Carlo & Tewksbury, 2014), the effect of fruiting phenophase length on emergence was dependent on the seed-addition treatment. This could explain why some plant species with a long fruiting phenophase had a high emergence in the experimental plots, whereas other plant species – also with long fruiting periods – showed a low emergence in the experimental plots. These results suggest that additional species-specific factors affecting seedling emergence (e.g., pre- and post-dispersal seed predation, competition, etc.) are yet to be quantified, although it remains a challenge to integrate a multiplicity of processes within a single experimental framework.

In summary, dispersal by birds resulted in distinct early-successional species assemblages compared to our two random seed-addition treatments of ornithochorous species, especially the Local treatment. Studies show that – through dispersal activities – frugivorous birds help create associations of ornithochorous plants (Verdú & García-Fayos, 2002; Lázaro et al., 2005; García & Martínez, 2012; Carlo & Tewksbury 2014). Our study offers a new perspective on how frugivorous birds finely structure these plant communities, given that vast numbers of ornithochorous species allow a myriad of combinatorial possibilities. This avian influence on the early-regenerating community is characterized by non-random groups of plant species whose representation is not purely driven by their relative abundances in the environment.

ACKNOWLEDGMENTS We thank L. Añeses, A. Rodríguez, A. Casas, D. Cianzio (RIP) and UPR-Mayagüez. R. Glenn, J.A. Colón, A. Morales, H. Kahl, M.A. Caraballo-Ortiz, R. Gallagher, N. Pérez, and Y. Gavilán helped to conduct field work. Dr. Matthias Schleuning and two anonymous referees made


Accepted Article

useful comments and suggestions on a previous version of this manuscript. NSF grants DEB1145994 award to T.A.C. funded the work. A.G-C. thanks Cabildo of Tenerife, Program TF INNOVA 2016-21 (with MEDI & FDCAN Funds).

AUTHORS’ CONTRIBUTION T.A.C. and S.Y. conceived the idea and designed the study. A.G.C & T.A.C. collected the data and all authors contributed to the analyses. T.A.C. led the writing with all authors contributing to the final version.

SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article. Appendix S1. Example of bootstrapping to make seed-additions. Figure S1. Aerial view of the study site. Figure S2. Rarefaction curve analysis. Figure S3. Average number of seeds added per plot and treatment. Figure S4. Average plant emergence per plot per treatment. Table S1. Ornithochorous plant species monitored during the study. Table S2. Parameter Estimates of the Generalized Linear Mixed Model.


Accepted Article

DATA ACCESSIBILITY Data deposited in the Dryad repository: doi:10.5061/dryad.55p11md (González-Castro et al. 2018)

REFERENCES Allen, J. A., Raison, H. E., & Weale, M. E. (1998). The influence of density on frequencydependent selection by wild birds feeding on artificial prey. Proceedings of the Royal Society B, 265, 1031–1035. Bascompte, J., & Jordano, P. (2007). Plant-animal mutualistic networks: the architecture of biodiversity. Annual Review of Ecology, Evolution, and Systematics, 38, 567–593. Caraballo-Ortiz, M. A., González-Castro, A., Yang, S., dePamphilis, C. W., & Carlo, T. A. (2017). Dissecting the contributions of dispersal and host properties to the local abundance of a tropical mistletoe. Journal of Ecology, 105, 1657–1667. Carlo, T.A., & Yang, S. (2011). Network models of frugivory and seed dispersal: Chalenges and opportunities. Acta Oecologica, 37, 619–624. Carlo, T.A. & Morales, J.M. (2016) Generalist frugivores kick-start diverse tropical forest regeneration via anti-apostatic seed dispersal. Ecology 97: 1819–1831. Carlo, T. A, & Tewksbury, J. J. (2014) Directness and tempo of avian seed dispersal increases emergence of wild chiltepins in desert grasslands. Journal of Ecology, 102, 248–255. Chazdon, R. L. (2014) Second Growth: the promise of tropical forest regeneration in an age of deforestation. Chicago, IL: The University of Chicago Press. Chesson, P. (2000) Mechanisms of maintenance of species diversity. Annual Reviews in Ecology and Systematics 31:343–366.


Accepted Article

Condit, R., Pitman, N., Leigh, E.G., Chave, J., Terborgh, J., Foster, R.B., Núñez, P.,Aguilar, S., Valencia, R. Villa, G., Muller-Landau, H.C., Lossos, E., & Hubbell, S.P. (2002) βdiversity in tropical forest trees. Science 295: 666–669. Daly, C., Helmer, E. H., & Quiñones, M. (2003). Mapping the climate of Puerto Rico, Vieques, and Culebra. International Journal of Climatology, 23, 1359–1381. Da Silva, J. M. C., Uhl, C., & Murray, G. (1996), Plant Succession, Landscape Management, and the Ecology of Frugivorous Birds in Abandoned Amazonian Pastures. Conservation Biology, 10, 491–503. Emer, C., Galetti, M., Pizo, M.A., Guimarães Jr, P.R., Moraes, S., Piratelli, A., et al. (2018). Seed-dispersal interactions in fragmented landscapes - a metanetwork approach. Ecology Letters, 21, 584–493. Ewel, J. J., & Whitmore, J. L. (1973) The ecological life zones of Puerto Rico and the US Virgin Islands. US Forest Service Research Paper ITF-18. Institute of Tropical Forestry, Rio Piedras, Puerto Rico. Foster, S. A., & Janson, C. H. (1985). The Relationship between Seed Size and Establishment Conditions in Tropical Woody Plants. Ecology, 66, 773–780. García, D. & Ortiz-Pulido, R. (2004). Patterns of resource tracking by avian frugivores at multiple spatial scales: two case studies on discordance among scales. Ecography, 27, 187–196. García, D., & Martínez, D. (2012). Species Richness matters for the quality of ecosystem services: a test using seed dispersal by frugivorous birds. Proceedings of the Royal Society B, 279, 3106–3113.


Accepted Article

González-Castro, A., Yang, S., Nogales, M., & Carlo, T. A. (2012). What determines the temporal changes of species degree and strength in an oceanic island plant-disperser network? PLoS One, 7:e41385. González-Castro, A., Yang, S., & Carlo, T. A. (2018). Data from: How does avian seed dispersal shape the structure of early successional tropical forests? Dryad Digital Repository. doi:10.5061/dryad.55p11md González-Varo, J. P., Carvalho, C. S. & Jordano, P. (2017). Unrevealing seed dispersal through fragmented landscapes: Frugivore species operate unevenly as mobile links. Molecular Ecology, 26, 4309–4321. Graham, L. L. B., & Page, S. E. (2012). Artificial bird perches for the regeneration of degraded tropical peat Swamp Forest: A restoration tool with limited potential. Restoration Ecology 20: 631–637. Gotelli, N. J. (2000). Null modell analysis of species co-occurrence patterns. Ecology, 81, 2606–2621. Holl, K. D. (1998). Do bird perching structures elevate seed rain and seedling establishment in abandoned tropical pasture? Restoration Ecology, 6, 253–261. Hothorn, T., Westfall, P., Heiberger, R. M., Schuetzenmeister, A., & Scheibe, S. (2017). Packacke ‘multcomp’. R package version 1.4-8. https://cran.rproject.org/web/packages/multcomp/multcomp.pdf Jordano, P. (2000). Fruits and frugivory. In M. Fenner (Ed.), Seeds: the ecology of regeneration in plant communities (pp. 125–166). Wallingford, UK: CABI Publishing. Lázaro, A., Mark, S., & Olesen, J. M. (2005). Bird-made orchards in northern Europe: nestedness and network properties. Oikos, 110, 321–329.


Accepted Article

Levey, D. J., & Martinez del Rio, C. (2001) It takes guts (and more) to eat fruits: lessons from avian nutritional ecology. Auk 118:819-831. Martinuzzi, S., Lugo, A. E., Brandeis, T. J., & Helmer, E. H. (2013). Case study: geographic distribution and level of novelty of Puerto Rican forests. In R. J. Hobbs, E. S. Higgs & C. M. Hall (Eds.), Novel ecosystems: Intervening in the new ecological world order (pp. 81–87). Chichester: John Wiley & Sons. Ltd. Moles, A. T, & Westoby, M. (2004). Seedling survival and seed size: a synthesis of the literature. Journal of Ecology, 92, 372–383. Muñoz, M.C., Schaefer, H.M., Böhning-Gaese, K. & Schleuning, M. (2017). Importance of animal and plant traits for fruit removal and seedling recruitment in a tropical forest. Oikos, 126, 823-832. Morales, J. M., García, D., Martínez, D., Rodríguez-Pérez, J. & J. M. Herrera. (2013). Frugivore behavioural details matter for seed dispersal: a multi-species model for Cantabrian thrushes and trees. PLoS Onee 8(6): e65216. Morán-López, T., Carlo T.A., Amico, G., & Morales J. M. (2018). Diet complementation as a frequency-dependent mechanism conferring advantages to rare plants. Functional Ecology. doi.org/10.1111/1365-2435.13152 Muller –Landau, H. (2008). Colonization-related tradeoffs in tropical forests and their role in the maintenance of plant species diversity. In W. P. Carson & S. A. Schnitzer (Eds.), Tropical forest community ecology (pp. 182–195). Chichester, UK: Wiley-Blackwell. Nathan, R. (2006). Long-distance dispersal of plants. Science, 313, 786–788. Norden, N., Daws, M. I., Antoine, C., Gonzalez, M. A., Garwood, N.C., & Chave, J. (2009). The relationship between seed mass and mean time to germination for 1037 tree species across five tropical forests. Functional Ecology, 23, 203–210.


Accepted Article

Oksanen, J., Blanchet, F.G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., et al. (2018). vegan: Community Ecology Package. R package version 2.4-6. https://CRAN.Rproject.org/package=vegan Olesen, J. M., Bascompte, J., Dupont, Y. L., Elberling, H., Rasmussen C., & Jordano, P. (2011). Missing and forbidden links in mutualistic networks. Proceedings of the Royal Society B, 278, 725–732. R Core Team. (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. Saracco, J. F., Collazo, J. A., & Groom, M. J. (2004). How do frugivores track resources? Insights from spatial analyses of bird foraging in a tropical forest. Oecologia, 139, 235– 245. Scheiner, S. M., & Gurevitch J. (1993). Design and analysis of ecological experiments. Chapman and Hall, New York, New York, USA. Schupp, E. W., & Fuentes, M. (1995). Spatial patterns of seed dispersal and the unification of plant population ecology. Ecoscience, 2, 267–275. ter Steege, H. T., Pitman, N. C. A., Sabatier, D., Baraloto, C., Salomão, R.P., Guevara, J. E., et al. (2013). Hyperdominance in the Amazonian Tree Flora. Science 342, 1243092. doi: 10.1126/science.1243092. Terborgh, J. (2012). Enemies maintain hyperdiverse tropical forests. The American Naturalist, 179, 303–314. Traveset, A., Rodriguez-Perez , J., & Pias, B. (2008). Seed trait changes in dispersers' guts and consequences for germination and seedling growth. Ecology, 89, 95–106. Verdú, M., & García-Fayos, P. (2002). Frugivorous birds mediate sex-biased facilitation in a dioecious nurse plant. Journal of Vegetation Science, 14, 35–42.


Accepted Article

Vellend, M. (2016) The theory of ecological communities. Princeton University Press, Princeton, NJ. Weiher, E., & Keddy, P. A. (1995). Assembly rules, null models, and trait dispersion: new questions from old patterns. Oikos, 74, 159–164. Yang, S., Albert, R., & Carlo, T. A. (2013) Transience and constancy of interactions in a plantfrugivore network. Ecosphere 4: Article 147, pp. 1-25.

Table 1. Effects table from a Generalized Linear Mixed Model (Type II ANOVA, for parameter estimates see Table S2) examining the effects of seed-addition treatment, plant species’ average seed mass (per seeds) of added plant species, the length of the phenophase (number of months with fruits), and two-way interactions terms, on the emergence of plants in experimental plots in Finca Montaña, Puerto Rico, during Feb 2013 - Sept 2014.

Seedling Emergence – Generalized Linear Mixed Model (binomial family)

Fixed Effects:

Df

χ2

P > χ2

Seed_Addition_Treat.

2

35.77

< 0.0001

Seed_Mass

1

10.35

0.0013

Phenophase

1

0.40

0.5269

Seed Addition_Treat:Seed_Mass

2

34.44

< 0.0001

Seed Addition_Treat:Phenophase

2

10.09

0.0064

Seed_Mass:Phenophase

1

0.02

0.7889


Accepted Article

Random Effects: Groups

Variance

Std. Dev.

Plant sp.

2.11

1.45

Site

0.58

0.76

FIGURE LEGENDS Figure 1. The experiment (A) compared the seed-additions and establishment of plant communities in natural Bird addition treatments (B) and manual seed-additions (C). Manual seed-additions simulated random seed dispersal of available fruiting species at a Local scale (i.e., 50 m radius from plots) or Landscape scale (i.e. whole study site). Patches started with bare and seedless soil (D). The structure of the plant communities in each treatment was quantified and compared after one year (E).

Figure 2. Mean species richness (± s.e.) of monthly seed-addition (A) in treatments across the ten experimental plots in Finca Montaña, Puerto Rico, from 2013-2014. (B) Species richness and fruit density of available fruits.

Figure 3. For each seed-addition treatment, the Jaccard’s index of similarity (average ± s.e.), was used to compare plant species composition between consecutive months. Different letters at the top of bars mean significant differences for pairwise comparisons between addition treatments.


Accepted Article

Figure 4. NMDS ordination plot of the community of recruited plants from seed-addition treatments (Bird, Landscape and Local). The dotted ellipses denote the 95% confidence intervals for centroids of each treatment. The plant species’ average position (names) and each experimental patch locations (symbols) are also shown.

Figure 5. (A) Average plant species richness (± s.e.) at emergence stage. (B) Average (± s.e.) proportion of plant species that were added as seeds but did not emerged as seedlings (i.e., species loss). (C) Average density (± s.e.) as seedlings per m2 of emergent plants. Letters indicate significant differences for pairwise treatment comparisons.

Figure 1


Accepted Article

Figure 2

Accepted Article

Figure 3


Accepted Article

Figure 4


Accepted Article

Figure 5
