SEARCH

SEARCH BY CITATION

Keywords:

  • Disturbance;
  • diversity;
  • long-term;
  • persistence;
  • sediment deposition;
  • sponge assemblages;
  • stability;
  • wind stress

Abstract

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

The importance of a long-term ecological perspective is well documented, yet long-term data are not widely available. This paper represents the first quantitative description of sublittoral sponge assemblages over long temporal scales (6 years) along the coast of the East Pacific Ocean (Mazatlan Bay), with the goal of describing their levels of intra- and inter-annual variability, and their relationship to changing environmental conditions. It was possible to detect an apparent short-term pattern (intra-annual), with a highly diverse and stable structure, usually in the drought season, which was consistent most of the years. However, only a few species showed regular (predictable) seasonal cycles. The species per square meter (from 0.1 to 0.5), total species per sampling (14–21), and H’ diversity using loge (1.4–2) also fluctuated greatly between years, suggesting that although a consistent short-term pattern was found most of the years, the inter-annual variability was also high. The univariate and multivariate results and regression models indicated that local winds played a key role in this short-term pattern (intra-annual). During the rainy season, or during the transition between the drought and rainy season, the winds coming from the southwest (WSW) produced an increase in the net sediment movement, which was an important factor for diversity. The long-term fluctuations (annual pattern of diversity) also correlated positively with wind speed (88% of the variance observed) and with sediment deposition (69% of the variance observed). In addition, the results indicated the existence of large-scale structuring factors, as the annual pattern of diversity also was correlated positively with the Southern Oscillation (SOI) and Multivariate ENSO (MEI) indices (82% and 88% of the observed variance, for MEI and SOI, respectively). In conclusion, this study suggests that sponge assemblages in Mazatlán Bay are extremely dynamic, and changes in this community are associated with processes that occur over the short- and long-time scales (several months to several years). The significant positive relationship between wind speed and sediment deposition showed that the main factors controlling the diversity in these shallow rocky ecosystems are the winds and the swell. A high proportion of coarse sand also indicated a very high energy in the environment, which, combined with the effect of silting, abraded and removed sponges and other organisms. These features contribute to the instability of the community by producing dramatic fluctuations in species abundance and preventing competitive processes from producing a more stable community.


Problem

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Many ecological studies have remarked on the need for long-term investigations (>5 years) (Buchanan & Moore 1986; Barry et al. 1995; Southward et al. 1995; Dye 1998; Estacio et al. 1999), yet much of the information about benthic ecosystems is derived from relatively short-term studies (Likens 1989). Thus, the temporal perspective in our understanding of marine benthic ecosystems is often a short one (Gray & Christie 1983).

The length of a study contributes significantly to its conclusions, generalizations and/or predictions, because longer studies have a greater probability of observing or helping to explain slow, rare or complex changes in natural systems (Fromentin et al. 1997;Levin 1992; Lundälv et al. 2004). With this perspective, long-term studies have a great potential for testing the generality of underlying mechanisms that structure the natural world (Buchanan et al. 1974; Ricciardi & Bourget 1999). In addition to contributing significantly to our general understanding of temporal stability in natural populations and communities, these studies have provided an invaluable perspective on fauna responses to habitat modification, flow regulation, climate change and introduced species (Likens 1989; Gido et al. 2004). Thus, they provide insights into other important long-term processes on marine ecosystems which could not be detected by short-term studies, such as the effects of global climatic fluctuations, anthropogenic disturbances and stochastic events (Fromentin et al. 1977; Hartnoll & Hawkins 1980; Lundälv & Christie 1986; Likens 1989).

Rocky marine assemblages are generally variable through space and time in an unpredictable manner. This variability is an integral part of the dynamics of all natural systems, making the prediction of ecological patterns very imprecise. In this sense, long-term studies are important for establishing confident predictions about the consistency of processes from one time to the next, and for understanding the variability of these natural systems (Levin 1992; Lundälv et al. 2004).

Although long-term (>5 years) studies of rocky systems are not common, the few that exist have illustrated the complex dynamics which characterize these environments (Pearson et al. 1986; Barry et al. 1995; Dye 1998; Kröncke et al. 1998). Several studies have quantitatively related long-term faunal changes to environmental variables such as temperature, storms and currents (Barry et al. 1995; Kröncke et al. 1998; Lundälv et al. 2004). Most of them indicate the existence of large-scale structuring factors probably linked to climatic events and climate variability (Lundälv & Christie 1986; Barry et al. 1995; Lundälv et al. 2004), or the interrelationship between these large factors and biological processes such as competition, predation and chemical defence (Lundälv et al. 2004). In addition, biological processes such as recruitment strategies and patterns have also been documented to influence long-term community trends (Lundälv & Christie 1986).

Sponges are widespread and abundant in benthic assemblages across different habitats, and are one of the most important groups in sublittoral rocky environments, where a number of factors control species distribution and assemblage structure (Sarà & Vacelet 1973; Bergquist 1978). In shallow-water coastal systems, these factors include hydrodynamic processes or sediment characteristics, which play a definitive role in structuring benthic assemblages (Carballo 2006; Carballo & Nava 2007). Sponges are ideal for studies of climate-driven community change because they are constantly exposed to the environment to an extent which is unique in the animal kingdom (Carballo & Naranjo 2002). There exist some data that have also related ENSO events and climatic changes directly to dramatic population change in sponges (Vicente 1989, 1990; Fromont & Garson 1999). Sponges also have the potential to provide unusually detailed information about oceans, and thus climate variations in the past (Ellwood & Kelly 2003a,b). In addition, sponges can be considered a model taxon to compare ecological processes across large spatial scales. Because they can influence other benthic organisms and ecosystem characteristics, this second aspect is also interesting as not much is known about the effect of large-scale environmental factors on the benthic community in the East Pacific Ocean.

Until recently, most research has been focused on short-term studies, or on local-scale explanations of community patterns (Wilkinson & Cheshire 1989; Diaz et al. 1990; Muricy 1991; Zea 1994; Sarà 1966; Bell & Barnes 2000a,b; Wulff 2005; Carballo 2006; Carballo & Nava 2007). Thus, the long-term dynamics of sponge assemblages are generally unknown, apart from broad responses to the physical environment (Sarà & Vacelet 1973; Bergquist 1978; Bakus & Green 1987; Barnes 1999; Carballo 2006). Only a few studies have considered long-term samplings or large-scale factors. The results of some recent ecological works about sponges (Bell et al. 2006; Carballo 2006; Carballo & Nava 2007) suggest that these assemblages are more dynamic than previously thought. This clearly contrasts with observations that the structure and dynamics of sponge communities exhibit an extremely slow rate of change in the Mediterranean area (Pronzato & Manconi 1995; Garrabou & Zabala 2001), even on a long-term scale (up to 10 years) (Pansini & Pronzato 1990; Marquez 2006; Corriero et al. 2007). The longest continuous study known for sponges also showed that the sponge community remained relatively constant over 16 years (Caribbean) (Hughes 1996). In contrast, another long-term study also undertaken in the Caribbean (14 years long) showed the disappearance of a high number of species during the study period (Wulff 2001, 2006). However, all the long-term sponge studies have been undertaken in Caribbean coral reefs and in rocky substrata from the Mediterranean Sea, and care should be taken when considering the dynamics of these assemblages because they may be different for Eastern Pacific rocky or coral reef ecosystems. Thus, there is a real need to describe the structure and dynamics of sponge assemblages across relevant temporal and spatial scales to ensure the validity, representativeness and generality of results.

In the Pacific Ocean, the preliminary results of a 16-month study showed that important short-term structural changes occurred in rocky littoral ecosystems from the Bay of Mazatlán, suggesting that the community is seasonally influenced by natural disturbances (Carballo 2006). However, we do not know about the consistency of these processes over a longer scale, or if this pattern occurs every year, which would allow us to establish a confident prediction about the physical processes that are shaping the structure of sponge assemblages, or about their stability and persistence.

In the present paper, the long-term variability of sponge assemblages at a site from the east Pacific Ocean was investigated from 2001 to 2006, most of the time on a monthly basis. We described the long-term variability in the assemblages and related these data to the environmental variability of several climate factors.

We addressed two main questions:

  • 1
    How do sponge assemblages, especially in shallow water, cope with environmental variability?
  • 2
    How much of the observed variability in these assemblages can be attributed to local variables and/or large-scale variables?

Material and Methods

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Study area

This study was carried out at a rocky coast located on Venados Island in the centre of the Bay of Mazatlán (México, Eastern Pacific Ocean) at a mean depth of 5 m (see previous papers; Carballo 2006;Carballo & Nava 2007). The shoreline of the island is composed of sandy beaches interrupted by rocky platforms that extend seaward 30–40 m or more. The sublittoral bottom consists of a gently sloping platform of a mosaic of relatively flat boulders between patches of sediment.

Meteorological conditions

The climate in the study area is tropical/sub-tropical, with two contrasting seasons in the year (drought and rainy). The average annual air temperature is 25 °C, and the average annual rainfall is 800 mm, occurring mainly during the rainy season from June to October (CNA). During winter, winds typically come from the northwest (NNW or NW) (data provided by the National Water Commission, CNA), and they produce a southward net sediment movement that redistributes the sand at the continental coast from the upper shoreline downward onto the platform. In contrast, during the summer the rains coincide with the winds from the southwest (WSW), producing a net sediment movement towards the north (Peraza 1985).

Sampling

The density and coverage of the sponges were estimated in three permanent 25 m long transects perpendicular to the shore, at 5 m depth. The transects consisted of a plastic line with small lead weights located at the same depth and in geomorphologically similar zones. At the beginning of the survey, a random point was chosen to attach the first transect, and the next two transects were situated 15 m apart and parallel to the first. Density was estimated by counting all sponges found inside an aluminium frame of 4 m2 (2 × 2 m), which was laid six times along the transect lines, resulting in a sampling area per transect of 24 m2 (6 frames × 4 m2). In total, 72 m2 were sampled monthly for the density (3 transects × 24 m2). Small samples of sponges that could not be identified in situ were taken for laboratory analysis. The area of surface coverage for each sponge specimen (cm2·m−2) was obtained by counting 1 cm2 meshes of an overlaying grid of 50 × 50 cm (0.25 m2), placed inside each aluminium frame to cover the total area.

The abundance of sponges was estimated monthly in 2001 and 2002. In 2003, 2004 and 2006 it was sampled every 4 months, because after we examined preliminary results we considered this sampling frequency sufficient to study the dynamics of the community and its relationship with the general environment. In early 2005, a different pattern was detected, and we returned to the original monthly sampling.

Most of the species reported in the Bay of Mazatlán are predominantly encrusting sponges, for which the density (as number patches per m2) and coverage area (as cm2 per m2) are adequate to estimate their abundance. An individual was defined as any specimen growing independently of its neighbours.

Analysis of patterns in sponge assemblages

The abundance, the number of species per square meter, the Shannon–Wiener index of diversity (H1 using loge), and the total number of species per sampling were used to describe changes in the sponge assemblages over time. Cluster analysis based on Bray–Curtis similarity matrices of root-transformed abundance of the species was used to study the assemblages’ similarities over time, using the species as variables. The clusters thus produced consisted of groups representing similar ‘states’ of the assemblages. In addition, non-metric multidimensional scaling ordinations (nMDS), based on Bray–Curtis similarity matrices of root-transformed abundance data, were used to create 2-dimensional representations of the temporal state of the assemblages (Field et al. 1982).

The use of permanent transects allows acquisition of information about individual specimens. Thus, interpretation of community dynamics was aided by monitoring the fusion, fission, mortality and interactions of the most conspicuous species. Increases in the area of the individual sponge were also considered when evaluating growth.

Local climatic variables

To examine the effects of some important local environmental variables on sponge assemblages, we used monthly means of rain, wind speed and direction (data provided by the Local Water Commission), water temperature and sediment deposition (see previous papers for more information about the methodology: Carballo 2006; Carballo & Nava 2007). The abundance and composition of the particles was quantitatively assessed, and the total amount of material sedimented was then expressed as kg DW m−2·day−1 (DW, dry weight). The abundance of different sizes of particles (in mm) was expressed as the percentage over the total number determined on preweighed meshes (Inman 1952).

A correlation-based principal component analysis (PCA) was applied to ordinate data from the general environment.

Global climatic variables

To examine the effects of wider variables we tested relationships between sponge assemblage changes and the following global climatic variables:

Southern Oscillation Index: El Niño/Southern Oscillation (ENSO) is the most important coupled ocean–atmosphere phenomenon to cause global climate variability on inter-annual time scales. Sustained negative values of the SOI indicate El Niño episodes. These negative values are usually accompanied by sustained warming of the central and eastern tropical Pacific Ocean and increased convection or cloudiness in the central tropical Pacific Ocean. Positive values of the SOI are associated with La Niña episode. Waters in the central and eastern tropical Pacific Ocean become cooler during this time. The Niña event is also associated to stronger than normal (easterly) trade winds across the Pacific Ocean (Philander 1990; Harrison & Vecchi 2001).

Multivariate ENSO Index (MEI): This index is based on six main variables observed over the tropical Pacific: sea-level pressure (P), zonal (U) and meridional (V) components of the surface wind, sea surface temperature (S), surface air temperature (A), and total cloudiness fraction of the sky (C) (Wolter 1987). The index has been developed mainly for research purposes as MEI integrates more information than other indices for the overall monitoring of the ENSO phenomenon and world-wide correlations with surface temperatures and rainfall. It reflects the nature of the coupled ocean–atmosphere system better than either component, and it is less vulnerable to occasional data glitches in the monthly update cycles (Wolter 1987; Wolter & Timlin 1993). The MEI is calculated as the first unrotated Principal Component (PC) of all six observed variables combined. This is accomplished by normalizing the total variance of each field first, and then performing the extraction of the first PC on the co-variance matrix of the combined fields (Wolter & Timlin 1993). The MEI is computed separately for each of 12 sliding bi-monthly seasons (Dec/Jan, Jan/Feb, . . . Nov/Dec). Negative values of the MEI represent the cold ENSO phase, La Niña, and positive MEI values represent the warm ENSO phase El Niño. To relate the MEI to monthly values of other variables (biological variables in our study), the MEI value of month (i−1) and month (i) is recommended as if it was the value for month (i) only (Wolter 1987; Wolter & Timlin 1993, 1998).

Relationship between the abiotic and biotic variables

The relationship between the abiotic and the biotic variables was studied by means of Spearman rank correlations. Linear and non-linear least-squares techniques and quadratic polynomial regression were also used to test the nature of these relationships. The non-linear models also included a power and an exponential function.

Mean annual species richness and diversity were also plotted against the annual historical average of SOI and MEI. The purpose of this part of the study was to test the hypothesis that the global abiotic variables such as the MEI and SOI could explain the long-term patterns observed in the sponge assemblages.

As there were years with 11–12 samplings (months), and years with 3–4 samplings, we calculated the mean annual values of the diversity using the total of sampling each year (from 3 to 12 samplings) and using only the common months sampled (3–4 samplings). Significant differences between the mean annual values from all the months and from only the shared months were analysed by ANOVA (Cochran’s C-test), and were further analysed using an a posteriori Student–Newman–Keuls (SNK) test at the 5% significance level.

Results

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

General environment

Local variables

Water temperature changed greatly over time, but a seasonal pattern was evident, with high values in spring /summer and low values in autumn/winter. Rainfall followed a similar pattern, with heavy rains from August to September when water temperature was higher. Rainfall also coincided with the winds from the southwest (WSW), but during winter, winds typically came from the northwest (NNW or NW). This area is also occasionally subject to strong wind–wave action caused by local winds and tropical storms.

Sediment deposition and wind speed were simultaneously represented (Fig. 1). The direction of the winds changed alternately over time; during winter and early spring (drought season) most of the winds came from the northwest (NNW or NW), and in the rainy season the winds came mainly from the southwest (WSW). In most cases, sediment deposition increased with winds coming from the southwest, or during the transition of the winds from north to southwest. This produced a sediment movement that increased deposition in rocky habitats. For example, in May 2001 the trap sediment placed in the study area collected a very large amount of material (up to 11 kg DW·m−2·day−1). Particles were mostly composed of medium sand (0.2–0.4 mm in diameter), but a high proportion of coarse sand (>1.4 mm in diameter) was also observed frequently in the summer months, suggesting a high resuspension of particles from the bottom.

image

Figure 1.  Variation of the wind speed (lines, left Y axis), and sediment deposition (bars, right axis) over time. The main wind direction is indicated with different types of lines. The bold line represents the winds coming from the southwest (W-WSW), and the discontinuous line represents winds coming from the northwest (NNW-NW).

Download figure to PowerPoint

A correlation-based principal component analysis (PCA) was applied to ordinate data from the general environment in Mazatlán Bay. The model separated the months of the drought season from the rainy season, and explained 69.8% of the variation (not represented). The main factors loaded for the first principal components, which explained 53.5% of the variations of the model, were air temperature (−0.42), winds coming from the southwest (−0.39) and water temperature (−0.39), showing that the general environment in the Bay is clearly influenced by the winds.

There is a significant positive linear regression between wind speed and the sediment deposition (Fig. 2). When we split the wind in its two main directions, we did not find a significant relationship for each one separately; however, the lines representing the ‘best fit’ for the data present a different trend, showing that the winds coming from the southwest are responsible for the increase in sediment deposition (Fig. 2).

image

Figure 2.  Relationship between sediment deposition and wind speed (top). The same but splitting the wind in its two main directions (bottom, NNW-NW, discontinuous line, and W-WSW, bold line).

Download figure to PowerPoint

Global variables (Southern Oscillation and Multivaried ENSO Indices)

The last major El Niño event occurred at the end of 1997, and remained very strong until the end of 1998, causing important changes in the benthic communities in our study area (Carballo et al. 2002). A cold La Niña event followed this El Niño event, in 1999 and 2000. After that, the years 2001, 2002 and 2003 were considered to have neutral conditions, although they showed strong variations in the SOI and MEI (Fig. 3). 2004 was also considered to have neutral conditions, but most of the months presented strong negative SOI values; see for example January (−12), April (−15) and June (−14). In 2005, the strongest negative value in the whole period of study (February, −29) was detected. But this year was also considered neutral (Fig. 3).

image

Figure 3.  Monthly fluctuation in the Southern Oscillation (top) and Multivariate ENSO (bottom) indices during the whole period of study.

Download figure to PowerPoint

During 2006, weak La Niña conditions were presented from January to April, with positive values ranging from 12.5 to 15. During May to July the SOI values were slightly negative, which means neutral conditions. From August to December, the SOI values remained negative but stronger, causing El Niño conditions (source, the climate update, NIWA). Positive temperature anomalies in the equatorial Eastern Pacific peaked in December 2006; the SOI remained steady during December. The MEI also fluctuated in a similar way as SOI, although here, negative values represent the cold ENSO phase, La Niña, and positive MEI values represent the warm ENSO phase El Niño (Fig. 3).

Biological patterns

Intra-annual pattern

A total of 58 species were found in this study, with an average of 18 species per sampling, a minimum of 8 (May 2001) and a maximum of 30 (November 2003 and June 2005).

The most frequent species over time was Microciona sp., followed by Mycale (Carmia) magniraphidifera van Soest, 1984, Cliona euryphylle Topsent, 1888, Cliona papillae Carballo, Cruz-Barra & Gomez, 2004 and Mycale (Carmia) cecilia de Laubenfels, 1936 (Fig. 4). These were the most persistent species throughout the whole study, as almost 30% of the species appeared once or twice during the whole study. When we plotted the cumulative number of species over time (Fig. 5, bottom), new species appear continuously; however, most of them are only occasional and transitory as they appeared once or twice in the whole study, such as Aplysilla glacialis (Merejkowski, 1877), Plakortis albicans Cruz-Barraza & Carballo, 2005, Scopalina rueztleri (Weidenmayer, 1977) and Chondrosia tenochca (Carballo et al., 2003). One might think that the sampling was not adequate to study the dynamics of these assemblages. However, when representing the cumulative number of species against the increase in the sampling area, the minimum area is reached before the 30 m2 (Fig. 5, top), a much smaller area than that sampled.

image

Figure 4.  Persistence of the different species found during the study. The persistence is expressed as percentage of the months in which each species appears. Species with a value of 100% show that they were present during the whole study period.

Download figure to PowerPoint

image

Figure 5.  Relationship between area sampled (cumulative area) and species richness at different times (top). Variation in the cumulative number of species over time (bottom).

Download figure to PowerPoint

Diversity fluctuated greatly, and ranged over time from 0.8 (August 2001) to 2.3 (January 2004). In general it showed a short-term consistent pattern, peaking in the drought season (see for example from February to April 2001, from January to March 2002, and in winter of 2003 and 2004) (Fig. 6). The lowest values during this long period were detected during the rainy season (warm months) (see from May to October 2001, from April to June 2002 and in May 2003).

image

Figure 6.  Variation in the mean H’ diversity (continuous line and dark circles, left axis), and number of species per square meter (discontinuous line and open circles, right axis) over time.

Download figure to PowerPoint

However, an apparently inverse pattern with diversity peaking in the rainy season was found in September 2005. In 2006 diversity starts to fluctuate as it did at the beginning of the study.

The species per square meter fluctuated from 0.08 (May 2001) to 0.9 (June 2005) in a consistent way with diversity, and it also showed the same trend (Fig. 6). However, it is also important to highlight the low number of species per square meter found during the study (less than one species per square meter).

Abundance (both density and coverage) also fluctuated greatly during the study: density fluctuated from 2.6 (May 2001) to 17.5 ind·m−2 (February 2006), and coverage from 3.4 (July 2001) to 163.3 cm·m−2 (February 2006) (Fig. 7). We only have information for coverage during part of the study, but both measures covaried positively and a significant correlation was found between density and coverage (r = 0.80, P = 0.001). So, we can describe the variation in the abundance with either of these parameters. In contrast with diversity, no clear pattern was found in the variation of abundance. This peaked at a different month each year; in 2001 it was highest in February (lowest in May), in 2002 it was highest in July (lowest in November) and in 2005 it was highest in September (lowest in February). Only Microciona sp., M. magniraphidifera, C. euryphylle, C. papillae and M. cecilia showed little inter-annual variation in their abundance (Figs 8 and 9). They changed seasonally, peaking in summer each year, with lower values in winter such as Microciona sp., or they maintained a constant value most of the year, such as Cliona spp. (Fig. 8). However, the temporal variation of most of the species was very heterogeneous, fluctuating in a different way each year, and therefore they presented important inter-annual differences in their abundance. Examples are Callyspongia californica Dickinson, 1945, Haliclona (Soestella) coerulea (Hechtel, 1945), Haliclona turquoisia de Laubenfels, 1945, Mycale aff. parishii (Bowerbank, 1875) and Tedania sp., which were very abundant at the beginning of the study (the first 3 months), and after November 2004 (Figs 8–10). There were also species that were present during the whole study, but disappeared at the end of 2004, such as Spirastrella decumbens Ridley, 1884, Haliclona (Reniera) tubifera (George and Wilson, 1919) and Desmanthus sp.

image

Figure 7.  Variation in the mean density (dark circles, left axis), and coverage (open circles, right axis) over time.

Download figure to PowerPoint

image

Figure 8.  Monthly (line plots, to the left) and mean annual (bar plots, to the right) variation of the abundance (density as black circles and bars and coverage as open circles and bars) of Cliona papillae (top), Microciona sp. (middle) and Cliona euryphylle (bottom).

Download figure to PowerPoint

image

Figure 9.  Monthly (line plots, to the left) and mean annual (bar plots, to the right) variation of the abundance (density as black circles and bars and coverage as open circles and bars) of Callyspongia californica (top), Adocia turquoisia (middle) and Mycale cf. parishii (bottom).

Download figure to PowerPoint

image

Figure 10.  Monthly (line plots, to the left) and mean annual (bar plots, to the right) variation of the abundance (density as black circles and bars and coverage as open circles and bars) of the association Haliclona coerulea/Jania adherens (top), Spirastrella decumbens (middle) and Tedania sp. (bottom).

Download figure to PowerPoint

Only some specimens of the most conspicuous species could be followed over time. These specimens lived between 4 and 10 months, but mortality in most specimens of small species occurs independently from that of other individuals. However, it is very difficult to split the mortality from the effect of fragmentation, as fragments from sponge breakage can reattach in some other place, thus masking mortality. Fusion and fission occurred in some species. It was more frequent in some cushion-shaped sponges such as C. californica, or M. cecilia, or in massive species such as H. coerulea, and it was rare in encrusting species such as Microciona sp. However, no particular seasonal trend in the occurrence of fusion and fission was found (Fig. 11). Overgrowth by other sponges was not observed. The specimens monitored varied in size but their growth appeared discontinuous and was not seasonal.

image

Figure 11.  Patterns of occurrence of fusions (plots to the left) and fissions (plots to the right) of Haliclona coerulea (a), Mycale cecilia (b), Microciona sp. (c), and Callyspongia californica (d).

Download figure to PowerPoint

The multivariate techniques are very useful to determine whether community structure varied among samplings, and to find similar assemblages over time. The cluster analysis differentiated clearly among two main groups; one included February, March and May 2001 together with all the samplings from November 2004 to the end of the study, and the other group included the rest of 2001, and the samplings corresponding to 2002, 2003 and 2004 (Fig. 12). The nMDS ordination also suggests that although there was considerable overlap in assemblage types between years, it was possible to distinguish between two main assemblages (Fig. 12); see all the sampling after November 2004 close to the first months of 2001 on the left of the plot, and the rest of 2001; and the samplings corresponding to 2002, 2003 and 2004 on the right, showing that a similar pattern in the structure of the assemblages was detected after November 2004, almost 3 years after the beginning of the study. The multivariate results agree well with those showed by diversity and species richness, as these two groups of months are also clearly separated in the diversity plot (Fig. 6), with November 2004 as the inflexion point between the two periods.

image

Figure 12.  Cluster analysis of samplings based upon group-average clustering using Bray–Curtis similarity (top). nMDS ordination for the sponge assemblages over time (bottom). The vertical line in the nMDS split the assemblages of the two main periods of study (see Discussion for more information).

Download figure to PowerPoint

Explaining the short-term (intra-annual) and long-term (inter-annual) patterns

Diversity usually decreased when the winds coming from the southwest increased (Fig. 13). However, the only significant correlation was that established between sediment deposition and diversity, suggesting that the short-term temporal pattern of the sponge assemblages was in part governed by this factor (59.9% of the observed variance) (Fig. 13).

image

Figure 13.  Relationship between wind speed split in its two main directions and H’ diversity (top). Gross line and dark circles represent the winds coming from the southwest (W-WSW), and dotted line and open circles represents the winds coming from the northwest (NNW-NW). Regression analysis between H’ diversity and sediment deposition (bottom).

Download figure to PowerPoint

The species per square meter (0.1–0.5), total species per sampling (14–21), and H’ diversity (1.4–2.0) also fluctuated greatly between years, suggesting that although a consistent short-term pattern was found most of the years, the inter-annual variability was also high (Fig. 14).

image

Figure 14.  Inter-annual variation of the H’ diversity (circles to the left y-axis), and species per square meter (bars to the right y-axis) during the period of study (top). The black circles and bars using all the months sampled each year. The open circles and bars using only the shared months (see Results for more information). Inter-annual variation of the mean density (bottom). The black bars using all the months sampled each year. The open bars using only the shared months (see Results for more information). Vertical lines indicate standard error.

Download figure to PowerPoint

The annual SOI and MEI average were also plotted against annual average diversity (both H’ diversity and species per square meter). The purpose of this part of the study was to test the hypothesis that the global abiotic variables, such as the MEI and SOI, could explain the long-term patterns observed in the sponge assemblages.

The best model obtained was polynomial quadratic, with species richness showing a unimodal response to both MEI and SOI (concave down parabola). This model explained most of the observed variance: 83% and 88% of the observed variance, for MEI and SOI respectively (Fig. 15).

image

Figure 15.  Quadratic polynomial regression between mean annual H’ diversity and mean annual wind speed in the area of study (top). Quadratic polynomial regression between mean annual species richness and mean annual Southern Oscillation Index (middle). Quadratic polynomial regression between mean annual species richness and mean annual Multivariate ENSO Index (bottom). Each point in the plots represent 1 year of the study.

Download figure to PowerPoint

The relationship between diversity and these indices was stronger than the relationships of the annual average diversity with local variables, and it is very interesting because the global variables integrate the most important factors related to the climate at global scale, such as rains, winds, water and air temperature, sea-level pressure and even the amount of cloudiness. So, as the local variables such as winds, waves and sediment deposition control part of the dynamics of benthic assemblages, it is expected that these indices more strongly control the inter-annual variability. In fact, the annual pattern of diversity also correlated positively with wind speed (88% of the variance observed) and with sediment deposition (69% of the variance observed).

Species peaked toward neutral values of both indices. Negative values of the MEI represent the cold ENSO phase, La Niña, and positive MEI values represent the warm ENSO phase, El Niño. In contrast, positive values of the SOI are associated with a La Niña episode, and negative values with an El Niño event.

The same trend is observed if we considered only the same months each year, as no significant differences were found for abundance and diversity when we use only 4 months per year in the analysis (Fig. 14).

At population level, the only species that presented interesting long-term relationships with environmental variables was Tedania sp., as the annual average abundance of this species correlated positively with water temperature (with the standard deviation) (r = 0.9, R2 = 0.8, P = 0.01, F-ratio = 18.2), and with SOI (r = 0.82, R2 = 0.67, P = 0.04, F-ratio = 8.34). This variation is quite interesting because the population decreased greatly since the beginning of the study, and not from May 2001 as other species did. This relationship with the water temperature and SOI suggests that water temperature may control the dynamics and regional distribution of this species. In fact, this is not a typical species from tropical areas; it is a temperate species and only appears at tropical latitudes when water is colder.

Discussion

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

Short- and long-term diversity patterns

Although some ecological processes are remarkably consistent, or vary with predictable seasonality, others vary on time scales with no clearly apparent patterns (Underwood 1999). However, when an appropriate scale of time is used, it is possible to explain part of the variability observed (Levin 1992; Lundälv et al. 2004).

Temporal variations of sponge assemblages have been studied for short periods of time (usually 1 year or less) (Wilkinson & Cheshire 1989; Diaz et al. 1990;Wulff 1990; Muricy 1991; Zea 1994; Sarà 1966; Bell & Barnes 2000a,b; Bell & Smith 2004; Carballo 2006; Carballo & Nava 2007). The longest studies known so far, 16 years (Hughes 1996) and 14 years long (Wulff 2006) were both undertaken in Caribbean coral reef ecosystems, and showed that the sponge community remained relatively constant. However, a community is persistent when it maintains its structure according to species composition. In this sense, the stability of the sponges at community level masked the real underlying dynamics of individuals as 78% and 51.3% of the species initially present disappeared during the initial years of study (Hughes 1996 and Wulff 2006, respectively), and there was a drop of 42.6% in total volume (Wulff 2001, 2006). This result is unexpected as the long-term monitoring studies from hard substrata in the Mediterranean Sea (Pansini & Pronzato 1990; Marquez 2006; 6 and 10 years, respectively) had shown a high and remarkable degree of persistence on the sponge community, with only two species disappearing from the study area and no new species entering (Pansini & Pronzato 1990). Corriero et al. (2007) recently reported that more than 80% of the Demospongiae reported by Sarà during 1954–1955 in the Venice Lagoon were found in 2001.

Similarly to the studies from the Caribbean, persistence was relatively low in Mazatlán Bay, as less than 20% of the species were present most of the time. The persistence is the result of a specialized strategy in the dynamics of the community, which is characterized by a high biomass, a low dispersal ability, and development in a highly predictable environment where catastrophic events are normally absent (Ott 1981). The sponges in Mazatlán Bay do not meet at least two of these conditions: high biomass and living in highly predictable environment. The same situation seems to occur in each of the zones along the Mexican Pacific coast that we have sampled (data not published).

The results also showed enormous variability in community dynamics, as the assemblages maintained a highly diverse and stable structure usually in the winter–autumn season (drought season in tropical areas). However, despite such variability, it was possible to detect an apparent short-term pattern (intra-annual), which was consistent most of the years. This pattern has also been documented previously for sponge assemblages, but over a shorter period of time (Carballo 2006).

When analysing mangrove root habitats we found the same variability in the dynamics of sponge assemblages. The sponges in Caribbean mangrove roots (Venezuela) showed remarkable stability, with species abundance and community composition changing little over an 18-month period (Sutherland 1980). This picture of a stable community is dramatically different in the mangrove roots communities of the Florida Keys, where a longer study (38 months) showed enormous variability (including sponges), despite being dominated by long-lived species with low, largely non-seasonal recruitment (Bingham & Young 1995). This level of variability was similar to that described by Ellison & Farnsworth (1992) for sponges in a Belizean mangrove habitat.

This lack of consistency between studies could be partly due to the different approaches of the authors, making it very difficult to judge whether a given pattern or process would be variable or consistent through time. For example, the depth varied from shallow water (this study) to 35 m (Hughes 1996). It is an almost indisputable fact that the shallow-water sponge populations (this study) are subjected to great temporal variations as a result of the constant changes taking place in their environments (Burton 1949; Sarà 1970). However, although our data support this assumption, this does not seem to be a general pattern, because some works showed that the composition of sponges also exhibits long periods of stability in shallow coral reef ecosystems from the east Pacific Ocean (see Wulff 2001, 2006). The sampling area varied from a few meters (Pansini & Pronzato 1990) to 72 m2 (this study), and the sampling intervals ranged from annual or bi-annual (Wulff 2001, 2006) to monthly or bi-monthly (this study). The latter is especially important, as analysis on longer scales used by other authors could give inaccurate impressions of community dynamics because a much more dynamic community is perceived at shorter sampling intervals (Bingham & Young 1995). The frequency of perturbation controls the temporal scale at which ecological transitions take place. In stable environments where biotic interactions are more important structuring agents, longer sampling intervals may be appropriate (for example annually or bi-annually) (Wulff 2001, 2006). However, in the rapidly changing sponge assemblages in Mazatlán Bay, frequent stochastic events destabilize the species assemblages and prevent longer-term ecological processes, such as competition, from structuring a more stable community. So, many censuses per year are required.

On the other hand, the temporal variation of most of the species in Mazatlán Bay was very heterogeneous, fluctuating in a different way each year and with little evidence of seasonality. Only a very few species, such as Microciona sp., presented a clear seasonal pattern of abundance. This lack of regularity is difficult to explain. It could be partly related to stochastic (non-predictable) processes. Each year, a considerable number of sponges are found on the beaches around the sampling sites, some of them detached together with their substrata (J. L. Carballo, unpublished data, Fig. 16). This occurs in the moments of intense swell provoked by the winds or tropical storms, affecting only the massive (such as H. coerulea), or branched species (such as Mycale aff. parishii), in a similar way to the hurricanes in the Caribbean (Wulff 1995). The preliminary information showed that these processes are variable both intra- and inter-annually, and they could be partly responsible for the irregular variation in the abundance of these species. Nevertheless, this is not the only possible explanation, because they do not seem to affect most of the encrusting sponges, which also fluctuated very irregularly and constitute more than the 80% of the species.

image

Figure 16.  Biomass of sponges found ashore on the beaches near to the sampling area during 2005 (open bars), and 2006 (black bars). Vertical lines indicate standard error. Data not published. Methodology not described in Material and Methods. The plot is shown only to highlight the effect of the winds and swell on the sponge assemblages (see Discussion for more information).

Download figure to PowerPoint

Searching for an explanation in the patterns: winds and waves

Equilibrium communities are generally thought to be structured by biotic interactions (primarily competition), whereas non-equilibrium community dynamics are controlled by physical disturbance and stochastic events (see Dayton 1984). The assemblages in Mazatlán Bay showed the high variability that is characteristic of a community controlled by disturbances. However, despite such variability, it was possible to detect an apparent short-term pattern (intra-annual) that was consistent most of the years.

But what destabilizes the sponge assemblages in Mazatlán Bay and produces the large fluctuations in abundance and diversity? Sponge diversity was consistently higher during the months previous to the change in the direction of the dominant winds, and the significant correlation established between sediment deposition and diversity suggests that this was one of the major factors controlling the short-term temporal pattern of the assemblages.

Interestingly, many marine ecosystems in different regions of the world show surprising parallels in this aspect. For example, Sandston reefs (Fortaleza, NE Brazil, Atlantic Ocean) are characterized by warm waters year round, but strong swells occur when the winds are stronger (August–January) (Silva et al. 2002). During this time, the abundant sand deposited on top and around the sandstone reefs may become suspended and increase stress for sponges, a common occurrence on beach-worn debris (Salani et al. 2006). The influence of large-scale natural disturbance from winter storms (northers) and river runoff are responsible for a naturally impoverished macroinfauna community from the Gulf of Mexico (Hernández-Arana et al. 2003). The northers season on the Campeche Shelf is consistently linked to strong winds of 20–30 cm·s−1 (Salas de León et al. 1992), and to the resulting lower biomass on the terrigenous shelf (Hernández-Arana et al. 2003). The action of storms is also a very common feature of rocky coasts worldwide, and they are an important factor structuring littoral benthic communities, as waves may travel large distances from where they are generated. Wave action and rain are also responsible for large fluxes of sediment along rocky cliffs in the Ligurian Sea (Italy) (peaking during Spring) (Bavestrello et al. 1991). Similarly, wave disturbance associated with winter storms on the California inner shelf is thought to be responsible for macroinfauna zonation and strong fluctuations in densities of opportunistic species (Oliver et al. 1980). In a region of Kenya, the rainy season (May–November) coincides with southeast monsoon winds, which entrain inshore currents and river discharge northward, affecting a high-diversity coral reef ecosystem (McClanahan 1988, 1990; McClanahan & Mutere 1994).

Wind stress also limits benthic secondary production and regulates benthic processes through disturbance of shallow shelf benthos (Emerson 1989), and this effect is not limited to shallow waters, as storms are presumed to re-suspend sediment sizes from 2.5 to 6 mm at depths of 50–80 m (Logan et al. 1969).

In the Antarctic waters, changes in the zonal winds during the 1982–1983 ENSO event were associated to long-term benthic biological dynamics. This change modified the circulation pattern within McMurdo Sound during that period, which accounted for the strong northward flow that coincided with the beginning of strong platelet ice formation, after a decade of low anchor ice and platelet ice formation (Dayton 1989).

Integrating local and large-scale factors

Marine ecosystems must endure both large-scale and local natural processes. However, until recently, most research focused on local-scale explanations of community patterns. Thus, the attempt to integrate the role of large-scale factors with local-scale processes remains a subject of major interest in ecology (Menge & Olson 1990).

A consistent short-term diversity pattern was found most of the years, which in part was controlled by local factors. In fact, the most important correlation regarding the biological pattern was the one established between sediment deposition and diversity (59.9% of the observed variance), which clearly shows how high levels of sediment deposition are detrimental to the diversity of sponge assemblages (Carballo 2006;Bell et al. 2002; Bell 2004). Raised levels of sedimentation have been proposed to explain impoverished sponge communities (Sarà & Vacelet 1973), because if siltation is excessive it can cause a reduction in filtration, or stop it completely (Reiswig 1971). It can also increase water turbidity, affecting the species living in association with photosynthetic symbionts (Sarà & Vacelet 1973).

Thus, the maintenance of the sponge assemblage in areas seasonally influenced by sand deposition may involve periods of growth and retraction (Stone 1970; Elvin 1976), as has also been documented for other organisms such as algae, which are restricted to a short growth season (Lieberman et al. 1979). In addition, the high proportion of coarse sand (>1.4 mm in diameter) also indicated a very high energy in the environment that, combined with the effect of silting, most likely abraded and removed sponges and other organisms from the rock surfaces, reducing abundance for most of the conspicuous taxa, and affecting the structure of the community by the elimination of species. In fact, a few of these conspicuous taxa are found on the beaches around the sampling sites, some of them detached together with their substrata. The persistent low number of species is likely a result of this combined physical process (silting and abrasion). The domination of encrusting and cushion-shaped species (>80% of sponges) also confirms it (Bell & Smith 2004). Similar studies have showed that a high sediment deposition influenced the diversity of benthic assemblages by excluding less tolerant species and by favouring monopolization of space by the most tolerant species (Engledow & Bolton 1994).

However, the assemblages also fluctuated largely between years, and the annual pattern of diversity also correlated positively with local variables like wind speed (88% of the variance observed) and sediment deposition (69% of the variance observed). Solid data that integrate local-scale with large-scale physical processes are lacking, despite evidence suggesting they are coupled. In McMurdo Sound (Antarctica), for example, zonal winds (a local process) associated with the 1982–1983 ENSO event (a large-scale process) were responsible for changes in the water circulation pattern, which modified the dynamics of the benthic community (Dayton 1989). Interestingly, the relationships between the annual pattern diversity and the global climatic variables (SOI & MEI) were also strong (82% and 88% of the observed variance, for MEI and SOI, respectively), showing that both large-scale and local natural processes are responsible for the pattern.

In ecological terms, it is important to identify the proportion of variance in biological time series which may be attributed to climate variability, and this study is consistent with predictions of changes in the benthic community associated with such variability, as these indices (MEI and SOI) integrate the most important variables related to the climate at a global scale, such as rains, winds, water and air temperature, sea-level pressure and even the degree of cloudiness (Wolter 1987; Wolter & Timlin 1993, 1998).

Responses of benthic assemblages to ENSO fluctuations are also well-known (Paine 1986). These indices have been proven to be related to biological processes in several papers, and interesting relationships have been previously documented between an ENSO event and diversity and biomass of macroalgal assemblages (Carballo et al. 2002), changes in the distribution and abundance of sponges (Dayton 1989), or with the annual numbers of green turtles (Chelonia mydas) breeding around northern Australia (Limpus & Nicholls 1988).

Sponges as bioindicators: Tedania sp. as bioindicator of change in the global surface water

The life habits of sponges make them excellent bioindicators for evaluating the health of the marine environment (Carballo & Naranjo 2002). Research has indicated that marine sponges are suitable as biomonitors of heavy metal pollution (Patel et al. 1985) and as biomarkers for the monitoring of physical and chemical stress (Pérez et al. 2002). In addition, sponges have the potential to provide unusually detailed information about the ocean, and thus climate variations in the past (Ellwood & Kelly 2003a,b).

Tedania sp. is a temperate species typical of the northern Sea of Cortez, where the average water temperature is lower than in Mazatlán Bay. It seems that only during colder temperatures, as occurred during the last two La Niña events, can this species reach more tropical latitudes. We could consider Tedania sp. as a good indicator of change in the global surface water temperature in the East Pacific Ocean, as we found an interesting relationship between the abundance of this species and the positive values of the Southern Oscillation Cycle.

Conclusions

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

We can conclude that shallow sponge assemblages in rocky tropical systems are dynamic, reflecting not only seasonal and annual variation in environmental conditions, but also the effects of local disturbances, which are also variable both in the short- as well as in the long-term (Carballo 2006; Carballo & Nava 2007). Species composition changed greatly on short time scales and structuring processes occurred on very short time scales (c. 4 months). The results showed a high species turnover both on a short- and a long-time scale, and at both scales the assemblage changes are largely consistent with local and global environmental changes. The paper also constitutes an example of how large-scale physical processes exert profound biological effects over long-time periods.

A critical factor when judging possible effects of the changes reported here on the coastal ecosystem is the degree of generality of these patterns; there was little direct evidence to judge whether these changes are of a general nature for the entire Eastern Pacific Ocean. Interestingly, a similar short-term pattern has been found almost 800 km away from our area of study for macroalgae assemblages and their associated fauna, with higher values of diversity in winter than in summer (Steller et al. 2003). Physical disturbance (such as wind stress, waves and storms) is probably a major structuring force not only for shallow water rocky bottoms from Mazatlán Bay, but also for coral reef ecosystems. Studies carried out on coral reefs in the Eastern Pacific also indicated that wave action and sediment movement have a strong control on coral reef dynamics (Glynn 1976).

Although the generalization of the response of sponge assemblages (and indeed the entire benthic community) to physical processes such as wind stress, waves and storms needs to be treated with caution, interestingly, the same contrasting pattern in the season (rainy versus dry) seems to drive ecological structure in benthic communities from the Gulf of Mexico (Hernández-Arana et al. 2003).

Acknowledgements

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References

We are grateful to the following sources of funding: CONABIO FB666/S019/99, CONABIO FB789/AA004/02, CONABIO DJ007/26 and CONACYT SEP-2003-C02-42550. This work was carried out with the permission of SAGARPA (Permit numbers: DGOPA.02476.220306.0985 and DGOPA.06648.140807.3121). We thank Clara Ramírez Jáuregui, Pedro Allende and Victoria Montes for help with the literature and images, German Ramírez Reséndiz and Carlos Suárez for their computer assistance, and Jose Salgado, Arturo Nuñez, Alberto Castro, Juan Toto and Sergio Rendón for their assistance in the field samplings. Our sincere gratitude to the anonymous reviewers who helped to improve the manuscript.

References

  1. Top of page
  2. Abstract
  3. Problem
  4. Material and Methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  • Bakus G.J., Green K.D. (1987) The distribution of marine sponges collected from the 1976–1978 Bureau of Land Management Southern California Bight Program. Bulletin of the Southern California Academy of Science, 86, 5788.
  • Barnes D.K.A. (1999) High diversity of tropical intertidal zone sponges in temperature, salinity and current extremes. African Journal of Ecology, 37, 424434.
  • Barry J.P., Baxter C.H., Sagarin R.D., Gilman S.E. (1995) Climate-related, long-term faunal changes in a California rocky intertidal community. Science, 267, 672675.
  • Bavestrello G., Cattaneo-Vietti R., Danovaro R., Fabiano M. (1991) Detritus rolling down a vertical cliff of the Ligurian Sea (Italy): the ecological role in hard bottom communities. PSZN: Marine Ecology, 12, 281292.
  • Bell J.J. (2004) Evidence for morphology-induced sediment settlement prevention on the tubular sponge Haliclona urceolus. Marine Biology, 146, 2938.
  • Bell J.J., Barnes D.K.A. (2000a) A sponge diversity centre within a marine “island”. Hydrobiologia, 440, 5564.
  • Bell J.J., Barnes D.K.A. (2000b) The distribution and prevalence of sponges in relation to environmental gradients within a temperate sea lough: inclined cliff surfaces. Diversity and Distributions, 6, 305323.
  • Bell J.J., Smith D. (2004) Ecology of sponge assemblages (Porifera) in the Wakatobi region, south-east Sulawesi, Indonesia: richness and abundance. Journal of Marine Biological Association of the United Kingdom, 84, 581591.
  • Bell J.J., Barnes D.K.A., Shaw C. (2002) Branching dynamics of two species of arborescent demosponge: the effect of flow regime and bathymetry. Journal of Marine Biological Association of the United Kingdom, 82, 279294.
  • Bell J.J., Burton M., Bullimore B., Newman P.B., Lock K. (2006) Morphological monitoring of subtidal sponge assemblages. Marine Ecology Progress Series, 311, 7991.
  • Bergquist P.R. (1978) Sponges. Hutchinson, London: 268 pp.
  • Bingham B.L., Young C.M. (1995) Stochastic events and dynamics of a Mangrove root epifaunal community. PSZN: Marine Ecology, 16, 145163.
  • Buchanan J.B., Moore J.J. (1986) Long-term studies at a benthic station off the coast of Northumberland. Hydrobiologia, 142, 121127.
  • Buchanan J.B., Kingston P.F., Sheader M. (1974) Long-term population trends of the benthic macrofauna in the off-shore mud of the Northumberland coast. Journal of the Marine Biological Association of the United Kingdom, 54, 785795.
  • Burton M. (1949) Observations on littoral sponges, including the supposed swarming of larvae, movement and coalescence in mature individuals, longevity and death. Proceedings of the Zoological Society of London, 118, 893915.
  • Carballo J.L. (2006) Effect of natural sedimentation on the structure of tropical rocky sponge assemblages. Ecoscience, 13, 119130.
  • Carballo J.L., Naranjo S. (2002) Environmental assessment of a large industrial marine complex based on a community of benthic filter-feeders. Marine Pollution Bulletin, 44, 605610.
  • Carballo J.L., Nava H. (2007) Comparison of sponge assemblage patterns between two adjacent tropical rocky habitats (tropical Pacific Ocean, México). Ecoscience, 14, 92102.
  • Carballo J.L., Olabarria C., Garza-Osuna T. (2002) Analysis of four macroalgal assemblages along the Pacific Mexican coast during and after 1997–1998 El Niño. Ecosystem, 5, 749760.
  • Corriero G., Longo C., Mercurio M., Marchini A., Occhipinti-Ambrogi A. (2007) Porifera and Bryozoa on artificial hard bottoms in the Venice Lagoon: spatial distribution and temporal changes in the northern basin. Italian Journal of Zoology, 74, 2129.
  • Dayton P.K. (1984) Processes structuring some marine communities: are they general? In: StrongD.R., SimberloffD., AbeleL.G., ThistleA.B. (Eds), Ecological Communities: Conceptual Issues and the Evidence. Princeton University Press, Princeton, NJ: 181197.
  • Dayton P.K. (1989) Interdecadal variation in an Antarctic sponge and its predators from oceanographic shifts. Science, 245, 14841486.
  • Diaz M.C., Alvarez B., Laughlin R.A. (1990) The sponge fauna on a fringing coral reef in Venezuela, II: Community structure. In: RützlerK. (Ed.), New Perspectives in Sponge Biology. Smithsonian Institution Press, Washington, D.C.: 367375.
  • Dye A.H. (1998) Community-level analyses of long-term changes in rocky littoral fauna from South Africa. Marine Ecology Progress Series, 164, 4757.
  • Ellison A.M., Farnsworth E.J. (1992) The ecology of Belizean mangrove-root fouling communities: patterns of epibiont distribution and abundance, and effects on root growth. Hydrobiologia, 247, 8798.
  • Ellwood M., Kelly M. (2003) Sponge “tree rings”: new indicators of ocean variability? Water & Atmosphere, 11, 2527.
  • Elvin D.W. (1976) Seasonal growth and reproduction of an intertidal sponge, Haliclona permollis (Bowerbank). Biological Bulletin, 151, 108125.
  • Emerson C.W. (1989) Wind stress limitation of benthic secondary production in shallow, soft-sediment communities. Marine Ecology Progress Series, 53, 6577.
  • Engledow H.R., Bolton J.J. (1994) Seaweed α-diversity within the lower eulittoral zone in Namibia: the effects of wave action, sand inundation, mussels and limpets. Botanica Marina, 37, 267276.
  • Estacio F., Adiego E., Carballo J.L., Sánchez-Moyano E., Izquierdo J.J., García-Gómez J.C. (1999) Interpreting temporal disturbances in an estuarine benthic community under combined anthropogenic and climatic effects. Journal of Coastal Research, 15, 155167.
  • Field J.G., Clarke K.R., Warwick M. (1982) A practical strategy for analyzing multispecies distribution patterns. Marine Ecology Progress Series, 8, 3752.
  • Fromentin J.M., Ibañez F., Dauvin J.C., Dewarumez J.M., Elkaim B. (1997) Long-term changes of four macrobenthic assemblages from 1978 to 1992. Journal of the Marine Biological Association of the United Kingdom, 77, 287310.
  • Fromont J., Garson M. (1999) Sponge bleaching on the West and East coast of Australia. Coral Reefs, 18, 340.
  • Garrabou J., Zabala M. (2001) Growth dynamics in four Mediterranean demosponges. Estuarine, Coastal and Shelf Science, 52, 293303.
  • Gido K.B., Schaefer J.F., Pigg J. (2004) Patterns of fish invasions in the Great Plains of North America. Biological Conservation, 118, 121131.
  • Glynn P.W. (1976) Some physical and biological determinants of coral community structure in the eastern Pacific. Ecological Monographs, 46, 431456.
  • Gray J.S., Christie H. (1983) Predicting long-term changes in marine benthic communities. Marine Ecology Progress Series, 13, 8794.
  • Harrison D.E., Vecchi G.A. (2001) El Niño and La Niña equatorial Pacific thermocline depth and sea surface temperature anomalies, 1986–1998. Geophysical Research Letters, 28, 10511054.
  • Hartnoll R.G., Hawkins S.J. (1980) Monitoring rocky-shore communities: a critical look at spatial and temporal variation. Helgoland Marine Research, 33, 484494.
  • Hernández-Arana H.A., Rowden A.A., Attrill M.J., Warwick R.M., Gould-Bouchot G. (2003) Large-scale environmental influences on the benthic macroinfauna of the southern Gulf of Mexico. Estuarine, Coastal and Shelf Science, 58, 825841.
  • Hughes T.P. (1996) Demographic approaches to community dynamics: a coral reef example. Ecology, 77(7), 22562260.
  • Inman D.L. (1952) Measures of describing the size distribution of sediments. Journal of Sedimentology and Petrology, 22(3), 125145.
  • Kröncke I., Dippner J.W., Heyen H., Zeiss B. (1998) Long-term changes in macrofaunal communities off Norderney (East Frisia, Germany) in relation to climate variability. Marine Ecology Progress Series, 167, 2536.
  • Levin S.A. (1992) The problem of pattern and scale in ecology. Ecology, 73, 19431967.
  • Lieberman M., John D.M., Lieberman D. (1979) Ecology of subtidal algae on seasonally devastated cobble substrate off Ghana. Ecology, 60(6), 11511161.
  • Likens G.E. (1989) Long-Term Studies in Ecology. Springer-Verlag, New York, NY: 214 pp.
  • Limpus C.J., Nicholls N. (1988) The Southern Oscillation regulates the annual numbers of green turtles (Chelonia mydas) breeding around northern Australia. Australian Journal of Wildlife Research, 15, 157161.
  • Logan B.W., Harding J.L., Ahr W.M., Willians J.D., Snead R.G. (1969) Carbonate sediments and reefs. Yucatan shelf, Mexico. American Association of Petroleum Geologist Memoir, 11, 1198.
  • Lundälv T., Christie H. (1986) Comparative trends and ecological patterns of rocky subtidal communities in the Swedish and Norwegian Skagerrak area. Hydrobiologia, 142, 7180.
  • Lundälv T., Larsson C., Lennart A. (2004) Long-term trends in algal-dominated rocky subtidal communities on the Swedish west coast – a transitional system? Hydrobiologia, 142, 8195.
  • Marquez E. (2006) Analyse de series photographiques à long terme: étude des traits de vie d’espèces longéves des communautés du coralligène de Méditerranée. Master thesis. Université de la Méditerranée (Aix-Marseille II). Centre d’Océanologie de Marseille: 28 pp.
  • McClanahan T.R. (1988) Seasonality in East Africa’s coastal waters. Marine Ecology Progress Series, 44, 191199.
  • McClanahan T.R. (1990) Kenyan coral reef-associated gastropod assemblages: distribution and diversity patterns. Coral Reefs, 9, 6374.
  • McClanahan T.R., Mutere J.C. (1994) Coral and sea urchin assemblage structure and interrelationships in Kenya reef lagoons. Hydrobiologia, 286, 109124.
  • Menge B.A., Olson A.M. (1990) Role of scale and environmental factors in regulation of community structure. Trends in Ecology and Evolution, 5, 5257.
  • Muricy G. (1991) Structure des peuplements de spongiaires autour de l’égout de Cortiou (Marseille, France). Vie Milieu, 41, 205221.
  • Oliver J.S., Slattery P.N., Hulberg L.W., Nybakken J.W. (1980) Relationships between wave disturbance and zonation of benthic invertebrates communities along a subtidal high energy beach in Monterrey bay, California. Fishery Bulletin, 78, 437454.
  • Ott J.A. (1981) Adaptative strategies at the ecosystem level: examples from two benthic marine systems. PSZN: Marine Ecology, 2, 97180.
  • Paine R.T. (1986) Benthic community-water column coupling during the 1982–1983 El Niño. Are community changes at high latitudes attributable to cause or coincidence? Limnology and Oceanography, 31, 351360.
  • Pansini M., Pronzato R. (1990) Observations on the dynamics of a Mediterranean sponge community. In: RützlerK. (Ed.), New Perspectives in Sponge Biology. Smithsonian Institution Press, Washington, D.C.: 404415.
  • Patel B., Balani M.C., Patel S. (1985) Sponge “sentinel” of heavy metals. Science of the Total Environment, 41, 143152.
  • Pearson T.H., Duncan G., Nuttall J. (1986) Long term changes in the benthic communities of Loch Linnhe and Loch Eil (Scotland). Hydrobiologia, 142, 113119.
  • Peraza V.R. (1985) Transporte Litoral de Arenas en Playas de la Costa Sur del Estado de Sinaloa. MSc thesis, Universidad Autónoma de México, México DF, 57 pp.
  • Pérez T., Sarrazin L., Rebouillon P., Vacelet J. (2002) First evidences of surfactant biodegradation by marine sponges (Porifera): an experimental study with a linear alkylbenzenesulfonate. Hydrobiology, 489, 225233.
  • Philander S.G.H. (1990) El Niño, La Niña and the Southern Oscillation. Academic Press, San Diego: 289 pp.
  • Pronzato R., Manconi R. (1995) Long-term dynamics of a freshwater sponge population. Freshwater Biology, 33, 485495.
  • Reiswig H.M. (1971) Particle feeding in natural populations of three marine demosponges. Biological Bulletin, 141, 568591.
  • Ricciardi A., Bourget E. (1999) Global patterns of macroinvertebrate biomass in marine intertidal communities. Marine Ecology Progress Series, 185, 2135.
  • Salani S., Lotufo T.M.D.C., Hajdu E. (2006) Sigmaxinella cearense sp. nov. from sandstone reefs off Fortaleza (Ceará State, Brazil) (Desmacellidae, Mycalina, Poecilosclerida, Demospongiae). Zootaxa, 1369, 4353.
  • Salas de Leon D.A., Monreal-Gomez M.A., Ramirez J.A. (1992) Periodos caracteristicos en las oscilaciones de parámetros meteorológicos en Cayo Arcos, Mexico. Atmosfera, 5, 193205.
  • Sarà M. (1966) Studio quantitativo della distribuzione dei Poriferi in ambienti superficiali della Riviera Ligure di Levante. Archivio Oceanografico e Limnologico, 14(3), 365386.
  • Sarà M. (1970) Competition and cooperation in sponge population. In: FryW.G. (Ed.), The Biology of the Porifera. Academic Press, London: 273284.
  • Sarà M., Vacelet J. (1973) Écologie des Démosponges. In: GrasséP.P. (Ed.), Traité de Zoologie, Anatomie, Systematique, Biologie. Masson et Cie., Paris: 463516.
  • Silva B.B.da., Alves J.J.A., Cavalcanti E.P., Dantas R.T. (2002) Wind energy potential for the prevailing direction in Northeast Brazil. Revista Brasileira de Engenharia Agricola e Ambiental, 6, 431439.
  • Southward A.J., Hawkins S.J., Burrows M.T. (1995) Seventy years’ observations of changes in distribution and abundance of zooplankton and intertidal organisms in the western English Channel in relation to rising sea temperature. Journal of Thermal Biology, 20, 127155.
  • Steller D.L., Riosmena-Rodríguez R., Foster M.S., Roberts C.A. (2003) Rhodolith bed diversity in the Gulf of California: the importance of rhodolith structure and consequences of disturbance. Aquatic Conservation: Marine and Freshwater Ecosystems, 13, 520.
  • Stone A.R. (1970) Growth and reproduction of Hymeniacidon perleve (Montagu) in Langstone Harbour, Hampshire. Journal of Zoology, 161, 443459.
  • Sutherland J.P. (1980) Dynamics of the epibenthic community on roots of the mangrove Rhizophora mangle, at Bahia de Buche, Venezuela. Marine Biology, 58, 7584.
  • Underwood A.J. (1999) History and recruitment in structure of intertidal assemblages on rocky shores: an introduction to problems for interpretation of natural change. In: WhitfieldM., MatthewsJ., ReynoldsC. (Eds), Aquatic Life Cycle Strategies: Survival in a Variable Environment. Institute of Biology, London: 7996.
  • Vicente P. (1989) Regional commercial sponge extinctions in the West Indies: are recent climatic changes responsible? PSZN: Marine Ecology, 10(2), 179191.
  • Vicente P. (1990) Response of sponges with autotrophic endosymbionts during the coral-bleaching episode in Puerto Rico. Coral Reefs, 8, 199202.
  • Wilkinson C.R., Cheshire A.C. (1989) Patterns in the distribution of sponge populations across the central Great Barrier Reef. Coral Reefs, 8, 127134.
  • Wolter K. (1987) The Southern Oscillation in surface circulation and climate over the tropical Atlantic, Eastern Pacific, and Indian Oceans as captured by cluster analysis. Journal of Climate and Applied Meteorology, 26, 540558.
  • Wolter K., Timlin M.S. (1993) Monitoring ENSO in COADS with a seasonally adjusted principal component index. Proceedings of the 17th Climate Diagnostics Workshop, Norman, OK, NOAA/N MC/CAC, NSSL, Oklahoma Clim. Survey, CIMMS and the School of Meteorology, University of Oklahoma: 5257.
  • Wolter K., Timlin M.S. (1998) Measuring the strength of ENSO – how does 1997/98 rank? Weather, 53, 315324.
  • Wulff J.L. (1990) Patterns and processes of size change in Caribbean demosponges of branching morphology. In: RützlerK. (Ed.), New Perspectives in Sponge Biology. Smithsonian Institution Press, Washington, D.C.: 425435.
  • Wulff J.L. (1995) Effects of a hurricane on survival and orientation of large erect coral reef sponges. Coral Reefs, 14, 5561.
  • Wulff J.L. (2001) Assessing and monitoring coral reef sponges: why and how? Bulletin of Marine Science, 69(2), 831846.
  • Wulff J. (2005) Trade-offs in resistance to competitors and predators, and their effects on the diversity of tropical marine sponges. Journal of Animal Ecology, 74, 313321.
  • Wulff J.L. (2006) Rapid diversity and abundance decline in a Caribbean coral reef sponge community. Biological Conservation, 127, 167176.
  • Zea S. (1994) Patterns of coral and sponge abundance in stressed coral reefs at Santa Marta, Colombia Caribbean. In: Van SoestR.W.M., Van KempenT.M.G., BraekmanJ.C. (Eds), Sponges in Time and Space: Biology, Chemistry, Paleontology. Balkema, Rotterdam: 257264.