Search

food. We experimentally examined how two factors influence various source-use decisions of Gray Jays Perisoreus canadensis exploiting locally abundant food sources. First, we examined the behavior of jays exploiting a food source during an initial "encounter" and during another encounter 3 days later. The jays made fewer caches during subsequent encounters with a source at a given location, around which the local density of previously made caches was high. They also compensated by spacing out their caches more widely and by increasing the rate at which they retrieved and redistributed previously made caches. Second, we investigated the effect of the composition of a food source (i.e., the value of its food items) on the jays' source-use decisions. The jays cached substantially more food items (raisins) from a large-item (X = 491) than from a small-item (328) source when sources of each type were made available on different days. The cumulative number of caches made by individual jays (over 10 h) was more nearly asymptotic in the smaller-item treatment. We propose that the jays' tendency to recache previously made caches accounts for their failure to cease hoarding from our experimental sources. This two-stage process of food storage (i.e., initial placement and subsequent recaching) may yield long-term average rates of storage throughout the territory exceeding those predicted by simple rate-maximization, patch-use models.

1. In warm climates many insects exhibit discrete generations, in the absence of obvious factors that could synchronize their age structure. It has been hypothesized that parasitoid wasps might be responsible for these oscillations in the host age structure, known as generation cycles. 2. We examine the role of the parasitoid Anagrus delicatus in the dynamics of the salt-marsh planthopper Prokelisia marginata. In particular, we evaluate the hypothesis that Anagrus contributes to the formation of generation cycles in the hopper, in the subtropical climate of Florida. 3. Two kinds of evidence are presented

A model is presented that concerns how a scatterhoarder should distribute food items taken from an ephemeral, locally abundant source among cache sites scattered throughout its territory, and how such an animal should adjust the way it distributes its caches in response to the presence of previously made caches. The model assumes that such an animal behaves in a manner that maximizes the rate at which recoverable (surviving) food energy is stored. One application of this model concerns how a scatterhoarder should respond when it encounters a food source in the same location as a previously exploited one. Such spatial overlap of food sources through time may be a common occurrence for animals such as Gray Jays Perisoreus canadensis which store thousands of food items throughout their territory during the summer months for use during winter. Under the conditions of the experiment, but not under all theoretical conditions, the model predicts that Gray Jays should space their caches more widely during subsequent encounters with a source at a given location. Consequently, they should take longer to cache a given number of food items, and, because fewer of these caches will survive, the number of recoverable caches made per unit of time spent caching should be reduced. A field experiment using Gray Jays is described. As predicted, jays compensated for the local density of extant caches by transporting food items farther from the source during second encounters with a source in a given location. These results indicate that the jays, at least for the conditions of this experiment, are not forced by a memory constraint to behave in some standard fashion when they encounter a point food source. The theoretical and empirical results begin to show how scatterhoarders, by flexibly adjusting their behavior in response to previously made caches, may approach the maximal rate of storage of recoverable food throughout their territory over the long term.

We present two models which blend different forms of parasitoid search behavior; aggregation independent of host density, and in response to host density. In the first model (Model A) parasitoid and host densities on a patch are assumed to be positively correlated, while in the second model (Model B) the parasitoids have a linear regression on host density, with an associated error term. In Model A, the dynamics were least stable when host and parasitoid were perfectly correlated. In Model B, the slope of the regression had only a moderate effect on stability, while large amounts of error strongly promoted stability. These results suggest that a weak and noisy aggregative response by the parasitoid may be strongly stabilizing, while a strong relationship between parasitoid and host density can sometimes produce instability. As a strong relationship also produces spatial density-dependence in parasitism rates, such a pattern in the field need not indicate stabilizing forms of parasitoid behavior.

(1) The California red scale, a citrus pest, is under successful biological control by its parasitoid Aphytis melinus. Scale abundance is relatively stable, in comparison to other insect populations, and the rate of change of total scale abundance is density dependent. (2) Using data from a lemon grove, we have examined the scale-Aphytis system for several potential stabilizing mechanisms. These were: (a) density-dependent parasitism by Aphytis, in which parasitism rates vary in response to changes in scale density through time, (b) density-dependent sex-ratios in Aphytis, where the proportion of female progeny is influenced by both scale and parasitoid density, and (c) a spatial refuge from parasitism. (3) No evidence was found for the first two mechanisms. Parasitism rates were influenced only by rainfall, while sex-ratio was not significantly affected by any of the variables examined. (4) Strong evidence exists for a refuge from parasitism, which may account for the apparent stability of the system.

of their models. Field experimental tests are not without problems, particularly where indirect effects of other interacting species may confound interpretations (Connell 1983, Quinn and Dunham 1983, Bender et al. 1984). Careful experimental designs, together with monitoring of changes in associated species that could produce indirect effects, probably remains the best way to test theories of succession. Walker and Chapin (1987) suggest that "it may be premature to synthesize our understanding of succession into a unified theory". However any synthesis of current knowledge into a theory could usually be judged premature, since knowledge is always incomplete. The models proposed by Walker and Chapin (1987: Fig. 1), like those of Connell and Slatyer (1977) are based upon incomplete knowledge, but are welcome because they stimulate thought and indicate directions for further investigation.