Comments from curator
Phenotyping data comes from the Dryad deposit of the paper (
doi:10.5061/dryad.k95p5, CC0), file
data for Dryad.xlsx (sheet "data for figure 1"; downloaded from its
Zenodo mirror, same file). For each of the 24 DGRP lines (column "line", the RAL number) it gives the percentage of females that flew under each of the three conditions of the paper. These per-line percentages are the only data published (no per-fly scores), so the study has a summary dataset only. The paper calls the lines "isofemale lines" of the DGRP and gives their identities only in the Dryad file; all 24 are DGRP lines of DGRPool.
Design: the 24 lines, chosen to span a wide range of genetic variation, were reared for two generations at controlled density (two females laying for 48 h per vial) on Bloomington standard medium at 25 °C, 50% humidity, 12:12 h light:dark. Female metamorphs were collected under CO
2; at 6-8 days of age they were transferred to individual vials, and the next day each fly was tested once, without anaesthesia, under one of three conditions: heat stress (39 °C, 21% O
2), hypoxic stress (25 °C, 12% O
2) or no stress (25 °C, 21% O
2). The fly was tapped from its vial into a 25.4 cm acrylic flight chamber (temperature set by a walk-in incubator, oxygen by a ROXY-1 controller) and scored as a fall if it landed on the floor within 10 cm of the point below the vial, otherwise as a flight. The paper reports 20 flies per line and condition; lines were tested in random order each day, and the conditions on alternating days. Only females were tested: sex is F.
The summary dataset has three phenotypes (unit %, summary type mean: the percentage of the line's tested females that flew), copied unchanged from the file (where they are rounded to 3 decimals, e.g. 38.095):
-
mn_Flight_Normox39C: heat stress (column normoxia39C), from 0% (DGRP_861) to 70% (DGRP_338, DGRP_563).
-
mn_Flight_Hypoxia25C: hypoxic stress (column hypoxia25C), from 45% (DGRP_799) to 100% (DGRP_338).
-
mn_Flight_Normox25C: no stress (column normoxia25C), from 70% (DGRP_799) to 100% (5 lines).
The paper's analysis (linear models of flight under heat stress on flight under hypoxia and/or no stress, compared by AICc and model averaging with the R package MuMIn) is reproduced from these values: the log likelihoods, AIC (which are AICc, n = 24), ΔAIC and Akaike weights of Table 1 are identical (e.g. P
hypoxia model: log likelihood -99.6, AIC 206.5, weight 0.74), and r
2 = 0.30 for heat vs hypoxia and 0.045 for heat vs no stress (paper: 0.30 and 0.04). The slope reported in the text (β = 0.99, s.e. = 0.35, p = 0.01) is not reproduced: the least-squares slope is 0.85 (s.e. 0.28, p = 0.006) and the model-averaged slope 0.84.
Of note, 13 of the 72 values are multiples of 1/21 rather than 1/20 (e.g. 38.095% = 8/21; 5 hypoxia values and 8 heat values, in 12 lines: DGRP_041, DGRP_088, DGRP_153, DGRP_321, DGRP_356, DGRP_373, DGRP_374, DGRP_563, DGRP_707, DGRP_796, DGRP_799, DGRP_852), so 21 flies rather than 20 were scored in these cases. All other values are multiples of 1/20 (0% and 100% fit either).