Effects of human trampling on a rocky shore fauna on the Sao Paulo coast , southeastern Brazil

Increased tourist activity in coastal regions demands management strategies to reduce impacts on rocky shores. The highly populated coastal areas in southeastern Brazil are an example of degradation caused by development of industry and tourism. Among different shore impacts, trampling has been intensively studied, and may represent a significant source of stress for intertidal fauna. A randomised blocks design was applied to experimentally study the effects of two different trampling intensities on richness, diversity, density and biomass of the rocky shore fauna of Obuseiro beach, Guarujá, southeastern Brazil. Blocks were distributed in two portions of the intertidal zone, dominated respectively by Chthamalus bisinuatus (Cirripedia) and Isognomon bicolor (Bivalvia). Blocks were trampled over three months, simulating the vacation period in Brazil and were monitored for the following nine months. Results indicate that Chthamalus bisinuatus is vulnerable to trampling impacts. Richness, diversity and turn-over index tended to be higher in trampled plots four months after trampling ceased. In general, results agree with previous trampling studies, suggesting that even low intensities of trampling may cause some impact on intertidal communities. Management strategies should include isolation of sensitive areas, construction of boardwalks, visitor education and monitoring programmes. In Brazil, additional data obtained from experimental studies are necessary in order to achieve a better understanding of trampling impacts on rocky shore communities.

Palavras-chave: costão rochoso, comunidades intertidais, pisoteio, impactos turísticos.located on the central coast of São Paulo state, in southeastern Brazil.Obuseiro beach is situated within the limits of a military area ("1 o Batalhão de Artilharia Anti-Aérea do Forte dos Andradas", Brazilian Army), and the access to the beach is granted only to some local fishermen.
The study site consists of a gently sloping large granite rock, dominated by Chthamalus bisinuatus on the upper-shore and Isognomon bicolor on the mid-shore.

Experimental design
Ten randomised blocks were equally distributed in two portions of the intertidal zone, dominated respectively by Chthamalus bisinuatus and Isognomon bicolor.Both species were selected for this study since they are conspicuous, perennial and widely distributed on rocky shores in southeastern Brazil.I. bicolor is an exotic species that has been recorded on Brazilian rocky shores for the last ten years, especially in the zone originally occupied by the mussel Perna perna (Linnaeus) that has been over-exploited as a food resource in many areas (Martins, 2000).
Blocks were trampled in a single day over three months, simulating the summer vacation period in Brazil, and were monitored for the following nine months.Data collection was always performed before trampling, so data from the first month relates to community conditions before trampling impacts.
Each block consisted of three experimental units (EU), 4,900 cm 2 each, separated by a 20 cm stripe.In each block, different trampling intensities were applied in the EUs, as follows: treatment 0 (no trampling, control units); treatment 1 (50 steps per month); treatment 2 (250 steps per month).Since the steps were applied randomly in a 4,900 cm 2 area and foot size was 200 cm 2 , it can be considered that on average each portion of EUs was trampled two times in treatment 1 and ten times in treatment 2. Trampling was performed by a 56 kg person using rubber-soled shoes.
An area of 900 cm 2 within the EU was selected for data collection.The position of this area and the treatment applied in each EU was randomly determined.The remaining portion of the EU in the I. bicolor assemblage blocks was used to perform total removals of individuals present in 100 cm 2 , every 3 months during one year.
These samples were washed in a 0.5 mm mesh (GRANUTEST) in order to select macrofaunal organisms only.All organisms were counted after identification to species level, with the exception of Ophiuroidea, Turbellaria, Nemertea and Polychaeta which were identified to higher taxonomic levels.
Different variables were analysed, according to assemblage characteristics.Variables and their collection methods are presented as follows: i) Cover of C. bisinuatus: cover estimates were monthly obtained through photographs taken from nine squares of 100 cm 2 for each EU.Photographs were scanned and C. bisinuatus

Introduction
Rocky shores are subject to increasing anthropogenic impacts originating from both the land and the sea.These habitats have been, and are currently, affected by pollution, over-collection of living resources, introduction of alien species, modification of coastal processes and global changes (Thompson et al., 2002).
Increased tourist activity may also represent a significant source of impact to rocky shore communities.The effects of human trampling on intertidal fauna have been intensively studied, especially in temperate zones (Povey and Keough, 1991;Brosnan and Crumrine, 1994;Fletcher and Frid, 1996;Keough and Quinn, 1998;Schiel and Taylor, 1999).Most studies have observed a reduction in abundance of some taxa, such as polychaetes (Brown and Taylor, 1999), bivalves (Beauchamp and Gowing, 1982) and barnacles (Brosnan and Crumrine, 1994) that may affect population dynamics and diversity of the entire community.
Different species of non-crustose algae were also vulnerable to trampling, showing reduction in values of abundance, cover and biomass (Povey and Keough, 1991;Fletcher and Frid, 1996;Keough and Quinn, 1998;Schiel and Taylor, 1999).Some groups, such as limpets and crustose algae, indirectly benefited from trampling, usually occupying empty areas created after the displacement of conspicuous species (Keough and Quinn, 1998).
Few studies have been developed on barnacle and bivalve assemblages, even though organisms in these groups are considered vulnerable to trampling, since they are small, sessile and easily squashed (Povey and Keough, 1991).Previous research has mostly focused on large conspicuous organisms, even though smaller cryptic animals also deserve attention due to their great abundance, high productivity and importance as food for higher trophic levels (Brown and Taylor, 1999).
The presence of beaches and other natural resources on practically the whole Brazilian coast attracts large number of tourists, especially at vacation periods and close to major cities, such as Sao Paulo.Guaruja municipality, for instance, has a population close to 230,000 people and receives more than 150,000 tourists during the summer season (IBGE, 2002).
The highly populated coastal areas in Brazil are an example of degradation caused by development of industry and tourism with several environmental impacts on coastal communities.However, the effects of human trampling on Brazilian rocky shores have not been previously examined.Therefore, this study aimed to investigate the effects of trampling on communities associated to Chthamalus bisinuatus Pilsbry (Cirripedia) and Isognomon bicolor (Adams) (Bivalvia) from a rocky shore in southeastern Brazil.

Study site
The study was conducted at Obuseiro beach in the city of Guarujá (24° 02' 19.0" S and 46° 17' 16.2" W), was applied in its exponential form (2 H' -Shannon apparent richness) which represents a good alternative to H' and allows direct comparisons to absolute values of richness.v) Turnover index of the community associated with I. bicolor: this index represents the sum of community losses and gains over time (Giordano, 2001) and it is calculated with the following formula: where: L = number of taxons present in T n and absent in T n+1 (lost taxons) G = number of taxons absent in T n and present in T n+1 (gained taxons) S Tn = number of taxons in T n time S Tn+1 = number of taxons in T n+1 time As the sum of two fractions, the turnover index can vary from zero to two, where zero represents cover percentage was measured in a central area of 25 cm 2 , using the software ArcView GIS 3.2 (ESRI, 1999).Therefore, for each EU, C. bisinuatus cover was estimated from a total area of 225 cm 2 (nine squares of 25 cm 2 ).ii) Cover of I. bicolor: cover estimates were obtained monthly through the division of the 900 cm 2 area in 225 small squares of 4 cm 2 .Each square was classified from zero to two (zero representing none to 35% cover of I. bicolor, one representing 36 to 70% cover and 2 representing more than 70% cover of I. bicolor).The sum of these classes (that could vary from 0 to 450) was then transformed into a percentage value that indicated the cover percentage of I. bicolor in each EU.iii) Richness of the community associated with I. bicolor: richness was calculated for each sample (100 cm 2 ), based on the number of taxons (mostly species) present in the sample.iv) Diversity of the community associated with I. bicolor: Shannon-Wiener diversity index (H') Recovery to original values of differences among treatments (Kruskal-Wallis test; february; p = 0.0493) was not observed even seven months after trampling ceased (Kruskal-Wallis test; November; p = 0.004).Repeated measures analysis also indicated that trampling promoted different patterns among treatments (Table 2), but the Tukey HSD test indicated significant differences only between treatments 0 and 2 (Tukey test; p = 0.0003).
Cover values of I. bicolor were usually high, ranging from 75 to 90% (Figure 2) and were not affected by trampling.However, mean values of percentages decreased with time in all treatments.
Mean richness values of the community associated to I. bicolor ranged from 13 to 18 taxa (Figure 3), while diversity values ranged from 4.5 to 5.8 (Figure 4).Both variables showed similar patterns four months after trampling ceased, with mean values for treatments 1 and 2 significantly higher when compared to original values (Table 3).
For all treatments, the turnover index values declined with time, changing from an initial pattern of gaining species (turnover positive values) to a pattern of losing species (turnover negative values).It is worth noting that while turnover values for treatment 0 were gradually reduced, treatments 1 and 2 values rose from May to August (Figure 5).
Density values varied widely among taxa, ranging from 0.2 to 413 indiviuals/100 cm 2 .Highest values of the persistence of the same taxa in the community, while two represents the complete substitution of community taxa over time.To better express the index descriptive value, a negative sign was attributed to the numeric value when the relative loss (L/S Tn ) surpassed the relative gain (G/S Tn+1 ).vi) Density of the macrofauna associated with I. bicolor: taxon density (number of individuals of the taxon/100 cm 2 ) was calculated for each collected sample.

Analyses
Considering the small numbers of replicates for each assemblage (n = 5), non-parametric tests were adopted in data analysis.Kruskal-Wallis and Mann-Whitney tests were used in order to identify significant differences between treatments.
Cover data from C. bisinuatus and I. bicolor followed repeated measures design and were analysed by a method proposed by Gurevitch and Chester Jr. (1986).The remaining variables were taken from independent samples collected every three months, and therefore could not be classified as repeated measures data.
Density values were analysed for each taxon separately, but only results that indicate some pattern are presented herein.

Results
Twenty-seven animal taxa were recorded in the studied assemblages (Table 1).Mean percentage cover of C. bisinuatus ranged from 15 to 50%.Trampling effects in C. bisinuatus cover were observed after the third month of trampling, especially for treatment 2 (Figure 1).density were presented by I. bicolor, indicating once again its dominance in the assemblage.
Trampling did not affect density values, that varied differently among taxa, but not among treatments.In general, three different patterns were observed: i) density reduction over time, presented by I. bicolor, Brachidontes solisianus, Lasea adansoni and Hyale nigra; ii) density increase over time, presented by Collisella subrugosa; and iii) constant values of density over time, presented by Syllidae polychaetes (Figure 6).

Discussion and Conclusions
Several studies indicated that invertebrate fauna associated with mussel communities may harbour from 32 to 99 species (Seed, 1996).However, in these studies all taxa were identified to species level, which may be responsible for the higher values found when compared to the I. bicolor community in this study (27 taxa).The structure of the bed may also interfere in these values, since mussel species usually form beds with several layers of individuals (Seed, 1996), increasing attachment and colonization surface for associated species while the I. bicolor bed is constituted by a single layer.
Trampling impacts on C. bisinuatus cover may be related to low resistance of individuals to direct effects of trampling.Chthamalus species were also affected by trampling on rocky shores in the United States, Australia and United Kingdom (Povey and Keough, 1991;Brosnan and Crumrine, 1994;Pinn and Rodgers, 2005), suggesting that morphological, physiological or reproductive aspects of the group may be responsible for its high vulnerability to trampling.Moreover, low cover values and indirect effects, such as limpets pasture could have also prevented C. bisinuatus recovery after trampling.
Resistance of I. bicolor to trampling may be explained by its high values of density and also by aspects, such as accommodation and lateral movement, that correspond to rapid colonization of small patches by adult individuals (Connell and Keough, 1985).
Increase in richness and diversity values after trampling have already been observed in previous studies    It is well established that trampling may indirectly increase abundances of a range of common herbivores, as a result of cover reduction of dominant species (Povey and Keough, 1991;Keough and Quinn, 1998).In the present study, the decrease in I. bicolor cover percentage over time may be responsible for the increase in C. subrugosa densities, even though none of these processes were related to trampling impacts.During grazing, herbivores usually dislodge barnacle young recruits, reducing their especially for intermediate levels of trampling (Liddle, 1975;Beauchamp and Gowing, 1982).
This pattern follows the disturbance hypothesis proposed by Connell (1978), in which intermediate level perturbations may interfere in the competitive exclusion process, changing community structure and favouring the occurrence of new stages on patch succession.This process allows the colonization of species rare or absent from well-established communities, consequently enhancing richness and diversity values.The analysis of the turnover index corroborates this hypothesis, by indi- survival rates and consequently affecting barnacle density values (Dayton, 1971;Safriel et al., 1994).Recolonization of bare spaces in barnacle assemblages depended mainly on planktonical larvae recruitment, in contrast to mussels and algae species that present lateral movements or vegetative growth.This may amplify the impacts of herbivore grazing on barnacle populations.
In this study, although density increases of C. subrugosa were not related to trampling, higher levels of mollusk grazing may have been an important factor preventing recovery of C. bisinuatus populations after trampling.
Other patterns, such as those observed for H. nigra, L. adansoni and B. solisianus were probably related to the natural decrease of I. bicolor cover and density, since these species are strongly associated to the I. bicolor bed.
In general, results agree with previous trampling studies, suggesting that even low intensities of trampling may cause some impact on intertidal communities and recovery may take several years.In this study, effects of human trampling were mainly observed in C. bisinuatus population, that occur in the upper portion of the intertidal zone, remaining exposed to trampling during several hours.Moreover, C. bisinuatus recovery was not observed until the next vacation period (nine months after trampling ceased), which suggests that impacts from several years may be accumulative in heavily trampled rocky shores.
Management strategies should include isolation of sensitive areas, construction of boardwalks, visitor education and monitoring programmes.In Brazil, additional data obtained from experimental studies are necessary in order to achieve a better understanding of trampling impacts on rocky shore communities, especially in the highly populated southeastern Brazilian shores.

Figure 2 .
Figure 2. Mean cover percentage of I. bicolor over time at all trampling intensities observed at Obuseiro rocky shore, São Paulo, Brazil.

Figure 3 .
Figure 3. Mean values of richness of the community associated with I. bicolor over time at all trampling intensities observed at Obuseiro rocky shore, São Paulo, Brazil.

Figure 4 .
Figure 4. Mean values of diversity of the community associated with I. bicolor over time at all trampling intensities observed at Obuseiro rocky shore, São Paulo, Brazil.

Figure 5 .
Figure 5. Mean turnover values of the community associated with I. bicolor over time at all trampling intensities observed at Obuseiro rocky shore, São Paulo, Brazil.

Table 1 .
Invertebrate intertidal fauna from a rocky shore in Obuseiro beach, SP, Brazil.Common taxa occured in more than 40 samples, frequent taxa occurred in 20 to 39 samples, and rare taxa occurred in less than 20 samples (total: 60 samples).
Figure 1.Mean cover percentage of C. bisinuatus over time at all trampling intensities observed at Obuseiro rocky shore, São Paulo, Brazil.

Table 3 .
Mann-whitney test results for richness and diversity values of the community associated with I. bicolor (p < 0.05)