Assessing the impact of deforestation of the Atlantic rainforest on ant-fruit interactions: a field experiment using synthetic fruits

Abstract

Ants frequently interact with fleshy fruits on the ground of tropical forests. This interaction is regarded as mutualistic because seeds benefit from enhanced germination and dispersal to nutrient-rich microsites, whereas ants benefit from consuming the nutritious pulp/aril. Considering that the process of deforestation affects many attributes of the ecosystem such as species abundance and composition, and interspecific interactions, we asked whether the interaction between ants and fallen fleshy fruits in the Brazilian Atlantic forest differs between human-created fragments and undisturbed forests. We controlled diaspore type and quantity by using synthetic fruits (a plastic 'seed' covered by a lipid-rich 'pulp'), which were comparable to lipid-rich fruits. Eight independent areas (four undisturbed forests, and four disturbed forest fragments) were used in the field experiment, in which we recorded the attracted ant species, ant behaviour, and fruit removal distance. Fruits in undisturbed forest sites attracted a higher number of species than those in disturbed forests. Moreover, the occurrence of large, fruit-carrying ponerine ants (Pachycondyla, Odontomachus; 1.1 to 1.4 cm) was higher in undisturbed forests. Large species (≥3 mm) of Pheidole (Myrmicinae), also able to remove fruits, did not differ between forest types. Following these changes in species occurrence, fruit displacement was more frequent in undisturbed than in disturbed forests. Moreover, displacement distances were also greater in the undisturbed forests. Our data suggest that fallen fleshy fruits interacting with ants face different fates depending on the conservation status of the forest. Together with the severe loss of their primary dispersers in human-disturbed tropical forest sites, vertebrate-dispersed fruits may also be deprived of potential ant-derived benefits in these habitats due to shifts in the composition of interacting ant species. Our data illustrate the use of synthetic fruits to better understand the ecology of ant-fruit interactions in variable ecological settings, including human-disturbed landscapes.

Figure 1. Study area location at the municipalities of Piedade and Tapiraí, São Paulo State, Brazil.

The rectangle with dotted borders defines the fragmented disturbed landscape where we selected four forest fragments of ca. 100‘Parque Estadual de Jurupará’. The white area situated within the four continuous forest sites represents the Jurupará Dam. Modified from Banks-Leite et al. (2010).

Figure 2. Behavioural interactions between ants and lipid-rich synthetic fruits in the Atlantic forest, Southeast Brazil.

(A) Recruited workers of a large Pheidole species (≥3 mm) displacing a ‘fruit’; individual workers of (B) Pachycondyla striata (∼1.2 cm) and of (C) Odontomachus chelifer (∼1.4 cm) carrying a synthetic fruit to the nest. See also Table S1.

Figure 3. Ant species per station baited with synthetic fruits in undisturbed and disturbed Atlantic forest sites.

Thirty stations were sampled per site. Central lines and borders in each box-plot indicate the mean value and the mean ± its standard error; whiskers delimit the range of the 95% confidence interval.

Figure 4. Non-metric multidimensional scaling ordination of the ant community in undisturbed and disturbed Atlantic forest sites.

Thirty stations baited with synthetic fruits were sampled in four undisturbed (white circles; UFs) and four disturbed forest sites (gray circles; DFs). Ordination was based on Bray–Curtis index of similarity between studied sites for presence-absence data.

Figure 5. Ant attendance and ant behaviour toward synthetic fruits.

(A) Occurrence of particular ant groups and (B) occurrence of beneficial behaviours at sampling stations. Experiments with synthetic fruits were carried out in two diverging forest types in the Atlantic forest, Southeast Brazil, undisturbed (white circles; four sites) and fragmented forest sites (gray circles; four sites). The ant groups (large ponerines and large Pheidole spp.) were those whose behaviours were considered as potentially beneficial to ‘seeds’ (either removing the entire ‘fruit’ or cleaning the ‘pulp’ in situ) during the 22-hour experiment. The number of stations (y-axis) in (B) corresponds to those stations having at least one of the five seeds either removed or cleaned by ants. Asterisks indicate significant differences (p<0.5) between forest types.

Figure 6. Dispersal of synthetic fruits by ants.

(A) Removed synthetic fruits and (B) ‘fruit’ displacement distance in a fragmented Atlantic forest landscape. Experiments with synthetic fruits were carried out in two forest types: four undisturbed sites and four disturbed forest sites. Thirty stations were sampled per site. Each sampling station received five synthetic fruits at experiment beginning. Medians are represented by squares, while the lower and upper whiskers represent the 25% and 75% quartiles respectively. Outliers (values smaller or larger than 2 times the 75% minus the 25% quartiles) are represented by unfiled circles.