Comparative analysis of tidepool fish species composition on tropical coastal rocky reefs at State of Ceará , Brazil

This study compared tidepool fish assemblages within and among habitats at Iparana and Pecém beaches, State of Ceará, Northeast Brazil, using visual census techniques. A total of 8,914 fishes, representing 25 families and 43 species were recorded. The most abundant taxon was Sparisoma spp, followed by Haemulon parra (Desmarest, 1823), Acanthurus chirurgus (Bloch, 1787) and Abudefduf saxatilis (Linnaeus, 1758). Haemulidae was the most abundant family in number of individuals, followed by Scaridae, Acanthuridae and Pomacentridae. Withinand betweensite differences in species assemblages probably reflected environmental discontinuities and more localized features, such as pool isolation episodes, or environmental complexity, both acting isolated or interactively. The locality of Iparana was probably subjected to a greater fishing pressure and tourism than Pecém, a potential cause for the observed lowest fish abundance and biodiversity. We conclude that tidepool ichthyofauna may be quite variable between and within reef sites. Thus, observations taken from or damages caused on one area may not be generalized to or mitigated by the protection of adjacent sites.

Tropical reef fish communities are rich and complex, and comprise species of different ecological strategies, at several life history stages.Coastal intertidal reefs and their associated tidepools are important habitats, especially for young-of-the-year of both coastal and oceanic fish species, some of which of special economic interest (CADDY & SHARP, 1986;GRIFFITHS et al., 2003).MORING (1986) studied the use of tidepools as breeding areas by many fish species, whereas UNDERWOOD & CHAPMAN (1996) observed that rock pools might provide refuges for many intertidal and subtidal invertebrates.In fact, ROSA et al. (1997) stressed the importance of tidepools for both coastal and marine fish populations, as these habitats maybe part of the home ranges for some species.
Tidepools are often used by swimmers, artisanal fishers, and may be impacted by development in the coastal zone.In the state of Ceará, artisanal fishers and seaweed collectors harvest on coastal reefs and tidepools, and have reduced considerably the abundance of some taxon (Dárlio A. Teixeira, pers. comm.).Uncontrolled fish and live rock (reef rocks containing sponges, corals, crabs, shrimps, algae and natural bacteria) collection for the aquarium trade have also negatively impacted fish communities of these habitats (WOOD & WELLS, 1988;MONTEIRO-NETO et al., 2003).Finally, reefs and tidepools have been impacted at several scales due to coastal development, resulting in coastal erosion, habitat loss and reduction of biodiversity (CAMPOS et al., 2003).Therefore, reef fish studies are essential for a precise evaluation of the damages caused by those activities and the elaboration of environmental management plans.
Few studies have addressed reef fishes in Northeast Brazil.Early studies on tidepool fishes were limited to species inventory (ARAÚJO et al., 2000).More recently, research focused on the distribution, abundance and diversity of reef fish fauna (FERREIRA et al., 1995;ROSA & MOURA, 1997;ROCHA et al., 1998;ROCHA &ROSA, 2001a andFEITOSA et al., 2002).ROSA et al. (1997) studied the abundance and diversity patterns of tidepool fishes at State of Paraíba and concluded, among other findings, that Gobiidae, Pomacentridae and Labrisomidae were the most abundant families, in terms of number of individuals.
Reef topography usually limits the array of sampling techniques to highly selective fishing strategies such as hook and line and spearfishing (FERREIRA et al., 1995).Therefore, several complimentary techniques are needed to fully assess reef fish communities (ZAPATA, 1992).Alternatively, underwater visual census broadens the limits of reef fish community studies, allowing the simultaneous data collection on species composition and abundance, ecology and behavior (FERREIRA et al., 1995).
One pitfall of visual census may be the underestimation of cryptic species.ACKERMAN & BELLWOOD (2000) observed that a significant proportion of the total number and biomass of fishes on the reef were not accounted for by visual census.Nevertheless, non-destructive techniques should be recommended whenever reef ecosystem conservation is at stake, due to their little environmental interference and the possibility of fish observation in their natural habitat (ROSA et al., 1997).
The aim of this study was to investigate aspects of the tidepool fish assemblages structure to: (i) provide information on the composition of species found in tidepools of tropical coastal rock reefs at state of Ceará, Northeast Brazil; (ii) quantitatively document the patterns of variation in fish assemblages' structure within and between reef localities; (iii) discuss which environmental variables may be important in influencing the structure of tidepool fish assemblages.

MATERIALS AND METHODS
Underwater censuses were performed in four tidepools located along intertidal reefs at the beaches of Iparana (3 o 41'21"S, 38 o 36'36"W) and Pecém (3 o 32'40"S, 38 o 48'52"W), situated nearly 40 km apart from each other in the coast of State of Ceará, NE Brazil (Fig. 1).Sandstone, Calcium Carbonate and Iron Oxide reefs extended from the beach face to nearly two hundred meters seaward, and were exposed at spring low tides (MORAIS, 1968).
Tidepools selected for this study were characterized by a sand-rubble bottom with sparsely distributed small boulders, and walls partially covered with macroalgae.The pools became isolated from the sea at 0.3 m low tides.
Pools 1 and 2 were located at Iparana, and 3 and 4 at Pecém reefs (Fig. 1).Each pool was visually examined for the percent cover of the substrate (sand-rubble, bare rock, rock covered with algae).The structural complexity was assessed based on frequency of occurrence of each of the three dimensions (presence of pebbles, crevices and holes) and put into one of the categories (high, medium, low).The surface area (A) of each tidepool was estimated by adding up the triangular areas to match the approximate surface area of the tidepool.Average depth (D) was estimated from three measurements taken at random locations along a longitudinal transect in order to estimate the volume (V = A x D).
Visual census methodology was adapted from CHRISTENSEN & WINTERBOTTOM (1981) in respect to observation time, and consisted of an intensive species search, by snorkeling around the internal rim of the tidepool.Each full search lasted less than 25 minutes to avoid repetitive counts.Species were identified and counted by trained divers, and data recorded on PVC boards.Visual censuses were conducted monthly during daytime, at tides lower than or equal 0.3 m (pools became isolated).A total of 96 censuses were conducted from June 1998 to June 2000.When species determination was uncertain in the field, specimens were captured for laboratory identification.These specimens were not included in the statistical analysis.Taxonomic identification was based on BURGESS et al. (1990), CERVIGÓN (1991CERVIGÓN ( , 1993CERVIGÓN ( , 1994CERVIGÓN ( , 1996)), FIGUEIREDO & MENEZES (1978, 1980, 2000), HUMANN (1995), MENEZES & FIGUEIREDO (1980, 1985), RANDALL (1996) and ROBINS (1986).The revisions of Anisotremus moricandi (Ranzani, 1842), Halichoeres brasiliensis (Bloch, 1791) and Halichoeres penrosei Starcks, 1913 were based on ACERO & GARZÓN (1982), ROCHA & ROSA (2001b) and ROCHA (2004) respectively, and were aided by specialists and visits to University and Museum Collections.Sparisoma spp.(S. axillare (Steindachner, 1878) or S. frondosum (Agassiz, 1831), following MOURA et al., 2001)) were left at the genus level due to the subtle visual identification of small individuals in waters of low visibility.The collected material was identified, photographed, preserved and deposited at the Collection of Aquatic Biology, Department of Engenharia de Pesca, Universidade Federal do Ceará.
Occurrence and abundance patterns and species associations were studied through cluster analysis and PCA ordination.Abundance data was distributed on a species (rows) by sample (columns) data matrix.Species representing less than or equal to 10% of the hypothetical abundance of 44 individuals (one individual per sample) were discarded to reduce bias and contagion as other authors have suggested (BOESCH, 1977;MONTEIRO-NETO et al., 2003).Thirty-two species were included in the analysis.
The Hellinger's Distance was used as a similarity measure for both cluster and ordination analyses.This distance reduces the effects of null abundances inherent of biological surveys and better resolves the proportional similarities between less common species and abundant ones (LEGENDRE & GALLAGHER, 2001;LEGENDRE & LEGENDRE, 1998).The distance was calculated by  -Brazil).
Differences between species and sample groups defined in cluster analysis were statistically tested using a Monte Carlo permutation model.The Hellinger's distance between all group centroids was calculated using the objects (species or samples) within group mean abundances, and set as benchmarks.Ten thousand permutations on the original abundance values were performed, and the proportion of null centroid distances (p) higher than or equal to the correspondent benchmarks was calculated.Groups were considered significantly different from each other if p ≤ 0.05.This test was performed on Excel's add in Resampling (Resampling Stats 1999).
To facilitate cluster interpretation, Q-and R-mode classifications were related to each other using nodal analysis, which determines species-collections coincidences (BOESCH, 1977).Nodal analysis related dendrograms to each other, by calculating the frequency of occurrence and transformed average abundance of species within sample groups, providing a near tridimensional space to interpret species-habitat relationships (ROSSI-WONGTCHOWSKI & PAES, 1993).
Indicator Values (IndVal) were calculated for all 32 species, in order to identify indicator species for typologies of ecological samples obtained by Q-mode classification.Species presenting significant IndVal lower than or equal to 0.5 (p≤0.05) were considered as indicators of the sample groups.Calculations used sample groups derived from cluster analysis and followed DUFRENE & LEGENDRE (1997).All values were tested with 10,000 Monte Carlo permutations.IndVal calculations and statistical tests were performed on PCOrd for Windows, version 3.0 (MCCUNNE & MEFFORD, 1997).
The transformed data matrix was also submitted to Principal Component Analysis where both sample and species gradients were analyzed along the resulting Canonical Axes.Analyses were carried on CANOCO for Windows v.4.0 (TER BRAAK & SMILAUER, 1998).
Shannon-Wiener (H) and Simpson (D) diversity (SIMPSON, 1949) were compared among classification groups and tidepools using the averaged species abundances within classification groups or pools.Renyi´s graphic ordination method was used to check whether groups or pools were comparable (RENYI, 1961).If the resulting diversity patterns are parallel to each other, species diversity may be compared, whereas when lines cross each other, no comparisons are possible.Differences among average diversity were statistically tested with bootstrapping following the model proposed by SOLOW (1993).For all procedures the Species Diversity and Richness v.2.3 software was used (Henderson and Seaby, Pisces Conservation, ltd., 1998).

RESULTS
The physical descriptors of each of the examined tidepools are presented in table I.All pools had nearly the same area, ranging between 185 to 194 m 2 , but pool volume differed considerably, especially between pool 1 (deepest) and 3 (shallowest).With respect to substrate cover, Iparana pools tended to present a greater proportion of bare rock and rock covered with algae, whereas Pecém pools had a greater proportion of rock covered with algae (pool 3) and sand (pool 4).Pools 1 and 4 apparently showed a greater habitat complexity represented by a greater frequency of occurrence of crevices and holes as compared to the other pools examined (Tab.I).
Iparana pools 1 and 2 showed lowest fish abundance (1,480 and 1,356 respectively) as compared to Pecém pools 3 and 4 (2,912 and 3,179 fishes recorded respectively).On the average, fish densities were three times greater at Pecém than at Iparana.Pecém pools also attained the highest species richness with 34 and 29 species, respectively (Tab.II).
Q-mode cluster analysis identified two major sample groups showing significant locality (Pecém vs. Iparana) related differences at similarities lower than 1.5.For exploratory purposes, these groups were subdivided into two groups each, as shown in the nodal diagram (Fig. 2).Groups A and B were formed by samples from Pecém Pools 3 and 4, respectively.Group C included nearly all Iparana samples, whereas group D was composed by samples from June, August and September in Pool 3 and June in Pool 4, showing less evident temporal patterns (Fig. 2).Species Groups (R-mode cluster analysis) were not statistically different from each other (Tab.III).However, distinct dominance could be detected among the surveyed pools (Fig. 2).For descriptive purposes we refer to the four species Groups discriminated in table II and  figure 2 Group 2, composed by Eucinostomus spp., Engraulidae, Lutjanus jocu (Bloch & Schneider, 1801) and Gymnothorax vicinus (Castelnau, 1855) was observed mainly at Iparana Pools, with G. vicinus and Eucinostomus spp.being sporadically found at Pecém.The presence of these species at Iparana was more pronounced in June, August and September (samples from Group D).
Although intra-site differences regarding species occurrence were not significant, some discrepancies could  The distribution of rare species, which were not included in the previous analyses, followed a pattern similar to the observed in the nodal diagram (Fig. 2 Principal Component Analysis explained approximately 57% of the total variance among samples and species on the first two axes.The analysis revealed gradients similar to the ones found on previous analyses in this study.The first canonical axis explained 47.0% of the total variance and evidenced locality differences, with positive scores for sample groups C and D (Iparana) and negative scores for groups A and B (Pecém).
The distribution of species groups in the canonical space evidenced the main patterns of fish assemblage segregation among sampling localities.Species Group 2 showed positive loadings on the first axis, and was mostly representative of Iparana pools, with highest positive loadings attained by Eucinostomus spp., Engraulidae, L. apodus, G. vicinus and L. jocu.Group 3 also attained positive loadings, with two exceptions (Haemulon steindachneri and Acanthurus coeruleus), in spite of their slightly higher abundance in Pecém pools (Fig. 3).Groups 1 and 4 did not segregate in the first canonical axis, despite of the small negative trend observed in the loadings of Group 1.
The second canonical axis explained 10.5% of the variance and reflected mainly within locality differences with Pecém pools 3 and 4 (sample groups B and A) showing positive and negative scores, respectively (Fig. 3).
Differences between pools at Iparana were less evident, and resulted from abundance peaks of species with time restricted occurrences (e.g.Eucinostomus spp.).Such peaks accounted for the higher scores of sample group D, generating an apparent segregation, due to short scale temporal changes, probably related to life history aspects of these estuarine forms.Species that were exclusive from, or presented higher abundances, on Pool 3 (A.surinamensis, Ortopristes rubber (Cuvier, 1830), S. flaviventris, Chilomycterus antillarum) attained positive loadings on the second axis, while negative values were observed for the accessory members of Group 4. Almost all species from group 2 presented positive loadings on the second axis as well, with the exception of Gymnothorax vicinus, which occurred on all surveyed pools.In terms of species diversity, bootstrap analysis indicated that both Simpson and Shannon Wiener indices were significantly smaller for group D, when compared to all other sample groups (Tab.Va).All diversity indices for Groups A, B and C were not significantly different (Fig. 4a), with an exception for Groups A and B, which were not inter-comparable.Iparana tidepools 1 and 2 showed very similar Renyi´s ordination patterns, whereas Pecém pools 3 and 4 were displayed apart (Fig. 4b).Bootstrap analysis showed that both Simpson and Shannon Wiener diversity were not significantly different between pools 1 and 2, but both differed from pool 3, and they all differed from pool 4 (Tab.Vb).Pools 1 and 2 showed the lowest species diversity, whereas pool 3 the highest, and pool 4 intermediate diversity values.Axis II -13.3%

DISCUSSION
Our work showed that the taxonomic composition of fish species at Pecém and Iparana was consistent with several studies on tidepool fish fauna in the brazilian coast (FERREIRA et al., 1995;ROSA & MOURA, 1997;ROCHA et al., 1998;FERREIRA & CAVA, 2001), but differed considerably from ROSA et al. (1997) andBARREIROS et al. (2004).
Differences may be attributed not only to species geographical distribution and faunal affinities, but also to processes related to pool size (area and/or volume), height of the pool on the shore, wave action (GIBSON, 1982;GROSSMAN, 1982;PROCHAZKA & GRIFFITHS, 1992; (1997) andBARREIROS et al. (2004), were conducted with anaesthetics or ichthyocides on small sized rock pools (1.44-20.5 m 2 , 13-49 cm deep), often in the high intertidal.Such methods may be more reliable on small areas since they kill all the fish and provide information on cryptic species such as representatives of Bleniidae, Gobiidae, Labrisomidae and Pomacentridae.However, larger fish may not stay in small high intertidal pools due to dramatic variation in environmental conditions during the day (GRIFFITHS et al., 2003).
The prevalence of coastal reef associated families such as Haemulidae, Sparidae and Acanthuridae indicated that a substantial amount of the fish species pool from Ceará's tropical coast was incorporated into the tidepool fish assemblages (LIMA, 1969;LIMA & OLIVEIRA, 1978;ARAÚJO et al., 2000), as many opportunist or secondary resident species (GRIFFITHS, 2003a) moved into the pools at high tide and became trapped at low tide (GIBSON & YOSHIYAMA, 1999).For instance, HORN et al. (1999) pointed out that nearly all coastal fish species may, in one time or another during their life cycle, use intertidal habitats for different purposes, including the search for food and shelter from predators.
Monte Carlo permutation model, cluster and nodal analysis showed significant differences between sites (Iparana x Pecém), and suggested some degree of intrasite heterogeneity, especially at Pecém pools (although the differences were non-significant).This may reflect both larger environmental discontinuities and more localized features, such as pool isolation episodes (GIBSON & YOSHIYAMA, 1999), or environmental complexity (BENNET & GRIFFITHS, 1984;PROCHAZKA & GRIFFITHS, 1992) provided by substrate type and algae cover, both acting isolated or interactively.For instance, the relative importance of estuarine related species (Eucinostomus spp.and Engraulidae) for community composition was stronger for Iparana pools, which are located near the mouth of the Ceará River estuary, whereas tropical reef species, such as Stegastes fuscus (Cuvier, 1830), Serranus flaviventris, Chaetodon ocellatus, C. striatus, Myrichthys ocellatus, were found exclusively at Pecém.The environmental gradient provided by freshwater discharge, and subsequent changes in seawater chemistry and turbidity, could not only directly influence the environmental conditions at the pool areas, but also have a localized effect over the species source-pool, from which the tidepool community would draw its components (FUJI, 1988;METAXAS & SCHEIBLING, 1993;DAVIS, 2000).
These results were further supported by PCA with the segregation of Iparana and Pecém along the first component, and Pecém pools 3 and 4 on the second.Iparana related groups 1 and 2 (not significantly different) had no Indicator species but the estuarine Eucinostomus spp.This further supported the hypothesis of a greater contribution from the Ceará river estuary to the fish composition of tidepool assemblages at the neighboring Iparana reef.However, Eucinostomus lefroyi was also an important component at Pecém and Bagre marinus, another estuarine related species, was found exclusively at that site.Mugil curema also occurred sporadically at both sites.Therefore, estuarine forms would thus be an important component of tidepools at both surveyed areas, but taxon might vary according to species turnover and local sporadic availability of some species.
On the other hand, the non-significant intra-site differences suggest that the fish access to the pools may be, to some extent, also relayed on a stochastic process (GIBSON & YOSHIYAMA, 1999).Factors such as algae or benthic cover and reef complexity may be more important for cryptic and site-attached species (BENNET & GRIFFITHS, 1984;PROCHAZKA & GRIFFITHS, 1992;SILBERSCHNEIDER & BOOTH, 2001;GRIFFITHS, 2003b), a portion of the fish community that was not properly assessed by the methodology employed here (ENGLISH et al., 1994;GIBSON, 1999).Furthermore, the high proportion of transient species recorded reinforces the idea that there are few, if none, habitat choices during the fish arrival into the pools at high tides.Site-attached fish may stay longer if the habitat conditions, or environmental complexity, are suitable for colonization, and could provide the nondetected intra-site differences.The overall environmental complexity, evaluated based on substrate complexity and algae cover, differed among the surveyed pools.However, despite the significant inter-site differences in species diversity, pools at both localities presented high substrate complexity (Iparana pool 1) and algae coverage (Pecém pool 3 and Iparana pool 2).This reinforces the idea that a proper assessment of the whole fish community should be made with more than one sampling method, in order to determine the true importance of cryptic forms to local diversity and community structure.
The flooding of tidepools during high tides, which provides the access to fish to this ecosystem, is dependent upon factors such as tidal amplitude, local reef topography and distance from shore line (METAXAS & SCHEIBLING, 1993;MAHON & MAHON, 1994;DAVIS, 2000).Our results showed that higher diversity and abundance was attained at Pecém, where pools were shallower and closer to shore.These results contrasted with the findings of GIBSON (1982), PROCHAZKA & GRIFFITHS (1992) and MAHON & MAHON (1994) who suggested that larger pools in the low intertidal might be richer and have a higher proportion of transient species than small high intertidal pools (GIBSON & YOSHIYAMA, 1999).UNDERWOOD & CHAPMAN (1996) observed that pool location in relation to tidal movement rather than pool size may be more important for determining community structure and biodiversity.GRIFFITHS et al. (2003) observed that the vertical elevation of rockpools in the intertidal zone influenced the fish assemblage structure of individual rockpools.
Seasonal trends were not detected in our results.This was expected since seasonal environmental changes in tropical areas are more subtle than in temperate waters.BARREIROS et al. (2004) found consistent seasonal trends in tidepool ichthyofauna at higher latitude sites in Brazil.Nevertheless, we have not looked into length frequency distributions of fishes.This could show that individual species population structure did not remain constant and that different life history stages used the tidepools throughout the year.In fact, recruits of Haemulidae occurred at certain times of the year (personal observation).Future studies should address this aspect and further relate patterns with local environmental seasonality.
Abundance estimates through visual census may be biased due to lower counts of cryptic species (ENGLISH et al., 1994;GIBSON, 1999).Nevertheless, the use of anaesthetics or ichthyocides may not be appropriate in larger pools due the high volume and cost of solution required to treat the pool, and the possible impacts on adjacent areas due to dispersion of the compounds.In such cases, visual census with snorkeling may be a more appropriate method.MORING (1993) and GIBSON & YOSHIYAMA (1999) observed that visual census surveys provided a reliable account of the fish community on tidepools, comprised mostly of young partial residents using these habitats as nurseries.Furthermore, the use of visual census may be very appropriate on easily accessible and highly impacted areas.These are non destructive, low cost, easy to conduct methods that may be replicated several times, at different locations for comparisons, providing an advantage for monitoring areas in which tourism and fisheries have already reduced fish populations (CHRISTENSEN & WINTERBOTTOM, 1981;ENGLISH et al., 1994;FERREIRA et al., 1995;WILLIS, 2001).
Although we could not relate these disparities to any of the previously mentioned factors, we believe that human impact could also be an important component determining the variability of fish assemblages in these localities.Intertidal fishes occupy a variety of microhabitats within the reef structure including tidepool waters, patches of seaweed and rock crevices (HORN et al., 1999).As much as reefs are susceptible to human access, intertidal fishes are quite vulnerable to overexploitation and habitat degradation due to pollution and people trampling over the reefs (BRYANT et al., 1998;HORN et al., 1999).On weekdays, artisanal fishers and fish collectors visit the reefs harvesting species from tidepools.We estimated an average of 25 people on the reefs on our monthly visits, either fishing or collecting bait, ornamental fishes or live rock.On weekends, at low tide, tourists and bathers use tidepools for swimming and snorkeling, doubling the number of people on the reef (personal observations).On the other hand, human influence was lower at the locality of Pecém, at the time of our study.We found average daily frequencies of five people visiting the reefs during our study and, other than fishers and occasional swimmers, no other form of exploitation was recorded.The lowest diversity at Iparana tidepools may be a response to the continuous exploitation of intertidal fishes and the reef fauna there.In fact, SYMS & JONES (2000) found that persistent habitat disturbance may cause irreversible changes in fish communities, leading to the extinction of marine populations, including fishes (DULVY et al., 2003).Also, GOMES et al. (1995) stated that any changes in the community structure might first be detected through the patterns of abundance of the rare and "satellite" species (sensu HANSKI, 1982).In this survey, both sites shared the same ubiquitous (Group 1), or better adapted species, which comprised the core of Iparana tidepools fish community.The lack of many species found at Pecém pools might be a reflection of a medium to long term impoverishment process, related not only to direct human impacts over the reef area, but also to indirect effects on the Ceará River, and the consequent loss of water quality at nearby coastal areas.
Another example of human impact on tidepool fish communities was the construction of an offshore port facility at Pecém, which started in 1996.The abrupt changes in the coastline altered the local dynamics of sediment transport, which resulted in beach accretion and sand deposition over the reefs, totally filling the tidepools in 2002.Reefs were not exposed since then (personal observation).Our results also showed that it also contributed to a subtle loss of a quasi pristine environment, as well as a moderately diverse fish community.Some consequences of these changes are predictable up to a point.As fewer habitats are available to tidepool fishes, we suggest that fish biodiversity may be reduced within certain limits.However, we may not predict, at this stage, what will be the consequences for the recruitment of certain species, or the maintenance of local populations of fishes.Since tidepool ichthyofauna may vary between and within reef sites, conclusions taken from one area may not be generalized to other adjacent sites, and damages caused to one area may not necessarily be mitigated by the maintenance and protection of another area.

Fig. 2 .
Fig. 2. Nodal diagram showing species and sample groups and relative mean abundance and frequency of occurrence of species per sample groups at Iparana and Pecém beaches, State of Ceará, Brazil, from June 1998 to June 2000.

Fig. 3 .
Fig. 3. Result of PCA analysis showing the distribution of tidepool samples (symbols) and species (anagrams and arrows) within the first two canonical axes at Iparana and Pecém beaches, State of Ceará, Brazil, from June 1998 to June 2000.
Fig. 1.Map of Brazil within South America, the State of Ceará within Brazil , and both studied localities within the coastal zone of the state, showing their distance relation between each other and among the state Capital Fortaleza.

Table I .
Descriptive physical characteristics of tidepools at Iparana and Pecém beaches, State of Ceará, Brazil, from June 1998 to June 2000.

Table II .
List of family/taxon and their respective codes and numerical abundance per tidepool, location and pooled total numerical (N) and percent (%) abundance.Bottom shows total abundance and total number of species at Iparana and Pecém beaches, State of Ceará, Brazil, from June 1998 to June 2000.Serranus flaviventris).Indicator species at Iparana groups were Lutjanus jocu, Eucinostomus spp.and the unindentified Engraulidae (Tab.IV).

Table III
. Results of Monte Carlo multiple comparisons between sample groups A-D (top) and species groups 1-4 (bottom) at Iparana and Pecém beaches, State of Ceará, Brazil, from June 1998 to June 2000 (**, highly significant; *, significant; ns, non significant; NC, non-comparable).

Table IV .
Indicator species values (INDVAL) for each taxonomic group within sample groups, and averages (Avg) among groups with respective probaility values (P; **, highly significant; *, significant; ns, non significant) Highlited cells show highest INDVAL values for each taxonomic group at Iparana and Pecém beaches, State of Ceará, Brazil, from June 1998 to June 2000.