D3411: Don’t leave readers in the dark! The MaxEnt “black box” is open, so show your work!

Species distribution models (SDM) are commonly used to map the potential distribution of species based on environmental suitability. MaxEnt is a dominantly used SDM. In 2017, Steven Phillips published an article, "Opening the black box: an open-source release of Maxent”, and now researchers can flexibly use MaxEnt to create a parameterized SDM suitable for their species. Parameterization can easily be done by statistically assessing which feature classes and regularization modifiers create the best-fitted model based on complexity. Other settings, however, are equally important but do not receive as much attention. We assessed how authors parameterized their SDM through a literature review of 100 recently published articles focusing on insects. Preliminary data show that 47% of papers do not optimize model complexity. Most authors did detail how they accounted for sampling bias (65%) and co-linear variables (81%), as well as stated the percent of data points used as a test sample (81%) and how the data was divided (61%). Most authors also specified the number of background points used (65%), but few mentioned the sampled area (26%). Additionally, 35% of papers incorporating an optimized model did not state the final model. Our analysis suggests that many authors are using the default settings, and many that do optimize to some degree, do not explicitly reveal their methods. These methods cannot be replicated; thus, it is unclear if they are appropriate for the study system. Researchers using SDM must state their work to keep readers out of the dark!