Predation in Insular Plant Dynamics: An Experimental Assessment of Postdispersal Fruit and Seed Survival, Enewetak Atoll, Marshall Islands

We studied predation upon the fruits of four common terrestrial plant species of atolls by excluding: 1) all predators; 2) large predators only; and 3) no predators. Each ofthese treatments was located within the vegetation at three distances from the shore: beach edge of fringing thicket; middle of the thicket; and inner forest. The plants were Terminalia catappa, Messerschmidia argentea, Scaevola taccada, and Guettarda speciosa. Predation loss over the year was significant (P < 0.01) for all four species. Location contributed significantly only for S. taccada. Amount of loss and predominant type of seed predator for a plant species were related to size and conspicuousness. The largest fruits, those of the widespread tree T. catappa, were destroyed completely at all sites; the damage was done by both insects and terrestrial crabs. The large fruits of G. speciosa were destroyed by insects inland and by crabs at the beach edge. The small fruits of M. argentea, a plant which is common along the beach, tended to have greater losses away from the beach edge; those losses were caused primarily by insects. For S. taccada, which occurred at the beach edge, fruit survival was highest in the middle of the fringe thicket; average damage and disappearance were high and caused primarily by large predators. In sum, the experiment demonstrated that predation by widespread, omnivorous large animals and insects on an atoll island was important in survival of fruits and enclosed seeds. We conclude that this interaction could have a significant influence on the dynamics of plants on atolls. SEED PREDATION BY ANIMALS iS common (e.g., Janzen,1971; Bohart and Koerber,1972; Hubbell, 1980), and mortality oftree seeds is often especially severe following dispersal from the parent (e.g., Watt, 1923; Janzen, 1971, 1983; Pinowski and Kendeigh, 1977; Howe and Smallwood, 1982). Such losses are often significant for plant demography (Harper 1969, 1977). For example, destruction of seeds and fruits can limit plant recruitment (Louda 1982a, b, 1983) or determine plant distribution (Janzen, 1975; Louda, 1982b). Atoll plants depend on propagules for recolonizations and persistence (Carlquist,1974). Thus, predation on invading propagules, both by cosmopolitan species such as terrestrial hermit crabs (Weins,1962) and by insects (Janzen, 1971), could be important for our understand' Received for publication 21 April 1984; revision accepted 9 October 1984. Janet and Phil Lamberson placed and maintained our equipment. Ted Turk retrieved the units for us. Tom Ebert and Adrian Wenner provided encouragement and advice at Enewetak. Diane Baxter, Mark Hay, Kathleen Keeler, Ken McKaye, Mike Salmon, Don Strong and Jasper Trendall discussed and improved the original manuscript. The United States D.O.E. (E.R.D.A.) and the Mid-Pacific Marine Laboratory provided funding. ing of insular plant dynamics and biogeography (MacArthur,1972). However, quantitative and experimental data on the fate of plant propagules on oceanic islands, and on the role of predators in that fate, do not exist (Weins,1962; Carlquist,1965,1970,1974; Howe and Smallwood, 1982). Although herbivory and seed predation have been hypothesized to be low on islands (Janzen, 1971, 1975; Carlquist, 1974), four observations suggested the opposite might be the case on Enewetak Atoll. First, evidence of damage by predators was abundant. We examined fruits collected from trees and from randomly placed quadrats under midcanopy of several common tree species. On the tree, the proportion damaged by predispersal insect predators varied from 0.0 on Scaevola taccada to 0.84 on Cordia subcordata (N = 50/species, pers. observ.). On the ground the proportion with serious external damage varied from 0.04 on C. subcordata to 0.66 on Guettarda speciosa; similar levels of damage were found on several islands of the atoll (pers. observ.). Second, terrestrial hermit crabs were common and especially dense near the beach edge ofthe vegetation (pers. observ.; Page and Willason, 1982). These animals have been ob-