Exposure to stressors can affect an organism’s physiology and behavior as well as that of its descendants (e.g. through maternal effects, epigenetics, and/or selection). We examined the relative influence of early life vs. transgenerational stress exposure on adult stress physiology in a species that has populations with and without ancestral exposure to an invasive predator. We raised offspring of eastern fence lizards (Sceloporus undulatus) from sites historically invaded (high stress) or uninvaded (low stress) by predatory fire ants (Solenopsis invicta) and determined how this different transgenerational exposure to stress interacted with the effects of early life stress exposure to influence the physiological stress response in adulthood. Offspring from these high- and low-stress populations were exposed weekly to either sub-lethal attack by fire ants (an ecologically relevant stressor), topical treatment with a physiologically-appropriate dose of the stress-relevant hormone, corticosterone (CORT), or a control treatment from 2 to 43 weeks of age. Several months after treatments ended, we quantified plasma CORT concentrations at baseline and following restraint, exposure to fire ants, and adrenocorticotropic hormone (ACTH) injection. Exposure to fire ants or CORT during early life did not affect lizard stress physiology in adulthood. However, offspring of lizards from populations that had experienced multiple generations of fire ant-invasion exhibited more robust adult CORT responses to restraint and ACTH-injection compared to offspring from uninvaded populations. Together, these results indicate that transgenerational stress history may be at least as important, if not more important, than early life stress in affecting adult physiological stress responses.

Abstract lacks sample sizes, effect sizes and p values. So, we get a little suspicious.

The write-up is somewhat messy. A flowchart of the sampling and subsampling would have been nice. But, as far as I can make out:

  1. They started with n=86 gravid/’pregnant’ first generation lizards collected from 6 sites in the USA, and then waited for them to lay their eggs, resulting in about n=280 second generation lizards. These were split into 3 groups: 2 treatment, 1 control using a split clutch design (what is that?). Treatment consisted of some kind of stress: either by being placed with fire ants and getting stung or having some hormone applied to their skin via oil. The control group were handled as the others and also had oil put on it but without any active ingredient in it.
  2. Then they waited 8-18 weeks.
  3. The lizards are now adults. Two new subsets were formed, both with n=31. One of these was exposed to the fire ants again, and the other had sham treatment. The lizards were returned to the flock and after 1 hour they took a blood sample.
  4. (another similar substudy with another subset of lizards)


The change in CORT following restraint (CORT reactivity) was not significantly affected by early life treatment ( Fig. 1; F 2,10 =0.29, p=0.752 ). However, CORT reactivity was significantly higher in offspring of lizards from fire ant – invaded populations than uninvaded populations (F 1,9 =5.2 4 , p=0.048 , d =0.87 ; baseline covariate F 1,15 =2.45, p=0.14 0 ). The effect of population invasion status did not differentially influence effect s of early life treatment on CORT reactivity (early life treatment x invasion status F 2,9=0.34 , p=0.718 ).

CORT concentrations were significantly higher in lizards following adult fire ant exposure compared to those placed in an empty arena ( Fig 2; fire ant exposure assay F 1 , 35 = 12 . 0 5 , p=0.001 , d =0.90 ; SVL covariate F 1,44 =10.50, p=0.002 ) . There was an effect of invasion status of the maternal population, but not early life treatment, on overall CORT concentrations across the two adult fire ant assays, with CORT concentrations being higher in lizards from invaded sites (invasion status F 1,23 =4.87, p=0.038 , d =0.57 ; early life treatment F 2,53 =0.86 , p=0.430 ) . CORT responses t o fire ant exposure were similar across maternal population s and early life treatment s ( invasion status x fire an t exposure assay F 1, 25 =0.08 , p=0.782 ; early life treatment x fire ant exposure assay F 2, 34 = 1 . 83 , p=0. 1 76 ; early life treatment x invasion status F 2,34 =2.02, p=0.1 48 ; early life treatment x invasion status x fire ant exposure assay F 2,34 =0.5 5 , p=0.58 4 ).

CORT concentrations following ACTH – inject ion were not affected by early life treatment ( Fig. 3; F 2,22 =2.39, p=0.115 ) but were significantly higher in offspring from fire ant – invaded source populations than in those from uninvaded populations ( F 1, 17 = 7 . 60 , p=0.013 , d =0.97 ) . Early life treatment and invasion status did not interact to affect CORT concentrations following ACTH injection (invasion status x early life treatment F 2,17 =1.4 1 , p=0.272).

The study was not pre-registered as far as I can tell.

So, we have a bunch of lizards from different sites, a lot of subsampling and two tests for a parental stress proxy that came out with p=.048 and p=.013. As far as I can make out, there was no control that these parental populations did not differ genetically in some relevant way. Their tests are somewhat unclear, so maybe they tried to adjust for site, maybe not.

The title is a typical NHST subgroup fallacy. The p for early life stress was not significant, but the one for parental stress proxy was. This doesn’t allow us to conclude that these were different, and neither does the lack of p < alpha allow the conclusion that early life stress had no effect (another NHST fallacy).

I think we will have to wait a little longer before being convinced of such claims.