Skip to main content

Anticipatory plasticity: frog embryos respond to environmental cues by producing an adaptive phenotype at hatching


Harmon, Emily; Evans, Boyce; Pfennig, David (2023), Anticipatory plasticity: frog embryos respond to environmental cues by producing an adaptive phenotype at hatching, Dryad, Dataset,


Developmental plasticity can occur at any life stage, but a context in which it might be crucial is when individuals that produce specific phenotypes early in development gain a competitive advantage at a later life stage. Here we asked if pre-hatching (embryonic) exposure to a nutrient-rich resource can impact hatchling morphology in tadpoles of Mexican spadefoot toads, Spea multiplicata. Induction of a distinctive carnivore morph can occur when a tadpole eats live fairy shrimp. We investigated whether cues from fairy shrimp, detected as embryos, determine hatchling morphology in a manner allowing individuals to take advantage of this nutritious resource. We found that hatchlings with embryonic exposure to shrimp were larger and had larger jaw muscles––traits that increase their ability to compete for shrimp. Thus, embryos can assess and respond to environmental cues by producing preemptive resource-use phenotypes. Such anticipatory plasticity may be an important but understudied form of developmental plasticity.


In July 2020, we added fairy shrimp eggs, Streptocephalus sp., to 5-gallon aquaria of distilled water placed by a window. Streptocephalus sp. occur in the natural ponds where our toads were collected. Their nauplii began hatching within 24 hours. We started feeding the shrimp nauplii Brine Shrimp Food (Northeastern Brine Shrimp LLC) when one day old. We harvested shrimp from aquaria and transferred them to a tub of dechlorinated tap water. We repeated this step to rinse away residual food particles and concentrate the shrimp's density.

In nature, fairy shrimp hatch within 48 hours of pond filling. Depending on the time of day a pond fills, S. multiplicata oviposit within 2-20 hours of pond filling (they breed only on the night after a pond fills). Because S. multiplicata eggs take another 48 hours to hatch, S. multiplicata embryos would typically be exposed to live shrimp for at least 2-20 hours. To mimic these conditions, we bred the frogs approximately 20 hours after the shrimp eggs were placed in water. We bred three pairs of S. multiplicata that were collected near Portal, Arizona, USA, and that had been part of a laboratory colony at UNC-Chapel Hill for up to four years. We induced breeding by injecting adults with 0.07 mL luteinizing hormone-releasing hormone at a concentration of 0.01 µg/µL and leaving the pairs in separate nursery tanks overnight. While the frogs were breeding, we filled 60 replicate plastic boxes (11 x 11 x 11 cm) with 900 mL of dechlorinated tap water. We assigned boxes to the shrimp exposure treatment or control, adding 10 mL of concentrated shrimp nauplii in dechlorinated water to the treatment boxes. Control boxes received 10 mL of dechlorinated water.

The next morning, we collected the frog eggs that had been laid overnight, placing a group of approximately 20 sibling eggs into each experiment box. We randomly distributed ten boxes per treatment per family across three shelves of a metal rack (N=30 replicates per treatment). Eggs were kept at 23 ℃ on a light-dark cycle of 14hL: 10hD. We removed dead shrimp from treatment boxes daily and replenished the boxes with new concentrated shrimp (treatment) or water (control) daily until the tadpoles hatched. All tadpoles hatched approximately two days later.

To assess whether pre-hatching shrimp exposure impacted tadpole morphology at hatching, we left 15 tadpoles per box to develop for another project and randomly sampled the remaining 0-day-old tadpoles from each box (N = 125 hatchlings per treatment). We euthanized the sampled hatchlings via immersion in a 0.1% aqueous tricaine methanesulfonate (MS-222) solution according to IACUC protocol ID 17-252.0-A and preserved them in 95% ethanol. We measured the preserved tadpoles using previously published methods. Specifically, using hand-held digital calipers, we measured each hatchling’s body size (snout-vent length; SVL) and orbitohyoideus (jaw) muscle width (OH; carnivores have larger OH). We also measured the length of its gut (carnivores have shorter guts) by removing and uncoiling the intestines against a ruler. Finally, we determined its Gosner developmental stage by visual inspection. Measurements were done blind with respect to treatment. We standardized OH width and gut length for body size by regressing ln(OH) and ln(gut length) on ln(SVL).

Because we were concerned that we did not have enough fairy shrimp in the first trial, we ran another trial in Fall 2022. This second trial differed from the first in the following three ways. First, two families of eggs were used from adults collected from the field two months before the experiment began (also near Portal, AZ). Second, our shrimp treatments contained both fairy and brine shrimp (Artemia sp.). Brine shrimp are more easily raised at the high densities required to trigger carnivore development; these brine shrimp were hatched from eggs in 25 ppt formulated seawater and fed a solution of baker’s yeast. Third, we sampled up to four hatchlings from approximately 20 replicate boxes per treatment (control N = 66 hatchlings from 23 boxes, treatment N = 49 hatchlings from 17 boxes). Across the two trials, we had 47 replicate treatment boxes and 53 replicate control boxes across five families, representing 366 hatchling tadpoles measured.

To determine if pre-hatching exposure to shrimp impacted the morphology of hatchlings, we used linear mixed-effects models in the R package ‘lme4’. Because there was not a blocking effect of trial date, we pooled data across trials. Our response variables were SVL, Gosner stage, SVL-corrected jaw width, and SVL-corrected gut length. Our fixed effect was treatment, and we specified experimental box nested within family as a random intercept. We used an F-test of fixed-effects terms with Kenward-Roger’s method for denominator degrees of freedom for inference ('lmerTest' package). We performed all analyses in R version 4.1.2.


National Science Foundation, Award: DGE 1650116

National Science Foundation, Award: DEB 1753865