CONFOUNDING IN-FIELD EXPERIMENTS - DIRECT AND INDIRECT EFFECTS OF ARTIFACTS DUE TO THE MANIPULATION OF LIMPETS AND MACROALGAE

This study addresses questions about the problem of artifacts in field experiments on plant-animal interactions on rocky shores. In a previo us study we suggested that grazing by limpets on filamentous algae fac ilitated the establishment of the red alga Rissoella verruculosa at mi d-shore levels on the rocky coast south of Livorno (Italy). This model predicted similar patterns of recovery of Rissoella in plots exposed to limpets and in plots where both limpets and filamentous algae were removed. Here we explore this prediction through a 2-way factorial exp eriment in which limpets and filamentous algae were manipulated by mea ns of barriers of epoxy-putty painted with copper paint and manual rem oval, respectively. The experimental results did not support this pred iction as the brown crust Ralfsia verrucosa, rather than Rissoella, do minated in plots were both herbivores and the filamentous algae had be en removed. These results, however, appeared confounded by artifacts d ue to the experimental procedure. The effects of barriers were tested by comparing patterns of recovery of species in plots surrounded by fr ames of epoxy-putty painted with discontinuous bands of copper paint, against control plots. This experiment revealed a significant positive effect of barriers on Ralfsia and a trend toward a negative effect on Rissoella. An additional experiment indicated that overgrowth by Ralf sia significantly reduced the expansion of the encrusting base of Riss oella. overall, these patterns suggested the occurrence of indirect ef fects of artifacts: barriers, by increasing the abundance of Ralfsia i ndirectly prevented the establishment of Rissoella in plots where both limpets and the filamentous algae were removed. This study offers an example of how artifacts propagating through indirect pathways may fur ther confound the results of field experiments. Copyright (C) 1997 Els evier Science B.V.