gesting a potential indirect cascading effect as a result of declining populations of apex predator species. Increases in Pomacentrids could have also been related to the massive decline in percent living coral cover as a result of post-bleaching mortality events in 1998 and 2005.

3.3. Spatio-Temporal Variation in Fish Community Structure

There was a highly significant change in fish community structure through time, across sites, and across depth zones (Table 6). Temporal change between 1997 and 2007, and between 1997 and 2012, was highly significant, but not between 2007 and 2012. All interactions were also highly significant. These results suggest that reef communities underwent significant change within the period of 1997 to 2007 when two significant sea surface warming episodes were followed by two massive coral bleaching events, and two mass coral mortality events. A species cumulative dominance plot (Figure 3(a)) showed a significant decline in cumulative dominance with time as a result of significant declines in dominant species, mostly fishery-targeted taxa. Principal Component Ordination (PCO) analysis (Figure 3(b)) showed that temporal variation in fish community structure at ICA in 1997 was the result of variation in herbivores Acanthurus coeruleus and juvenile stages of Sparisoma sp., planktivores Abudefduf saxatilis and Thalassoma bifasciatum, and corallivore Chaetodon capistratus. Increasing biomass of territorial damselfishes Stegastes leucostictus and Microspathodon chyrurus explained the difference in 2007 and 2012. This solution explained 44% of the observed variation. PCO analysis at PLT (Figure 3(c)) showed that 1997 patterns were explained by herbivores Scarus vetula (which is also an important fishery-targeted species) and Sp. aurofrenatum, and by C. capistratus. Patterns observed in 2007 were explained by A. coreuleus, St. leucostictus, St. diencaeus, and Sc. iserti. Patterns documented in 2012 were explained by pelagic piscivore Carangoides ruber, and by benthic invertivores Holocentrus adscensionis and Halichoeres bivittatus. This solution explained 33% of the observed variation. Patterns at DIA were explained in 1997 by C. ruber, A. saxatilis, Sp. viride, Ha. pictus, and St. partitus. Patterns in 2007 were explained by Ha. bivittatus, St. leucostictus, and St. variabilis, while in 2012 by Sp. aurofrenatum, and piscivore Ocyurus chrysurus, a highly fishery-targeted species. This solution explained 29% of the observed variation. Most of these indicator species were indeed non-targeted taxa.

3.4. Spatio-Temporal Variation in Fish Dominance

SIMPER analysis based on fish biomass showed that the top five indicator species of 1997 reefs at ALCNR were Sparisoma viride (18.5%), Acanthurus coeruleus (12.8%), Ocyurus chrysurus (8.3%), Abudefduf saxatilis (7.9%), and Chaetodon capistratus (6.5%), with a cumulative contribution of 54%. Average similarity of all reefs surveyed was 33%. The top five indicator species of 2007 were Sp. aurofrenatum (16%), Ac. bahianus (14.3%), Sp. viride (12.8%), Ac. coeruleus (12.7%), and Scarus iserti (7.8%), with a cumulative contribution of 63.5%. Average similarity of all reefs surveyed was 46%. The top five indicator species of 2012 were Ac. coeruleus (12.4%), Sp. aurofrenatum (10.1%), Sp. viride (9.8%), Ac. bahianus (9.6%), and Microspathodon chrysurus (7%), with a cumulative contribution of 48.9%. Average similarity of all reefs surveyed was 54%. Differences between 1997 and 2007 were mostly explained by declining schools of browser herbivore Ac. coeruleus and of planktivore Ab. saxatilis. Declining abundance of Ac. coeruleus also explained differences between 1997 and 2012. The reappearance of O. chrysurus, particularly at PLT and DIA, did explain most of the variation between 2007 and 2012. Average dissimilarity between 1997 and 2007 was 70%, and between 1997 and 2012 was 69%. Average dissimilarity between 2007 and 2012 was 49%.

Observed temporal trends showed that fish communities were becoming more similar across sites and that observed gradients in piscivore and generalist predator abundance in 1997 were rapidly lost after the 1998 and 2005 massive bleaching and mass coral mortality events, and following 15 years of continuing fishing impacts. Dominant fish species (by biomass) at ICA were Ac. coeruleus and Sp. viride, while Sp. viride and Ocyurus chrysurus were dominant at PLT, and Sp. viride and Ac. coeruleus were dominant at DIA, with overall similarity averaging 36% across sites. Higher abundance of Ac. coeruleus at ICA explained most of the difference observed between ICA and PLT (73% dissimilarity), and between ICA and DIA (69% dissimilarity). Higher abundance of Ab. saxatilis at DIA explained most of the difference between PLT and DIA (71% dissimilarity).

3.5. Decline in Fishery-Targeted Species

There is significant evidence that critical fishery-targeted species showed significant signs of decline through time across most sites (Figure 4). Important targeted members of groupers (subfamily Epinephelinae) basically disappeared after 1997 from most sites, including the Red Hind (Epinephelus guttatus), Rock Hind (E. adscensionis), Nassau grouper (E. striatus), and Coney (Cephalopholis fulva). No critically-depleted Goliath grouper (E.

itajara), or deeper water grouper species of genus Mycteroperca were observed during the study. This pattern was very similar for representative members of snappers (Lutjanidae) (Figure 5). This included the Gray snapper (Lutjanus griseus), Schoolmaster (L. apodus), Mahogany snapper (L. mahogoni), and Yellowtail snapper (Ocyurus chrysurus). Although juveniles of Mutton snapper (L. analis) and Dog snapper (L. jocu) were sporadically observed across sites, no individuals were documented during visual censuses. Rapidly declining grouper and snapper populations coincided with increasing presence of invasive Lionfish (Pterois volitans), particularly during the 2012 fish counts. Lionfishes could have become a key mortality factor for juvenile fishes.

Important targeted members of parrotfishes (family Scaridae) also declined after 1997 across most sites (Figure 6). This included the Redtail parrotfish (Sparisoma chrysopterum), Stoplight parrotfish (Sp. viride), Queen parrotfish (Scarus vetula), and Blue parrotfish (Sc. coeruleus). No individuals of rare and highly overexploited species such as the Rainbow parrotfish (Sc. guacamaia) and the Midnight parrotfish (Sc. coelestinus) were observed during any of the surveys.

3.6. Characterization of Stakeholder Groups

Commercial fisherman relied on the reserve mainly on a seasonal basis and less so for key, commercial species such as Spiny lobster (Pannulirus argus), Queen conch (Lobatus gigas), and Yellowtail snapper (O. chrysurus). Also, fishermen did not exhibit any direct conflicts with other user groups. They also perceived a long-term decline in key, natural resource conditions in the reserve related to their livelihoods (i.e. commercial fisheries), and over 80% of them were generally in favor of a no-take MPA that exists on a de facto basis—ICA to LOB (and perhaps PLM)—due to the high volume of recreational use within and between those islands (Table 6). If a no-take MPA is to be implemented using stakeholder participation, the following should be considered: participation should involve open meetings that allow complete participation, over technical workshops and representative councils; fishermen’s views on site location and allowable uses should be included in the information gathering and decision-making process; and, regardless of the governmental agency that would administer the no-take

MPA, management should facilitate dialogue among the commercial fishing communities in the region and the PRDNER such that the former may improve trust in the agency.

Concessionaires were comprised of diverse interests, including dive and snorkel operators, catamaran and other large vessel operators, and fishing and other mixed-trip charters. The entire group relied extensively (in many cases, exclusively) on the reserve’s coastal and marine resources for their livelihoods, and there was a majority view among respondents that the reserve’s coral reef and related resources have declined over the past decade or longer. Most concessionaires also identified ICA and PLM as areas that experience high volumes of use and which are centers of use conflict, particularly from recreational divers and private recreational vessels. Due to these and other related factors (i.e., overfishing, water quality decline), over 70% of the concessionaires were in favor of a no-take MPA within the ALCNR (Table 7). The reserve, as would be designed by most respondents, would exclude commercial and recreational fishing but would allow non-consumptive uses, including diving, snorkeling, and cruising. The areas that were most often identified as candidate sites were those which were the waters around the most heavily visited islands, including ICA and LOB. Although less popular, concessionaires also identified DIA and PLM as islands around which to set up a no-take MPA. Concessionaires believed that reserve designation should involve mainly open meetings that allow the participation of all stakeholder groups and the public, while many concessionaires also favored having other forms of public participation, including technical workshops. Concessionaires were also more amenable than the commercial fishing group to have PRDNER manage the no-take MPA site, but most still preferred a US Federal Agency, mainly because of the latter’s enforcement capability and financial capacity.

Registered vessel operators who visited the ALCNR were prolific boaters, taking five trips per month, each of which lasted over half a day (4.5 hours); thus, they represent a group that is most likely knowledgeable about the region and its resources, if not its designated status. That is, the results also demonstrated that while the operators took a majority, or almost 90%, of their trips to the reserve, over a third of the respondents were unaware of the reserve or its boundaries. If the group were to be engaged in a process to set up a no-take MPA, part of the

process would have to involve boater (and, indeed, general public) education on the existence of the reserve and its present boundaries and regulations. The results also determined that a smaller percentage of registered vessel operators (52%), compared to the corresponding percentages of commercial fishermen (80%) and concessionaires (73%) supported a no-take MPA designation (Table 7). However, this lower support for a no-take MPA may have been a result of the methodology involved, which consisted of a formal, self-administered survey instrument that did not allow for respondents to select the types of uses that could be restricted and choices between designation processes and different management agencies. Notwithstanding those constraints, over 90% of those who did identify a recommended no-take MPA in the reserve selected ICA, which was more popular than the adjacent LOB and PLM, but which together garnered support for closure from over 70% of the registered vessel operators.

3.7. Geospatial Rankings of Candidate No-Take MPA Sites

Table 8 summarized the multi-criteria ranking analysis of candidate no-take zones within ALCNR. Overall, SDP ranked as the most significant site for a no-take MPA designation, with an 83% combined score. It was followed by DIA with a score of 75%, and PLM and LOB, with 74%, each one. PLT followed with 70%, ICAW with 65%. ICAE scored 47%. These results suggest that based on a combination of criteria, three different complexes emerged as potential candidate no-take zones: SDP-DIA, PLM-PLT, and ICAE-ICAW-LOB. SDP-DIA ranked highest under most criteria, but particularly under biological criteria. These sites were located at approximately 10 km off the Fajardo coast and showed the lowest density of visitors and the most diverse and abundant fish assemblages. Distance and strong oceanographic conditions is a major constraint largely reducing recreational activities east of LOB towards DIA. Nonetheless, these are the sites with potentially stronger conflicts with commercial fishermen. PLM-PLT is an area that also showed a moderately high ranking by a combination of criteria. Most stakeholders supported its designation as a no-take zone due to several reasons, including proximity to marinas and hotels at the eastern coast of PR (<6 km), easy access, usually protected conditions at frequented areas, and because of its overfished state. Similar justifications were presented for supporting the designation of the ICAE-ICAW-LOB complex.

There was a strong agreement between stakeholder perceptions of the status of ALCNR’s natural resources (i.e., fish communities, benthic communities, water quality, cleanliness, etc.) and empirical data obtained in this study regarding the status of fish communities within the reserve that strongly support the immediate designation of no-take zones. After careful consideration of empirical evidence regarding the depauperate condition of fish communities within ALCNR with increasing time in comparison to the 1997 baseline, we strongly recommend the designation of three zones as no-take MPAs as a consensus option (Figure 7): 1) ICAE-ICAW-LOB complex, 2) PLM-PLT island complex, and 3) SDP-DIA. This alternative will protect significant essential fish habitats, will facilitate compliance and enforcement of no-take zones as units can be easily identified and demarcated with buoys. This alternative adjusts better to a consensus model more compatible with the options supported by fishermen and other user groups, while simultaneously protecting overexploited resources by overfishing and by recreational activities, as well as protecting critical areas that still support healthy fish populations.

Rank scores: 0 = Low ranking; 1 = Moderate ranking; 2 = High ranking. Total maximum score was 22 cumulative points.

At the same time it also maintains important areas subjected to very strong ocean circulation between SDP and LOB open to artisanal fishing to protect the traditional livelihoods of local fishing communities. Respecting traditional uses within these communities is a critical element necessary to build trust with local stakeholders, and develop and compliance within any no-take MPA. The proposed network design will help improve connectivity among proposed no-take zones and will improve fish spillover effects to adjacent areas open to fishing.

4. Discussion

4.1. Ecosystem-Based Model Output

Since its designation in 1980, recreational and commercial fishing activities across ALCNR have been poorly managed. Populations of the most significant fishery-targeted species were significantly depleted, apex predators were largely absent from most reefs, particularly during the 2007 and 2012 surveys. The most significant predators were small or medium-sized commercial predators, and herbivore guilds were dominant across most sites. Several fish functional groups were largely depleted through most of the study sites, particularly in areas subjected to very intense recreational activities, including non-regulated spearfishing. Piscivore guilds were the most affected. Most grouper (Serranidae), snapper (Lutjanidae), and barracuda (Sphyraenidae) populations were significantly depleted or completely absent at most sites. This pattern was evident also for some grunts (Haemulidae) and parrotfishes (Scaridae). Further, most individuals of fishery-targeted species observed belonged to smaller size categories. Other significant fishery-targeted groups such as triggerfishes (Balistidae), porcupinefish (Diodontidae), highly-prized species such as Nassau Grouper (Epinephelus striatus), or entire genera such as Mycteroperca spp. were totally absent during the entire study. Such trends suggest an unequivocal long-term impact of fishing of apex predators and of heavy exploitation of primarily carnivorous trophic level fishes similar to those observed elsewhere [2] [34] -[37] . Fishing impacts can also have significant indirect effects on benthic community structure [9] [38] [39] . The actual status of coral reef fish communities within ALCNR is a major cause of concern that requires rapid action to significantly reduce or eliminate consumptive uses through the establishment of a network of small, discrete no-take MPAs, with strong community-based participation and support through a co-management model.

4.2. The Ticking Clock of Climate Change: Impacts on Reef Fisheries

Climate change has also become a critical driver of coral reef decline. The northeastern Caribbean region underwent two significant sea surface warming episodes in 1998 and 2005 that resulted in massive coral bleaching and post-bleaching mass coral mortality events with paramount negative impacts on coral reefs [40] -[44] . Declining habitat conditions have resulted in altered fish community structure [11] -[14] . Climate change and exploitation can significantly interact in their effects, such that climate may cause failure in a fishery management scheme [45] , particularly under weak governance conditions and limited enforcement scenarios. Fishery exploitation may also disrupt the ability of a resource population to withstand, or adjust to, climate changes. Therefore, climate-related habitat decline and exploitation could result in altered demography, parental effects, altered migration or declining connectivity of important species [45] . At the ecosystem level, reduced complexity by elimination of species (i.e., fishing), may be destabilizing, and it may result in declining functional redundancy [46] , and could lead to reduced resilience to disturbance [47] . Differential exploitation of marine resources and climate change could also promote increased turnover rates in marine ecosystems [48] , which would exacerbate the effects of environmental changes. Declining coral reef biodiversity will also likely lead to a reduction in the resilience and to an increase in the response of populations and ecosystems to future climate variability and change [45] . If detrimental human impacts to coral reefs are reduced and key ecological processes are enhanced, pulse disturbances and ecological variability may provide opportunities for returning to a coral-dominated state, which could enhance ecosystem resilience, benefits and services [49] . Future management strategies will have to consider the structure and functioning of populations and ecosystems in a wider sense in order to maximize their ability to respond to changing climate and significantly improve their resilience. In that sense, no-take MPAs can become critical tools to restore and conserve populations and ecosystems structure and functioning.

4.3. Community-Based Participatory Model Output

All stakeholder groups agreed that coral reef conditions in the reserve had declined, as had associated resources such as water quality which has affected the health of the coral reefs and fisheries which depended on healthy coral reefs. While the stakeholder groups believed that there were a myriad of causes for the decline, there was general consensus that overfishing (resulting from commercial and recreational fishing) and land-based source pollution (especially as related to coastal and marine tourism, sedimentation, and water quality) had been responsible. Importantly, stakeholders did not recognized climate change and sea surface warming as threats to coral reefs and fisheries which raises a major concern regarding the lack of a public educational and outreach program regarding the consequences and impacts of climate change. The stakeholders mostly accepted the solution of implementing a no-take MPA and may actually have been addressing their concerns over resource decline by identifying the most heavily used areas (the ICA-LOB-PLM complex) as those which deserve the highest protection. While it is clear that non-consumptive stakeholder groups stand to gain the most by restricting access to all other kinds of uses within a no-take MPA, the study revealed that even consumptive groups such as commercial fishermen, fishing charters, and consumptive dive charters generally did not oppose the implementation of a no-take MPA. Their perceptions were consistent with dive tourist perceptions in Belize which suggested that adequate enforcement of no-take MPAs should improve coral reef conditions [50] . All groups agreed that a no-take MPA will help recover depleted fisheries, but did not recognize the value of no-take MPAs as a tool to potentially buffer against climate change-related impacts by maintaining higher coral reef ecosystem resilience.

As important as reaching consensus on the location and characteristics of a no-take MPA was the determination of the process to be used to foster public participation in a format that stakeholders considered would be fair and equitable and the identification of the management agency which stakeholders believed would be best positioned to ensure enforcement and management efficacy. While not discussed in any detail in this study, it was found via commercial fishermen and concessionaire interviews that the stakeholder groups held a dim view on public participation; that is, members of both groups often felt that meetings addressing resource management issues were often poorly advertised and held at hours when they could not attend. Others believed that public participation, while allowed and even promoted, made little difference in influencing the final decisions. However, stakeholders still preferred holding meetings as part of the decision-making process for a no-take MPA instead of other formats, such as technical workshops or representative councils. In terms of identifying the agency that could best implement a no-take MPA, most stakeholders were in favor of a US Federal Agency, particularly the US Fish and Wildlife Service (USFWS). It is likely that the USFWS was most commonly cited because it is the primary federal, natural resource agency that most stakeholders are aware of in the region, through experiences with the Vieques National Wildlife Refuge and the Culebra National Wildlife Refuge. Thus, at the federal level at least, stakeholders are not fully aware of other models, including national parks and, in particular, national marine sanctuaries, which have never been implemented in PR.

Also, there remains the need to better understand and ameliorate the mainly negative views that many stakeholders, and especially commercial fishermen, hold towards the local government enforcement agency, PRDNER. Many stakeholders interviewed as part of the study believed that the local government did not have the financial or enforcement capacity to manage a no-take MPA. Others felt that the agency was draconian and thus did not foster stakeholder confidence in being fair in the management of a no-take MPA. Finally, a few respondents perceived the government as having failed to adequately protect the regional natural reserves, including Canal Luis Peña No-take Natural Reserve, in Culebra Island, and ALCNR, and argued that the enforcement agency could not handle additional management tasks. These examples are raised here to highlight the pervasive views held by many stakeholders concerning the government and to recommend that any no-take MPA designation process consider improving stakeholder understanding of agency’s missions and objectives, its management actions and accomplishments, and an overall rehabilitation of the agency’s image in relation to stakeholder trust.

4.4. Ecosystem-Based and Community-Based Model Integration

Information from ecosystem-based and community-based participatory models was successfully merged to create a multi-criteria data matrix to evaluate and rank candidate no-take MPAs. MPA designation typically relies in scientific information (i.e., oceanographic, biological, ecological), but rarely in societal, which might explain why so many MPAs may fail to meet their objectives [15] . Because the placement, design, and management of no-take MPAs are all related to the intended goals, the most crucial information is that related to the specific objectives the MPA is designed to achieve which is essentially societal [51] . Literature concerning the human dimensions of no-take MPAs has demonstrated both the need to consider human uses in MPA management processes [33] [52] -[54] , and to incorporate public participation at the various stages of the MPA cycle, from designation [55] [56] , to reassessment and monitoring [57] -[59] . Uses and concerns, when left unidentified, can undermine the effectiveness of no-take MPAs, manifested as incomplete characterizations of socioeconomic uses [55] , inter-group conflicts [56] -[61] , and mistrust of management intentions or processes [62] [63] . Conversely, when human dimensions are integrated into the decision-making process in no-take MPA designation stage and throughout the management process, MPAs often perform to their expectations rather than being relegated as “paper parks”. Successful no-take MPA designation approaches, such as that used for the designation of the Dry Tortugas Ecological Reserve in the Florida Keys National Marine Sanctuary [64] , are those that accept stakeholder knowledge, acknowledge stakeholder concerns, and engage stakeholder participation in a transparent and consensus-building process. A combination of base community support, management measures and enforcement of regulations contributed towards positive ecological trends in MPAs [65] . Societal integration and participation has also been identified as key elements for successful implementation of ecosystem-based management of MPAs [66] and for climate change adaptation [67] . MPAs often attract increased numbers of tourists, often resulting in increased water-based visitation and impacts [68] . Ecosystem-based and community-based participatory model integration will allow for reaching consensus regarding the identification of key problems, potential solutions to diminish such impacts, and for the implementation of management strategies.

5. Conclusions

This study addressed the spatio-temporal changes in fish community structure at ALCNR within a 15-year time span (1997-2012), identified areas of convergence between different stakeholder groups, ranked candidate no-take MPA sites based in multiple criteria, evaluated the preferred methods of public participation within and between community groups, and determined community expectations of no-take MPA benefits and costs. Information gathered contributed to devising and prioritizing strategies by which to maximize coral reef-associated fisheries protection while enabling public participation and maximizing community support for no-take MPAs. No-take MPA implementation linked to habitat protection and management can be an important tool to recover already depleted fish populations and significantly depleted coral reefs within ALCNR that have also been impacted by climate change. We strongly recommend the designation of three areas within the reserve as no-take zones, based on a consensus model more compatible with the options supported by fishermen and other user groups, which can simultaneously protect overexploited resources from fishing and by a myriad of recreational activities. We particularly recommend protecting critical areas that still support healthier fish populations, and at the same time protecting local fishermen’s livelihoods by maintaining their principal fishing grounds open to fishing.

A final recommendation is that there remain several critical research needs. The first is to understand the coral reef ecosystem configurations possible under different future scenarios of stress levels, habitat types and biogeographic location [69] . There is also a need to understand how ecosystem processes will be influenced by changing environmental conditions and species compositions, particularly in the context of typical non-sustainable coastal and tourism development practices in tropical islands [70] which results in significant sediment loads to coral reefs [71] -[74] . Finally, there is a need to assess which management approaches will be most appropriate for novel coral reef future scenarios with potentially reduced species diversity, minimal living coral cover, lower functional redundancy, and compromised resilience. There would be a need to reassess the effects the effects and appropriateness of current management approaches, and trialing innovative management options that manage exploited but functional systems [69] . Such approaches include ecosystem-based and community-based participatory model integration.

The approach used in this study conferred a series of benefits and advantages, including: 1) a relatively open-ended interviews allowing for a broad discussion on MPA management strategies; 2) important baseline biological data and stakeholder and interest group participation in the development of management strategies for the MPA; 3) baseline information useful to develop an ecosystem-based long-term ecological monitoring program; 4) a study that targeted at specific user groups with a variety of views, including those of consumptive and non-consumptive stakeholder groups, special interest and public interest groups, and others; 5) an approach that incorporated both qualitative and quantitative aspects of social science in that it used multiple tools such as semi-structured interviews, focus group sessions, as well as participatory surveys, to complete a comprehensive characterization of community and visitor views on an MPA; and 6) a study that relied on inputs at various levels and disciplines; the amount of information collected was maximized, and results are applicable to similar areas through the Caribbean. In the context of rapid climate change impacts and increasing exploitation of fishery resources, the integration of ecosystem-based and community-based models fostering strong stakeholder participation has become an important strategy to restore depleted resources and rehabilitate coral reef ecosystem resilience, functions and benefits. Coral reefs have changed in unprecedented ways. There is a need to explore new management approaches, assess changes in ecosystem services, and investigate how human societies can adapt and respond to novel futures [69] . Ecosystem-based and community-based model integration may be critical to foster long-term community-based integration to develop and implement mitigation strategies for climate change impacts, as well as to improve governance. Failing to take action now may compromise important reef resources in the near future.

Acknowledgements

This study was made possible thanks to the support provided by NOAA’s Coral Reef Conservation Program (NA05NMF4631050). Support was also provided to E.A.H.D. by the Caribbean Coral Reefs Institute of the University of Puerto Rico (UPR) (NA04NOS4260206), and by the National Science Foundation through the Center for Applied Tropical Ecology and Conservation of UPR (HRD #0734826). Our appreciation to Robert Matos (Former Director, Reserves and Refuges Bureau, PRDNER) and Héctor Horta (Former Officer Manager, ALCNR, PRDNER) for logistical support during some phases of the project.

References

NOTES

^*Corresponding author.