Trevelin Leonardo Carreira, Novaes Roberto Leonan Morim, Colas-Rosas Paul François, Benathar Thayse Cristhina Melo, Peres Carlos A
Programa de Pós-graduação em Zoologia, Museu Paraense Emílio Goeldi/ Universidade Federal do Pará, Belém, PA, Brazil.
Fiocruz Mata Atlântica, Fundação Oswaldo Cruz, Rio de Janeiro, RJ, Brazil.
PLoS One. 2017 Mar 23;12(3):e0174067. doi: 10.1371/journal.pone.0174067. eCollection 2017.
The advantages of mist-netting, the main technique used in Neotropical bat community studies to date, include logistical implementation, standardization and sampling representativeness. Nonetheless, study designs still have to deal with issues of detectability related to how different species behave and use the environment. Yet there is considerable sampling heterogeneity across available studies in the literature. Here, we approach the problem of sample size optimization. We evaluated the common sense hypothesis that the first six hours comprise the period of peak night activity for several species, thereby resulting in a representative sample for the whole night. To this end, we combined re-sampling techniques, species accumulation curves, threshold analysis, and community concordance of species compositional data, and applied them to datasets of three different Neotropical biomes (Amazonia, Atlantic Forest and Cerrado). We show that the strategy of restricting sampling to only six hours of the night frequently results in incomplete sampling representation of the entire bat community investigated. From a quantitative standpoint, results corroborated the existence of a major Sample Area effect in all datasets, although for the Amazonia dataset the six-hour strategy was significantly less species-rich after extrapolation, and for the Cerrado dataset it was more efficient. From the qualitative standpoint, however, results demonstrated that, for all three datasets, the identity of species that are effectively sampled will be inherently impacted by choices of sub-sampling schedule. We also propose an alternative six-hour sampling strategy (at the beginning and the end of a sample night) which performed better when resampling Amazonian and Atlantic Forest datasets on bat assemblages. Given the observed magnitude of our results, we propose that sample representativeness has to be carefully weighed against study objectives, and recommend that the trade-off between logistical constraints and additional sampling performance should be carefully evaluated.
雾网捕法是迄今为止新热带地区蝙蝠群落研究中使用的主要技术，其优点包括操作的便利性、标准化以及采样的代表性。尽管如此，研究设计仍需处理与不同物种行为和环境利用方式相关的可探测性问题。然而，文献中现有研究的采样存在相当大的异质性。在此，我们探讨样本量优化问题。我们评估了一种常识性假设，即前六个小时是几种蝙蝠夜间活动的高峰期，从而可得到整个夜晚的代表性样本。为此，我们结合了重采样技术、物种累积曲线、阈值分析以及物种组成数据的群落一致性，并将它们应用于新热带地区三个不同生物群落（亚马逊地区、大西洋森林和塞拉多）的数据集。我们发现，将采样限制在夜间仅六个小时的策略常常导致所调查的整个蝙蝠群落的采样代表性不完整。从定量角度来看，结果证实了所有数据集中都存在显著的样本面积效应，不过对于亚马逊地区的数据集，六小时策略在外推后物种丰富度明显较低，而对于塞拉多数据集，该策略效率更高。然而，从定性角度来看，结果表明，对于所有三个数据集，有效采样的物种身份将受到子采样时间表选择的内在影响。我们还提出了一种替代的六小时采样策略（在采样夜的开始和结束时），在对亚马逊地区和大西洋森林蝙蝠群落数据集进行重采样时表现更好。鉴于我们所观察到的结果规模，我们建议必须仔细权衡样本代表性与研究目标，并建议应仔细评估后勤限制与额外采样性能之间的权衡。
