Sampling bias and the fossil record of planktonic foraminifera on land and in the deep sea
Sampling bias and the fossil record of planktonic foraminifera on land and in the deep sea
复制标题
DOI：
10.1666/11041.1
复制
发表时间：
2012-09-01
期刊：
影响因子：
2.7
通讯作者：
Smith, Andrew B.
中科院分区：
文献类型：
--
作者：
Lloyd, Graeme T.;Pearson, Paul N.;Smith, Andrew B.
迄今为止，浮游有孔虫多样性的大规模趋势是基于天气生物地层范围图的利用。尽管这种方法确保了所使用数据的分类一致性和质量，但它没有正式考虑化石记录中可能存在的任何采样偏差。我们证明，原始文献中记录的浮游有孔虫的出现数据因抽样而存在很大偏差。我们通过证明源自陆地和深海记录的原始多样性曲线显着不同来做到这一点，但它们都与各自环境中的采样强度相关，因此最终受到每个环境中地质记录结构的控制。由于中生代记录的采样在我们的陆地记录中是最好的，而新生代记录的采样在我们的深海记录中是最好的，因此我们将两者结合起来，根据这些数据生成对物种和属多样性随时间变化的最有支持的估计。我们使用股东法定人数二次抽样和建模方法来纠正抽样偏差。然后对数据进行转换，生成物种丰富度的范围图，并将其与多样性历史的两个早期估计进行比较，在这些估计中可以恢复可比较的箱内物种数据。在这三个估计值之间没有发现稳健的统计相关性。尽管幅度差异是可以预料的，但曲线实际形状的差异却令人惊讶。我们的结论是，这些差异源于数据本身的性质，即采用的分类方案和使用的分类覆盖范围。
Large-scale trends in planktonic foraminiferal diversity have so far been based on utilization of synoptic biostratigraphic range charts. Although this approach ensures the taxonomic consistency and quality of the data being used, it takes no formal account of any sampling biases that might exist in the fossil record. We demonstrate that the occurrence data of planktonic foraminifera, as recorded in the primary literature, are strongly biased by sampling. We do this by demonstrating that raw diversity curves derived from the land-based and deep-sea records are strikingly different, but that they each correlate with the intensity of sampling in their respective environments, and thus are ultimately controlled by the structure of the geological record in each setting. Because sampling of the Mesozoic record is best in our land record whereas sampling of the Cenozoic is best in our deep-sea record, we combine the two to generate the best-supported estimates of species and genus diversity over time from these data. We correct for sampling bias using shareholder quorum subsampling and a modeling approach. The data are then transformed to generate a range-through plot of species richness that is compared with two earlier estimates of the diversity history where comparable species-in-bin data can be recovered. No robust statistical correlation is found among the three estimates. Although differences in amplitude are to be expected, differences in the actual shape of the curve are surprising. We conclude that these differences stem from the nature of the data themselves, namely the taxonomic scheme adopted and the taxonomic coverage used.