Effects of Environmental Stress on Ascidian Populations in Algeciras Bay (Southern Spain). Possible Marine Bioindicators?

MARINE ECOLOGY PROGRESS SERIES Mar Ecol Prog Ser Published December 5 Effects of environmental stress on ascidian populations in Algeciras Bay (southern Spain). Possible marine bioindicators? S. A. Naranjo*, J. L. Carballo, J. C. Garcia-Gomez Laboratorio de Biologia Marina, Dpto. de Fisiologia y Biologia Animal, Facultad de Biologia, Apdo 1095, E-41080 Sevilla, Spain ABSTRACT: The distribution and abundance of littoral ascidians were analyzed with respect to thelr possible relationships with environmental stress. As part of a multidisciplinary research project on the benth~c communities in Algeciras Bay, southern Spain, a suite of environmental variables was measured (hydrodynamism, silting, suspended solids and organic matter). After displaying the similalities of fauna through clustering and ordination of sampling sites, the relationships between community differences and changes in the abiotic component were established based on the BIO-ENV procedure and Canonical Correspondence Analysis. Hydrodynamism and the percentage of organic matter in the silt is the variable combination that best explains (Spearman correlation of 0.82) the biotic structure. While all ascidians show a certain tolerance to diverse environmental factors, some species such as Ciona intestinalis. Diplosoma spongiforme, Phallusia mammillata, Microcosmus squamiger, Styela plicata and Synoicum argus could be considered as indicators of areas which have been subject to intense stress (substrate transformation, water stagnation and sedimentation excess) over long periods of time, whereas others such as Aplidium conicurn, Aplidium punctum. Clavelina dellavallei, Halocynthia papillosa and Stolonica socialis, which live only in natural and non-perturbed rock areas, could be categorized as species very sensitive to stress, as well as ind~cators of good conditions. KEY WORDS: Ascidian communities . Environmental stress . hdultivariate community measurements B~oindicators INTRODUCTION There is now widespread recognition that chemical monitoring of pollution alone is not enough, and that pollution is essentially a biological phenomenon because of its impact on living organisms (Wright et al. 1994). The analysis of changes in benthic community structure has now become one of the mainstays in detecting and monitoring the biological effects of marine pollution (Warwick & Clarke 1993). Nevertheless, knowledge of the distribution patterns of benthic organisms along natural environmental gradients (e.g. from turbulent to calm condi.tions) is necessary before a possible anthropogenic disturbance can be proved. Algeciras Bay features important industrial developments (with chemical industries, refineries, thermal power plants, ironworks and paper mills) as well as intense harbour activity. All of this results in a highly transformed coastline (mainly in the central zone) where shipyards, piers and breakwaters, among other port constructions, affect the normal water flow. Some areas that are better conserved and have natural rocks exist in the outer zone. Recent studies show the amount of direct waste in this area to be close to 2800 m"', with much more urban sewage than industrial waste-waters (Wait et al. 1990). consequently, Algeciras Bay can be considered to be a patchwork of small areas under different levels of stress due to the influence of environmental and anthropogenic factors. Thus, a multitude of small environments can be identified and, correspondingly, a similar level of heterogeneity can be expected in the structure of the benthic communities within these environments. Algeciras Bay is considered to be an especially suitable area for biomonitoring studies (Nai-anjo & GarciaO Inter-Research 1996 Resale of full article not permitted Mar Ecol Prog Ser 144: 119-131, 1996 Gomez 1993, Carballo et al. 1994, Conradi & Cervera 1995, Carballo et al. 1996). In this sense, a multidisciplinary research program was conducted from 1991 to 1995 in order to gain knowledge of the macrobenthic fauna and its relation to major environmental variables in this area. Among all sessile organisms studied, the ascidians are a particularly interesting group in this respect, since they display both of the main ecological strategies-solitary and colonial-and a variety of morphological types which are of adaptive significance (Jackson 1977a, 1979) and which correlate with distinct abilities for colonizing new surfaces. Accordingly, it is a good diagnostic group whose d~stribution may reflect some of the prevailing structuring factors affecting benthic assemblages (Turon 1990) Ascidians are found on all submerged hard surfaces, including concrete, iron, rope and plastics. Some species are most abundant in highly transformed and pnll~~teri ~nvirnnrnentq 5~1r-h as ports, harhors and industnal areas, where the excess of particulate organic matter and the proliferation of bacteria are a source of food for filter-feeding invertebrates, particularly ascidia.ns (Monniot et al. 1991). Ascidians can also take up various metals from the water (vanadium is one of the most common) and some toxic substances (heavy metals and hydrocarbons). Therefore, ascidians contribute to the cleaning and purification of waste-waters. For this reason, numerous studies have outlined the importance of this group as pollution bioindicators (Papadopoulou et al. 1972. Papadopoulou & Kanias 1977, Monniot 1978, Bell et al. 1982). In spite of this, little is known about the influence of local environmental factors (such as water movement, suspended solids or silting) on the presence or absence of the species in a given habitat. In this sense, we have studied the composition and abundance of the ascidian communities in Algeciras Bay, as well as their level of adaptation to abiotic components. MATERIAL AND METHODS Ascidians were sampled at 11 coastal stations distributed around Algeciras Bay, at depths from 3 to 15 m. Their locations are shown in Fig 1. Due to th.e heterogeneity of the hard-bottom habitats (natural rock formations, port constructions, shipyards and other artificial substrates), the stations were selected on the basis of environmental variability. Inventories of species per station were taken of the specimens collected by SCUBA-divers along continuous transects (approximately 50 m long). These inventories were completed by visual and photographic sampling along the same transects, with a similar immersion time at Fig. 1 Location of the sampling stations in Algeciras Bay, southern Spain all stations (there were no replicates on any occasion). All ascidians sighted at a distance of 1 m on either side of the transects were collected and/or photographed. Surveys were taken from September 1992 to November 1993. Table 1 shotvs the species abundance in terms of semi-quantitative abundance codes. Coding can be useful when different sampling methods are utilized, and in this study it permitted more objective comparisons of the faunistic composition at different stations (Carballo et al. 1996). Furthermore, this procedure is presumed to have an effect sim~lar to standardization by columns, which minimizes data variability arising from sampling heterogeneity (Maldonado S1 Uriz 1995). The establ~shment of environmental variability among stations was based on the analysis of 5 abiotic variables that could be related to environmental stress. These were hydrodynamism, silting, suspended solids, suspended organic matter and organic matter in the silt (SOM) gathered in collecting bottles. The methods used to estimate these parameters are detailed in Carballo et al. (1996); measurement was made as in Gambi et al. (1989) for hydrodynamism, Moore (1972) for silling and Strickland & Pdrsons (1960) for the other three. As a preparatory step to the statistical analysis, these abiotic variables were log-transformed. Naranjo et al. Effects of environmental stress on ascidlans 121 - Comparisons among sampling stations (on the basis of species abundance) were conducted according to the strategy outlined by Field et al. (1982) for the analysis of data on community structure: (a) The biotic relationship between any 2 samples (sampling stations in our case) was distilled into a coefficient measuring similarity (or dissimilarity) in species composition. The Bray-Curtis lndex (Bray & Curtis 1957), which is not a function of joint absence, was chosen for this purpose. It is recognized as an efficient measure for the evaluation of affinities on semi-quantitative data from large faunistic assemblages (Gamito & Raffaelli 1992). (b) Stations were classified into groups (using the triangular matrix of similarity obtained between every pair of stations) either by hierarchical agglomerative clustering, wlth group-average linking (e.g. Sneath & Sokal 1973), or by mapping the station interrelationships into ordination using non-metric multidimensional scaling (MDS) (Kruskal & Wish 1978). (c) Relationships between species were analyzed by transposing the data matrix and repeating the classification and ordination methods on the new similarity matrix between every pair of species. Species which were indicative of the dissimilarity between groups of stations were determined as in Clarke (1993), using the SIMPER computer program. In order to link the multivariate community structure (reflected in the faunistic heterogeneity of the different sampling stations) to environmental variables (which also change with each station), we used 2 different, but complementary, statistical techniques: (1) BIO-ENV procedure (Clarke & Ains~vorth 1993). In summary, this consists of a separate compar~son of the among-station similarity matnx for the biota with equivalent triangular matrices for all combinations of abiotic variables, and the consequent choice of the subset of environmental variables which provides a good match between the 2 configurations. A match was measured by the Spearman rank correlation coefficient (p,), which was computed for the dissimilarity matrices of the biotic and abiotic data. Although no specific plot is provided showing the result, a diagram displaying the biotic MDS (based on the Bray-Curtis dissimilarity index) in conjunction with the abiotic MDSs (computed from Euclidean distances) for the subset of envlronnlental variables selected by the BIO-ENV procedure as optimal would be appropiate, especially when high values of p, are obtained (e.g. p, % 0.8) (Clarke & Ainsworth 1993). (2) Canonical Correspondence Analysis (CCA). This is a direct gradient technique, and represents a special case of multivariate regression (Palmer 1993). In the resulting ordination diagram, environmental variables can be represented by arrows along with the species and/or station scores. Details of how to interpret CCA diagrams are given in ter Braak (1986) and Palmer (1993). The software used was: PRIMER version 3.1 from Plymouth Marine Laboratory, UK, folcluster, MDS, SIMPER and BIO-ENV analysis; STATGRAPHICS 6.0 for regression analysis; and CANOCO 3.12 for Correspondence Analysis, CCA and Montecarlo permutation tests. RESULTS Station groupings based on species composition Table 1 indicates large variability in the nature, type and availability of hard substrates characterizing the sampling stations. A preliminary study of the local distribution of the most common ascidians indicated 3 basic groups: (a) species exclusively distributed in the outer zone of the bay which settle mainly on natural rock substrates; (b) species more abundant in the inner zone of the bay which settle mainly on artificial substrates; (c) species found throughout the bay. Analysis of the dendrogram in Fig. 2 shows the dlfferences among the stations as stated above. This classification technique separates the stations into 3 major groups (A, B, C) which are related, in part, to the nature of the substrate. In group A, the stations are characterized by the presence of extensive natural rock formations (in the extremes of the bay). Groups B and C include the rest of the stations (In the inner parts of the bay), where artificial substrates are dominant (walls and columns in ports, breakwaters, shipyards, etc.). The average dissimilarity between groups A and B 1s 59.74%. As is shown in Table 2, 17 of the 38 specles explain the dissimilarity of 70% between groups A and B. Pseudodistorna obscururn, Polycitor adriaticurn, Fig 2. Dendrogl-am of stations using group-average clusterlng from Bray-Curt~s similarity on specles abundances The 3 groups of stations [A to C) separated at a 50% similarity threshold (dotted line) are indicated 122 Mar Ecol Prog Ser 144 119-131, 1996 Table 1. Asctd~an species abundance In Algec~ras Bay and substrate character~st~cs at sampl~ng statlons Sem~quant~tat~ve abundance codes, 0 = species abhent, 1 = rare (1 or 2 speclmens recorded), 2 = frequent (from 3 to 10 specimens per ~rnmers~on) 3 = very common (an average of more than 10 speclmens per immersion) Subsirate onentation H = horizontal surfaces, V = vertical surfaces, 0 = overhangs U = under boulders; Cc = crevlces and cantles. E = ep~b~on Natural rock types a = large submerged boulders, b = stnall stones on sandy bottoms; c = rock slabs, d = biodetr~t~c bottoms, e = Caulerpa prolrfera weeds Artlfic~al substrate types a = piling of concrete blocks in breakwaters, b = vertlcal surfaces In port walls, c = plers in sh~pyarcis, d = others artlfic~al structures scattered on bottom Depth (m): 3 to 7 8 to 15 3 to 5 3 to 10 0 to 4 2 to 6 10 to 18 3 to 9 2 to 5 l to 6 4 to 10 Substrate Art~flc~al substrate: - - - b,c a,c a,c,d b,c.d a,d d a,c,d b,d onentatlon Natural rock a,b a,b a,b,c c c c,d c,d c c,e c,d a,c Stn 1 Stn 2 Stn 3 Stn 4 Stn 5 Stn 6 Stn 7 Stn 8 Stn 9 Stn 10 Stn 11 Apli d~ um conicum Aplidiurn elegans Apl~diurn pallidum Apl~d~um punctum Ascldia ment~lla Asadiella aspersa Ascidjella sca bra Botnllus leach1 Rotryllus schlossen Qon<~ edwards~ Ciona ~ntestlnalis Clavelina de!Javallej Clavel~na lepadiformis Clavel~na nana Didernn urn conaceurn Didemnum maculosun~ D~plosoma listenanun~ Diplosoma spong~iorme Dlstom~ls variolosus Halocynth~a pap~llosa Mic~ ocosm us squamiger Molgula bleize! Molgula occldentalrs Phallus~a fumigata Phallusla ingeria Phallusla mammillata Polyutor adrial~cum Polycitor cr~staljn Llm Polysyncraton lacazei Pseudodlstoma obscurum Pyura rnlcrocosrn us Rhopalaea neapol~tana Stoloiuca soclal~s Styela canopus Styela phcata Synoicum a]-gus Syno~curn blochmann~ Trld~demnurn cei-eum A4icrocosmus squamiger and Synoicum argus can be considered as good discr~.minators because they contnbute the highest values to the average disslmilanty (6) and because they have reasonably high values of the S,/SD(~,) ratio (Clarke 1993) The stations whlch make up group C (5 and 9) show the lowest levels of specific richness in the bay and have a faunistical composition typical of harbour areas (h/licrocosmus squamiger, Styela plicata, Synoicum argus, etc.); thus, they are more closely related to the stations in group B (40 %, similarity). MDS ordination was perfolmed using the smlaritles between species In the resulting 2-dimensional conflguratlon a def~nite species aggregation can be observed, but a direct intrepretation could be inaccurate Molgula bleizel and Ascidiella aspersa are abundant but exclusive to Stn 9, so their similarity wlth the other species is very low The removal of these outlier specles 1s recommended (Gauch 1982, Clarke & Green 1988) in order to obtain a more reliable picture of the remaining species Although stress tends to decrease as dlmenslonal~t~ is increased (Field et a1 1982), when MDS was performed Naranjo et al.: Effects of environmental stress on ascidlans 123 in 3 dimensions IFia. 3). the swecies' relaTable 2. Averaqe abundance (7) of important ascidian species In qroups A [l , ., r. tionships became more evideit, rn F,~, 3 it to 3, I l) and B (4, 6 to 8, 10) at.kllgeciras Bay stations. species areiisted in 0;- der of their contribution (8,) to average dissimilarity (8) (= 59.74) between is to identify of the 2 groups (only partly given) Species names given In full in Table 1 species based on the maior trends of distriwhich live on natural rock are on the right S. argus 2.8 0.25 4.15 1.38 3.01 side of the diaaram IA~lidium ~allidum, l? adriaticum 0.6 3.0 4.13 2.41 1.71 13.86 695 1 bution in the bay: species mainly distributed at the outer part of bay (Stns 1 to 3) - - Species Ys YA 6, SD(6,) 8, ,'SD(6,) E$,(%8) gata, Rhopalaea neapolitana, Synoicum I A. eleqans 2.0 1.25 2 14 1.73 123 39.41 1 d . . Stolonica socialis, Claveljna dellavallei, Tndidenlnum cereurn, Pseudodistoma cristallinum, Aplidium punctum, Phallusia furnlM. s4Uamiger 2.6 0.25 4.02 2.18 20.58 P obscururn 0.0 2.5 3 97 0.64 6.25 cristallnus 0,0 63 ,73 27.24 1 52 31.63 A, 1.2 2.5 251 1.90 132 35.84 conjcum. Ciona edwardsi, A~lidium eleI D. sponqiforme 1.8 1.0 186 1.10 170 59.48 1 blochmanhi, ~seudodistoma obscurum). Species which are considered to have a medium level of tolerance in relation to the substrate (Polycitor adriaticum, Aplidium S. b~~it~mannj 0.4 1.5 2.13 1.20 1.78 42.98 A. punct"m 0.2 2.06 1.95 1.06 46.44 C.dellavallei 0.0 1.5 2.02 2.11 0.96 49.82 C, lepadiform,s 2,2 1.99 1.46 1.37 53.15 P mammillata 1.4 0.5 1 92 1.52 1.26 56.36 natural rock, but they can also colonize , . gans, Didemnum n~aculosum, Clavelina nana, Botiyllus leachi, etc.) are grouped in the middle; they are more abundant over artificial substrates. These species are dis- . - cereum 0.2 1.25 173 1.61 1 08 62.38 D. variolosus 1.6 1.0 1.71 1.18 1.45 D, maculosum 65.24 1.58 1.17 1.35 67.90 D, l;sterjanum 1.4 0.75 1.55 1.11 1.40 70.49 tributed throughout practically all of the bay, except for influence of hydrological factors along its margins is unthe most internal areas (piers and shipyards). Finally, even. The 5 environmental variables recorded showed a species grouped on the left (Ciona intestinalis, Micronotable variability at both spatial and temporal levels cosmus squamiger, Synoicum argus, Styela plicata, (the latter is not taken into account in this work). Fig. 4 Phallusia mammillata and Clavelina lepadiformis) are (b to g) shows a multiple box and whisker plot for each typical species of harbour areas and colonize every type variable analyzed. In spite of some sampling stations beof artificial surface ing placed differently, the relationships among stations were similar to the ones in Carballo et al. (1996). Due to the proximity of the Straits of Gibraltar, the Relationship between species and environmental hydrodynamic conditions depend on major coastal variables currents and, in consequence, on tidal flows and prevailing winds. Extreme values of hydrodynamism were Abiotic characteristics of the stations registered at outer (and more exposed) stations (1 to 3 and 11). The high hydrodynamism recorded at Stn 8 Because of the many coastal transformations in Algecan be explained by the influence of 2 littoral currents ciras Bay (due mainly to industrial development), the which appear during low-tide periods (Camiiias 1987). Fig. 3. Three-dimensional MDS configuration of species slmilanty matnx (stress = 0.09). Values of Asc~diella aspersa and Molgula ble~zei were removed previous to analyses. Species names are abbreviated from Table 1 124 Mar Ecol Prog Ser 144: 119-131, 1996 l D,, = 4 Stat~on Station Fig 4 (a) Specif~c nchness (D,,,) of ascidians at the sampl~ng stations; (b to g) mult~ple box and wh~sker 'O plot for each abiotic factor registered at the 11 sampling stations. (d) as (c] but without Stn 5 Stat~on Station In general, hydrodynamism is inversely related to silting, but at Stn 5 these parameters seem occasionally parallel (e.g. during the rainy season). The unusual conditions of this area are due to the influence of the Palmones River (sedimentation is increased by the river's flow) Thus, both the extreme and the highest average annual values for silting (195.6 and 74.04 g m-2 mo-', respectively) and for SOM (13.2 and 7 g m-2 mo-', respectively) were registered here. On the other hand, suspended solids and suspended organic matter reached the highest values at Stn 9, where water renewal is limited by a partially enclosed harbour zone. Hydrodynamism seems to influence the other factors. Several environmental relationships can be established among the set of vanables measured (and logtransformed) by using correlation analysis. Whereas hydrodynamism and suspended solids show a negative correlation (r = -0.72; p = 0.018), a positive linear relationship exists between silting and suspended solids (r = 0.54; p < 0.1), as well as between both variables and their organic matter contents (absolute values only, see Fig. 5). Univariate and multivariate community measures After the variation levels of major hydrological factors were recorded, different statistical methods were carried out in order to establish a relationship between abiotic heterogeneity and fauna1 distribution (from ascidian species abundance) throughout the survey area. As a first approximation, the biotic information relative to each sampling station was summanzed by simple univariate measures such as the number of species found (S) or the specific richness. For this purpose, specific richness has been established from the Margalef index: D*,, = S - l/ln(N), where N is the total number of specimens collected (e.g. Bakus 1990). Although S and DM, are usually correlated (for this case: r = 0.67; p < 0.05), specific richness gives a quantitative measure of diversity avoiding some problems inherent to other indexes such as Shannon's or Simpson's (Magurran 1989). Fig 5 Llnear correlations between the organic matter (0 M ) content (total and percentage) with silting and suspended solids + positive, and - negatlve correlations 'p < 0 1, '"p < 0 001 Naranjo et al.: Effects of environmental stress on ascidians 125 - Fig. 4a shows the specific richness throughout the bay. The low values for this parameter (DblEI < 3.1) obtained at Stns 5, 6 and 9 are remarkable, as well as the highest specific richness (Db1, > 7) at Stn 2. This biotic information provided no significant relationship with the set of environmental variables considered here; however, a definite positive relationship between specific richness and hydrodynamism and an inverse one with respect to suspended solids is suggested (r = 0.53 and r = -0.51, respectively; p < 0.1). The biological data are best represented using a multivariate summary, such as an MDS ordination (Clarke & Warwick 1994). Fig. 6 shows the spatial ordination of stations based on species abundances for the set of abiotic variables. In these plots, especially in the first one, the ordination of stations with regard to the horizontal axis can be explained on the basis of a hypothetical hydrodynamism gradient, so that the sites most exposed to currents are clustered to the left and those in calm conditions to the right. However, Stn 5 modifies this general pattern and, especially with silting, its extreme values can lead to a misinterpretation of the plots (Gamito & Raffaelli 1992). The representations of the rest of the abiotic variables do not explain the station groupings in a consistent manner either. The limitations of the MDS technique are evident, but it can be used previous to more detailed analyses. The BIO-ENV procedure is considered to be a better approach to the problem. Table 3 shows the combinations of environmental variables which give rise to the largest rank correlation (p,) between biotic and abiotlc sample dissimilarities. The combination of variables which best grouped the stations, in a manner consistent w~th the ascidian composition, involves hydrodynamism together with SOM (ps = 0.82), and the combination of these with suspended solids (p, = 0.78). Hydrodynamism is the single factor which reaches the maximum matching coefficient (p, = 0.64), and when further explanatory variables are added p, increases. However, if silting or suspended organic matter are included the coefficient decreases, probably due to their minor effects on community structure (Clarke & Ainsworth 1993). If the most important environmental factors for structuring the community were known, the localities having rather similar values for these factors would be expected to present rather similar ascidian composition, and an ordination based on this environmental data would group stations in the same way as for the biotic plot (Clarke & Warwick 1994). Fig. ?a shows the MDS of species abundance at 10 stations (Stn 5 removed). The remaining plots in this figure are those for the best 2and 3-variable combinations selected by BIO-ENV and that for the complete set of variables measured. As Clarke & Ainsworth (1993) stated, for consistency of presentation, these plots are also MDS ordinations, yet based on the Euclidean distance of normalised abiotic variables. This approach would therefore be acceptable, since the small number of variables lead to only minor differences between MDS and the corresponding Principal Components Analysis, which is expected when the higher-dimensional structure is well-represented in 2 dimensions (note the low values of stress obtained). Fig. 6. Two-dimensional MDS ordination of stations. Superimposed circles represent the annual averages for each of the abiotic variables (stress = 0.05) hydrodynamism b b.. 36 c, b ? s~lltng am o 0 SOM suspended sol~ds ' 0 00 0 '0 0 0 suspended organlc maller -0 "63 O @ 0 126 Mar Ecol Prog Ser 144: 119-131, 1996 Table 3. Results of BIO-ENV analysis combinations of environmental variables, taken k at a time, y~elding the 'best matches' of biotic and abiotic similarity matnxes for each k, as measured by standard Spearman coefficient p,. Bold type indicates the combination w~th maxlmum p, overall. Environmental factors hyd - hydrodynamism; susOM = suspended organic matter (percentage), SOM = organlc matter in the silt (percentage); sus = suspended solids; silt = s~lt~ng k 1 hyd (0.64) 2 hyd, SOM (0.82) 3 hyd, sus, SOM (0.78) 4 hyd, susOM, SOM, sus (0.67) 5 hyd, susOM, SOM, sus, silt (0.62) Best variable comb~nat~ons (p,) SOM SUS susOM silt (0.50) (0.21) (0.17) (-0.07) SOM, sus SOM, susOM hyd. sus . . . (0.66) (0.56) (0 491 hyd, SOM, susOM hyd, SOM, sllt (0.74) (0.64) hyd, susOM, SOM, silt hyd, susOM, SOM, silt (0.65) (0.61) The high degree of concordance between biotic and abictic plots in Fig. 7 is rerr?arkab!e, particularly ordinations (a) and (b), which group the sites in a very similar fashion. The cluster formed by Stns 1 to 3 and 11, which are located in the outer zone of the bay, are both consistent in the first 3 plots as is the poor relationship of Stn 9 to the others. Table 3 shows how p, declines slightly when further variables, apart from those most relevant, are added; this is also reflected in the corresponding MDS. In fact, the BIOENV procedure has a natural stopping rule for the cases where p, decreases with inclusion of unimportant variables (Clarke & Ainsworth 1993). This is shown in the MDS based on complete abiotic data (Fig. ?d), where the lack of concordance with the biotic MDS is obvious. hydrodynamlsm + SOM 77 hydrodynamism + SOM + sus~ended solids all envlronmenlal variables 10 On the other hand, the highest p, obtained (close to 1) shows a strong influence of the abiotic variables on the ascidian populations, but it is difficult to see if they are true environmental gradients, since their level of influence on each species remains cryptic. A CCA could be useful for solving this problem, given that in the resulting biplots the species ordination is directly related to the variation of abiotic factors (ter Braak 1986). In order to avoid distortions which can be caused by rare data (both biotic and abiotic variables), all data from Stn 9 and of AscidieUa aspersa and Molgula bleizei were removed. Furthermore, as suspended solids and the respective organic matter proportions were strongly correlated (r = -0.92; p c 0.051, the former variable was omitted so that multi-colinearity problems would not occur. In this manner, CCA was conducted on a revised set of data from 4 environmental variables recorded at 10 stations with a total of 36 species found. The biplot of species and hydrological factors provided by CCA is shown in Fig. 8. Axes I and I1 expla.in 73.4 O/n of total variance in the data, with eigenvalues of 0.32 and 0.11, respectively. The abiotic factors most related to axis I are hydrodynamism and SOM, whereas silting is close to axis I1 (see Table 4). The robustness of the analysis was determined using the Monte-Carlo permutation test (Hope 1968, in ter Braak 1988). The results were as follows: F-ratio = 2.1; p c 0.01. The outcome of CCA was also checked by comparing the species-factor correlations with those obtained by Correspondence Analysis (CA) and Detrended Correspondence analysis (DCA), where no evident Fig. 7. MDS plots of 10 sampling statlons based on: (a) abundance data of 38 ascidian specles; (h) and (c) the best 2and 3-variable combinations of differences were found. transformed environmental variables: and Id1 all environmental var~ables In the biplot, specles are expressed as ~, (see text). Stress values are: 0.04; 0; 0.01; and 0.03, respectively points and environmental factors as Naranjo et al.: Effects of environmental stress on ascidians SOM 1 0 C ~nt 0 I A.palo suspended organic Fig. 8. Species-factor biplot based on Canonical Cormatter respondence Analysis of a 36 species data set with respect to the abiotic variables (represented by arrows). Each arrow points in the direction of maximum change of the respective environmental variable silting across the diagram, and ~ts length IS proportional to the rate of change in this direct~on. Specles names are abbreviated from Table 1 arrows. The length of an arrow indicates the importance of this factor. Each arrow determines a direction or axis in the diagram, obtained by extending the arrow in both directions. The projections of a species on this axis shows its preference for high or low values of this environmental gradient (ter Braak 1986). We thus infer that Pseudodistoma obscurum, Polycitor cristalinus, Polycitoiadriaticurn, Synoicum blochmanni, Halocynthia papillosa and Rhopalaea neapolitana prefer areas highly exposed to currents and high levels of SOM; species such as Aplidium punctum, Trjdidemnum cereum, Clavelina dellavallei and Stolonica socialis have the highest weighted average with respect to suspended organic matter; while Ciona intestinalis, Ascidiella scabra, Styela plicata, Synoicum argus, Microcosmus squamiger, Ascidia mentula and Phallusia mamrnillata are most abundant in calm sites where highest contents of SOM (absolute values) are reached. The remaining specles prefer average values or show no clear preferences. Table 4. Canonical coeff~cients of and intraset correlations of DISCUSSION Relation to abiotic variables In benthic communities, physical disturbance, predation and competition are considered to be the most important factors affecting the abundance and distribution of species although, in relation to ascidians, the significance of larval behaviour must also be taken into consideration (Buss 1979, Svane & Lundalv 1982, Davis 1987, Svane & Young 1989). Along with these factors, the substrate features (nature, size and availability) determine the composition of benthic systems (Jackson 197713). In Algeciras Bay, we find that the heterogeneity of the environments investigated was conditioned by the harbour works and urban developments existing along its littoral. This physical variability, along with the different types of urban and industrial wastes, clearly affects the number and composition of species present, which have different levels of adaptation and tolerance. Therefore, the distinct influence that the maior 1 . P 1 and location of substrates in the bay. Those species env~ronn~ental variables with the first 2 axes of CCA. susOM: environmental factors have along the coastline should suspended organic matter (percentage); SOM: organic matter in the silt (percentage) be considered before the principal trends in the spatial distribution of ascidians can be established. Variables Canonical coefficients Intraset correlations Axis 1 Axis 2 Axis 1 Axis 2 and Clavelina lepadiformis) were more abundant in The similarity analyses show a clear correspondence between the levels of species grouping and the type Silting -O 039 -0.f13 -0.041 -O 770 SUSOM -0 223 -0.298 -0.236 -0 322 Hydrodynamlsm -O 695 0,144 -0,737 0,156 SOM -0 779 0.260 -0.827 0.281 which prefer light, shallow environments and which colonize artificial surfaces (Microcosmus squamiger, Styela plicata, Phallusia mamrnillata, Synoicum argus