A field ecologist’s adventures in the v\irtual world: using simulations to design data collection for complex models

Field data collection can be expensive, time consuming, and difficult; insightful research requires statistical analyses supported by sufficient data. Pilot studies and power analysis provide guidance on sampling design but can be challenging to perform, as ecologists increasingly collect multiple types of data over different scales. Despite a growing simulation literature, it remains unclear how to appropriately design data collection for many complex projects. Approaches that seek to achieve realism in decision-making contexts, such as management strategy evaluation and virtual ecologist simulations, can help. For a relatively complex analysis, we develop and demonstrate a flexible simulation approach that informs what data are needed and how long those data will take to collect, under realistic fieldwork constraints. We simulated data collection and analysis under different constraint scenarios that varied in deterministic (field trip length, travel, and measurement times) and stochastic (species detection and occupancy rates and inclement weather) features. In our case study, we fit plant height data to a multispecies, three-parameter, nonlinear growth model. We tested how the simulated data sets, based on the varying constraint scenarios, affected the model fit (parameter bias, uncertainty, and capture rate). Species prevalence in the field exerted a stronger influence on the data sets and downstream model performance than deterministic aspects such as travel times. When species detection and occupancy were not considered, the field time needed to collect an adequate data set was underestimated by 40%. Simulations can assist in refining fieldwork design, estimating field costs, and incorporating uncertainties into project planning. We argue that combining data collection, analysis, and decision-making processes in a flexible virtual setting can help address many of the decisions that field ecologists face when designing field-based research.