This article discusses how to predict environmentally driven phenology of an organism that partially regulates its microclimate.
Phenological models representing physiological and behavioral processes of organisms are used to study, predict, and optimize management of ecological subsystems. One application of phenological models is the prediction of temporal intervals associated with the measurable physiological development of arthropods, for the purpose of estimating future time points of interest such as the emergence of adults, or estimating past time points such as the arrival of ovipositing females to new resources. The second of these applications is of particular use in the conduct of forensic investigations, where the time of a suspicious death must be estimated on the basis of evidence, including arthropods with measurable size/age, found at the death scene. Because of the longstanding practice of using necrophagous insects to estimate time of death, standardized data and methods exist. We noticed a pattern in forensic entomological validation studies: bias in the values of a model parameter is associated with improved model fit to data, for a reason that is inconsistent with how the models used in this practice are interpreted. We hypothesized that biased estimates for a threshold parameter, representing the lowest temperature at which insect development is expected to occur, result in models’ accounting for behavioral and physiological thermoregulation but in a way that results in low predictive reliability and narrowed applicability of models involving these biased parameter estimates. We explored a more realistic way to incorporate thermoregulation into insect phenology models with forensic entomology as use context, and found that doing so results in improved and more robust predictive models of insect phenology. (Publisher Abstract Provided)
- Solicitation Overview: FY 2023 DNA Capacity Enhancement for Backlog Reduction (CEBR) Program—Competitive and Formula
- A Paradigm Shift in Forensic Toxicology Screening: The Development and Validation of Two Automated Sample Preparation Techniques for the Comprehensive Screening of Biological Matrices Using High Resolution Mass Spectrometry with Comparison to Conventional
- Machine learning approaches for person identification and verification