Data for: Modelling time-temperature dependent mortality of pest flies in cold storage to support the management of trade-related biosecurity risks

The data in the attached spreadsheet were used to create the resource dataset for the paper titled "Modelling time-temperature dependent mortality of pest flies in cold storage to support the management of trade-related biosecurity risks" in the Journal of Pest Science.\nLineage: We generated a resource dataset from the published scientific literature testing the effect of cold temperature on pest fly mortality, against which to fit a linear mixed model. We extracted pest fly mortality data from 28 publicly available studies (consisting of 289 trials) published from 1988 to 2024. The 28 studies encompassed 10 species of fly pests with up to 4 developmental stages (egg, larval stages 1, 2, and 3), 13 different host commodities, and exposed temperatures ranging from 0 °C to 7 °C. For each trial in each study, we extracted host commodity (fruit type, including cultivar), fly species, developmental (or life) stage, exposed temperature, exposed time duration at each temperature, and the mortality value. Where the developmental stage was given as “larvae” we assigned it as “larval stage 2”, young larvae as “larval stage 1” and mature larvae as “larval stage 3”. We grouped the different cultivars of the same fruit type together, e.g. Valencia and Navel as “oranges”, and Ellendale and Murcott as “mandarins”. Studies in which the flies were fasted or given a lab diet were excluded. Studies that looked at the effect on multiple host commodities but did not specify which fruit the mortality rate results from were also excluded (e.g. Nel (1936)). We excluded pupae from the model since we found only one study (Kim et al., 2018) with limited data regarding their mortality rates. Mortality rates from trials within each study were generally given as a pooled value across replicates within each trial. Data on individual replicates or standard error values were rarely given. We therefore recorded the mortality rate as it was provided, either for individual replicates or the overall mean. Exact 0% and 100% mortality rates were removed from the data before model fitting as probits of 0 and 100% equate to probit values of negative and positive infinity, respectively. Some datasets were generated using a digitising tool to extract data from graphs in publication pdf files provided by the R package ‘metaDigitise’ (Pick et al., 2019) and DataThief (Tummers, 2006). Where necessary, data presented as tables in publications were digitised using the R package tesseract (Ooms, 2024).