The inoculation of seeds with nitrogen-fixing and plant-growth promoting bacteria is a well-established agricultural practice that has been increasingly adopted worldwide, decreasing costs and environmental impacts of food production. Most of the globally commercialized inoculants are for the soybean crop, and a method for recovery of Bradyrhizobium cells from inoculated soybean seeds for subsequent counting has been adopted by several laboratories of South America, especially to investigate the bacterial survival on seeds treated with pesticides. However, the use of inoculants containing Azospirillum brasilense in cereal crops has exponentially increased, requiring investigation about the recovery and counting of cells from inoculated seeds. We first verified that the method used for recovery and counting of viable cells of Bradyrhizobium from soybean seeds was not applicable for maize seeds inoculated with A. brasilense. We then modified several steps of the method, aiming at succeeding in recovering Azospirillum viable cells. The main limitation was identified in the nature of the seed tegument, dry and poor in nutrients, resulting in A. brasilense cell aggregation. Pre-hydration of seeds for 2 h in sterile distilled water, followed by shaking for 30 min in sterile distilled water with Tween 80 allowed proper counting of A. brasilense cells recovered from maize seeds. The method was successfully applied to count Azospirillum cells recovered from pre-inoculated maize seeds, and to estimate the impact of seed treatment with pesticides on cell survival.

An adaptability and stability study was carried out using soybean yield data from several locations in Paraná State, obtained from 1990 to 1999. The main objectives were: a) to check the efficiency of the Embrapa Soja breeding program for selecting the highest yielding lines with specific (regional) or broad adaptation; b) to analyze the performance of the control cultivars under favorable and unfavorable conditions; c) to identify the best stability methodology for inbred line evaluation. The evaluated lines were classified into early (L), semiearly (M) and medium (N) maturity groups. A randomized complete block design with four replications was used in each location and all analyses were carried out by maturity group. The number of M maturity lines that scored higher than the controls BR-16 and Embrapa 4 increased along the years, especially in unfavorable environments. On the other hand, the number of L and N maturity groups scoring higher than the controls either remained constant or fluctuated (increasing or decreasing) with time. The controls FT-Guaíra, Embrapa 4, BR-16 and FT-10 and the great majority of the tested lines showed broad adaptation. IAS 5 and FT-Abyara showed adaptation to poor environments and also broad adaptation. The analysis of the means in favorable and unfavorable environments was the most efficient method for soybean line evaluation. A methodology to classify soybean genotypes was proposed, based on these means.