Seasonality, stochasticity and population cycles
Seasonality, stochasticity and population cycles
复制标题
季节性、随机性和人口周期
DOI：
--
复制
发表时间：
1998
期刊：
影响因子：
--
通讯作者：
B. Finkenstädt
中科院分区：
文献类型：
--
作者：
B. Grenfell;B. Finkenstädt
寻找模式和过程一直是种群动态史上的中心奥德赛。最近的一个主要偏离是应用现代的、计算机密集的统计方法来解决生态问题(Manly 1997)。人口动力学为统计方法的应用提供了一个特别富有成效的领域，这些方法可以解决依赖密度的过程的复杂性和影响(Hassell等人)。1976年5月；格伦费尔等人。以及它们与环境和人口随机性的相互作用(Sugihara和1990年5月；兰德和威尔逊1991年；格伦费尔等人)。(1994年)。除了这一机遇，不可重复性和测量噪声的问题也使短生态时间序列成为发展新的统计方法的挑战。描述时空动力学和估计种群间耦合强度的关联问题是一个特殊的挑战。这篇专题文章提供了一个综合统计方法和小哺乳动物生态学思想的特别好的例子。我们特别集中在Bjornstad等人的论文上。(1998)和Stenseth等人。(1998年)。他们一起分析了北开店田鼠种群时空变异起源的模式和过程。这项工作与最近的研究平行，两者都是关于野生绵羊种群的动态(Clutton-BRock等人。)和儿童流行病的时空动态(格伦费尔和哈伍德，1997)；因此，我们重点比较田鼠的动态和这些其他系统。Bjornstad等人。(1998)使用频域方法显示田鼠周期的频率和幅度的地理趋势。考虑到许多时间序列都很短，这是一个很难解决的问题，因为由此产生的周期图非常稀疏。他们解决了这个问题，通过平滑和一种新的生态应用函数数据分析的多变量估计光谱集。这揭示了一个明显的趋势，从北海道东南部的相对稳定的动力学到西北部的大振幅周期。这项研究由Bjornstad等人报道。(1998)为今后研究动力学的时空变化提供了一种有用的方法。这里的一种可能的改进方法可能是扩展他们的方法，以考虑地理和其他变异的模式，在中心S S种群之间的相关性。原则上，这可以通过对相干谱(本质上是不同频率的串联之间的R2)中的图案的泛函分析来完成。然而，对这种模式的解释将复杂得多，特别是因为很难从天气的同步效应中剔除空间耦合的影响(格伦费尔等人)。1998年)。在第二篇论文中，Stenseth et al.(1998)使用时间序列模型来寻求对Bjornstad等人揭示的动力学地理模式的解释。(1998年)。Stenseth等人。(1998)使用一个线性现象学模型来探索密度依赖强度的地理变化如何影响人口周期的周期和幅度。他们的基本生物学假设是，季节性强度的地理变化推动了观察到的田鼠动态的变化。通过对模型的巧妙的重新参数化，他们表明这近似于对单个种群的直接密度依赖约束的强度的变化。将该模型与田鼠序列进行拟合，表明了与他们的假设一致的直接密度依赖模式。对该模型的随机模拟加强了这一点，它在周期周期中产生与观测到的一致的地理变化。这些结果为动力学中的地理变化提供了一种有趣且异常一致的解释。然而，它们也提出了一些问题，并与其他已知噪声、非线性和季节性相互作用的系统进行了比较。
The search for pattern and process has been the central odyssey in the history of population dynamics. A major recent departure has been the application of modern, computer-intensive statistical methods to ecological problems (Manly 1997). Population dynamics provides an especially fruitful area for the application of statistical methods which can resolve the complexity and impact of density-dependent processes (Hassell et al. 1976; May 1976; Grenfell et al. 1992) and their interaction with environmental and demographic stochasticity (Sugihara and May 1990; Rand and Wilson 1991; Grenfell et al. 1994). As well as this opportunity, the problems of unreplicability and measurement noise make short ecological time series a challenge for the development of new statistical approaches. Characterizing spatio-temporal dynamics and the linked problem of estimating coupling strengths between populations is an especial challenge. The papers in this Special Feature provide a particularly well-integrated example of the synthesis of statistical methodology and ecological ideas in small mammal ecology. We concentrate in particular on the papers by Bjornstad et al. (1998) and Stenseth et al. (1998). Together, they analyze pattern and process in the origin of spatio-temporal variations in Hokkaidian vole populations. This work parallels recent research, both on the dynamics of feral sheep populations (Clutton-Brock et al. 1997) and the spatio-temporal dynamics of childhood epidemics (Grenfell and Harwood 1997); we therefore focus on comparisons between the vole dynamics and these other systems. Bjornstad et al. (1998) use frequency domain methods to demonstrate a geographical trend in the frequency and amplitude of vole cycles. Given the shortness of many of the time series, this is a hard problem, since the resulting periodograms are very sparse. They solve this problem with smoothing and a novel ecological application of functional data analysis to the multivariate set of estimated spectra. This reveals a clear trend, from relatively stable dynamics in southeastern Hokkaido, to largeamplitude cycles in the northwest. The study reported by Bjornstad et al. (1998) provides a useful methodology for future studies of spatio-temporal variations in dynamics. One possible refinement here might be to extend their methodology to consider geographic and other variations in the pattern of c r o s s correlation between populations. In principle, this could be done by a functional analysis of patterns in the coherency spectrum (essentially the R 2 between series at different frequencies). However, the interpretation of such patterns would be much more complex, especially since it would be hard to factor out the impact of spatial coupling from the synchronizing effects of weather (Grenfell et al. 1998). In the second paper, Stenseth et al. (1998) use a time series model to seek explanations for the geographical pattern in dynamics revealed by Bjornstad et al. (1998). Stenseth et al. (1998) use a linear phenomenological model to explore how geographical variations in the strength of density-dependence affect the period and amplitude of population cycles. Their underlying biological hypothesis is that geographical variations in the strength of seasonality drive the observed variations in vole dynamics. By a neat re-parameterization of the model, they show that this approximates to variations in the strength of direct density-dependent constraints on individual populations. Fitting the model to the vole series indicates patterns of direct density-dependence consistent with their hypothesis. This is reinforced by stochastic simulations of the model, which generate geographical variations in cycle period consistent with those observed. These results provide an interesting and unusually consistent interpretation of geographical variations in dynamics. However, they also raise a number of questions and comparisons with other systems in which noise, nonlinearity and seasonality are known to interact.