Ann. Zool. Fennici 41: 729–745 ISSN 0003-455X Helsinki 20 December 2004 © Finnish Zoological and Botanical Publishing Board 2004

Recognition systems and biological organization: The perception component of social recognition

Jill M. Mateo

Committee on Human Development, 5730 S. Woodlawn Avenue, University of Chicago, Chicago, IL 60637, USA (e-mail: [email protected])

Received 30 Apr. 2004, revised version received 12 Sep. 2004, accepted 12 June 2004

Mateo, J. M. 2004: Recognition systems and biological organization: The perception component of social recognition. — Ann. Zool. Fennici 41: 729–745.

Recognition of conspecifics is necessary for differential treatment of individuals in a variety of social contexts, such as territory establishment and defense, dominance hier- archies, reciprocal , mate choice, parent–offspring interactions and nepotistic contexts, to name a few. Here I first review various categories of perceptual mecha- nisms of social discrimination, focusing largely on the extensive literature on the per- ception component of kin recognition, although the ideas presented here can and have been used for analyses of recognition at many levels of social organization. I then dis- cuss a range of recognition mechanisms observed in a model species, Belding’s ground squirrels, and how socio-ecological factors influence the development and expression of each mechanism. Finally, I address several theoretical and empirical controversies in the kin-recognition literature which pertain to the perceptual component of recogni- tion, as well as areas in need of additional investigation.

Introduction of cognitive processing or awareness) whereby animals become familiar with conspecifics and An understanding of recognition systems can later remember them and treat them accordingly, be important for explaining interspecific vari- based on the nature of those previous interac- ation in population cycles, nepotistic patterns, tions. Recognition can also be based on cues of dispersal and avoidance (Charnov those individuals, on inanimate objects, or on & Finerty 1980, Blaustein et al. 1987, Hepper other proxies for identity such as spatial location. 1991a, Pusey & Wolf 1996, Sherman et al. While social recognition of individuals is not 1997), particularly in group-living species with required for , mate choice, tolerance on-going social interactions. Recognition abili- of neighbors or parental care, it can facilitate ties would be expected when animals interact appropriate behavior in these contexts and favor repeatedly over time and when discrimination behaviors directed toward particular individu- among multiple familiar individuals is benefi- als. From a functional perspective, the benefits cial, such as with nepotism, reciprocal altru- of social recognition are clear, yet the processes ism and dominance hierarchies (Hamilton 1964, underlying it can be quite complex. Trivers 1971). I define social recognition as a Kin recognition is an internal process of cognitive process (without implying any level assessing genetic relatedness that can be inferred 730 Mateo • ANN. ZOOL. FENNICI Vol. 41 through kin discrimination, the observable, dif- all are actually mechanisms for recognizing ferential treatment of conspecifics based on cues kin (as versus perceptual abilities serving some that correlate with relatedness. Recognition at other purpose), and whether distinctions between most levels of social organization (but not all; see mechanisms are largely conceptual rather than “Context-based recognition”, below) involves biologically relevant, as described below. Before three components: the expression of unique phe- proceeding, I note that these mechanisms are not notypic cues, or labels (also referred to as the mutually exclusive or necessarily exhaustive, ‘production’ component), the perception of these and that an individual can use several mecha- labels and their degree of correspondence with nisms depending on the recognition context. a “recognition template”, and the action taken by the animal (the ‘evaluator’) as a function of the perceived similarity between its template Context-based recognition and an encountered phenotype (or ‘cue-bearer’; Beecher 1982, Holmes & Sherman 1982, Reeve Similar to Barnard’s (1990) “discrimination by 1989, Gamboa et al. 1991, Sherman et al. 1997, non-conspecific cues”, I group together several Liebert & Starks 2004, Tsutsui 2004). Expression modes of discrimination which are based on and perception components comprise the mecha- contextual cues rather than cues borne by indi- nism of recognition, both of which are necessary viduals themselves. Previously I have focused for discrimination of conspecifics before the on spatial cues alone (Mateo 2002, 2003), but action component can be expressed, resulting recent studies suggest a need for an expansion of in differential treatment of conspecifics (e.g., this category. nepotism, mate choice, avoidance, cooperation). First, animals may be recognized indirectly However, social discrimination — of kin, neigh- based on spatial cues, with individuals encoun- bors, mates, allies, competitors, etc. — is not an tered in a particular area (e.g., nest or burrow) inevitable result of social recognition. treated as kin regardless of true relatedness (e.g., Thus recognition labels and perceptual abili- mother–offspring recognition via natal bur- ties provide proximate or mechanistic explana- rows; Holmes & Sherman 1982). During the tions for social recognition, whereas the action last two decades this mechanism has received component provides a functional or adaptive scant empirical attention, in part because some explanation for recognition because it involves (e.g., Barnard 1990; reviewed in Tang-Martinez behaviors with fitness costs or benefits (such as 2001) have suggested it is not a true, direct form agonism or avoidance of inbreeding). However, of kin recognition, since animals are responding it is important to note that natural selection (and to locations rather than phenotypes of individu- perhaps ) can operate independ- als. However, spatial cues may correlate reliably ently on each component of recognition, such with relatedness, as when females lay their eggs that if kin labels and perceptual abilities have not or give birth in a nest to which other individuals both evolved, differential treatment of individu- do not have access, and when communal nesting als cannot be selected regardless of how favora- and brood parasitism do not occur. Thus when ble it might be (Hamilton 1964; see also Holmes parents return to the nest, they by default invest & Mateo 2006). exclusively in their own offspring. Likewise, those young will only encounter their parents in that location, and so treating any adult found Mechanisms of kin recognition there as kin would be accurate. For example, bank swallow (Riparia riparia) parents use a The evolution of kin-selected behaviors of course spatially based mechanism to recognize their requires the evolution of mechanisms that make young confined to the nest and later switch recognition of kin possible (Hamilton 1964; see to a familiarity-based mechanism (described Table 1). Several mechanisms for the perception below) just before young fledge (Beecher et al. component of recognition have been described, 1981). Male black-tailed prairie dogs (Cynomys although there is disagreement about whether ludovicianus) appear to treat young-of-the-year ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 731 differentially depending on whether young are Before young leave the nest and begin inter- encountered in an area where the male copulated acting with other conspecifics, spatially based (Hoogland 1995). recognition is an accurate and adaptive mecha-

Table 1. Perceptual mechanisms of social recognition.

Mechanism Contexts Type of cues Advantages Disadvantages

Context- Spatial Kin (and only Location (nest, Family cues do Risk of mis- based cues kin) are reliably burrow, territory) not need to be identifying encountered in a learned non-kin particular area encountered in familiar location as kin Mating Social partner has Evaluator’s Directed parental Cuckolded access no opportunity to memory of investment in parent may mate with others exclusive offspring even if invest in mating with young do not non-kin target’s parent produce kin labels Cohort Young in a Age of Paternal half- Recognition sharing cohort sired by evaluator and siblings mistakes if one or a few target recognized many males adults without need successfully sire for production young or perception of kin labels Recognition Precise Cue-bearer’s 3 components Risk of treating alleles recognition phenotypic of recognition non-kin with without learning traits can evolve shared trait family traits together as kin Prior Short-lived, little Cue-bearer’s Can recognize May not be association overlap of phenotypic familiar kin able to generations, traits matched regardless of discriminate few social against location among familiar interactions; exemplars in individuals; little fitness evaluator’s cannot benefit for template recognize recognizing previously unfamiliar kin unfamiliar kin Phenotype Fitness benefits Cue-bearer’s Can recognize Liberal matching for recognizing phenotypic previously matching classes of traits matched unfamiliar kin algorithm may unfamiliar kin; to prototype lead to multiple mating derived from acceptance of by males creates referents in non-kin or unfamiliar evaluator’s rejection of kin paternal half- template siblings Self- Multiple mating Cue-bearer’s Can discriminate Linkage referent creates unequally phenotypic among full and disequilibrium phenotype related but equally traits matched half-siblings; could cause aid matching familiar siblings; to evaluator’s can recognize to be given to precise estimates own cues kin if template non-kin of relatedness are is ‘lost’ needed 732 Mateo • ANN. ZOOL. FENNICI Vol. 41 nism for parent–offspring recognition, as long as tify likely paternal half siblings and perhaps even other females (or researchers) do not add non-kin to assess likely paternity. If breeding in a social to the nest. The disagreement as to whether this group is seasonal and is dominated by one or a is a true form of kin recognition (see also below) few males, then most young born into a cohort stems, in part, from confounding the process of are likely to be offspring of that male(s). For recognition with the outcome of such recognition. cohort sharing to be a reliable mechanism for If selection favors discrimination of conspecifics recognition, the timing of births must be episodic based on genetic relatedness, and all individuals enough for distinctions between separate cohorts in an area are reliably kin, then the rule-of-thumb to be clear to all in the group. From the breeding ‘treat anyone in area X as kin’ will also be favored male’s perspective, he could use either mating (see also Hamilton 1964: p. 22). That is, spatially effort or cohort sharing to identify his putative based recognition is the most parsimonious proc- offspring. Cohort sharing could explain the dif- ess for recognizing kin in this context, while other ferential treatment of paternal half siblings and mechanisms would be favored when individuals non-kin observed in savannah baboons (Papio in an area are not all related, yet both scenarios cynocephalus) and rhesus macaques (Macaca lead to the same outcome: preferential treatment mulatta; Altmann 1979, Widdig et al. 2001, of kin or discrimination against non-kin. Smith et al. 2003; called social familiarity and A second contextual cue that can reliably cor- age proximity by these authors). This mecha- relate with is mating effort. For example, nism would not lead to accurate recognition of male dunnocks (Prunella modularis) adjust their kin groups if many males achieve reproductive parental investment in nestlings according to success, and thus may only be favored when kin how much time they spent exclusively with a discrimination has low costs, such as in groom- female during the mating period. Although the ing dyads, play-partner preferences, or foraging males may not be discriminating between their proximity, rather than in mate-choice, nepotistic young in the nest and those of other males, time or coalition contexts. in proximity to a female is a good proxy for a male’s relatedness to the clutch and thus he can adjust his total investment in the clutch (Davies Recognition-allele mechanism et al. 1992). Similarly, if the male partner in a tree-swallow (Tachycineta bicolor) pair is exper- Second, recognition could be mediated by “rec- imentally removed during the female’s fertile ognition alleles” which cause expression of a period, replacement males will invest in the phenotypic cue, recognition of that cue in others, offspring in the nest. If the partner is removed and preferential treatment of individuals bearing during incubation, replacement males do not the cue (Hamilton 1964, Dawkins 1976, Holmes invest in the young and even commit infanticide, & Sherman 1982). For example, altruism in as there is no chance that they were the sire of social amoebas (Dictyostelium discoideum) is any of the young (Robertson 1990, Whittingham mediated by the csA gene which guides molecu- et al. 1993). Recognition of offspring in these lar recognition, cell adhesion and cooperative cases is based on contextual cues which correlate aggregations (Queller et al. 2003). An allele with relatedness — using degree of paternal cer- in fire (Solenopsis invicta) causes workers tainty as a rule of thumb — rather than traits of bearing that allele to kill queens which do not kin themselves. This mating-effort mechanism of bear that allele, with discrimination occurring discrimination will be favored if it reliably leads through olfactory cues (Keller & Ross 1998). to preferential treatment of kin, or in avoidance This recognition mechanism has not been the of or aggression toward non-kin (Holmes & focus of much research because, conceptually, Mateo 2006, see also Neff & Sherman 2002). it is expected to result in cooperation with non- A third contextual cue I refer to as “cohort kin that happen to express that cue, and so such sharing” can help young animals to recognize single-allele cue-based cooperation is unlikely to their paternal half siblings (and perhaps other spread (Dawkins’ 1976 “green-beard” effect; see classes of unfamiliar kin), and for adults to iden- also Blaustein 1983). For example, planktonic ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 733 larval ascidians (Botryllus schlosseri) settle near learning involved in social recognition is usually and eventually fuse with kin more than non-kin, unknown. With prior association, animals can and these kin-biased aggregations are beneficial recognize familiar individuals, but not unfamil- because colonies composed of kin grow and iar kin such as paternal half-siblings or cousins. reproduce faster than aggregations of distant or For example, animals may learn the traits of their non-kin. Larval settlement patterns are influenced parent(s) and same-aged siblings, as well as their by cellular recognition, with larvae fusing with older or younger siblings if dispersal is delayed, another individual if they share a histocompat- and recognize these individuals later regardless ibility allele but rejecting those with a different of where they are encountered. Prior association allele, even if those larvae are kin (Grosberg & can be established or learned in the nest itself or Quinn 1986). Thus in this system the expression, in the family’s territory, if territory use is largely perception and action components all appear to exclusive and non-kin are not encountered there. be influenced by histocompatibility alleles. If development is precocial and young mix with The vertebrate major histocompatibility com- non-kin shortly after hatching or birth, learning plex, a large and highly polymorphic set of genes should occur rapidly, to restrict establishment involved in immune functioning, influences the of familiarity with kin only (e.g., Gottlieb 1981, production of kinship-correlated odors (reviewed Gubernick 1981, Poindron & Lévy 1990). in Penn & Potts 1998). In addition, olfactory- Prior association, or familiarization, is a receptor genes are found in the MHC (although proxy for relatedness and as such is sufficient for it is unclear presently whether these genes are accurate kin recognition when relatives reliably functional; Younger et al. 2001), thus suggest- interact in the absence of non-kin during early ing the MHC as candidate recognition alleles. learning, such as at nest sites or in exclusive To date there is no evidence that the MHC also home ranges (and therefore differs from a cohort- causes preferential treatment of individuals with sharing mechanism). Through direct association, similar MHCs, so although the MHC influences individuals learn and become familiar with their the production of genetically distinct odors, these early associates (typically relatives) and can later odors were likely co-opted for kin-recognition recognize them as kin even if encountered in a purposes independently. (If the MHC genes do different spatial location. Recognition can also influence odor perception as well as expres- be mediated through prior association with a sion, this would fit Hepper’s (1991b) criteria for third individual, as when siblings born at differ- recognition genes and Crozier’s (1987) criteria ent times both interact with their parent (Holmes for “genetic kin discrimination”, which assume & Sherman 1982) or when a parent’s behavior that other genes regulate any response that might toward young promotes similar behavior by its result from recognition.) Most research on kin older offspring (e.g., helpers at the nest; Richard- recognition, both theoretical and empirical, has son et al. 2003). Prior association is commonly focused on prior-association and phenotype- implicated in parent–offspring recognition, when matching mechanisms, so I will not discuss the litter, brood or clutch size is > 1 or when young recognition-allele mechanism further. remain with their natal group. This mechanism is also favored when discrimination among equally familiar individuals (e.g., full and half maternal Prior-association mechanism siblings) or equally unfamiliar individuals (kin and non-kin) is not advantageous, such as when Third, recognition may be based on familiarity animals are short-lived with little overlap of gen- via prior association, when animals learn the phe- erations or when there are few opportunities for notypes of individuals during early development social interactions with other kin. However, prior (e.g., siblings and parents), and later discrimi- association will lead to recognition mistakes nate these familiar relatives from unfamiliar ani- if non-kin are encountered during the learning mals. This learning can occur through a number phase (e.g., communal nesters), or if close kin of processes, such as habituation, or are not encountered until some time later (e.g., associative learning, although the actual type of young born in different years). 734 Mateo • ANN. ZOOL. FENNICI Vol. 41

1.00 a kin (Beecher 1982, Holmes & Sherman 1982, 0.75 Reeve 1989). That is, unfamiliar kin are “rec- 0.50 ognized” as belonging to a particular kin class.

accepted 0.25 This recognition occurs through generalization from recognition templates, with the degree of Percentage of transfers 0.00 12 3 6 8 10 16 20 22 26 30 32 36 44 50 Age at transfer (days) match between an encountered phenotype and 30 an individual’s template indicating the degree of b Rel-Fam Rel-Unfam 25 relatedness between the two animals. Unrel-Fam Unrel-Unfam 20 Phenotype matching is favored when kin are 15 encountered after early development because it time (s) 10 allows discrimination among individuals with- out prior association, by comparing their cues

Mean (± SE) retrieval 5

0 to a learned recognition template. This mecha- Day 1 Day 8 Day 15 Day 22 Age of young nism would be expected when there is multiple mating by males (for recognition of paternal Fig. 1. — a: Proportion of young S. beldingi accepted half-siblings), communal nesting (so females by unrelated adult females in a field cross-fostering can discriminate against familiar but unrelated experiment. Free-living S. beldingi mothers readily retrieved into their natal burrow unrelated young placed young), natal or breeding dispersal (for males to at their burrow entrance, but only until their own young recognize their older brothers or fathers), overlap reached about 25 days of age, which coincides with of generations, particularly in long-lived spe- when mothers’ own young emerge aboveground from cies (Bekoff 1981, Holmes & Sherman 1982) their natal burrow. From Holmes and Sherman (1982). or when there is no parental care (e.g., Göth & — b: Latency (mean s + SE) to retrieve pups by captive adult female S. beldingi. Mothers discriminate among Evans 2004). Phenotype matching would also be their young and those of other females just prior to the favored in cases of intra- or inter-specific parasit- age of natal emergence. Rel = related; Unrel = unre- ism (see Göth & Hauber 2004). Finally, com- lated; Fam = familiar; Unfam = unfamiliar. From Holmes parison of unfamiliar conspecifics’ cues to own (1984). cues (self-referent phenotype matching) might be especially important when females mate mul- tiply so that young can discriminate among their Phenotype-matching mechanism maternal full and half siblings (Mateo & John- ston 2000a; see also below). Fourth, through an extension of prior association, animals can learn their own phenotypes or those of their familiar kin, and later they can compare An example of multiple mechanisms or match the phenotypes of unknown animals expressed within one species to this learned recognition template (“pheno- type matching”; also referred to as comparing Species can utilize several different recognition phenotypes and signature matching; Alexander mechanisms, depending on the social context. 1979, Holmes & Sherman 1982, Beecher 1988). For example, Belding’s ground squirrel moth- Phenotype matching requires a correlation ers (Spermophilus beldingi) use spatial cues to between phenotypic and genotypic similarity so discriminate among young, caring for any pup in that individuals with traits that most closely their underground nest, until just before young match an animal’s template are its closest kin. leave the nest (Fig. 1a). After emergence from Both prior association and phenotype matching the nest, spatial cues no longer reliably cor- involve a comparison between templates and relate with relatedness, as young may now mix unfamiliar or familiar phenotypes, but as noted with other litters. However, at this age offspring above prior association leads to recognition are beginning to produce unique recognition only of previously encountered familiar indi- odors and mothers can use a prior-association viduals, whereas phenotype matching permits mechanism for discriminating among own and discrimination among unfamiliar kin and non- unfamiliar young (Fig. 1b; Holmes & Sherman ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 735

** a b Full siblings 12 a 16 ** * ** 3/4 siblings Non siblings 12 8 ** * ** 8 4 4 *

0.375 0.125 0 per pair ± 95% CI 0

Time (s) spent investigating 3/4 Paternal Non- 0 Play Nasal investigation sibling cousin kin Mean frequency of behavior

** Fig. 3. Mean frequencies (+ 95% CI) of (a) play and 10 b c * (b) nasal investigations between pairs of juveniles as a function of relatedness (3/4 siblings are offspring of two sisters who mated with the same male). Horizontal 5 bars and asterisks represent differences in behavior frequencies (* p < 0.05, ** p < 0.01) based on Poisson regression analyses. Mean frequencies are adjusted 0.375 0.125 0.25 0.19 0 statistically by the regression model for effects of sex Time (s) spent investigating 3/4 Paternal Grand- Half- of each juvenile in a pair and the weight difference sibling 1/2 sibling mother aunt between juveniles in each pair. From Mateo (2003). Fig. 2. Duration of investigation (mean + SE) of S. beldingi odors collected from subjects’ unfamiliar rela- tives during preference tasks. — a: Investigation of oral-gland odors of unfamiliar kin, collected from sub- nasal investigation, therefore leads to preferen- jects’ 3/4 sibling (offspring of two sisters mated with the tial social interactions with kin (Fig. 3; Holmes same male), cousin (offspring of the subjects’ moth- 1997, Mateo 2003). er’s brother) and non-kin. — b: Investigation of oral S. beldingi can also discriminate among non- odors collected from subjects’ grandmother and aunt. kin, such as territory neighbors or potential mates, — c: Investigation of dorsal-gland odors collected from subjects’ half aunt (mother’s half sister) and non-kin. using phenotype matching to recognize unfamil- Numbers inside bars are estimated coefficients of relat- iar relatives of non-kin (Mateo 2002), and prior edness between subjects and odor donors. Horizontal association to discriminate among individual bars and asterisks represent significant differences in non-kin (J. M. Mateo, unpubl. data). However, investigation of odors (* p < 0.05, ** p < 0.01) based this recognition of non-kin seems to disappear on repeated-measures ANOVA or paired t-test. An increase in investigation as relatedness of the odor during hibernation, as animals who could distin- donor to the subject decreases indicates that odors guish between the odors of familiar and unfamil- of distant kin are perceived as less familiar to sub- iar non-kin at the end of the summer no longer jects than odors of close kin. Significant differences in show such an ability after overwintering (Fig. investigation of odors from various kin classes indicate 4a; Mateo & Johnston 2000b). Torpor does not discrimination of those odors, and thus that the subject can recognize the difference between, for example, its impair the retention of recognition of littermates distant and non-kin. From Mateo (2002). (Fig. 4b), indicating that recognition of unrelated conspecifics is lost but that some level of kin rec- ognition is maintained across hibernation. Yet if 1982, Holmes 1984; J. M. Mateo, unpubl. data). animals forget the odors of familiar, unrelated S. Juveniles and adults can also use phenotype beldingi, they may also forget the odors of their matching to discriminate among unfamiliar indi- littermates. That is, the recognition templates as viduals based on genetic relatedness (Fig. 2; e.g., a whole may be lost during hibernation, in which Holmes 1986a, 1986b, Mateo 2002). This recog- case animals must use their own odors via self- nition ability influences social behaviors, includ- referent phenotype matching to recognize kin ing nasal investigations (allowing perception each spring (see also Holmes 1986b and below). of kin-distinct oral-gland odors; Mateo 2002) Thus S. beldingi use a variety of perceptual and play-partner preferences (thought to influ- mechanisms to recognize their social partners, ence the development of adult kin preferences; and during their lifetimes the perceptual compo- Holmes 1994). Discrimination among unfamiliar nent, in particular recognition templates, change kin and non-kin, as evidenced by differential and need to be continually updated. 736 Mateo • ANN. ZOOL. FENNICI Vol. 41

8 a b Littermate odor 15 Non-littermate odor Familiar odor 6 Unfamiliar odor 10

4

5 investigating 2

Mean time (sec + SE) spent 0 0 Pre-hibernation Post-hibernation Post-hibernation Testing period Fig. 4. Tests of odor discrimination before and after hibernation. — a: Duration of investigation (s + SE) of oral- gland odors of familiar and unfamiliar juveniles by Belding’s ground squirrels prior to hibernation (left) and after hibernation (right). — b: Duration of investigation (s + SE) of odors of previously familiar littermates and non-lit- termates after hibernation. Horizontal bars represent differences in responses to the two odor types (p < 0.01). Redrawn from Mateo and Johnston (2000b).

Direct and indirect recognition mon. Further, Tang-Martinez (2001) suggested that discrimination resulting from spatially The recognition mechanisms discussed above based cues is not kin recognition, and therefore have been categorized by some as direct and it should not be considered a recognition mecha- indirect. Waldman (1987) asserted that site-spe- nism. Unfortunately, recognition is an unobserv- cific spatial recognition is “indirect” recogni- able, internal cognitive or neural process, and tion because non-phenotypic cues serve as the we know too little about this general process basis for discrimination, whereas prior associa- to determine whether animals actually do or tion, phenotype matching and recognition alleles do not recognize their kin when encountered result in “direct” recognition because discrimi- in various locations. However, as I discussed nation is based on bearers’ traits such as odors, above, if individuals encountered in a particular plumage or vocalizations (see also Table 1). location are treated as kin, and if those individu- Barnard (1990) differentiated between direct and als routinely are kin, then selection will favor a indirect cobearer discrimination; in the former, spatially based mechanism of recognition and alleles recognize copies of themselves in conspe- discrimination. cifics, whereas in the latter, alleles use kinship as The distinctions between direct and indi- an indirect means of identifying who is likely to rect recognition described above, as proposed share copies of themselves. In a slightly differ- by Blaustein (1983), Waldman (1987), Porter ent vein, Porter (1988) distinguished between (1988), Barnard (1990), Hepper (1991b) and direct familiarization, when recognition requires Tang-Martinez (2001), are in fact differences previous interactions with the animal to be iden- in expression components rather than percep- tified, and indirect familiarization, which does tion components. These researchers catego- not require prior experience with the animal to rize the four common recognition mechanisms be identified (but does require direct familiarity based on cues to recognition, confounding the with shared relatives). expression and perception components. Since Tang-Martinez (2001) contended that spa- the process of recognition and the cues for it tially based recognition actually represents evolve separately, as long as cues (contextual errors, as kin encountered away from the home or phenotypic) vary with relatedness, any of the area would be treated as non-kin. Yet this mech- above perceptual mechanisms can be used for anism will only be favored by selection when kin-recognition purposes. Thus the distinction relatives reliably interact in a particular area, and between direct and indirect cues for recognition not in other areas (or when preferential treatment may be heuristic for understanding the expres- of kin in those other areas is not favored), and sion component, but less so for the perception thus these “errors” would be extremely uncom- component. ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 737

Distinction among recognition In their historical development and as currently mechanisms conceived, both mechanisms assume that learn- ing occurs exclusively when kin are present, In a recent review of recognition mechanisms, usually during early development, which is often Tang-Martinez (2001) proposed that there is only the case in burrows or nests where single lit- one recognition mechanism — learning as a ters, broods or clutches are reared. However, result of familiarity — with only the cues that the prior-association and phenotype-matching are used for recognition differing among recog- mechanisms use different perceptual processes nition processes. She further suggested that prior to match conspecifics to recognition templates, association and phenotype matching are involved with the former involving an exact match to the in all recognition situations. That is, prior asso- template and the latter generalizing from the ciation involves learning familial phenotypes template as a gestalt representation of kin. In and later matching encountered phenotypes to addition, referents, or the familiar phenotypes those learned cues. And phenotype matching learned early in development against which requires prior association with kin during some encountered phenotypes are matched, are likely period of development during which familial encoded in templates differentially as well, with cues are learned, so that later encountered phe- prior association “storing” referents as individual notypes can be matched to the learned template. exemplars (e.g., mother, father, and each separate This is how most kin-recognition researchers sibling) and phenotype matching likely using the have historically characterized the mechanisms, referents to form prototypes (e.g., a family-wide yet Tang-Martinez (2001) suggested the only cue). As discussed in Holmes and Mateo (2006), difference between the two processes is that I use “prototype” as cognitive psychologists use prior association involves learning of individu- it to refer to a set of features that are most com- ally distinct cues from kin whereas phenotype monly present in all members which belong to a matching involves learning of kin- or family- particular category (Smith & Medin 1981, Estes distinct cues, which all members of a kin class 1994, Hampton 1995). Prototypes develop when share apparently without individual variation. animals experience individual instances within The reason for such a distinction is unclear, par- a category (e.g., a family’s odors) and abstract ticularly since the same cues are often used for the common attributes of those instances, stor- both kin recognition and individual recognition ing them as a generic representation of the (e.g., Brown et al. 1990, Rendall et al. 1996, common attributes of the category as a whole. Todrank et al. 1998, Mateo 2002; J. M. Mateo, With exemplars, animals experience individual unpubl. data). Furthermore the two mechanisms, instances within a category and store those indi- as originally proposed, mediate very different vidual instances rather than abstracting the com- recognition outcomes. As noted above, with phe- monalities among them. In sum, although both notype matching animals can recognize their prior association and phenotype matching origi- familiar kin (e.g., individuals such as parents or nate from learning the cues of close kin during siblings encountered when recognition templates early development, how those cues are matched are learned) as well as previously unfamiliar kin, against conspecifics’ cues and whether unfamil- allowing for discrimination of kin and non-kin, iar kin can be recognized differ. as well as categories of kin (e.g., half-sibling, Finally, Waldman (1987) and Porter (1988) cousins, and second cousins). The two mecha- suggested that prior association by default nisms likely have different processes too, in involves matching of phenotypes; that is, the particular how the phenotypes learned during familiar individual’s traits are compared or early development are stored in memory or in a matched to an animal’s recognition template. template (see “Recognition templates”, below). While this is technically true (see next section), Prior association and phenotype matching such an argument obscures the important distinc- not only have distinct outcomes in terms of tion between these two recognition mechanisms. which kin can be recognized, but also likely have The perceptual matching algorithms fundamen- distinct physiological and perceptual processes. tally differ, with prior association allowing only 738 Mateo • ANN. ZOOL. FENNICI Vol. 41 exact matches, whereas phenotype matching per- out the genome, then all alleles, including those mits graded matches to the templates, and thus not involved in recognition, would benefit from recognition of individuals of varying kin classes. recognizing corresponding alleles in conspecif- In sum, I view these two recognition processes ics. as proximately and functionally different mecha- To demonstrate self-matching for kin recog- nisms, although their outcomes — differential nition, animals must be raised so that they do not treatment of conspecifics according to genetic experience kin cues other than their own during relatedness — can be similar. development. An ability to use their own cues as a referent for recognition would be demonstrated if they can later discriminate among unfamiliar Self-referent phenotype matching kin and non-kin. However, it is critical that ani- mals be reared without exposure to any kin cues Self-referent phenotype matching — the ability except their own, otherwise one would not be of animals to learn and use their own phenotypes able to rule out the possibility that their siblings’ as referents for recognition of relatives (termed or parents’ cues were learned (cf. Heth et al. the “armpit effect” by Dawkins 1982) — would 1998, see Mateo & Holmes 2004 for details). be the most accurate way of assessing related- Holmes and Sherman (1982) and Blaustein ness, as an animal’s own cues better reflect its (1983) noted the difficulty in differentiating genotype than those of its close kin. Self-match- between self-matching and recognition alleles as ing should be favored in species with multiple the mechanism underlying a particular recogni- paternity or maternity to discriminate among full tion event. First, our current knowledge does not and half siblings or when individuals commonly allow us to determine whether alleles involved encounter older (or younger) siblings after disper- in the expression and perception of recognition sal (Holmes & Sherman 1982, Sherman 1991). It cues also code for behaviors based on that rec- would also be favored when there is a risk of ognition, which would be necessary to conclude learning from non-kin, such as in brood parasitic recognition alleles as they are currently con- or communal nesting situations, as animals would ceptualized. Second, to demonstrate self-match- not have to learn the cues of familiar individuals ing one would have to manipulate an animal’s but instead could rely solely or largely on its own recognition cues during template formation, or cues. I have proposed elsewhere (Mateo & John- prevent it from learning its own cues, as it ston 2000a) that self-referent phenotype match- is assumed this mechanism involves learning ing would be especially useful in nepotistic situa- whereas recognition alleles do not (e.g., Hauber tions, when animals need to identify their closest et al. 2000). Blaustein also acknowledged that, kin, whereas in mate-choice or nest-defense con- in general, phenotype matching is a more parsi- texts use of additional referents, such as parents monious mechanism given the theoretical argu- and siblings, would be sufficient to avoid close ments against recognition alleles, but added that, inbreeding or to prevent entry of non-kin into the empirically, current data are consistent with both nest (see Hauber & Sherman 2001). The likeli- mechanisms. hood of a self-matching mechanism in nepotistic (compared with mating) contexts is controver- sial. Alexander (1990) argued against the evolu- Contextual and developmental changes tion of self-matching in nepotism because alleles in the perception component underlying such recognition would be “genetic outlaws”, benefiting themselves at a cost to the Different recognition mechanisms may be used remainder of the genome, and thus would be in different social contexts. For example, an suppressed by unlinked alleles not involved in adult may use spatial cues to identify its depend- the recognition process. Others (Dawkins 1982, ent young, prior association to avoid escalat- Hamilton 1987, Sherman 1991) have countered ing agonistic interactions with an aggressive that if alleles involved in both generating and neighbor, and phenotype matching to cooperate perceiving recognition cues are spread through- with a cousin. Later, when its young become ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 739 more independent and begin to move about the inclusion of kin and non-kin in recognition tem- environment, it may start to use prior association plates is advantageous. North American beavers to distinguish between its young and those of establish monogamous pairbonds and defend ter- another adult. For some animals, this ability to ritories along waterways. Natal dispersal also switch among recognition mechanisms may have occurs along these waterways, so it is likely that a developmental component. Because recogni- an unfamiliar younger sibling will pass through tion templates develop from early associations an older sibling’s territory. Not only can beavers with familiar kin (and/or with self), prior-asso- use phenotype matching to recognize these sib- ciation-based recognition abilities are expected lings, but their mates can as well, exhibiting dis- to precede phenotype-matching recognition crimination among its partner’s relatives (Sun & abilities. For example, young Belding’s ground Müller-Schwarze 1997). Mice and ground squir- squirrels discriminate among the cues of familiar rels can also use phenotype matching to discrim- and unfamiliar adults as early as 15-d of age, yet inate among non-kin (e.g., unfamiliar siblings of do not appear able to recognize unfamiliar kin an familiar but unrelated rearingmate, or unfa- via phenotype matching until 30 d (J. M. Mateo miliar kin of an unrelated odor donor; Holmes unpubl. data). This developmental pattern makes 1986b, Porter 1988, Aldhous 1989, Mateo 2002). adaptive sense, as young ground squirrels do not Phenotype matching among non-kin, encoun- encounter unfamiliar kin or non-kin until they tered after early development, may facilitate leave their natal burrow, at about 27-d of age. It social relationships within a colony, formation is important to recall that an animal may have of winter groups of communally nesting , the ability to recognize particular kin, yet not and recruitment of dispersing individuals into express that ability. For example, among com- new groups (e.g., Holekamp 1983, Wolff 1985, munally nesting and nursing species, a young Hare 1992). One should bear in mind, however, animal may know which female is its mother that incorporation of non-kin into recognition and yet nurse indiscriminately. If there is no cost templates may hinder accurate discrimination of nursing from unrelated or distantly related among classes of kin, unless some perceptual females, then selection will not favor expression mechanism or decision algorithm keeps kin and of recognition; that is, preferential nursing from non-kin phenotypes separate in the template. its mother (see also Roulin 2002). As characterized above, templates are formed during early development, when individuals Is phenotype matching true recognition interact exclusively or almost exclusively with of kin? kin such as parents and siblings. Such learning need not be confined to early development, how- Some researchers (e.g., Waldman 1987, Alexan- ever. It is possible, and often advantageous, to der 1990) have argued that “recognition” of unfa- incorporate new kin like older siblings or grand- miliar kin via phenotype matching is in fact an offspring into recognition templates. Non-kin error, as animals mistakenly identify a stranger, may also be learned, such as mates or territorial whose cues match those in the template, as a par- neighbors. Thus, learning of familiar individu- ticular familiar related individual (called “imper- als could continue throughout the lifespan, with fect social learning” by Alexander). In other social-recognition templates continually updated words, a paternal half-sibling would be treated as needed, so that the prior-association mecha- as kin because the identifying animal mistakes nism can be used to recognize individual famil- it for its familiar maternal half-sibling. Waldman iar kin and non-kin. Updated templates can also goes on to suggest that such mistakes would lead permit expanded phenotype matching, although to graded responses to unfamiliar kin, as more the extent to which this would be favored is distantly related kin would match the template unclear because there is a risk of incorporating less well than close kin, and would be treated non-kin into the template and thus a failure to accordingly. Yet this behavior is precisely what distinguish among kin and non-kin accurately. Hamilton’s rule would predict (assuming equal However, there may be circumstances when costs and benefits) if these unfamiliar kin were 740 Mateo • ANN. ZOOL. FENNICI Vol. 41 recognized correctly. Functionally, it does not prototype-based, depending on the recognition matter whether kin are recognized correctly or context (see also Breed & Bennett 1987). That mistaken for other kin, since the outcome of rec- is, individual representations may be used to dis- ognition would be favored regardless of its proc- criminate between a mother and an older sibling ess, as long as individuals are treated according or between two co-foundresses, whereas gestalt to coefficients of relationship. representations would be utilized when interact- ing with unfamiliar kin and non kin. When animals attempt to identify strangers Areas for future study by way of phenotype matching, which referents should they use for comparison? Multiple indi- Recognition templates viduals’ cues may be “stored” in a recognition template, such as those of siblings, parent(s), The word “template” has been used to refer to and even the animal itself, but one of the most animals’ memories of familiar kin, presumably notable gaps in our knowledge of recognition learned at some early point in development, yet mechanisms is which of these referents are uti- the meaning of the word is unclear. Do tem- lized and in which contexts (see also Mateo & plates exist in neural structures, like face-recog- Holmes 2004). That is, when an unfamiliar nition cells or song-nuclei circuits (e.g., Perrett attempts to enter a nest, does the guard bee com- et al. 1988, Margoliash 2002)? Or is “template” pare the stranger’s cues with those of the queen, simply a heuristic for conceptualizing recogni- itself, or its fellow workers? If multiple referents tion abilities? One can examine whether tem- are used, are they weighted equally, or are some plates consist of exemplars (e.g., specific memo- given more importance than others? As another ries of individuals’ phenotypes) or prototypes (a example, if extra-pair copulations are common, single amalgam or gestalt of several individuals’ young might weight their parents’ cues or their phenotypes), as described above. Many species own cues more than those of their siblings, some can discriminate among specific kin, such as of which may have been sired by a different sibling A and sibling B (e.g., Rendall et al. 1996, male, and therefore would be less accurate refer- Todrank et al. 1998, Mateo 2002; J. M. Mateo, ents for relatedness. unpubl. data), suggesting that templates contain Further, different referents could be used in some specific representation of these individuals. different situations. Cross-fostering studies in Yet when an unfamiliar relative is encountered, which halves of litters are transferred between such as a paternal half-sibling, do animals refer mothers have shown that young incorporate into to these separate representations for recognition their templates cues of the adult female, of their (necessitating multiple comparisons between genetic siblings, and of their foster siblings. In each representation and the stranger’s pheno- discrimination tasks fostered animals can rec- type), or do they utilize a gestalt template com- ognize their familiar nestmates, and can rec- prised of those representations (requiring just ognize unfamiliar relatives of their genetic and one global comparison)? Although the answers of their foster siblings (reviewed in Mateo & to such questions would further our understand- Holmes 2004). Although cross-fostering creates ing of the proximate bases of recognition, func- an unnatural situation, it can have biological rel- tionally, the outcome would be the same regard- evance, particularly when a clutch, brood or litter less of whether exemplars or prototypes are is multiply sired. By attending to differences in used, particularly if the referents in a template recognition cues produced by its full siblings and are weighted according to their relatedness to its half siblings, an animal can treat full siblings the animal (e.g., full-siblings’ traits are empha- preferentially, or it might be able to differentially sized more than half-siblings’ traits in either recognize its unfamiliar maternal and paternal template scenario). Since some animals can both kin, especially cues of their siblings are weighted recognize individual kin (via prior association) differentially. I note that such weighting might and “recognize” unfamiliar kin (via phenotype require, firstly, preferential weighting of its own matching), templates may be both exemplar- and cues, since these would be shared with its full ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 741 siblings more than with its half siblings (see also template, producing a threshold response (match Mateo & Johnston 2000a for an empirical exam- or not match), whereas unfamiliar individuals ple of weighting). would have to be evaluated against the gestalt family cues, with some decision algorithm used to produce a graded response according to relat- Neural correlates of perception edness.

What do the empirical results of recognition studies tell us about the perception component? Dissociation of recognition components Across taxa, unfamiliar stimuli are usually attended to longer than familiar stimuli (John- Recall that the components of recognition evolve ston 1981, Halpin 1986, Stoddard 1996). For separately, and that the action component — example, animals tend to antennate, smell or preferential treatment of kin, aggressive behav- look at unfamiliar individuals or most distantly iors toward non-kin, or mate choice — can only related kin longer and/or more often than famil- evolve if the expression and perception compo- iar individuals or close kin (e.g., Getz & Smith nents of the recognition process have already 1986, Dahbi & Lenoir 1998, Bull & Cooper evolved. Thus when kin-biased behaviors are 1999, Fadao et al. 2000, Mateo 2002). Although expected, but not observed, it could be due there are exceptions to this tendency, such inves- to a failure of the expression, perception, or tigatory behaviors suggest that it takes longer action components, and each possibility should to identify or recognize unfamiliar individuals be explored empirically. As an example, con- than familiar ones, as if additional time is needed sider the problems faced by male birds which for unknown cues to be compared or matched are unsure of their relatedness to nestlings in to templates to make an identification. A recent the nest shared with their social partner. Several study indicates that this may be an inaccurate studies have failed to find preferential investment conclusion, at least in some rodents. Uchida in chicks sired by the male (see Kempenaers & and Mainen (2003) showed that psychophysical Sheldon 1996), from which some have concluded discrimination of simple odors by trained rats that Hamilton’s rule (rB > C) is not satisfied. occurs in about 200 ms, regardless of the simi- That is, the risk of not caring for their own chicks larity of the odor stimuli and thus the difficulty outweighs the potential benefits of discriminating of the task. It appears then that fine-tuned per- against chicks sired by another male and thus the ceptual discrimination of odors does not require action component (rejection of non-kin) has not additional neural processing time (although how evolved. However, it is possible that chicks do quickly behavioral discrimination will be evi- not produce kin labels (concealing information dent is unclear). If these findings apply to other about their true father; Beecher 1991, Johnstone species as well, then how should we interpret 1997, but see Bonadonna et al. 2003 for possible differential investigation times? Although specu- olfactory cues to social recognition and Price lative, perhaps an unknown phenotype is quickly 1999 for possible auditory cues) or that males do perceived and characterized (e.g., in the olfac- not have the ability to make use of kin labels to tory bulb or antennal lobe), but additional inves- recognize their own chicks (Beecher 1988, Kem- tigation is required to match it to the recognition penaers & Sheldon 1996, see also Hatchwell et template and accurately determine the degree of al. 2001). Until such studies are done, it remains relatedness between the two individuals. Alter- unclear which of the three recognition compo- natively, prolonged investigation may serve to nents have or have not evolved when kin-differ- make unfamiliar stimuli familiar. Indeed, dis- ential behaviors are expected but not observed. crimination, and investigation duration, may be An additional corollary of this potential dis- quicker for identifying individuals via exem- association of recognition components is that plar-based template matching than for recogniz- an absence of behavioral discrimination or of ing unfamiliar kin via prototype-based template differential treatment of conspecifics cannot be matching. Familiar individuals would match the interpreted as a lack of recognition ability. Ani- 742 Mateo • ANN. ZOOL. FENNICI Vol. 41 mals may have the ability to recognize classes of the expression component, and perhaps the per- kin and non-kin, yet do not exhibit preferential ception and action components as well. treatment or avoidance of those classes in all social contexts. In those contexts, Hamilton’s rule may not favor discrimination despite accu- Conclusions rate recognition of kin (Gamboa et al. 1991). For example, asocial golden-mantled ground The perceptual component of social recogni- squirrels (S. lateralis) exhibit no kin-directed tion has a long theoretical and empirical his- behaviors as adults, such as cooperative territory tory. From amphibians to rodents and fish to defense or nepotistic alarm calls, unlike sym- social insects, animals use a variety of recogni- patric Belding’s ground squirrels (S. beldingi; tion mechanisms to discriminate among classes Sherman 1981, Michener 1983). Yet both species of conspecifics, such as kin, neighbors, colony are able to discriminate quite accurately among members and mates. These mechanisms vary not classes of relatedness, indicating that the expres- only among species but also among recognition sion and perception components have evolved in contexts. Future research should extend our cur- both species, but the action component involving rent knowledge of the perceptual mechanisms nepotistic behaviors has been elaborated only involved in recognition, to understand the neural in S. beldingi. It is possible that S. lateralis use processes involved in discrimination, the devel- their kin-recognition abilities in some other con- opment of recognition templates, the algorithms text, such as mate choice (Mateo 2002). used to determine identity, and how decision Another paradoxical example of an apparent rules are used to act (or not act) upon the result- lack of differential nepotism is found in social ing recognition. insects. In colonies with multiple matrilines (more than one queen, and thus more than one sire) or a single queen with multiple patrilines, Acknowledgments one might expect workers to tend to full sis- ters more so than half sisters, given the poten- I thank Phil Starks for inviting me to contribute to this spe- tially significant influence it would have on their cial issue. I also thank Warren Holmes and Paul Sherman for long discussions of kin-recognition mechanisms and Warren, mother’s reproductive success (and the work- Phil and an anonymous reviewer for comments on a prior ers’ indirect fitness). Yet empirical studies have version of this manuscript. My research is supported by the consistently failed to demonstrate unequivocally National Science Foundation and the National Institutes for differential treatment of larvae or workers based Mental Health. on within-colony differences in relatedness (e.g., Breed et al. 1994, Keller 1997, Strassmann et al. 2000, Tarpy et al. 2004). Recognition cues — in References the form of cuticular lipids — are acquired by all colony members from the nest, yet if individu- Aldhous, P. 1989: The effects of individual cross-fostering on the development of intrasexual kin discrimination in als’ cues retain some distinctiveness (for exam- male laboratory mice, Mus musculus L. — Anim. Behav. ple because of when or where in the nest they 37: 741–750. developed), then they could theoretically use Alexander, R. D. 1979: Darwinism and human affairs. — themselves as referents for discriminating within University of Washington Press, Seattle. the colony (Carlin 1989, Keller 1997). Chemical Alexander, R. D. 1990: Epigenetic rules and Darwinian algo- rithms. — Ethol. Sociobiol. 11: 241–303. analyses of hornet and paper- lipids indicate Altmann, J. 1979: Age cohorts as paternal sibships. — Behav. they are variable enough to facilitate discrimi- Ecol. Sociobiol. 6: 161–164. nation among matrilines within a nest, but are Barnard, C. J. 1990: Kin recognition: Problems, prospects, unlikely to permit accurate discrimination in and the evolution of discrimination systems. — Adv. nests composed of different patrilines (Dani et Stud. Behav. 19: 29–81. Beecher, M. D. 1982: Signature systems and kin recognition. al. 2004). Thus current data suggest that the lack — Am. Zool. 22: 477–490. of expressed within-colony discrimination by Beecher, M. D. 1988: Kin recognition in birds. — Behav. some social insects could be due to constraints in Genet. 18: 465–482. ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 743

Beecher, M. D. 1991: Successes and failures of parent-off- man, San Francisco. spring recognition in animals. — In: Hepper, P. G. (ed.), Estes, W. K. 1994: Classification and cognition. — Oxford Kin recognition: 94–124. Cambridge University Press, University Press, New York. Cambridge. Fadao, T., Tingzheng, W. & Yajun, Z. 2000: Inbreeding Beecher, M. D., Beecher, I. M. & Hahn, S. 1981: Parent- avoidance and mate choice in the mandarin vole (Micro- offspring recognition in bank swallows. II. Development tus mandarinus). — Can. J. Zool. 78: 2119–2125. and acoustic basis. — Anim. Behav. 29: 95–101. Gamboa, G. J., Reeve, H. K. & Holmes, W. G. 1991: Concep- Bekoff, M. 1981: Vole population cycles: Kin-selection or tual issues and methodology in kin-recognition research: familiarity? — Oecologia 48: 131. A critical discussion. — 88: 109–127. Blaustein, A. R. 1983: Kin recognition mechanisms: pheno- Getz, W. M. & Smith, K. B. 1986: kin recogni- typic matching or recognition alleles? — Am. Nat. 121: tion: learning self and nestmate phenotypes. — Anim. 749–754. Behav. 34: 1617–1626. Blaustein, A. R., Bekoff, M. & Daniels, T. J. 1987: Kin Göth, A. & Evans, C. S. 2004: Social responses without early recognition in vertebrates (excluding primates): Mecha- experience: Australian brush-turkey chicks use specific nisms, functions and future research. — In: Fletcher, D. visual cues to aggregate with conspecifics. — J. Exp. J. C. & Michener, C. D. (eds.), Kin recognition in Ani- Biol. 207: 2199–2208. mals: 333–357. John Wiley & Sons, Inc., New York. Göth, A. & Hauber, M. E. 2004: Ecological approaches to Bonadonna, F., Hesters, F. & Jouventin, P. 2003: Scent of species recognition in birds through studies of model and a nest: discrimination of own-nest odours in Antarctic non-model species. — Ann. Zool. Fennici 41: 823–842. prions, Pachyptila desolata. — Behav. Ecol. Sociobiol. Gottlieb, G. 1981: Roles of early experience in species- 54: 174–178. specific perceptual development. — In: Aslin, R. N., Breed, M. D. & Bennett, B. 1987: Kin recognition in highly Alberts, J. R. & Petersen, M. R. (eds.), Development of eusocial insects. — In: Fletcher, D. J. C. & Michener, perception, vol. 1: 5–44. Academic Press, New York. C. D. (eds.), Kin recognition in animals: 243–285. John Grosberg, R. K. & Quinn, J. F. 1986: The genetic control and Wiley & Sons Inc., New York. consequences of kin recognition by the larvae of a colo- Breed, M. D., Welch, C. K. & Cruz, R. 1994: Kin discrimi- nial marine invertebrate. — Nature 322: 456–459. nation within honey bee (Apis mellifera) colonies: An Gubernick, D. J. 1981: Parent and infant attachment in mam- analysis of the evidence. — Behav. Process. 33: 25–40. mals. — In: Gubernick, D. J. & Klopfer, P. H. (eds.), Brown, R. E., Roser, B. & Singh, P. B. 1990: The MHC and Parental Care in : 243–305. Plenum Press, individual odours in rats. — In: McDonald, D. W. & NY. Natynczuk, S. (eds.), Chemical signals in vertebrates V: Halpin, Z. T. 1986: Individual odors among mammals: Ori- 228–243. Oxford University Press, Oxford. gins and functions. — Adv. Stud. Behav. 16: 39–70. Bull, C. M. & Cooper, S. J. B. 1999: Relatedness and Hamilton, W. D. 1964: The genetical evolution of social avoidance of inbreeding in the lizard, Tiliqua rugosa. behaviour, I. II. — J. Theor. Biol. 7: 1–52. — Behav. Ecol. Sociobiol. 46: 367–372. Hamilton, W. D. 1987: Discriminating nepotism: Expect- Carlin, N. F. 1989: Discrimination between and within colo- able, common and overlooked. — In: Fletcher, D. J. C. nies of social insects: Two null hypotheses. — Neth. J. & Michener, C. D. (eds.), Kin recognition in animals: Zool. 39: 86–100. 417–437. John Wiley and Sons, NY. Charnov, E. L. & Finerty, J. P. 1980: Vole population cycles: Hampton, J. A. 1995: Testing the prototype theory of con- A case for kin-selection? — Oecologia 45: 1–2. cepts. — J. Memory Lang. 34: 686–708. Crozier, R. H. 1987: Genetic aspects of kin recognition: Hare, J. F. 1992: Colony member discrimination by juvenile Concepts, models, and synthesis. — In: Fletcher, D. J. Columbian ground squirrels (Spermophilus columbia- C. & Michener, C. D. (eds.), Kin recognition in animals: nus). — Ethology 92: 301–315. 55–73. John Wiley & Sons, Inc., New York. Hatchwell, B. J., Ross, D. J., Fowlie, M. K. & McGowan, Dahbi, A. & Lenoir, A. 1998: Nest separation and the dynam- A. 2001: Kin discrimination in cooperatively breeding ics of the Gestalt odor in the polydomous Cata- long-tailed tits. — Proc. R. Soc. Lond. B 268: 885–890. glyphis iberica (, Formicidae). — Behav. Hauber, M. E. & Sherman, P. W. 2001: Self-referent pheno- Ecol. Sociobiol. 42: 349–355. type matching: theoretical considerations and empirical Dani, F. R., Foster, K. R., Zacchi, F., Seppä, P., Massolo, A., evidence. — Trends Neurosci. 24: 609–616. Carelli, A., Arevalo, E., Queller, D. C., Strassmann, J. E. Hauber, M. E., Sherman, P. W. & Paprika, D. 2000: Self- & Turillazzi, S. 2004: Can cuticular lipids provide suffi- referent phenotype matching in a brood parasite: The cient information for within-colony nepotism in ? armpit effect in brown-headed cowbirds (Molothrus — Proc. R. Soc. Lond. B 271: 745–753. ater). — Anim. Cogn. 3: 113–117. Davies, N. B., Hatchwell, B. J., Robson, T. & Burke, T. 1992: Hepper, P. G. 1991a: Kin recognition. — Cambridge Univer- Paternity and parental effort in dunnocks Prunella mod- sity Press, Cambridge, United Kingdom. ularis: how good are male chick-feeding rules? — Anim. Hepper, P. G. 1991b: Recognizing kin: ontogeny and clas- Behav. 43: 729–745. sification. — In: Hepper, P. G. (ed.), Kin recognition: Dawkins, R. 1976: The selfish gene. — Oxford University 259–288. Cambridge University Press, Cambridge. Press, New York. Heth, G., Todrank, J. & Johnston, R. E. 1998: Kin recogni- Dawkins, R. 1982: The extended phenotype. — W. H. Free- tion in golden hamsters: Evidence for phenotype match- 744 Mateo • ANN. ZOOL. FENNICI Vol. 41

ing. — Anim. Behav. 56: 409–417. rels. — Anim. Behav. 59: 491–499. Holekamp, K. E. 1983: Proximal mechanisms of natal dis- Michener, G. R. 1983: Kin identification, matriarchies, and persal in Belding’s ground squirrels (Spermophilus beld- the evolution of in ground-dwelling sciurids. — ingi). — Ph.D. thesis, University of California, Berkeley. In: Eisenberg, J. F. & Kleiman, D. G. (eds.), Advances in Holmes, W. G. 1984: Ontogeny of dam-young recognition in the study of mammalian behavior: 528–572. American captive Belding’s ground squirrels (Spermophilus beld- Society of Mammalogists, Shippensburg, PA. ingi). — J. Comp. Psychol. 98: 246–256. Neff, B. D. & Sherman, P. W. 2002: Decision making and Holmes, W. G. 1986a: Identification of paternal half-siblings recognition mechanisms. — Proc. R. Soc. Lond. B 269: by captive Belding’s ground squirrels. — Anim. Behav. 1435–1441. 34: 321–327. Penn, D. & Potts, W. K. 1998: How do major histocompat- Holmes, W. G. 1986b: Kin recognition by phenotype match- ibility complex genes influence odor and mating prefer- ing in female Belding’s ground squirrels. — Anim. ences? — Adv. Immunol. 69: 411–436. Behav. 34: 38–47. Perrett, D. I., Mistlin, A. J., Chitty, A. J., Smith, P. A. J., Holmes, W. G. 1994: The development of littermate prefer- Potter, D. D., Brönnimann, R. & Harries, M. H. 1988: ences in juvenile Belding’s ground squirrels. — Anim. Specialised face processing and hemispheric asymmetry Behav. 48: 1071–1084. in man and monkey: evidence from single unit and reac- Holmes, W. G. 1997: Temporal aspects in the development tion time studies. — Behav. Brain Res. 29: 245–258. of Belding’s ground squirrels’ litter-mate preferences. Poindron, P. & Lévy, F. 1990: Physiological, sensory and — Anim. Behav. 53: 1323–1336. experiential determinants of maternal behaviour in Holmes, W. G. & Mateo, J. M. 2006: Kin recognition in sheep. — In: Krasnegor, N. A. & Bridges, R. S. (eds.), rodents — issues and evidence. — In: Wolff, J. O. & Mammalian parenting: biochemical, neurobiological, Sherman, P. W. (eds.), societies. University of and behavioral determinants: 133–156. Oxford Univer- Chicago Press, Chicago. [In press]. sity Press, New York, NY, USA. Holmes, W. G. & Sherman, P. W. 1982: The ontogeny of kin Porter, R. H. 1988: The ontogeny of sibling recognition in recognition in two species of ground squirrels. — Am. rodents: Superfamily Muroidea. — Behav. Genet. 18: Zool. 22: 491–517. 483–494. Hoogland, J. L. 1995: The black-tailed prairie dog: social Price, J. J. 1999: Recognition of family-specific calls in life of a burrowing . — University of Chicago stripe-backed wrens. — Anim. Behav. 57: 483–492. Press, Chicago. Pusey, A. & Wolf, M. 1996: in animals. Johnston, R. E. 1981: Attraction to odors in hamsters: An — Trends Ecol. Evol. 11: 201–206. evaluation of methods. — J. Comp. Physiol. Psychol. Queller, D. C., Ponte, E., Bozzaro, S. & Strassmann, J. 95: 951–960. E. 2003: Single-gene greenbeard effects in the social Johnstone, R. A. 1997: Recognition and the evolution of dis- amoeba Dictyostelium discoideum. — Science 299: tinctive signatures: When does it pay to reveal identity? 105–106. — Proc. R. Soc. Lond. B 264: 1547–1553. Reeve, H. K. 1989: The evolution of conspecific acceptance Keller, L. 1997: Indiscriminate altruism: unduly nice parents thresholds. — Am. Nat. 133: 407–435. and siblings. — Trends Ecol. Evol. 12: 99–103. Rendall, D., Rodman, P. S. & Emond, R. E. 1996: Vocal Keller, L. & Ross, K. G. 1998: Selfish genes: a green beard in recognition of individuals and kin in free-ranging rhesus the red fire ant. — Nature 394: 573–575. monkeys. — Anim. Behav. 51: 1007–1015. Kempenaers, B. & Sheldon, B. C. 1996: Why do male birds Richardson, D. S., Komdeur, J. & Burke, T. 2003: Altruism not discriminate between their own and extra-pair off- and infidelity among warblers. — Nature 422: 580. spring? — Anim. Behav. 51: 1165–1173. Robertson, R. J. 1990: Tactics and counter-tactics of sexually Liebert, A. E. & Starks, P. T. 2004: The action component of selected infanticide in tree swallows. — In: Blondel, recognition systems: a focus on the response. — Ann. J., Gosler, A., Lebreton, J. D. & McCleery, R. (eds.), Zool. Fennici 41: 747–764. Population biology of passerine birds, and integrated Margoliash, D. 2002: Evaluating theories of bird song learn- approach: 381–390. Spring-Verlag, Berlin. ing: implications for future directions. — J. Comp. Roulin, A. 2002: Why do lactating females nurse alien off- Physiol. A 188: 851–866. spring? A review of hypotheses and empirical evidence. Mateo, J. M. 2002: Kin-recognition abilities and nepotism — Anim. Behav. 63: 201–208. as a function of sociality. — Proc. R. Soc. Lond. B 269: Sherman, P. W. 1981: Kinship, demography, and Belding’s 721–727. ground squirrel nepotism. — Behav. Ecol. Sociobiol. 8: Mateo, J. M. 2003: Kin recognition in ground squirrels and 251–259. other rodents. — J. Mammal. 84: 1163–1181. Sherman, P. W. 1991: Multiple mating and kin recognition by Mateo, J. M. & Holmes, W. G. 2004: Cross-fostering as a means self-inspection. — Ethol. Sociobiol. 12: 377–386. to study kin recognition. — Anim. Behav. 68: 1451–1459. Sherman, P. W., Reeve, H. K. & Pfennig, D. W. 1997: Rec- Mateo, J. M. & Johnston, R. E. 2000a: Kin recognition and ognition systems. — In: Krebs, J. R. & Davies, N. B. the ‘armpit effect’: evidence of self-referent phenotype (eds.), Behavioural ecology: an evolutionary approach: matching. — Proc. R. Soc. Lond. B 267: 695–700. 69–96. Blackwell Scientific Publications, Oxford, United Mateo, J. M. & Johnston, R. E. 2000b: Retention of social Kingdom. recognition after hibernation in Belding’s ground squir- Smith, E. E. & Medin, D. L. 1981: Categories and concepts. ANN. ZOOL. FENNICI Vol. 41 • Perception component of kin-recognition mechanisms 745

— Harvard University Press, Cambridge. — Q. Rev. Biol. 46: 35–57. Smith, K. B., Alberts, S. C. & Altmann, J. 2003: Wild female Tsutsui, N. D. 2004: Scents of self: The expression compo- baboons bias their social behaviour towards paternal nent of self/non-self recognition systems. — Ann. Zool. half-sisters. — Proc. R. Soc. Lond. B 270: 503–510. Fennici 41: 713–727. Stoddard, P. K. 1996: Local recognition of neighbors by ter- Uchida, N. & Mainen, Z. F. 2003: Speed and accuracy of ritorial passerines. — In: Kroodsma, D. E. & Miller, E. olfactory discrimination in the rat. — Nat. Neurosci. 6: H. (eds.), Ecology and evolution of acoustic communica- 1224–1229. tion in birds: 356–376. Comstock Publishing Associates, Waldman, B. 1987: Mechanisms of kin recognition. — J. Ithaca, NY. Theor. Biol. 128: 159–185. Strassmann, J. E., Seppa, P. & Queller, D. C. 2000: Absence Whittingham, L. A., Dunn, P. O. & Robertson, R. J. 1993: of within-colony kin discrimination: foundresses of the Confidence of paternity and male parental care: an social wasp, Polistes carolina, do not prefer their own experimental study in tree swallows. — Anim. Behav. larvae. — Naturwissenschaften 87: 266–269. 46: 139–147. Sun, L. & Müller-Schwarze, D. 1997: Sibling recognition in Widdig, A., Nürnberg, P., Krawczak, M., Streich, W. J. the beaver: a field test for phenotype matching. — Anim. & Bercovitch, F. B. 2001: Paternal relatedness and Behav. 54: 493–502. age proximity regulate social relationships among adult Tang-Martinez, Z. 2001: The mechanisms of kin discrimina- female rhesus macaques. — Proc. Natl. Acad. Sci. tion and the evolution of kin recognition in vertebrates: U.S.A. 98: 13769–13773. A critical re-evaluation. — Behav. Process. 53: 21–40. Wolff, J. O. 1985: Behavior. — In: Tamarin, R. H. (ed.), Biol- Tarpy, D. R., Gilley, D. C. & Seeley, T. D. 2004: Levels ogy of New World microtus: 340–372. Special Publica- of selection in a social insect: a review of conflict and tion, The American Society of Mammalogists. cooperation during honey bee (Apis mellifera) queen Younger, R. M., Amadou, C., Bethel, G., Ehlers, A., Lin- replacement. — Behav. Ecol. Sociobiol. 55: 513–523. dahl, K. F., Forbes, S., Horton, R., Milne, S., Mungall, Todrank, J., Heth, G. & Johnston, R. E. 1998: Kin recogni- A. J., Trowsdale, J., Volz, A., Ziegler, A. & Beck, S. tion in golden hamsters: Evidence for kinship odours. 2001: Characterization of clustered MHC-linked olfac- — Anim. Behav. 55: 377–386. tory receptor genes in human and mouse. — Genome Trivers, R. L. 1971: The evolution of . Res. 11: 519–530.

This article is also available in pdf format at http://www.sekj.org/AnnZool.html