Seed Dispersal of a Neotropical Myrmecochore: Variation in Removal Rates and Dispersal Distance

Whether vertebrate seed predation is a contemporary selective force in Calathea ovandensis is determined and the removal rates in seed depots open to all dispersal agents and predators with those in cages which exduded vertebrates are compared.

Abstract

We investigated small-scale spatial variation in dispersal success of the ant-dispersed herb Calathea ovandensis, and the potential effect of seed predators on seed dispersal. We found no evidence that vertebrate seed predators influenced seed dispersal probabilities. Seeds placed in experimental seed depots which excluded vertebrate seed predators (but not ants) were removed at the same rate as those in depots accessible to both ants and vertebrates. Five ant species were attracted to the seeds, but only two, Pachycondyla harpax and P. apicalis, regularly moved seeds beyond the parent plant canopy. P. harpax, the most abundant ant, moved seeds about 75 cm; the rarer P. apicalis moved seeds 12 times further. Other ants recruited to seeds without effecting dispersal; in particular, Solenopsis geminata and Wasmannia auropunctata interfered with seed dispersal. Their frequency varied (0-36%) among sites, resulting in significant variation among sites in seed removal rates. VARIATION AMONG POTENTIAL MUTUALISTS may constitute an important selective force in plant-animal interactions that could result in evolutionary specialization of plants to particular animals (Horvitz & Schemske 1984, Schemske & Horvitz 1984). However, there is little information on the magnitude and causes of spatial variation in mutualisms in general, and in seed dispersal systems in particular. In this paper we examine ant-seed interactions in the tropical herbaceous myrmecochore, C. ovandensis Matuda (Marantaceae) (Horvitz 1980, 1981; Horvitz and Beattie 1980). We conducted observations and experiments to answer the following questions concerning variation in dispersal success: (1) Do ant species vary in dispersal distance? (2) What are the frequencies of different ant species at different sites? (3) Does the percent of seeds removed vary among sites as a result of spatial variation in the ant community? Because the probability of successful seed dispersal in myrmecochores is often affected by both ants and vertebrate seed predators (Culver & Beattie 1978, Heithaus et al. 1980, O'Dowd & Hay 1980, Heithaus 1981, Beattie 1983), we determined whether vertebrate seed predation is a contemporary selective force in Calathea. We compared the removal rates in seed depots open to all dispersal agents and predators with those in cages which exduded vertebrates. This experimental design does not investigate the historical selective forces that shaped the evolution of ant dispersal, but does examine the contemporary selective effects of vertebrate seed predators in an ant-dispersed species. STUDY SPECIES AND SITE C. ovandensis (Marantaceae) is an acaulescent tropical perennial herb that is deciduous during the dry season (see Horvitz 1980, 1981; and Horvitz & Schemske 1984 for a detailed description). Fruit capsules, containing a maximum of three seeds each, dehisce at maturity in the late rainy season, and the large (0. 5 cm) seeds fall to the forest floor near the parent plant. Each seed bears a large (0.4 cm), white, lipid-rich aril that is used for food by ants. The seeds have innate dormancy and germinate at the beginning of the next rainy season, ca. 270-300 days after capsule dehiscence. Our study site was located in a secondary forest at Laguna Encantada, near San Andres Tuxtla, Veracruz, Mexico (Horvitz & Schemske 1984). C. ovandensis is abundant throughout this forest, varying among sites in density, population stage structure, and population dynamics (Horvitz & Schemske, pers. comm.). MATERIALS AND METHODS DISPERSAL CHARACTERISTICS OF DIFFERENT ANT SPECIES.-To determine how far ants carry seeds and where seeds are taken by ants, we directly observed ant-seed interactions. Each observer placed six fresh seeds on the forest floor and subsequently observed ant activity continuously for 90 min, recording which ant species came to the seeds, where seeds were taken by ants, and any interactions among ant species. Six seeds was chosen because fruit capsules contain a maximum of three seeds each and no more than two capsules per plant dehisce on the same day (Horvitz, pers. comm.); therefore, six represents a maximum number of seeds that dispersal agents are likely to find at one spot at one time. Twenty-nine such obserI Received 26 April 1984, revision accepted 26 June 1985. 2 Present address: Department of Biology, University of Miami, Coral Gables, Florida 33124, U.S.A. BIOTROPICA 18(4): 319-323 1986 319 This content downloaded from 157.55.39.95 on Sat, 23 Apr 2016 04:56:22 UTC All use subject to http://about.jstor.org/terms TABLE 1. Dispersal characteristics of ant species. % f Relative frequency (%) drop-site types ant-seed Root interactionsa Dispersal distance, or Ant species (N = 134) cm x (SD, max) rock Litter Tunnel Log Crevice Pachycondyla harpax 54.5 76 (43) [2471 56 23 9 3 0 P. apicalis 5.9 925 (635) [20501 86 14 Pheidole spp. 17.2 0 Solenopsis geminata 19.4 7.9 (8.5) [371 Shallow burial in situ Wasmannia auropunctata 3.0 4.0 (2.9) [71 a Each seed is scored for only one ant-seed interaction; see Methods. vations were made over a 3-wk period during the fruiting season (September-October 1983) and near fruiting plants. We watched a total of 174 seeds for 90 min and recorded all ant activity at these seeds. There were occasionally visits by several ants to the same seed within the observation period. In these cases we counted the ant having the greatest effect on that seed as the relevant ant-