Tropical Tree Growth and Longevity: Validation of Growth Simulation,a Bootstrapping Model

ABSTRACT

Growth Simulation, an analytical modeling technique, has been increasingly used in ecological studies and practical forestry applications where dendrochronology is not applicable. The technique uses randomly sampled diameter increments from tagged trees over a known time interval to assemble a statistical sample of lifetime growth trajectories. We carried out a validation of Growth Simulation using a temperate species in order to compare indirect model outputs with direct tree ring analysis. Rings were measured on sample disks cut from 55 pine branches ranging in age from 8–36 years. Assessments included lifetime growth rates, growth rate with respect to diameter and age, periodic annual increment (PAI), cross-referencing of rings by date, and autocorrelation of growth over successive periods. Tree ring analysis and Growth Simulation showed close correspondence for all parameters tested (maximum, median, and minimum growth rates; longevity estimates). Growth Simulation is found to be a robust and informative technique for studies of tropical tree growth, and is especially useful where analysis of tree rings is not feasible or when bootstrapping analysis of ring data is of interest.     

Height–diameter equations for boreal tree species in Ontario using a mixed-effects modeling approach

Height–diameter relationships based on stand characteristics (trees/ha, basal area, and dominant stand height) were investigated for balsam fir, balsam poplar, black spruce, jack pine, red pine, trembling aspen, white birch, and white spruce using data from permanent growth study plots in northern Ontario, Canada. Approximately half the data were used to estimate model parameters with the rest used for model evaluation. Multiple Chapman–Richards functions with parameters expressed in terms of various stand characteristics were fit to determine the best models for predicting height.     

Economic evaluation of forest drainage and fertilization in northern Ontario peatlands with an investment decision model

Drainage and fertilization of an existing peatland black spruce (Picea mariana (Mill.) B.S.P.) stand and 24 possible methods of planting a clear-cut area near Cochrane, Ontario were evaluated by means of a recently developed forestry investment decision model, FIDME.Results indicate that, on the basis of the cost estimates used and assumptions made, drainage of the existing stand is the most economical treatment if it can reduce the rotation age by 30 years or more. Drainage combined with fertilization is the second best alternative if this treatment can reduce the rotation age by 40 years or more. Fertilization of the undrained site ranks a distant third, and then only if it reduces the rotation age by 10 years; otherwise, it will be uneconomical to apply.In the case of the clear-cut area, planting bare-root stock on a mechanically prepared but undrained site without weed control ranked first among the 24 alternatives considered; the same treatment, but with weed control included, ranked second. Planting containerized seedlings on a mechanically prepared but undrained site with weed control ranked third, followed by planting bare-root stock on a mechanically prepared and drained site with weed control. In contrast, planting containerized seedlings on an unprepared site and with no weed control ranked as the least cost-effective method considered.     

Electrophysiological and Behavioral Responses of Adult Anoplophom glabripennis (Motschulsky) to Volatile Components of Host-Plant

Electrophysiological and behavioral responses of adult A. glabripennis (Motsch.) to volatiles from ashleaf maple (Acer negundo L.) were investigated to identify semiochemicals involved in host location. Measurable electroantennogram (BAG) responses were elicited to all compounds tested, the most effective antennal stimulants were trans-2-hexen-1-al,decyl aldehyde and trans-2-hexenl-ol.These profiles were similar between males and females. In Y-tube olfactometer bioassays, above three compounds with certain concentration, trans-2-hexen-l-al (1%), trans-2-hexen-l-ol (1%) and decyl aldehyde (10%), were significantly attractive to the adults in laboratory. The results show that either BAG or olfactory responding to a particular volatile compound are markedly influenced by the concentration.     

论一元立木材积模型的研建方法

对建立一元立木材积模型的两种常规方法进行了深入分析,提出了以树高—胸径模型为基础的二阶回归估计方法。利用该方法既可以提高模型切合性能,又能有效控制模型的拟合误差,并以西藏自治区的四旁树一元立木材积模型为例进行了说明。     

畜禽养殖疾病诊断智能传感技术研究进展

畜牧业是我国农业的重要组成部分,目前我国畜牧业向着规模化、集约化发展,同时也增加了畜禽疾病诊断的难度。为提高畜禽养殖中的动物福利水平,并降低畜牧养殖中因动物疫病与健康异常带来的经济损失与公共卫生安全风险,近年出现了一批通过数字化、智能化手段实现畜禽疫病诊疗的自动化方法,如机器视觉分析、动物音频分析、红外温度感知、深度学习分类等,这些方法可以有效提高对患病或异常畜禽动物的诊断效率、缩短诊断周期、降低畜牧养殖中人工巡检劳动力。畜禽疫病自动诊疗方法不同于常规的基于病理学知识的诊断方法,其主要通过各类传感器自动获取畜禽在养殖过程中的图像、声音、体温、心率、排泄物等各类特征信息,而后通过梅尔倒谱系数、Logistics回归分析等数学模型和支持向量机、深度学习等智能算法对采集的信息进行综合分析与处理,并对动物的健康状态做出评价与预测。文章分别从畜禽形态诊断技术、行为诊断技术、声音诊断技术、体温诊断技术、其他生理参数诊断技术等几个方面总结阐述了目前动物疫病智能诊断技术研究的进展和一些基础的方法原理,这些方法基于动物外型与体尺、行为与动作、鸣叫与声音、体温、排泄物、呼吸与心率等数字化特征,通过数学模型对传感器采集到的特征进行实时分析与分归类,基本实现了对理想环境下动物健康状态的评价。目前的畜禽动物疾病自动诊疗技术研究成果丰富,但相关诊断方法大多是在理想环境下进行,而实际的生产养殖环境中干扰因素很大,目前的诊断方法大多无法很好地排除干扰并精确提取出所需特征信息;并且目前的数字化禽畜疾病诊断方法多是基于禽畜的一种特征信息进行分析诊断,这使得诊断系统的诊断准确度受到影响,诊断结果说服力不足。同时目前的大多数数字化禽畜疾病诊断方法还存在诊断泛化能力差、抗干扰能力差等问题,这些问题制约了其推广与应用。未来畜禽疾病自动诊断的研究重点是提高其传感算法的精度和数学模型的适用性与鲁棒性,并进一步发展基于多种特征耦合与数据融合的智能化畜禽疾病诊疗专家系统,争取实现实时、高效、智能、精准的畜禽健康诊断。     

基于深度学习与激光点云的橡胶林枝干重建及参数反演

树木的几何建模在林木性状评价、森林动态经营管理与可视化研究中具有重要意义。现今,从激光雷达（Light Detection And Ranging,LiDAR）数据中重建树体三维模型并精准获取林木空间枝干结构参数是数字林业发展的必然趋势。该研究提出了一种深度学习与计算机图形学相融合的树木骨架重建与参数反演方法。该方法以PR107、CATAS 7-20-59、CATAS 8-79三个品种的橡胶树为实验对象,首先,采用背包移动激光雷达获取三个橡胶树品种的样地数据,并通过体素剖分和数据增广策略来构建橡胶树训练样本集。其次,构造由四层特征编码层和特征解码层所组成的点云分类深度学习网络,并包含优化的PointConv模块与不同尺度的特征插值模块,以实现在多尺度条件下,全面考虑点云的全局和局部优化特征,引导网络实现枝叶点云的精确分类。最后,面向分类后的枝干点云,运用计算机图形学的空间连通性算法与圆柱拟合策略,重建树木骨架模型,并自动解决叶子点云与对应的一级枝干归属问题,进而在叶团簇尺度下开展对单株树的精细描述与参数反演。通过对三块橡胶树测试样地的验证和与实测值的比对表明,该研究提出的深度学习网络枝叶分类总体准确率在90.32%以上。骨架重建与叶团簇分析结果显示,PR107品种橡胶树具有较为发散的树冠、最大的分枝夹角和叶团簇体积;CATAS 7-20-59品种橡胶树冠呈花瓶型,分枝夹角和叶团簇体积较小;而CATAS 8-79品种橡胶树尽管胸径最粗,但不耐寒害处于落叶期导致冠积最小。同时,反演得到的橡胶树一级枝干直径与实测值比对为：决定系数R2不低于0.94,均方根误差（Root Mean Square Error,RMSE）小于3.01 cm;主枝干与一级枝干的分枝角为：决定系数R2不低于0.91,均方根误差RMSE不高于4.94°。同时发现橡胶树一级枝干的直径与对应的叶团簇体积呈正相关分布。该研究将人工智能的理论模型应用于林木的激光点云数据处理中,为林木激光点云的智能化分析与处理提供了新颖的解决思路。     

Milk Yield of Holstein Cows Induced into Lactation Twice Consecutively and Lactation Curve Models Fitted to Artiifcial Lactations

Nineteen multiparous barren Holstein cows were subjected to an induction of lactation protocol for 21 d administering estradiol cypionate （2 mg kg-1 of body weight （BW） d-1, on day 1 to 14）, progesterone （0.10 mg kg-1 of BW, on day 1 to 7）, lfumethasone （0.03 mg kg-1 of BW, on day 18 to 20） and recombinant bovine somatotropin （rbST;500 mg per cow, on day 1, 6, 16 and 21）. At the end of lactation and with a minimum of a 2-mon dry period, the same cows were again hormonally induced into lactation. Cows in both lactations were not artiifcially inseminated, they were milked 3 times daily and received rbST throughout lactation. Mean accumulated milk yield at 305 d in milk （DIM） did not differ between the ifrst and second induced lactations （（9 710 ＆#177;1 728） vs. （9 309＆#177;2 150） kg;mean＆#177;SD）. Total milk yield （（12 707＆#177;3 406） vs. （12 306＆#177;4 218） kg;mean＆#177;SD） and lactation length （（405＆#177;100） vs. （410＆#177;91） d;mean＆#177;SD） were not different between the ifrst and second induced lactations. In a second study, 15 empirical models including exponential, power law, yield-density, sigmoidal and miscellaneous models were compared for their suitability by modeling 12-mon （n=334）, 18-mon （n=164） and 29-mon （n=22） lactation cycles of Holsteins cows induced into lactation and treated with rbST throughout the lactation. Hoerl （Y=ab1/xxc）, Wood （Y=axb exp（cx）） and Dhanoa （Y=ax（bc）exp（cx）） models were equally suitable to describe 12-mon lactations. An exponential model with ifve parameters （Y=exp（a＋bx＋cd2＋e/x）） showed the best ift for milk yield for 18-mon lactations. The rational model （Y=a＋bx/1＋cx＋dx2） was found to produce the closest ift for 29-mon lactations. It was concluded that, with the protocol used in the present study, multiparous cows respond favorably to a second cycle of induced lactation, with milk yield similar to that experienced during the ifrst cycle. Thus, dairy producers might be able to lengthen the productive life of infertile high producing cows with a renewal of artiifcial lactation, which would imply an overall reduction in voluntary culling of cows. Also, various equations used to describe the lactation curves demonstrated the potential for iftting monthly milk records of Holstein cows with prolonged lactations and induced hormonally into lactation.