Habitat suitability modelling of coralligenous in the Northern Adriatic Sea and potential distribution under climate change scenario

Climate change is deeply affecting oceanographic, biogeochemical, and hydrological processes and consequently influencing the ecological patterns of ecosystems. Indeed, marine habitats and species are facing many alterations in their structure, functioning, and in their capacity of providing ecosystem services. To investigate and explore the distribution and potential variation of habitats and species under future climate change scenarios, Habitat Suitability Models (HSMs) have been widely applied during the last years for their recognized ability in predicting the suitability of a location for species and habitats in correlation with the environmental conditions.

With the application of two of the best-known HSMs (Random Forest and MaxEnt), this research intends to investigate the distribution of the coralligenous, a widespread habitat in the Northern Adriatic Sea threatened by the effects of climate change, and identify its potential variation in a severe future scenario. The analysis consisted in examining the correlation between the habitat distribution with environmental parameters obtained from online databases and a set of dedicated ocean model simulations applied in recent past conditions and under RCP 8.5 climate change scenario. Furthermore, to explore the potential uncertainty of the environmental variables in future conditions, a sensitivity analysis has been implemented by running additional HSMs simulations set up with variables' increments and decrements resulting from projections modeled by other research.

The models perform very well in predicting habitat distributions. The prediction under the climate change scenario shows that opportunistic species (e.g. turf-dominant algae) find more suitable conditions in the area being more tolerant to stressful conditions and alterations of the environmental variables. As a result, calcareous macroalgae appear to be more vulnerable to climate change effects, including increases in temperature, nutrient concentrations, salinity, and velocity. Overall, the results of the sensitivity analysis confirmed the results predicted by models; however, Random Forest also shows a higher sensitivity to uncertainty than MaxEnt.

In conclusion, this study gives a sight of the likely ecological behavior in correlation with past environmental conditions and future alterations due to climate change. Besides, HSMs confirm to be very useful tools to develop adequate conservation strategies and/or identify priority areas to protect. Thanks to the sensitivity analysis, additional hints about the models’ behavior according to the environmental uncertainties are extrapolated, allowing to consider with consciousness the results and understanding of the potentialities of the models according to the data in possession.