Evolutionary rates and stabilizing selection in large-bodied opossum skulls (Didelphimorphia: Didelphidae)

Evolutionary rates and stabilizing selection in large-bodiedopossum skulls (Didelphimorphia: Didelphidae)

Bernardo Lemos1,2, Gabriel Marroig1,2* and Rui Cerqueira1

1 LaboratoÂrio de Vertebrados, Departamento de Ecologia, Universidade Federal do Rio de Janeiro, CP 68020, CEP 21941-590,Rio de Janeiro, RJ, Brazil

2 Departamento de GeneÂtica, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ, Brazil

(Accepted 14 November 2000)

Abstract

Phenotypic evolutionary rates are reported for cranial characters of eight species of large-bodied

Neotropical marsupials: Didelphis aurita, D. marsupialis, D. albiventris, Philander opossum, P. frenata,

Lutreolina crassicaudata, Chironectes minimus, and Metachirus nudicaudatus. These rates were found to be

lower than expected if cranial diversi®cation in these opossums had occurred by mutation and genetic drift

only, and it is clear that a greater diversi®cation was prevented. As all parameters used in estimating rates

were very conservative, the conclusion that stabilizing selection has predominated during the evolution of

the skull of large-bodied opossums is fairly robust. We also show that directional selection sustained for

150 generations (158 years) or less is capable of producing differences of the same magnitudes as those

found between various pairs of species. Therefore, we conclude that even where a particular differentiation

has been caused by directional selection, neutral rate tests are unlikely to infer it. This is because following

a morphological shift, stabilizing selection will progressively erase evidences of directional selection.

Key words: stabilizing selection, phenotypic evolution, neutral rates, adaptive landscapes, Marsupialia

INTRODUCTION

One of the main goals of evolutionary biology is theevaluation of the relative role played by evolutionaryforces in producing current patterns of morphologicaldiversity. Evolutionary change ultimately results fromcommon microevolutionary forces, such as genetic driftand natural selection, and several measures of the ratesof such change at the morphological level have beenproposed and used for inferring past evolutionary pro-cesses (Haldane, 1949; Lerman, 1965; Lande, 1977;Cherry et al., 1982; Gingerich, 1983; Charlesworth,1984; Turelli, Gillespie & Lande, 1988; Lynch, 1990;Martins, 1994). Neutral rate models developed fromquantitative genetic theory extend the neutralist hypoth-eses of molecular evolution (Kimura, 1968) to thephenotypic level, thus relating phenotypic evolutionaryrates to population models based on mutation andgenetic drift (Turelli et al., 1988; Lynch, 1990). Thesemodels test, in polygenic characters, the null hypothesisof neutral evolution by comparing an empirically calcu-lated evolutionary rate with a theoretical one predicted

exclusively by mutation and genetic drift (Turelli et al.,1988; Lynch, 1990).

Neutral rates were primarily devised as statisticaltools for distinguishing between adaptive vs non-adaptive evolutionary hypotheses (Lande, 1977; Turelliet al., 1988). However, these rates have been unable todetect directional selection operating to differentiateextant species (Cheetham, Jackson & Hayer, 1993;Savalli, 1993; Spicer, 1993). Notwithstanding this, theyhave been successfully used to infer the operation ofstabilizing selection and, consequently, the punctuatedphenotypic evolution of all tested organisms (Lande,1976, 1977; Turelli et al., 1988; Lynch, 1990; Spicer,1993).

Here evolutionary rates for phenotypic cranialcharacters of eight species of large-bodied Neotropicalmarsupials are reported: Didelphis aurita and D. marsu-pialis (common black-eared opossums), D. albiventris(white-eared opossum), Philander opossum andP. frenata (common grey four-eyed opossums), Lutreo-lina crassicaudata (thick-tailed opossum), Chironectesminimus (water opossum) and Metachirus nudicaudatus(brown four-eyed opossum). These species are in-cluded in a monophyletic clade with relatively well-established phylogenetic relationships (Kirsch et al.,1993; Kirsch & Palma, 1995; Patton, Reis & Silva, 1996)

*All correspondence to: Gabriel Marroig, LaboratoÂrio deVertebrados, Departamento de Ecologia, Universidade Federal doRio de Janeiro, CP 68020, CEP 21941-590, Rio de Janeiro, RJ, Brazil.E-mail: [email protected]

J. Zool., Lond. (2001) 255, 181±189 # 2001 The Zoological Society of London Printed in the United Kingdom

and are ecologically and morphologically differentiated(AstuÂa de Moraes et al., 2000; Cerqueira & Lemos,2000). We show here that despite their differencesthese taxa are much more similar to one anotherthan would be expected if they had diverged by muta-tion and random drift only. While our analyses haveidenti®ed a pattern that is consistent with stabilizingselection and occasional bursts of diversi®cation, wecaution that our approach is inferential and based onextant material.

The ability of neutral models of phenotypic evolutionto distinguish between microevolutionary forces thathave driven differentiation events are discussed in thelight of Wright's adaptive landscape (Wright, 1932,1982) and its implications for the tempo of phenotypicdifferentiation. We stress that failure of tests of rate todisentangle the process responsible for phenotypicdifferentiation results from their inability to detect theunderlying microevolutionary forces when applied atthe macroevolutionary scale as in comparisons amongspecies.

MATERIALS AND METHODS

Data

Eighteen skull and lower jaw measurements (Fig. 1) of 8species of large-bodied opossums were taken using adigital calliper. Measurements, formally de®ned byGardner (1973) and Cerqueira (1980), were taken onlyfrom wild-caught adult animals belonging to age classes6 and 7 (Tyndale-Biscoe & Mackenzie, 1976). Samplesizes for each species along with a list of measuredspecimens and their museum identi®cations are given inthe Appendix. Separate analyses were carried out oneach sex as ANOVAs showed sexual dimorphism in alltaxa herewith studied. Ln-transformed variables wereused in all analyses. Normality and homogeneity ofvariances were tested through Kolmogorov±Smirnov(K±S) and Levene tests, respectively.

General size and shape factors were estimated respec-tively by the ®rst and second eigenvectors of the pooledcovariance matrix extracted in a principal componentanalysis (PCA). Individual scores of ®rst and secondprincipal components and the 18 measured characterswere used as separate variables in the evolutionary rateestimations. To avoid in¯ated within-species variancesdue to geographic variation, we tried to maximize thesample size of each species, taking into account therestriction that specimens came from a single or fewclose localities. For D. marsupialis and P. opossum, asingle locality series was used, thus eliminating geo-graphic variation while, for other species, a pooledlocality sample was used. In this case, an ANOVA waspreviously done for evaluating sample heterogeneityrelated to geographic variation. Discriminant analysiswas performed and the Mahalanobis distance (D2)calculated to evaluate the degree of differentiationamong large-bodied opossums.

Rate models and their parameters

The MDE model developed by Turelli et al. (1988) wasused following the modi®cation of BjoÈrklund (1991).This model tests the neutral hypotheses by F statistics:F = S2/ts2

m, (with n71 degrees of freedom in thenumerator and in®nite degrees of freedom in the de-nominator), where t is the total number of generationsseparating lineages, s2

m the rate of input of new geneticvariance due to mutation, and S2 the between-speciesvariance determined from a one-way ANOVA in thefollowing way: S2 = (MSb7MSw)/no, where MSb is thebetween-population mean square, MSw the within-population mean square, and no the average sample size(Lynch, 1988; BjoÈrklund, 1991; Savalli, 1993). Lowerand upper 95% con®dence limits of S2 resulting frommutation and drift were determined as:

s2m t F0.025[n71,?] / s2

p < genetic drift< s2

m t F0.975[n71,?] / s2p

(Turelli et al., 1988; BjoÈrklund, 1991; Savalli, 1993).Values of between-species variance above or below

the 95% con®dence neutral interval indicate directionalselection and stabilizing selection, respectively.

Rate estimates were also calculated following themodel described by Lynch (1990). This model estimatesthe rate statistic: D=s2

b / (s2w t), where t is the total

number of generations separating 2 lineages and s2b

and s2w are the observed between- and within-species

components of phenotypic variance of log-transformedmeasures (Lynch, 1990). This rate is directly comparableto estimates of mutational heritability (s2

m/s2e) (Lynch,

1990).Both neutral models of phenotypic evolution rely on

the following parameters: (1) polygenic mutational var-iance; (2) number of generations per year occurring ineach taxon; (3) time of divergence among taxa (Turelliet al., 1988; Lynch, 1990).

As extensions of the neutral hypotheses of molecularevolution, both models of phenotypic evolution makeuse of the neutral mutational variance as a crucialparameter. While these estimates are experimentallydif®cult to obtain for quantitative characters, they areusually replaced by estimates of total mutational var-iance. The rate of input of new genetic variance pergeneration, s2

m, when scaled by the environmentalvariance, s2

e, falls between 1073 and 1072 for mammal-ian skeletal traits (Lynch, 1988). This mutationalheritability, s2

m /s2e is then used in neutral rate models.

Following Lynch (1990), 1074 was considered as aminimum conservative estimate of mutational varianceand 1072 as the maximum expected value of divergenceunder mutation and genetic drift. It must be noted thatthe total mutation heritability (s2

m /s2e) was used

instead of the smaller neutral mutation heritability thatis supposed to be used in neutral models.

Generation time was estimated as the averagebetween the ages of ®rst and last female reproduction(Pianka, 1994). Gentile et al. (1995) reported thatfemales of D. aurita and P. frenata can reproduce up to

B. Lemos, G. Marroig and R. Cerqueira182

2 and 3 times, respectively, in the same reproductiveseason that occurs once a year, while studies ofD. albiventris show that females have an average of2 litters per year (Cerqueira, 1984; Regidor &

Gorostiague, 1996). Usually, females born in the begin-ning of a reproductive season will give birth 6 monthslater, but still during the same season. Rarely, femaleswill be reproducing 1 year after the ®rst reproduction

183Evolutionary rates in large-bodied opossums

Fig. 1. Eighteen skull and lower jaw measurements. GLS, greatest length of skull; CB, condylobasal length; PL, palatal length;

ZB, zygomatic breadth; BBC, breadth of brain case; BAM, breadth across molars; CAN, breadth across canines; SM, length of

upper molar series; SMI, length of lower molar series; MAX, length of maxillary toothrow; MAD, length of mandible; NAS,

length of nasal; ROS, breadth of rostrum across jugals; RAF, breadth of rostrum across frontals; GBN, greatest breadth of

nasals; IC, interorbital constriction; POC, postorbital constriction; BPS, breadth of palatal shelf. Measurements formally

de®ned by Gardner (1973) and Cerqueira (1980).

(Cerqueira, 1984). Therefore a conservative estimate ofthe generation time in these taxa would be 9 months, orthe age of ®rst reproduction (6 months) plus the age ofthe last reproduction (12 months) divided by 2. Similarreports on the reproductive behaviour of M. nudicau-datus, L. crassicaudata, and C. minimus are presentlyunavailable; a generation time of 9 months is hereassumed as a conservative estimate in these taxa.

The divergence times reported by Patton et al. (1996)were initially used for estimating evolutionary rates.Whenever these evolutionary rates fell below the lowercon®dence limits of neutral evolution, estimates ofminimal divergence time were drawn between large-bodied opossum species in order to be more conserva-tive. These estimates of minimal divergence time wereused for recalculating evolutionary rates. Since for adetermined amount of phenotypic differentiation, diver-gence time and evolutionary rate are inverselycorrelated, an underestimate of divergence time conse-quently resulted in more conservative over-estimates ofevolutionary rates that were, therefore, moved towardsthe null hypothesis of neutral evolution. Minimumdivergence times were estimated by re-analysing eithercytochrome b±mitochondrial DNA sequence data re-ported by Patton et al. (1996) or distance estimatesbased on DNA±DNA hybridization data (Kirsch et al.,1993; Kirsch & Palma, 1995). This was done by using apoint of calibration corresponding to the divergence ofD. virginiana from other extant Didelphis species at2 million years before present. This point was chosen forcalibration because D. virginiana is a North Americanspecies which presumably speciated in North Americaduring Pleistocene climatic ¯uctuations (Gardner,1973), probably following the uplift of the Panamabridge connecting North and South America between2 and 4 million years before present (Marshall &

Sempere, 1993). For estimates of divergence time usingcytochrome b sequences, we used gamma correctedKimura 2-parameter distances among the availablespecies as calculated by PHYLTEST (Kumar, 1996)using the gamma value of 0.76 obtained with the soft-ware GAMMA (obtained from S. Kumar). DNA±DNAhybridization distances directly taken from Kirsch et al.(1993) and Kirsch & Palma (1995) were also used.Cytochrome b was used for estimating minimum diver-gence time of D. albiventris, P. opossum, P. frenata, andM. nudicaudatus, whereas DNA±DNA hybridizationwas used for D. aurita, D. marsupialis, L. crassicaudataand C. minimus.

The null hypothesis in both rate tests used hereassumes that the differentiation of any 2 taxa from acommon ancestor occurred by mutation and drift inboth branches of a dichotomous tree. This is a Brow-nian motion model of divergence, assuming that theancestral phenotype was exactly intermediate betweenthe 2 descendants, and results in a minimum rate ofchange. However, an opposite scenario used by usconceived differentiation as a completely asymmetricprocess, occurring in only 1 descendent lineage (theother, remaining with the same phenotype as that of theancestral one). In this case, only half of the totalnumber of generations separating the descendent taxa isconsidered, resulting in the highest possible estimates ofdivergence rates under a Brownian motion model.

Using the rate model of Lynch (1990), we estimatedthe number of generations required for each possiblemorphological shift that could have occurred in thedifferentiation of the extant large-bodied opossum'scranium. We hypothesised that each shift in the differ-entiation of sister groups resulted from one pulse ofdirectional selection operating only in 1 lineage andproducing a rate of differentiation of 1072, that in turn


Fig. 2. Phylogenetic hypotheses and minimum divergence times used on evolutionary rate calculations. Divergence times in

years. Cytochrome b (Patton et al., 1996) was used for estimating minimum divergence time of P. opossum, P. frenata, and

M. nudicaudatus, whereas DNA±DNA hybridization (Kirsch et al., 1993; Kirsch & Palma, 1995) was used for D. aurita,

D. marsupialis, L. crassicaudata, and C. minimus.

characterizes directional selection under Lynch's (1990)model. This was achieved by lowering estimates ofdivergence time to the point where differentiation ratesreached the highest limit (1072) of neutral rates (Lynch,1990).

RESULTS

Table 1 shows Mahalanobis distance (D2) betweenspecies. All distances were signi®cant at P < 0.0001.These data are in agreement with AstuÂa de Moraes et al.(2000) indicating substantial skull differences amongNeotropical opossum species. Estimated minimum di-vergence times used in the evolutionary rate calculationsare shown in Fig. 2. This dating places the morpho-logical evolution of large-bodied opossums within thePliocene±Pleistocene. Mean, standard deviation andsample size for each character per species (not shown)

can be obtained upon request. The K±S test for nor-mality within species revealed two non-normal samplesout of 288 (eight species; 18 characters; two sexes). TheLevene test of homoscedasticity showed that 13 out of36 tests performed were heterogeneous among samples.This heterogeneity of the within-species variance was aresult of signi®cant size differences among species andintraspeci®c geographic variation of some samples.ANOVA among localities on pooled samples did notreveal any systematic deviation of speci®c characters orlocalities in any species. PCA revealed a general sizefactor on the ®rst component accounting for 88.1% and88.0% of the variance in females and males, respectively.The second principal component, accounting for 5.2%of variation in both sexes, was considered to be asso-ciated with variation of cranial shape.

Table 2 presents the estimated MDE rates and thecon®dence intervals for accepting the neutral hypoth-esis; Table 3 shows average rates calculated using


Table 1. Mahalanobis (D2) distances between large-bodied opossum species centroids. Values for males (below diagonal) andfemales (above diagonal)

D. aurita D. marsupialis D. albiventris P. frenata P. opossum C. minimus L. crassicaudata M. nudicaudatus

D. aurita 3.2 86.2 196.4 106.0 47.8 164.8 229.0D. marsupialis 5.9 51.6 123.8 75.7 45.4 117.3 141.6D. albiventris 73.6 73.5 73.7 37.8 23.8 73.5 135.6P. frenata 189.7 182.6 53.3 18.9 18.0 54.2 50.8P. opossum 88.3 88.3 26.2 26.0 22.0 66.5 63.3C. minimus 60.0 65.3 28.4 25.7 26.6 22.5 33.6L. crassicaudata 242.2 212.3 110.3 72.5 95.0 48.1 118.9M. nudicaudatus 196.4 192.9 98.3 58.9 64.3 35.8 157.0

Table 2. MDE rate values for each character in males (M) and females (F); and con®dence intervals for the neutral hypothesesof evolution by mutation and genetic drift assuming minimum and maximum mutational variances (m) of 0.0001 and 0.01,respectively. Divergence time in generations

Character MDE rate Divergence time Inferior limit (= 0.0001) Superior limit (= 0.01)

F M

BAM 18.16 13.66 50 0.001 1.0BBC 6.90 5.70 100 0.003 2.0BPS 7.04 5.03 250 0.008 5.0CAN 10.30 9.06 500 0.01 10.0CB 13.03 10.24 5000 0.1 100.4GBN 8.23 5.83 250 000 7.7 5023.9GLS 13.10 9.77 500 000 15.5 10 047.9IC 8.92 5.57 1 000 000 31.0 20 095.9MAD 16.13 11.65 2 000 000 61.9 40 191.8MAX 20.34 15.47 3 000 000 92.9 60 287.7NAS 12.89 12.79 4 000 000 123.8 80 383.6PL 11.55 11.75 5 000 000 154.8 100 479.6POC 10.25 9.25 6 000 000 185.8 120 575.5RAF 7.53 6.19ROS 6.51 4.90SM 23.21 13.92SMI 13.07 21.16ZB 8.91 6.55PC1 12.87 10.20PC2 4.48 4.10

Lynch's (1990) model for the above described asymme-trical divergence scenario. Rates derived from this,varied among characters, ranging from two to ®ve levelsof magnitude lower than the one predicted by theneutral model; most of them were around 1078. MDErates were in agreement with those estimated by Lynch'smodel (1990), both indicating that the cranium of large-bodied opossums has undergone a long period of stasis.As all parameters used for calculating evolutionaryrates were very conservative, this result is fairly robust.

Lowering estimates of divergence time to the pointwhere differentiation rates reached the highest limit(1072) of neutral rates (Lynch, 1990), revealed thattransitions between any two morphotypes could beaccounted for by processes of directional selection main-tained by very few generations. Assuming a rate ofdivergence of 1072, the differentiation betweenD. aurita and D. marsupialis could be explained by ®vegenerations of directional selection. An event of 50generations of directional selection at a rate of 1072

could also account for the transition of D. albiventris(D. aurita; D. marsupialis) and one for 10 generationscould explain divergence between P. opossum andP. frenata. Among all tested hypothetical transitions ata rate of 1072, the most prolonged one would be thatoccurring between D. marsupialis and M. nudicaudatusmorphotypes, requiring some 150 generations or113 years assuming a generation length of 9 months.

DISCUSSION

A general conclusion from applying neutral rate tests tolarge-bodied opossums is that all cranial charactersshowed an evolutionary rate that is consistent with theoperation of stabilizing selection. As all parameters usedin estimating rates were very conservative, the conclu-sion that stabilizing selection has been predominantduring the evolution of large-bodied opossum skulls isfairly robust. Although short periods of strong direc-tional selection might have occurred, it is clear that agreater diversi®cation was prevented. This does notmean that the crania of large-bodied opossums have notdiverged from one another (Table 1; AstuÂa de Moraeset al., 2000), but only that their divergence is in the

order of magnitude below that expected by the neutralmodel of evolution. Interestingly, the divergence timeshere estimated for large-bodied opossums are within thePliocene±Pleistocene, which was a very dynamic periodof South American history (for a review see Marroig &Cerqueira, 1997) with plenty of opportunities formorphological differentiation of these taxa. In fact,AstuÂa de Moraes et al. (2000) showed substantial differ-ences in the crania of some Neotropical opossum speciesthat were also correlated to the ecological diversi®cationof opossums in South America. Moreover, AstuÂa deMoraes et al. (2000) emphasized that Neotropicalopossums diverged much more in size than in shape. Inagreement with their conclusion, MDE rate values forsize divergence (PC1) is three times higher than ratevalues for shape divergence (PC2) (Table 2).

Our results are consistent with a punctuated tempo ofmorphological evolution where short periods of rapidevolutionary rates are separated by long periodswithout morphological change (Eldredge & Gould,1972). This is also consistent with the observation thatduring periods of rapid evolutionary differentiation, ataxon goes through a morphological shift, from oneposition of stabilizing selection optimum (i.e. adaptivepeak) to a different one (Simpson, 1953; Kirkpatrick,1982). As a metaphor to the way evolution proceeds, theadaptive landscape concept predicts that morphologicalevolution will be punctuated (see Kirkpatrick, 1982;Coyne, Barton & Turelli, 1997; Wade & Goodnight,1998).

It is important to note that this analysis is consistentwith the punctuated equilibrium model in showing apunctuated evolutionary tempo but does not necessarilyimply a speciation mode to morphological evolution. Infact, the low differentiation of crania between D. auritaand D. marsupialis (Cerqueira & Lemos, 2000) as well asbetween species within Philander (pers. obs.) imply alack of direct relationship between speciation andmorphological evolution. The very few generations ofdirectional selection necessary to explain their univariatecranial differences illustrate the extent of their cranialsimilarity. This indicates that speciation and morpho-logical evolution are not correlated in large-bodiedopossums and that punctuated phyletic evolution is apreferable alternative to a process where speciation is


Table 3. Average divergence rate among 18 characters used in each species comparison. Values for males (below diagonal) andfemales (above diagonal). Rates were calculated according to Lynch (1990) and using one half of the number of generations(asymmetric divergence) separating each species pair

D. aurita D. marsupialis D. albiventris P. frenata P. opossum C. minimus L. crassicaudata M. nudicaudatus

D. aurita 3.16E-08 4.85E-07 8.51E-07 5.13E-07 2.46E-07 6.99E-07 6.05E-07D. marsupialis 2.60E-08 8.53E-07 1.20E-06 8.58E-07 4.01E-07 8.65E-07 9.01E-07D. albiventris 6.92E-07 7.24E-07 1.86E-07 6.83E-08 1.20E-08 1.76E-07 1.52E-07P. frenata 6.94E-07 7.45E-07 4.08E-08 5.86E-08 4.10E-08 1.44E-08 2.64E-08P. opossum 5.54E-07 6.26E-07 4.75E-08 2.60E-08 2.00E-08 7.94E-08 1.05E-07C. minimus 2.47E-07 2.16E-06 1.32E-08 2.50E-06 1.55E-08 2.42E-08 7.25E-08L. crassicaudata 6.68E-07 6.85E-07 1.20E-07 1.64E-08 5.81E-08 2.30E-08 5.31E-08M. nudicaudatus 5.80E-07 7.02E-07 1.29E-07 2.88E-08 9.28E-08 6.45E-08 5.76E-08

formally equated to morphological divergence (seeLarson, 1989).

Neutral rate models were primarily devised for distin-guishing between evolutionary change mediated bydirectional selection from neutral change exclusivelyresulting from mutation and drift. However, thesemodels have been unsuccessful for inferring directionalselection in the differentiation of extant taxa, even inextreme cases where there is strong independent evi-dence that a particular evolutionary change has beencaused by directional or sexual selection, as in widow-birds and bishops (Anderson, 1982; Savalli, 1993).Dif®culty in detecting directional selection might resultfrom Wright's (1932) adaptive landscapes, whichimpose a punctuated tempo to evolutionary change. Ataxon might undergo a substantial shift through severalstandard deviations in its morphological attributeswhen crossing an adaptive valley (Kirkpatrick, 1982),but that will probably be followed by a long periodwithout change (Eldredge & Gould, 1972). Stasis occursbecause populations caught at any peak will be main-tained around that position in the adaptive landscapeby stabilizing selection, which will usually act againstany movement from the local point where averagepopulation ®tness is maximized. On the other hand,valleys in the adaptive landscape correspond to unstablesituations, and populations displaced from their adap-tive peak will rapidly evolve to the nearest point whereaverage ®tness is maximized. Therefore, morphologicalchanges that involve shifts from one adaptive peak toanother will usually be very fast, even by the standardsof population genetics, taking only a few generations(see Kirkpatrick, 1982). This is exempli®ed by the veryfew generations of directional selection required foreach cranial shift that could have occurred within thelarge-bodied opossums. Following such morphologicalshifts, the prevailing stabilizing selection will progres-sively erase evidence of directional selection that mightbe detected by neutral rate tests.

The longer the time that has elapsed since a differ-entiation event, the slower the rates are likely tobecome, because of stasis and evolutionary reversals(Gingerich, 1983). Therefore, once an evolutionary shiftin morphology has taken place, our chances of detectingthe forces accounting for it are inversely related to thetime elapsed since its occurrence. As morphologicaltransformations driven by microevolutionary forcestake place, they can only be understood from a micro-evolutionary perspective where processes are stillperceptible. Only at the microevolutionary level canwe effectively understand the forces accounting formorphological shifts, and the attempt to explain macro-evolutionary patterns in terms of microevolutionaryforces will frequently be meaningless. Hence, thedetection of directional selection with neutral rate testsat the macroevolutionary level, such as interspeci®c (orinter-higher taxa) comparisons, do not offer optimisticprospects. On the other hand, neutral rate tests might bevaluable when applied to recent diversi®cation events,as between populations of the same species.

Approaches that are intended to understand theevolutionary process responsible for a particular mor-phological shift can only be productive at amicroevolutionary level, and neutral evolutionary ratesmight be effective in making inferences on the morpho-logical differentiation processes if applied tomicroevolutionary phenomena. However, it must benoted that even at a microevolutionary level, an infer-ence of directional selection can be dif®cult to attain.For instance, in the cases reported here, directionalselection would only be inferred if the neutral rate testhad been applied at the very beginning of speciesdivergence. Hence, the temporal scale of phenotypictransformation imposes a serious dif®culty to any ratemodel of morphological evolution whose goal is tounderstand the forces that produce morphologicaldifferentiation.

Acknowledgements

Special thanks are given to H. SeuaÂnez for carefulreading and valuable suggestions on early drafts of themanuscript, and to E. Hingst-Zaher and M. Weksler formeasuring some specimens for us. We thank D. AstuÂade Moraes and R. Schama for helping to improve theEnglish, and S. Kumar for providing us the softwareused for gamma parameter estimation. We also thanktwo anonymous reviewers for valuable suggestions andthe following curators for allowing us to examine speci-mens under their care: C. Alho (UNB), R. Barquez(CML), J. Contreras (CJC), I. G. Fox (MNHNP),P. Jenkins (BM), A. Mones (MNHN), L. F. de Oliveira(MN), M. J. Piantanida (MACN), J. S. Silva Junior(MPEG), D. Verzi (MLP), M. de Vivo (MZUSP), andR. Voss (AMNH). This work was supported by grantsfrom Conselho Nacional de Desenvolvimento CientãÂ®coe TecnoloÂgico, FundacËaÄo de Amparo a Pesquisa do Riode Janeiro, Projeto de ConservacËaÄo e UtilizacËaÄo Susten-taÂvel da Diversidade BioloÂgica to Rui Cerqueira, and anAmerican Museum Natural History Collection StudyGrant to B. Lemos.

REFERENCES

Anderson, M. (1982). Female choice selects for extreme tail lengthin a widowbird. Nature (Lond.) 299: 818±820.

AstuÂa de Moraes, D., Hingst-Zaher, E., Marcus, L. F. &Cerqueira, R. (2000). A geometric morphometric analysis ofcranial and mandibular shape variation of didelphid marsu-pials. Hystrix 10: 115±130.

BjoÈrklund, M. (1991). Evolution, phylogeny, sexual dimorphismand mating system in the grackles (Quiscalus spp.: Icterinae).Evolution 45: 608±621.

Cerqueira, R. (1980). A study of neotropical Didelphis (Mammalia,Polyprotodontia, Didelphidae). PhD thesis, University CollegeLondon, London.

Cerqueira, R. (1984). Reproduction de Didelphis albiventris dansle nord-est du BreÂsil (Polyprotodontia, Didelphidae). Mam-malia 48: 95±104.

Cerqueira, R. & Lemos, B. (2000). Morphometric differentiation


between Neotropical black-eared opossums, Didelphis marsu-pialis and D. aurita (Didelphimorphia, Didelphidae). Mammalia64: 319±327.

Charlesworth, B. (1984). Some quantitative methods for studyingevolutionary patterns in single characters. Paleobiology 10:308±318.

Cheetham, A. H., Jackson, J. B. C. & Hayer, L. C. (1993).Quantitative genetics of bryozoan phenotypic evolution.I. Rate tests for random change versus selection in differentia-tion of living species. Evolution 47: 1526±1538.

Cherry, L. M., Case, S. M., Kunkel, J. G., Wyles, J. S. & Wilson,A. C. (1982). Body-shape metrics and organismal evolution.Evolution 36: 914±933.

Coyne, J. A., Barton, N. H. & Turelli, M. (1997). Perspective: acritique of Sewall Wright's shifting balance theory of evolution.Evolution 51: 643±671.

Eldredge, N. & Gould, S. J. (1972). Punctuated equilibria: analternative to phyletic gradualism. In Models in paleobiology:82±105. Schopf, T. J. M. (Ed.). San Francisco: Freeman,Cooper & Co.

Gardner, A. L. (1973). The systematics of the genus Didelphis(Marsupialia: Didelphidae) in North and Middle America.Special Publications of the Museum Texas Tech University.Lubbock: Texas Tech Press.

Gentile, R., D'Andrea, P. S. & Cerqueira, R. (1995). Age structureof two marsupial species in a Brazilian restinga. J. trop. Ecol.11: 679±682.

Gingerich, P. D. (1983). Rates of evolution: effects of time andtemporal scaling. Science 222: 159±161.

Haldane, J. B. S. (1949). Suggestions as to quantitative measure-ment of rates of evolution. Evolution 3: 51±56.

Kimura, M. (1968). Evolutionary rate at the molecular level.Nature (Lond.) 217: 624±626.

Kirkpatrick, M. (1982). Quantum evolution and punctuated equi-libria in continuous genetic characters. Am. Nat. 119: 833±848.

Kirsch, J. A. W., Bleiweiss, R. E., Dickerman, A. W. & Reig,O. A. (1993). DNA/DNA hybridization studies of carnivorousmarsupials. III. Relationships among species of Didelphis.J. Mammal. Evol 1: 75±97.

Kirsch, J. A. W. & Palma, R. E. (1995). DNA/DNA hybridizationstudies of carnivorous marsupials. V. A further estimate ofrelationships among Opossums (Marsupialia: Didelphidae).Mammalia 59: 403±425.

Kumar, S. (1996). PHYLTEST: a program for testing phylogenetichypothesis. Version 2. 0. University Park: Pennsylvania StateUniversity.

Lande, R. (1976). Natural selection and random genetic drift inphenotypic evolution. Evolution 30: 314±334.

Lande, R. (1977). Statistical tests for natural selection on quanti-tative characters. Evolution 31: 442±444.

Larson, A. (1989). The relationship between speciation and mor-

phological evolution. In Speciation and its consequences:579±598. Otte, D. & Endler, J. A. (Eds). New York: Sinauer.

Lerman, A. (1965). On rates of evolution of unit characters andcharacters complexes. Evolution 19: 16±25.

Lynch, M. (1988). The rate of polygenic mutation. Genet. Res. 51:137±148.

Lynch, M. (1990). The rate of phenotypic evolution in mammalsfrom the standpoint of the neutral expectation. Am. Nat. 136:727±741.

Marroig, G. & Cerqueira, R. (1997). Plio-Pleistocene SouthAmerica history and the Amazon Lagoon hypothesis: a piece inthe puzzle of Amazonian diversi®cation. J. comp. Biol. 2:103±119.

Marshall, L. G. & Sempere, T. (1993). Evolution of the Cenozoicland mammal fauna in its geochronologic, stratigraphic, andtectonic context. In Biological relationships between Africa andSouth America: 329±392. Goldblatt, P. (Ed.). New Haven: YaleUniversity Press.

Martins, E. P. (1994). Estimating the rate of phenotypic evolutionfrom comparative data. Am. Nat. 144: 193±209.

Patton, J. L., Reis, S. F. & Silva, M. N. F. (1996). Relationshipsamong didelphid marsupials based on sequence variation in themitochondrial cytochrome b gene. J. Mammal. Evol. 1: 3±29.

Pianka, E. R. (1994). Evolutionary ecology. New York: Harper-Collins.

Regidor, H. A. & Gorostiague, M. (1996). Reproduction in thewhite eared opossum (Didelphis albiventris) under temperateconditions in Argentina. Stud. Neotrop. Fauna Environ. 31:133±136.

Savalli, U. M. (1993). An application of the neutral model to theevolution of tail length in the genus Euplectes (Aves, Ploceidae).Evolution 47: 696±699.

Simpson, G. G. (1953). The major features of evolution. NewYork: Columbia University Press.

Spicer, G. S. (1993). Morphological evolution of the Drosophilavirilis species group as assessed by rate tests for naturalselection on quantitative characters. Evolution 47: 1240±1254.

Turelli, M., Gillespie J. H. & Lande R. (1988). Rate tests forselection on quantitative characters during macroevolution andmicroevolution. Evolution 42: 1085±1089.

Tyndale-Biscoe, C. H. & Mackenzie, R. B. (1976). Reproductionof Didelphis marsupialis and D. albiventris in Colombia.J. Mammal. 57: 249±265.

Wade, M. J. & Goodnight, C. J. (1998). The theories of Fisherand Wright in the context of metapopulations: when naturedoes many small experiments. Evolution 52: 1537±1553.

Wright, S. 1932. The roles of mutation, inbreeding, crossbreeding,and selection in evolution. Proc. VI Int. Genet. Congr. 1:356±366.

Wright, S. 1982. Characters change, speciation, and the highertaxa. Evolution 36: 427±44


Appendix

Museum acronyms: Museu Nacional do Rio de Janeiro,Rio de Janeiro (MN); Museu de Zoologia da USP, SaÄoPaulo (MZUSP); Natural History Museum-BritishMuseum, London (BM); American Museum of NaturalHistory, New York (AMNH); Museu Paraense EmilioGoeldi, BeleÂm (MPEG); ColecËaÄo da Universidade deBrasãÂlia, BrasãÂlia (UNB); Museo Argentino de CienciasNaturales `Bernardino Rivadavia', Buenos Aires(MACN); ColeccioÂn de MamãÂferos Lillo, TucumaÂn(CML); ColeccioÂn Julio Contreras, Corrientes (CJC);Museo Nacional de Historia Natural del Paraguay,Assuncion (MNHNP); Museo Nacional de HistoriaNatural, Montevideo (MNHN); Museo de La Plata, LaPlata (MLP).Collection of the LaboratoÂrio de Verteb-rados, LaboratoÂrio de Vertebrados, Rio de Janeiro(LV).

List of specimens used:Didelphis albiventris (n = 113): MN no. 1481, 6732,

6733, 17203, 17207, 17208, 17209, 17210, 17211, 17212,17217, 17218, 17219, 17221, 17222, 17223, 17230, 17239,17240, 17248, 17252, 17253, 17255, 17257, 17258, 17259,17261, 17267, 17268, 17269, 22857, 22860, 22864, 22865,22866, 22871, 22875, 22880, 22885, 23540, 23541, 23542,23543, 23545, 23546, 23547, 23548, 23549, 23551, 23552,23553, 23554, 23555, 23556, 23557, 23558, 23559, 23559,23560, 23560, 23561, 23561, 23562, 23564, 23565, 23566,23567, 23570, 23572, 23580, 23581, 23582, 23585, 23586,23587, 23588, 23590, 23593, 23595, 23631, 23632, 23660,23669, 23681, 23685, 23823, 23824, 25197, 25458, 25463,25489, 25549, 26025, 26027, 26029, 26041, 27347, 27740,27753, 27799, 30039, 30040, 30041, 30049, 30056, 30156,30379, 30566, 30567; MZUSP no. 16533, 16537, 16538,16539.

Didelphis marsupialis (n = 46): MPEG no. 8081, 8090,8090, 8512, 8512, 8512, 12758, 12814, 12840; MN no.20053, 20057, 20072, 20074, 20086, 20087, 20088, 20089,20090, 20091, 20092, 20093, 20097, 20098, 20103, 20106,20110, 20111, 20124, 20128, 20130, 20133, 20137, 20138,20139, 20140, 20144, 20148, 20153, 20159, 20165, 20166,20169, 20192, 20264, 20267, 20274.

Didelphis aurita (n = 82): BM no. 34.1.17.6, 34.1.17.7;LV no. CRB832, EDH69, EDH77, EDH78, EDH80,FU06, GL21, JDM01, LBCE325, LBCE98, MC185,MC270, MC281, SU176; MN no. 1179, 3793, 3854,3856, 3858, 3860, 3863, 5614, 5615, 5616, 5688, 5739,5740, 5793, 5795, 5796, 5808, 5811, 5812, 5856, 7149,7150, 7151, 7152, 7153, 7154, 7155, 7156, 7157, 7160,7163, 7164, 7165, 7166, 7167, 7169, 7171, 7172, 7175,7177, 7180, 7614, 10482, 10483, 10485, 10486, 10486,10488, 10490, 10491, 10494, 10503, 20950, 23721, 24935,25677, 27827, 28896, 30276, 30423, 33425, 42776, 42795,43753, 43966, 47776.

Philander frenata (n = 105): LV no. EDH228, EDH75,GL42, GL43, LBCE174, LBCE175, LBCE309,LBCE327, LBCE99, ORG06, ORG28, SJ04, SPA02,SPA03, SPA04, SU200, SU45; MN no. 1202, 1207,3986, 3988, 3989, 3996, 5596, 5597, 5604, 5605, 5690,

5698, 5750, 5753, 5778, 5804, 5807, 6219, 6220, 6242,6243, 6420, 6421, 7183, 7185, 7186, 7204, 7211, 7212,7213, 7214, 7215, 7216, 7291, 7320, 8221, 8223, 8224,8226, 8227, 8228, 8445, 10374, 10376, 10377, 10379,10382, 26431, 26783, 26797, 26800, 26805, 26831, 26838,26863, 28817, 28819, 28842, 28899, 28902, 28917, 28918,28952, 29810, 29950, 30136, 30410, 30420, 30422, 30425,30427, 30429, 30430, 30432, 30434, 30435, 30440, 30461,42779, 42781, 42782, 42785, 42786, 42802, 42803, 43751,43960, 43961.

Philander opossum (n = 64): MPEG no. 8076, 8534,12871, 12878, 12879, 12880, 12882, 12883, 12884, 12885,12886, 12887, 12889, 12890, 12891, 12892, 15246; MNno. 20004, 20006, 20007, 20008, 20009, 20010, 20011,20012, 20015, 20016, 20018, 20019, 20021, 20023, 20024,20025, 20026, 20028, 20029, 20030, 20031, 20032, 20033,20035, 20036, 20039, 20046, 20050, 20176, 20179, 20181,20189, 20190, 20193, 20194, 20196, 20200, 20239, 20243,20244, 20245, 20247, 20249, 20250, 20251, 20252, 20253.

Lutreolina crassicaudata (n = 84): AMNH no. 143886,254511, 254512, 254519; BM no. 17.1.25.62, 17.1.25.65,17.1.25.68, 17.1.25.69, 17.5.2.20, 17.5.2.21, 17.5.2.23,17.5.2.24, 20.12.18.35, 20.12.18.36, 24.9.1.82, 25.5.1.13,26.11.13.1, 85.11.26.10, 85.11.26.9, 97.1.1.2, 97.1.1.1,99.11.1.13; CJC no. 1748, 3169, 3189, 3301, 3368, 3375,3397, 3871; CML no. 807, 918, 919, 1353, 2895; MACNno. 13.265, 13.266, 13.285, 13.713, 15.365, 17.268,17.270, 18.735, 19.190, 31.258, 32.197, 49.242, 49.37,52.79; MLP no. 667, 1676; MNHN no. 716, 717, 718,719, 720, 885, 1068, 1121, 1122, 1123, 1141, 1142, 1247,1248, 1249, 1250, 1251, 1290, 1291, 1292, 1293, 1294,1295, 1296, 1297; MNHNP no. 924, 931, 932, 1100; MNno. 24447, 24448, 24450, 24452.

Chironectes minimus (n = 48): AMNH no. 30752,33027, 34121, 37482, 37483, 47190, 62365, 72019, 72020,96758, 96760, 97319, 97659, 123288, 129703, 129704,133198, 141918, 182939, 264571, 264572; BM no.25.5.14.3, 26.1.8.21, 28.10.11.58, 29.6.6.73, 29.6.6.74,84.9.d; CML no. 1542; MPEG no. 1527, 1756; MACNno. 52.69, 13053; MNHNP no. 477, 824; MN no. 1280,1280, 1281, 1281, 1283, 1283, 5897, 5897, 10507, 23752,23752, 25738; UNB no. 399, 439.

Metachirus nudicaudatus (n = 123): LV no. MC258,LBCE312, MC272, MC323, MC327, ORG06, ORG34,ORG39, ORG42; MN no. 2598, 3980, 3981, 3982, 3983,3985, 5223, 5578, 5581, 5590, 5598, 5599, 5600, 5601,5603, 5777, 5786, 5805, 5806, 7316, 7318, 7581, 7583,7584, 7585, 7586, 7587, 10245, 10246, 10247, 10248,10250, 10253, 10255, 10258, 10259, 10260, 10261, 10263,10264, 10265, 10266, 10267, 10268, 10269, 10271, 10272,10274, 10276, 10277, 10279, 10280, 10281, 10283, 10285,10286, 10287, 10395, 10396, 10397, 10398, 10399, 10400,10401, 10403, 10404, 10406, 10407, 10408, 10409, 10410,10411, 10413, 10414, 10415, 10419, 10420, 10421, 10422,10423, 10424, 10426, 10427, 10428, 10429, 10430, 10439,10443, 10445, 10447, 10449, 11714, 24958, 25304, 25305,25991, 26612, 26615, 28811, 28937, 28958, 29951, 31405,31425, 32453, 33428, 33430, 42774, 42792, 42817; UNBno. 467, 469, 472, USP no. 22351.


Documents

Evolutionary rates and stabilizing selection in large-bodied opossum skulls (Didelphimorphia: Didelphidae)