Key points are not available for this paper at this time.
In the lowland deciduous and riparian evergreen forests of Guanacaste Province, Costa Rica, the insect seed predators are highly host-specific and display a variety of traits suggesting that the ways they use hosts are not randomly distributed. The vertebrate seed predators may feed on many species of seeds over the course of a year or within the animal population, but, at any one time or with any one individual, strong facultative host-specificity occurs. Furthermore, the vertebrates have a variety of species-specific behaviors that suggest specialization to overcome the defenses of particular seed species. Even the dispersal agents kill seeds as they pass through their guts, and the details of seed-defecation patterns should be important to the seed. Within this forest, leaf-eating caterpillars seem to be either specialized on one or a few species of plants, or spread over many. While an entire forest is never defoliated at one time, defoliations and severe herbivory occur with many plant species in various seasons or years. Herbivory by large herbivores is probably trivial when compared to that of the insects. Furthermore, the defenses that the large herbivores have to overcome may well have been selected for by both insects and extinct Pleistocene large herbivores. I suspect that many of the animal-plant interactions in this forest are not coevolved, and those that are coevolved will be difficult to distinguish from those that are not. THERE ARE TWO QUITE DIFFERENT KINDS of patterns of herbivory.' First there is the way an individual herbivore bites a plant and the influence of this on other herbivores (Janzen 1973) and on the fitness, health, and growth patterns, et cetera of the bitten plant (e.g., Simberloff et al. 1978; Oppenheimer and Lang 1969; McKey 1978; Chester 1950; Janzen 1976a, b; Rockwood 1973). Much more research is needed on the allocation responses, economics of retention versus shedding of damaged parts, and strategic damage versus absolute damage. For example, the production of induced chemical defenses may raise the fitness of a plant not through the lowering of the absolute amount of herbivore damage done, but rather by causing the damage to be distributed in a manner that harms the plant less (Janzen 1979). However, this essay is not directed at this pattern of herbivory, but rather at the second one. This is the distribution of the species and quantities of herbivores across the mosaic of plant species, developmental stages, plant parts, habitats, defense types, and time. I focus on this pattern of herbivory in the deciduous forests, and their contained riparian vegetation and old pastures, in Santa Rosa National Park (SRNP), northwestern Guanacaste Province, Costa Rica (0-300 m elevation). INSECT SEED PREDATORS The lowlands of Guanacaste Province exclusive of the Nicoya Peninsula (about 3200 kiM2) have about 975 species of broad-leafed plants at latest count (Janzen and Liesner 1980). In an ongoing study of the insects that eat the seeds of these plants, at least 110 species of beetles (Bruchidae, Curculionidae, Cerambycidae) were reared from more than 3700 samples of fruits and seeds (Janzen 1980a). The seeds of at least 100 of the 975 species of plants were regularly found to have beetle larvae seed predators that develop in the seeds at the stage between nearly full-sized and mature (preand post-dispersal). Most of the species of beetles were specific to a particular plant species: 83 species (75%) preyed on one plant species, each of 14 preyed on only two, nine on three, and two on four. The bruchid Stator pruininus preyed on six species and Stator limbatus on eight species, and there was no overlap in their two prey lists (although all prey were legumes). Of the 100 prey species, 63 were in the Leguminosae, and the remainder were spread among 17 other plant families; however, only 17 percent of the broadleafed plant species in the Guanacaste lowlands are
Daniel H. Janzen (Tue,) studied this question.