Due to inadequate efforts to reinforce nitrogen fixation capability of bean via symbiosis with rhizobia, improvement of bean productivity is still highly dependent on chemical fertilization. An advanced understanding of agro-ecosystem-bean-Rhizobium interaction is required to improve symbiosis efficiency. Thus, seasonal development of rhizobial nodulation was characterized according to 20 agro-ecological properties for 122 commercial bean fields. Principal component analysis identified soil texture as a major descriptor of agrosystem-bean-disease-Rhizobium interaction. Nonparametric correlation analysis indicated significant associations of root nodulation with bean class, fungicidal treatment of seed and soil, Fusarium root rot index, planting date and depth, soil texture, clay and sand content. Ordinal regression analysis demonstrated that rhizobial nodulation was improved by applying initial drought, heavier soil textures with greater organic matter and neutral pH, using herbicides and manure, growing white beans, irrigating every 7–9 days, later sowing in June, reducing disease and weed, shallower seeding, sowing beans after alfalfa, avoiding fungicidal treatment of seed and soil, and omitting urea application. This largescale study provided novel information on a comprehensive number of agronomic practices as potential tools for improving bean-Rhizobium symbiosis for sustainable legume production systems.
From 2009 to 2018, a total of 80 wheat crops were studied at plot and regional scales to predict stripe rust epidemics based on influential climatic indicators in Kermanshah province, Iran. Disease onset time and epidemic intensity varied spatially and temporarily. The disease epidemic variable was classified as having experienced nonepidemic, moderate or severe epidemics to be used for statistical analysis. Principal component analysis (PCA) was used to identify climatic variables associated with occurrence and intensity of stripe rust epidemics. Two principal factors accounting for 70% of the total variance indicated association of stripe rust epidemic occurrence with the number of icy days with minimum temperatures below 0°C (for subtropical regions) and below −10°C (for cool temperate and semi-arid regions). Disease epidemic intensity was linked to the number of rainy days, the number of days with minimum temperatures within the range of 7−8°C and relative humidity (RH) above 60%, and the number of periods involving consecutive days with minimum temperature within the range of 6−9°C and RH% > 60% during a 240-day period, from September 23 to May 21. Among mean monthly minimum temperatures and maximum relative humidity examined, mean maximum relative humidity for Aban (from October 23 to November 21) and mean minimum temperature for Esfand (from February 20 to March 20) indicated higher contributions to stripe rust epidemic development. Confirming PCA results, a multivariate logit ordinal model was developed to predict severe disease epidemics. The findings of this study improved our understanding of the combined interactions between air temperature, relative humidity, rainfall, and wheat stripe rust development over a three-season period of autumn-winter-spring.
In this paper, a new Multi-Layer Perceptron Neural Network (MLP NN) classifier is proposed for classifying sonar targets and non-targets from the acoustic backscattered signals. Besides the capabilities of MLP NNs, it uses Back Propagation (BP) and Gradient Descent (GD) for training; therefore, MLP NNs face with not only impertinent classification accuracy but also getting stuck in local minima as well as lowconvergence speed. To lift defections, this study uses Adaptive Best Mass Gravitational Search Algorithm (ABGSA) to train MLP NN. This algorithm develops marginal disadvantage of the GSA using the bestcollected masses within iterations and expediting exploitation phase. To test the proposed classifier, this algorithm along with the GSA, GD, GA, PSO and compound method (PSOGSA) via three datasets in various dimensions will be assessed. Assessed metrics include convergence speed, fail probability in local minimum and classification accuracy. Finally, as a practical application assumed network classifies sonar dataset. This dataset consists of the backscattered echoes from six different objects: four targets and two non-targets. Results indicate that the new classifier proposes better output in terms of aforementioned criteria than whole proposed benchmarks.