In Matthews et al. (2016), errors were found in Abstract section on page 841 and power (log–log) model section on page 852, due to the following reasons:

To compare c‐values (from the `power log‐log species–area relationship model), the area units should be the same. However, in the analyses comparing the c‐values of the power model between habitat islands and different categories of true islands, the authors mistakenly did not convert the island areas of the habitat island datasets (originally in ha) into km² (the area units of the true island datasets).

When the c‐values were calculated after converting habitat island areas to km², the median habitat island c‐value was 3.16 (Q1 and Q3 = 2.33 and 4.10). c‐values were significantly lower for oceanic islands (median = 1.45; Q1 and Q3 = 0.33 and 2.45; P < .001), continental‐shelf islands (median = 2.20; Q1 and Q3 = 0.80 and 3.39; P < .001), and inland water‐body islands (median = 2.81; Q1 and Q3 = 1.60 and 3.78; P = .04) whereas in the original article we (incorrectly) reported that c‐values were significantly higher for inland water‐body islands, than for habitat islands. Hence, in the Abstract section, under results, the fourth sentence should say “Average c‐values were significantly lower for oceanic islands, inland water body islands, and continental‐shelf islands, than for habitat islands.”

In addition, making these changes affect two parts of two figures in the supporting information Appendix S3 (Figures S2a and S6a). The new figures indicate that, when converting habitat island areas to km², the mean habitat island c‐value increases slightly; this makes sense as you can fit more species in 1 km² than 1 ha.

All of the remaining analyses in the paper are unaffected by the habitat island areas being in the units of hectares, and are thus unchanged.

The supporting information Appendix S3 has been corrected and available online.

The authors apologize for this error and any confusion this may cause.

在马修斯等。（2016），由于以下原因，在第841页的“摘要”部分和第852页的“功率（对数-对数）模型”部分中发现了错误：

为了比较c值（来自“幂对数-对数种类-面积关系模型”），面积单位应相同。但是，在比较栖息地岛屿和不同类别的真实岛屿之间的幂模型的c值的分析中，作者错误地没有将栖息地岛屿数据集的岛屿面积（最初以ha为单位）转换为km ²（面积单位）真实的岛屿数据集）。

当将栖息地岛面积转换为km ²后计算c值时，栖息地岛的c值中位数为3.16（Q1和Q3 = 2.33和4.10）。大洋岛屿（中位数= 1.45; Q1和Q3 = 0.33和2.45; P <.001），大陆架岛屿（中位数= 2.20; Q1和Q3 = 0.80和3.39; P <.001）的c值明显较低，以及内陆水体岛（中位数= 2.81； Q1和Q3 = 1.60和3.78；P= .04），而在原始文章中，我们（错误地）报告说，内陆水体岛的c值明显高于生境岛的c值。因此，在摘要部分的结果下，第四句应为“与生境岛屿相比，大洋岛屿，内陆水体岛屿和大陆架岛屿的平均c值要低得多。”

此外，进行这些更改会影响辅助信息附录S3（图S2a和S6a）中两个图形的两个部分。新的数字表明，将栖息地岛面积转换为km ^2时，平均栖息地岛c值略有增加。这很有意义，因为您可以在1 km ^2中容纳比1 ha更多的物种。

本文中的所有其余分析均不受以公顷为单位的生境岛面积的影响，因此未发生变化。

支持信息附录S3已更正并可以在线获得。

作者对此错误以及由此引起的任何困惑表示歉意。