The larvae of many sessile marine invertebrates are able to locate and settle selectively in physical microrefuges where the survival of the post-larval stage may be enhanced. However, it has been suggested that clonal encrusting organisms with unlimited attachment to the substratum do not rely heavily on microrefuges for post-larval survival, since they quickly outgrow refuges and then suffer partial rather than total mortality. Consequently, the larva of such organisms is predicted to show little selectivity at the time of settlement. This prediction, verified for some bryozoans and ascidians, has also been extrapolated to sponges despite the fact that there IS virtually no field information on patterns of settlement and post-settlement mortality in this group. In the present study, we investigated patterns of settlement and juvenile mortality in 2 common Mediterranean encrusting sponges, Crambe crambe and Scopalina lophyropoda. We placed a set of grooved plates (providing microrefuges for larval settlement) and a set of smooth plates (without refuges) in the natural community, mapped larval settlement on plates at 3 d intervals, and recorded the fate of each settled larva for 10 wk. The total number of larvae of C. crambe that settled on the plates was significantly higher than that of S lophyropoda, and both species settled preferentially in the microrefuges of the grooved plates. On each of the grooved plates, larvae settled preferentially on the shaded sides of outer ridges, avoiding equivalent ridge sides that were highly exposed to light. This pattern suggests that microrefuges are located by larvae on the basis of photic cues. The analysis of survival time revealed that survival of post-larval stages protected in grooves was higher in both species compared to unprotected post-larvae. The survival curve of the juveniles did not clearly follow an exponential model. Survival decreased more or less rapidly for the first 2 mo, but the chance of mortality was not markedly higher 1 wk after settlement than during subsequent weeks. Because we could not exert control over the various coincidental mortality factors that affected the sponges settled on the experimental plates, we tested the following hypotheses in the laboratory: (1) that the abundant echinoderms inhabiting these assemblages cause significant mortality by both unselective ingestion and abrasion, and (2) that microrefuges substantially reduce sponge vulnerability to this mortality factor. In this experiment, we investigated differences in mortality among small sponges as a function of microhabitat exploitation (sponges protected in grooves versus unprotected sponges on smooth surfaces) and the presence/absence of potential predators (the sea urchin Paracentrotus lividus, the starfish Echinaster sepositus). It was found that E. sepositus caused negligible mortality, irrespective of sponge microhabitat. In contrast, P. lividus caused significant mortality, indicating that bulldozing by urchins may have substantial negative effects on sponge recruitment in temperate hard-bottom communities where urchins are ubiquitous. Our results also indicate that exploitation of physical microrefuges, such as grooves and crevices, reduces, but does not prevent, sponge mortality due to bulldozing by urchins. The enhancement of juvenile survival in microrefuges has probably operated as a selective force, favoring larval mechanisms that facilitate selective settlement in refuges.This research was supported by 2 grants (DGCYT PB-94- 0015-C02-01C,I CYT. 21AR95-1764) and funds of the Generalitat de Catalunya (1997SGR00084)Peer reviewed