Assessment, monitoring and modelling of the abundance of Dunaliella salina Teod in the Meighan wetland, Iran using decision tree model.

The microalga Dunaliella salina has been broadly studied for different purposes such as beta-carotene production, toxicity assessment and salinity tolerance, yet research on the habitat suitability of this alga has rarely been reported. The present research aims to apply a suitable monitoring and modelling methods (two critical steps in ecological researches) to predict the abundance of D. salina. The abundance of D. salina was predicted by decision tree model (J48 algorithm) in 10 different monitoring sites during 1-year study period (2016-2017) in the Meighan wetland, one of the valuable hypersaline wetlands in Iran. The abundance of alga (as output of model) together with various water quality and physical-habitat wetland characteristics (as inputs of model) were monthly and repeatedly monitored in two different depths (one from the surface layer and another one from the depth of maximum 50 cm) which in total resulted in 240 instances (120 instances for each depth). Based on trial and error, a sevenfold cross-validation resulted in the highest predictive performances of the model (CCI > 75% and Cohen's Kappa > 0.65). According to the model's prediction, the number of sunny hours might be one of the most important driving parameters to meet the habitat requirements of alga in the hypersaline wetland. Model also predicted that an increase in dissolved oxygen and sodium concentrations might increase the abundance of D. salina in the salt wetland. In contrast, an increase in total suspended solids concentration and monthly precipitation might lead to a decrease in the abundance of alga. Chi-square test of independence showed a significant difference between the abundance of the D. salina and spatio-temporal patterns in the wetland (Pearson chi-square statistic = 221.7, p = 0.001) so warm seasons (spring and summer) had more contribution to the sampling of the species than cold seasons (autumn and winter). The difference in the abundance of the species in different sampling sites can be attributed due to the various anthropogenic activities.