Conspecific Oophagy by Tadpoles: Conditions for Its Occurrence and Importance as a Source of Anuran Egg Mortality
Abstract
The prevailing view of tadpoles as mostly herbivorous organisms is changing with multiple reports of carnivory and ontogenetic, spatial, and temporal changes in tadpole diet. One increasingly recognized aspect of tadpole diet is conspecific oophagy of living fertilized eggs. As reports of this behavior accumulate, patterns remain unanalyzed, and implications for population dynamics remain unexplored. We reviewed reports of conspecific oophagy by tadpoles and found that the behavior is taxonomically, phylogenetically, and regionally widespread, with reports worldwide, except Africa, and across 13 families. Additionally, we report the first observations of conspecific oophagy in threatened Chiricahua Leopard Frogs (Rana chiricahuensis). Conspecific oophagy is predominantly reported in species using ephemeral lentic environments, with cannibalistic tadpoles predominantly in advanced developmental stages (Gosner stages 31–38). We found that species engaging in conspecific oophagy share similar reproductive and larval traits, with a few intriguing outliers including R. chiricahuensis and other leopard frogs. We observed conspecific oophagy in R. chiricahuensis in the wild and assessed its effects by monitoring egg masses and evaluating the relative importance of conspecific tadpole abundance and size in predicting egg mass mortality. The size of nearby conspecific tadpoles was the only statistically significant predictor of egg mass survival. Our results underscore the importance of ontogenetic diet changes in tadpole development, which facilitate conspecific oophagy.
Cannibalism is common in nature and occurs in a variety of taxonomic groups, including anurans (Fox, 1975; Polis, 1981; Polis and Myers, 1985). However, the occurrence of cannibalism may be underrecognized in tadpoles. All tadpoles use a buccal pumping system for feeding but vary in feeding mechanisms. Tadpole feeding can be broadly separated into microphagous suspension feeding and macrophagous feeding, which uses jaw sheaths and teeth to remove and break up material from substrate (Wassersug and Hoff, 1979; Alford, 1999). Across these feeding types, tadpoles are traditionally categorized as herbivores or detritivores (Montaña et al., 2019). However, reports are accumulating that more anurans have carnivorous and omnivorous tadpoles than previously thought (Montaña et al., 2019). These reports include records and studies of cannibalistic behavior in tadpoles, including those of some spadefoot toad tadpoles (Scaphiopodidae), which may develop into carnivore morphs in the presence of conspecifics (Polis and Myers, 1985; Pfennig and Frankino, 1997). However, much remains unknown about tadpole cannibalism. In particular, understanding con-specific oophagy by tadpoles may have substantial implications given the significance of embryonic and tadpole survival to anuran population dynamics (Govindarajulu, Altwegg, and Anholt, 2005; Rose et al., 2021; Terrell et al., 2023).
Conspecific oophagy by tadpoles was initially thought to be limited to maternal provisioning of unfertilized eggs to oophagous arboreal tadpoles of some anuran families (Crump, 1983). However, we now know that tadpoles of various species feed upon living fertilized conspecific eggs that are not actively provisioned, a behavior referred to hereafter as “conspecific oophagy by tadpoles” (Petranka and Thomas, 1995; Summers, 1999; Wickramasinghe, Oseen, and Wassersug, 2007). Causal explanations for this behavior often consider costs and benefits to fitness of the cannibalizing tadpoles.
Conspecific oophagy likely has few direct individual fitness costs, though it is possible that consumption of eggs could condense cannibalistic tadpoles in space and increase their risk of predation. However, cannibalism of half-siblings and even full siblings could have indirect fitness costs to the average group fitness of a cannibalistic tadpole and its relatives (Crossland et al., 2011). Intraclutch conspecific oophagy by tadpoles occurs in at least one anuran species, Philoria pughi, so these impacts to group fitness are possible (Gould, Mahony, and Mahony, 2023). In addition, overwintering tadpoles could consume kin from a subsequent cohort produced by their parents. Such group fitness costs are unlikely in settings where probability of consuming relatives is low. For example, many anuran species with short breeding periods produce many nests from multiple females that are available to potential cannibalistic tadpoles. Fitness costs of conspecific oophagy by tadpoles could be more probable in species with smaller clutches, infrequent breeding, low local abundance, and asynchronous breeding. In almost all instances, there would be few costs to cannibalistic tad-poles. Because of this, the benefits of conspecific oophagy have received more attention in causal explanations of the behavior.
There are two leading hypotheses on how conspecific oophagy benefits tadpoles: the nutrient benefit hypothesis and the reduced competition hypothesis (Crossland et al., 2011). The nutrient benefit hypothesis proposes that tadpoles consume conspecific eggs to acquire necessary protein and nutrients (Crump, 1983; Crossland et al., 2011). Reports proposed the nutrient benefit hypothesis in conjunction with observations showing that many reports of conspecific oophagy by tadpoles occur in low-nutrient ephemeral habitats (Crump, 1983; Summers, 1999; Gould, Clulow, and Clulow, 2020). However, feeding experiments have been inconclusive in determining nutritional benefits of conspecific oophagy (Summers, 1999; Crossland, et al. 2011). Summers (1999) found no difference in growth rates between tadpoles fed conspecific eggs and those that were not fed at all. However, tadpoles in the study were in early developmental stages (Gosner stage 25), and Summers pointed out that tadpoles in later developmental stages may benefit more from conspecific oophagy (Summers, 1999). Late-stage development of tadpoles may require nutrients that cannot be found in algae and other aquatic vegetation, and consequently, tadpoles may undergo ontogenetic diet shifts (Montaña et al., 2019).
The reduced competition hypothesis proposes that tad-poles cannibalize eggs to reduce intraspecific competition (Crossland et al., 2011). In some instances, predation can result in higher cohort survivorship by reducing population density of tadpoles, especially in ephemeral conditions with a high risk of breeding sites drying out (Dayton and Fitzgerald, 2011). Additional support for the reduced competition hypothesis is based on findings that some tadpoles perform conspecific oophagy but avoid heterospecific oophagy. Intraspecific competition can be more intense than interspecific competition in tadpoles (Skelly, 1995; Smith, Dingfelder, and Vaala, 2004; Crossland et al., 2011), supporting the idea that conspecific oophagy may reduce competition. For example, reduced competition may result from conspecific oophagy in Cane Toad tadpoles (Rhinella marina; Crossland et al., 2011), which are known to avoid heterospecific oophagy. However, conspecifics are not universally more impactful competitors, and reduction in competition is specific to particular anuran species (Kupferberg, 1997; Smith, Dingfelder, and Vaala, 2004). Support for the reduced competition hypothesis could gain more traction if studies can show that conspecific oophagy occurs more widely and with a corresponding increase in cohort survival (Dayton and Fitzgerald, 2011).
Ultimately, reduced competition and nutrient benefits do not have to be mutually exclusive factors in driving tadpoles to cannibalize conspecific eggs. Conspecific oophagy could both lower competition and provide easier nutrient uptake than consuming other food sources. Together, these hypotheses have similar predictions, specifically that conspecific oophagy is more likely to occur in tadpoles in later developmental stages and in low-nutrient habitats. However, it is unknown whether patterns of conspecific oophagy fit these predictions. To gain insights into the broader context of conspecific oophagy, we reviewed reports of anuran species performing conspecific oophagy in the tadpole stage in the context of abiotic and biotic environmental factors.
We also report the first observations of conspecific oophagy by Chiricahua Leopard Frog tadpoles (Rana chiricahuensis). The behavior in R. chiricahuensis does not appear to fit predictions of conspecific oophagy. Rana chiricahuensis tadpoles inhabit water bodies with long hydroperiods that are not nutrient-limited (Degenhardt, Painter, and Price, 2005). Intrigued by our observations, we also questioned the effect of this behavior on R. chiricahuensis population dynamics. To understand these broader consequences, we asked whether tadpole conspecific oophagy substantially contributed to R. chiricahuensis egg mass mortality in our study population. This question is particularly relevant as R. chiricahuensis is listed as Threatened in the United States (US Fish and Wildlife Service, 2002) and Vulnerable by the IUCN Red List (IUCN SSC Amphibian Specialist Group, 2022). In addition to our direct observations of cannibalism, we hypothesized that higher abundance of more developed tadpoles near egg masses should correlate with reduced survival of egg masses. We also predicted that tadpole abundance and development are more important predictors of egg mass survival than attributes of the riverscape.
We thus took a three-pronged approach to better understand conspecific oophagy by tadpoles: (1) we analyzed patterns in reports of conspecific oophagy to understand the conditions and potential causes of this behavior; (2) we described our field observations of R. chiricahuensis tadpoles cannibalizing conspecific eggs; and (3) we tracked the fate of egg masses in a population of R. chiricahuensis and modeled the importance of tadpole conspecific oophagy in relation to the mortality rate of the eggs.
Materials and Methods
Patterns in Conspecific Oophagy by Tadpoles
We conducted a literature review to analyze patterns in conspecific oophagy by tadpoles. We excluded reports of tadpoles consuming unfertilized (provisioned) eggs or deceased eggs. We conducted searches for all years of publication in Web of Science and Google Scholar using the following terms: (Tadpole* OR Anuran Larvae) AND (Feeding OR Diet OR Eating OR Foraging) AND (Conspecific OR Cannibal*) AND (Egg* OR Egg Mass). The Google Scholar search covered any time up to 2024, and the Web of Science search covered the years from 1900 to 2024. We confirmed relevance of each result by reading the text and removed studies that were not pertinent. We also reviewed Alford (1999) and all volumes of the journal Herpetological Review for observations of conspecific oophagy by tadpoles, collecting articles and notes that either reported or cited conspecific oophagy. We used the snowballing method, scanning citations of our relevant articles, to find additional reports (Cook et al., 2022).
For each report of conspecific oophagy, we noted year and region of the observation and any of the following variables if available: ecosystem(s) (lentic, lotic, or both), hydroperiod(s), and Gosner stage(s) (Gosner, 1960). We compared the frequency of the occurrence of conspecific oophagy among anuran families, continents, ecosystems, and hydroperiods. To measure the increase in reports over time, we conducted a regression analysis in R Statistical Software (4.1.2; R Core Team, 2021) to test for a statistically significant association between the time of publication and the number of reports of conspecific oophagy by tadpoles. Finally, we calculated descriptive statistics on Gosner stages of tadpoles performing conspecific oophagy. We used either the single number provided for Gosner stage for reports that did not provide a range or the median value for reports in which a range of Gosner stages was reported.
We conducted a principal component analysis (PCA) of larval and reproductive traits of the species reported to perform conspecific oophagy as tadpoles. The traits we used were tad-pole maximum total length, minimum larval period, maximum larval period, minimum clutch size, maximum clutch size, and longest reported breeding season. We extracted data from literature searches and anuran trait databases such as AmphiBIO (Oliveira et al., 2017) (Table S1). We converted all values in our dataset to their z-scores. We excluded other anuran species and focused only on species in which conspecific oophagy is known to occur. As such, our PCA provided information on whether there is convergence or dispersion in multivariate trait space for current records of conspecific oophagy by tadpoles. We did not attempt a comparison between anuran species that have tad-poles performing conspecific oophagy and species that do not because many species may perform conspecific oophagy but have not yet been recorded doing so. Using the classification system of Altig and Johnston (1989), we also obtained information on ecomorphology of each species known to engage in con-specific oophagy. In addition to Altig and Johnston (1989) and Altig and McDiarmid (1999), we used other sources to identify ecomorphology of the species under consideration. We examined PCA groupings to assess whether they were associated with ecomorphology guilds. We performed the PCA and plotted output using R Statistical Software (4.1.2; R Core Team, 2021).
Conspecific Oophagy and Egg Survival in Chiricahua Leopard Frogs
We made observations of R. chiricahuensis and collected population data at a creek in Gila National Forest, New Mexico, USA. We withhold the precise location for the security of this threatened species, which is susceptible to chytrid infections. Breeding activity of R. chiricahuensis varies, but at sites with breeding seasons of suitable duration, different cohorts of R. chiricahuensis tadpoles coexist with conspecific egg masses (Jennings, 1988; Jennings, 1990). This is the case at our study creek because a pulse of continuous breeding occurs from April to June, with a lesser pulse in late summer to early fall. It is probable that a cohort of tadpoles overwinters, as well, and that this overwintering cohort is present at the time of the first oviposition of each reproductive season (Jennings, personal communication).
We collected data daily on R. chiricahuensis egg masses along a 1,220 m transect from 16 May 2022 to 20 June 2022 (Table S2). For each egg mass, we recorded the date first observed, geographical coordinates, and its fate (either date of hatching or date of egg mass mortality).
At 80 m intervals along the transect, two observers searched for R. chirachuensis tadpoles by dip-netting for 5 min. We also dip-netted for 5 min in a connected backwater that harbored many egg masses and an abundance of tadpoles, which brought the number of survey sites up to 16. For each tadpole, we measured its total length (nearest mm) and noted its Gosner stage (Table S3). Bury and Corn (1991) found teams of two surveyors were optimal, and the 5-min dip-netting duration allowed each interval site to be visited more frequently. We surveyed the entire transect every third day.
We measured water velocity, stream width, and stream depth at each interval site (Table S4). We measured water velocity as the time it takes for a low-density floating tennis ball to move 1 m (Dobriyal et al. 2017). We used these data to calculate the area (using width and mean depth) and volumetric flow rate (using area and water velocity) of the stream at these spatial points.
Using R, we conducted a logistic regression model with egg mass survival as the response variable and mean tadpole size, abundance, and volumetric flow rate as predictors (4.1.2; R Core Team 2021) (Table S5). We removed water velocity, mean water depth, and stream segment area as variables because they were all highly correlated with volumetric flow rate. Tadpole development was also removed as a variable due to its correlation with tadpole size (Table S6). Tadpole statistics and stream variables associated with each egg mass were taken from the sampling point closest to the egg mass and the sampling event closest in time to the existence of the egg mass.
Results
Taxonomic, Spatial, and Temporal Distribution of Conspecific Oophagy
Forty-one species within 13 anuran families (22% of described families) have reports of conspecific oophagy by tadpoles (Table 1) (Frost, 2023). Reports of the behavior were relatively evenly distributed among families except for Ranidae, Bufonidae, and Hylidae, which together accounted for approximately 56% of observations (Table 1).
Conspecific oophagy by tadpoles is spatially widespread with reports from anuran species from all continents where anurans are present except Africa. There were some differences in the proportion of reports from each region, with most reports coming from Europe, North America, and South America (Table 1). We aggregated reports into 5-year intervals. When including the first report in 1944, reports have significantly increased over time (P = 0.00038, Table 1). However, the 1944 report was a considerable early outlier in reporting; the next report does not appear until 1965. When we excluded the 1944 outlier, growth in observations of conspecific oophagy by tadpoles was not statistically significant (P = 0.089). The number of new species reports decreased in recent periods of 2015–2019 and 2020–2024, indicating a lack of substantial growth in the reporting rate on conspecific oophagy by tadpoles.
Conditions of Occurrence
Most occurrences of conspecific oophagy by tadpoles were in lentic ecosystems (Table 1; ~93% of events, n = 38 of 41), with only four reports from lotic environments (~10% of events, n = 4 of 41). Rhinella marina performed the behavior in both ecosystem types. Although not all reports included information on the hydroperiod (26 of 41 reports had associated hydroperiod data), most in situ occurrences were in ephemeral conditions (~69%, n = 18). Rhinella marina was also the only species to perform conspecific oophagy in both hydro-period conditions (ephemeral and nonephemeral). Overall, across ecosystems and hydroperiods, most occurrences of con-specific oophagy by tadpoles were in ephemeral lentic environments (~69%, n = 18). To the best of our knowledge, only R. marina and Nannophrys ceylonensis, alongside our observations of R. chiricahuensis tadpoles, have been observed performing conspecific oophagy in nonephemeral lotic environments.
As with hydroperiod data, not all reports provided Gosner staging of tadpoles performing conspecific oophagy, with ~56% (n = 23 of 41) including these data (Table 1). There is a pattern across anuran species in the developmental stage of larvae engaging in conspecific oophagy. Tadpoles consuming conspecific eggs had a mean Gosner stage of ~32, a median Gosner stage of ~33, and a mode Gosner stage of ~25 (Table 1; range = 24–39, standard deviation [SD] = 5.23, coefficient of variation = 0.16). However, when excluding ex situ experiments, mean Gosner stage was ~34, median was ~36, and mode was ~37 (Table 1; range = 24–39, SD = 4.04, coefficient of variation = 0.12). Most tadpoles cannibalizing eggs are thus in or past the grouping of Gosner stages 31–38, as all descriptive statistical values from in situ observations were within this range. Although it was an outlier regarding hydroperiod and ecosystem conditions, our observation of conspecific oophagy by R. chiricahuensis tadpoles follows the pattern of developmental stages of cannibalistic tadpoles.
Distribution in Multivariate Trait Space
We acquired relevant traits for most species reported to perform conspecific oophagy in the tadpole stage (Table 2). Leptodactylus knudseni, Nanohyla arboricola, Philoria pughi, Phyllomedusa vaillantii, Pleurodema bufoninum, Smilisca sordida, and Ranitomeya ventrimaculata were the only species for which complete trait coverage was unavailable. We therefore excluded these species from our PCA.
The first three axes of our PCA accounted for 81% of the variance in traits under consideration (Table 3). We projected these axes onto each other and evaluated the distribution of species in multivariate trait space. Projecting PC I onto PC II, we identified three groupings of species and two outliers (Fig. 1A). One grouping consisted of most species in the PCA (~76%, n = 26 of 34) and, except for Myobatrachidae, included members from every anuran family reported to exhibit con-specific oophagy. A second grouping consisted of leopard frogs from the American Southwest, and a third grouping consisted of most bufonids present in our PCA. The leopard frog grouping differed from other groups based on tadpoles taking longer to reach metamorphosis and reaching larger sizes. The bufonid group differed from other groups based on having larger clutch sizes and, consequently, reaching higher numbers of conspecific larvae in an area. There were also two outlier species that did not group with any other species, Mixophyes balbus and Rana catesbeiana. Mixophyes balbus was closest to the leopard frog grouping but is larger in tadpole size and longer in larval period. Rana catesbeiana was an outlier in every trait category (Fig. 1A). Projecting PC I onto PC III and PC II onto PC III revealed that species were not grouped by length of breeding season. However, in these projections, the other traits continued to differentiate species into the same groupings and outliers found in the projection of PC I onto PC II (Figs. 1B, C).
Citation: Journal of Herpetology 59, 3; 10.1670/2678037
Groupings and outliers from PCA were not correlated with tadpole guilds (Fig. 1). We found most species that are currently known to perform conspecific oophagy are primarily in the benthic guild (Table 2; 65%, n = 22 of 34). All groupings outside of the main grouping consisted of benthic species.
Chiricahua Leopard Frog Field Observations
At our site in Gila National Forest, on 23 May 2022 at 2204 h, we observed R. chiricahuensis tadpoles surrounding a conspecific egg mass. This living fertilized egg mass contained embryos at Gosner stages 15–17 (Gosner, 1960). The tadpoles surrounded the mass and fed upon the embryos within (Fig. 2A; Supplemental Video S1). Although there were other conspecific tadpoles nearby that may also have contributed to consumption, we observed R. chiricahuensis tadpoles at or beyond Gosner stage 37 (Gosner 1960) feeding on the egg mass, which was gone by the following morning (roughly 12 hours after observation). On 28 May 2022, we observed a second instance of R. chiricahuensis tadpoles feeding on an egg mass (Fig. 2B; Supplemental Video S2). This egg mass contained embryos at Gosner stages 13–15. Similar to our first observation, the tadpoles involved in feeding were at or beyond Gosner stage 37, and the egg mass was gone by the following morning.
Citation: Journal of Herpetology 59, 3; 10.1670/2678037
In addition to the direct observations of tadpoles eating eggs, we observed further instances where large conspecific tad-poles surrounded R. chiricahuensis egg masses (n = 6). These egg masses appeared to be losing structural integrity. It is possible that damage to the egg masses may have been unrelated to predation by conspecific tadpoles because ranid egg masses naturally disintegrate (Beattie 1980). However, daily egg surveys showed these egg masses intact the day before we saw them surrounded by tadpoles. Furthermore, the egg masses had decreased in size since our daily egg survey. Each time we saw a group of large conspecific tadpoles around an egg mass at night, those masses disappeared by morning. Therefore, we consider these observations to be oophagy events.
Chiricahua Leopard Frog Egg Mass Mortality Factors
Overall egg mass mortality was 53.70% (n = 29 of 54), not including 6 egg masses lost in a flood at the end of our study. All predictor variables (tadpole abundance, mean tadpole size, and flow rate) had a negative relationship with egg mass survival. However, only mean tadpole size was a statistically significant predictor of egg mass survival (Table 4). At the largest mean tadpole sizes observed, associated egg mass survival was as low as 25%, whereas at the smallest mean sizes of tadpoles, egg mass survival was close to 95%. Furthermore, although the relationships between egg mass survival and all predictor variables were negative, the relationship between flow rate and egg mass survival was weaker than relationships between tadpole variables and egg mass survival (Table 4).
Precipitation and stream discharge were relatively stable during the study period (Fig. 3). A flood at the end of our study scoured six remaining egg masses. We excluded these egg masses from our logistic regression because we were interested in determining the influence of conspecific oophagy on mortality of R. chiricahuensis egg masses under typical conditions. Nevertheless, it is clear that stochastic events, such as egg mass scouring by high stream flow, can be high sources of egg mortality at this site.
Citation: Journal of Herpetology 59, 3; 10.1670/2678037
Discussion
Although it remains difficult to distinguish which factors cause tadpoles to cannibalize eggs, our results provide support for the nutrient benefit hypothesis and evidence against the competition reduction hypothesis as a widespread explanation. However, outliers also indicate that causal factors may differ between anurans performing this cannibalistic behavior. Regardless of the cause, our study on R. chiricahuensis egg mortality indicates tadpole cannibalism can be critical to anuran population dynamics and their accurate modeling.
Our analysis shows that conspecific oophagy occurs predominantly in ephemeral lentic water bodies. The spatial extent of these lentic water bodies where oophagy occurred was usually not reported. Size of the local habitat can have a positive effect on nutrient abundance. Most water bodies associated with con-specific oophagy have ephemeral conditions and presumably are relatively small in area. Although measurements of nutrient concentrations were not reported, ephemeral lentic water bodies are typically nutrient poor (Northington and Webster, 2017). This pattern supports the nutrient benefit hypothesis. As noted by the first researchers to observe conspecific oophagy by tadpoles in ephemeral lentic water bodies, egg masses likely serve as important nutrient pulses to developing tadpoles in these environments (Crump, 1983). Although our observations took place in a lotic system, that site and other outliers in which conspecific oophagy occurred in nonephemeral or lotic environments may better resemble ephemeral lentic environments than would be expected. For example, our observation of conspecific oophagy by R. chiricahuensis tadpoles was in a stream with an upper portion drying out. Flow continued, but nutrient input may have decreased. Although it is possible that conspecific oophagy occurs more frequently in ephemeral lentic environments due to increased contact with eggs, anurans in lotic environments often restrict reproduction to small sections of these habitats, which could facilitate frequent contact between eggs and tadpoles in these conditions as well (Kupferberg, 1996).
Another potential supporting piece of evidence we found for the nutrient benefit hypothesis is the tendency for tadpoles at advanced Gosner stages to perform conspecific oophagy. A previous study found no positive growth impacts for tadpoles feeding on conspecific eggs, but tadpoles in that study were in early developmental stages. The authors suggested that later-stage tadpoles could benefit from conspecific oophagy due to ontogenetic diet changes (Summers, 1999). Our finding that conspecific oophagy almost exclusively occurs past a threshold of Gosner stage 31 of tadpole development supports this idea. Additionally, in the R. chiricahuensis population we studied, conspecific tad-pole size was a better predictor of egg mortality than conspecific tadpole abundance. However, an ontogenetic diet shift leading to oophagy may not be exclusively due to changes in nutrient needs and could simply be driven by a developmental increase in gape or development of the chondrocranium, which can facilitate tadpole macrophagy. Experimental work could elucidate why cannibalistic tadpoles are typically in advanced Gosner stages. Excluding tadpoles from eggs in cages with various mesh sizes would isolate the effect of size. Moreover, to directly test the nutrient benefit hypothesis, experiments should measure nutritional benefit and quantify effects on the growth rate in tadpoles that feed on conspecific eggs.
Our findings on the importance of tadpole developmental stage to the occurrence of conspecific oophagy also potentially refute the competition reduction hypothesis. The substantial developmental difference between predated eggs and predatory tadpoles, along with evidence for ontogenetic diet changes occurring, implies that if conspecific oophagy did not occur, there would likely be minimal competitive impact on tadpoles who did not perform the behavior. This is because larger and more developed tadpoles typically have competitive advantage in intraspecific competition (Chen et al., 2001). Therefore, we would not expect conspecific oophagy to be closely linked to development if competition reduction was the primary benefit to tadpoles.
In addition to indicating influences of specific abiotic and biotic conditions in favoring conspecific oophagy, our findings also confirm that the behavior is spatially and taxonomically widespread. Although we found no reports from Africa, we suspect this finding is due more to spatial biases in research and publications rather than a lack of occurrence in the region. The three most prominent anuran families exhibiting conspecific oophagy in the larval stage (Ranidae, Hylidae, and Bufonidae) all occur in Africa. Furthermore, all prerequisites for conspecific oophagy by tadpoles are present within the continent. Just focusing on Northwestern Africa, many native anurans exist in an ecological guild in which eggs and tadpoles often occupy lentic habitats, especially ephemeral lentic habitats (Escoriza and Ben Hassine, 2017). Moreover, tadpoles of a member of that guild, Mediterranean Painted Frogs (Discoglossus pictus), have been reported performing conspecific oophagy in Europe (Licata, Anzá, and Mercurio, 2015). Regardless, these findings, especially our finding that conspecific oophagy by tadpoles is taxonomically widespread, add to a growing body of literature reporting tadpoles to be more diverse in diet than previously thought (Montaña et al., 2019).
We found that most species known to perform conspecific oophagy as tadpoles share similar larval and reproductive traits (Fig. 1A). Guild membership was typically benthic. Jaw sheaths and labial teeth arrangement of benthic tadpoles may influence the ability to perform conspecific oophagy. Habitat use by benthic tadpoles could be an alternative or additional reason for the frequency of benthic tadpoles performing conspecific oophagy because it may facilitate encounters with conspecific eggs (Altig and Johnston, 1989). Regardless, we also found exceptions to the general multivariate trait grouping of species recorded performing conspecific oophagy, and these exceptions pose questions about whether the behavior may have the same general cause (nutritional benefit) while varying in form or whether other factors drive the behavior in some anurans, such as competition being a valid cause for a subset of cannibalistic tadpoles.
One notable exception was leopard frogs of the American Southwest. According to our PCA, leopard frog tadpoles have longer larval periods and reach larger sizes than other anurans known to perform conspecific oophagy (Fig. 1A). These trait differences along with the fact that leopard frogs utilize aquatic environments of permanent hydroperiods and lotic conditions (Degenhardt, Painter, and Price, 2005) may indicate that these tadpoles consume conspecific eggs for nutrient benefit but for different reasons than most anuran species. Instead of being a critical singular or limited pulse helping tadpoles with fast larval periods reach metamorphosis, conspecific eggs may be regular pulses that benefit leopard frog tadpoles for longer survival in the larval stage, including for overwintering.
The other main outlier group in our PCA consisted of many bufonids. These differ from other anuran species based on their larger clutch sizes (Fig. 1A). As previously discussed, evidence for tadpole competition reduction as a result of con-specific oophagy exists in one species of Bufonidae, R. marina (Crossland et al., 2011). This previous evidence and their distinguishing trait of clutch size indicates that, unlike for most anurans, competition reduction may be the primary causal factor of conspecific oophagy in R. marina and other bufonids.
Overall, outlier anuran groupings in our PCA raise questions for future research on differences in this behavior between anuran groups. At the same time, it is likely that additional anuran species perform conspecific oophagy as tadpoles but have not yet been documented doing so. Inclusion of such species could alter our results, potentially causing existing groupings to disappear or become more clearly separated. Thus, to understand potential alternative causal factors and strategies involved in conspecific oophagy by tadpoles, future work should continue to add species records of the behavior through field observations and laboratory experiments. Investigations should especially focus on identifying the behavior in species that are intermediate between our PCA groupings and outliers.
Because nearby conspecific tadpole size was the only statistically significant predictor, our multiple logistic regression analysis on R. chiricahuensis provides evidence that conspecific oophagy by tadpoles was the greatest source of egg mass mortality during this time frame for the R. chiricahuensis population we studied. It is unlikely that other potential egg mass predators found in the past at this site (water bugs, water boatmen, and small fish) were responsible for the mortality events we recorded, which were eliminations of entire egg masses. Water bugs consume ranid eggs but only in small amounts (Henrikson, 1990). Water boatmen consume anuran eggs even less (Henrikson, 1990). Although fish predate on some anuran eggs, studies have found little consumption of ranid eggs (Grubb, 1972). This, however, raises the question of what attributes allow tadpoles to readily consume ranid egg masses whereas other potential predators cannot. Future research should investigate exactly how tadpoles consume conspecific eggs and why some anuran egg predators are more successful than others.
Additional support that R. chiricahuensis egg predation in our study population is from conspecific tadpoles and not other predators comes from findings on a population of river-ine Foothill Yellow-Legged Frogs (Rana boylii). Egg predation was rare in that population unless dry conditions exposed eggs to terrestrial predators (Kupferberg, 1996). None of our R. chiricahuensis egg masses were exposed in this way. The study on R. boylii did find high egg mortality due to stochastic flooding scouring egg masses (Kupferberg, 1996). As previously mentioned, a similar degree of flooding mortality occurred in our studied R. chiricahuensis population at the end of our study. This fact indicates flooding is a potentially high source of egg mortality in riverine-breeding frog populations. However, our analysis supports that most egg mortality in this R. chiricahuensis study resulted from cannibalism by conspecifics. Though we sampled a subset of annual reproductive output, we suggest that this rate of cannibalism may occur over much of the reproductive season. There were relatively low stream flows over most of the reproductive season (March 2022 to May 2022), and large tadpoles were present throughout this time (Fig. 3). This finding has potentially important consequences and implications for understanding anuran population dynamics.
Embryo and early tadpole survival rates can be highly sensitive and elastic vital rates of anuran population dynamics, affecting population growth rate in American Bullfrogs and other species (Govindarajulu, Altwegg, and Anholt, 2005; Terrell et al., 2023). Thus, if conspecific oophagy reduces egg abundance and lowers embryonic survival, the behavior could impact population dynamics. A study on cannibalism in R. marina tadpoles found decreases in the survival of 99% of eggs and immobile hatchlings (DeVore, Crossland, and Shine, 2021). We showed conspecific oophagy by R. chiricahuensis eliminated entire egg masses, and our model results indicated it was the greatest source of egg mortality at our site, excluding the stochastic flood event. Thus, our study offers more support for the claim that conspecific oophagy can have substantial impacts on embryo survival and anuran population dynamics. Population models on species known to perform the behavior should therefore consider its effects on vital rates and overall population growth rate. Better understanding in general of cannibalistic interactions and cannibalism frequency will produce more accurate population forecasting (Govindarajulu, Altwegg, and Anholt, 2005). To determine its impact on anuran population dynamics, population modeling should be done comparing dynamics with and without consideration of mortality from conspecific tadpoles.
Incorporating the behavior into population modeling is particularly relevant to anuran conservation translocations and reintroductions. Many factors influence translocation and rein-troduction success (Wren et al., 2023), and determining which life stages to stock, at what times, and at what frequencies can be a critical factor in establishment success and new population viability (Scheele et al., 2021; Hossack et al., 2022; Thompson, Lundskog, and Dittmer, 2022). For anurans that perform con-specific oophagy as tadpoles, consideration of this behavior and incorporation into models can help address these questions.
For example, our focal species, R. chiricahuensis, is a species of conservation concern, and conservation translocations and reintroductions have been ongoing in the species’ historical range since at least 1995 and have varied in success (Hossack et al., 2022). Monitoring of R. chiricahuensis translocations and reintroductions shows stocking tadpoles increases reintroduction and translocation success when compared to stocking other life stages (Hossack et al., 2022). On the other hand, stocking egg masses has had low success. There could be sources of mortality of these masses, like conspecific oophagy by tadpoles, that have not been accounted for.
The statistically significant negative correlation between conspecific tadpole size and egg survival in R. chiricahuensis indicates that, at least in some systems, conspecific oophagy has an influence on the number of R. chiricahuensis eggs that hatch. Addressing these mortality sources could enhance the viability of stocking egg masses for translocations and rein-troductions. For example, after determining the size at which tadpoles become a cannibalistic concern, cages of specific mesh width could be deployed to exclude these potential egg predators. Another consideration is that reintroduction success from tadpole stocking may be increased by accounting for tadpole dietary needs and ontogenetic shifts. Although experiments are needed, conspecific oophagy by tadpoles may increase tad-pole survival rates and could have carry-over effects to post-metamorphic life stages. Thus, managers may want to consider supplementing a population that already has tadpoles present by providing eggs for those tadpoles to consume.
Understanding and considering implications of conspecific oophagy by tadpoles may also help manage introduced anuran species. Some species that we found perform conspecific oophagy as tadpoles, including R. marina, American Toads (Anaxyrus americanus), Midwife Toads (Alytes obstetricans), and D. pictus, are introduced species in some regions (Crossland et al., 2011; Escoriza, Ben Hassine, and Boix, 2014; Kelly et al., 2017; Goodman et al., 2023). Incorporating conspecific oophagy by tadpoles into population and planning models could benefit research by determining the most efficient life stages on which to focus removal efforts (Govindarajulu, Altwegg, and Anholt, 2005).
Species groupings in plots from a PCA of 6 reproductive and larval traits of 34 anuran species recorded performing conspecific oophagy in the tadpole stage. (A) Projection of PC II onto PC 1. (B) Projection of PC III onto PC I. (C) Projection of PC III onto PC II. Ellipses indicate grouping or outlier: red = majority of species; yellow = leopard frogs of the American Southwest; green = most bufonids; purple = Rana catesbeiana; and blue = Mixophyes balbus. Symbols indicate guild assigned to the species: open upside-down triangle = benthic; open triangle = benthic-arboreal; filled triangle = benthic-facultative carnivore; filled circle = carnivore; filled square = nektonic; open diamond = semiterrestrial.
Conspecific oophagy by Rana chiricahuensis tadpoles observed in Gila National Forest, New Mexico, USA. Two direct observations were made on (A) 23 May 2022 and (B) 28 May 2022.
Hydrograph from the San Francisco River in Glenwood, New Mexico, USA, from 1 March 2022 to 30 June 2022, a range that captures the first reproductive pulse of Rana chiricahuensis at our site. This is the closest US Geological Survey water monitoring location to our study creek. Data are from the US Geological Survey National Water Information System. The earliest known historical oviposition at our site is in April, but tadpoles can be present throughout this time period because they can overwinter at the study site. It is probable that cannibalistic tadpoles in our study are from the previous reproductive season.
Contributor Notes
