Feeding behaviour of yaks on spring, transitional, summer and winter pasture in the alpine region of the Qinghai-Tibetan plateau

Of the present estimated world population of 14.2 million yaks, approximately 13.3 million occur within Chinese territories (Food and Agriculture Organization of the United Nations, 2003). Although there is an extensive bibliography covering the species, few studies have been conducted in the area of foraging behaviour. The present study was conducted at pasture during the spring, transitional, summer and winter seasons to determine the daily temporal patterns of grazing and ruminating behaviour by yaks. During each study period, two 24 h recordings were undertaken with each of six mature dairy yaks. One study period was conducted on each of the transitional, summer and winter pastures, whereas, due to the considerable changes occurring in the morphology of the spring pasture, three separate studies were completed during March, April and May. During the second of these studies (April), the effect of level of concentrate supplementation on grazing and ruminating behaviour was also examined. Behaviour recordings were made using solid-state behaviour recorders. Short-term intake rates (IR, g min(-1)) were calculated by weighing yaks before and after approximately 1 h of grazing, retaining the faeces and urine excreted and applying a correction for insensible weight loss. Yaks spent less time grazing during the dry season (the early period on the spring pasture) compared with the later green swards (the later period on the spring pasture, the transitional pasture and the summer pasture) (P < 0.05). When the forage quality improved, but there was still insufficient mass (the later period on the spring pasture), the yaks extended their grazing time at the expense of other activities. During the early periods on the spring pasture, the short-term IR by yaks was up to 53 g DM min(-1), significantly higher than at other times (P < 0.05). The level of concentrate offered had little or no effect on grazing or ruminating time. The total eating time of the yaks offered 0.5 or 1.0 kg concentrate was 2.9 and 4.5 h day(-1) respectively, significantly lower than unsupplemented yaks (6.8 h) (P < 0.05). In general, yaks can regulate their foraging behaviour according to the changes of sward conditions in order to achieve optimal grazing strategies. (C) 2007 Published by Elsevier B.V.

Reproductive ecology of the Qinghai-Tibet Plateau endemic Gentiana straminea (Gentianaceae), a hermaphrodite perennial characterized by herkogamy and dichogamy

The combined occurrence of both herkogamy and dichogamy in a hermaphrodite species has been considered to strongly favour outcrossing. In this study, we investigated in detail the reproductive ecology of Gentiana straminea Maxim. (Gentianaceae), a hermaphrodite perennial endemic to the Qinghai-Tibet Plateau. In a series of observations and experiments over four consecutive years, we examined whether the combination of dichogamy and herkogamy in individual flowers completely prevents geitonogamous pollen transfer in this species. The mode of floral development clearly indicates that autonomous self-pollination is completely avoided through herkogamy and dichogamy in individual flowers. This implication was confirmed by the breeding experiments, since no seed was produced when flowers were isolated. However, this gentian proved to be highly self-compatible when geitonogamous selfing was artificially induced. Many flowers opened simultaneously on individual plants, the ratio of male to female phase flowers was close to 2:1 in each inflorescence, at the full anthesis phase, and they were randomly distributed amongst the upper, middle and lower parts of each stem's inflorescence. On average, Bombus sushikini Skorikov, the most frequent visitor and only legitimate pollinator of G. straminea, visited nearly two flowers per inflorescence, and four flowers per plant. Among the pollinators' foraging bouts, the proportions of geitonogamous visits to inflorescences or flowers within an individual plant were 29% and 37%, respectively. Therefore, despite the strict dichogamous and herkogamous characteristics of the individual flowers, geitonogamous selfing might still prevail in G. straininea because of the size of its floral displays and the continuous visiting behavior of B. sushkini. (C) 2005 Elsevier SAS. All rights reserved.

RFID系统优化模型与智能算法研究

射频识别技术（Radio Frequency Identification, RFID）作为采集与处理信息的高新技术和信息化标准的基础，被列为本世纪十大重要技术之一。但是，RFID技术的大规模实际应用仍处于探索阶段，RFID系统的应用基础技术还存在着大量尚未解决的关键问题，其中RFID系统优化是RFID技术研究和应用的重要课题。由于RFID系统本身的动态性和不确定性， RFID系统优化面对的一般是非线性、多目标、大规模的复杂优化问题，传统的数学优化算法在处理这些问题时，存在困难。为此，研究新的优化算法成为RFID技术实际应用和理论研究中必须解决的课题。 智能计算方法是求解复杂RFID系统优化问题的一种可供选择的算法。智能计算作为一个新兴领域，其发展已引起了多个学科领域研究人员的关注，目前已经成为人工智能、经济、社会、生物等交叉学科的研究热点和前沿领域。智能计算的各类算法已在传统NP问题求解及诸多实际应用领域中展现出其优异的性能和巨大的发展潜力。 本文旨在对RFID系统的各种优化问题进行深入研究和探讨，面向RFID技术的实际应用需求构建其优化模型，并基于智能计算思想设计能够有效求解这些复杂模型的新型智能优化算法。具体研究内容包括： 首先，进行了RFID读写器网络的调度问题研究。在深入分析RFID网络中读写器冲突类型和成因的基础上，考虑RFID网络中的读写器冲突约束，以最小化系统中的频道数量、时隙分配以及总处理时间建立了RFID读写器网络调度的数学优化模型。从生物学的角度出发提出基于生态捕食模型的改进PSO算法（Particle Swarm Optimizer based on Predator-prey Coevolution, PSOPC），在一定程度上解决了PSO算法在迭代后期随着多样性丧失而陷入局部最优的缺点。应用PSOPC设计了求解RFID读写器网络调度模型的智能求解算法，分别给出算法的求解框架、关键步骤的实现机制。通过在不同规模的RFID读写器网络上进行实例仿真，验证了算法的有效性和模型的正确性。 其次，进行了基于菌群自适应觅食算法RFID网络规划问题的研究。考虑RFID系统在不同应用环境下的系统需求，建立了RFID网络规化的数学模型，其目标函数分别为：RFID网络标签覆盖率的最大化目标函数、RFID读写器冲突的最小化目标函数、RFID网络运行的经济效益最大化目标函数、RFID网络运行的负载平衡目标函数以及同时考虑全局目标的混合目标函数。将自然界生物觅食所采用的自适应搜索策略与细菌的趋化行为和群体感应机制相集成，提出了适合求解复杂RFID网络规划问题的菌群自适应觅食算法（Adaptive Bacterial Foraging Optimization, ABFO）。通过仿真实验基于ABFO算法分别对RFID网络规划模型中的五个目标函数进行了实例求解和分析，测试结果与标准PSO算法和遗传算法进行了比较分析。 再次，进行了基于系统智能方法的RFID网络规划分布式决策模型研究。采用分布式决策的思想建立了RFID网络规划的层次模型，在一定程度上缓解、分散了RFID网络规划问题的复杂性，以解决具有混合变量（包括离散变量和连续变量）的多目标RFID网络规划问题。针对层次模型求解的复杂性，以复杂适应系统理论为指导思想设计了一种新型系统智能优化算法对RFID网络规划的层次模型进行求解。系统智能算法将群体智能中的单层群体系统概念扩展为多层涌现系统，仿真实验表明新提出的算法显著提高了智能计算方法的寻优能力，以及算法的适应性、鲁棒性和平衡性等性能。 最后，进行了RFID网络目标跟踪系统中的数据融合研究。以基于RFID技术的目标定位与跟踪系统为应用背景，提出了基于模糊聚类方法的多RFID读写器数据融合模型框架。通过深入分析蜜蜂采蜜的基本生物学规律，对蜜蜂的个体行为及群体行为进行模拟，提出了一类新型群体智能优化算法-蜂群优化算法（Bee Swarm Optimization, BSO），并将BSO算法嵌入RFID目标定位跟踪系统，作为其模糊聚类的基本算法。仿真研究表明，提出的融合模型能够有效的过滤读写器对跟踪目标的错误监测数据，显著提高目标定位与跟踪的精度。