Do Adult Frogs Remember Their Lives as Tadpoles and Behave Accordingly? A Consideration of Memory and Personality in Anuran Amphibians

Abstract

Memory is a fundamental neurological function, essential for animal survival. Over the course of vertebrate evolution, elaborations in the forebrain telencephalon create new memory mechanisms, meaning basal vertebrates such as amphibians must have a less sophisticated system of memory acquisition, storage, and retrieval than the well-known hippocampal-based circuitry of mammals. Personality also appears to be a fundamental vertebrate trait and is generally defined as consistent individual behavior over time and across life history stages. In anuran amphibians (frogs), personality studies generally ask whether adult frogs retain the personality of their tadpole stage or whether personality shifts with metamorphosis, an idea behavioral ecologists term adaptive decoupling. Using a multidisciplinary perspective and recognizing there are ~7843 species of frogs, each with some molecular, morphological, physiological, or behavioral feature that makes it unique, we review, clarify, and provide perspective on what we collectively know about memory and personality and their mechanisms in anuran amphibians. We propose four working hypotheses: (1) as tadpoles grow, new telencephalic neurons become integrated into functional networks, producing behaviors that become more sophisticated with age; (2) since carnivores tend to be more bold/aggressive than herbivores, carnivorous anuran adults will be more aggressive than herbivorous tadpoles; (3) each amphibian species, and perhaps life history stage, will have a set point on the Shy–Bold Continuum; and (4) around this set point there will be a range of individual responses. We also suggest that several factors are slowing our understanding of the variety and depth of memory and personality possibilities in anurans. These include the scala natura approach to comparative studies (i.e., the idea that one frog represents all frogs); the assumption that amphibians are no more than simple reflex machines; that study species tend to be chosen more for convenience than taxonomic representation; and that studies are designed to prove or disprove a construct. This latter factor is a particular hindrance because what we are really seeking as scientists is not the confirmation or refutation of ideas, but rather what those ideas are intended to produce, which is understanding.

1. Introduction

1.1. Memory

Memory, defined as the storage and use of learned knowledge [1] or more generally the ability to use the past in the service of the present [2], is a fundamental neurological function essential for animal survival [3]. But the ability to remember does not imply its neurological basis is the same across the animal kingdom, or even across the seven classes of vertebrates. Murray and colleagues [4] point out that because memory appears similar among taxa, researchers have tended to use common descriptions of comparable phenomena, giving the erroneous impression there are shared memory circuits and processes. Comparative neuroanatomical studies demonstrate this cannot be true [5,6]. Instead, Murray et al. [4] suggest that in the hundreds of millions of years of vertebrate evolution, elaborations in the telencephalon have created new memory mechanisms culminating in the mammalian hippocampal formation and its neocortical connections. This perspective implies that basal vertebrates such as fishes and amphibians have a less sophisticated system of memory acquisition, storage, and retrieval than the well-known circuitry of mammals.

1.2. Personality

The insights Murray et al. [4] offer about vertebrate memory mechanisms can also be applied to animal personality. The concept of animal personality is one component of a recent behavioral ecological approach that features syndromes or constructs intended to group different types of animals based on ethological similarities [7,8]. Personality is generally defined as consistent individual behavior over time and across life history stages [9,10,11]. In amphibians, most personality studies have asked whether personality remains consistent or changes across metamorphosis. The results have been mixed: roughly half demonstrate behavioral consistency [9,10,11]; the remainder show personality shifts [12,13,14,15]. Here we explore these two perspectives.

Given the everyday human-based use of the term ‘personality,’ the behavioral ecological application of the term to amphibians is at first disconcerting. Certain characteristics of human personality, for example, empathy, humility, patience, modesty, and integrity [16], do not readily align with observations of basal vertebrate behavior. In our experience [17,18,19] and in the literature [9,10,11,12,13,14,15], personality in amphibians tracks along a two-dimensional axis ranging from aggressive/daring/exploratory on one end to timid/reluctant/reticent on the other, known to both psychologists and behavioral ecologists as the Shy–Bold Continuum [20]. Paralleling Murray et al.’s [4] observations on memory mechanisms, the casual application of the term ‘personality’ across vertebrates has given the erroneous impression that personality processes and expressions are shared across vertebrate classes. However, as vertebrates evolved, the mammalian, and particularly the human cortex (both allocortex and neocortex centered in limbic regions and prefrontal cortex), created new personality mechanisms that dramatically increased the range of expression available to organisms, vertebrates in particular.

The large differences in the neuronal basis and concomitant range of expression between the personalities of amphibians and humans leads to some unusual perspectives when applying the term personality without qualification to amphibians. For example, if after narrowly escaping being eaten by a snake, a bold-tending frog suddenly becomes timid, can it be said to have developed a personality disorder? Indeed, some researchers appear so uncomfortable with applying the term ‘personality’ to amphibians they avoid its use [15].

1.3. Anuran Diversity

Using a multidisciplinary perspective, we review, clarify, and provide perspective on the insights scientific studies have revealed about memory and personality and their mechanisms in anuran amphibians. But before we begin, it is important to comprehend the morphological and ecological diversity anurans represent. Anuran amphibians comprise 57 families encompassing ~7843 species [21] (in comparison there are ~6750 species of mammals [22]). Each frog species has some feature—molecular, morphological, physiological, or behavioral—that makes it unique [21,23]. Anuran novelty includes aquatic, terrestrial, arboreal, or subterranean lifestyles; locomotory capabilities that include swimming, walking, jumping, and gliding; internal or external fertilization; egg masses that are laid aquatically, terrestrially, or arboreally; parental care that consists of maternal provisioning, egg carrying, tadpole transport, and egg incubation; generalist or specialist approaches to feeding; pigmentation patterns that emphasize camouflage, are aposematic, or vary to match the background; calling synchronously or asynchronously; philopatry; toxicity; cocoon building; and water conservation measures that use behavior (microsite selection) or behavioral physiology (e.g., waxy lipid secretions the frog spreads over its body) [24,25,26]. This adult variation is compounded by differences in tadpole morphology, behavior, and ecology, which also varies widely [27,28,29,30,31,32]. For example, Orton [27] describes four basic types of tadpoles (see also [30]) while Amin and colleagues [31] have proposed an additional Type V based on the tadpoles of Lepidobatrachus laevis.

1.4. Anuran Life History Diversity

The basal anuran life history pattern consists of an egg that hatches into an aquatic, typically algivorous larva that after some period undergoes a rapid metamorphosis into a terrestrial, carnivorous juvenile that matures into an adult. But this pattern does not hold for all species [32,33]. For example, in some species (e.g., in the genus Xenopus) aquatic larvae metamorphose into aquatic adults. In other species (e.g., the genus Eleutherodactylus) free-living tadpoles never form; instead both embryonic and larval life history stages are completed inside the egg capsule (i.e., hatching is delayed until metamorphosis) [34]. In a variation of this approach, females of the two now-extinct Gastric Brooding Frogs (genus Rheobatrachus [35]) swallowed their eggs and raised their tadpoles in their stomachs. And some free-living tadpoles (e.g., Zaire Dwarf Clawed Frogs, Hymenochirus boettgeri; Seep Frogs, Occidozyga baluensis; [36,37]) are carnivorous. Each of these life history variations offers untapped insights into our understanding of memory and personality in amphibians.

2. Anuran Memory

Larval and adult life history stages of anuran amphibians with complex life histories develop and evolve autonomously in response to functional/ecological demands [27,28,29,30,31,32]. This phenomenon is called adaptive decoupling [7,10] and in part forms the basis of ‘Starrett’s Rule,’ which states that the plainest adult frogs often have the most bizarre tadpoles while the strangest adults typically have the most mundane tadpoles [38,39]. Such divergent life histories question the value of long-term memory, or the storing of learned knowledge [1] in these animals.

Field observations and laboratory experiments demonstrate that both tadpole and adult anurans form memories. For example, tadpoles remember the odor of predators (kairomones) and alarm cues (Schreckstoff) [40,41,42,43,44,45,46,47,48,49]; adult Crawfish Frogs (Rana areolata) remember the location of their primary burrows [50,51]; adult male American Bullfrogs (Rana catesbeiana) remember and habituate to the breeding calls of adjacent calling males [52]; and female Jamaican Laughing Frogs (Osteopilus ocellatus) must remember the locations of the epiphytic bromeliad clusters hosting their arboreal tadpoles so they may return to provision them with trophic eggs [53].

The proportion of its lifespan a frog spends as a free-living tadpole, which influences the potential for tadpole memories to influence adult behavior, ranges from zero to substantial, depending on species and location. For example, in the United States six species of Eleutherodactylus frogs exhibit direct development [18]. As mentioned above, these animals delay hatching until metamorphosis, meaning an individual’s embryonic and larval stages are contained within an egg capsule. The only memories these tadpoles can form are associated with being confined to an egg. A similar statement could have been made about the tadpoles living in the stomachs of the two now-extinct species of gastric brooding frogs in the genus Rheobatrachus [35]. Such memories can be of little value following metamorphosis when juveniles are free-living and navigating a landscape, finding prey, and avoiding predators.

Among U.S. anuran species with free-living tadpoles, larval periods can be as short as 7.5 days (e.g., Couch’s Spadefoot Toads, Scaphiopus couchii [54,55]). Since S. couchii can live up to 13 years in the wild [55,56], their larval period, and thus the time available to form larval memories, represents a minute 0.15% of their lifespan. In contrast, the tadpole stage of Southern Mountain Yellow-legged Frogs (Rana muscosa) and Sierra Nevada Yellow-legged Frogs (Rana sierrae), species adapted to high-elevation montane lakes, can extend up to four years in the northern portions of their range [57,58]. Since both species of yellow-legged frogs live over a decade in nature [59,60] their larval period and the memories formed during this time can encompass as much as 40% of their lifespan.

Whether their tadpole life history stage represents a large or small proportion of their lifespan, little about tadpole life needs to be remembered following metamorphosis. For example, in the seasonal or semi-permanent wetlands most species of tadpoles inhabit, the only relevant navigational challenges are swimming between deep and shallow water to thermoregulate [61], and swimming from poor to rich algal beds to feed [40]. Such movements are likely reactions to proximal temperature and olfactory cues, respectfully, and therefore need not be remembered. In contrast, larval memories with post-metamorphic value include visually mediated experiences centered on detecting the presence of predators [62]. Most of the collective tadpole sensory apparatus is geared toward detecting and avoiding predators [63,64]. Avoiding detection is likely the best defense a tadpole employs, because even with the escape advantages given by their Mauthner cell-based C-starts [65,66,67,68], tadpole swimming is so inefficient that for most species once a predator is locked onto a specific tadpole, its probabilities for survival become low.

In anurans, metamorphosis typically occurs rapidly, within hours or days, and as such challenges a hardwired vertebrate nervous system. During metamorphosis, tadpole primary sensory and motor systems and their secondary processing centers are lost or reworked and rewired [69,70]. For example, tails and fins are resorbed; mouths are reconfigured for carnivory; forelimbs emerge from beneath opercular flaps; lateral line mechanoreceptive organs and their peripheral nerves are lost; the motor systems underlying aquatic suction feeding and undulatory swimming are reconfigured for biting and walking/jumping; the gut transforms from a double-coiled tube into a specialized digestive system consisting of stomach, intestine, and colon; and animals become temporarily deaf as the auditory apparatus forms ossicles and a tympanum [63,71,72].

The brain becomes modified as well. At metamorphosis the anuran telencephalon elongates and the diencephalon thickens [69,70,73]. These gross morphological differences are the product of neurogenesis and the radial migration of new neurons away from periventricular growth zones [69,70]. The telencephalon, being terminal, has the cranial capacity to add neurons rostrally and thus elongates, while the diencephalon, being bounded rostrally by the telencephalon and caudally by the midbrain, adds neurons that radiate and therefore widens. Neurons generated at or after metamorphosis are not likely to be involved in the retention or recall of tadpole memories.

Given the vast differences between tadpoles and post-metamorphic frogs in morphology, behavior, ecology, and brain capacity, it is reasonable to assume not much of what a tadpole remembers carries forward after metamorphosis. The rationale here is twofold. First, as mentioned above, little of what a tadpole learns as a swimming herbivore offers fitness benefits to a jumping carnivorous juvenile or an adult. Second, considering the high metabolic costs of maintaining a vertebrate central nervous system, retaining unnecessary memories imposes an energetic burden. This logic suggests a metamorphosing frog should exercise selective retrograde amnesia and purge any now superfluous memories of its aquatic larval life. Karpicke and Coverdale [74] address the advantages of forgetting former ecological relationships:

“If a location no longer has food, if a water supply has dried up, or if a predator that prowled an area is now deceased, it would be costly to continue to remember that there had once been food, water, or a predator in those locations. Thus, it may be adaptive to prioritize forgetting of those fitness relevant items.”

Consistent with this interpretation, recent studies suggest anuran personality can transform at metamorphosis [10,15]. We discuss this phenomenon further below.

Modern research has re-kindled interest in the advantages of forgetting. For example, Fawcett and Hulbert [75] offer “Every-day forgetting represents a feature of a well-designed memory system.” And, as pointed out above, Karpicke and Coverdale [74] propose that for it to work optimally, a memory system should jettison information that is outdated or inappropriate under current conditions. Further, Nørby [76] suggests that forgetting ensures that an animal orients information towards the present and the future, and that by performing so facilitates its ecological sensitivity.

3. The Nature of Our Knowledge of Memory

Much of what we understand about memory derives from the human experience as talking, feeling, thinking beings. Humans possess declarative memories for names, facts, and events, and procedural memories for motor patterns such as riding a bicycle or playing a musical instrument. Our declarative memory has short- and long-term components mediated by a variety of cell types connected through a complex neural circuitry in the telencephalic hippocampal formation.

Memories are acquired through interoceptive and exteroceptive sensory systems. In humans, and likely all mammals, short-term or working memories lasting seconds to minutes are held in primary, secondary, and tertiary sensory cortices in communication with frontal lobe association cortex. These memories involve ongoing electrical activity of neurons, alterations in intracellular Ca²⁺ and other ions, and changes in second messenger systems. Long-term memories, in contrast, are uploaded by the allocortical hippocampal formation located in the medial portion of the temporal lobe. Long-term memories rely on protein phosphorylation and other covalent modifications as well as changes in gene transcription and translation resulting in structural changes in proteins and neurons (summarized in [77]). One school of thought suggests memories are stored in the form of discrete neural substrates called engrams [1,2,78].

While the hippocampal formation is responsible for the uploading and retrieval of declarative memories, it does not store memories. As mentioned above, long-term memories in humans and other mammals are stored in heteromodal neocortices associated with the sensory systems employed in acquiring the memory. In humans, this process of uploading, storing, and recalling memories is disrupted in diseases such as Alzheimer’s, Lewy Body, and Frontotemporal dementias [79,80].

Memory persistence is tied to its emotional salience in humans, with both positive and negative experiences searing memory (although we cannot recall pain sensations or perceptions) [81,82,83]. Positive events associated with rewards that promote the survival of the individual or the species are mediated at least in part through the striatal nucleus accumbens using the neurotransmitter dopamine, while negative experiences are mediated through the amygdaloid complex using a variety of neurotransmitters including glutamate, serotonin, and noradrenaline, which are released in response to stress, and GABA (gamma-aminobutyric acid), an inhibitory neurotransmitter.

4. The Amphibian Neural Substrate for Memory

In amphibians the telencephalon is divided into a dorsal pallium and a ventral subpallium [84]. The pallium is divided into medial, lateral, ventral, and perhaps dorsal nuclei [6,85,86,87,88,89]; Figure 1B–F. The medial pallium (Mp in Figure 1B–F) is thought to be a homolog of the amniote hippocampus and subiculum [88,90]—the structures that, based on the ‘homologous nuclei have homologous functions’ rule [6], upload long-term memories but probably do not store them [6,88,89,90,91]. Working with the frog Discoglossus pictus (Alytidae) and the salamander Plethodon jordani (Plethodontidae), Westhoff and Roth [90] demonstrated that neurons in the ventral medial pallium project to the ipsilateral dorsal pallium (Dp in Figure 1C–F), while neurons in the dorsal pallium reciprocally connect to the ipsilateral medial pallium as well as the lateral septum, nucleus accumbens (Acc in Figure 1C), and medial amygdala (Apm in Figure 1D,E). Neurons in the dorsal medial pallium project bilaterally to widespread telencephalic and diencephalic nuclei. These results suggest the ventral medial pallium is the true homolog of the mammalian hippocampal formation and the dorsal pallium is homologous to mammalian heteromodal association cortex. Addressing this question, Westhoff and Roth [90] offer: “…the most parsimonious interpretation is that [amphibians] have a bipartite medial and a bipartite dorsal pallium and that this represents the ancestral tetrapod condition.” Striedter and Northcutt [6] call into question the presence of a dorsal pallium in amphibians, but memories in amphibians must be stored somewhere, and the telencephalic regions reciprocally connected to the ventral medial pallium are the most likely candidates (for details of amphibian pallial connections see also [88,91]).

The amphibian subpallium consists of a medial septum [92,93,94,95,96] and a lateral region containing the basal ganglia (in part Strd, Strv, Strc in Figure 1C,D) and the amygdaloid complex (Apl and Apm in Figure 1C,D) [97,98]. Unlike memory circuits, the circuitry of the basal ganglia in anurans is essentially identical to its circuitry in all vertebrates [5,99,100]. In humans, breakdowns in basal ganglia circuitry cause diseases such as Parkinson’s and Huntington’s. Through its ability to initiate and terminate motor patterns, the basal ganglia in amphibians appear to be the highest-order brain nuclei controlling motor function (the extrapyramidal circuit of Loonen and Ivanova [101]).

Located partially in the pallium, partially in the striatum of the basal ganglia, the amygdaloid complex is typically divided into several regions that collectively receive direct inputs from both conventional and vomeronasal olfactory systems, and indirect inputs from visual, auditory, somatosensory, and gustatory nuclei filtered through the dorsal thalamus [102]. Amygdaloid nuclei in turn send outputs to the hypothalamus where they effect autonomic, somatic, and endocrine responses [103,104,105,106].

Also located in the striatum and considered to be part of the outer shell of the amygdaloid complex, the dopaminergic nucleus accumbens (Acc in Figure 1C) traffics in goal- and reward-directed behavior [107]. The nucleus accumbens receives multimodal inputs from external and internal senses [108] and sends outputs to the amygdala, hypothalamus, and other forebrain and brainstem nuclei, including the serotonergic raphe nuclei [109].

The anuran amygdaloid complex, including the nucleus accumbens, plays an important role in the salient labeling of experiences as either threatening (amygdala; [101]) or beneficial (nucleus accumbens; [107]). It is presumed that through connections with the medial pallium, these qualitative labels are remembered and condition future behavioral responses when experiencing similar situations [91,103,110,111,112,113,114,115].

In addition to these telencephalic memory pathways, cells in the amphibian midbrain [62] and hindbrain [116] form memories, reinforcing the idea that many different types of memory mechanisms exist in the vertebrate central nervous system.

The components of the amphibian telencephalon responsible for memory are much simpler histologically and connectively than the components of the mammalian hippocampal formation, as follows:

-

While the mammalian hippocampal formation comprises two, three-layered allocortices containing specialized cell types including pyramidal, granule, and polymorphic neurons, the amphibian ventral medial pallium is unlaminated and composed of comparatively homogeneous neurons [6,88,90].

-

While the mammalian hippocampus contains specialized place cells [4], as well as landmark-vector cells [117], grid cells, boundary cells, and head-direction cells that affect the activity of place cells [118,119], the mammalian medial pallium is composed of neurons that are relatively homogeneous morphologically [6,88,90]. It has yet to be determined if they have specialized functions.

-

While mammals have direct perforant and indirect alvear pathways through the hippocampal formation that form complex connections with heteromodal association cortex, amphibians have relatively simple reciprocal connections between the medial pallium and other forebrain regions both ipsilaterally and contralaterally [88,90].

5. Behavioral Evidence for Amphibian Memory

Adult frogs and toads employ at least three types of non-procedural memory: habituation, reinforcement learning, and navigational learning [4]. Habituation occurs when a repeated stimulus is neither beneficial nor threatening it becomes ignored. It is the simplest behavior requiring memory. Most animals habituate, including primitive invertebrates such as the compost pile-inhabiting nematode Caenorabdis elegans [120]. Among anuran amphibians, for example, adult male American Bullfrogs (Rana catesbeiana) habituate to the breeding calls of adjacent “neighbor” males, known as the “Dear Enemy” effect [52].

Both reinforcement and navigational memories fall broadly under the category of declarative memories. Reinforcement memories tag experiences with valence values (reward or penalty) and when recalled under novel but similar experiences guide subsequent behavior in a presumably adaptive way [4]. Among the best examples of reinforcement learning in amphibians are memories of kairomones (predator odors) and Schreckstoff (alarm cues) in tadpoles [40,41,42,43,44,45,46,47,48,49]. In lab experiments with adult frogs and toads, Argentine Toads (Bufo arenarum) learned tasks based on water rewards [89]; African Clawed Frogs (Xenopus laevis) and Cane Toads (Bufo marinus) learned to avoid negative stimuli (i.e., electric shock) [121,122], and American Bullfrogs (Rana catesbiana) learned to inhibit their righting reflex to avoid negative stimuli [123]. As with humans [3,124], there is evidence in tadpoles that emotional salience strengthens reinforcement memories and reinforcement memories can be reconsolidated [47].

Navigational learning relies on animals either having a cognitive map of their home range or a remembered sequence of landmarks that gets played out as they travel [4]—see also [125,126]. Examples of navigational learning in amphibians include recalling the locations of upland burrows in Crawfish Frogs (Rana areolata), optimal foraging habitat in Ozark Zigzag Salamanders (Plethodon angusticlavius), and bromeliad plants holding tadpoles that must be fed trophic eggs in hylid and dendrobatid species [50,51,53,127,128].

As pointed out above, amphibians form long-term reinforcement and navigational memories without the histologic specialization and circuit elaborations present in the mammalian hippocampal formation. Absent such complexity, how long can a long-term memory last in an amphibian? Most of the work on memory, cognition, and personality in anuran tadpoles has been performed on spring-breeding ranid species (Rana sylvatica in North America [46,47,48] and R. arvalis [15] and Pelobates cultripes tadpoles in Europe [49]), which complete their larval stage in a matter of weeks to no more than a few months [129]. This puts a limit on the potential duration of larval memories held by tadpoles.

Much less is known about memory retention in adult frogs. It can be inferred from maternal provisioning of trophic eggs to their arboreal tadpoles that female hylid frogs remember the locations of the bromeliad tanks, tree cavities, and other locations for times ranging up to weeks or perhaps months. Our work on Crawfish Frogs (Rana areolata), which are obligate crayfish burrow dwellers, demonstrates they return to their home burrow following breeding in wetlands that can be more than a kilometer distant. To accomplish this, females must remember the location of their burrows for at least a week or two, males for weeks up to a month or more [19]. Crawfish Frogs will occupy the same burrow for up to five years, but they need not remember the location of their burrow this long. Each year may constitute a separate event, with frogs only needing to remember the location of the burrow they recently left, days or weeks before. Supporting this conclusion, when Crawfish Frog adults abandoned burrows for new ones, they did not revisit the locations of their former burrows [19].

Given these observations on both tadpoles and adults, it may be that long-term memory in frogs lasts at most on the order of months, and much of what needs to be remembered centers on locations (egg deposition and retreat sites). This generalization is consistent with the neurological simplicity of their telencephalon [88,90]. It also suggests the anuran ability to assess and respond to current situations is more valuable than remembering former but now obsolete situations. As Karpicke and Coverdale [74] point out, a well-designed memory system should jettison outdated information that is inappropriate under current conditions.

One might think that anuran adults would benefit from remembering the location of breeding sites from year to year, but especially for members of the spring breeding guild, the fishless seasonal and semipermanent wetlands they use for reproduction are transient features of the landscape that change in availability and quality from year-to-year across decadal-long drought cycles [130]. A frog has no guarantee the wetland it bred in last year will be suitable for breeding this year: under drought conditions there may be no water; under hydric conditions fish may have colonized. When choosing where to breed, rather than relying on memory, which could lead them to a known but unusable site and waste that year’s breeding effort, reproductive success would be better served if frogs used current cues such as the availability of water, its depth, and the absence of fishes.

6. Anuran Personality

Personality in amphibians might best be defined as a position along the Shy–Bold Continuum [20]. Unlike memory, the neurology underlying personality cannot be explained using comparatively simple telencephalic nuclei and their connections. Instead, personality involves reflexive anoetic and cognitive noetic circuits [131] as well as hormonal and mood states, and perhaps experience. How so? An animal’s personality is the aggregate of its behaviors. In amphibians, behaviors can be reflexive [62] or cognitive [23,91,112,113,132]; they can also vary seasonally, especially under social situations such as breeding, when behavior is influenced by sex hormones [133,134,135]. Further, personality can be altered in ecological contexts such as when attacking prey or being attacked as prey when “fight-or-flight” hormones such as adrenaline alter behavior. As well, motivational hormones such as dopamine and serotonin are secreted and can influence moods and therefore personality [136,137,138,139,140].

The literature on amphibian personality is small and is in general agreement that personality broadens as tadpoles develop [10,69,70]. Results are split about whether personality is altered across metamorphosis. Some studies suggest it does [12,13,14,15], others that it does not [9,10,11]. Koenig and Ousterhout [7] report personality does not change across metamorphosis in Spotted Salamanders [Ambystoma maculatum], but this is not surprising. The difference between a limbed sit-and-wait, sometimes stalking carnivorous larval salamander and a limbed sit-and-wait, sometimes stalking adult salamander is not as great as between a swimming herbivorous frog tadpole and a hopping carnivorous frog adult.

Research conducted in Sweden [15] suggests in the northern portion of their range, Moor Frog (Rana arvalis) populations have bold tadpoles and timid froglets while southern populations have timid tadpoles and bold froglets. There are two takeaways from this study. The first is amphibian personalities can vary across populations, perhaps in response to learned local ecological challenges [9,10,48,141]. If true, to truly know the personality capacitance of an amphibian species, you must census or survey a range of populations. The second takeaway is given these behavioral gradients, at locations between these northern and southern populations there must be populations where tadpoles and adults exhibit similar moderate responses to ecological challenges.

To truly understand amphibian personality, questions should encompass more than single-species changes over time. For example, we can envision studies addressing comparative personalities between different species, for example, the personality of adult Eastern Gray Treefrogs (Hyla versicolor) compared to adult Northern Leopard Frogs (Rana pipiens). To our knowledge this research has not been undertaken, but if appropriate species are chosen, such studies might lead to greater behavioral–ecological insights than following a single species across metamorphosis.

7. A Few Considered Hypotheses

Given the variation in ecologies, life histories, and natural histories across thousands of anuran species and the potential personality traits necessary to successfully accommodate this variation, it becomes useful (similar to the approach used by Mettke-Hofmann [142]) to develop testable hypotheses about amphibian personalities based on neuroethological principles. As a starting point, we propose four hypotheses.

First, neuroanatomical studies show tadpoles are continuously adding neurons as they grow [69,70]. As these new telencephalic neurons become recruited and integrated into functional networks, the working hypothesis is tadpole behavior will become more sophisticated over time [10,48,141].

Second, since carnivores tend to be more bold/aggressive than herbivores, the working hypothesis is that carnivorous anuran adults will be more aggressive than herbivorous tadpoles. Here, it must be noted that while aggressiveness and movement are equated in some studies of amphibian personality, they need not be correlated. Most amphibian adults are sit-and-wait predators. They are either stationary (most of the time) or violently striking at prey (in lunges lasting milliseconds). In contrast, grazing tadpoles tend to swim at a slow, constant pace to always be feeding in new areas. Tadpoles move more (i.e., might be considered more aggressive) than adults sitting and waiting for prey; tadpoles move less (i.e., might be considered less aggressive) than adults striking at prey. Context matters.

Third, as mentioned above, we hypothesize each amphibian species will have a set point on the Shy–Bold Continuum. For example, in its simplest form, predator avoidance in anurans can be divided into tortoise and hare strategies. Most adult frogs use the hare strategy—a predator approaches, they begin aggressively jumping to get as far away as fast as they can. On the other hand, Crawfish Frogs use a tortoise strategy (using crayfish burrows much like a turtle uses its shell, as defensive armament [19]). During our field studies, while assessing Crawfish Frogs at drift fences surrounding breeding wetlands, we would pick up a frog, take a measurement, set the frog down, record the observation, pick up the frog, make another measurement, set the frog down, and so on. All Crawfish Frogs we observed behaved in this compliant and convenient way [19]. If we had been conducting the same study on syntopic Southern Leopard Frogs, the moment we set down the frog, it would have first jumped into the tallgrass prairie surrounding the wetland and in a few hops been lost to us. Tadpole behavior also situates along the Shy–Bold Continuum. Moving objects, overhead shadows, loud noises, big waves, or being touched cause tadpoles of some species to move (an escape reaction [143]), while tadpoles of other species will cease moving (to not draw attention to themselves [144,145,146]).

Fourth, we hypothesize that around the species set point on the Shy–Bold Continuum there will be a variation—a range of individual responses. During our two-year radiotelemetry study of Crawfish Frogs [50,51], at least one member of our lab visited the home burrow of each frog every day except during the winter. Three male frogs collectively exhibited the range of possible reactions to our presence. As we approached, one male (Shack) was usually outside his burrow and in response to either our presence or the beeping of the telemetry receiver would begin calling, an aggressive territorial response [147]. This male was behaving boldly. When we approached the burrow of a second male (460) we never saw the frog, but our wildlife cameras indicated he was almost always out of his burrow except when we approached; to avoid us he would dive into his burrow. This male was behaving shyly. When we approached a third male (Romeo), he acted as if we were not there. He did not vocalize, nor did he dive into his burrow to avoid us. This male was behaving indifferently.

8. Factors Inhibiting Our Understanding of Anuran Personality

Several factors have slowed our understanding of the variety and depth of responses to environmental challenges—including components of personality—anurans have at their disposal. The first has been the scala natura approach to comparative neuroethological studies, manifested in terminology such as “the fish,” “the frog,” and so on. This perspective has been severely criticized by Northcutt and Gans [148]; see also [6]. The literature is beginning to show cracks in this “one frog represents all frogs” approach to anuran research [91,112,113,149,150,151], but the ‘one for all’ notion remains an attractive paradigm in many disciplines.

Next is the assumption by influential early workers that amphibians are simple reflex machines [97,152]. Ewert [62] described how post-World War II neuroethologists envisioned this reflex-based system working. They believed key stimuli (innate releasing mechanisms) activated fixed patterns of behavioral responses, much like a key fitting a lock [153,154,155,156,157]. That is, once sensory systems identified an object and a location—say a prey item to the right at 45°, 15 cm distant—a pre-programmed “address” in the midbrain tectum was “dialed up” (the metaphor here being a rotary telephone, reflecting the technology of the day). Further, they felt individual tectal addresses existed for all possible combinations of stimulations and locations. This model of amphibian behavior relegates the telencephalon to a braking system, partially or totally inhibiting fixed motor responses by dampening them. In these studies neuroethologists allowed for learning (i.e., the acquisition of learned key stimuli triggering new motor programs). But rather than evoking contextual telencephalic memory circuits, they offered a convoluted explanation where novel stimuli triggered various subcomponents of pre-set brainstem circuits to produce the desired (i.e., adaptive) response [62,153].

Contrast this view with the late-eighteenth century conclusions of William James. James [158] found that absent a telencephalon, frogs acted as if they were “a machine… By applying the right sensory stimulus [to the animal]… we are almost as certain of getting a fixed response as an organist is of hearing a certain note when he pulls out a certain stop.” This nineteenth century description anticipates the neurophysiological experiments of the 1970s mentioned above. In contrast, in normal frogs James [158] observed spontaneous and complex acts of behavior “as if moved by what in ourselves we should call an idea.” In other words, the experimental work of 1970s neurophysiologists was accurate and true when considering the proximal mechanisms underlying the stimulus-response circuitry of amphibians, but by minimizing the role of the telencephalon and the cognitive abilities of anurans, they overlooked the decision-making processes ultimately responsible for amphibian behavior.

To illustrate this difference in perspective, it is necessary to consider the auditory system of American Bullfrogs (Rana catesbeiana). Simmons and Buxbaum [159] describe mating call processing and production in Bullfrogs by addressing auditory circuitry at the midbrain level (i.e., torus semicircularis). And much like the data from Ewert’s [62] experiments, while every stimulus-response result was repeatable and true in the laboratory, Simmons and Buxbaum [159] did not consider the complexity of mating call communication encountered in nature [132]. During their breeding season, American Bullfrog males gather in the shallows of a lake or permanent wetland and begin calling to attract females. Females arrive and choose a male based in large part on male call characteristics; they are especially attracted to the low frequency calls produced by large males [160,161]. Small males, however, will attempt to breed by intercepting females as they approach large males [160,161]. These small males, called parasites [160,161] or satellites [162], tend to be young, first-time breeders. The choice for these young male Bullfrogs is whether to call and defend a territory or to stay silent and become a satellite/parasite. They make this choice based on social factors including the number of old, large males present and the number of females available [159]. Following James [158], this ‘choice,’ is based on telencephalic circuitry and hinges on a population-level social context not incorporated into laboratory electrophysiological experiments [62,163].

A further factor slowing our understanding of the range and depth of responses to environmental challenges anurans have at their disposal is research subjects tend to be chosen more for convenience (i.e., availability and/or ease of care) than taxonomic representation. Again, it is the ‘one frog is just as good as another’ assumption. Why is Xenopus laevis so often the subject of experimental studies in frogs? It is not because they are a true representative of the Order Anura (they are not; X. laevis is about as far away from a generalized frog as an animal can be and still be a frog); it is because these frogs became widely available after the Hogben Test for human pregnancy was replaced by immunological assays. Also, colonies are convenient to maintain: the aquatic egg, larval, and adult life history stages of X. laevis require only aquaria; most other anuran species need both aquaria (for eggs and tadpoles) and terraria (for juveniles and adults). Perhaps the real issue here though is not so much the species chosen but a failure by researchers to situate their study species within the context of possible anuran life histories and natural histories. If you tally up the studies addressing amphibian personality considered here, four assert personality changes across metamorphosis [12,13,14,15], three assert no change [9,10,11]. If the ‘one-frog-is-all-frogs’ perspective is true, half these studies are wrong. If the results of all these studies are correct (a fact we accept; these manuscripts represent solid science and have been through peer review), the one-species-represents-all-species assumption is wrong.

There is another way to gain perspective on this issue. Given there are 7843 species of frogs, we can ask with what confidence can any one species of anuran be considered representative of the entire group? And if authors feel their species does represent the Order Anura, what criteria do they use (presented in the context of the morphological, behavioral, and ecological variation known to occur)? Any study that does not follow this protocol and claims their results represent all 7843 frog and toad species is likely to be proven wrong by subsequent studies on anurans with different life histories or natural histories.

The final factor slowing our understanding of the range and depth of responses to environmental challenges anurans employ is how behavioral syndromes have been considered by some workers. While such syndromes create useful and insightful hypotheses or models, they are constructs and therefore derivative ideas that become less grounded when fundamental biology is ignored. For example, Koenig and Ousterhout [7] report personality does not change across metamorphosis in Spotted Salamanders (Ambystoma maculatum). Based on this finding (on one species out of 8800 amphibians), they reject the adaptive decoupling model for amphibians. But the observation that personality does not change across metamorphosis in Spotted Salamanders is not surprising. As we point out above, the difference between a salamander larva, which is a limbed sit-and-wait, sometimes stalking aquatic carnivore, and a salamander adult, which is limbed sit-and-wait, sometimes stalking terrestrial carnivore, is not as great as between a swimming algivorous tadpole and a hopping carnivorous frog. That is, the only decoupling between a larval and an adult salamander is between an aquatic and a terrestrial animal, while decoupling in anurans entails this factor plus the enormous difference in morphology, physiology, and behavior that accompanies herbivory versus carnivory. The problem with syndromes is they can take on a life of their own and become the primary focus. As George E. P. Box, a modeler, once said: “all models are lies; some of them are useful” [164]. From this perspective, adaptive decoupling models are more useful when studying frogs than salamanders, but even when applying them to frogs, one should use caution in being too literal.

The questions of when and where adaptive decoupling occurs in anurans is too large for any single empirical or experimental study. Instead of designing such studies with the idea of proving or disproving this or that construct, we suggest studies be conducted with the notion that results are contributing to a larger dataset that will ultimately reveal which life history and natural history conditions either drive personality shifts or maintain stasis. Why? Because what we really seek as scientists is not the confirmation or refutation of ideas, but rather what those ideas are intended to produce, which is understanding.

9. Conclusions

The study of memory and personality in anuran amphibians is in its infancy. Hampered by (1) a scala natura perspective; (2) the assumption that amphibians are simple reflex machines; (3) study species chosen more for convenience than taxonomic representation; and (4) questions designed to prove or disprove constructs rather than achieve understanding, we have only begun to grasp the capacity for memory and personality exhibited by the ~7843 species of anurans. Here, we propose four testable hypotheses that when, applied across species representing the true morphological, behavioral, and ecological diversity of anurans, should serve as a starting point for such an understanding: (1) as tadpoles grow, new telencephalic neurons are recruited and integrated into functional networks and behavior will become more sophisticated; (2) since carnivorous animals tend to be more bold/aggressive than herbivores (i.e., carnivores initiate attacks, herbivores react in ways that increase their probability of surviving), carnivorous anuran adults will be more aggressive than herbivorous tadpoles; (3) each amphibian species will have a set point on the Shy–Bold Continuum which might be termed a species personality; and (4) around this Shy–Bold Continuum set point there will be a range of singular responses, reflecting individual personality.