Antonio Ortiz’s research while affiliated with University of Alicante and other places

Jellyfish blooms cause important ecological and socio-economic problems. Among jellyfish, cubozoans are infamous for their painful, sometimes deadly, stings and are a major public concern in tropical to subtropical areas; however, there is little information about the possible causes of their outbreaks. After a bloom of the cubomedusa Carybdea marsupialis (Carybdeidae) along the coast of Denia (SW Mediterranean, Spain) in 2008 with negative consequences for local tourism, the necessity to understand the ecological restrictions on medusae abundance was evident. Here we use different models (GAM and zero-inflated models) to understand the environmental and human related factors influencing the abundance and distribution of C. marsupialis along the coast of Denia. Selected variables differed among medusae size classes, showing different environmental restriction associated to the developmental stages of the species. Variables implicated with dispersion (e.g. wind and current) affected mostly small and medium size classes. Sea surface temperature, salinity and proxies of primary production (chl a, phosphates, nitrates) were related to the abundances of small and large size classes, highlighting the roles of springtime salinity changes and increased primary production that may promote and maintain high densities of this species. The increased primary (and secondary) production due to anthropogenic impact is implicated as the factor enabling high numbers of C. marsupialis to thrive. Recommendations for monitoring blooms of this species along the study area and applicable to Mediterranean Sea include focus effort in coastal waters where productivity have been enriched by anthropogenic activities.

Collinearity analysis for all environmental variables. Lower-left panels show scatter plots with a loess smoothing line (in red) to help visualize patterns. Central (Diagonal) panels show histograms of the environmental variables and upper-right panels show the values of the Pearson’s correlation coefficients for each pair of variables, with text size proportional to its value. (TIFF)

Descriptive statistics for the number and density (in parentheses; indiv. m-3) of three sizes (Small, Medium, Large) of Carybdea marsupialis medusae captured by site (Almadrava ‘AL’, Molins ‘MO’, Blay beach ‘BB’, Rasset ‘RA’, Marineta Cassiana ‘MC’ and Les Rotes ‘RO’) along the coast of Denia. Minimum 'Min', maximum 'Max', sum 'Sum', mean 'Mean' and standard deviation 'SD' values are showed for each class size and sampling site. (DOC)

Optimal models showing the partial effect on abundance of the significant environmental variables for a) small, b) medium and c) large Carybdea marsupialis medusae collected along the coast of Denia. Central (bold) line show the best fit and the shaded areas show the 95% confidence intervals of the GAM-NB model. Bottom vertical lines represent observations. (TIFF)

Mathematical model formulation for the generalized additive models with Poisson family distribution (GAM-P), with negative binomial family distribution (GAM-NB), for the zero-inflated Poisson (ZI-P) and zero-inflated negative binomial (ZI-NB) models (source Zuur et al. 2009). (DOC)

Descriptive statistics for the selected environmental variables from the beginning of July 2010 to the end of July 2011for all sites surveyed (Almadrava ‘AL’, Molins ‘MO’, Blay beach ‘BB’, Rasset ‘RA’, Marineta Cassiana ‘MC’ and Les Rotes ‘RO’) on the coast of Denia, Spain. Values are shown by season. (DOC)

Model comparison for the generalized additive models with a Poisson family distribution (GAM-P), generalized additive models with a negative binomial family distribution (GAM-NB), zero-inflated Poisson (ZI-P) and zero-inflated negative binomial (ZI-NB) models. The Pearson correlation coefficient (r), Spearman rank correlation (p), intercept (α) and slope (b) of a linear regression between the observed versus fitted values, Akaike Information Criterion (AIC), log likelihood (Log lik) and degrees of freedom (Df) are showed for each size class of Carybdea marsupialis medusae. NA = Not Available. (DOC)

Carybdea marsupialis at the sampling site shows patchy distribution (hundreds of meters) and high densities. We find that low salinity is positively correlated to C. marsupialis distribution as showed by other authors (Gordon et al 2004; Cutress and Studebaker, 1973). Low salinity in the study area was not due to any river runoff but to sub-surface aquifer discharge (~6 106m3·y-1, and about 120-170mg NO3=·l-1), and seems that the response of the cubomedusae to low salinity is the same independently of the low salinity origin. We also detect a positive correlation with low visibility and sunny days. Sea bottom had no any influence on C. marsupialis distribution, in contrast to other authors that stated that this species was associated with sandy bottoms (Di Camilo et al 2006).

Jellyfish blooms have demonstrated to be an important driving force of changes in local biodiversity as well as change in ecosystems due to their interactions with other species or with ecosystem services. Recently, high numbers of the cubomedusae Carybdea marsupialis (Carybdeidae) have been found in the coast of Denia (SW Mediterranean, Spain). Some negative interactions with tourist and local people due to their painful sting have been found and ecological consequences probably occur because of the high density found in those beaches. Due to the geographic specificity of this phenomenon we considered the niche modeling approach, as a tool to elucidate which environmental restrictions influence their presence in the area. We used BIOMOD R-Package which also allows taking into consideration some possible spread ranges of this cubodemusae species. Variables as sea temperature and salinity, current direction and percentage of sandy bottom, were the most important variables which explained its distribution. Differences on the environmental restrictions between C. marsupialis juveniles (less than 5mm) and adult (more than 2 cm) organism were found, reflecting their particular role on the ecosystem. Different models tested gave similar outputs on spread predictions to geographically neighbor areas.