Prediction of Dorema ammoniacum density in degraded rangelands with using Neural Network

The Dorema ammoniacum of Apiaceae family is a medicinal, and forage species that wide range of its habitats have been destroying since last 4 decades. At southwest of Sabzevar, was conducted a study for predicting the Kandal density in degraded lands by neural network model in 2014. Independent variables consist of (soil texture, EC, pH, SAR, N, P, K, cations, organic matter and lime) and the dependent variable was Kandal density. After drawing the rangeland border on topographical map and satellite photo, we collected 70 systematic-random samples consisted of independent variables and the dependent variable from Kandal habitat. Of this number, 50 samples for training, 10 samples for cross validation and 10 samples for testing were assigned. Then 8 samples were collected from the degraded rangeland and they input to the model. The results showed that Kandal plant has no even relation with all of environmental agents, and Plant density correlated with EC (-93%), potassium (95%), cations (-92%), SAR (-79%), soil texture (80%), and alkalinity (4%). the least Kandal density with 0.12/m^2 belonged to degraded rangelands that they plowed over the past 45 years, and the most plant density with 0.23/m^2 belonged to degraded rangelands that they have plowed since 3 years ago too. The prediction of the Kandal density showed that the more intense destruction rangeland is increased, the more dispersal will be created in soil agents, and this gradually resulted in migration and extinction of kanal in the future.