Predation on a common Anolis lizard: Can the food-web effects of a devastating predator be reversed?

We artificially invaded a set of small islands with the large lizard Leiocephalus carinatus to determine effects on food-web elements including an intermediate predator, the lizard Anolis sagrei; the latter was previously found to have major, mostly direct effects on web spiders, as well as detectable indirect effects on aerial arthropods (including parasitoids) and leaf damage. In addition to these food-web elements, we monitored ground-surface arthropods; they are expected to be affected directly, as well as indirectly via A. sagrei, by L. carinatus. Five islands were randomly selected for invasion while six others served as controls. In addition, four islands without A. sagrei were monitored to determine the natural state of the sagrei-free food web. We also monitored a variety of A. sagrei's traits, from behavioral through physiological and demographic to morphological, to elucidate the mechanisms whereby L. carinatus might affect the smaller lizard and to evaluate trait-mediated as well as density-mediated effects. Comparison of unmanipulated islands with and without A. sagrei showed that A. sagrei appeared to be affecting web-spider density and diversity, numbers of small aerial arthropods, number of large ground-surface arthropods, and two types of leaf damage, scars and mines. All effects were negative save one: Small aerial arthropods were more abundant with A. sagrei; this variable and parasitoid abundance were negatively related to web-spider density, implicating an indirect effect pathway from A. sagrei via web spiders. Introduction of L. carinatus had major and immediate effects on A. sagrei density and perch height; effects on perch diameter, percentage hatchlings, and the width of the body-size distribution followed over time. Behavioral shifts in habitat use continued to the end of the experiment, after changes in population density had mostly leveled off. Adult-male body condition became poorer after introduction of L. carinatus. Among islands, percentage use of the ground was correlated negatively with percentage vegetated area of the island and positively with premanipulation relative hindlimb length. We hypothesize that longer-legged individuals do better on the ground because they are faster there. Conditional evidence for an increase in adult-male relative hindlimb length suggested that, when the ground-inhabiting L. carinatus was introduced, such individuals were differentially able to escape predation. Introduction of L. carinatus significantly reversed A. sagrei's effect on number of species and density of web spiders but had no other major effects on lower level food-web elements. Thus, a relatively short-chained (two links) effect, with a relatively strong second link (A. sagrei on spiders), showed the greatest indirect effect of the manipulation. Another potentially strong indirect effect, that on large ground-surface arthropods, did not occur, possibly because of compensatory (direct) predation by L. carinatus. Thus presence of omnivory may also be relevant to whether reversal of effects occurs. Although most A. sagrei effects were not reversed in the time available for the experiment, this may be typical, as natural L. carinatus colonizations can dwindle away to extinction over about the same time span, producing at best only short-term indirect effects. Population-density estimates from other sites and comparisons to other experimental studies suggest that, in general, predators on Anolis may have erratic effects in space and time.