Spatial and temporal variation of natural toxicity in cnidarians , bryozoans and tunicates in Mediterranean caves *

Corals and sponges are the most studied groups of benthic invertebrates in marine chemical ecology. This is mainly because they are abundant in all seas (Jackson, 1977; Uriz et al., 1992). However, active secondary metabolites are also found in virtually all benthic invertebrates, including bryozoans (e.g. Walls et al., 1993; Shellenberger and Ross, 1998), tunicates (e.g. Costa et al., 1997; Vervoort et al., 1998; Koulman et al., 1999), echinoderms (e.g.

In this paper we studied the patterns of spatial and temporal variation of natural toxicity in benthic invertebrates of several Phyla.We analysed the natural toxicity of the most abundant species of cnidarians, bryozoans and tunicates in distinct communities of two Mediterranean caves.Moreover, to assess seasonal variation of natural toxicity, analyses were done in June (spring) and November (autumn).

Study site
Sampling was performed by SCUBA diving along a horizontal transect in two Mediterranean caves in June and November 1996 and 1997.These caves are located in the Cabrera (Balearic Islands) and the Medes (Catalan littoral) Archipelagos respectively.In the Cabrera cave, three communities were sampled from the entrance to the inner part: a sciaphilic seaweed community (SSC) at the entrance, an external semi-dark cave community (ESC), and an internal semi-dark community (ISC).In the Medes cave, four communities were sampled: a hemisciaphilic seaweed community (HSM) located outside the cave, an external semi-dark cave community (ESM) at the entrance, a more internal semi-dark community (ISC), and a dark cave community (DCM) in the innermost part.The communities studied corresponded to zones 2, 3 and 4 in the Cabrera cave and zones 1, 2, 3 and 4 in the Medes cave; these zones have been described previously in (Martí et al., 2004a).
Whenever available biomass allowed, we collected replicates for each species.For small cnidarians, such as Polycyathus muellerae (Abel, 1959), Leptopsammia pruvoti (Lacaze-Duthiers, 1897), and Parazoanthus axinellae (Schmidt, 1862), replicates consisted of groups of adjacent polyps taken from different patches at least 1 m apart.For bryozoans and colonial tunicates, as each colony was clearly delimited, replicates were distinct individuals or colonies of the same species located at least 1 m apart.Lissoclinum perforatum (Giard, 1871), was the exception because of the small size of the colonies, so that the replicates consisted of 13 spatially close colonies that were then pooled.

Chemical extraction
In the laboratory the samples were cleaned of epibionts and separated from the substratum and any foreign material.They were then frozen, lyophilised and stored for posterior chemical extraction.A known weight of each lyophilised sample was extracted three successive times in 10 ml of methanol for 5, 10 and 15 minutes respectively, using an ultrasonic bath.The solvent from these extractions was filtered, pooled and evaporated under reduced pressure and nitrogen stream.The dry crude extracts were then weighed and kept at -20 Cº until dilution in artificial seawater for the Microtox ® assay.
To compare the results of the different samples, we used the gamma value (toxicity) of the concentration corresponding to a 1 mg sample DW ml -1 .The toxicity of this concentration was obtained by using the resulting regression equation of the Microtox ® assay (see Martí et al., 2003).By referring the results to the same concentration relative to sample DW, we could compare all the samples even when their DWs differed and the proportions of crude extract DW -1 varied in the different species.
We selected a threshold to discern between toxic and non-toxic species by comparing two distinct toxicity analyses: the Microtox ® and sea-urchin test.All species assayed from different Phyla showed activity against sea-urchin embryos at 0.5 gamma units of Microtox ® assay, therefore we selected this value as our threshold (see Martí et al., 2004b, for a more detailed explanation).

Statistical analyses
Seasonal changes in toxicity were statistically analysed for each species.We performed t-tests to find differences between seasons for the species present in only one community.Species in more than one community in a single cave were analysed by two-way ANOVA with season and community as the factors.Unless otherwise stated, Tukey post-hoc tests were used for a posteriori comparisons.In all cases, when data did not meet the assumptions of normality (Kolmogorov-Smirnov test) and homoscedasticity (Bartlett test), rank transformation was carried out and parametric analyses were performed on ranked data (Conover and Iman, 1981;Potvin and Roff, 1993).All the analyses were done with the Systat v 5.0 and Statistica v 4.0 packages.

The Cabrera cave
We analysed 4 cnidarian species from SSC, which accounted for 83% and 86% of the cnidarian coverage in June and November respectively; 4 species from ESC, which accounted for 100% of coverage in the two seasons; and 4 species from ISC, which represented 83% and 99% of coverage in June and November respectively (Table 1).
Specimens of Clavularia crassa (Milne-Edwards, 1848), and Eudendrium sp. were only analysed in November.Both species were toxic (Table 2).Madracis pharensis (Heller, 1868), was never toxic.The behaviour of P. muellerae differed in each community and season: its toxicity increased from June to November in SSC, decreased during the same period in ESC and it was not toxic in ISC.
For species with replicate samples, L. pruvoti was only toxic in ISC, where it showed decreased toxicity from June to November (Fig. 1).The results from the ANOVA showed that community and season had significant effects on toxicity (Table 2).Post-hoc analysis for differences among communities indicated that SSC and ESC were similar (p > 0.05), but they differed from ISC, which was more toxic (p < 0.05 and p < 0.005 respectively).The season was also significant (p < 0.05): toxicity was higher in June than in November in all communities.
P. axinellae was the most toxic cnidarian (Fig. 1).This species was significantly more toxic in ISC than in ESC (Table 2).It is interesting to note that the standard error of the mean toxicity for all the cnidarians VARIATION OF NATURAL TOXICITY IN INVERTEBRATES 487 analysed was notably smaller than that observed for algae (Martí et al., 2004b), and sponges (Martí, 2002), under the same environmental conditions, which indicates a lower genotypic variability.

The Medes cave
We analysed Corallium rubrum (Linnaeus, 1758), and L. pruvoti in the Medes cave (Table 1).In ESM, these two species accounted for 94% of the total cnidarian coverage in June and 96% in November.For ISM, the percentages were also high: 80% in June and 93% in November.Only L. pruvoti was analysed in DCM (where it had low coverage).C. rubrum was never toxic.L. pruvoti was toxic only in June in ISM and DCM.Differences among communities and between seasons (Fig. 1) were lower than those detected in the Cabrera cave.Although species tended to show higher toxicity in June, statistical differences were not detected between seasons (Table 2).

The Cabrera cave
In SSC, two species accounted for 69% and 62% of the bryozoan coverage in June and November respectively.Five species from ESC, which accounted for 61% of coverage in June and 64% in November, were also analysed.Only Frondipora verrucosa (Lamouroux, 1821), was analysed from ISC, and it accounted for 12% and 15% of coverage.Most species experienced high seasonal changes in coverage (Martí et al., 2004a) and were detected only in one season (Table 3).
Schizomavella sp. and Smittina cervicornis (Pallas, 1766) were never toxic.Sertella sp. and Myriapora truncata (Pallas, 1766) were more toxic in November than in June.For the species for which we had replicates (Fig. 2), Scrupocellaria sp. did not show significant differences in toxicity between communities (p > 0.05), but this species was clearly toxic in June, and its toxicity was close to 0.5 gamma in November.F. verrucosa was always toxic but high variability prevented significant differences being detected either between communities or between seasons (Fig. 2 and Table 4).

The Medes cave
As most bryozoans were extraordinarily small encrusting or branching forms, it was difficult to obtain enough material for the toxicity analysis.Only 3 species were analysed in the two seasons (Table 3).
The species from ESM accounted for a low percentage of total bryozoans present in the communi-   ty (6.7%).Only the small species Crisia spp. was toxic in ESM in June.

Tunicates
The Cabrera cave Tunicates were scarce and very small in this cave.We only studied L. perforatum from ISC and found it to be extraordinarily toxic, with a mean toxicity of 212.2 gamma units in June, and in November all the test bacteria were killed at the concentrations tested.The toxicity of this species was one of the highest detected in this study, and is comparable to that of the highly toxic sponge Crambe crambe (Schmidt, 1862) (Becerro et al., 1997;Martí, 2002).

The Medes cave
There were sufficient quantities of two species to perform toxicity analyses in the Medes cave in both seasons (Table 5).Didemnum sp. was not toxic in any season.Toxicity of Cystodytes dellechiajei (Della Valle, 1877), did not vary significantly (Table 6) between communities and seasons (Fig. 3).

DISCUSSION
Most of the cnidarians analysed (71%) in this study were toxic in at least one cave, community or season.None of the species had epibionts (Uriz et al., 1991 and personal observation), therefore toxic-ity may be an antifouling strategy, although further experimental evidence is required.Previous studies have addressed toxicity variations in cnidarians.The secondary metabolites of some species can vary quantitatively and qualitatively, depending on the biogeographic zone (Harvell et al., 1993), while other species have similar compositions of these metabolites in different habitats (Puglisi et al., 2000).The toxic behaviour of some cnidarians in this study is comparable to that of other cnidarian 490 R. MARTÍ et al.  species.For instance, the toxicity of P. axinellae in the Cabrera cave varied significantly in the different communities: it was more toxic in ISC than in ESC.
If we consider that the innermost zones of caves are comparable to deep habitats (Vacelet et al., 1993), this toxicity pattern is in accordance with that of a gorgonian species (Harvell et al., 1993).This study reported a significantly higher concentration of secondary metabolites in gorgonian specimens transplanted to deep habitats than in those that remained in the shallow habitats.A total of 71% of the bryozoans were toxic in one of the communities, caves or seasons.There are few studies on the bioactivity of bryozoans (Martín and Uriz, 1993;Walls et al., 1991Walls et al., , 1993;;Shellenberger and Ross, 1998).In the Mediterranean Sea, bryozoans are one of the most active taxa along with sponges and tunicates (Uriz et al., 1991).Martín and Uriz (1993), found strong biocide and anti-settlement activities in the extracts of Myriopora truncata and Sertella beaniana, while anti-mitotic or cytotoxic activities were moderate.F. verrucosa showed high intra-specimen variation in toxicity, with some highly active specimens compared to other bryozoans and most species of the other Phyla.
The bryozoans analysed here were usually quite free of epibionts and the antifouling role is the most commonly reported function for chemical defence in this group.Walls et al. (1993), and Shellenberger and Ross (1998), reported a negative correlation between the presence of secondary metabolites, the antibacterial activity of the extracts and a reduction of fouling, which might indicate an antifouling function for secondary metabolites.These authors also hypothesised that secondary metabolites in bryozoans contribute to inhibiting surface microbial films, which in turn influences the type of organisms that settle on them.The antifouling function of secondary metabolites in bryozoans is also reported in other studies (Walls et al., 1991).There is only one analysis of the variation of natural compound production in bryozoans.The secondary metabolites called bryostatins, which strongly inhibit human cancer, are synthesized only by some populations of Bugula neritina (Linnaeus, 1758) (Pettit, 1991).We also detected seasonal and spatial variation in toxicity in some bryozoans.
Antitumour, antileukemic, cytotoxic, antiviral and immunosuppressive molecules have been isolated from tunicates (Costa et al., 1997;Koulman et al., 1999).Tunicates are one of the Phyla with a higher percentage of toxic species in the Mediterra-nean (Uriz et al., 1991).We detected a large interspecies variation in toxicity in the three species analysed: one was highly toxic (i.e.L. perforatum), another mildly toxic (i.e. C. dellechiajei) and a third was not toxic (i.e.Didemnum sp.).L. perforatum was the most active organism in this study.
We conclude that all the Phyla considered have some toxic representatives but that species toxicity does not show a common pattern between caves, communities and seasons.Toxicity varied with season and/or community in most cnidarians and bryozoans, whereas the tunicates remained toxic throughout the communities and seasons.
FIG. 1. -Mean toxicity and standard error of cnidarians in the Cabrera and Medes caves.The dashed line indicates the 0.5 gamma threshold between toxic and non-toxic species.The abbreviations of communities are SSC, ESC and ISC: sciaphilic seaweed community and external and internal semi-dark cave communities in the Cabrera cave respectively.ESM, ISM and DCM: external and internal semi-dark cave communities and dark cave community in the Medes cave respectively.
FIG.2.-Mean toxicity and standard error of bryozoans in the Cabrera cave.The dashed line indicates the 0.5 gamma value threshold between toxic and non-toxic species.The abbreviations of communities are ESC and ISC: external and internal semi-dark cave communities respectively.

TABLE 1 .
-Toxicity (gamma units) of cnidarians from the Cabrera and Medes caves in both seasons.SSC, ESC and ISC = sciaphilic seaweed community and external and internal semi-dark cave communities in the Cabrera cave respectively.ESM, ISM and DCM = external and internal semi-dark cave communities and dark cave community in the Medes cave respectively.
» indicates species for which replicates were obtained and thus toxicity values are means

TABLE 2 .
-Two-way ANOVA results on toxicity of cnidarians from the Cabrera and Medes caves.

TABLE 3 .
-Toxicity (gamma units) of bryozoans from the Cabrera and Medes caves in both seasons.SSC, ESC and ISC = sciaphilic seaweed community and external and internal semi-dark cave communities in the Cabrera cave respectively.ESM and DCM = external semi-dark cave community and dark cave community in the Medes cave respectively.

TABLE 4 .
-Two-way ANOVA results on toxicity of Frondipora verrucosa from the ESC and ISC in the Cabrera cave.

TABLE 6 .
-Two-way ANOVA results on toxicity of Cystodytes dellechiajei from HSM and ESM in the Medes cave.