This article appears in the following ICES Journal of Marine Science issue: European Symposium on Marine Protected Areas as a Tool for Fisheries Management and Ecosystem Conservation [View the issue table of contents]
Considering multiple-species attributes to understand better the effects of successive changes in protection status on a coral reef fish assemblage
1 IRD-UR CoRéUs/EMH Ifremer, BP A5, 98848 Nouméa Cedex, New Caledonia
2 Université de la Nouvelle-Calédonie, LIVE, BP R4, 98851 Nouméa Cedex, New Caledonia
3 Université de la Méditerranée, Centre d'Océanologie de Marseille, UMR CNRS 6540, Campus de Luminy, Case 901, 13288 Marseille Cedex 09, France
4 Aquarium des Lagons, Nouméa, New Caledonia
5 IRD-UR CoRéUs, Université de Perpignan, 52 Avenue Paul Alduy, 66860 Perpignan, France
6 UMR5244 CNRS-EPHE-UPVD, Université de Perpignan, 52 Avenue Paul Alduy, 66860 Perpignan Cedex, France
Correspondence to B. Preuss: tel: +687 260 791; fax: +687 264 326; e-mail: bastien.preuss{at}ird.fr
Preuss, B., Pelletier, D., Wantiez, L., Letourneur, Y., Sarramégna, S., Kulbicki, M., Galzin, R., and Ferraris, J. 2009. Considering multiple-species attributes to understand better the effects of successive changes in protection status on a coral reef fish assemblage. – ICES Journal of Marine Science, 66: 170–179.The response of fish assemblages to changes in protection status is a major issue for both biodiversity conservation and fishery management. In New Caledonia, the Aboré reef marine reserve harbours more than 500 fish species, and has been subjected to changes in protection status since 1988. The present study investigates the impact of these changes on a wide subset of species (213), based on underwater visual counts collected before the opening and after the closure to fishing of this marine protected area (MPA). We analysed the spatial and temporal variability in fish assemblage attributable to protection status, explicitly considering habitat. To understand the successive responses of fish assemblage to fishing and protection, the assessment models included four criteria defining species groups that partition the fish assemblage: trophic regime, adult size, mobility, and interest for fishing. We could therefore identify the negative impact of opening the MPA to fishing on piscivores and highly mobile species. Surprisingly, target species were not affected more than non-target species. Model results were used to identify species groups that respond to fishing and protection. These results utilize fisheries-related criteria to provide new insight into the response of fish assemblages to protection from the perspective of MPA monitoring.
Keywords: assessment model, coral reef ecosystem, fish assemblage, fishing effect, MPA effect
Received 9 September 2007; accepted 6 August 2008.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
A key issue in tropical areas is the impact of the huge diversity found on tropical reefs on the effect of management measures. In particular, one can question the relative impact of such measures on the species composition of fish assemblages and their capacity for adaptation and their resilience, knowing that these highly diverse assemblages are subject to a number of anthropogenic and natural disturbances (Hughes, 1994; Connell, 1997; Nyström and Folke, 2001). In New Caledonia, recreational and subsistence fishing activities, mining and industrial activities, a growing population, and the development of tourism have all affected the reef systems negatively (Labrosse et al., 2000).
Marine protected areas (MPAs) are acknowledged as a major tool for biodiversity conservation and fishery management (Agardy, 1994; Sumaila et al., 2000), but they can also be seen as an instrument for an experimental approach aimed at improving our understanding of how communities respond to fishing. In this respect, MPAs are a tool for actively adaptive management in the sense of Walters and Hilborn (1976).
Many studies have focused particularly on marine reserves and assessed their effects on fish assemblages and on marine organisms (see reviews by Roberts and Polunin, 1991; Russ, 2002; Halpern, 2003; Pelletier et al., 2005). In most papers, the impact of reserve or fishing experiments was assessed on a taxonomic basis (usually families; e.g. Alcala, 1988; Jennings et al., 1996; McClanahan and Kaunda-Arara, 1996; Russ and Alcala, 1998) or for species having particular importance in the context of the study (e.g. García-Rubiés and Zabala, 1990; Letourneur, 1996; Edgar and Barrett, 1999; Johnson et al., 1999). Classical methods have tested differences in density, biomass, and species richness between the reserve and a comparable zone. Few studies have evaluated reserve effects by grouping species other than at a taxonomic level; most have considered only a limited number of species or sometimes trophic groups (e.g. Russ and Alcala, 1996).
Assessment for a given species or species group does not provide a synoptic view of the impact of the reserve. Assessing the impact of a reserve at the fish-assemblage level is more preferable for providing scientific elements for an ecosystem approach to management (Botsford et al., 1997; Jennings and Kaiser, 1998). This is particularly true in coral ecosystems, where diversity is particularly high (more than 600 observable species in New Caledonia according to Kulbicki et al., 2007). The structure of the assemblage can be analysed based on the taxonomic, ecological, or economic grouping of species. This approach has been at the heart of several studies on the effects of MPAs on the reef fish assemblages of New Caledonia (Amand et al., 2004; Ferraris et al., 2005; Kulbicki et al., 2007). Those first studies demonstrated the advantage of multispecies groups over single-species approaches in these highly diverse systems.
The objective of this study is to assess the effects of successive changes in protection status on the fish assemblage of the Aboré reef reserve. Located in the South Lagoon of New Caledonia, southwestern Pacific Ocean, this MPA had been in place for three years, when part of it was opened to fishing. Two years later, it was closed to fishing again. Fish assemblages were surveyed before and after the opening and the closure. We anticipated that changes in the fish assemblage would vary according to environmental factors and hypothesized that species attributes such as diet, species size, home range, or interest for fishing would be important factors in this variation. For instance, highly targeted species may be more affected by fishing than less-valued ones, and mobile species may be less affected than sedentary ones.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The Aboré reef reserve MPA has been studied by Amand et al. (2004), Ferraris et al. (2005), and Kulbicki et al. (2007), and the present study contributes distinct datasets and time-frames. Earlier studies used data from censuses where all fish species were recorded, resulting in more than 400 species observed, but it could only be performed by highly trained divers, and the number of replicates was necessarily low (69) because these censuses were time-consuming. Data used in the present study pertain to a restricted list of species (213), but correspond to a much larger number of transects (212) and could therefore be performed by less-trained divers.
Several factors that might explain the variations of observed fish counts independently of protection status were investigated. Then, the impacts of opening the area to fishing and the second closure of the area were tested using a range of biological responses, such as species richness, abundance, biomass, and mean individual size. The fish assemblage was partitioned, based on mobility, trophic regime, adult size, and interest for fishing, and each response was calculated according to the different partitions. By crossing variables and partitions of fish assemblage, the metrics per species group could be analysed, and the effects of changes in protection status at the fish-assemblage level were assessed simultaneously. We were able to demonstrate the benefits of our approach, by comparing these results with an overall approach considering responses averaged over all species. The partitioning provided insight into the effects of such changes on the fish assemblage.
Study area
The Nouméa lagoon, located in New Caledonia, southwestern Pacific Ocean (Figure 1), is a large lagoon seascape, including coral reefs, where several marine reserves (no-take zones) were established in the 1980s to protect the coral reef ecosystem from the impacts of fishing. The present study took place in the Aboré reef reserve, located on a barrier reef that is 25 km long and 20 km off Nouméa, and representing about 15 000 ha. The area was closed to fishing from 1988 to 1993; then two-thirds of the reef (area B) was reopened to fishing from August 1993 to July 1995 for a fishing experiment. The opening of the reserve immediately resulted in intense fishing pressure; in the first two weeks, 800 boats were observed, and the fish yield was 8.7 t, as estimated from a sampling of 57% of these boats (Sarramégna, 2000). These levels more or less corresponded to what had been observed previously over an entire year (Sarramégna, 2000). The whole reef has been closed to fishing since August 1995 (Figure 2; Table 1).
|
|
|
Sampling protocol
The impact of allowing fishing in the reserve and the restoration effect after the final closure on fish assemblages were monitored from three surveys. In July 1993, a survey was performed using 60 transects in five locations spaced along the reef (Figure 2), just before resumption of fishing on two-thirds of the reef (area B; Figure 2), whereas one-third remained closed (area A). In July 1995, a second survey of 110 transects was conducted on both area A, closed to fishing, and area B, opened to fishing. Area B was closed again to fishing in September 1995, resulting in the complete closure of the Aboré reef for the second time. Six years later, in 2001, a third survey of 42 transects was conducted on the Aboré reef MPA (Table 1). The experimental design stratified the reef into two geomorphological zones: the inner reef flat and the inner reef slope (Figure 2). The reef flat is a very shallow area ranging from 0.7 to 1.5 m, whereas the inner reef slope is an intermediate zone between the reef flat and the sandy bottom lagoon, with inner spurs and grooves (Battistini et al., 1975). For each of the three years surveyed, the five sampling locations, regularly spaced along the reef, were selected to ensure a good longitudinal coverage where both geomorphological zones were present (Figure 2). At each location for each geomorphological zone, at least two transects, 500 m apart, were sampled (see Table 2 for sampling design by year, area, and geomorphological zone).
|
Our study is based on the so-called commercial transects, during which 213 species corresponding to 32 families from a restricted list were counted, including all fished species plus a number of species of scientific interest. Quantitative estimates of abundance of coral reef fish, using the distance-sampling method (Buckland et al., 2001), were made using an underwater visual census (UVC; Kulbicki and Sarramégna, 1999). Transects of 50 m were marked by lines set on the bottom. A diver swam along the transect line and recorded fish from the species list mentioned above. For each observation, the diver recorded the species and size of the fish. Biomass of each fish was calculated through available length–weight relationships (Kulbicki et al., 2005; Kulbicki, 2006).
Partitioning the fish assemblage
Choosing criteria for constructing species groups raised the question of how to define partitions of the fish assemblage that would be relevant to the impact of protection status. Four criteria were used: mobility, interest for fishing, trophic guild (feeding habits), and adult size. Following Grimaud and Kulbicki (1998), four mobility groups were defined: (i) territorial species with a very restricted range (usually <10 m2); (ii) sedentary species with a restricted range (ten to several hundred square metres); (iii) weakly mobile species, often distributed over the entire reef area (up to several thousand square metres); and (iv) highly mobile species, usually foraging over very large areas. Species groups corresponding to distinct interests for fishing were defined based on I. Jollit and colleagues (unpublished data): (i) highly targeted by spearfishing, (ii) moderately targeted by spearfishing, (iii) incidentally targeted by spearfishing, (iv) highly targeted by linefishing, (v) moderately targeted by linefishing, (vi) bycatch of linefishing, and (vii) not fished. The trophic groupings were based on diet composition following the results of Ferraris et al. (2005): (i) piscivores, (ii) macrocarnivores, (iii) microcarnivores, (iv) coral feeders, (v) herbivores, (vi) microalgae feeders and detritivores, and (vii) zooplankton feeders. Five size classes were defined based on adult sizes (Kulbicki, 2006): (i) 0–7, (ii) 8–15, (iii) 16–30, (iv) 31–50, and (v) >50 cm.
Data analysis
We aimed at testing the impact on the whole fish assemblage of both the fishing effect after removal of reserve status and the definitive closure of the B area. Therefore, we assessed these effects at two levels: (i) overall variables per transect, namely species richness, abundance, and biomass summed over all species; and (ii) metrics computed per species group, namely species richness, abundance, biomass, and mean size per species group. Both the criteria to group the species and the variables define a set of metrics.
The methods used to assess the effects of changes in protection status upon the fish assemblage involved both exploratory and inferential techniques. Exploratory techniques relied on the use of non-metric, multidimensional scaling (MDS) to display graphically the similarities between transects measured by the Bray–Curtis coefficient. The representativeness of the plots was evaluated by the stress value (Clarke and Warwick, 2001). Using the PRIMER software, we could visualize the influence of a factor of interest (such as year or habitat) to explain differences among transects.
Inferential techniques include one-way analysis of similarities (ANOSIM), which rests on a permutation procedure to test whether or not a factor significantly explains differences between groups of transects (Clarke and Warwick, 2001). To perform this test (using the PRIMER software), similarities between transects were calculated using the Bray–Curtis coefficient. ANOSIM tests were based on 999 permutations of the transects between factor levels. In addition to ANOSIM, generalized linear models (GLMs; McCullagh and Nelder, 1989) were performed using R software (R Development Core Team, 2007), with suitable data distribution, depending on the variable modelled to test and assess the effects of both removal and closure of area B in 1995, while accounting for other factors that were relevant to explain the variability of the fish assemblage.
Because the UVCs were performed by several divers, we first explored potential diver effects on variations in visual counts, using MDS plots and ANOSIM applied to abundance, biomass, and mean size data. Significant diver effects were detected, so an observer term was included as a factor in subsequent analyses to allow consideration of the variability of counts caused by the diversity of divers.
In addition to the four factors considered for models of overall metrics (year, area, habitat, and diver), models of metrics per species group also included the species group factor, with levels depending on the species-grouping criterion, as described earlier, i.e. mobility (four levels), interest for fishing (seven levels), trophic regime (seven levels), and size class (five levels). These models included first-order interactions between factors (except for the diver factor), and additionally, second-order interactions between year, area, and species group, to assess possible species-group-specific effects of protection status. For overall metrics, one model was fitted for each variable, whereas for metrics per species group, one model was fitted for each combination of variable and species-grouping criterion.
All metrics per species group were modelled in two steps: a binomial model for presence–absence and a lognormal model based on non-zero values of corresponding metrics, following the procedure proposed by Stefánsson (1996). This method is suited for quantitative data with large proportions of zero values that make them unable to meet the assumptions of regular GLMs. Modelling non-zero values led to 16 model fits, crossing four metrics (abundance, biomass, mean size, and species richness per species group) with four grouping criteria. The goodness-of-fit of each model was assessed through adjusted R2 and global Fisher's tests, and the conformity of model residuals to linear model assumptions was checked from standard residual plots and tests (Venables and Ripley, 1997). Once validated, models were selected to eliminate non-significant terms, based on the Akaike information criterion (Akaike, 1974). The significance of each effect was evaluated using the analysis of variance table based on the Type III sums of squares.
Regarding protection status, we were interested in both the fishing effect after removal of reserve status and the definitive closure of area B. The effects of these changes in protection status were assessed and tested through the interaction between the year and the area (A/B) factors (year x area) for overall metrics, and in addition through the interaction between the year, area, and species group factors (year x area x species group) for metrics per species group. When only the first-order interaction between area and year was significant, all species groups responded in the same way to changes of protection status. The magnitude and direction of the effect was quantified by computing adjusted means per area, year, and species group. Adjusted means correspond to predictions of the modelled metric, based on the significant effects of the model, thus leaving aside residual variations. In the analysis, adjusted means were computed for the year, area, and species group factors, while controlling for the diver effect. Multiple comparisons were performed using the Bonferroni correction for the following differences in adjusted means: (i) spatial difference between areas A and B in 1993, i.e. before the removal of reserve status; (ii) temporal variation in areas A and B between 1993, 1995, and 2001; (iii) spatial difference between areas A and B in 1995 and 2001. When the second-order interaction between area, year, and species group was significant, simultaneous confidence intervals were constructed per species group.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Assessment of changes in protection status on overall metrics per transects
Variations in species richness, abundance, and biomass per transect depended on the area (Figure 3; Table 3). Mean species richness increased between 1993 and 1995 in area A, then decreased in 2001, whereas in area B, it increased over the whole period. Overall abundance per transect consistently decreased in both areas over the same period, with a stronger decrease between 1993 and 1995 in area B and a stronger decrease between 1995 and 2001 in area A. Similarly, mean biomass decreased in both areas, with a much more important decline in area A between 1993 and 1995.
|
|
The observed trends were validated by significant models for overall metrics and for metrics per species group. In all models, residuals (not reported) conformed well to linear model assumptions, the global F-test was highly significant, although the variance explained by these models was quite low (Table 4). For these overall metrics, the diver effect was significant and well accounted for by the models (Table 4). There was no significant interaction between area and year, meaning that no effect of the 1993 fishing event or the 1995 closure could be detected from the species richness, abundance, or biomass data. In contrast, both abundance (p = 0.0003) and biomass (p = 0.0027) decreased over the years surveyed, as observed previously (Table 4). Species richness varied by area (p = 0.0003; Table 4), and both biomass and species richness were significantly lower in the inner reef flat zone (p = 0.0003 and p < 0.0001, respectively).
|
Changes in protection status and species attributes
The probability of occurrence of all adult size groups was lower in area A than in area B in 1993, whereas the difference was the other way around in 1995, revealing a significant effect of protection status (p < 0.02; Table 5). In all presence–absence models, the species group factor was highly significant (p < 0.01), illustrating differences in average occurrence between species groups, irrespective of how they were defined.
|
For non-zero values, model fits were generally much better than with overall metrics (Table 6). In particular, the R2 values of models ranged from 0.34 to 0.77, with best fits for abundance and species richness. The increase in explanatory power was accomplished by retaining a much larger set of significant effects in the selected models, and in particular, the interactions between two or more factors. To rank factors according to their influence, we used the number of significant effects in the 16 models fitted (Table 7). Diver identity and species group effects were significant in all models, so explained more variance than spatial or temporal effects. First-order interactions involving the species-group factor were also highly significant, underscoring the differences in response between the groups to the effects of year, area, and habitat.
|
|
The change in protection status significantly affected mean size in the groups defined by fishing interest (p < 0.01). Protection status also significantly affected abundance in the different mobility grouping (p < 0.01) as well as biomass in the different trophic groupings (0.05 < p < 0.01). The adjusted means (Figures 4–6), based on year, area, and species group effects (see Material and methods section), illustrate the species group effects and the additive diver effect (Table 5).
|
Regarding the mobility criterion, between 1993 and 1995 (fishing period in area B), the abundance of sedentary species rose in area B and declined in area A, whereas the abundance of all other groups declined, particularly between 1995 and 2001 (Figure 4), but the only significant variation over time was the decline of highly mobile fish in area B between 1993 and 1995, which can be explained by the fishing impact. The abundance of this group remained stable after 1995, corroborating a relative protection effect.
Considering trophic groups (Figure 5), a decline in biomass between 1993 and 1995 was observed for all groups in both areas except zooplankton feeders. Yet, this decline was only significant for piscivores in area B in relation to the opening to fishing (note that 43% of piscivores were highly or moderately targeted by both spear- and linefishing). Macrocarnivores demonstrated a global decline of biomass in both areas, with biomass significantly lower in area B (fished) than in area A (no-take) in 1995. Between 1995 and 2001, all trophic groups tended to decline, except piscivores whose biomass increased slightly. Similar trends in abundance were also observed in area A. Zooplankton feeders displayed an inverse pattern in area B. Yet, almost half of the species in this group are highly or moderately targeted by spearfishing.
|
Finally, mean size in the different fishing interest groups demonstrated a variety of patterns over time in areas A and B (Figure 6). Between 1993 and 1995, mean size decreased for most groups in both areas, except for species moderately targeted by linefishing, bycatch of linefishing, and for unfished species. As expected, unfished species were not affected by the opening to fishing, and species moderately targeted by linefishing and bycatch of linefishing declined in area B while remaining stable in area A between 1993 and 1995. After the final closure of area B between 1995 and 2001, the mean size of several groups declined further in both areas. The decline was mitigated in both areas for species highly targeted by linefishing, and in area B for bycatch and moderate targets of linefishing. For these two groups, the decline in mean size was larger in area A. The mean size of species caught by spearfishing declined but not significantly and over the 3 years in both areas.
|
Note that no species-group-specific protection status effect was found for the adult size criterion (Table 6).
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Spatio-temporal variations of the fish assemblage
First-order interaction of year, area, and habitat with species groupings demonstrated the complexity of variations in the measured abundance, species richness, and biomass of the fish assemblage. The factors that determined significant single effects in the models were the year, habitat, and diver factors. The significance level of the diver factor required the explicit inclusion of this factor in our models to control for observer variability. This aspect is often ignored or omitted in the literature.
Regarding temporal variations, a significant effect involving at least the year factor was found in 18 of the 23 fitted models in the whole study. Such variations corresponded to decreases over time of the studied metrics, a result that was also noted by Kulbicki et al. (2007), using a different approach. The causes of such temporal variations remain poorly understood and are obviously linked to factors that cannot be accounted for in such models. Undoubtedly, environmental fluctuations and events explain some of these variations and mask the effects of changes in protection status. For example, the large-scale oceanographic and climatic features such as El Niño Southern Oscillation (ENSO) events are known to influence water conditions greatly (temperatures, for example) and may seriously affect habitats and disrupt benthic populations and their reproductive success (Allison et al., 2003). Cyclonic events are also likely in the studied area, and although no critical event took place during the period under study, consequences of such events may be observed on longer time-scales. Other events, such as the strong winds that prevailed during the 1995 survey, could also explain part of the variations under consideration. However, without sampling outside of the MPA and over a long temporal series, it is difficult to detect effects of long-term phenomena.
On the basis of inferential models, we could assess the effects of changes of protection status through the interaction between year and area factors for a range of metrics. Overall abundance, biomass, and species richness proved not to be sensitive to changes in protection status. Regarding metrics per species group, a few metrics, namely piscivore biomass, the abundance of highly mobile species, and the mean size of linefishing bycatch, displayed significant variations that were consistent with changes in protection status endured by the fish assemblage. Several other metrics demonstrated non-significant variations, which tended to be consistent with these changes, e.g. the biomass of herbivores, the mean size of species moderately targeted by linefishing, and species highly targeted by linefishing. Other variations could not be easily related to changes in protection status and, overall, there were few significant year x area interactions. Often, declining patterns observed in both areas A and B suggest a strong connection between these two adjoining areas. Such exchanges would inevitably reduce spatial differences in fish assemblage between areas A and B and therefore contribute to the lack of significant protection effects. The spillover of individuals from area A to area B would mitigate the decrease in area B and/or the regeneration in area A. Larval dispersion and larval settling depend on hydrodynamics that operate at scales larger than the MPA and may also contribute, but are poorly known in this area. Fish movements may be particularly important in such coral reef formations consisting of linear barriers and multiple islands and coral patches. This point is confirmed by Chateau and Wantiez (2009), who recently demonstrated that fish mobility in the Caledonian lagoon is more important than previously believed.
Habitat was a determining factor for explaining variations of the fish assemblage, even approximated at the scale of the geomorphological zone (a significant factor in 14 models of the 23 fitted). Geomorphological zones indeed correspond to distinct depth and coral type (Ferraris et al., 2005). Therefore, environmental variations affecting coral cover and reef structure may ultimately be reflected in spatio-temporal variations of fish assemblages. Fine-scale changes in habitat structure may help to improve the explanatory power of the models (Ferraris et al., 2005), but such data were not available for inclusion in our analysis. Ferraris et al. (2005) demonstrated that although accounting for fine-scale habitat data improved model fits, the degree of significance of the effects was only improved marginally. In addition, poaching may occur in the MPA, mitigating the effect of protection. Another hypothesis is that the general decline of fish populations in the whole southwestern lagoon is related to the increase in human population (occurring in the Nouméa region) and the evolution of fishing methods. Fish in the reserve migrate to zones outside of the MPA as these other areas become depopulated. The simultaneous decline in abundance of most of the groups, fished or not, argues against the direct impacts of fishing.
Partitioning the fish assemblage
One of our objectives was to utilize species attributes to understand better the consequences of changes in protection status on the fish assemblage. Partitioning the fish assemblage according to a range of criteria provided a variety of insights and is a step towards an ecosystem approach to MPA assessment. In fact, the models of overall metrics displayed few significant effects and none in relation to changes in protection status. In contrast, metrics per species groups revealed a larger number of significant effects, most of which included the species-group factor, which means that, for a given species-grouping criterion, the variation of the metric modelled differed across the species groups. Therefore, corresponding models explained a much larger fraction of variance (up to 77%), than models of overall metrics (<30%). Therefore, the inclusion of species attributes in the models improved the assessment. This method revealed that piscivores and highly mobile species were groups that react the most to opening to fishing (Figures 4 and 5).
However, some of our results are not intuitive. Hence, significant effects of changes in protection status were detected for very mobile species, which a priori can move easily between areas A and B. This kind of effect is generally not expected. It is important to note, therefore, that 58% of the species of the highly mobile group are also species targeted or highly targeted by spear- or linefishers. Yet, the abundance of these fishing groups was not significantly affected by changes in protection status. One could hypothesize that, after the opening, highly mobile species may have left area B for other reef areas with lower fishing pressure, including (but not exclusively) area A. Likewise, one could argue that partitioning species according to their interest for fishing gave few striking results; species strongly targeted by the main fishing gears demonstrated little sensitivity to the opening of fishing. Beyond considerations of area connectivity, fish mobility, and interest for fishing, these counter-intuitive results raise the question of defining species groups that are relevant to assessing the effects of changes in protection status. One could therefore consider defining groups based on the combination of two or more criteria.
Sampling considerations
Given the complexity of the dataset, models appropriate to test the effects of changes in protection status had to include a relatively large number of explanatory factors: year, area, habitat, and diver. Although the number of observations was overall large (212 transects), it may not have been sufficient to unravel the variability attributable to these four factors.
In other studies of the Aboré reef fish assemblage, Ferraris et al. (2005), Amand et al. (2004), and Kulbicki et al. (2007) used a different set of observations based on the census of all observed species for only two years, 1993 and 1995. The dataset included only ca. 70 transects vs. more than 200 (out of which 170 were for both 1993 and 1995) in the present study. A larger number of significant interactions involving year and area were found in these two studies, including intuitive results. Although the data had been collected by several divers, subsequent variability did not mask the effects.
Note that the works mentioned above only dealt with the 1993–1995 variation, a fishing effect that was probably more conspicuous than a restoration effect. Yet, in the present study, this fishing effect was not so obviously detected. This lack of significance raises the question of additional sources of variability, such as the diver effect (between-diver variability in our dataset and difference in divers between the other references and the present study). It may also be related to the issue of the species list retained for the visual counts. Note that the stratification of the sampling scheme and the geographical range of stations were the same in both datasets.
The present results lead us to conclude that, in this case, additional sources of variation prevented us from detecting the effects of changes in protection status (particularly the restoration effect), such as those mentioned above. When it comes to assessing restoration effects between 1995 and 2001, data are only available for 2001, with a smaller number of transects than the other two dates. Monitoring the restoration of the fish assemblage would require observations collected at several dates after the final closure.
Assessing the response of fish assemblages to changes in protection status is a major issue for fishery management. This work gives original insight into the issue of designing adequate protocols for monitoring MPA in terms of conservation of biodiversity and sustainable exploitation of resources. In this respect, considering species attributes was useful, and the partitioning criteria that are considered provided a variety of insights into the effects of such changes on the fish assemblage. Although based on the case of a coral reef ecosystem, our findings may be applied to other contexts.
|
Acknowledgements |
|---|
Data collection was funded by the Province Sud (DRN), IRD, and the Programme National pour l'Environnement Côtier (PNEC). The authors thank the other scientists who helped collect the data, in particular G. Mou-Tham (IRD Nouméa), P. Labrosse and E. Clua (CPS Nouméa), and C. Chauvet (LERVEM, Université de Nouvelle-Calédonie). We pay special homage to Pierre Thollot who tragically died in a helicopter accident in November 2000. This work was made possible through a grant by the PNEC.
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Agardy M. T. Advances in marine conservation: the role of marine protected areas. Trends in Ecology and Evolution (1994) 9:267–270.[CrossRef]
Akaike H. A new look at statistical model identification. IEEE Transactions on Automatic Control (1974) 19:716–722.[CrossRef]
Alcala A. C. Effects of marine reserves on coral fish abundances and yields of Philippine coral reefs. Ambio (1988) 17:194–199.
Allison G. W., Gaines S. D., Lubchenko J., Possingham H. P. Ensuring persistence of marine reserves: catastrophes require adopting an insurance factor. Ecological Applications (2003) 13:S8–S24.[CrossRef][Web of Science]
Amand M., Pelletier D., Ferraris J., Kulbicki M. A step toward the definition of ecological indicators of the impact of fishing on the fish assemblage of the Aboré reef reserve (New Caledonia). Aquatic Living Resources (2004) 17:139–149.[CrossRef][Web of Science]
Battistini R., Bourrouilh F., Chevalier J. P., Coudray J., Denizot M., Faure G., Fisher J. C., et al. Eléments de terminologie récifale indopacifique. Téthys (1975) 7:1–111.
Botsford L. W., Castilla J. C., Peterson C. H. The management of fisheries and marine ecosystems. Science (1997) 277:509–515.
Buckland S. T., Anderson D. R., Burnham K. P., Laake J. L., Borchers D. L., Thomas L. Introduction to Distance Sampling: Estimating Abundance of Biological Populations. (2001) Oxford, UK: Oxford University Press. 432.
Chateau O., Wantiez L. Movement patterns of four coral reef fish species in a fragmented habitat in New Caledonia: implications for the design of marine protected area networks. ICES Journal of Marine Science (2009) 66:50–55.
Clarke K. R., Warwick R. M. Change in Marine Communities: an Approach to Statistical Analysis and Interpretation (2001) 2nd edn. Plymouth, UK: PRIMER-E.
Connell J. H. Disturbance and recovery of coral assemblages. Coral Reefs (1997) 16:S101–S113.[CrossRef][Web of Science]
Edgar G. J., Barrett N. S. Effects of the declaration of marine reserves on Tasmanian reef fishes, invertebrates and plants. Journal of Experimental Marine Biology and Ecology (1999) 142:107–144.
Ferraris J., Pelletier D., Kulbicki M., Chauvet C. Assessing the impact of removing reserve status on the Aboré Reef fish assemblage, New Caledonia. Marine Ecology Progress Series (2005) 292:271–286.[CrossRef][Web of Science]
García-Rubiés A., Zabala M. Effects of total fishing prohibition on the rocky fish assemblages of Medes Islands marine reserve. Scientia Marina (1990) 54:317–328.
Grimaud J., Kulbicki M. Influence de la distance à l'océan sur les peuplements ichtyologiques des récifs frangeants de Nouvelle-Calédonie. Comptes Rendus de l'Académie des Sciences de Paris. Sciences de la Vie (1998) 321:923–931.
Halpern B. The impact of marine reserves: do reserves work and does reserve size matter? Ecological Applications (2003) 13:S117–S137.[CrossRef][Web of Science]
Hughes T. P. Catastrophes, phase shifts, and large-scale degradation of a Caribbean coral reef. Science (1994) 265:1547–1551.
Jennings S., Kaiser M. J. The effects of fishing on marine ecosystems. Advances in Marine Biology (1998) 34:202–351.
Jennings S., Marshall S. S., Polunin N. V. C. Seychelles' protected areas: comparative structure and status of reef fish communities. Biological Conservation (1996) 75:201–209.[CrossRef][Web of Science]
Johnson D. R., Bohnsack J. A., Funicelli N. A. The effectiveness of an existing no-take fish sanctuary within the Kennedy Space Center, Florida. North American Journal of Fisheries Management (1999) 19:436–453.[CrossRef]
Kulbicki M. Ecologie des poissons lagonaires de Nouvelle Calédonie. (2006) Ecole Pratique des Hautes Etudes: University of Perpignan. 194. PhD thesis, + annexes.
Kulbicki M., Guillemot N., Amand M. A general approach to length–weight relationships for Pacific lagoon fishes. Cybium (2005) 29:235–252.
Kulbicki M., Sarramégna S. Comparison of density estimates derived from strip transect and distance sampling for underwater visual census: a case study of Chaetodontiidae and Pomacanthidae. Aquatic Living Resources (1999) 12:315–325.[CrossRef][Web of Science]
Kulbicki M., Sarramégna S., Letourneur Y., Wantiez L., Galzin R., Mou-Tham G., Chauvet C., et al. Opening of an MPA to fishing: natural variations in the structure of a coral reef fish assemblage obscure changes due to fishing. Journal of Experimental Marine Biology and Ecology (2007) 353:145–163.[CrossRef][Web of Science]
Labrosse P., Fichez R., Farman R., Adams T. New Caledonia. In: Seas at the Millennium: An Environmental Evaluation—Sheppard C., ed. (2000) Amsterdam: Elsevier Science. 723–736. 920 pp.
Letourneur Y. Réponse des peuplements et populations de poissons aux réserves marines: le cas de l'île de Mayotte, Océan Indien occidental. Ecoscience (1996) 3:442–450.[Web of Science]
McClanahan T. R., Kaunda-Arara B. Fishery recovery in a coral-reef marine park and its effect on the adjacent fishery. Conservation Biology (1996) 10:1187–1199.[CrossRef][Web of Science]
McCullagh P., Nelder J. A. Generalized Linear Models (1989) 2nd edn. London: Chapman and Hall/CRC. 511.
Nyström M., Folke C. Spatial resilience of coral reefs. Ecosystems (2001) 4:406–417.[CrossRef]
Pelletier D., Garcia-Charton J. A., Ferraris J., David G., Thébaud O., Letourneur Y., Claudet J., et al. Designing indicators for evaluating the impact of marine protected areas on coral reef ecosystems: a multidisciplinary standpoint. Aquatic Living Resources (2005) 18:15–33.[CrossRef][Web of Science]
R Development Core Team. R: A Language and Environment for Statistical Computing. (2007) Vienna, Austria: R Foundation for Statistical Computing. http://www.R-project.org.
Roberts C. M., Polunin N. V. C. Are marine reserves effective in management of reef fisheries? Reviews in Fish Biology and Fisheries (1991) 1:65–91.[CrossRef]
Russ G. R. Yet another review of marine reserves as reef fishery management tools. In: Coral Reef Fishes—Sale P. F., ed. (2002) Elsevier Science. 421–443.
Russ G. R., Alcala A. C. Marine reserves: rates and patterns of recovery and decline of large predatory fish. Ecological Applications (1996) 6:947–961.[CrossRef][Web of Science]
Russ G. R., Alcala A. C. Natural fishing experiments in marine reserves 1983–1993: roles of history and fishing intensity in family responses. Coral Reefs (1998) 17:399–416.[CrossRef][Web of Science]
Sarramégna S. Contributions à l'étude des réserves marines du lagon sud-ouest de Nouvelle-Calédonie: influence des différents statuts de protection sur la structure des peuplements ichtyologiques. (2000) Nouméa, New Caledonia: Université de Nouvelle-Calédonie. PhD thesis.
Stefánsson G. Analysis of groundfish survey abundance data: combining the GLM and delta approaches. ICES Journal of Marine Science (1996) 53:577–588.
Sumaila U. R., Guénette S., Alder J., Chuenpagdee R. Addressing ecosystem effects of fishing using marine protected areas. ICES Journal of Marine Science (2000) 57:752–760.
Venables W. N., Ripley B. D. Modern Applied Statistics in Splus (1997) 2nd edn. New York: Springer. 548.
Walters C. J., Hilborn R. Adaptive control of fishing systems. Journal of Fisheries Research Board of Canada (1976) 33:145–159.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||