DIRECT, INDIRECT, AND CUMULATIVE EFFECTS OF 3 NATIVE HERBIVORES ON A SHARED HOST PLANT

The effects of three herbivores on growth, survivorship, and fruit production of smooth sumac (Rhus glabra) were measured over the 1984-1986 growing seasons. The herbivores, a specialist chrysomelid beetle (Blepharida rhois), a specialist cerambycid beetle (Oberea ocellata), and whitetail deer (Odocoileus virginianus), feed on different plant parts and are active during different times of the year. My goals were to determine (1) the effects on ramets of each herbivore separately and in combination over several years, (2) whether herbivore effects were consistent from year to year, (3) whether the history of attack affects subsequent herbivory, both within and between seasons, (4) whether herbivores interact directly, and/or indirectly, through changes in the host plant, and (5) whether any herbivore(s) emerge as particularly prominent selective agents on the plant, considering both their direct and indirect effects. Selective exclosures were used to manipulate herbivory histories of ramets. All manipulations were done in the field under natural densities of all herbivores. Herbivory treatments were cumulative over 3 yr, so that by 1986, each treatment comprised a history of 3 yr of absence/presence of browsing and beetle attack. Differing combinations of attack allowed me to detect how herbivory in one year affected ramet growth in following years. The effects of deer, chrysomelids, and cerambycids on sumac differed not only in their magnitude and direction, but also in their effects in combination. Chrysomelid herbivory (randomly assigned, using natural densities of beetles) over 3 yr was always injurious to sumac ramets. Damage in 1984 was especially injurious, and was still detectable 2-3 yr later in ramet growth, despite the absence of any subsequent herbivory. Cumulative effects of chrysomelid damage were also found, i.e., the more years a ramet was exposed to chrysomelid herbivory, the more likely it was to die. In contrast to chrysomelid damage, deer browsing was generally associated with increased ramet growth and/or reproduction. These effects were weaker and had no detectable long-term effects. The positive effects associated with browsing may reflect selectivity on the part of deer rather than positive effects of browsing, since browsing was not randomly assigned (Strauss 1988a). I found nonadditive effects of damage by different herbivores on ramet growth or mortality. For example, there was no effect of browsing on the magnitude of chrysomelid effects on ramet growth; however, browse effects on ramet growth disappeared on chrysomelid-attacked ramets, but were still apparent on ramets protected from chrysomelid herbivory. Many interactions between herbivores were found. Every pair of herbivore species, despite very different feeding habits, exhibited at least one interaction. Most interactions resulted from temporally separated herbivory events and were thus mediated by the host plant. In no cases were interactions symmetrical. From 3 yr of observations, chrysomelids were always the most injurious herbivore, deer were never injurious (and were potentially beneficial), and cerambycids, although injurious alone, were even more so in conjunction with the chrysomelid. Although the strengths of selection exerted by each herbivore varied from year to year, their relative effects on the plant did not change. From this result, I argue that selective effects of herbivores on sumac at Cedar Creek are not diffuse, but are consistent from year to year.