Adequação do teste de tetrazólio e fenotipagem computacional da qualidade fisiológica de sementes de Amburana cearensis A.C. Smith (Fabaceae)

Abstract

The production of seeds of native forest species is widely used in obtaining seedlings, for forest plantations and for the restoration of degraded areas, since the analysis of physiological quality and of factors that may affect the germination power and vigor of seeds is fundamental for obtaining high quality seedlings for recovery projects and conservation, in situ and ex situ, of the genetic resources. Among the procedures used, the tetrazolium test stands out for its speed and amount of information generated, with precise diagnosis of the main causes of quality reduction. However, a problem inherent in the test is the subjectivity of the analyst in the proper interpretation of the results. Thus, the objective with this work was to establish the appropriate procedure to evaluate the quality of Amburana cearensis (Allemão) A.C. Smith (Fabaceae) seeds by the tetrazolium test; to establish a low cost methodology for the evaluation of the test and classification of seed viability and vigor through the automation of image analysis and compare with traditionally used tests. The first part involved the adequacy of tetrazolium, application of the methodology in three seed lots (I, II and III) and comparison with physical and physiological quality tests. For this, two pre-preparation methods (scarified and non-scarified seeds) and three soaking temperatures (30, 35 and 40 degrees Celsius) were tested, and evaluated in a double factorial scheme (2x3). The seeds were exposed to tetrazolium solution (0.05%) for 2h30 hours at 30 °C. It was proceeded the visual analysis, classifying seeds into four classes of viability and vigor. As for the image analysis, were collected, from the images of each repetition, different shades of red and brown, later used in the segmentation of seeds as the percentage of dead tissue and medium tones of red from living tissue for the classification of viability and vigor by RNA's. It was proceeded the analysis of variance, mean test (Tukey = 0.05%) and the parameters were correlated. For the tetrazolium standardization, the method of preparation using scarified seeds stood out, allowing better removal of the integument. The temperatures tested did not differ statistically, however it was observed greater unviability of the embryos at 40 ° C. Lot I stood out for the tetrazolium viability and germination vigor tests. There was a significant positive correlation between Tz viability and Tz vigor in relation to physical and physiological quality parameters. The evaluation by image analysis could quantify the percentages of dead tissue (PM1 to PM5) and living tissue or the medium tones of red (V1 to V5), present in each seed, according to its location and to classify them, individually, in vigorous, viable and unviable with 100% accuracy. When correlated with germination test parameters, seed vigor by image classification positively correlated, in a significantly way, with germination percentage, Emergency Speed Index (IVE), Length and dry mass of epicotyl (CME and MSE) and hypocotyl (CMH and MSH). From the correlation estimates it was observed that the highest germination (%), IVE, CME and CMH are related to the higher percentage of living tissue, or tones of red, in the vital regions of the embryo (V1 and V2). Dry mass, however, is related to lower percentages of dead tissue in all regions of the seed. Seeds scarified and soaked in water for 24 hours at 30 °C, with subsequent tegument removal and exposure in tetrazolium solution at 0.05% / 2: 30 hours at 30 °C, is efficient in evaluating the viability of seeds of A cearensis. Computational image analysis using Artificial Neural Networks is an effective alternative to evaluate the tetrazolium test for the physiological quality of A. cearensis seeds.