Contrasting Manual and Automated Assessment of Thermal Stress Responses and Larval Body Size in Black Soldier Flies and Houseflies

Abstract

Within ecophysiological and genetic studies on insects, morphological and physiological traits are commonly assessed and phenotypes are typically obtained from manual measurements on numerous individuals. Manual observations are, however, time consuming, can introduce observer bias and are prone to human error. Here, we contrast results obtained from manual assessment of larval size and thermal tolerance traits in black soldier flies (Hermetia illucens) and houseflies (Musca domestica) that have been acclimated under three different temperature regimes with those obtained automatically using an image analysis software (Noldus EthoVision XT). We found that (i) larval size estimates of both species, obtained by manual weighing or by using the software, were highly correlated, (ii) measures of heat and cold tolerance using manual and automated approaches provided qualitatively similar results, and (iii) by using the software we obtained quantifiable information on stress responses and acclimation effects of potentially higher ecological relevance than the endpoint traits that are typically assessed when manual assessments are used. Based on these findings, we argue that automated assessment of insect stress responses and largescale phenotyping of morphological traits such as size will provide new opportunities within many disciplines where accurate and largescale phenotyping of insects is required.

Scoring thermal limits in small insects using open-source, computer assisted motion detection

“…Greater deviation between CP and visual estimates, however, suggests that this metric may be measuring a different biological response. Numerical differences between computational and visual scoring techniques is a known phenomenon in distance-based automated estimation of HKDT (Laursen et al, 2021; MacLean et al, 2022), although the deviation in these cases could originate from the sliding window approach used to summarize tracking data.…”

“…Thus, methods for automated scoring of thermal tolerance are necessary for improving repeatability and reducing strain on investigators. Recent work has applied particle tracking software (i.e., EthoVision XT) to investigate thermal tolerance in insects, by estimating the onset of coma or recovery using subject displacement, summarized in time intervals, as a proxy for movement (Laursen et al, 2021; MacLean et al, 2022). In Laursen et al (2021), discrepancies between scored estimates were attributed to the lack of sensitivity of distance-based methods to detect subtle movements of appendages.…”

“…Recent work has applied particle tracking software (i.e., EthoVision XT) to investigate thermal tolerance in insects, by estimating the onset of coma or recovery using subject displacement, summarized in time intervals, as a proxy for movement (Laursen et al, 2021; MacLean et al, 2022). In Laursen et al (2021), discrepancies between scored estimates were attributed to the lack of sensitivity of distance-based methods to detect subtle movements of appendages. In MacLean et al (2022), estimates derived using speed recapitulate the expected biological trends between treatments but present numerical discrepancies against those scored manually.…”

Welfare considerations for farming black soldier flies, Hermetia illucens (Diptera: Stratiomyidae): a model for the insects as food and feed industry

“…These data contrast previous field data that suggest nearly 100% of oviposition occurs between 27.5-37.5 °C (Booth and Shepherd, 1984). Studies of critical thermal maxima suggest that 44-46 °C is lethal for adults (Laursen et al, 2021). Age and size (but not sex) determined preferred temperature for adults: young adults (1 day after eclosion) preferred warmer temperatures (28.7 °C) than 7-day-old adults (19.2 °C), while smaller adults preferred cooler temperatures than larger adults (Addeo et al, 2021).…”

Estimation of genetic parameters for the implementation of selective breeding in commercial insect production

“…The relatively short life-cycle of most insects compared to other livestock adds further complexity, since the window for data recording and intervention is short and numerous individuals with synchronised life cycles have to be phenotyped simultaneously. Handling and phenotyping thousands of insects is labour intensive [ 28 , 29 ] and this step is a major bottleneck. When phenotypic records from large populations are required, novel automated approaches need to be used to ensure accurate, fast, unbiased, and reproducible phenotyping across individuals, families, and generations.…”