There were major problems in executing the MaxEnt model – after reprojecting and resampling data and fitting it to the study area, it was unable to process the data and execute the model in a manageable time frame and frequently crashed. As a result, I had to start over from scratch and did not project or fit my data to a study area in order to ensure the model worked, which of course hinders the accuracy, precision and continuity of the model. The processing times that did execute took vast amounts of time (20-30 mins to add input data to the model and an entire hour to execute each model) thus gave little room for error, was very costly and put me on a tight deadline. I would note that ArcGIS’s program is a powerful system but it does not do well with certain projections, setting a study area and having finer resolutions. Cutting down on the number of explanatory variables such as the various climate data variables or resampling the data to a coarser resolution may have reduced processing times but of course, this comes at the cost of fewer inputs and therefore a potentially less accurate model.
Sampling bias is inherent to most presence data and impacts the results of the analysis. The Spatial Thinning parameter can be used to help reduce this impact, however, while spatial thinning is a useful remediation to reduce the effects of sampling bias, it is recommended that data from structured surveys be used to further minimize the impact of sampling bias. I did not have any structured surveys to minimize the impact of my sampling bias, and I believe that sampling bias was a large factor in my study, so large that Model 1 identified urban areas as an influencing factor to tick presence.
This study used annual climate variables, not seasonal or monthly climate variables (which were available). Using additional variables would likely lengthen the processing time and would not be feasible for this study however in future iterations of this study additional seasonal or monthly variables may strengthen the predictive capacity and accuracy of the model (Eisen et al., 2016). Using surface temperature data would also have been beneficial to this study because the tick spends most of their lifetime at the ground level, thus surface temperature (rather than air temperature) is a better predictor of survival (Cheng et al., 2017).
This study used only one climate model to obtain climate variables, however there was availability of multiple climate model to choose from. An assemblage of the models would likely help create better predictive data (Eisen et al., 2016). However, future climatic data are also simply predictions and their level of accuracy dwindles with increasingly farther time periods (eg. 2070). In fact, with an ever-changing climate, these extrapolations may not be relevant at all in a few years due to the uncertainty of climate change.