empirical and theoretical constraints on the evolution of

Empirical and Theoretical Constraintson the Evolution of Lactation

V. HAYSSENDepartment of Biological Sciences

Smith CollegeNorthampton, MA 01063

ABSTRACT

For 829 mammalian species, data onage at weaning or age at first solid foodwere analyzed with respect to bodymass, phylogeny, habitat, diet, length ofgestation, basal metabolism, and neona-tal development. The primary influenceon lactation length is female mass, butphylogenetic constraints are important.Thus, lactation can be characterized asshort (earless seals and baleen whales),long (marsupials, bats, and primates), oraverage (remaining eutherians). Amongaverage mammals, lagomorphs haveshort lactations.

Lactation may have different func-tions, evolutionary constraints, and phys-iological control depending on whetheryoung first eat solid food near weaningor well before it. First solid food eatennear weaning occurs in polytocous spe-cies with altricial young; in this case,lactation has a clear energetic role. Incontrast, first solid food well beforeweaning is common for mammals withsingle, precocial offspring. For these spe-cies, the energetic and nutritional con-straints on lactation may be less impor-tant than the benefits of maintainingcontact between mother and young, suchas reduced juvenile mortality and in-creased opportunities for learning socialor foraging patterns. Thus, the age atfirst solid food relative to the age atweaning may indicate the function oflactation within the reproductive biologyof a given mammal.

Delayed development and implanta-tion alter the timing of energetic invest-

ment during gestation, so too, the age atfirst solid food may alter or reflect therate of energetic investment during lacta-tion. Thus, the age at first solid foodrelative to the age at weaning may indi-cate the function of lactation within thereproductive biology of a given mam-mal. Testing these hypotheses will re-quire data from diverse species on thenutritional and energetic value of milkbefore and after first solid food as wellas on the mechanics and consequences ofnursing or suckling during the course oflactation.(Key words: evolution of lactation, al-lometry, reproductive effort, parental in-vestment)

Abbreviation key: BMR = basal metabolicrate.

INTRODUCTION

Lactation is the quintessence of mammals.Like the evolution of mammals, the evolutionof lactation has incorporated many organiza-tional changes. Biochemical changes involvecarbohydrate, fat, and protein synthesis in thefemale and their catabolism in her young.Gland structure and location are anatomicallyand developmentally modified in females, asare cheek musculature and tooth growth inyoung. Physiological changes include modifi-cations not only of hormonal control and me-tabolism, but also of the behaviors of mothersand their young in order to implement and tointegrate nursing and suckling. Lactation altersthe timing of reproduction to mesh ecologi-cally, energetic demands with seasonal re-source abundance. To achieve this organiza-tional integration, the genetic frameworksupporting lactation and other reproductivetraits is likely to be replete with pleiotropy andpolygeny. Thus, the selection pressures on lac-

Received July 31, 1992.Accepted November 23, 1992.

1993 J Dairy Sci 76:3213-3233 3213

3214 HAYSSEN

tation are, and have been, numerous and com-plex.

Most mammalian traits (e.g., body length,number of toes, and dentition) are characteris-tics of individuals, but lactation and otherreproductive characteristics are not. In particu-lar, the length of lactation is not determinedsolely by the mother or by the young but ratherdepends on their interaction. Selection acts onat least two genetic complements simultane-ously. Reproductive characteristics, such aslactation, are characteristics of family groups,not of individuals. The evolutionary processesthat affect reproductive characteristics have notbeen delineated, but trait group or interdemic(between populations) selection may be impor-tant (11). The study and characterization oflactation solely on the molecular, cellular, orindividual level miss the essential character ofthis phenomenon.

This paper explores lactation across mam-malian species and assesses the relative impor-tance of diverse biological aspects (for exam-ple, habitat, diet, body mass, and phylogenetichistory) to the temporal and energetic charac-teristics of lactation. Temporal aspects will beemphasized because the energetic aspects (milkquality and quantity) have already receivedextensive analysis and, more importantly, be-cause time itself is a limiting resource and apotent evolutionary pressure. Reproductionmust fit into a seasonal climatic cycle, wheresuch a cycle exists. Additionally, reproductivesuccess depends on the production of as manyviable offspring as possible over a lifetime; thesooner that these offspring can be produced,the greater is the selective advantage (2). Timeis essential for parents, for offspring, and forpopulations during periods of exponentialgrowth and in saturated environments. Thus,the duration of lactation is an important aspectof the reproductive biology of mammals. Anadditional advantage of this particular analysisis to provide the means to estimate the lacta-tion lengths of extinct, rare, or unstudied mam-mals based on maternal mass and phylogeny.

MATERIALS AND METHODS

Data on lactation lengths have beenrecorded for about one-fifth of the knownmammals, representing 19 or 20 taxonomicorders. Of these 829 species, age at weaning

(lactation length) is available for 753 species,and age at first solid food is available for 419species. These and additional data on milkcomposition, length of gestation (including di-apause and implantation delays), developmentat birth, litter size, and adult female mass weresynthesized from over 12,000 sources (1, 5, 7).Common log transformations were performedon the age at weaning, age at first solid food,length of gestation, and adult female mass toimprove the symmetry of each distributionacross species and the uniformity of spreadacross orders (8). Not all data were availablefor each species. The age at first solid food ismore precisely and uniformly defined than theage at weaning. Unfortunately, the age at firstsolid food is recorded about half as often.More observations of the age at first solid foodare needed.

Operational definitions of the length of lac-tation (the age of weaning) vary according tothe needs of the investigator. Unless this varia-bility is correlated with a more natural aspectof mammalian biology (for instance, bodymass or phylogeny), subsequent analysis ofrelationships, although not biased, will bemore difficult to elucidate. In some cases, theneeds of investigators probably are confoundedwith phylogeny (e.g., primatologists study pri-mates; the whaling industry studies whales).Unfortunately, the biases inherent in thesedifferent approaches to mammalian groupshave not been precisely delineated. Thus, theireffect on comparative analyses are unknown.

Several temporal units are used to recordthe length of reproductive stages: days, weeks,months, and years. As units, weeks andmonths have little biological relevance becausemost species are unaware of our human meas-urement of time. Measurement of lactation andother reproductive stages in these units is notbiologically meaningful and should beavoided. The use of months as a unit is espe-cially problematic because the measure cannotbe accurately rescaled into another time frame.A single month can range from 28 to 31 d. Forthis report, all temporal measures have beenconverted to days, and a month is assumed tobe 30 d.

For this paper, lactation length is consideredto be a dependent variable, and description ofits variance with respect to a variety of othervariables is the major goal. Statistical treat-

Journal of Dairy Science Vol. 76, No. 10, 1993

SYMPOSIUM: LIMITS TO LACTATION 3215

ment was by least squares regression or analy-sis of covariance (5, 9). Least squares regres-sion was chosen over reduced major axisregression to maintain a constant dimensional-ity of slopes across different data sets. Statis-tics for significant regression analyses arereported. For each set of regressions, a sig-nificance of .05 across sets was approximatedby reducing the significance for individualregressions by the number of regressions per-formed for that analysis. Thus, to be reported,individual regressions for analysis by in-fraclass needed P < .0167 (.05/3 infraclasses),whereas ordinal comparisons needed P < .0025(.05/20 orders), and familial regressions neededP < .0004 (.05/136 families).

Several dimensionless ratios were calcu-lated to explore mammalian lactation. Al-though they are not amenable to rigorous para-metric analysis, these ratios, nevertheless, areextremely useful for illustrating multidimen-sional trends. The ratio of age at first solidfood to age at weaning may be an estimate ofthe relative nutritional dependency of the off-spring on lactation. The ratio of lactationlength to the period between conception andweaning (gestation plus lactation) is an esti-mate of the relative contribution of lactation toa female’s temporal investment in reproduc-tion. Although the energetic costs of construc-tion and maintenance of placental structuresare not included, the ratio of litter mass at birthto female mass is an estimate of the energeticinvestment into a given reproductive effort be-fore lactation begins. Similarly, the ratio of anindividual neonate’s mass to maternal mass isan estimate of the energetic investment into asingle offspring before lactation begins and anestimate of the reliance on lactation for thephysical growth of an individual offspring. Ofcourse, for single litters, litter mass and neona-tal mass are identical.

Taxonomic Representation

The ordinal representation of the species inthis data set is roughly similar to that formammals. Small mammals (Insectivora,Chiroptera, and Rodentia) are underrepresentedby 5 to 10%, whereas marsupials, primates,and carnivores are overrepresented by similaramounts. For each mammalian order, the num-ber of species with data on the age at weaning

or first solid food and the common names ofrepresentative species follow: infraclass Pro-totheria, order Monotremata (3 species; echid-nas and platypus); infraclass Metatheria, orderMarsupialia (89 species; opossums, phalangers,kangaroos, wombats, and koala); infraclass Eu-theria, orders Xenarthra or Edemata (9 species;anteaters and sloths), Insectivora (38 species;hedgehogs, shrews, and moles), Scandentia (3species; tree shrews), Dermoptera (no data; fly-ing lemur), Chiroptera (75 species; bats), Pri-mates (64 species; lemurs, monkeys, apes, andhumans), Carnivora (128 species; wolves,bears, raccoons, weasels, genets, mongooses,hyena, cats, fur seals, walrus, and earlessseals), Cetacea (28 species; dolphins, por-poises, and whales), Sirenia (3 species; dugongand manatees), Proboscidea (2 species;elephants), Perissodactyla (9 species; zebras,tapirs, and rhinoceroses), Hyracoidea (3 spe-cies; hyraxes), Tubulidentata (1 species; aard-vark), Artiodactyla (82 species; swine, camels,hippopotamus, deer, giraffe, cattle, antelope,sheep, and goats), Pholidota (3 species; pango-lins), Rodentia (270 species; squirrels, gophers,kangaroo rats, beaver, voles, hamsters, gerbils,mice, rats, and porcupines), Lagomorpha (15species; pikas, hares, and rabbits), and Macro-scelidea (4 species; elephant shrews). The in-fraclasses Metatheria and Eutheria are referredto jointly as the taxon Theria. The Appendixprovides selected familial common names.

RESULTS AND DISCUSSION

Univariate Statistics

Age at Weaning. The length of lactation inmammals (Figure 1) ranges almost three ordersof magnitude, from 4 to 5 d in hooded seals(Cystophora cristata, family Phocidae, orderCarnivora), spiny rats (Proechimys guairae,family Echimyidae, order Rodentia), and ele-phant shrews (Macroscelides proboscideus, fa-mily Macroscelididae, order Macroscelidea) toover 900 d in some of the great apes (chimpan-zees, Pan troglodytes, and orangutan, Pongopygmaeus, family Pongidae, order Primates).Although extremely short lactation lengths(<10 d) are rare, long lactation lengths (>500 d)are common for large-bodied species with sin-gleton offspring, such as kangaroos, great apes,walruses, whales, sirenians, elephants, and

Journal of Dairy Science. Vol. 76, No. 10, 1993

3216 HAYSSEN

rhinoceroses. Overall, 50% of known lactationlengths range from 29 to 135 d. The medianlactation length is 60 d, the geometric mean is64.56 d (1.81 log10 d, SD = .428), and thearithmetic mean is 108.4 d (SD = 132.1). Thepeak at about 30 d may be an artifact of usinghuman-biased units such as weeks or monthsrather than more biologically meaningful unitssuch as days.

Generally, lactation lengths for prototheri-ans (monotremes) and metatherians (marsupi-als) are longer than those for eutherians (me-dian lactation length: prototherians, 150 d, 3species; metatherians, 120 d, 75 species; eu-therians, 50 d, 675 species), although lactationlengths are longest in eutherians. Lactation inmonotremes and marsupials, but not eutheri-ans, encompasses the exponential period ofoffspring growth. Additionally, lactation inmarsupials is distinctly biphasic. During thephase of continuous teat attachment, marsupialmilk is high in carbohydrates and low in lipids,whereas the reverse holds when suckling isintermittent (3, 10). Occasionally, the teat at-tachment phase of marsupial lactation isequated with the chorioallantoic phase of eu-therian gestation. Superficial similarities exist,but this synonomy ignores the vastly differentproximate mechanisms (metabolic and physio-logical control) and ultimate causes (selectivepressures and evolutionary constraints) thatdistinguish gestation and lactation from eachother.

Age at First Solid Food. Fewer data existfor the age at first solid food. Solid food is

eaten within wk 1 after birth (occasionally onthe day of birth) in hystricomorph rodents,such as guinea pigs, porcupines, coypus, andagoutis, as well as in other mammals, such assloths, zebras, hyraxes, and bovids. In contrast,first solid food is nearly coincident with wean-ing in many sciurid (e.g., squirrels) and muroid(e.g., rats or gerbils) rodents and may evenoccur after weaning, as with earless seals. Ageat first solid food ranges from 1 to 300 d; 50%of the observations are from 14 to 50 d. Themedian age is 30 d, the geometric mean is27.54 d (1.44 log10 d, SD = .455), and thearithmetic mean is 45.1 d (SD = 50.9). Themedian age at first solid food is later formetatherians (75 d, 72 species) than for eu-therians (22 d, 345 species). Sufficient data onfirst solid food for prototherians are lacking.

Comparisons and Analysis. For both age atweaning and age at first solid food, the medianand the geometric mean are within 10% ofeach other, but the arithmetic mean is at least50% larger. Thus, the log10-transformed datamore closely approximate Gaussian distribu-tions for both variables.

The interquartile range of the age at firstsolid food (36 d, 419 species) is about a thirdthat of the age at weaning (106 d, 753 species).The smaller sample size for age at first solidfood may account for its smaller range, or theage at first solid food may be less flexible andmore uniform across mammals than the age atweaning.

Age at weaning and age at first solid foodare positively correlated (regression statistics: r= .58; R2 = .34; F = 176; sample size, 343species). Species that are older at weaning arealso older at first solid food. Age at weaningand age at first solid food are both positivelycorrelated with female mass, although the ef-fect is weak for age at first solid food (seeTable 1, Equation [1] for age at weaning statis-tics; regression statistics age at first solid food:r = .19; R2 = .03; F = 15; sample size, 408species).

Analysis of the ratio of age at first solidfood to age at weaning effectively removes theconfounding influence of body mass. Ingeneral, the longer the lactation length, thesooner first solid food occurs relative to wean-ing. For eutherians with lactation lengths of atleast 1 yr (21 species), most (90%) have an ageat first solid food within the first third of the

Figure 1. Frequency of lactation lengths in 753 speciesof mammals. The peak at 30 d may be an artifact of theuse of anthropocentric units (e.g., weeks or months) ratherthan biologically meaningful units (e.g.. days).



lactation period. For 62% of these species,solid food is eaten within the first fifth oflactation. In contrast, for eutherians with lacta-tion lengths <50 d (123 species), few (11%)have an age at first solid food within the firstthird of lactation. For 48% of these species,first solid food occurs during the last third oflactation. Thus, infants depend completely onmaternal resources for a larger proportion ofthe lactation period in species with short lacta-tions, whereas offspring may be less reliant onmaternal nutrients when lactation lengths arelong. The timing of first solid food relative toweaning is also correlated with litter size andwith neonatal development as described morefully later.

Allometry of Lactation: The Relationshipof Lactation to Maternal Mass

For mammals, the length of lactation ispositively correlated with adult female mass(Figure 2; Table 1, Equation [1]). In general,larger species have longer lactation lengths,although several interesting exceptions to thispattern exist.

Earless seals (family Phocidae) and baleenwhales (infraorder Mysticeti), but not othermarine mammals, have much shorter lactationlengths than expected for their body masses.The extremely short lactation lengths of earlessseals appear to be related to the unpredictableand short existence of the ice floes where theybreed. The relatively short lactation lengths ofbaleen whales may reflect the timing of theirannual migrations, because a lactation lengthappropriate to their large mass would not besynchronized with the seasonal nature of their

food and activity. Both groups (especially thebaleen whales) exert great statistical influenceon the computed relationship between lactationlength and body mass. Removal of these twogroups from the calculations generates a morerepresentative numerical relationship for mam-mals in general (Table 1, Equation [2]).

The remaining mammals are still not ahomogeneous group, because several mam-malian taxa have distinctly long lactationlengths for their mass. Bats, primates, andmarsupials together form a homogeneousgroup of mammals with lactation lengthslonger than those of other mammals with simi-lar maternal mass. Covariance analysis con-firms this observation (Table 1, Equations [3]and [4]). The slopes of the two regression linesare not statistically distinguishable; thus, therelationship between lactation length and bodymass is not different between the two groups.However, the intercepts of the equations aredifferent. Bats, marsupials, and primates, as agroup, have lactation lengths 50% longer (on alog10 scale) than other therians of comparablemasses. The Prototheria (monotremes) are in-termediate. Although the reproductive biologyof marsupials is generally different than that ofeutherians (6), marsupials are indistinguishablefrom bats and primates with respect to lacta-tion length. Thus, for lactation length, in-fraclass comparisons are not appropriate.

Why should mammals with long and aver-age lactation lengths have similar allometricrelationships? One explanation is that thesimilarity of slope between long and averagemammals is accidental. In other words, bats,marsupials, and primates each have a separate,selective advantage for long periods of lacta-

mammalian groups.1TABLE 1. Allometric relationships between the age at weaning (days) and adult female mass (grams) for various

Slope SE Constant SE R2 n F

Mammalia .184 .008 1.275 .026 .43 732 543Mammalia2 (no short) .214 .008 1.212 .025 .51 705 743Long lactations3 .227 .012 1.501 .033 .62 201 329Average lactation4 .250 .006 .977 .021 .75 504 1517

1All variables are log10-transformed. Sample size is number of species.2Therian mammals without earless seals (Phocidae), baleen whales (Mysticeti).3Includes marsupials (Marsupialia), primates (Primates), and bats (Chiroptera) only.4Eutherians, excluding primates (Primates), bats (Chiroptera), earless seals (Phocidae), and baleen whales (Mysticeti).


3218 HAYSSEN

tion, but, because the three taxa are adjacentand slightly overlapping with respect to bodymass, they form a single group with the sameslope as that for other mammals. However, ifthe slope is regarded as a true biologicalphenomenon (i.e., that an ancestral allometryof mass with lactation length exists for mam-mals), then the convergence of marsupials,bats, and primates needs explanation.

Given the fetal state of their neonates, therelatively long lactation lengths of metatheri-ans (marsupials) are not surprising; however,neither bats nor primates have such tiny off-spring relative to maternal mass. Some meta-bolic parameter or growth rate may possiblyexplain the allometric convergence. Thelengthy primate lactations may be related tothe more extensive psychosocial growth anddevelopment characteristic of the order. Asimilar developmental phenomenon may occurin bats that is related to the extensive neu-romuscular coordination required for food ac-quisition in flight. Perhaps newborn bats andprimates are as “fetal” in their neurologicaldevelopment relative to their adult needs as

neonatal marsupials are physically. Compara-tive data on the developmental similarities ofmarsupials, bats, and primates and on the qual-ity and quantity of milk over the course oflactation in these groups, especially bats, areneeded to explore this empirical relationship.

Thus, mammalian lactation lengths relativeto maternal mass can be characterized as short(earless seals and baleen whales), long (marsu-pials, bats, and primates), or average (the re-maining eutherians). For this last (average)group, 75% of the variation in the length oflactation is related to maternal mass. Thecorrelation of lactation length with body massis stronger for larger species (>1 kg vs. <=1 kg),but this result is confounded by phylogeny,because about 75% of the small mammals withaverage lactation lengths are rodents.

Eutherian Ordinal Analyses. Although lac-tation is strongly correlated with maternalmass across mammalian species, within in-dividual eutherian orders, the relationship be-tween female mass and lactation length is lessobvious (Figure 3). Many systematically dis-tinctive orders (tree shrews, sirenians,

Figure 2. Allometric relationship of lactation length with female mass (732 species). Marsupials, bats, and primatesform a homogenous group with long lactations. Baleen whales and earless seals have unusually short lactations. Key toorders: infraclass Prototheria; order Monotremata (P), infraclass Metatheria, order Marsupialia (+); infraclass Eutheria,orders Artiodactyla (A), Chiroptera (B), Carnivora (C), Proboscidea (E), Pholidota (F), Hyracoidea (H), Insectivora (I),Scandentia (K), Lagomorpha (L), Primates (M), Rodentia (R), Sirenia (S), Tubulidentata (T), Cetacea (W), Xenarthra (X),Macroscelidea (Y), and Perissodactyla (Z). Common names are given in Materials and Methods.



Figure 3. Box plots of lactation length (log10-transformed age at weaning in d) for 753 mammalian species groupedby order and ranked by median maternal mass (provided beneath name of order). Sample size in number of species islisted in the right column. Lagomorphs have relatively uniform and short lactations. Carnivore lactation lengths have thebroadest range. Median (+), interquartile range ( • ), approximate 95% confidence interval for median ( ), outliers, thevalues distant from the median by >1.5 times the interquartile range (*), and “whiskers”, the range excluding outliers (—).Names of common representatives of each order are given in Materials and Methods.


3220 HAYSSEN

elephants, hyraxes, aardvarks, pangolins, andelephant shrews) have too few species for anal-ysis. For others, such as xenarthrans, peris-sodactyls, and lagomorphs, the absence of astatistical relationship between lactation lengthand female mass may be an artifact of thenarrow range of body masses present acrossspecies within these orders. However, therange of body masses in each of several otherorders (insectivores, bats, carnivores, and ceta-ceans) spans three to four orders of magnitude;however, for these groups, lactation length alsois not correlated with maternal mass. Only forprimates, artiodactyls, and rodents is lactationsignificantly related to maternal mass, and onlyfor primates and artiodactyls does body massexplain >50% of the variance in lactationlength (Table 2). For eutherian ordersrepresented in the data set by at least 3 fami-lies or 15 species, descriptions of the allometryof lactation length follow (Figures 4 to 7).

The four families of xenarthrans (which in-cludes anteaters, two- and three-toed sloths,and armadillos) are each remnants of previ-ously diverse taxa. Data for 6 of the 29 species(21%) are available. Although lactation lengthsare highly varied among species with similarfemale mass (Figure 4a), overall, they are nearthose expected for average mammals ofequivalent size.

The six families of the order Insectivora(which includes hedgehogs, tenrecs, shrews,moles, etc.) are a taxonomic mixture with littleclose phylogenetic relationship. Data for 36 ofthe 379 species (10%) are available. Maternalmass varies over almost three orders of magni-tude, but lactation lengths are nearly all be-tween 20 and 50 d (median 28 d). Thus, withinthis order, the duration of lactation and femalemass are not correlated (Figure 4b). However,phylogeny confounds this analysis: shrews(family Soricidae) tend to be small; moles(family Talpidae) are larger; tenrecs (familyTenrecidae) are intermediate; and hedgehogs(family Erinaceidae) are often the largest. Al-though Insectivora generally have lactations ofaverage duration for mammals of their mass,the small soricids (shrews <10 g) tend to havelonger lactation lengths than expected.

With at least 17 families, 175 genera, andover 900 species, bats (order Chiroptera) aresecond only to rodents in phylogenetic diver-sity. Unfortunately, data on lactation are avail-able for only 8% of the species (representing37 genera and 10 families). For 85% of theselimited observations, lactation length is be-tween 30 and 90 d. Maternal mass spans nearlythree orders of magnitude but accounts foronly 8% of the variation in the length of

TABLE 2. Regression statistics for significant allometric relationships between the age at weaning (days) and adultfemale mass (grams) for various mammalian taxa.1

Slope SE Constant SE R2 n F

InfraclassMetatheriaEutheria

OrderMarsupialiaPrimatesArtiodactylaRodentiaFissipedia2

FamilyDasyuridaeMacropodidaeMustelidaeBovidae

.197 .017 1.628 .047 .64 74 129

.191 .008 1.210 .026 .48 655 608

.197 .017 1.628 .047 .64 74 129

.334 .043 1.104 .148 .51 59 61

.271 .032 .831 .154 .51 69 73

.150 .015 1.153 .032 .30 242 103

.180 .026 1.238 .098 .36 85 49

.137 .020 1.771 .038 .64 27 46

.340 .038 1.172 .141 .85 15 82

.168 .031 1.314 .100 .53 25 28

.237 .040 1.001 .195 .45 42 34

1For each set of regressions, a significance level of .05 was approximated by reducing the significance for individualregressions by the number of regressions performed for that analysis: infraclass regressions (P < .0167); ordinalregressions (P < .0025); and familial regressions (P < .0004). Sample size is number of species. All variables are log10-transformed.

2Carnivora, excluding the Pinnipedia (Otariidae, Odobenidae, Phocidae).



Figure 4. Allometric relationships of age at weaning with maternal mass for a) edentates, b) insectivores, and c) bats.Species from the same family have the same letter code. Key to families within each order: a. Edentata or Xenarthra:Bradypodidae (B), Choloepidae (C), Dasypodidae (D), and Myrmecophagidae (M): b. Insectivora: Chrysochloridae (C),Erinaceidae (E), Tenrecidae (N), Solenodontidae (O), Soricidae (S), and Talpidae (T); c. Chiroptera: Emballonuridae (E),Phyllostomidae (F), Megadermatidae (G), Rhinolophidae (L), Molossidae (M), Noctilionidae (N), Pteropodidae (P),Rhinopomatidae (R), Vespertilionidae (V), and Nycteridae (Y). Familial common names are given in the Appendix.


3222 HAYSSEN

Figure 5. Allometric relationships of age at weaning with maternal mass for a) primates, b) carnivores, and c)Cetaceans. Species from the same family have the same letter code. Key to families within each order: a. Primates:Cebidae (B), Cercopithecidae (C), Daubentoniidae (D), Galigonidae (G), Hominidae (H), Indridae (I), Cheirogaleidae (K),Lemuridae (L), Callimiconidae (M), Lorisidae (O), Pongidae (P), and Callitrichidae (Y); b. Carnivora: Protelidae (A),Canidae (C), Felidae (F), Herpestidae (H), Mustelidae (M), Otariidae (O), Procyonidae (P), Phocidae (S), Ursidae (U),Viverridae (V), Odobenidae (W), and Hyaenidae (Y); c. Cetacea: Balaenopteridae (B), Delphinidae (D), Eschrichtidae (E),Monodontidae (M), Phocoenidae (P), Platanistidae (R), Physeteridae (S), and Ziphiidae (Z). Familial common names aregiven in the Appendix.



lactation (Figure 4c). Insectivora and Chirop-tera have a similar range of female mass, andboth orders exhibit little correlation of bodymass with lactation length; however, the me-dian lactation length for bats (60 d) is overtwice that of the Insectivora (28 d).

Primates also lactate for long periods rela-tive to most mammals. Maternal mass isstrongly and positively correlated with lacta-tion length (Figure 5a) and accounts for >50%of the variation in the length of lactation (Table2). Although the ordinal trend (based on 33%of the known species) is clear, no strong corre-lation between mass and lactation length existswithin any primate family.

Of mammalian orders with >20 species,carnivores and artiodactyls are the best known.Each has lactation length data recorded for43% of its species. The increasing variance inlactation length with body mass in carnivoresis distinctive. Nearly all female carnivores <10kg lactate from 30 to 130 d. However, of thosecarnivores >10 kg, some ursids (bears), mostotarids (fur seals), and odobenids (walruses)have lactation lengths >130 d, whereas manyphocids (earless seals) wean their young <30 dafter birth. Exclusion of the pinnipeds im-proves the homoscedasticity and generates asignificant allometric equation (Fissipedia, Ta-ble 2). Lactation lengths for mustelids (e.g.,weasels, skunks, and badgers) are strongly andpositively correlated with maternal mass (Table2), and the few data for procyonids (raccoons)and viverrids (genets) also suggest a positivecorrelation with female mass. For the remain-ing carnivore families, lactation length appearsto be relatively independent of maternal mass(Figure 5b).

Few of the data on cetacean lactationlengths are from direct observations of livemother and calf pairs over the course of lacta-tion. Most data are estimates from the seasonalpresence of milk in adult females and in calveskilled by the commercial whaling industry.These data represent 36% of the known speciesand suggest that the large baleen whales lactatefor shorter periods than the smaller toothedwhales. Because smaller cetaceans have lacta-tion lengths more in keeping with their bodymass, the overall trend is an inverse relation-ship of lactation with body mass (Figure 5c).Like bats, toothed whales must catch in-dividual food items in a dynamic, three-

dimensional environment, and their longer lac-tation lengths may reflect this developmentalconstraint. Clearly, more precise data areneeded to confirm the unusual negative corre-lation of lactation and body mass in cetaceans.

Like the xenarthrans, the three families ofperissodactyls (which include zebras, tapirs,and rhinoceroses) are remnants from previ-ously more diverse taxa. Lactation data areavailable for 9 of 18 species (50%). The largerrhinoceroses have the longest lactation lengths,but, overall, the duration of milk production isonly weakly correlated with female mass(Figure 6a).

In contrast, over 50% of the variation inlactation length is correlated with maternalmass for artiodactyls (which include pigs, pec-caries, camels, hippopotamuses, deer, giraffes,and antelope; Figure 6b; Table 2). Bovids(antelope, cattle, sheep, and goats) constitute60% of the artiodactyla data set, and lactationlengths of bovids are strongly and positivelycorrelated with maternal mass (Table 2).

Lagomorphs come in three sizes (100- to500-g pikas, 500- to 1500-g rabbits, and2- to 3-kg hares), but lactation ranges onlybetween 17 and 28 d for the entire order,represented by 15 of 62 species (24%). Notonly do lagomorphs tend to have relativelyinvariant periods of lactation, they also haveshort lactation lengths compared with othermammals of similar mass (Figure 7a). Theleporids (hares and rabbits) have neonateswidely divergent in developmental condition atbirth; the larger hares produce precocial young,whereas the smaller rabbits produce altricialyoung. Perhaps as a result, first solid food isabout 1 wk earlier for precocial hares than foraltricial rabbits. Hares have shorter lactationsthan expected given their mass, but whetherthis difference is due to the production ofprecocial young or to a general trend towardshort lactation lengths within lagomorphs isnot clear. Delineation of the differences be-tween hares and rabbits in the physiology andecology of lactation could be especially usefulbecause the two groups are both polytocous,they have similar diets and phylogenetic ances-try, and they can have sympatric populations(i.e., they coexist in the same location). Thus,the two groups form a natural control for theeffects of diet, habitat, and phylogeny on theproduction of altricial versus precocial young.


3224 HAYSSEN

Rodents are the largest mammalian order,and lactation data are available for about 15%of the >1700 species. Maternal mass rangesfrom 6 g to 35 kg, but almost 90% of thelactation lengths are between 10 and 50 d, andhalf are between 21 and 38 d (Figure 7b).Thus, for rodents, lactation is only modestlycorrelated with maternal mass (Table 2).Sciurid (e.g., squirrels) rodents have long lacta-tion lengths relative to maternal mass.Cricetids (e.g., hamsters, gerbils, and deermice)may have shorter lactation lengths than murids(e.g., rats and house mice) of similar size,although the overlap between them is large.

Reproductive Effort and the Length of Lactation

Temporal Investment. Time and energy aretwo components of maternal investment. Mam-mals can dissociate the temporal and energeticcomponents of reproductive investment duringgestation, through diapause and delayed im-plantation, and during lactation, via early firstsolid food and altered milk composition orproduction. Although the energetic demands ofa current reproduction effort clearly affect fu-ture reproductive success, temporal aspects arealso important, especially in temperate or otherclimatically seasonal environments. Appropri-ate timing of the energetic demands of

Figure 6. Allometric relationships of age at weaning with maternal mass for a) odd-toed ungulates (perissodactyls) andb) even-toed ungulates (artiodactyls). Species from the same family have the same letter code. Key to families within eachorder: a. Perissodactyla: Equidae (E), Rhinocerotidae (R), and Tapiridae (T); b. Artiodactyla: Bovidae (B), Camelidae (C),Cervidae (D), Giraffidae (G), Hippopotamidae (H), Tragulidae (M), Suidae (S), and Tayassuidae (T). Familial commonnames are given in the Appendix.



reproduction to coincide with periods of re- importance of females and offspring in thesource abundance and environmental suitabil- evolution of a particular reproductive pattern.ity increases reproductive success. The length A female’s investment in her offspring be-of lactation is determined by the timing of gins at conception; thus, lactation is only oneweaning. Yet, mothers and their offspring may component of reproductive effort, and the tim-be in conflict regarding the timing of weaning ing of weaning depends, in part, on the lengthrelative to the seasonality of resource abun- of gestation. Marsupials may spend >90% ofdance. The energetic demands on females the period between conception and weaning onusually are largest before weaning, but the lactation, and phocid seals may devote <5% ofperiod after weaning is the most demanding their temporal investment on lactation, butfor offspring. The length of lactation and the these groups are exceptional (Figure 8). Giventiming of weaning may reflect the relative that mammalian gestation and lactation lengths

Figure 7. Allometric relationships of age at weaning with maternal mass for a) pikas, hares, and rabbits (lagomorphs)and b) rodents. Species from the same family have the same symbol or letter. Key to families within each order: a.Lagomorpha: Leporidae (L) and Ochotonidae (O); b. Rodentia: Muridae (+), Sciuridae (o), Cricetidae (x), Arvicolidae (A),Bathyergidae (B), Castoridae (C), Dasyproctidae (D), Echmyidae (E), Agoutidae (F), Gliridae (G), Hystricidae (H),Erethizontidae (I), Dipodidae (J), Heteromyidae (K), Aplodontidae (L), Myocastoridae (M), Octodontidae (O), Pedetidae(P), Hydrochaeridae (Q), Rhizomyidae (R), Spalacidae (S), Thryonomyidae (T), Capromyidae (U), Caviidae (V), Chinchilli-dae (Y), and Zapodidae (Z). Familial common names are given in the Appendix.


3226 HAYSSEN

Figure 8. Box plots of the proportion of the period from conception to weaning, which is lactation for 653mammalian species grouped by order and ranked by median gestation length (provided beneath name of order). Samplesize in number of species is listed in the right column. For most eutherians, lactation occupies about half of the periodbetween conception and weaning. For most prototherians and metatherians, lactation is usually over 75% of the parentalinvestment period. Median (+), interquartile range ( • ), approximate 95% confidence interval for median ( ), outliers, thevalues distant from the median by >1.5 times the interquartile range (*), and “whiskers”, the range excluding outliers (—).Names of common representatives of each order are given in Materials and Methods.



range over nearly three orders of magnitude,the percentage of time that a female spends inlactation relative to the total period ofreproductive investment is surprisingly con-stant across mammals. Generally, lactation oc-cupies about half of the period between con-ception and weaning (X = 49%; SD = 17%;median 48%, 653 species). Approximately50% of these species have lactation lengthsbetween 40 and 60% of the time betweenconception and weaning (lower quartile,39.3%; upper quartile, 57.3%). Thus, mostmammals spend approximately the sameamount of time in gestation as in lactation.One practical consequence is that the gestationlength of a mammal is a rough estimate of itslactation length.

Mammals with extremely long or short lac-tation lengths do not have consistent patternsof temporal investment. Marsupials have longlactation lengths, and for them lactation is alarge percentage (75 to 95%) of the periodfrom conception to weaning. However, batsand primates also have long lactation lengths,but lactation for these groups is not an extraor-dinary proportion of the time between concep-tion and weaning. Lactation lengths of bats,although relatively long for their mass, arerelatively short, <40% of their total parentalinvestment with none >65%. However, lacta-tion lengths of most primates are about 40 to60% of parental investment, and for about 10species lactation is >65% of the period be-tween conception and weaning.

Mammals with short lactation lengths aresimilarly diverse. Earless seals have short lac-tation lengths (4 to 40 d), and lactation is also asmall temporal component of their reproduc-tive investment (gestation lasts about 1 yr).Similarly, baleen whales have short lactationlengths for their mass (.5 to 1 yr), but thecontribution of lactation to the period betweenconception and weaning in these mammals isnot relatively small; gestation lasts .8 to 1.2 yror about the same length as that of earlessseals. Although the data are limited, milk com-position also appears to be similar between thetwo groups. Milks from both groups are highin fat and protein. In contrast, the age at firstsolid food is probably very different. Earlessseals first ingest solid food after lactation isover, whereas baleen whale calves probablyfirst ingest krill well before weaning. Thus, the

temporal and energetic demands of lactationare different for the two mammalian groupswith the shortest relative lactation lengths.

Although a trade-off between gestation andlactation might be expected, such that longperiods of lactation would compensate forshort periods of gestation and vice versa, thistrend is valid only for marsupials and earlessseals. For most mammals, after the effects offemale mass are removed, the lengths of gesta-tion and lactation are positively, not inversely,correlated, although the effect is not strong(Table 3). Thus, mammals with long gestationlengths tend to have long periods of lactation,and mammals with short in utero developmenttend to wean their offspring early.

Energetic Investment at Birth. The timingof first solid food relative to weaning mayindicate the extent to which lactation is impor-tant to the nutritional or energetic needs of theoffspring. Thus, the timing of first solid foodrelative to weaning may be of greater biologi-cal interest in the study of lactation than theabsolute age at either weaning or first solidfood. When the age at first solid food occurslong before weaning, lactation may have sig-nificant functions beyond nutrition or energy.

The relationship of relative nutritional de-pendency (age at first solid food divided byage at weaning) to the relative mass of off-spring at birth (litter mass divided by femalemass) can be examined either with respect tolitter size (Figure 9a) or neonatal development(Figure 9b). A clear pattern emerges, eventhough neonatal development and litter size areconfounded. (Monotocous eutherians usuallyhave precocial young, whereas polytocous spe-cies often have altricial or semi-altricial neo-nates.)

TABLE 3. Statistics for the multiple regression of bodymass (grams) and the length of gestation (days) on thelength of lactation (days).l

Variable Coefficient SE

Constant .911 .057Log female mass .122 .012Log gestation .295 .040

1For 634 species, after the effects of female mass areremoved gestation and lactation lengths are positivelycorrelated (R2 = .47; F = 285). All variables are log10-transformed.


3228 H a y s s e n

Figure 9. Relative nutritional dependency (age at first solid food divided by age at weaning) versus relative energeticinvestment at birth (litter mass at birth divided by maternal mass) broken down by a) litter size (litter sizes of 8 or moreare coded as 8; 280 species) for all mammals or b) developmental state at birth for eutherians (70 species). Key todevelopmental state: altricial (1), semi-altricial (2). semi-precocial (3). and precocial (4). Single, precocial young are asmaller proportion of maternal mass at birth and rely less on lactation for nutritional support than the altricial young ofpolytocous species.



Monotocous, precocial eutherians have rela-tively small offspring at birth (small percentageof adult mass), and these offspring eat solidfood early in lactation. The young are furred,mobile, and complete in sensory function (eyesand ears open) soon after birth, but they remainwith their mothers for long periods. For thesespecies, concurrent gestation and lactation sel-dom occurs. Overall, the relationship amongthese reproductive characteristics—early firstsolid food, precocial singleton young, andsmall relative mass at birth—is distinctive andmay be a highly derived condition.

Conversely, polytocous species with altri-cial or semi-altricial young are far more varia-ble. Litter mass can be a small or large propor-tion of female mass, and age at first solid foodcan occur early or late in lactation but usuallyoccurs closer to weaning. In addition, gestationand lactation are commonly concurrent. Theproduction of multiple offspring with a largelitter mass at birth, altricial development, andlate first solid food may be the ancestral eu-therian condition.

The consequences of each constellation ofreproductive characteristics are varied. The lit-ter mass at weaning of muroid rodents (altriciallitters with late first solid food) is often largerthan that of the mother, whereas a calf or ahuman infant at weaning (single precocial off-spring with early first solid food) is only asmall fraction of maternal mass. In addition,the mouse and vole weanlings have relied al-most entirely on their mothers for nutrients andenergy during lactation, whereas the humaninfant and the calf supplement their diets withsolid food for a large portion of lactation. Notonly are the demands of lactation enormouswhen multiple young are dependent for alltheir nutritional and energetic needs on milk,as with mice and voles, but, with concurrentgestation, the female may also be meeting thenutritional needs of a second, in utero, litter.Such a mother’s energetic budget and potentialfor future reproductive success might betightly constrained.

For species with single precocial offspring,lactation may play a far different role in theallocation of temporal and energetic resourcesthan it does in polytocous species with altricialyoung. Milk production may not be the pri-mary function of lactation; instead, selectivepressures may operate on other aspects of off-

spring growth and development via main-tenance of the parent and offspring bond. Sub-sequent benefits include predator defense,educational opportunity, and social facilitation.Predators may be less apt to attack a smalljuvenile if a large adult is nearby, because thatadult may directly intervene to protect thejuvenile. Extended parental care can also pro-vide an opportunity for young to learn fromtheir parents about the location and quality offood, nest sites, or potential threats. Finally,few mammals with single precocial young aresolitary. Most are born into a larger socialarena of colony, troop, tribe, or herd. Thesurvival and reproductive success of thesemammals depends on their integration intoappropriate social roles. Long lactations thatcontinue after nutritional or energetic needs aremet facilitate this social integration.

Differences in the energetic demands,evolutionary constraints, and functional role oflactation between altricial, polytocous speciesthat have late first solid food and precocial,monotocous species that have early first solidfood suggest additional differences in the con-trol of lactation, its resistance to environmentalchange, and the composition and output ofmilk over time. For example, the frequency ofsuckling and the duration of lactation may beunder maternal control in polytocous specieswith late first solid food, whereas offspring aremore apt to control nursing and weaning whensolid food is ingested early in lactation. Thecost to a mother of continuing to nurse a singleoffspring a fraction of her size that alreadyobtains a proportion of its nutritional require-ments on its own is much less than that as-sociated with nursing multiple young that col-lectively weigh as much or more than shedoes. Thus, the cost to benefit ratios associatedwith weaning are different.

Although lactation often extends or delaysgestation when the two are simultaneous, theduration of lactation is also determined bygestation, because the birth of a subsequentlitter forces the weaning of the first. Not onlyare successive litters in conflict regarding thetiming of weaning, but littermates competethroughout lactation for current maternalresources. Maternal control over milk compo-sition and production over the course of lacta-tion allows females to fine tune their energeticinvestment as environmental conditions vary.


3230 HAYSSEN

The composition of milk over the course oflactation may be more closely tied to the needsof the offspring when milk is the sole sourceof energy and nutrients than when milk supple-ments metabolic intake. In sum, maternal con-trol over the timing of weaning and the alloca-tion of metabolic resources may reduceintralitter and interlitter competition and, thus,allow a more selectively advantageous distri-bution across current and future reproductiveinvestment.

Lactation may be under tighter genetic con-trol (lower heritability) in species for whichage at first solid food occurs close to weaningrather than close to birth because heritabilitiesare small for traits of critical importance toreproductive fitness, and appropriate allocationof resources during reproduction is more criti-cal to survival and to reproductive success forpolytocous species with ingestion of solid foodlate in lactation. Thus, lactation may be lessresponsive to current offspring demands inspecies for which age at first solid food is nearweaning. Responses to changing environmen-tal conditions may be more important to fitnessin these mammals. Lactation may be most tiedto the survival of current offspring in monoto-cous species with precocial young through thephysiological regulation of gestation (includingthe timing of birth) and the flexibility of itsgenetic program across species.

To test these hypotheses, data on the nutri-tional and energetic value of milk before andafter first solid food and data on sucklingfrequency, duration, initiation, and biochemicalconsequences over lactation and in a variety ofenvironmental conditions are needed for anarray of species. The captive breeding popula-tions of diverse mammals in various zoos,especially those near research institutions thatcan analyze milk composition, provide an op-portunity not only to collect these data but alsoto facilitate interdisciplinary exchange.

Other Influences on the Length of Lactation

Habitat and Diet. Preliminary analyses ofthe relationships of habitat and diet to the

Thus, for some species, the nutritional value

length of lactation (4) suggest that, when theyof milk may be of least equal importance to

are broadly defined, habitat and diet do notthe thermoregulatory, social, or protective

influence the length of lactation across mam-functions of lactation. Milk quality and quan-tity may reflect these additional functions of

mals (Table 4). However, many taxonomic ord- lactation and not be closely tied to offspringers of mammals have distinctive habitats (e.g., growth and development.

TABLE 4. Analysis of the variance in lactation length (380species) with habitat and diet when broadly defined habitatand diet have little influence on the length of lactation.1

Cumulative

Independent variable F df R2

Adult mass 595.93 1 .45Infraclass 96.91 2 .59Order (infraclass) 13.58 13 .72Habitat 1.85 5 .73Diet 2.19 2 .73

1For further details, see the study by Hayssen (4).

volant and marine) as well as specific diets(e.g., herbivores and carnivores); thus, habitatand diet analyses are confounded withphylogeny. The effects of habitat or diet onlactation may be more apparent within ordersor families rather than across higherphylogenetic categories. Such detailed analysisis outside the realm of this review. Overall, formammals, body mass and phylogenetic ances-try have greater influence on lactation lengththan do current habitat and diet.

Milk Composition. When the primary roleof lactation is social or protective, milkproduction and composition may be less cor-related with the long-term nutritional needs ofthe offspring than with support of immediateenergetic demands. Thus, the low fat and pro-tein and the high sugar content of primate andungulate milks may reflect the alternativefunctions of lactation in these mammals.

Milk may have a thermoregulatory functionfor some mammals. Certainly, the high fatcomposition of cetacean milk serves a ther-moregulatory function in the offspring. Moredirect thermoregulatory advantages may alsoaccrue. Many altricial offspring are unable tomaintain high body temperatures, and milkwith high specific heat coming from endother-mic mothers may serve to warm the young,especially newborns for which ingested milkcan be a relatively large proportion of neonatalmass.



Basal Metabolism. The relationship be-tween basal metabolic rate (BMR) and lacta-tion is complicated (Table 5). Data for age atweaning, female mass, and BMR exist for only180 species, including the three monotremes.After the effects of mass are removed, BMR isnegatively correlated with lactation for thesespecies; i.e., a higher resting metabolism isrelated to an earlier age at weaning. However,separate analyses of species with long (44 spe-cies) and average (131 species) lactationlengths indicate no significant relationships ofBMR with lactation in either group after theeffects of mass are removed. Separate analysesof metatherians (29 species) and eutherians(148 species) suggest that BMR has no rela-tionship to lactation length for marsupials butis inversely related to the age of weaning inplacental mammals, again, after the effects ofmass are removed.

Basal metabolic rate is defined for non-reproductive, nongrowing, feed-deprived, inac-tive animals. During lactation, neither mothersnor offspring fit these criteria. Unfortunately,measures of metabolism in mother and off-spring pairs during lactation are lacking. Untilsuch estimates are available for an array ofmammals, the relationship of metabolism tolactation will remain as opaque as milk.

CONCLUSIONS

Lactation defines mammals. As such, un-derstanding lactation is crucial to understand-ing the biology of mammals. However, stan-

dard biological texts give scant attention to thiscentral mammalian characteristic, and moststudies of the phenomenon itself are on thecellular and molecular aspects. In this com-parative investigation, temporal and energeticaspects of lactation have been analyzed usingspecies, not cells or molecules, as units.

For mammals, allometric constraints are themost important determinants of lactationlength. Within smaller phylogenetic subsets ofmammals (orders and families), allometric con-straints are reduced, or they disappear. Apartfrom female mass, some reproductive charac-teristics (litter size, neonatal development, andrelative nutritional dependency) may operate inconcert to influence lactation. For instance,species with single, precocial young have earlyfirst solid food relative to weaning regardlessof body mass. The age at first solid food mayalter the rate or timing of energetic investmentduring lactation, as do delayed implantationand dispause during gestation.

The coordination and integration ofreproductive characteristics have a profoundimpact on the evolution of unitary phenomena,for example, the lipid content of milk. Theamount of fat in milk may be influenced notsolely by nutritional needs, but also by theselective pressures on a more varied, yet cohe-sive, array of reproductive characteristics.

Pleiotropy and polygeny, as well as thesocial nature of reproduction in general andlactation in particular, weave the molecularand cellular aspects of lactation into the largerfabric of a deme’s reproductive biology. Simi-

TABLE 5. Relationship of lactation length to adult female mass (grams) and basal metabolic rate (BMR) (whole body,milliliters of oxygen per hour) for various mammalian groups.1

SlopeMASS SE SlopeBMR SE Constant SE R2 n F

Mammalia .518 .075 -.471 .101 1.528 .077 .40 180 61Long lactations2 .206* .216 -.006* .276 1.570 .113 .54 44 26Average lactations3 .210 .062 -.020* .086 1.111 .070 .58 131 90

P = .00lMarsupialia .454* .310 -.354* .399 1.744 .147 .56 29 19Eutheria .332 .075 -.227 .102 1.339 .079 .40 148 49

P = .0271All variables are log10-transformed. Sample size is number of species. Coefficients followed by an asterisk are not

significantly different from zero. Unless separately indicated, all other coefficients are statistically significant (P < .0005).2Includes marsupials (Marsupialia). primates (Primates), and bats (Chiroptera) only.3Eutherians, excluding primates (Primates), bats (Chiroptera), earless seals (Phocidae), and baleen whales (Mysticeti).


3232 HAYSSEN

lar patterns (reproductive strategies) mayemerge, although the particular threads(specific mechanisms) within the design willvary.

Ancestral features of mammalian lactation(and mammalian reproductive biology ingeneral) are the basis for all other futuremodifications. Only primitive aspects will bethe same across a majority of mammals. De-rived (selectively advantageous) aspects of lac-tation evolve independently in various line-ages. To ascertain the general applicability of aparticular molecular or physiological process,its ancestry must be determined. If the processis primitive, it may be widespread across mam-mals; if it is derived, its utility is apt to bespecific to that lineage. Parallel modificationsmay occur independently in other lineages sub-jected to similar selective pressures. However,the advanced features of lactation in a particu-lar group are germane only for that group, andnot of general applicability.

The hormonal and metabolic controls oflactation in species with widely disparatesuites of reproductive characteristics (for ex-ample, mice and cattle) have evolved undervery different selective regimens. On themolecular and cellular levels, only primitivefeatures are expected to be similar amongmammals with differing reproductive patterns.Physiological or biochemical processes identi-fied as having adaptive value in mice may nothave similar utility for cows or humans. Broadphysiological similarities in the hormonal con-trol and metabolism of lactation most likelyrepresent primitive features of milk productionrather than recent adaptive modifications.

Lactation is the quintessence of mammals.Mammals are defined by this multifacetedcomponent of their reproduction, which hasinfluenced their biochemistry, physiology,anatomy, behavior, sociality, ecology, and evo-lution. Understanding how these componentsare integrated and how they have changed overtime offers an unparalleled opportunity forsynthesis across biological disciplines that areusually disparate. Bridging disciplinary barri-ers, as this symposium does, will eventuallycontribute to a multidimensional understandingof lactation as cohesive and intricate as thephenomenon itself.

ACKNOWLEDGMENTS

This article is dedicated to the memory ofmy brother, Sandy, whose multifaceted ap-proach to both life and death was as compli-cated and important to me as the evolution oflactation is to mammals.

D. Blackbum, B. E. Horner, and S. Tilleyprovided informed and helpful reviews of themanuscript.

REFERENCES

1 Case, T. J. 1978. On the evolution and adaptivesignificance of post-natal growth rates in the terrestrialvertebrates. Q. Rev. Biol. 53:243.

2 Cole, L. C. 1954. The population consequences of lifehistory phenomena. Q. Rev. Biol. 29:103.

3 Green, B. 1984. Composition of milk and energeticsof growth in marsupials. Symp. Zool. Soc. Lond. 51:369.

4 Hayssen, V. 1985. A comparison of the reproductivebiology of metatherian (marsupial) and eutherian(placental) mammals with special emphasis on sexdifferences in the behavior of the opossum, Didelphisvirginiana. Ph.D. Diss., Cornell Univ., Ithaca, NY.

5 Hayssen, V., and R. C. Lacy. 1985. Basal metabolicrates in mammals: taxonomic differences in the allom-etry of BMR and body mass. Comp. Biochem. Phys-iol. A Comp. Physiol. 81:741.

6 Hayssen, V., R. C. Lacy, and P. J. Parker. 1985.Metatherian reproduction: transitional or transcending.Am. Nat. 126:617.

7 Hayssen, V., A. van Tienhoven, and A. van Tien-hoven. 1993. Asdell’s Patterns of MammalianReproduction. 3rd ed. Cornell Univ. Press, Ithaca.NY.

8 Hoaglin, D. C., F. Mosteller, and J. W. Tukey. 1983.Understanding Robust and Exploratory Data Analysis.John Wiley & Sons, New York, NY.

9 Snedecor, G. W., and W. G. Cochran. 1980. StatisticalMethods. 7th ed. Iowa State Univ. Press, Ames.

10 Tyndale-Biscoe, H., and M. Renfree. 1987. Reproduc-tive Physiology of Marsupials. Cambridge Univ.Press, Cambridge, Engl.

11 Wilson, D. S. 1980. The Natural Selection of Popula-tions and Communities. Benjamin/Cummings Publ.,Menlo Park, CA.

APPENDIX

Familial common names (or the most popu-larly known representative) are listed for orderspresented in Figures 4 through 7.

Edentata or Xenarthra: Bradypodidae (three-toed sloths), Choloepidae (two-toed sloths),Dasypodidae (armadillos), and Myrmecophagi-dae (anteaters).

Insectivora: Chrysochloridae (goldenmoles), Erinaceidae (hedgehogs), Solenodonti-



dae (solenodons), Soricidae (shrews) Talpidae(moles), and Tenrecidae (tenrecs).

Chiroptera: Emballonuridae (sheath-tailedbats), Megadermatidae (false vampire bats),Molossidae (free-tailed bats), Noctilionidae(bulldog bats), Nycteridae (slit-faced bats),Phyllostomidae (leaf-nosed bats), Pteropodidae(flying foxes), Rhinolophidae (horseshoe bats),Rhinopomatidae (mouse-tailed bats), andVespertilionidae (common insectivorous bats).

Primates: Callimiconidae (Goeldi’s mar-moset), Callitrichidae (tamarins), Cebidae(New-world monkeys), Cercopithecidae (Old-world monkeys), Cheirogaleidae (mouselemurs), Daubentoniidae (aye-aye), Galigonidae(galagos), Hominidae (humans), Indridae (sifa-kas), Lemuridae (lemurs), Lorisidae (lorises),and Pongidae (great apes).

Carnivora: Canidae (dogs) Felidae (cats),Herpestidae (mongooses), Hyaenidae (hyenas),Mustelidae (weasels), Odobenidae (walruses),Otariidae (fur seals), Phocidae (earless seals),Procyonidae (raccoons), Protelidae (aard-wolves), Ursidae (bears), and Viverridae (ge-nets).

Cetacea: Balaenopteridae (rorquals), Del-phinidae (marine dolphins), Eschrichtidae (greywhales), Monodontidae (white whales), Pho-

coenidae (porpoises), Physeteridae (spermwhales), Platanistidae (river dolphins), andZiphiidae (beaked whales).

Perissodactyla: Equidae (zebra), Rhinocerot-idae (rhinoceroses), and Tapiridae (tapirs).

Artiodactyla: Bovidae (antelope), Camelidae(camels), Cervidae (deer), Giraffidae (giraffes),Hippopotamidae (hippopotamuses), Suidae(pigs), Tayassuidae (peccaries), and Tragulidae(chevrotains).

Lagomorpha: Leporidae (hares) andOchotonidae (pikas).

Rodentia: Agoutidae (pacas), Aplodontidae(mountain beavers), Arvicolidae (voles),Bathyergidae (African mole rats), Capromyidae(hutias), Castoridae (beavers), Caviidae (guineapigs), Chinchillidae (chinchillas), Cricetidae(gerbils), Dasyproctidae (agoutis), Dipodidae(jerboas), Echmyidae (spiny rats), Erethizonti-dae (New-world porcupines), Gliridae (dor-mice), Heteromyidae (kangaroo rats), Hyd-rochaeridae (capybaras), Hystricidae (Old-world porcupines), Muridae (rats), Myocastori-dae (nutrias), Octodontidae (degus), Pedetidae(spring hares), Rhizomyidae (bamboo rats),Sciuridae (squirrels), Spalacidae (blind molerats), Thryonomyidae (cane rats), and Zapodi-dae (jumping mice).


empirical and theoretical constraints on the evolution of

Documents