1. Mutualistic interactions between frugivorous birds and fleshy fruited plants are key processes for both natural plant regeneration and the maintenance of birds relying on fruit resources. However, seed dispersal effectiveness (SDE) has been frequently studied only from the plant’s perspective, i.e. the contribution of animals to plant fitness.

2. Using a sample of Atlantic rainforest avian frugivores, this study presents a first comparative, empirical study of fruit effectiveness as a nutritional food resource for seed dispersing birds through the use of Resource-Provisioning Effectiveness (RPE) models. RPE is the product of the amount of fruits a frugivore can consume (quantity component) and the quality of the ingesta in terms of energy and nutrients obtained (quality component).

3. Our results show wide variation in Resource-Provisioning Effectiveness among fleshy-fruited plant species. Energy-rich fruits consistently show a smaller quantity component, while energetically-poor fruits are consistently the most consumed, with fruit species spanning a gradient from these two extremes.

4. The specific RPEi resulting from a pairwise fruit-frugivore interaction is positively correlated to the total RPE (RPET) that a given fruit species has for the whole frugivore assemblage. RPE therefore appears to be a characteristic feature of the fruit species, rather than of the specific frugivore partner.

5. Only the fruit’s specific energy content showed a significant phylogenetic signal, suggesting potential constraints for free covariation between RPE and SDE of fruits and frugivores.

6. Synthesis: We analyze variation in the effectiveness of fleshy fruit food provisioning to avian frugivores by explicitly redefining Resource-Provisioning Effectiveness within the Seed Dispersal Effectiveness framework. We found ample variation in RPE among plant species, showing differences in both quantity and quality components of fruit resources rewards for the frugivores. Our findings help unravel how seed-dispersing birds may discriminate among alternative fruit resources and to understand the configuration of mutual dependencies among mutualistic partners.

Using a sample of Atlantic rainforest avian frugivores we analyze fruit effectiveness as a nutritional food resource for seed dispersing birds, i.e., Resource-Provisioning Effectiveness (RPE), an overlooked aspect of seed dispersal effectiveness. We found ample variation in RPE among plant species, given not just variation in fruit design and nutrient content but also in fruit availability, accessibility and other features that determine how rewarding a fruit resource is for the frugivore. Yet fruit species show consistent RPE values despite highly variable outcomes of interactions with different avian frugivores.

The rationale for our study includes four steps: 1) selecting a representative sample of frugivore species illustrating both the full range of body mass and phylogenetic diversity of the Brazilian Southeastern Atlantic forest frugivores; 2) compiling a literature dataset on fruit foraging and usage data for these frugivore species; 3) analyze variation in RPE across selected plant species by building effectiveness landscapes from the frugivore’s perspective; and 4) experimentally assess RPE subcomponents (i.e. fruit pulp digestibility) that refine the estimates of RPE variation among plant species.

To estimate the RPE for fruit species consumed by each frugivore species we have defined its two components: the rate at which the fruits are ingested/handled by the animal (feeding rate) as the quantity component, and the energy gained from eating the fruit (fruit energy) as the quality component. The feeding rate is calculated by multiplying the bird visitation frequency to the plant by the number of fruits consumed in each visit; from a frugivore perspective these subcomponents of RPE-quantity illustrate potential encounter rates with the fruit resource (visitation) and potential handling/ingestion rates once the fruits are encountered (fruits/visit). Fruit-pulp energy yield is assessed using the specific energy (KJ/g) multiplied by the fruit pulp dry mass. Since the nutrient proportional contents of the fruit pulp are calculated based on its dry mass, we use this value as the profitable fruit mass per fruit (Herrera 1981). Specific energy was quantified using the following energy conversion factors for fruits (FAO, 2002): 14.1 KJ/g for proteins, 35 KJ/g for lipids and 15.1 KJ/g for carbohydrates. Ultimately, by multiplying the two components we obtained the total effect value for each specific interaction (RPE; Schupp, 1993; Schupp, Jordano, & Gómez, 2017):

RPE_{[g energy assimilated]} = [no fruits consumed/obs time (h)] × [g energy accumulated/fruit consumed]                (1)

Note that the RPE estimate for a given fruit species is specific for its interaction with a frugivore, so we used the average RPE values of a given fruit of all the frugivore species it provides food to, to characterize its overall or total effectiveness (RPE_T):  

RPE_{T= sum(}RPE_i)/N                                                (2)

where, for a given focal fruit species, N is the number of pairwise interactions with RPE data available, and RPE_i is the RPE value for a specific pairwise interaction i. Thus, RPE_T is fruit species-specific, while each pairwise interaction has a specific RPE value (RPE_i) depending on the identity of the specific frugivore partner.

Data collection

RPE Quantity component

Most of the data used to calculate the quantitative component have been obtained from available bibliographic sources. Data compilation comes from a total of 51 studies and four databases for frugivore-plant interactions from the Brazilian Atlantic Forest (see Data Sources section for a list of data sources used and Supp. Mat. Fig. S1 for a map of study locations). Variables collected from the bibliography were: number of visits, observation time and number of fruits consumed per visit. To reduce bias we divided the total number of visits to the plants by observation time to control for the different duration of each study. The fruit mass ingested per visit was positively correlated with frugivore body mass (Pearson’s correlation r = 0.588, p < 0.001, n = 541 distinct pairwise interactions; Supp. Mat. Fig. S2). Fruit mass-body mass correlation allowed the estimation of the number of fruits consumed per visit for avian species with no data on fruit consumption rates available; this was done for special cases when avian species had limited number of records in most studies (n = 7 fruit species for Aburria jacutinga, n = 2 fruit species for Penelope obscura, n = 3 fruit species for Procnias nudicollis).

We have referred to a fruit as all the dispersing and consequently ingested units (i.e. diaspora). Diaspora and fruit are often the same thing; however, in some cases such as Cabralea canjerana or Virola spp., an aril (i.e. a fleshy covering in some seeds) acts as a diaspore, being smaller than the actual fruit. Therefore, when the diaspore was actually different from the fruit, the energy in the qualitative component has been measured accordingly. In other cases frugivores may peck pieces and ingest just a part of the whole fruit or infructescence (e.g. catkins of Cecropia spp. or syconia from Ficus spp.). For those large infructescences that birds do not consume whole, we used the number of pecks and assessed the percentage of a single fruit actually consumed, corresponding to a given number of pecks and the beak size of the frugivore. Data on each species’ gape size and fruit length (obtained from Bello et al., 2017 and Galetti et al., 2013) allowed us to estimate the total number of pecks needed to consume an infructescence (Supp. Mat. Table S1).

RPE Quality component

All fruit pulp dry masses and nutrient proportions were obtained from available databases (Pizo, 2017 unpublished; Jordano 1995, 2007; Bello et al., 2017). In a few cases, pulp dry mass values were missing; these were estimated using additional species-specific data such as pulp fresh mass or water proportion (6 species records imputed in this way; see Supp. Mat. Table S2). However, when species-specific data were unavailable, we estimated the specific values by averaging the data available for congeneric species. This type of data imputation was done for a few cases and only when essential to retain a given species in the dataset (9 species records were imputed this way; see Supp. Mat. Table S3).

When calculating specific energy, we preferably used the non-structural carbohydrate content for the carbohydrate value; however, when this information was not available total carbohydrate content or total sugar content was used (see Suppl. Mat. Table S4).

Due to the diverse origin of nutritional data, variation in the techniques and devices used for their analyses is likely. This variation in methods together with the use of different carbohydrate content variables have limitations when comparing energetic values among fruits. We therefore want to highlight these possible limitations and advise to take results cautiously. While the quality component gives us an approximate idea of the nutritional value of each fruit it does not offer a detailed ranking.

A total of 315 different fruit species were reported to be eaten by the frugivore species considered in this study, of which 36 fruits had enough data to calculate the quantity and quality component.

Please visit the GitHub repository for this dataset to find updates and additional information: https://github.com/PJordano-Lab/MS_JEcol_Resource-provisioning-effectiveness.