Additive genetic variance for traits least related to fitness increases with environmental stress in the desert locust, Schistocerca gregaria

Abstract Under environmental stress, previously hidden additive genetic variation can be unmasked and exposed to selection. The amount of hidden variation is expected to be higher for life history traits, which strongly correlate to individual fitness, than for morphological traits, in which fitness effects are more ambiguous. However, no consensual pattern has been recovered yet, and this idea is still debated in the literature. Here, we hypothesize that the classical categorization of traits (i.e., life history and morphology) may fail to capture their proximity to fitness. In the desert locust, Schistocerca gregaria, a model organism for the study of insect polyphenism, we quantified changes in additive genetic variation elicited by lifetime thermal stress for ten traits, in which evolutionary significance is known. Irrespective of their category, traits under strong stabilizing selection showed genetic invariance with environmental stress, while traits more loosely associated with fitness showed a marked increase in additive genetic variation in the stressful environment. Furthermore, traits involved in adaptive phenotypic plasticity (growth compensation) showed either no change in additive genetic variance or a change of moderate magnitude across thermal environments. We interpret this mitigated response of plastic traits in the context of integrated evolution to adjust the entire phenotype in heterogeneous environments (i.e., adaptiveness of initial plasticity, compromise of phenotypic compensation with stress, and shared developmental pathway). Altogether, our results indicate, in agreement with theoretical expectations, that environmental stress can increase available additive genetic variance in some desert locust traits, but those closely linked to fitness are largely unaffected. Our study also highlights the importance of assessing the proximity to fitness of a trait on a case‐by‐case basis and in an ecologically relevant context, as well as considering the processes of canalization and plasticity, involved in the control of phenotypic variation.


1-Procedures for minimizing parental environmental effects in the experiment
Although our dataset was obtained from a nature-derived laboratory population, our protocol was designed to minimize non-genetic parental effects on our genetic estimates. First, we removed the main environmental source of parental effects, i.e. crowding, for four successive generations, by rearing individuals under the same isolation conditions in 1L individual plastic cages. Indeed, the desert locust can experience phase polyphenism: a suite of profound, transgenerational, and plastic changes, in response to dramatic increases in local population densities, caused by scattered heavy rains that result in local concentration of food resources (Pener & Simpson, 2009). Table S1 summarizes both literature-based evidence for parental and lifetime effects of population density on the 10 studied traits. The main hypothesis explaining the proximal causes of maternal effects related to population density, involves a crowd-mediated maternal factor either controlling primary egg size (and thus the amount of yolk) which in turn influences hatchling size and color (Maeno & Tanaka, 2010), or released from the reproductive accessory glands in the egg foam and influencing offspring black pigmentation and behavior (McCaffery, Simpson, Islam, & Roessingh, 1998;Simpson & Miller, 2007).
Second, we standardized rearing and maintenance during the whole experiment, in order to equalize parental and fore-parental environments across individuals, within our population. Standardized rearing conditions were a temperature at 34.0°C, 55% humidity, photoperiod of 12/12 hours, and ad libitum feeding with fresh wheat shoots and bran for three successive generations prior to phenotypic measurements. Desert locust populations experience highly variable thermal conditions in the wild (Roffey & Magor, 2003), with air temperature in deserts varying drastically between seasons (from an average of about 30°C to 15°C) and between day and night (from over 50°C to below freezing; Ward 2009). Table S1 summarizes literature-based evidence for parental and lifetime phenotypic effects of temperature, humidity and food on the 10 studied traits. Overall, parental effects induced by these environmental factors are scarce in comparison to those mediated by population density (or gregariousness).
Through this strict standardization of the rearing conditions during four laboratory generations, the remaining parental effects should be strongly restricted to pure genetic variation among parents, (hard to control) micro-environmental variation and to gene-byenvironment interactions. Furthermore, phenotypic measurements were performed on integrative growth traits and early adult traits that are less sensitive to maternal effects than traits involved in early development (and survival). Indeed, maternal effects are expected to be larger for early offspring traits than for late traits, even if they can persist into adulthood (McAdam et al., 2014). Although whether maternal effects detected in hatchlings would persist in later stages is unknown, it has been shown in locusts that the colour of the hatchlings can change drastically in the second instar, depending on the rearing density experienced during the first instar (Injeyan & Tobe, 1981;Tanaka & Maeno, 2006). This suggests that the susceptibility to maternal effects of early and late nymphal development is significantly decoupled in S. gregaria.
Finally, we used a half-sib / full-sib quantitative genetics design with a paternal crossing scheme, which allows for estimates of V A that are not inflated by common environmental effects, especially maternal effects (e.g., nutritional resources provided in the egg by the mother). With all these precautions taken, we can safely assume that our experimental design was efficient in minimizing non-genetic parental effects in our dataset. In this context, adding a maternal effect (cancelled out by an experimental control) in the animal model would lead to statistical overfitting: a model too complex for the data captures too much random noise, is expected to perform poorly, and can lead to unreliable estimates. In other words, since the animal model specifying only a genetic effect captures most of the complexity of our data, it will make the best prediction. Under this modelling strategy, we are only limited by the size of our dataset, which may generate some measurement errors (see the simulation-based power analysis, especially Figure 3, in the main text). The directional changes shown here are from optimal to stressful environments (i.e. gregarious density, low temperature, low humidity and low quality of food). Note that there are interaction terms between temperature and humidity not detailed here. Temperature, density, humidity and food were controlled for in our quantitative genetics experiment. Parental influences summarized here concern early stages while in our study, all morphological phase traits were measured in late stages. Body shape is assessed through the four morphometric ratios E/F, F/C, F/V and O/V with E: Length of the fore wing; F: Length of the hind femur; C: Maximum width of the head; H: Height of the pronotum; P: Length of the pronotum; O: Vertical diameter of eyes; V: Width of the vertex between eyes (see main text and section 2 of the Supporting Information for further details). NA: no data, controversial data or no effect. 1. Elliot et al., 2003;2. Nolte, 1962;3. Dudley, 1964;4. Stower, Davies & Jones, 1960 ;5. Hunter-Jones, 1958;6. Nickerson, 1956;7. Nolte, 1965;8. Dirsh, 1953;9. Bouaichi & Simpson, 2003;10. Islam et al., 1994a;11. Islam et al., 1994b12. Maeno & Tanaka, 201013. Maeno & Tanaka, 2009;14. McCaffery et al., 1998;;15. Maeno & Tanaka, 2011;16. Manchanda, Sachan & Rathore, 1980;17. Hamilton, 1936;18. Hamilton & others, 1950;19. Husain & Ahmad, 1936;20. Gündüz & Gülel, 2002;21. Wardhaugh et al., 1969;22. Papillon, 1968a;23. Papillon, 1968b;24. Maeno & Tanaka, 2008;25. Van Huis et al., 2008.

2-Pedigree visualization
Figure S1 shows our pedigree for the 483 G5 individuals. Five laboratory generations are depicted, including the first generation (G1) issued from the nine egg-pods laid by wild females collected in Mauritania in December 2010. The parentage relationships of these G1 offspring were inferred based on multi-locus genotyping as described in Pélissié et al. (2016).
Other laboratory generations are the three generations of environmental standardization (G2 to G4) and the 5 th generation of thermal treatments and phenotypic measurements. While the number of offspring varied substantially between families and between juvenile (2 to 29 offspring/family/trait/treatment) and adult (1 to 17 offspring/family/trait/treatment) traits, the high nymphal mortality (46%) was random across our sample and the family structure remained constant between traits and treatments (i.e., 13-15 dams and 8 sires). Figure S1. Drawing of the pedigree for the 483 G5 individuals.

3.1-Methods for nymphal pronotum color data
Last-instar nymphs were anesthetized with CO 2 and photographed in lateral view. Pictures were corrected for white balance, using the PhotoFiltre 7© software improved by an external module (plugin "wbadjust"), and analyzed using ImageJ v1.52a (Abràmoff et al., 2004). We selected the whole lateral surface of the pronotum, using the "Polygone selection" function.
The pronotal color pattern is known to correlate to the color pattern of the rest of the body (e.g., head, thorax) (Hunter-Jones 1958). We measured each color channel (RGB) as a 8-bit display mean value in the range 0-255, using the "Color Histogram" function, and calculated the percentage of green color (Green Pigmentation) as G/(R+G+B) (see Fig.S2 for illustrations). We differentiated the set of pixels for which brightness was in the 25% lower range (i.e., 0-64) from the background, using the "Color Threshold" function, and calculated the percentage of dark color (Dark Pigmentation) as number of dark pixels/ number of total pixels (see Fig.S2 for illustrations). In order to provide a better approximation to the Gaussian distribution for statistical analysis, we used a logit transformation on the dark pixel data: log((number of dark pixels)/( number of total pixels -number of dark pixels)). In order to avoid zeros, we preliminary added half a pixel to numbers of dark pixels and of total pixels.

4-Fixed effects of temperature, sex, extra-molting, and hatching weight on the 10 traits measured in this study
Mean ±SD for each level and of categorical factors and statistical values for each factor are presented in Table S2 and S3, respectively. Effects of temperature on color, shape, growth and nymphal life history traits are described in the main document. We described below other effects than temperature associated 4 bits of information against the null hypothesis

5-Pairwise phenotypic correlations
Phenotypic correlations among color, shape, growth and nymphal life history traits are shown in pigmentation. This is agreement with correlations predicted in phase polyphenism (see references in Table S1). The single exception was Dark Pigmentation that was associated with both a low body weight and a slow growth. These correlations were moderate, with absolute Pearson coefficient values ranging from 0.14 to 0.55. The single high value (>0.8) was found for the overall growth rate and development time, traits that mathematically derived from each other. 6-Simulation analysis of the sensitivity of our quantitative genetics analysis to the presence of a low level of maternal effects. Figure S5: Sensitivity of our quantitative genetics analysis to a low level of maternal effects. We show median and interquartile range (y-axis) for heritability estimates (A) and Svalues for the model with an additive genetic variance (B) as a function of simulated heritability (x-axis). We set 11 levels of heritability (i.e., 0. 0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0) and, for each level, we simulated 1,000 phenotypic datasets based on our experimental design at the low temperature, i.e. with exactly the same pedigree as for the subset of individuals phenotyped either for the morphometric ratio O/V (light grey) or the Nymphal Viability (dark grey) (i.e., the minimum and maximum sample size for this dataset, respectively). The level of maternal effect was set to 0.1.