Current Biology 19, 206���212, February 10, 2009 ��2009 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2008.11.055 Report Linking Social and Ecological Systems to Sustain Coral Reef Fisheries Joshua E. Cinner,1,* Timothy R. McClanahan,2 Tim M. Daw,3 Nicholas A.J. Graham,1 Joseph Maina,2 Shaun K. Wilson,1 and Terence P. Hughes1 1Australian Research Council Centre of Excellence for Coral Reef Studies James Cook University Townsville 4811 Australia 2Wildlife Conservation Society, Marine Program Bronx, NY 10460-1099 USA 3School of Development Studies University of East Anglia Norwich NR4 7TJ UK Summary The ecosystem goods and services provided by coral reefs are critical to the social and economic welfare of hundreds of millions of people, overwhelmingly in developing coun- tries [1]. Widespread reef degradation is severely eroding these goods and services, but the socioeconomic factors shaping the ways that societies use coral reefs are poorly understood [2]. We examine relationships between human population density, a multidimensional index of socioeco- nomic development, reef complexity, and the condition of coral reef fish populations in five countries across the Indian Ocean. In fished sites, fish biomass was negatively related to human population density, but it was best explained by reef complexity and a U-shaped relationship with socioeconomic development. The biomass of reef fishes was four times lower at locations with intermediate levels of economic development than at locations with both low and high devel- opment.Incontrast,averagebiomassinsidefisheryclosures was three times higher than in fished sites and was not asso- ciated with socioeconomic development. Sustaining coral reef fisheries requiresan integrated approach thatuses tools such as protected areas to quickly build reef resources while also building capacities and capital in societies over longer time frames to address the complex underlying causes of reef degradation. Results and Discussion Effectively confronting the coral reef crisis will require us to link social and ecological systems so that we can better under- stand and address the complex socioeconomic drivers that influence how societies use and ultimately govern their use of coral reefs [2, 3]. It is generally held that human use, driven primarily by population density, is a principal cause of coral reef degradation [4���7]. However, less is known about how other socioeconomic factors such as economic development shape societies��� impacts on coral reefs [8, 9]. Sociological perspectives on human-environment interactions emphasize how socioeconomic development can affect a society���s impact on the environment, often in nonlinear and sometimes positive ways [10, 11]. To explore these linkages in coral reef fisheries, we collected data on a composite index of village- level infrastructure (as a proxy for local-scale socioeconomic development), human population density, and structural complexity of reef habitat (rugosity) in 19 fished sites and 11 fishery closures across five countries in the western Indian Ocean. We evaluated these drivers��� influence on the biomass of reef fishes, which is a variable sensitive to management and human impact [12]. First, we examined whether the biomass of reef fishes targeted in the multispecies fishery could be explained inde- pendently by human population density, structural complexity, and socioeconomic development. In fished sites, human population numbers had a significant but weak negative rela- tionship to the biomass of target reef fishes (n = 19, r2 = 0.28, p = 0.02 Figure 1A), and the benthic structural complexity had a moderate positive relationship (n = 16, r2 = 0.54, p = 0.001 Figure 1B), consistent with previous studies on reef fishes [4, 7, 13, 14]. Our novel finding is that the strongest relationship to fish biomass was the quadratic function of the socioeconomic-development index, which displayed a U-shaped relationship (n = 19, r2 = 0.77, p 0.001 Figure 1C). Second, we tested candidate models with all possible combinations of the three factors to determine the best combi- nation of variables for explaining fish biomass in fished sites. We included country as a random effect to account for nonin- dependence of samples within countries [15]. A key and surprising finding from this study is that the best model included the quadratic socioeconomic-development index and reef structural complexity, but did not include human pop- ulation density (likelihood-ratio test of nested models with and without this term ratio = 0.166, p = 0.684) (Table 1). The quadratic term of the development index was highly significant in the selected model (likelihood ratio = 14.5, p 0.001). Thus, fish biomass is highest where community development is very low or high, but low where development is intermediate (Figure 1C). Fish biomass (6 the standard error of the mean) at the bottom of the curve (Takaungu, Kenya) was 77 6 11.9 kg/ha, approximately 1/4 of the biomass of the sites with the highest and lowest levels of development (336 6 52 kg/ha for Anse Volbert, Seychelles and 294 6 57.3 kg/ha for Ambodilaitry, Madagascar, respectively) (Figure 1C). These findings are consistent with the environmental Kuz- nets curve hypothesis, which predicts that increasing socio- economic development results in ecological degradation until a point when environmental conditions improve as societies become increasingly affluent and begin to demand environ- mental quality (creating a U-shaped relationship between affluence and local environmental conditions) [10, 16, 17]. The causal mechanisms behind a Kuznets curve relationship are generally classed in three broad categories: (1) a technique effect, whereby societies may change the technologies used to produce goods and services, which may have differing levels of impact on the environment (2) a composition effect, whereby the composition of the economy could change to *Correspondence: joshua.cinner@jcu.edu.au

be less destructive to the local environment, for example, by switching from primary-resource extraction to a service industry and (3) a scale effect, whereby wealthier societies displace local impacts, for example, by drawing resources from other areas, often those poorer or less regulated [16, 18]. The parallel sociological perspective of ecological modernization suggests that it is not economic development per se that leads changing environmental conditions, but rather the accompanying institutional changes, such as invest- ments in scientific and natural-resource management organi- zations [19]. We used socioeconomic survey data from these communi- ties to further examine how a combination of the technique, composition, and scale effects, and also aspects of local sociocultural institutions, may play a role in our observation of a Kuznets relationship for coral reef fishes in the western Indian Ocean (Table 2). Sites with low levels of development are characterized by high levels of dependence on fishing as a primary occupation, minimal engagement in salaried employment, and few boats with engines (Table 2 and Figure 2A). Although these low-development sites tend to have weak national governments [20], the presence of customary sociocultural institutions such as taboos may act to restrict fishing effort (although this later indicator was only suggestive at p = 0.054 Table 2). Together, these factors suggest that in low-development sites, technological con- straints and social institutions may limit people���s exploitation of marine resources. Reduced dependence on marine resources, variable access to boats but increasing access to engines and other technologies, high use of spear guns, and a lack of customary management institutions characterize communities with intermediate levels of development (Table 2 and Figure 2B). Factors such as reduced dependence on marine resources and increased technological efficiency can break down customary sociocultural institutions that may be critical in managing marine resources [21]. For example, in Kenya, which has some sites with the poorest fishery condi- tions, customary institutions were once widespread, but they have largely broken down in recent years [22], with destructive fishing techniques now practiced in some of these locations [6]. Sites with high socioeconomic development are generally characterized by effective national government [20], low dependence on fishing, reduced use of potentially damaging gear such as gill nets and higher use of more benign gear Figure 1. Fits of Reef-Fish Biomass Fits of reef fish biomass as a function of (A) human population density (r2 = 0.28), (B) habitat rugosity index (r2 = 0.54), and (C) community-level socioeco- nomic-development index (r2 = 0.77). Solid lines show curves fitted from linear (A and B) and quadratic (C) regressions. Data are distinguished by country as follows: MD, Madagascar SZ, Seychelles KY, Kenya MS, Mauritius TZ, Tanzania. Table 1. Comparison of Candidate Models Model Fixed Model Terms df n AICc BIC DAICc DBIC AICc weight 1 no fixed terms 3 16 178.5 178.8 3.6 8.7 10% 2 quadratic development 5 16 177.9 175.7 3.1 5.6 13% 3 habitat rugosity 4 16 179.7 179.1 4.9 9.0 5% 4 log population density 4 16 179.5 179.0 4.7 8.9 6% 5 habitat rugosity + quadratic development 6 16 174.8 170.1 0.0 0.0 61% 6 log population density + quadratic development 6 16 182.8 178.1 8.0 8.0 1% 7 log population density + habitat rugosity 5 16 182.6 180.5 7.8 10.4 1% 8 log population density + habitat rugosity + quadratic development 7 16 181.3 172.7 6.5 2.6 2% Comparison of candidate models with three fixed effects for reef-fish biomass: a quadratic function of our socioeconomic-development index, habitat rugosity index, and natural log of human population density. All models include a random effect of country. Model 5, including the development index and habitat complexity, has the lowest BIC and AICc scores, confirming it as the best fit. The following abbreviations were used: df, degrees of freedom n, sample size AICc, Akaike information criterion corrected for small sample sizes BIC, Bayesian information criterion DAICc and DBIC, difference from the criterion scores of the most favored model AICc weight, likelihood weight based on the AICc values of all tested models [45]. Coral Reef Social-Ecological Systems 207