Integrative and Comparative Biology, volume 50, number 4, pp.



Evolutionary Loss of Animal Regeneration: Pattern and Process
Alexandra E. Bely1
Biology Department, University of Maryland, College Park, MD 20742, USA
From the symposium ‘‘Animal Regeneration: Integrating Development, Ecology and Evolution’’ presented at the annual
meeting of the Society for Integrative and Comparative Biology, January 3–7, 2010, at Seattle, Washington.

Synopsis The ability to regenerate lost or damaged body parts is widespread among animals and provides obvious
potential benefits. It is therefore perplexing that this ability has become greatly restricted or completely lost in many
lineages. Despite growing interest in the cellular and molecular basis of regeneration, our understanding of how and why
regenerative abilities are lost remains rudimentary. In an effort to develop a framework for studying losses of regener-
ation, here I outline an approach for rigorously identifying such losses, review broad patterns of regenerative ability
across animals, describe some of the clearest examples of regeneration loss, discuss some possible scenarios by which
regeneration may be lost, and review recent work in annelids that is providing new insights into loss of regenerative

Introduction approach. It will be necessary to identify unambigu-
The evolutionary loss of regenerative ability repre- ous losses of regeneration through comparative
sents a fundamental and perplexing problem in bi- studies interpreted in a phylogenetic context, to
ology. Although the complex question of how many reveal the ecological correlates and developmental
times regeneration has evolved anew among animals basis of such losses, and to formulate and test
remains to be resolved, it is evident that regenerative specific models for how and why loss of regeneration
ability is widespread. Regeneration provides obvious occurs. Investigating a broad range of animals in this
benefits to an injured individual, yet regenerative way will also be essential in order to assess the extent
abilities appear to have been greatly restricted or to which regeneration loss proceeds in a similar way
completely lost in many animal lineages (Bely and across different groups.
Nyberg 2010). Why should regenerative abilities As a step towards focusing efforts on this fasci-
ever be lost and how do such losses occur? nating phenomenon, here I begin developing a
Although regeneration has long fascinated humans framework for identifying and investigating regener-
(Morgan 1901; Dinsmore 1991), undoubtedly in ation losses in the Metazoa, review available data
part because of our own frustratingly limited regen- bearing on this topic, and suggest important avenues
erative potential, and although the process of regen- for future research. Specifically, I outline how to
eration has now been analyzed in considerable detail identify regeneration losses rigorously; review broad
in a few model organisms (Sa´nchez et al. 2006; patterns of regenerative ability across animals; de-
Carlson 2007; Brockes and Kumar 2008), our under- scribe some of the clearest examples of regeneration
standing of why and how regenerative abilities are loss; discuss some possible scenarios by which regen-
lost (and possibly gained) through evolution remains eration may be lost; and review recent data in anne-
extremely limited. lids that are providing new insights. Although
Developing a good understanding of the ultimate regeneration can occur at a range of biological
and proximate mechanisms of the loss of regenera- levels of organization (cells, tissues, internal organs,
tion will require a targeted and multidisciplinary structures) (Bely and Nyberg 2010), here I focus

Advanced Access publication August 19, 2010
ß The Author 2010. Published by Oxford University Press on behalf of the Society for Integrative and Comparative Biology. All rights reserved.
For permissions please email:

gains of regeneration. 1) (Bely and Nyberg 2010). growth. this may itself be 1981. tails). distribution of regeneration are gleaned from Controls are especially important in trials of species single-species laboratory studies (typically develop- suspected of not regenerating. young/ regenerative ability remains limited. Wagner and Misof 1992. For many old. and under dif. First. A loss of regen- phylogenetic framework for interpreting the pattern eration is most easily identified when non- of the evolution of regeneration.g.. If it is possible to across species.g. as loss of such structures the absence of regeneration. simply no information on regenerative ability ferent environmental conditions (e. most available data on the stages of regeneration loss (see next section). When the on a range of species in the target group and homol- phylogenetic distribution of the presence/absence of ogous regions of the body should be removed in each species to ensure that amputations represent a regeneration makes the polarity of change less obvi- comparable challenge across the species tested. E. A number of chal- broad range of biologically relevant conditions. Once thorough experiments on regenerative ability Identifying lineages that have lost regenerative abili. animal groups. Furthermore. However. different (Fig. For example. Thus. reproduction). ure of regeneration) may be necessary in order to demonstrating the absence of regenerative potential confidently reconstruct the direction of evolutionary is inherently problematic. Identifying losses of re- necessarily be assumed to represent losses (Brockes generation requires both thorough comparative ex- et al. further studies been followed. have been completed. such that under the experimental conditions so comparisons amputations and conditions are often inconsistent can be made with cut animals. and if documen- temperatures) to determine whether the success of tation is available. of different nutri. across a range of related species (e. Bely of considerable interest for understanding early and Sikes 2010). if ampu- occurs widely in nature and thus structure-level re.. fins. non-regenerating species should not been identified rigorously. there is tional status (e. Bely specifically on the regeneration of structures (e. starved/well fed). ditions permissive of regeneration. If such follow-up studies demonstrate a few studies have systematically performed studies that. it is often available for only a tiny regeneration is merely a function of one of these fraction of the species in a group. tion losses thus far. regeneration losses have rarely generation... a of regenerative ability using comparable amputations lineage has experienced a great narrowing in the con. Instead. since ‘‘absence of evidence change. 2001). Follow-up studies to confirm the ab- assess during the trial the investment by individuals sence of regeneration under a range of conditions are into other developmental processes (e. heads. Although loss of regeneration is thought to have reconstruct the pattern of evolution of regeneration. tated animals continue to grow or reproduce during generation has a clear ecological and evolutionary the trial. additional taxonomic sampling or further work successful regeneration ensues. If ous. Scadding 1977. even as they fail to regenerate. is not evidence for absence. amputations should be performed on indi. rather than completely losing regeneration.g. occurred in many groups of animal and the optimal Although the loss of regeneration is generally as- approach for recognizing such cases is straightfor- sumed to be much more likely than the gain of re- ward in principle. For lenges have hindered the identification of regenera- example. the approach outlined above has rarely occur (or does not occur normally). . only variables. although a number of losses of should be performed to confirm the inability of the regeneration have been tentatively identified. on the developmental basis of regeneration (or fail- eration is demonstrated unambiguously. most species in question to regenerate successfully under a still await rigorous confirmation. the resulting data should be ties is an obvious first step in studying regeneration interpreted in the context of a robust phylogeny to loss. published information on viduals at different life-history stages (e.516 A.’’ If regeneration does not In practice. even some entire phyla. this can also strengthen the case for limbs.g.g. are certainly possible. this demon- context.g. sexually immature/mature).. although poorly periments on regenerative ability and a robust documented. Comparative exper- regenerating taxa are nested well within clades that iments on regenerative ability should be performed otherwise have the ability to regenerate. Vollrath 1990. rarely conducted. strates that individuals have sufficient resources available for these other processes and that these re- Identifying losses of regeneration sources can be mobilized under the experimental How to identify losses of regeneration conditions.. the presence of regen. Uncut individuals mental or physiological studies) or field studies re- should be maintained to gauge the rate of mortality porting naturally regenerating individuals.

. or simply not worth publishing. echinoderms. Finally. A relat- from the phylum). appear unable to wound-heal or even to survive body amputation because high internal pressure causes internal organs to be propelled from the body following a breach of the body wall. cessity to make assumptions about the relative and no evidence for the presence of regeneration. such as after a trans- verse body amputation (although this speci es is capable of axonal regeneration. although not necessarily simultaneously. Comparing regenerative potential across . ‘‘?’’ indicates probability of gains versus losses. 1 Phylogenetic distribution of regenerative ability across the among lineages. 2001). this species. based on the Second. leading from Bely and Nyberg (2010). The fact that regenerative ability appears common in a group does not imply that it is necessarily ancestral. Whole-body 2010).g. even for groups for which a considerable same information (Brockes et al.g. a gain of regeneration may be a Metazoa. The absence and not easily applied more broadly. and excludes regeneration of internal organs or tissues. Third. Even if rea- amount of published information is available. Indeed. different authors to sometimes make opposite con- clusions about the polarity of change. One group in which considerable information regeneration is defined as the ability to regenerate every part of is available but no formal mapping has been per- the body. there sonable estimates of the probabilities of transition has undoubtedly been a strong publication bias can be obtained.. of regeneration is often interpreted as inconclusive. or siphon) their legs and many species cannot (Vollrath 1990). 2004). and possibly nematodes in gen- eral. Many species can regenerate regeneration is defined as the ability to regenerate a multicellular structure of the organism (e. Regeneration is scored at two levels of organization: more likely reconstruction (e. limb.g. as- cidians) are uniformly good regenerators. or the developmen- tal basis of regeneration differs in fundamental ways Fig. Yanik et al. reconstructing evolutionary changes in re- generative ability has been hampered by the lack of formal mapping onto robust phylogenies in some groups. Even if good information is available on the regenerative ability of many species in a group. As an when different species exhibit high morphological example. The bias against publishing failure of regeneration has likely been pervasive. a head. Modified limited knowledge of this for regeneration. tail. and this should be considered care- fully whenever information on regenerative ability is gleaned solely from the published literature.Regeneration loss in animals 517 model organism Caenorhabditis elegans cannot regen- erate at the ‘‘structure’’ level. it is widely known that the nematode disparity. if the most deeply branching lineages are all non-regenerating. I could find no published documentation for this inability to survive body amputation or to regenerate. whole-body and structure-level regeneration. elegans is one of the most intensively studied animals in biology. the problem of homology is significant uninteresting. Ignoring this potential bias may lead to the false impression that certain groups (e. these are likely to be taxon-specific against reporting failure of regeneration. There is extremely uncertainty in the phylogenetic position of this lineage.. however. White indicates that there is at least one ed problem inherent in character mapping is the ne- substantiated report for the lack of regeneration in the phylum. planaria. Garza-Garcia et al. Structure-level formed is spiders. despite the fact that C. without an estimate of their relationships it is simply not possible to reconstruct evolutionary changes in regenerative ability. Black Until these data are mapped onto a robust phyloge- indicates that regenerative abilities are present in the phylum ny. viewing non-regenerating species as representing (regardless of whether non-regenerating species are also known losses should be considered tentative at best.

leeches). At one extreme. RNA interference. cnidarians.g. Goss 1969. and birds are the ability to regenerate specific.g. Vorontsova some groups such as sea stars. Given this information. for instance. ecdysozoans. losses of homologous developmental processes. Goss 1969). Bely different animal phyla. considerable financial investment. and lophotrochozoans. Bely and Nyberg 2010). E.. birds. but determining when tion for evaluating where losses and gains of regen. but the ability to regenerate this body parts. and birds appear to be example of regeneration in deuterostomes. some tion among animals (Vorontsova and Liosner 1960. sponges. regenerative abilities can surely increase Overview of regenerative abilities across the as well as decrease. amphibians. The evolutionary . 2010). many lizards Goss 1969. 1) (Needham 1952. many simple. leeches. sponges.518 A. homologous body homologous. mammals cannot. regeneration of siphons of (e. and birds cannot (Needham 1952. Bely and Nyberg 2010). Available ev- ative ability across the Metazoa is useful background idence suggests that cartilaginous and ray-finned fish for identifying losses of regeneration. is inherently hydrozoans have representatives capable of regener- problematic because homologizing body parts is dif. and acoels each have representatives that can ‘‘homology problem’’ is not easily overcome. or of no regeneration at next-generation sequencing). Among animals. placozoans. even Comparisons of regeneration across highly divergent severe ones. highly variable becomes increasingly well understood in a few model (Fig. The tails plans are so divergent that it is difficult to compare of fish. regenerate following a range of amputations. the phores. which can be ad. yet it is nearly or entirely unable to regenerate any structure. For of the original. planaria. were good regenerators. generators. mammals. The alternative scenario.g. ating a complete individual from just a tiny fragment ficult at best. can. Even so. dently.. generation at some level. although this possibility cannot mation requires sustained. Assuming the to branch out from a model system to investigate the ancestral animal could regenerate well. seems likely that the common ancestor of animals. As discussed above. evaluating broad patterns in regener. As the process of regeneration regenerative abilities are. groups example. With the advent of ascidians. targeted efforts and a be formally excluded based on available data. while at the other extreme. This is because. While re- models will become a key source of information for generation of structures occurs in all three of these inferring how regenerative capabilities have evolved bilaterian clades (e. not possible to homologize sea star arms in any Are there broad patterns to this variation? simple way to structures in other deuterostomes. all of the outgroup lineages to the such as hemichordates or chordates. seems unlikely.. Garza-Garcia et al. and it is thus cur. of only limited regeneration. it is becoming feasible all (e. that most or all of these bearing directly on how gains and losses of regener. ities. and and Liosner 1960. by contrast. source of evidence supporting or refuting the hy. Such studies regenerating bilaterians must represent losses of re- will undoubtedly provide a rich source of informa..g. heads and tails of molecular technologies that can be applied across a planarians). region varies considerably among them. cteno- dressed relatively easily through targeted efforts. but not the bony elements). reptiles. comparative datasets. and all have representatives with the po- taxa may need to focus specifically upon gains and tential to regenerate every structure in the body (Fig. these non- regenerative process in close relatives. relatives is not necessarily straightforward. losses occurred relative to their closest regenerating eration have occurred across animals. comparing regenerative ability The distribution of regeneration of tails across major across phyla is inherently problematic because body groups of vertebrates illustrates this point. soft-bodied animals today have high regenerative abil- pothesis of homology of regeneration across taxa. amphibians can. regeneration of sea star arms is a classic such as nematodes. although this phenomenon remains poorly documented. animal lineages acquired regenerative abilities indepen- ation are reconstructed. 1). many taxa in all three clades are capable wide array of organisms (e. and can be downright nonsensical. limbs of arthropods. comparing findings across these 1960. Vorontsova and Liosner organisms. it regenerate homologous morphological structures. Unlike the protostomes and deuterostomes include powerful re- first three challenges listed above. comprised of the rently not possible to use this approach across large deuterostomes. and the pattern and timing of Metazoa gains and losses can therefore be difficult to evaluate. however. Even a cursory cannot regenerate their tails (the tail fin of some fish survey of regenerative ability reveals marked varia. Within the main bilaterian clade. Obtaining this kind of infor. can regenerate. These groups also have few or no known species rather than upon gains and losses of the ability to incapable of regeneration. It should be noted that the cellular and molecular and early animals in general. basis of regeneration can serve as an important Consistent with this conclusion. nematodes.

including the cyprinodontiforms. groups. nemerteans. Further support for tail regeneration being chozoans each possess many representatives capable ancestral (and thus for the lack of tail regeneration of ‘‘whole-body’’ regeneration (defined as being able being derived) comes from the regeneration process to regenerate every part of the body). and lophotro- the tail. 2008). While three independent losses of the ability to regenerate non-bilaterians. the cyprinodontiforms. even within vertebrates. amphisbaenids. as in et al. For exam. recent phylogeny of the group (Vidal and Hedges a broad. geckos. data on the developmental basis of (which together likely represent the sister group to regeneration. having been reported in the tuatara (the many of the non-regenerating taxa represent losses sister group to the rest of the lepidosaurs). as it must involve multiple Arnold 1984. In vertebrates. Comparative experiments on a recent summary of fish phylogeny (Nelson 2006). many other bilaterian groups in which regenerative The distribution of regeneration of fins in fish ability is variable. and the per- clitellates. this distribution strongly suggests that tail re. carti. and some Finally. When regeneration of the tail occurs in lepi. regeneration ation is quite strong. annelids. Brockes relatives). in lepidosaur vertebrates (lizards and close 1977. such that amputa. In the context of penultimate section). 2008). Bely 2006). Vollrath 1990. although comparisons of regeneration agamids (Gans 1978. siphons) can potentially regenerate in the laginous rod at its center rather than an articulated other two main bilaterian clades (the deuterostomes . erate a specific body region. Bellairs and among phyla are inherently problematic. the molting clade of bilaterians. loss of regeneration. Arnold 1984. better sampling and formal phyloge- skinks. Vorontsova and Liosner derstanding of phylogenetic relationships. 2001. the regenerated tail has a continuous. Fin regeneration occurs in lungfish (an out- ditional sampling. a pattern suggestive of Although this study represents the first to explic. nevertheless suggests that an important restriction generation is ancestral for lepidosaurs and that the in regenerative ability occurred near the base of the non-regenerating groups likely represent at least Ecdysozoa.Regeneration loss in animals 519 pattern of regeneration of tails is clearly not simple column composed of vertebrae (Etheridge 1959. Tsonis 2000. the spiny-rayed teleosts). 1). notably. rigorous phylogenetic mapping. it consistently produces a regenerated tail Furthermore. platyhelminths. the Naidinae (Bely and Sikes 2010. in each. while there is ample evidence that a that is distinctly different from the original. most broad range of body structures (e. but it appears to be absent in a few evidence for this. group to the teleosts). 1981. abnormal fin) in representatives that multiple losses of the ability to regenerate the of at least three groups of spiny-rayed teleosts: the head have occurred within a subfamily of aquatic scorpaeniforms. some agamids. confidently reconstructing the pat. and amphibians. the ability to regenerate the tail is wide. Mapping results onto a molecular phylogeny to possess a mix of species (sometimes even very indicated that three losses of the ability to regenerate closely related species) that can and cannot regen- the head have occurred in this group. acoels. Stephan-Dubois and Bautz 1975. A number of additional animal groups are known cies. see ciforms (Wagner and Misof 1992). salmoniforms and esociforms and optimally. Based on a the extreme morphological disparity among groups. including snakes. One of the clearest examples of fins appears to be absent or heteromorphic (pro- comes from a recent study on annelids indicating ducing a misshapen. Ste´phan-Dubois 1960. and perciforms (Wagner and Misof In some animal groups.. Although it is likely that spread. ities are known from ecdysozoans (Fig. regenerative ability were performed on a range of this distribution would suggest that regeneration of species and. chameleons. from several other groups suggest additional likely among others (Needham 1952. Scadding ple. multiple losses of normal fin regeneration. Seligmann et al.g. deuterostomes. urochordates. owing to Bryant 1985. tails. atheriniforms. heads. provides another likely example of regeneration tern of the evolution of regeneration will require ad. iguanids. Kolmayer and losses when interpreted in the context of current un. limbs. 1960. tions represented a comparable challenge across spe. the case for loss of regener. 1992 and references therein). et al. of regeneration. and numerous species of spiny-rayed teleosts (the most diverse group of Some likely losses of regeneration living teleosts). Bellairs and Bryant 1985. and netic mapping are needed in most cases to provide lacertids. data hemichordates. Seligmann gains and/or multiple losses. phylum-level view of animal regeneration 2005). However. dosaurs. means of reconstructing the pattern of losses. itly map regenerative ability onto a phylogeny as a hexapods. Such groups include arachnids. loss. the species-specific number of fins is ancestral for fish and that there have been ‘‘head’’ segments was removed. no such abil- itself.

tures of regeneration loss are most likely to still be Doing so will require collecting and integrating func- detectable. arthropods. Some of the broader ques- an organism’s function more than the total absence tions we should be aiming to address with such data of the structure (Vollrath 1990). 2003). and we therefore know it. can With regard to the ultimate. evidence that even traits recently under strong selection though by no means all. if the latter.. sents the oldest major loss of regeneration repre. Wagner regeneration and another process (e. or so low that the structure is not worth replacing little about why and how this process occurs. such that redeployment of the remains a real need to define and evaluate specific latter’s developmental program is no longer possible evolutionary and developmental scenarios that could in the adult stage (Galis et al. no the wild becomes relaxed (Lahti et al. ideally by Distinguishing between these possible ultimate ex- investigating very recent losses in which the signa. Goss 1992.520 A. Lawrence 2010. tional. straightforward scenario under which regeneration tion of arthropods’ appendages (limbs and antennae) might be neutral is if the structure in question is infre- (Vorontsova and Liosner 1960. Regeneration could also be neutral if the functional importance of The process of loss of regeneration the structure in question is so high that the animal Very little work has been aimed at understanding the cannot survive without it long enough to regenerate process of regeneration loss. selection could directly disfavor regeneration Uncovering the proximate causes of regeneration loss because it somehow contributes to lower fitness. preferably Why might regeneration be lost? closely related ones. or evolutionary. as involves deciphering which steps of regeneration could occur if a partially regenerated structure impairs have become abrogated. Bely and Nyberg 2010). regeneration can occur in individuals after their ter. evaluate such scenarios using actual data. however.. This scenario account for this phenomenon. 2009). Sublethal predation is a prevalent cause of regeneration minal molt.g. and. To begin developing a could explain decreases in regenerative ability that framework for investigating this process. longer being actively maintained. Neutral loss of regeneration could also occur about the factors leading to loss of regeneration if previously tight developmental pleiotropies between (Morgan 1901. Regarding How might regeneration be lost? the first. the only structure-level re. such as growth (e. but there simply break down. There is ample Regeneration of appendages occurs in many. A generation known from ecdysozoans is the regenera. generation particularly prone to becoming blocked? er process. Wulff Does loss of regeneration occur all at once or grad- 2010) and investment in the latter becomes favored ually. advantage. Nyberg 2010). The first step towards addressing these questions is If regeneration confers no significant selective to determine when and how the regenerative process . It will be important to morphology. al. a common selective force maintaining regeneration. and developmental data for both regenerating and non-regenerating species. and thus this is presumably ability near the base of the Ecdysozoa likely repre. for example if there is different groups of animals? Are certain steps of re- an energetic trade-off between regeneration and anoth. however. and may have been decreases the frequency of sublethal predation in a related to the evolution of the protective cuticle population could therefore lead to regeneration no characteristic of this clade. E.g. Adaptive explanations of the process look like? need not be invoked for loss of regeneration. Only by tackling this problem with hypothesis testing and actual data. For given the costs of replacement (Goss 1969. quently lost or damaged in nature. In summary. Bely and lophotrochozoans). ecological. The inferred restriction of regenerative in nature (Lindsay 2010). A simple change in predator-prey dynamics that sented among metazoans. planations for loss of regeneration is no small task. Regeneration could include: Does regeneration fail in different ways in also be disfavored indirectly. or because it is a neutral trait. then it could be lost as a neutral trait. below I dis. Reichman well over a century there has been speculation 1984). and because can be rapidly lost or modified when selection in regeneration in arthropods depends on molting. regeneration could be lost either because it is selected against in some way. causes we begin moving away from the pervasive specula- of regeneration loss. Goss 1969). a range of possible explanations tion that has thus far characterized discussions about have recently been outlined and discussed (Bely and the ultimate causes of regeneration loss. what do intermediate stages at the expense of the former. correlate with changes in morphology or changes cuss some possible scenarios for both why and how in the developmental processes underlying the regeneration could be lost. embryogenesis) and Misof 1992. Dinsmore 1991.

Graphs on the left (A1. This could occur if. Solid lines denote the ancestral condition. energy available in an organism. when regeneration of fish fins fails. dashed and dotted lines denote populations progressively farther along on the trajectory to total loss of regeneration. such as the Fig. It is already ma (Bely 2006. within a given lineage. 1981). B2. or the functionality of the regenerated structure. losses might occur need to be developed. animals abilities. such cessively more advanced stages of regeneration loss that no obvious structures are formed (Scadding showing increasingly restricted or poorer regenerative 1977. Thus. that there may be phyloge. and environmental conditions). it is clear group. and C2) show scenarios in which loss of regeneration is insensitive to these factors. interestingly. available evidence suggests that regenera- structure missing certain functionally important tion may fail in predictable ways within a taxonomic elements? Although data are still limited. a structure occur by the evolution of a single mutation that com- often regenerates but is heteromorphic and misshapen pletely abrogates regeneration. that failure to regenerate can manifest itself in a variety certain steps of the regenerative process are particularly of ways and. Does a blastema (the mass of undifferentiated cells such that animals leak body fluid and die from the from which new structures form) not develop? Is the amputation. graphs on the right (A2. For ex. and with spider known that speed or success of regeneration can be legs. Bely and Sikes 2010). never properly wound-healing (Vollrath blastema improperly patterned? Is the regenerated 1990). with populations at suc- blastema often forms but is poorly patterned. Although loss of regeneration theoretically could ample. prone to becoming blocked. with amphibian limbs. 2 Some hypothetical models for the gradual loss of regenerative abilities in a lineage. a regeneration is lost gradually. . such as the frequency of successful regeneration. it is also possible that (Wagner and Misof 1992). Does wound-healing fail? is positioned somewhere other than at a limb joint. with annelids’ heads or tails. ‘‘Regeneration success’’ could be measured a number of ways.Regeneration loss in animals 521 halts or becomes abnormal. Specific hypotheses for how such gradual typically wound-heal but fail even to develop a blaste. and C1) show scenarios in which loss of regeneration is sensitive to a particular factor (developmental stage. B1. wound-healing itself often fails if the amputation sensitive to a variety of factors. the morphological fidelity of the regenerated structure. Curves for populations that have completely lost regenerative abilities would lie along the x-axis. netic trends in how regeneration tends to fail.

Some groups. lies (representing over one quarter of the families for tion loss. regenerate the head is much more variable. Groups in which regenera- population. recently evolved cases of regeneration loss. but the available data quantifiable factors. the fidelity of regen. in all but the most ic resources of the individual (e. in a lineage that is gradually losing regenerative abilities. making them useful models for understand- developmental flexibility. 2007. they will Figure 2 shows several possible scenarios for be extremely useful for elucidating the earliest steps how regeneration success could change through time of loss of regeneration at the molecular level. Other models Zrzavy´ et al. split black/white circles). Lawrence 2010). lost in only a few taxa (Bely 2006). if so. such as quently being lost many times. The ability to regenerate a new the energetic threshold for investment in regeneration tail is widespread across the phylum. Lawrence 2010). are common (Maginnis 2006. the pointing the first block(s) responsible for the success of regeneration could be contingent upon such abrogation of regeneration will be challenging at factors as well. which there are relevant data) possess both anteriorly tingency of regeneration success on internal and regenerating and anteriorly non-regenerating species external factors should be evaluated in taxa that have (Fig. (2007). One possibility is that successful regeneration Annelids as a model for studying loss becomes increasingly restricted across the life cycle. able uncertainty about the deep-level phylogeny of regeneration might only occur within an increasingly annelids (Rousset et al. For instance. success in regenera. the number of suggested losses is serve as a proxy for the ancestral. however. 3A. tion has been lost relatively recently are particularly The holy grail for understanding how regeneration useful for investigating the mechanisms underlying has been lost is to identify the actual genetic the loss of regeneration. such as growth. out within the family Naididae (former Tubificidae generation non-functional. Although this is an important goal. having been gradually increases (Fig. E. 2B1). 3A). but if such species can be identified. the con. fully-regenerating taxa which can ability is increased. To identify recent regeneration losses rigorously. 2008. Muller et al. it is fine-scale sampling and detailed phylogenies must essential to recognize that once a block to normal be obtained. rapidly accumulate. are consistent with the ability to regenerate anterior tion may correlate with the degree of pleiotropy segments being ancestral for the phylum and subse- between regeneration and a core process. and pin- are in the process of losing regenerative abilities. 2007. For example. Such work has recently been carried regeneration has evolved in a lineage. Yet another possibility is that the envi. The ability to tween regeneration and other process. Bely developmental age of the stump tissue and the energet. 2A1). are large this pleiotropy is breaking down. suggesting at least lost or appear to be losing regenerative abilities. rendering re. perhaps because Annelids exhibit extensive variation in regenerative younger tissues are comprised of cells with greater ability. Energetic tradeoffs be. Another possibility is that ing regeneration loss. 2008]). most embryogenesis. regeneration could become restricted to ear- lier ontogenetic stages (Fig. By mapping this informa- investment into regeneration could increase to the tion onto a molecular phylogeny for the group from point that investment in regeneration effectively Struck et al. these taxa time. As sampling for regenerative as in closely related. and suggesting numerous losses. generation. clades thought to be comprised entirely of species erated structures might gradually decrease through that cannot regenerate anterior segments.. additional blocks may [Erse´us et al. There is still consider- become increasingly narrow (Fig. tion. regenera- Han et al. as well this many recent losses. the energy threshold for the phylum (Bely 2006).522 A. change(s) responsible for the original failure of re. Thus. For of regeneration example. identifying the specific factors best. In populations that tion is likely to fail for multiple reasons. focusing specifically on the . However. Currently no species are known that show nat- that are still permissive of regeneration may point to an ural. Evidence for regenera- if investment in other processes becomes favored at tion of anterior segments was recently compiled for the expense of regeneration. fully-regenerating likely to increase as well. seven annelid fami- might differ in lineages at different stages of regenera.g. Models such as the ones shown in Fig. pable of regenerating anterior segments are scattered ronmental conditions permissive of regeneration could across the phylum (Fig. 2009) and greater sampling of regener- for a gradual loss of regeneration may involve less easily ative ability is clearly needed. To test the validity of such models. If regeneration is being lost because notably the Hirudinida and Nereididae. Struck et al. 1999. and. narrow range of temperature or salinity. I show here that annelids inca- never occurs. 2C1). intraspecific variation in success of regenera- underlying mechanism for the loss. 2 provide likely represent relatively old losses of the ability to clear predictions about how success of regeneration regenerate anteriorly.

that other phylogenetic studies of . and Amphichaeta one naidine species. Tubifex. Mapping these lent abilities to regenerate both the anterior and pos. It should be regenerate anterior segments. subfamily Naidinae (Bely and Sikes 2010). a process thought to within the naidines and found that six of the have evolved from regeneration (Bely and Wray 18 naidine species investigated are incapable of 2001). Because several naidine species possess excel. split black/white circles denote groups in which some species have been shown to regenerate anterior segments while others have been shown to be incapable of doing so. In (A). In (B). Branchiura. 3 Phylogenetic distribution of anterior regeneration in the annelids across (A) the entire phylum and (B) the clitellate subfamily Naidinae. Phylogenetic relationships are based on (A) Struck et al. however. it had long been assumed that species represent three independent losses of the this entire group consisted of excellent regenerators. Data on regeneration are based on (A) Bely (2006) and (B) Bely and Sikes (2010).Regeneration loss in animals 523 Fig. Black circles denote groups in which all taxa investigated have been shown to have the ability to regenerate anterior segments. even though it could noted. five-gene molecular terior ends. and because naidines are all capable of phylogeny of the group indicates that these six asexual reproduction. all species shown are naidines with the exception of the outgroups Monopylephorus. could not each representing a loss (Fig. Bely (1999) demonstrated that at least genera Paranais. Pristina. although not all. with the However. 3B). Paranais litoralis. Chaetogaster. Following up on this are a group of small freshwater oligochaetes that re. study. (2007) and Erse´us and Ka¨llersjo¨ (2004) and (B) Bely and Sikes (2010). comparative data onto a robust. white circles denote groups in which all investigated taxa lack the ability to regenerate anterior segments. ability to regenerate the head region. Bely and Sikes (2010) sampled more widely produce asexually by fission. all taxa are annelids. and most. regenerating the head segments. and Lumbriculus. groups represented are families. Naidines regenerate posterior segments.

Thus. 2008). This regeneration than does P. 2A1 by a flat line along the x-axis. wound. Bely naidines. and growth). riorly has surfaced until now likely speaks to the Chaetogaster diaphanus. Erse´us et al. diaphanus actually represents an older loss of blocked about the time of blastema formation. lower lids similarly fail at this same point: after right). E. close relatives) typically reproduce asexually by para. regenerative. Specifically. most a block at this phase does not impair other critical notably some vertebrates. gaining regeneration. in which a new head and tail form in generally more permissive of regeneration. Thus. developing tissue). However. this species appears to that regeneration fails at approximately the same have completely lost the ability to regenerate anteri- point in the process in the non-anteriorly regenerat. was found to also lack regenerate. and that the middle of the body (Fig. Interestingly. yet still generation. A few other taxa. the data are consistent with a known to regenerate digit tips as adults.. Fig. C.. comprised of young developing tissue (Fig. embryogenesis. glimpse at an intermediate phase of the loss of re. After anterior amputation. higher regenerative abilities during the embryonic healing. or if initiation of regeneration is evidence for regeneration being lost in an uneven the least pleiotropic phase of regeneration. Determining or juvenile phases of the life cycle than in the adult the underlying cause for the repeated evolution of phase (Vorontsova and Liosner 1960. Envall diately behind the developing head (leaving a stump et al. However. is a common and abundant ing anteriorly from adult tissue. can regenerate ante- species and one of the naidine subjects of an early. Instead. even though The study by Bely and Sikes (2010) provides a rare adults of these species are not (Illingworth 1974.g. Han et al. for a control. although in generation from adult tissue. suggesting an intermediate phase of regener- wound-healing and before formation of the blastema ation loss as represented by the dotted or dashed (Bely 2006). a close naidine outgroup capable of regenerat- Chaetogaster diaphanus. human children and mouse embryos warrants further investigation. litoralis. then. and there is model in which regeneration becomes gradually re. amputations can be this may be manifest in lineages that are losing re- performed within young developing tissue. regeneration (e. This suggests that. litoralis. Although it is not known whether regeneration is either preferentially. indicate the pos. 2010). lines of Fig.524 A. no extant mammals are generation. 4B. as exemplified by groups such as ascid- veloping head (leaving a stump comprised of young ians which may have extremely high regenerative . it may be that actively developing tissue is tomic fission. other non-regenerating anne. are capable of regenerating digit tips. these species wound-heal Paranais litoralis. sents the ancestral. early phases of the life cycle are not necessarily more formed on three species. 2A1. One of the species of old adult tissue). or experiencing no within the context of an otherwise fully formed change in regeneration. which cannot regenerate publication bias against reporting failure to anteriorly in the adult. Results showed that Pristina found incapable of regenerating anterior segments. if selection is acting to halt C. diaphanus has lower regenerative potential by regeneration as early as possible (e. regeneration (Dehorne 1916). specifically bisecting the de. Because naidines (and some of their a phase of regeneration loss is problematic at best. early-acting blocks to regeneration in annelids clearly For example. no evidence that the mammalian ancestor could do stricted to earlier ontogenetic stages. interpreting the vertebrate pattern as in Fig. 2A1 and likely repre- mention of this species’ inability to regenerate ante. are known to possess processes that are being maintained (e. cutting imme- sibility of fewer losses (Bely and Wray 2004. and would be represented on the graph in eration loss may be predictable within this group. based on smaller datasets.g.g. Wallace 1981). or most easily. if this phase represents the largest The study of naidines is the first to provide clear mutational target). this species lesser extent. a naidine incapable of anterior re- but fail to form a detectable blastema. fully-regenerating condition. at least in annelids. suggesting that some aspects of regen. the ability to regenerate anteriorly from young The study by Bely and Sikes (2010) also found developing tissue. Such amputations were per. the fact that could occur. 4B. 2A1. such that way across the life cycle. such as asexual reproduction). leidyi. to a tissue. 2006. stump. It should also be noted that adult (Fig. or. regardless of the developmental age of the ing naidines. if initiation on the trajectory to regeneration loss than is of regeneration is the most susceptible phase of P. That no matches the solid line of Fig. orly. if optimal several additional criteria (Bely and Sikes 2010) is allocation of energy favors investment in another consistent with this lineage being ‘‘farther along’’ process. 4A). riorly from a stump comprised of young developing thorough description focusing on fission and.. was found still to be one lineage post-amputation cell proliferation is still capable of anterior regeneration when the stump was initiated. right). for example. At a first approximation. as diagrammed so either.

be integrated for the purpose of testing alternative embryonic developmental capabilities of naidines hypotheses about loss of regeneration. bottom left. as this will allow studies of lost through indirect negative selection should the functional biology. 2005). while removal of two head segments (bottom) or one head segment (not shown) from the fission zone can result in regeneration of a complete head. also be evaluated in the future. Whether regeneration is ac- sublethal predation (De Vlas 1979a. when anteriorly amputated. studies. Paranais litoralis is known to accelerate the fission Why. Clavier tually delayed or inhibited when a fission zone is 1984. developing a good understand- regenerating species studied. ultimately. since there are and developmental genetic basis of regeneration to likely energetic tradeoffs between the multiple post. fission. in- cause of a decrease in the direct selective advantage tegrative approach. including the six non-anteriorly wide range of factors. Developing animal models that for anterior regeneration remains to be evaluated. Lindsay 2010). regen. Worms eventually separate and can reinitiate the process. There is ample tached offspring more quickly than do unamputated evidence that some annelids experience high levels of individuals (Bely 1999). Results are from Bely and Sikes (2010). comparative sion at the cost of regeneration. litoralis cannot regenerate if the amputation leaves a stump consisting entirely of old segments (white) (diagrams at top left. 2009. Diagrams on the right portray amputations adjacent to. Berke et al. the new head made within the fission zone is comprised of three head segments. are amenable to a range of investigations will be The possibility that regeneration may have been particularly important. However. (B) P. An into regeneration has traditionally been carried out anteriorly amputated naidine can potentially invest on a few. In P. and top right) but can regenerate if the amputation leaves even a small zone of actively developing tissue (light gray) at the distal end of the stump (diagram at bottom right). are critical for investigating resulting in regeneration loss as diagrammed in why and how regenerative abilities are lost. and if standing the ultimate and proximate causes of regen- optimal allocation of energy favors investment in fis. eration loss requires a shift in approach. there is no Future directions information on the incidence of sublethal predation Because regenerative ability can be influenced by a for most naidines. which are devoid of dorsal chaetae (bristles). the hypothe. ecology. under- in several different developmental processes. evolutionary history. or even a single segment (not shown) result in wound-healing but not in regeneration. forming a transiently linked chain of individuals.Regeneration loss in animals 525 Fig. and thus it present should be studied in both anteriorly regen- is likely that regeneration is selectively favored in erating and anteriorly non-regenerating species to many annelid species. litoralis (diagrammed here). and the asegmental cap of tissue bearing the mouth. abilities as adults yet none demonstrated during the Fig. regeneration has been lost in process such that amputated individuals produce de- several naidines is as of yet unclear. possibly non-regenerating taxa. regeneration). ing of the mechanisms responsible for evolutionary sis that regeneration loss in naidines occurred be. Groups . loss of regeneration will require taking a broad. 4 Paratomic fission in naidines (A) and the ability to regenerate anteriorly in the naidine Paranais litoralis (B). two head segments (bottom). Therefore. Kaliszewicz et al. for example. 1979b. focused on closely related regenerating and eration could be indirectly selected against. Diagrams on the left portray amputations in adult tissue: removal of all three head segments (top). or within. the fission zone: removal of all three head segments forming within the fission zone (top) results in wound-healing but not in regeneration. 2B1. While research (tissue turnover. larval phase (Vorontsova and Liosner 1960). growth. been demonstrated in a naidine species (Kaliszewicz 2003. (A) During paratomic fission. distantly related model organisms. a new tail (dark gray) and a new head (light gray) are intercalated in the middle of the body of an adult worm. Indeed. Sublethal predation has even further investigate this possibility.

1991. editor. Production due to regeneration by Euclymene reveal separate steps in the process of regeneration oerstedi (Polychaeta: Maldanidae) in the maritime basin of the Rance (Northern Brittany). systems for study. J Anat 199:3–11. Marchese M. Ka¨llersjo¨ M. Integr Comp Biol. capabilities in an asexual oligochaete. Grimm V. 2008. Driscoll PC. Hydrobiologia Gans C. The characteristics and affinities of the 406:243–51. Erse´us C.1093/ Bely AE. The the Society for Developmental Biology. De Vlas J. Erse´us C. Funding Dehorne L. Decoupling of fission and regenerative and mammals? Evol Dev 5:208–20. Illingworth CM. among close relatives represent particularly attractive Burlington (MA): Academic Press. Etheridge R. Yang X. J Exp Mar Biol Ecol loss and suggest which aspects may be predictable. Brockes JP. 2001. Wray GA. Annual food intake by plaice and flounder Acknowledgments in a tidal flat area in the Dutch Wadden Sea. p. New York: Wiley. 2010. Bryant S. 1916. 2010. 18S rDNA phylogeny of Clitellata (Annelida). Evidence the Annelida. Mol Phylogen Evol 40:570–84. Uchmanski J. 301–410. eration possible in amphibians but not in reptiles. The evolution of regeneration: adaptive or sion in annelids: insights from engrailed. Goss RJ. Development 128:2781–91. Evolutionary Developmental the non-monophyletic status of Naidinae (Annelida.526 A. Sublethal predation on Stylaria lacustris: implications. 2003. Velloso CP. Allan CH. 2005. Regeneration as an Predation effects on the evolution of life-history traits in evolutionary variable. A. Biology of the Galis F. Envall I. Tubificidae). Trapped fingers and amputated finger Berke SK. Principles of regeneration. 2007. 1959. Muneoka K. Lee J. Biology. Nyberg KG. with special I thank L. Amphisbaenia. inherent? J Theor Biol 159:241–60. Trans Zool Soc Lond 34:347–416. De Vlas J. Principles of regenerative biology. Distribution of segment regeneration ability in Garza-Garcia AA.and orthodenticle. regeneration in two onuphid polychaetes: patterns and Kaliszewicz A. . Clitellata). 1984. Sublethal predation and tips in children. A. Evolution of animal regeneration: regeneration in salamanders. 2006. Johst K. Kaliszewicz A. lizards and their relatives.). symposium was supported by the National Science Dinsmore CE. Lizard caudal vertebrae. Arch Zool exp ge´n 56:25–157. Les naı¨dimorphes et leur reproduction asex- Participation in the SICB 2010 Animal Regeneration ue´e. Trends Ecol Evol 25:161–70. The American Microscopical Society. Gustavsson L. Ka¨llersjo¨ M. Shapiro for comments on the manuscript reference to consumption of regenerating parts of macro- and members of the Bely laboratory for helpful benthic prey. 2008. I. Evolutionary aspects of tail shedding in 1744:66–8. 1978. 2001. Why is limb regen- reptilia.E. 1969. Wetzel MJ. Advance re-emergence of a field. Kumar A. and Ecology and Clitellata. discussions. Kumar A. Dev Biol 315:125–35. Jockusch EL. 1984.B. Sikes JM. systematic position of Opistocystidae (Annelida. 1985. Bely AE. 2009. 2004. Bely AE. (IOS-0920502). 2008. a study of regenerative capabilities. 1992. Annu Rev Cell Dev Biol 24:525–49. 1979b. Billett F. 2010. Autotomy and regeneration in 1967:699–721. New York: Proc Natl Acad Sci USA 107:1464–9. Molecular evidence for and Cell Biology. as these ‘‘natural replicates’’ may Clavier J. reptiles. Bely in which regeneration has been lost multiple times Carlson BM. the Society for Integrative and Comparative Cambridge: Cambridge University Press. 75:97–106. editors. 2004. cropped by flatfish. A history of regeneration Foundation (IOS-0940753 to Sara Lindsay and research: Milestones in the evolution of a science. Osman RW. Mol Phylogen Evol 30:50–63. ICZN rules: a References farewell to Tubificidae (Annelida. Zool Scr 33:187–96. In: Gans C. Evolution.E. Biol Bull 217:242–52. Goss RJ. following recent evolutionary loss in asexual annelids. Wagner GP. 1999.’s re. doi:10. Neth J Sea Res 13:362–93. class gene expression. Am Nat 166:409–17. for the local evolution of mechanisms underlying limb Bely AE. 2010. Hydrobiologia Brockes JP. including the divisions of Developmental Envall I. birds. Han M. Integr Comp Biol 46:508–18. Evolution of regeneration and fis. Cruz V. 2010. a clonal oligochaete. regeneration. Invertebrate Zoology. Wray GA. and Erse´us C. Erse´us C. Bely AE.B. Brockes JP. Access published online on April 21. Zootaxa Arnold EN. Molecular phylogeny of naidid Development and regeneration of the neonatal digit tip in worms (Annelida: Clitellata) based on cytochrome oxidase mice. J Pediatr Surg 9:853–8. Biology. Academic Press. Comparative aspects of animal 501:83–92. Clitellata) search is funded by the National Science Foundation revealed by DNA data. Bely AE. 2003. 1974. Mol Phylogen Evol 54:309–13. 1979a. Neth J Sea Res 13:117–53. J Nat Hist 18:127–69. 2006. Gustavsson L. Copeia Bellairs A. Latent regeneration abilities persist icb/icq022. Secondary production by tail regeneration in a tidal flat population of lugworms (Arenicola marina). E.

Scadding SR. London: Pergamon Press. Biol J Linn Soc 93:721–43. Johnson NA. Tsonis PA. Pia´lek L. Rˇı´ha P. Jin Y. Asexual propagation and Muneoka K. Bleidorn C. Bull Br Arachnol Morgan TH. Arch Anat Microsc Morphol Exp 64:75–89.1093/icb/ McHugh D. Foster SA. Vidal N. Trends Ecol Evol 24:487–96. Vorontsova MA. Erse´us C. on teleost pectoral fin regeneration. doi:10. Access published online on May 6. 2010. New York: Sons. Ste´phan-Dubois F. Regeneration in vertebrates. Advance Access A molecular phylogeny of annelids. icq100. (lizards. Siddall ME.Regeneration loss in animals 527 Kolmayer S. John Wiley & Sons. implications for orb weaver phylogeny. 1990. Frequency of injury and the ecology of Tsonis PA. logical strategies? Integr Comp Biol. Energetic costs of loss and regeneration anterieure. Dev Biol regeneration in marine benthic invertebrates. snakes. Otto SP. 2007. logique des neoblastes et des cellules fixes du parenchyme Relaxed selection in the wild. ation gap: genetic insights from diverse animal models.1093/icb/ Sa´nchez Alvarado A. . 2009. J Embryol Exp Morphol regeneration in the ‘living fossil’ Sphenodon (Reptilia: 8:376–86. Liosner LD. buds and digit tips. J Exp Zool 261:62–78. Annelid phylogeny and icq027. BMC Evol Biol 9:189. Stephan-Dubois F. Nat Yanik MF. New York: John Wiley & Wallace H. 2010. Wagner GP. 2005. 1960. Bridging the regener. Integr Comp 221:273–84. 1977. Limb regeneration in adult amphibia. 1981. 2010. 1981. Ne´oblastes et Seligmann H. regenerative capability in vertebrates: a comparative study London: Methuen & Co. 1984. Phylogeny Scadding SR. Misof BY. Janousˇkovec J. Hendry AP. Semin Cell Dev Biol 10:405–13. 2010. Muller TL. Donohue K. Reginelli A. London: Macmillan & Co. Can of Annelida (Lophotrochozoa): total-evidence analysis of J Zool 59:34–46. morphology and six genes. Nelson JS. Coss RG. Advance Access published online on August 7. published online on August 17. Ngo-Muller V. 2004. Kusen T. Vertebrate limb regeneration. Regeneration and wound-healing. Advance Struck TH. Cinar H. Functional regeneration after laser axotomy. Hickman E. The costs of autotomy and regeneration nuclear protein-coding genes. Regeneration of sponges in ecological context: Am Nat 123:752–63. 2006. Soc 8:177–84. Hedges SB. Evolution of regeneration capabilities. 2008. Regeneration in higher vertebrates: limb regeneration. Biol. Taylor G. Cladistics 23:41–63. 2009. limitation du pouvoir de re´ge´ne´ration ce´phalique chez la functional and evolutionary aspects of tail autotomy and planaire Dendrocoelum lacteum. in animals: a review and framework for future research. Ajie BC. Cinar HN. BMC Evol Biol 7:57. 2010. Integr Comp Biol. Werner YL. Evolutionary modification of Needham AE. Etude histologique et cyto- Blumstein DT. Morphological. tion capacity in adult Amphibia. 2006. 2007. Halanych KM. Vollrath F. Nature 432:822. Rhynchocephalia). 1992. J Exp Zool 202:57–68. Wulff J. of arms in stellate echinoderms. Zrzavy´ J. Regeneration. 1952. Schult N. 1901. Anderson R. 1975.1093/icb/icq099. Phylogenic distribution of limb regenera. is regeneration an integral part of life history and morpho- Rousset V. Lindsay SM. doi:10. the status of Sipuncula and Echiura. C R Biol 328:1000–8. Reichman OJ. Bautz A. Yakar A. Rouse GW. The phylogeny of squamate reptiles doi:10. Ben- Rev Genet 7:873–84. 1999. Leg regeneration in web spiders and its Behav Ecol 17:857–72. 2006. Chisholm AD. Fishes of the world. Lahti DC. 2010. 1960. Pleijel F. Moravec J. 2000. and amphisbaenians) inferred from nine Maginnis TL. chez des planaires Dendrocoelum lacteum privees de region Lawrence JM.