Decadal-scale observations of marine reserves suggest that indirect effects on taxa that occur through cascading trophic interactions take longer to develop than direct effects on target species. Combining and analyzing a unique set of long-term time series of ecologic data in and out of fisheries closures from disparate regions, we found that the time to initial detection of direct effects on target species (±SE) was 5.13 ± 1.9 years, whereas initial detection of indirect effects on other taxa, which were often trait mediated, took significantly longer (13.1 ± 2.0 years). Most target species showed initial direct effects, but their trajectories over time were highly variable. Many target species continued to increase, some leveled off, and others decreased. Decreases were due to natural fluctuations, fishing impacts from outside reserves, or indirect effects from target species at higher trophic levels. The average duration of stable periods for direct effects was 6.2 ± 1.2 years, even in studies of more than 15 years. For indirect effects, stable periods averaged 9.1 ± 1.6 years, although this was not significantly different from direct effects. Populations of directly targeted species were more stable in reserves than in fished areas, suggesting increased ecologic resilience. This is an important benefit of marine reserves with respect to their function as a tool for conservation and restoration. T he current global trend to increase the number of no-take marine reserves is a phenomenon with complex ecologic, scientific, and socioeconomic dimensions (1-3). Stakeholders want to know how rapidly changes will occur after protection, even if natural variability can be large and difficult to predict. Patterns of variation in recovery rates of harvested species determined from long-term empirical studies can provide these important ecologic insights. Studies that have quantified the rate at which recovery of targeted species may take place have found the main factors affecting the recovery rates of populations in reserves to be the following: initial population size, intrinsic rate of increase (r), stock recruitment relationships, size of reserve, metapopulation structure, relationships with source locations, annual variations in success of individual recruitment events, the success of reducing fishing mortality (F) in the reserve (4-6), and the degree to which fishing has affected populations. Most of these factors relate to population growth, suggesting that recovery is a cumulative process. In addition, the design of reserves and rates of movement across reserve boundaries frequently play a strong role (4

A frequent expectation of the use of marine reserves in management of coral reef fisheries is maintenance or enhancement of yields to areas adjacent to reserves by adult (post-settlement) movements from reserve to fished areas (the so-called 'spillover effect'

ABSTRACT-Fishing is the most important exploitative activity on coral reefs. Despite this, there are few direct tests, utilizing manipulative or natural experiments, of the impact of fishing on coral reef fish assemblages. This paper documents a natural experiment in the central Philippines: a dramatic increase in fishing pressure within a 750 m long marine reserve, previously protected from fishing for 10 yr (1974 to 1984). In 1983, the site had a significantly higher abundance of fishes, particularly those considered to be favoured targets of fishermen (e.g. serranids, lutjanids, lethrinids), than similar sites which were fished. Abundances were estimated by visual census within the reserve and at 3 control sites. In early 1984, protective management broke down and fishing by up to 100 municipal fisherman began, using traps, hand-lines, gill-nets, spears and occasionally more destructive, non-selective fishing methods, such as explosives and drive nets. The reserve and control sites were recensused after 18 mo of fishing in the Sumilon Island reserve. There were decreases in abundance of favoured targets of fishermen and a significant change in community structure -including significant decreases in both species richness and density -of the coral reef fish assemblage inside the reserve but not at the 3 control sites. Species which contributed most to the change in community structure were not favoured targets and had not constituted a large proportion of yield in previous years Significant decreases in abundance of small, schooling planktivores and benthlc feeding chaetodontids, and significant increases in abundance of small labrids and large scarids, were related largely to the effectiveness of explosives and drive nets against planktivores and herbivores and degradation of the shallow benthic habitat caused by both these methods. Thus, the intense fishing pressure had both direct and indirect effects and a far wider impact on the fish assemblage than the predicted significant decreases in abundance of target species.