Effects of tourist visitation and supplementary feeding on fish assemblage composition on a tropical reef in the Southwestern Atlantic

The effects of tourist visitation and food provisioning on fish assemblages were assessed by visual censuses (stationary technique) carried out in a tropical reef in Northeastern Brazil. Comparisons of species abundance, richness, equitability, and trophic structure in the presence (PT) and absence (AT) of tourists suggest that tourist visitation and supplementary food influenced the structure of the fish assemblage, as follows: (a) diversity, equitability and species richness were significantly higher on the AT period, while the abundance of a particular species was significantly higher during PT; (b) trophic structure differed between the AT and PT periods, omnivores being more abundant during the latter period, while mobile invertivores, piscivores, roving herbivores and territorial herbivores were significantly more abundant on AT. Reef tourism is increasingly being regarded as an alternative to generate income for human coastal communities in the tropics. Therefore, closer examination of the consequences of the various components of this activity to reef system is a necessary step to assist conservation and management initiatives.


Introduction
Tourism is the fastest growing economic activity of the world, natural area visitation being its most rapidly growing segment.These areas are visited to "get away from it all", and this involves many recreational activities (Newsome et al., 2002).
Wildlife viewing and equivalent types of nature-based tourism are often promoted as the option of choice for the direct use of wildlife (Honey, 1999), a view that is increasingly being embraced by the promoters of reef tourism -a form of marine recreational tourism that according to Alevizon (2004), generally encompasses activities to "facilitate client viewing and/or other human-wildlife interactions, such as touching and handling".
Food provisioning by tourists is among the different forms of human-wildlife interaction in reef areas (Newsome et al., 2004), and has been considered a source of disruption or alteration to the natural distribution and abundance patterns of marine fishes (see Perrine, 1989;Cole, 1994;Hawkins et al., 1999), as well as of the "fed" individuals and the ecosystems of which they are part of (Alevizon, 2004).Furthermore, Newsome et al. (2002) have drawn attention to a real danger of wildlife becoming accustomed to, and dependent on humans for food in tourism areas, due to the potentially serious health and behavioral implications, particularly for a rare species or a species with a restricted population.
Despite the evidence that visitation may be harmful to overall reef communities when not organized and/or controlled (see Salvat, 1987;Roberts & Harriot, 1994;Prior et al., 1995;Allison, 1996;Harriot et al., 1997), reef tourism is often regarded as an important income source to fishermen communities where fisheries have declined (Cesar, 1996;Vogt, 1996), while helping to protect reefs by providing an incentive to their conservation (Hawkins et al., 1999).In fact, different human coastal communities in the tropics are already in the process of shifting from dependence on fishing to dependence on tourism (e.g., Diedrich, 2007).
In this paper we examine the relationships between fish assemblage composition, tourist visitation, and food provisioning in a tropical reef in Northeast Brazil.The study site (Picãozinho reef) is a highly sought after tourist destination where fish are usually fed by visitors with bread, cookies or fish food (MII, pers.observ.).
Our main goal was to assess the effects of supplementary feeding on the reef fish assemblage structure by comparing its abundance and diversity between two periods (presence versus absence of visitors).

Study site
Picãozinho (07°06'15"S, 34°48'45"W) is a coastal tropical reef with a maximum depth of 6 m, located about 1500 m off the coast of João Pessoa, Paraíba, Northeastern Brazil.The seascape in the study area is mostly characterized by seaweeds such as Caulerpa racemosa, Halimeda opuntia and Dyctiopteris delicatula, while corals such as Palythoa caribaeorum, Zoanthus spp., Siderastrea stellata and Musimilia hartii are the dominant components of the benthic community (Young, 1986;Medeiros et al., 2007).Recreational tourism has been occurring at Picãozinho reef since the mid 80's, without any supervision.Supplementary feeding, trampling, boat anchoring and water turbidity increase are problems commonly observed in the area during visitation (Ilarri et al., 2007).
The fish assemblage was assessed using the stationary visual census method adapted from Bohnsack & Bannerot (1986).At each point sampled, the numerical abundance of active reef-associated fishes was estimated.Sites were randomly chosen within the area mostly visited by tourists, from a set of sites with comparable zonation and tourist influence.All censuses were carried out at low tide (-0.2 to 0.3 m).In order to avoid bias due to tidal amplitude, censuses were conducted within small intervals from one another.
Visual censuses were carried out from February 2006 through August 2006 (dry season), at depths between 0.5 and 3 m.In total we performed 72 censuses (2.5 m radius), 36 during tourist visitation (PT) and 36 when tourists were not present (AT).Prior to visitation, samples were classified as AT, and subsequently as PT.No distinction was made between weekdays and weekends, as tourism took place throughout the week at the reef.The reef was open for tourist visitation while the census data of PT were being undertaken and for the AT, we sampled the exact same area.
Numerical abundances of targeted species were determined by counting all individuals sighted, while size (total length) of individual fish was visually estimated and assigned to one of the following five size classes: 1-5 cm, 6-10 cm, 11-15 cm, 16-20 cm and 21-25 cm.Species were visually identified in situ or through underwater photographs, which were compared with the descriptions found in Humann & Deloach (2002) and Carvalho-Filho (1999).
Species were grouped according to trophic categories well-established in the literature (Randall, 1967;Hobson, 1975;Ferreira et al., 2004) Species richness is expressed as the number of species (S), while diversity and homogeneity were estimated by the Shannon Wiener's diversity index (H´), using the software PRIMER 5.0.Since most data departed from normality (Shapiro-Wilk's W test), comparisons between PT and AT were made using the non-parametric Mann-Whitney U-test (in some cases Z-adjusted values and P-values were considered).A non-metric multidimensional scaling (MDS) was applied to the similarity matrix to visualize the relationships among samples within PT and AT, considering the abundance data of all fish census replicates.Prior to the analysis, a double square-root transformation was applied to the abundance in order to normalize and avoid skew in the data set.Additionally an analysis of similarity percentages of particular species (SIMPER) was conducted to examine potential differences in fish assemblage structure between PT and AT, following Field et al. (1982).Essentially, this procedure computes the average dissimilarity between all pairs of inter-group (PT vs. AT) periods, and then breaks this average down into the separate contributions from each species.These analyses were carried out using the software PRIMER 5.0.
The MDS analysis revealed some segregation between the periods, with most of the assemblage samples from PT located in the middle of the plot, indicating that the samples which are closer together are less distinct; conversely, the AT samples exhibited a more heterogeneous pattern with dissimilarity between samples during tourist visitation at the study site (Fig. 1).The SIMPER analysis revealed that A. saxatilis contributed more to the dissimilarity (61.73%) than any other species (Table 2).A comparison of the fish assemblage attributes indicated that fish diversity, equitability and richness were lower in the presence tourist, while only the total number of fish was higher (Table 3).
The mean number of individuals in each trophic category differed between PT and AT periods as follows: omnivores were nearly four times more abundant during PT than during AT, while five out of the eight trophic categories recorded (mobile invertivores, omnivores, piscivores, roving herbivores and territorial herbivores) exhibited significant differences between periods (Table 4).
During PT, fishes in the 6-10 cm size class comprised over 31.95% of the total abundance.The 11-15 cm size class contained 30.36% of the fishes, while the 1-5 cm size class made up over 26.11%.During AT, fishes in the 6-10 cm size class made up over 32.15% of the fishes, the 1-5 cm size class contained 27.38% while the 11-15 cm size class constituted 24.24% of the total abundance.The least abundant size class both during PT and AT was the 21-35 cm class, contributing only 0.73% and 1.21%, respectively.The abundance of three size classes (6-10 cm, 11-15 cm and 16-20 cm) was significantly higher at PT period, while the other size classes showed no statistical differences between the periods (Table 5).

Discussion
The consequences of ill-planned, uncontrolled recreational tourism to coastal ecosystems have been increasingly acknowledged, however there have been few attempts to specifically examine the possible effects of supplementary feeding on reef fish assemblages.Our findings corroborated the results of previous studies (Cole, 1994;Millazzo et al., 2005), which have shown that food provisioning plays a role in shaping the structure of the fish assemblages.The consistent discrepancies in fish abundance, diversity, evenness and size classes abundance between the AT and PT periods at the study site were mainly related to the extremely high abundance of a single species (Abudefduf saxatilis), commonly known as sergeant major.Although usually abundant in many tropical Atlantic reefs (Humann & Deloach, 2002), A. saxatilis is seldom found in high abundances on Brazilian Northern reefs (Molina et al., 2006).Due to its generalist omnivorous feeding habits, that species is able to feed on a diverse array of food items and normally shows clear patterns of opportunistic behavior (Deloach & Humann, 1999), being also capable of shifting between food sources as a result of seasonal and historical environmental disturbances (Ferreira et al., 2004).
At Picãozinho reef, A. saxatilis was the only fish species observed feeding on the supplementary food offered by tourists, while the majority of the other species vanished during visitations.Over the years, the supplementary food may have benefited the population of A. saxatilis, thus leading to the high densities found in our study.This could be related to behavioral aspects of this species, since aggressive species tend to benefit from artificial feeding, dominating over non-aggressive species (Orams, 2002).Such changes have the potential to alter Table 1.Trophic categories, number of individuals (N), frequency and relative abundance (% of total number of individuals) of reef fishes in the two periods, at presence of tourists (PT) and at absence of tourists (AT).The families are listed in phylogenetic order, following Nelson (2006); species are alphabetically organized within each family.Rov.Herbiv.= Roving herbivore; Mob.Invert.= Mobile Invertivore; Plankt.= Planktivore; Ter.Herbiv.= Territorial herbivore.the physiological and, possibly, the genetic constitution of fish communities, by promoting natural selection on the most aggressive individuals (Orams, 2002).According to Ferreira et al. (2004), Northeastern Brazilian fish assemblages are dominated by roving herbivores and mobile invertivores.Yet, in our investigation we recorded a remarkable dominance of omnivores -a difference perhaps directly related to the type of supplementary food provided by tourists, which could act selecting species of that trophic group.Differential effects caused by food provisioning on trophic groups have also been found in the studies by Millazzo et al. (2005) and Cole (1994), who detected an increase in the abundance of carnivores and invertivores at their study sites.Furthermore, at the Picãozinho reef, sergeant majors were sighted actively following divers and snorkelers while feeding -a behavioral alteration also possibly triggered by food provisioning.Such behavioral shifts have been documented in the literature (Cole, 1994) and need to be further investigated.
Although the behavior of reef fish species could lead to a differential spatial distribution, probably triggered by olfactory cues and visual stimuli (Bond, 1979), it is very likely that uncontrolled tourist visitation contributed to the shift in the reef fish assemblage structure pattern at the study site.These shifts probably have started primarily as a mere concentration of A. saxatilis individuals in a short period of time, due to the supplementary feeding by the tourists.However, over the years this pattern probably has been incorporated to the fish assemblage structure of the reef, leading to the absolute dominance of A. saxatilis, even in the absence of tourists.This pattern was highlighted by Medeiros et al. (2007), which compared fish assemblages between Picãozinho and a neighboring reef submitted to a different level of tourism influence, and which presented a remarkably distinct fish assemblage structure.
It should be noted, however, that artisanal fishery -which has historically occurred in Picãozinho reef (Ilarri et al., 2007) -may have also played a role in structuring fish assemblage at the study site.Nevertheless, the negative consequences of fishing will be apparent regardless of tourist presence, thus, not influencing our results, where differences between AT and PT are strongly correlated to tourism alone.Reefs are vitally important to the tourism industry in Brazil, particularly in the Northeastern region of the country.Recognizing this, and the need to address the problems resulting from the increased use of reef resources, continued efforts to assess the condition of Brazilian reefs at local, regional, and national scales should be prioritized, particularly through long-term programs.Local management action at the study site should encompass, at minimum, setting areas closed for visitation, limiting the number of visitors, and monitoring of recreational activities.

Fig. 1 .
Fig. 1.Non-Metric Multi-Dimensional Scaling (MDS) plot of fish assemblages samples of the Picãozinho reef in each of the two studied situations.Dark triangles = PT (presence of tourists) and light triangles = AT (absence of tourists).

Table 2 .
SIMPER percentages of the top seven fish species that most contributed to the differences between PT (presence of tourists) and AT (absence of tourists) periods.

Table 3 .
Summary of reef fish community structure variables (median ± SD) plus the Mann-Whitney U-test results of comparisons between the two periods, at presence of tourists (PT) and at absence of tourists (AT).Statistical significance: p < 0.001(*).

Table 4 .
Differences in number of species and abundance (median of the number of individuals ± SD) per census of the eight trophic groups between periods with the presence of tourists (PT) and absence of tourists (AT).Statistical significance: p < 0.001(*); P < 0.05 (**), P > 0.05 (ns, not significant).