Statistics Netherlands, P.O. Box 24500, 2490HA The Hague, The Netherlands.
Ecol Appl. 2011 Oct;21(7):2510-20. doi: 10.1890/10-1786.1.
The survival of many species in human-dominated, fragmented landscapes depends on metapopulation dynamics, i.e., on a dynamic equilibrium of extinctions and colonizations in patches of suitable habitat. To understand and predict distributional changes, knowledge of these dynamics can be essential, and for this, metapopulation studies are preferably based on long-time-series data from many sites. Alas, such data are very scarce. An alternative is to use opportunistic data (i.e., collected without applying standardized field methods), but these data suffer from large variations in field methods and search intensity between sites and years. Dynamic site-occupancy models offer a general approach to adjust for variable survey effort. These models extend classical metapopulation models to account for imperfect detection of species and yield estimates of the probabilities of occupancy, colonization, and survival of species at sites. By accounting for detection, they fully correct for among-year variability in search effort. As an illustration, we fitted a dynamic site-occupancy model to 60 years of presence-absence data (more precisely, detection-nondetection) of the heathland butterfly Hipparchia semele in The Netherlands. Detection records were obtained from a database containing volunteer-based data from 1950-2009, and nondetection records were deduced from database records of other butterfly species. Our model revealed that metapopulation dynamics of Hipparchia had changed decades before the species' distribution began to contract. Colonization probability had already started to decline from 1950 onward, but this was counterbalanced by an increase in the survival of existing populations, the result of which was a stable distribution. Only from 1990 onward was survival not sufficient to compensate for the further decrease of colonization, and occupancy started to decline. Hence, it appears that factors acting many decades ago triggered a change in the metapopulation dynamics of this species, which ultimately led to a severe decline in occupancy that only became apparent much later. Our study emphasizes the importance of knowledge of changes in survival and colonization of species in modern landscapes over a very long time scale. It also demonstrates the power of site-occupancy modeling to obtain important population dynamics information from databases containing opportunistic sighting records.
在人类主导、破碎化的景观中，许多物种的生存依赖于复合种群动态，即适宜生境斑块中灭绝和定居的动态平衡。为了理解和预测分布变化，了解这些动态至关重要，为此，复合种群研究最好基于来自多个地点的长时间序列数据。可惜的是，这种数据非常稀缺。另一种选择是使用机会数据（即，在没有应用标准化野外方法的情况下收集的数据），但这些数据在不同地点和年份之间存在野外方法和搜索强度的较大差异。动态站点占有率模型提供了一种调整可变调查工作的通用方法。这些模型将经典复合种群模型扩展到考虑物种的不完全检测，并得出物种在站点的占有、定居和生存概率的估计值。通过考虑检测，它们完全纠正了搜索工作随年份的变化。作为一个例证，我们拟合了一个动态站点占有率模型，该模型基于 60 年的荷兰石楠荒地蝴蝶 Hipparchia semele 的存在-缺失数据（更确切地说是检测-未检测）。检测记录来自一个包含 1950-2009 年志愿者数据的数据库，未检测记录是从数据库中其他蝴蝶物种的记录推断出来的。我们的模型表明， Hipparchia 的复合种群动态在该物种分布开始收缩前几十年就发生了变化。从 1950 年开始，定居概率就开始下降，但这被现有种群存活率的增加所抵消，其结果是分布稳定。只有从 1990 年开始，存活率才不足以补偿定居率的进一步下降，占有率开始下降。因此，似乎是几十年前起作用的因素引发了该物种复合种群动态的变化，最终导致了占有率的严重下降，而这种下降直到很久以后才变得明显。我们的研究强调了在非常长的时间尺度上了解现代景观中物种生存和定居变化的重要性。它还证明了站点占有率模型从包含机会性目击记录的数据库中获取重要种群动态信息的能力。
