Patterns of SSR variation in bread wheat (<Emphasis Type="Italic">Triticum aestivum</Emphasis> L.) seeds under ex situ genebank storage and accelerated ageing

Maintaining seed viability and germplasm integrity is a challenging task in conservation of plant genetic resources, as seeds under storage will lose viability and genetic changes will occur. Attempt was made to analyze the patterns of genetic changes in wheat germplasm under ex situ genebank storage and accelerated ageing treatments. A set of 16 naturally aged wheat accessions under ex situ genebank storage since 1994 were sampled. Four recently regenerated wheat accessions were selected, four random seed samples were chosen from each accession, and three of them were exposed to three different accelerated ageing treatments. These 32 seed samples in two germplasm sets displayed a range of germination rates from 4 to 98 %. Thirty-seven microsatellite markers representing 21 wheat chromosomes were applied to screen 12 seeds of each sample and 449 SSR alleles were scored. Large SSR variation was found in each germplasm set. There was 73.1 % of the total SSR variation present among the naturally aged samples and 78.2 % present among the accelerated ageing samples. Several analyses for genetic association consistently revealed no clear genetic separations among samples of high or low germination rates in both germplasm sets. Samples under different accelerated ageing treatments did not show much genetic differentiations from the original sample of each accession. Mantel tests revealed non-significant associations between SSR variability and sample germination rates for both germplasm sets. These findings are useful for understanding seed deterioration under different ageing conditions and suggest that genome-wide SSR variability may not provide sensitive markers for the monitoring of wheat seed viability.     

基于卷积神经网络与迁移学习的油茶病害图像识别

传统的植物病害图像识别准确率严重依赖于耗时费力的人工特征设计。该文利用深度卷积神经网络强大的特征学习和特征表达能力来自动学习油茶病害特征,并借助迁移学习方法将Alex Net模型在Image Net图像数据集上学习得到的知识迁移到油茶病害识别任务。对油茶叶片图像进行阈值分割、旋转对齐、尺度缩放等预处理后,按照病害特征由人工分为藻斑病、软腐病、煤污病、黄化病和健康叶5个类别。每个类别各选取750幅图像组成样本集,从样本集中随机选择80%的样本用作训练集,剩余20%用作测试集。利用随机裁剪、旋转变换和透视变换对训练集进行数据扩充,以模拟图像采集的不同视角和减少网络模型的过拟合。在Tensor Flow深度学习框架下,基于数据扩充前后的样本集,对Alex Net进行全新学习和迁移学习。试验结果表明,迁移学习能够明显提高模型的收敛速度和分类性能;数据扩充有助于增加数据的多样性,避免出现过拟合现象;在迁移学习和数据扩充方式下的分类准确率高达96.53%,对藻斑病、软腐病、煤污病、黄化病、健康叶5类病害的F1得分分别达到94.28%、94.67%、97.31%、98.34%和98.03%。该方法具有较高的识别准确率,对平移、旋转具有较强的鲁棒性,可为植物叶片病害智能诊断提供参考。     

基于Web投票机制的免疫协同过滤推荐技术研究

针对投票机制的个性化推荐系统,利用人工免疫网络模型的动态适应性、自组织性等特性,通过构建用户的兴趣模型来设计用户投票行为的抗体（抗原）编码及交叉克隆算子,并通过人工免疫网络模型来产生高效的推荐结果。     

The effect of land cover/vegetation on soil water dynamic in the hilly area of the loess plateau,China

Severe soil and water loss has lead to widespread land degradation in China's loess plateau. During the past decades, a great deal of effort was made on vegetation restoration to reduce soil and water loss in the loess plateau. However, due to water shortage the efficiency of vegetation restoration was not as satisfactory as expected. As part of a vegetation restoration project, we conducted research aiming to understand the relationship between vegetation pattern and soil water dynamics. The goal was to find vegetation types appropriate for the loess plateau with scarce water resources. In 1986, fifteen plots of land were planted with five vegetation types: pine woodland, shrubland, sloping cropland, alfalfa and semi-natural grassland. Soil water content, runoff, soil erosion were measured for each plot. Environmental variables, such as rainfall, evaporation and temperature, were recorded simultaneously by an automated meteorological station. The relationship between land cover pattern and soil water dynamic was evaluated by using statistical models. We found that: (1) soil water loss occurred during the growing season, and it reached the maximum in the second half of July; (2) soil water was not fully replenished from rainfall during the rainy season; (3) pine woodland induced the largest water loss to surface runoff, followed by sloping cropland, alfalfa, semi-natural grassland and shrubland; the poor capability of pine woodland for water conservation may be attributed to soil compaction and poor ground coverage under the tree; (4) in most cases, soil water of the five vegetation types was low except for shrubland and semi-natural grassland where it was moderate-high during a few periods. These conditions inhibit sustainable vegetation growth in the semi-arid loess hilly area of the loess plateau, China.     

毛壳菌生物防治机制研究进展

毛壳菌应用于生物防治,具有安全、高效、低毒的优点。介绍了毛壳菌的生物防治机理,包括产生具有抗菌性能的代谢产物、产生细胞壁降解酶、菌寄生、菌竞争和诱发植物防卫反应物5种类型,最后,展望了毛壳菌在生物防治领域的应用前景。     

方法检出限对水质量评价等级制约的研究

为了研究不同仪器、方法对同一水样检测的结果是否有明显差异,是否得到不同的评价参数,选取河南省新乡市地表水和地下水各3个样本,分别用原子吸收分光光度法（AAS法）和电感耦合等离子发射光谱法（ICP-OES法）检测同一样本中6种金属元素的浓度,并进行检测的差异性分析;评价其水质等级。结果表明,所获得的36对检测结果数据均有明显差异,其中14个指标的污染等级不同。仪器与方法检出限较高,导致所得监测结果受检出限限制,不是真实、准确测得的结果,而是取检出限的一半的值得出,影响了环境质量评价结果的准确、真实程度,影响环境管理措施的制定和实施,应当选择检出限低于最高评价等级数值的仪器和方法。     

响应面法优化碱性蛋白酶制备酪蛋白抗菌肽的研究

酪蛋白抗菌肽作为一种新型抗菌剂,分子量较小、热稳定性强、水溶性好、无免疫原性、对人无毒副作用等特点,可广泛应用于食品、药品、化妆品、饲料行业中。因此本研究利用碱性蛋白酶水解酪蛋白,以碱性蛋白酶浓度、酶解时间、酶解温度为影响因子,在单因素试验结果的基础上,应用Centure-Composite Design中心组合方法进行三因素三水平的实验设计,以酪蛋白水解液抑菌圈直径为响应值,运用响应面法对水解条件进行进一步的优化。结果表明：酶解温度、酶浓度、酶解时间对酪蛋白水解液抑菌圈直径影响显著;酪蛋白抗菌肽制备的最佳工艺为：水解温度为55.73℃,酶添加量为0.67%,酶解时间为66.07 min。回归方程预测酪蛋白水解液抑菌圈直径理论值可达到21.66 mm,3次验证实验的平均抑菌圈直径为21.67 mm,与理论最大值接近,可见该模型能较好地预测酪蛋白抗菌肽制备的最佳条件。     

牧草根土复合体锯切力学模型和试验研究

通过分析切根刀工作时的受力状况,建立根土复合体锯切力学模型。在万能试验机上,用自制的锯式切根刀对白羊草构成的根土复合体进行锯切力学性能试验。分析结果表明:经回归拟合获得单位刀宽工作阻力的响应面数学模型,并与锯切力学模型进行比较,得到一常数,使得工作阻力可通过力学模型乘以该常数来计算;贯入深度与贯入阻力之间呈正比关系;切根刀在往复运动情况下,切根作业时上行和下行工作阻力存在差别,下行时的工作阻力大于上行工作阻力。     

基于MATLAB鸡舍春季温度的分析与预测

在蛋鸡饲养中,温度作为蛋鸡鸡舍的一个重要参数,对蛋鸡的健康、产蛋量、鸡蛋大小和蛋壳质量有着极大的影响。为此,结合鸡舍1个月内每天各个时段的实测温度,应用MATLAB软件对鸡舍温度进行了分析、数据的拟合(最小二乘法)及作图,并对结果进行分析预测,以此得出鸡舍模拟温度公式。结果表明,该方案设计合理,误差较小,研究成果具有较强的可操作性,便于实际应用。     

发酵糟渣旋流闪蒸干燥及其热物性参数的快速测定

介绍了利用旋流闪蒸干燥器干燥发酵糟渣时，气流与物料间放热系数及糟渣物料热传导率、热扩散系数的简便快速测定方法。