Figure 1.
This nest houses the ants Camponotus femoratus and Crematogaster levior and the epiphytic plants Peperomia macrostachya and Codonanthe uleana Fritsch (purple fruit). Such gardens are established when ants embed seeds of AG epiphytes into their arboreal carton nests.
Table 1.
Ant species recorded at general food baits and at ant-garden seed baits on transects in southeast Perú.
Figure 2.
Fate of P. macrostachya seeds placed at 108 sampling stations.
Most P. macrostachya seeds remained undispersed during the bait study. Three ant species removed 390 seeds, and the AG ant C. femoratus was responsible for the vast majority of observed dispersal.
Figure 3.
Arthropod and vertebrate exclusion experiments.
(A) Close-up of a seed tray treated with both Tanglefoot and vertebrate-exclusion cage. (B) Experimental design showing the four treatments presented in random positions in 1 m2 plots.
Figure 4.
Removal of P. macrostachya seeds during exclusion experiments.
Bars are proportion of seeds removed from each treatment±SE, based on untransformed counts of seeds removed. Black bars (A) represent plots placed within AG territories, and gray bars (B) represent plots placed outside of AG territories. Each bar represents results for 240 seeds (15 seeds per exclusion treatment per block per day, over four blocks and four days).
Table 2.
Results of ANOVA testing for fixed effects of cage, Tanglefoot and C. femoratus (“AG ants”) in the exclusion experiment.
Figure 5.
Actual and putative seed fate pathways for P. macrostachya seeds.
Estimated probability of events documented in the bait study, exclusion experiment, and garden census are noted by percent values next to solid lines (see footnotes below). Solid lines lacking percent values have been reported anecdotally in the literature or observed by E.Y., while pathways represented by dotted lines are proposed but undocumented. Seeds may be dispersed directly from AG plants by both ants and mammals, or they may fall to the ground. Seeds on the ground in AG territories are retrieved primarily by C. femoratus. If seeds undergo long distance dispersal, as when they are consumed by flying or arboreal mammals, they may also be deposited on the forest floor where they can die, be retrieved by non-AG ants or AG ants far from the original colony. Incorporation into AG carton represents a seed's best, but still unlikely, opportunity for survival. Sources of percent values: a. 65% of P. macrostachya seeds were removed from within AG territories in the one-day bait study; 89% of P. macrostachya seeds were removed from within AG territories in the four-day exclusion experiment. b. 35% of seeds were not removed from within AG territories in the bait study; 11% were not removed in the exclusion experiment. We assume that un-removed seeds germinate in place and die. c. In the bait study, we did not observe non-AG ants removing P. macrostachya seeds within AG territories. However, Dolichoderus bispinosus may have done so, and we have occasionally seen Pheidole astur removing P. macrostachya seeds from near C. femoratus foraging trails. d. In the exclusion experiment, 3% of seeds were removed from trays treated with Tanglefoot, and from trays treated with both Tanglefoot and mesh cage, suggesting that mammals were of minimal importance in seed removal from AG territories. e. Here we assume that seeds deposited in vertebrate feces would occur randomly inside and outside AG territories, and would be treated the same way as seeds that had not passed through a digestive system. Hence we apply the numbers from our bait study, including bait stations inside and outside of AG territories. f. Numbers taken from our estimate of seed survival upon arrival in AG carton.