The movement decisions of frugivores shape spatial patterns of seed deposition, with considerable impacts on plant individuals, populations, and communities. However, we have a limited understanding of the mechanisms determining the foraging strategies behind the movements of frugivorous animals and their interaction with the landscape. We investigated foraging strategies of red-bellied lemur (Eulemur rubriventer) groups dispersing fruits of the tropical tree Harungana madagascariensis. We used spatially-explicit mechanistic simulations to examine movements and seed dispersal generated by four theorized foraging strategies in an agent-based model: frugivores move directly to the 1) most rewarding plant in the landscape, 2) nearest fruiting plant, or towards the most rewarding plant in the landscape, with stops at any fruiting plants encountered along the way according to a local attraction distance of 3) 10 m and 4) 20 m. We then compared 12 metrics and data distributions for seed dispersal and movement patterns from these models to empirical observations. Simulations of foraging strategies with directed movement toward highly rewarding resources (strategy 1) produced the longest seed dispersal distances and best-matched observed patterns. Observed data of lemur movements and directionality best-matched results from three strategies, simulations toward highly rewarding resources without stops (strategy 1) and with either stop along the way with a 10 m (strategy 3) or 20 m (strategy 4) attraction distance. Collectively, our results supported the strategies in which lemurs moved consistently towards the small number of trees that produced the most fruit on the landscape. Because these trees were located far from each other, the landscape interacted with lemur foraging strategies to produce the longest dispersal distances compared to the strategy of moving towards the closest fruiting trees regardless of crop size trees that produced fewer fruits nearer lemur locations. Our work supports incorporating a mechanistic understanding of animal foraging, movement behavior, and landscape patterns in fruit production for the prediction of seed movement. Explicit considerations of these mechanisms can improve our understanding of the maintenance and structure of diversity in plant communities, with implications for the conservation of forest ecosystems relying heavily on animals for seed dispersal.

2.1 Study Area

Our study took place at Valohoaka camp in Ranomafana National Park, Madagascar. This park encompasses 41,000 hectares of mid-elevation (500-1500 m) rainforests (average annual rainfall 2,835 mm), set aside in 1991 to protect a particularly biodiverse ecosystem within a Malagasy biodiversity conservation hotspot (Myers et al., 2000; Wright & Andriamihaja, 2002). Valohoaka is in a relatively intact and homogeneous portion of the protected primary forests of Ranomafana National Park (Razafindratsima, 2017). Research permits for field work were provided by the Malagasy Ministry of Environment, Ecology, and Forests. As this was an observational study, involving no handling or manipulation of the animal’s environment, ethical approval was not necessary.

2.2 Data Collection

Movement and Dispersal Observations

We collected data on frugivory, movement, and seed dispersal over three months (May - July 2018) by following two habituated groups of the frugivorous lemur, Eulemur rubriventer. This is a large-bodied (1.58 kg – 2.63 kg) (Terranova & Coffman, 1997) species with a diet consisting of 80% fruit (Overdorff, 1993). Individuals live in small nuclear family groups that stay close to each other while they move and forage (average group size of 3.44 ± SD 0.55)(Razafindratsima et al., 2014). Daily surveys, which began at 06:30 and concluded at 15:00 hrs, were carried out by a dedicated team for each lemur group. Observation teams, which followed the lemurs on foot, were composed of three highly trained lemur follow technicians who recorded data and always kept the lemurs in sight. On each survey day, we cycled through group individuals to select a focal individual for a given day, then recorded every instance of frugivory by the focal individual.

Fieldwork resulted in 590 hours of observation, with 431 frugivory events at 210 individual plants. Using these data, we calculated the time, distance, and direction of travel between successive foraging events. Throughout the survey period, we also recorded any instance of seed dispersal by all group individuals, identified all seed species, and recorded the location of the event using a Garmin eTREX GPS device (Garmin, Olathe, KS, U.S.A). We recorded 346 dispersal events for all plant species. Gut retention time was calculated for Harungana madagascariensis from these data using the 25 dispersal events (423 individual seeds) which occurred a minimum of 5 hours past the start of the survey, and for which there was a single possible source plant. We selected the cut-off of five hours based on a previous gut retention study on this species so that we could eliminate the risk of incorrectly identifying the source plant (Razafindratsima et al., 2014). We restricted the following analysis to the dispersal of H. madagascariensis seeds (Fig 1), as this species represented most of the dispersal events that could be sourced to a single-parent plant. In previous work, lemurs have been found to feed primarily on this species in the dry season (Dagosto, 1995) (comprising ~ 25% of their frugivory events in our study), and it is considered a keystone resource (Overdorff, 1996). We obtained gut retention estimates like those calculated by Razafindratsima et al. (2014) for this species.