Field scientists collect data in a noisy heterogeneous environment, where the value of additional data for characterizingthe natural system is weighed against the time and money involved in data collection. This is analogous to foraging forfood data is the resource and its collection can be optimized based on energy costs. Here we conduct a novel simulateddata foraging study to elucidate how spatiotemporal data collection decisions are made in field sciences, and how search isadapted in response to in-situ data. Expert geoscientists were asked to evaluate a hypothesis by collecting environmentaldata using a mobile robot. At any point, participants were able to stop the robot and change their search strategy ormake a conclusion about the hypothesis. We identified previously unrecognized spatiotemporal reasoning heuristics, towhich scientists strongly anchored, displaying limited adaptation in response to new data. We analyzed two key decisionfactors: variable-space coverage, and fitting error to a given hypothesis. We found that, despite varied search strategies, themajority of scientists made a conclusion as the fitting error converged. Scientists who made premature conclusions, eitherdue to insufficient variable-space coverage or before the fitting error stabilized, were more prone to incorrect conclusions.We believe the findings from this study could be used to improve field science training in data foraging, and aid in thedevelopment of technologies to support data collection decisions.