Supplement 1. R script files to perform the simulations of Figs. 1, 2, and 3.

File List methods.R (MD5: a6dbce8775f7e95f915c257b09df8eb9) Figure1.R (MD5: 771bfccf830ed45207ad7f42e1dc3b89) Figure2and3.R (MD5: 6030b26a7b56c6a4a1a87afdef477e33) Description The fiel methods.R contains implementations of five methods for fitting logistic models using presence-only data: Expectation-Maximization (EM), Lancaster-Imbens (LI), Lele-Keim (LK), Scaled binomial loss (SB), and Steinberg-Cardell (SC). All methods take as input a matrix of presence data and a matrix of background data, with columns corresponding to environmental predictors. The EM, SB and SC methods also take a third argument: an estimate of species prevalence. Figure1.R contains code for making Fig. 1 in the paper. Comment lines explain the code, describe how to run it, and describe the arguments to the methods. Figure2and3.R contains code for making plots as shown in Figs. 2 and 3 in the paper. Comment lines explain the code, describe how to run it, and describe the arguments to the methods.

文件清单：
methods.R（MD5：a6dbce8775f7e95f915c257b09df8eb9）、Figure1.R（MD5：771bfccf830ed45207ad7f42e1dc3b89）、Figure2and3.R（MD5：6030b26a7b56c6a4a1a87afdef477e33）

描述：
本数据集的methods.R文件实现了5种基于仅存在数据（presence-only data）拟合逻辑斯蒂回归模型的方法：期望最大化（Expectation-Maximization, EM）、兰开斯特-因本斯（Lancaster-Imbens, LI）、莱莱-凯姆（Lele-Keim, LK）、尺度化二项损失（Scaled binomial loss, SB）以及斯坦伯格-卡德尔（Steinberg-Cardell, SC）。所有方法均以存在数据矩阵与背景数据矩阵作为输入，矩阵的列对应环境预测变量。其中EM、SB与SC方法还需额外输入第三个参数：物种出现率的估计值。

Figure1.R文件包含生成论文中图1的代码，文件内的注释行对代码进行了解释、说明了运行方法，并阐述了各方法的输入参数。

Figure2and3.R文件包含生成论文中图2与图3的绘图代码，文件内的注释行同样对代码进行了解释、说明了运行方法，并阐述了各方法的输入参数。