The paper presents a study on the impact of weather factors on yielding of yellow lupin (Juno, Parys) in 1991-2008 and yields forecasts for 2050-2060 with the use of weather-yield models and three climate change scenarios.

The source material was the results of cultivar experiments and meteorological observations obtained from the COBORU research stations for cultivar testing and evaluation situated in Eastern Poland (Głodowo, Marianowo, Sulejów, Uhnin). The analyses included the correlation and multiple progressive stepwise regression methods. The fitting of the models was evaluated with the coefficient of determination R², adjusted coefficient of determination R²_adj, standard estimation error and coefficient of determination R²_pred calculated with the Cross Validation procedure. The significance of regression equations was determined with the Snedecor F-test. The selected equations were used to predict  lupine yielding under conditions of double CO₂ content in the atmosphere (years: 2050-2060). To this end, 100 potential annual daily courses were generated with the WGENK model for each station for the following meteorological factors: total radiation, maximum temperature, minimum temperature and precipitation assuming three climate change scenarios for Central Europe: GISS model E, HadCM3 and GFDL.

The impact of meteorological factors on the yielding of yellow lupin was diversified depending on the station location and on the cultivar. Most frequently, the yields of both species were significantly and negatively influenced by the maximum temperature during flowering, to a lesser extent by the average temperature. Yielding of yellow lupin also depended on the amount of rainfall in the period from germination to the beginning of flowering. The estimated sums of total radiation had a great impact on the yield of yellow lupin (especially the Juno variety); also significant were the sums calculated for the periods germination-beginning of flowering and flowering.

The strongest correlation between yield and meteorological factors was obtained for the Marianowo station; coefficients of determination R²_pred of the models created for the majority of the studied varieties (Juno, Parys) exceeded the value of 0.70.

The studies demonstrated that the predicted climatic changes would exert a beneficial impact on lupin yielding. Simulated yields were higher than those observed in 1991-2008, with the scenario HadCM3 being the most beneficial.