As the utilization of rangelands in Iran has exceeded 2.2 times its carrying capacity, increasing numbers of livestock, especially in drought years, have dramatic effects on rangelands. In this regard, the prediction of forage production is an important management strategy to mitigate the consequences of drought. In Plour and Saveh Range Experimental Sites, climatic factors such as precipitation, temperature (mean, maximum and minimum), evapotranspiration, standardized precipitation index, reconnaissance drought index, and/or a combination of these factors were used to construct a predictive model of forage production. For each climatic variable and/or index, 33 time periods of 1–4, 6 and 9-months were specified. We used principal components analysis, stepwise regression and best subset regression, to reduce the number of variables and then the appropriate time periods selected. To select a model, assessment statistics of correlation coefficient, mean of bias error, root mean of square error, mean of absolute relative error and ideal point error were used. Finally models derived from combined climatic factors and drought indices were selected for the prediction of forage production at both study areas. In the arid region of Saveh, production is more effected by precipitation, while in the humid and colder region of Plour, minimum temperature had more effects on plant growth. In Saveh, due to a lack of rainfall in February and minimum temperatures in March, production has direct and indirect relationships with drought and with maximum temperatures, respectively. In the Plour, production has a direct relationship with the maximum temperatures of March–June and drought of February–March and has an indirect relationship with the maximum temperatures of May–June.

Keywords