Behavior, neuroanatomy, and social organization in two species of voles (Microtus ochrogaster and Microtus montanus)


Material Information

Behavior, neuroanatomy, and social organization in two species of voles (Microtus ochrogaster and Microtus montanus)
Physical Description:
iv, 145 leaves : ill. ; 28 cm.
Shapiro, Lawrence Elliott, 1945-
Publication Date:


Subjects / Keywords:
Voles -- Behavior   ( lcsh )
Sexual behavior in animals   ( lcsh )
Social behavior in animals   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1987.
Includes bibliographical references (leaves 129-144).
Statement of Responsibility:
by Lawrence Elliott Shapiro.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000951360
notis - AER3602
oclc - 16955553
System ID:

Full Text

(Microtus ochrogaster AND Microtus montanus)







ABSTRACT............................... ....... ........... iii


I INTRODUCTION....................................... 1

Behavioral Profiles and the Study of Mating
Systems.......................................... 3
Comparative Neurobehavioral Studies................ 6
Goals and Rationale................................ 8
Prairie and Montane Voles as Model System.......... 8

II SEMINATURAL ENCLOSURE.............................. 16

Experiment 1. Social Behavior....................... 16
Experiment 2. Vaginal Smears ....................... 28


Experiment 3. Baseline Differences................. 44
Experiment 4. The Effects of Species Cross-Pairing. 52
Experiment 5. The Effects of Cross-Fostering....... 58
Experiment 6. The Effects of Morphine and Naloxone. 68

IV BRAIN/BEHAVIOR RELATIONSHIPS ....................... 81

Experiment 7. Species Differences in Brain Size.... 82
Experiment 8. Cingulate Cortex Differences........ 97
Experiment 9. Maternal Behavior .................... 105


Species Profiles: Prairie and Montane Voles........ 113
Comparative Neurobehavioral Analyses............... 121
Conclusions ........................................ 126

REFERENCES....................................... ....... 129

BIOGRAPHICAL SKETCH...................................... 145

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

(Microtus ochrogaster AND Microtus montanus)


Lawrence E. Shapiro

May 1987

Chairman: Donald A. Dewsbury
Major Department: Psychology

The purpose of this research was to delineate

differences in the social and reproductive behavior of

prairie voles and montane voles and to identify

neuroanatomical correlates of those behavioral differences.

In a seminatural enclosure, prairie vole trios, consisting of

one male and two females, displayed higher levels of contact

proneness and lower levels of male/female aggression than did

montane voles. There were no species differences in levels

of female/female aggression. Vaginal smears, taken daily

during the 10-day observation period, indicated female/female

suppression of estrus in prairie voles but not in montane


Male/female pairs were placed in a test cage for 2 hours

and huddling duration was recorded during the third hour.

Prairie vole pairs spent a mean of approximately 30 of


the 60 minutes in contact with each other, whereas montane

voles spent less than 2 minutes. Montane vole pairs, which

had been cross-fostered to prairie voles, huddled for less

than 1 minute; none of the prairie vole pups crossed to

montane voles were alive at weaning. Injections of naloxone

had no effect on huddling in either species. Morphine had no

effect on huddling in montane voles; 10 mg/kg significantly

decreased huddling in prairie voles.

The mean brain weight for prairie voles was 12% heavier

than that for montane voles. There were no species

differences in body weight. Prairie voles had a larger

cross-sectional area in one region of the cingulate cortex

than did montane voles. Previous research has indicated that

the cingulate cortex is involved in the mediation of social

contact and maternal behavior. In the present research,

montane voles showed lower levels of both social contact and

maternal behavior relative to prairie voles.

Species-characteristic behavioral differences are the

result of fine-tuning by natural and sexual selection. The

study of these differences can provide insight into proximate

and ultimate mechanisms underlying differences in social

organization. The use of species such as prairie and montane

voles for neurobehavioral analyses also affords the

opportunity to elucidate "naturally selected" neural

mechanisms that mediate species differences in social



Ever since Darwin (1871) first discussed mating

systems from an evolutionary perspective, i.e., as

adaptations, field workers and theoreticians alike have

attempted to elucidate the selective pressures and

evolutionary mechanisms that underlie variation in this and

other aspects of a species' social organization. One of the

most powerful methods of studying adaptations is to compare

large numbers of closely related species in an effort to

determine how differences in behavior reflect differences in

ecology. Classical sociobiological studies, therefore,

correlated mating systems with habitat for large numbers of

related species. Crook (1964) studied 90 species of

weaverbirds in an effort to correlate the variety of mating

system types with certain key ecological variables. Similar

studies were done with primates (Clutton-Brock & Harvey,

1977; Crook & Gartlan, 1966), ungulates (Jarman, 1974),

blackbirds (Orians, 1969), and mammals (Eisenberg, 1966).

The results of these classical studies indicated 1) that

certain mating systems occurred predictably with certain

habitats and not others and 2) that the type of mating system

adopted by a population was more closely correlated with

ecological variables than with phylogenetic constraints.

Among the ecological variables identified as important

selection pressures were resource abundance and

distribution, predation, habitat cover, etc. Crook (1964),

for instance, found that species of weaverbirds living in

the forest were insectivorous and experienced high levels

of predation while species living in savannah habitats

tended to eat seeds and experience relatively low levels of

predation. Moreover, forest species tended to be

monogamous, whereas savannah species tended to be

polygamous. Factors such as food distribution and

predation were thus offered as "ultimate explanations" for

the observed differences in the mating systems of forest

versus savannah-living species.

Although many of these early evolutionary explanations

were the result of post hoc correlations, they have, in

fact, stood the test of time (Krebs & Davies, 1984).

Indeed, with the knowledge of only a small number of

environmental variables, many aspects of a species' social

organization can now be accurately predicted (e.g., Emlen &

Oring, 1977). In spite of considerable progress, however,

the study of mating systems has, until recently, remained

at a qualitative level of analysis (Vehrencamp & Bradbury,

1984). Whereas our ability to predict certain

habitat/mating system correlations is impressive,

assumptions as to why many of these correlations occur

remain largely untested. Even more limited is our

understanding of those mating systems, such as polyandry,

for which no ecological correlates have been discerned (see

Vehrencamp & Bradbury, 1984).

The aim of most comparative studies is to reveal

cause-effect relationships between selection pressures and

particular mating systems. A primary difficulty in

establishing these cause-effect relationships is the fact

that environmental pressures do not select mating systems

per se. Selection pressures act on the traits and

propensities of individuals not on groups or populations

(e.g., Dawkins, 1976, 1981; Williams, 1966; Wilson, 1975),

and it is through the more immediate mechanisms operating

on individual organisms that selection mediates adaptive

patterns of spacing, mating, demography, etc. Thus, much of

the difficulty in understanding the evolution of mating

systems can be traced to a simple failure to recognize the

importance of the proximate, behavioral mechanisms that

intervene between selection pressures and mating systems.

Behavioral Profiles and the Study of Mating Systems

A mating system can be characterized as a particular

ensemble of behavioral options (Vehrencamp & Bradbury,

1984). This set of options can vary among populations of

the same species and, even within a population, individuals

can display alternative strategies depending upon

environmental or social circumstances (e.g., Sachser,

1986). Nevertheless, species do display characteristic

forms of social organization that are the result of stable

species differences in behavior (e.g., Dewsbury, in press;

Mason, 1974). When individuals from species that display

different forms of social organization are examined under

controlled conditions in the laboratory, species-typical

behavioral profiles emerge. Specific differences in the

profiles of closely related species are the result of

fine-tuning by natural and sexual selection. These traits

and propensities underlie and determine which assortment of

behavioral options are expressed by a population in a

particular ecological and social context. The comparative

analysis of species-characteristic profiles can thus

provide valuable insight into both how (proximate) and why

(ultimate) mating systems evolve.

"How" Mating Systems Evolve

Social organization is an idealized concept (Mason,

1974). The social or mating system of a population is an

emergent property reflecting the traits and propensities of

the individuals that make up that population

(West-Eberhard, 1979). Different suites of traits at the

individual level translate into different sets of

behavioral options, or mating systems, at the population

level. Thus, in order to understand how (i.e., the

immediate, proximate mechanisms by which) mating systems

evolve, fine-grained behavioral differences must be

correlated with equally fine-grained differences between

mating systems. Indeed, the proximate, behavioral

mechanisms that underlie even the same set of behavioral

options (monogamy, for example) can be quite different in

two species which are adapted to different habitats (e.g.,

Mitani, 1984). Species-typical profiles can reveal such

differences. It is at this fine-grained level of analysis

that differences in behavior must be discerned if species

comparisons are to be genuinely revealing.

"Why" Mating Systems Evolve

In their discussion of the problems and shortcomings

of past studies, Vehrencamp and Bradbury (1984) concluded

that a "third-generation" of studies is required if we are

to understand why mating systems evolve, i.e., their

adaptive significance. The goal of these new studies is to

utilize empirical findings in conjunction with mathematical

modelling techniques, such as game theory, to establish

quantitative (not qualitative) functional relationships

between all possible habitat variables, all possible

behavioral options, and all relevant fitness components

(i.e., such factors as copulation rate, female encounter

rate, juvenile survival, etc.). They admit that this is

difficult, if not impossible, to accomplish in practice.

Nevertheless, the importance of their proposal is that it

underscores the importance of quantifying exactly how

variability in the behavior of individuals translates into

variablity in fitness.

The development of behavioral profiles for closely

related species such as prairie voles and montane voles can

greatly facilitate the practical application of

third-generation techniques which Vehrencamp and Bradbury

(1984) advocate. Behavioral profiles, generated in the

laboratory, can often enable one to accurately determine,

even in the absence of hard to obtain field data, exactly

which ecological factors select for which behavioral

options. These data can then be used to rank and quantify

the various fitness components which are a function of these

ecology/behavior correlations, the critical step in

third-generation studies.

Comparative Neurobehavioral Analysis

In-depth behavioral profiles of closely related species

are not only useful in understanding mating system

diversity. Their use in conjunction with neuroanatomical

differences in closely related species represents a powerful

experimental paradigm for elucidating fundamental

neurophysiological processes and mechanisms (e.g., Ewert,

1984; Prohazka, Novak, & Meyer, 1986). The comparative

neuroanatomical work presented in the final section of this

dissertation is the first in a series of more detailed

analyses which will capitalize on the existence of two

closely related species for which a great deal of behavioral

data have been compiled. Whereas the comparative method has

revealed a wealth of data on behavior, there are as yet few

comparative efforts aimed at the neuroanatomical or

neurochemical level (cf. Pohazka et al., 1986). This is not

surprising since neuroscientists rarely compare naturally

occurring species differences when investigating the neural

bases of social behavior (Ewert, 1984). Traditionally, the

research strategy has been as follows. First, only a single

species or strain is utilized. Next, particular areas of the

brain are lesioned or ablated (or stimulated) in an attempt

to eliminate (or elicit) a particular behavior. Finally,

based on the results of these manipulations, an inference is

made as to the functional role of the particular brain

structure or neural circuitry in question.

What is relevant here is that many of the behaviors

affected by lesions or ablations within a single strain or

species exist in the naturally contrasting behavior of

montane and prairie voles. Therefore, these two species

afford a unique opportunity to verify whether or not areas

of the brain previously implicated in the control of

certain behaviors actually prove to be different when their

intact, naturally selected brains are compared. Most

importantly, if differences are found, one can then

ascertain the precise neuroanatomical/neurochemical nature

of those differences and thus greatly augment the results

of previous lesion/stimulation studies.

Goals and Rationale

The goals of the present research are 1) to delineate

ecologically relevant differences in the social behavior of

prairie and montane voles and 2) to identify

neuroanatomical correlates of those and other ecologically

relevant behavioral differences between the species. The

rationale behind the development and comparative analysis

of behavioral and neuroanatomical differences in closely

related species is twofold. First, the comparative

analysis of in-depth, species-typical behavioral profiles

can be helpful in understanding both how and why mating

systems evolve. Second, the comparative neurobehavioral

analysis of closely related species such as prairie and

montane voles represents a highly useful experimental

paradigm for elucidating fundamental brain/behavior


Prairie and Montane Voles as a Model System

Prairie and montane voles are ideal for comparative

analyses. They breed well in the laboratory and can be

tested under carefully controlled conditions. In addition,

a large body of literature now exists that documents

important differences in their social and reproductive

behavior and, as the next two sections illustrate,

behavioral profiles for these species are well under way.

The following sections briefly survey some of the relevant

data that have accumulated on these species.

Field Studies and Natural History

Voles of the genus Microtus are generally found in

grassland and tundra ecosystems (Getz, 1985). Most species

of this genus are capable of occupying a wide range of

habitats and all experience periodic fluctuations in

population density. In general, prairie voles inhabit the

relatively continuous grassland habitat which spreads

throughout much of the central United States (Getz, 1978,

1985). The vegetation consists of grasses and sedges with a

considerable amount of forbs and woody shrubs in certain

parts of the range (Getz, 1985). Montane voles inhabit the

grass and sedge dominated mountain valleys of the

northwestern United States (Jannett, 1980; Getz, 1985).

Both of these species occur in dense vegetative cover as

well as more sparsely covered areas. Where cover is low,

because of overgrazing, prairie voles live in subterranean

burrows and maintain large and deep runway systems (Thomas

& Birney, 1979). Montane voles also make extensive use of

tunnels and surface runway systems. Both species utilize

grasses for surface and subsurface nest construction and

both species feed on seeds, roots, dead vegetation and even

insects in some situations.

Field data on patterns of association and nesting

indicate that prairie voles are generally monogamous while

montane voles are polygamous. Getz, Carter, and Gavish

(1981) provided evidence from the field that prairie voles

display a monogamous mating system at low to moderate

population densities. Using multiple-capture, live-trapping

procedures they recorded stable male/female associations

which lasted throughout the non-breeding season. These

findings were confirmed in a more recent study by Getz and

Hofman (1986) in which 50% of the breeding pairs that were

monitored via radio transmitters and repeat trapping were

found to jointly occupy a nest and to share a home-range.

Moreover, the same pairs were found to occur in both the

breeding and non-breeding season indicating that male/female

associations were stable and long-term.

Jannett (1980, 1982) has studied the social

organization of montane voles in the fields and valleys of

northwestern Wyoming. He has used a variety of field

techniques including live-trapping and radio tracking of

animals implanted with irradiated tantalum wires. Jannett

found that, unlike prairie voles, male and female montane

voles maintained intrasexually exclusive territories.

Females were observed to chase away other females but not

other males. Adult males maintained territories which

overlapped with the territories of several females but were

seen to drive away other males. Males were also observed

to shift their activity to be in the vicinity of estrous

females; male and female montane voles were not observed to

co-nest at any time of the year (Jannett, 1980).

Laboratory Analyses

The data indicating contrasting mating systems in

natural populations of prairie and montane voles are

supported by a growing body of laboratory evidence. Thomas

and Birney (1979) investigated the mating system of prairie

voles by experimentally manipulating the ratio of males to

females in various groups and following their social and

reproductive behavior over time. In each group the area of

the pen was subdivided so that there were as many

compartments as voles and thus each individual vole had the

opportunity to nest separately; each vole also had several

mating possibilities available to it, depending upon the

group's particular male/female composition. The results of

this extensive study were that voles in 26 of the 27 groups

mated monogamously regardless of composition. Each of the

females which produced one or more litters during the 80

day observation period was observed to pair exclusively

with only one male. In addition, the male with which the

female had paired was observed to participate in all of the

activities associated with the care and rearing of young.

The results of this experiment provide convincing

evidence for a behavioral propensity characteristic of many

monogamous species, i.e., pair formation between males and

females (e.g., Wittenberger & Tilson, 1980). The term

"pairbond" is difficult (if not impossible) to adequately

define (see Dewsbury, in press for review). In the present

studies the term pairbond is used simply to indicate the

preferential association of a male and a female.

Pairbonding in prairie voles has been studied by Getz and

Carter (1980) and Carter et al. (1985) who examined the

effects of familiarity on mating preferences and

postcopulatory patterns of aggression in prairie voles.

Males and females were housed across a barrier from each

other until the female came into estrus. Females were then

given a choice between two tethered males--the familiar

male with whom she had been housed and an unfamiliar one.

Females showed no preference in either association or

mating behavior between the two males. In contrast, males

and females that have mated become aggressive toward

strange adults of both sexes while showing high levels of

contact behavior toward each other (Carter et al., 1985).

In an experiment directly comparing the pairbonding

propensities of prairie and montane voles Shapiro et al.

(1986) allowed a female to copulate with a tethered male

for one ejaculatory series. The female was then removed, a

second male was tethered across from the first and the

female was free to copulate with either the unfamiliar male

or the one with whom she had recently mated. Female

prairie voles spent more time in contact with and copulated

more with the familiar than with the unfamiliar male.

Montane vole females, on the other hand, showed no

preference for either male.

In a second phase of this same study, familiarity

without copulation was examined for its effect on

subsequent mate choice. Females were housed across a

wire-mesh barrier from a male for 2 weeks and then allowed

to choose between the familiar and an unfamiliar male. The

unfamiliar male was then housed across from the female for

2 weeks and the female was again allowed to choose between

the two males. In neither test did housing proximity have

an effect on mate choice. However, female prairie but not

montane voles did preferentially mate with the male from

which they received their first ejaculate 2 weeks earlier

in the first test. These results highlight the relative

importance of mating versus familiarity. In the monogamous

prairie vole, copulation and/or ejaculation can evidently

function as an evolved behavioral mechanism whereby the

identification of and preference for a particular male is

"stamped in" or imprinted. These studies from our

laboratory not only extend and support the findings of

Carter et al. (1985), they also demonstrate the lack of

such an effect in montane voles.

Controlled studies have also provided evidence for

species differences in the mating behavior of males.

Fuentes and Dewsbury (1984) compared the sexual behavior of

prairie and montane voles in multi-female test situations

and found that when a male prairie voles was placed with

four receptive females one of the four received over 75% of

the male's intromissions. Montane vole males placed in the

same test situation distributed their copulatory behavior

more evenly among the females. These differences in male

sexual behavior are consistent with earlier findings that

male prairie but not montane voles displayed the "Coolidge

effect," i.e., the tendency of males that have reached

sexual satiety with a given partner to regain copulation if

the original female is replaced by a different receptive

female (Dewsbury, 1973; Gray & Dewsbury, 1973). These

laboratory findings are exactly what would be predicted

based on the proposed mating systems of these two species.

Perhaps the best documented behavioral difference

between these two species revealed in the laboratory is

parental care. Results are consistent in demonstrating

that, even within the confines of the laboratory, prairie

vole males and females display marked differences in their

parental behavior which reflect their proposed mating

systems in the wild (e.g., Dewsbury, 1985; Hartung &

Dewsbury, 1979; McGuire & Novak, 1986; Oliveras & Novak,

1986; Wilson 1982a, 1982b).

In an effort to synthesize many of the behavioral

indicators of monogamy proposed by Kleiman (1977), Dewsbury

(1981) developed a composite "monogamy scale" on which he

ranked 42 species of muroid rodents. Species were ranked

on such traits as sexual dimorphism, the Coolidge effect,

parental care etc. More recently, Dewsbury (in press) has

summarized many of the behavioral differences which have

been compiled for these two species.

The following studies are divided into three sections,

each one leading logically into the next. In Section II,

the species will be compared in a complex social situation

in an effort to highlight salient differences in behavior.

Section III is focused on a key species difference, contact

proneness; this behavioral "propensity" is subjected to

detailed experimental analyses. In the final section,

Section IV, the brains of prairie voles and montane voles

are compared in an effort to elucidate neuroanatomical

differences that might underlie this and/or other relevant

behavioral differences.


Experiment 1: Behavioral Profile


This experiment was designed to quantify differences in

the behavior of prairie voles and montane voles in a complex

social situation. As a result of observing both species

under identical conditions, it was hoped that key behavioral

propensities, such as patterns of nesting and pairing, inter-

and intrasexual aggression, etc., would be identified which

function to generate differences in the social organization

of these two species.


Animals. A total of 15 one-male/two-female prairie

vole trios (i.e., 30 females and 15 males) and a total of 11

montane vole trios (i.e., 22 females and 11 males) were

observed in this experiment. All animals were laboratory

bred and at least 60 days of age at the start of testing.

Animals were housed in clear polycarbonate cages measuring 29

x 19 x 13 cm with wood shavings used as bedding. They had

continuous access to Purina rabbit chow and water. There was

a 16L:8D photoperiod of white flourescent light with light

onset at 2,000 hours; dim red lights were on at all times.

Animals were sexually inexperienced at the start of testing.

Apparatus. Two seminatural test enclosures were used

in this study; the two apparatuses were similar in size but

not identical in construction; in one enclosure the base was

plywood in the other it was plexiglass. Both were

constructed on a 4 x 8 ft base (238 x 117 cm) and were 64 cm

high. In both apparatuses three walls were of plywood and

one, the front, was plexiglass. A plywood partition divided

each apparatus into two compartments of equal size; two trios

were thus simultaneously run in each apparatus. A substrate

of 50% peat and 50% wood chips was used in each apparatus.

Several rocks and branches were placed about each

compartment. In addition, each compartment (i.e., one-half

of the apparatus) contained three individual-sized

polycarbonate cages (29 x 19 x 13); plastic cylinders 18.4 cm

long and 5.1 cm in diameter were employed as entrances to

each cage. These cages were present so that each animal had

the opportunity to nest separately. Purina rabbit chow was

present ad lib and there was one water bottle available at

all times at the rear of each compartment.

Procedure. Females were observed in the seminatural

enclosure alone for the first 3 days, trios (i.e., one male

and two females) for the final 7 days. On the morning of the

first day of testing all animals were weighed and marked for

identification. A small patch of fur was clipped from the

neck of one female and the rump of the other; males remained


Observation periods of 10 min in length were made three

times daily by a single observer. The first observation

period was made between 900 and 1200 hours; observation

period 2 was made between 1200 and 1800 hours; the third

observation period was carried out between 1800 and 2100

hours. Behavioral observations were recorded by hand on

prepared data sheets. The positions of all animals were

noted as well as any instances of copulatory behavior. The

specific behaviors measured and their definitions were as



Rough &



Body-nosing contacts were generally
very brief naso-body contact episodes
lasting less than 2 sec. No attempt was
made to establish different categories of
body-nosing, e.g., naso-anal,
naso-genital etc. Frequency of
occurrence was recorded for this

This behavior was scored when two
animals were engaged in vigorous
fighting with both animals tumbling
end-over-end while attempting
to bite and claw each other. Frequency was

This behavior was scored when one animal
vigorously pursued another. Frequency
was recorded.

Boxing was scored when two animals were
engaged in vigorous and rapid
forepaw-forepaw contact. Animals were
almost exclusively in an upright position,
i.e., standing on their hindlimbs,
when engaged in this behavior.
Frequency was recorded.

Huddling Duration rather than frequency was
recorded for this behavioral category.
Huddling was scored when two animals
rested in contact with each other.
Animals had to remain stationary
and in direct contact for at least 5 sec
before they were considered to be
huddling. Duration was then recorded
using a stop watch. Scores were
recorded for three classes of animals:
females (f/f), males and females (m/f),
and three animals huddling together (trio).


Species differences recorded in the seminatural

apparatus are summarized in Table 1. There were no

significant differences between the species on any behavioral

measure when just female/female (f/f) interactions were

considered. That is, for all categories considered during

the 10 days of observation in the seminatural apparatus

(i.e., body-nosing, rough and tumble, chases, boxing, and

huddling duration) a comparison of female interactions

revealed no differences between species. When interactions

between males and females (m/f) were considered, however,

prairie voles displayed significantly lower frequencies of

body-nosing, chases, and boxing than did montane voles. When

composite frequencies of aggression (i.e., the sum of chases

+ rough and tumble + boxing) are plotted across time (Figure

1) it is evident that aggression remained fairly constant for

prairie voles throughout the 10-day observation period; for

montane voles aggression was highest on day 4, when the male

was introduced, and decreased rapidly after that.

Table 1

Total instances of behavior and huddling duration (min) during
10-day observation period for prairie and montane vole groups
in the seminatural apparatus (mean + SE)

Measure Prairie voles Montane voles p

Body-nosing (f/f) 2.6 + 1.0 2.4 + 0.8 NS
Body-nosing (m/f) 7.8 + 1.9 21.6 + 4.9 <0.006

Rough & tumble (f/f) 0.2 + 0.1 0.8 + 0.7 NS
Rough & tumble (m/f) 0.1 + 0.1 0.8 + 0.7 NS

Chases (f/f) 1.4 + 0.6 1.4 + 0.8 NS
Chases (m/f) 0.4 + 0.1 5.7 + 2.3 <0.008

Boxing (f/f) 0.6 + 0.4 0.2 + 0.1 NS
Boxing (m/f) 0.1 + 0.1 3.6 + 0.9 <0.001

Huddling (f/f) 23.8 + 5.6 10.3 + 3.0 NS
Huddling (m/f) 59.7 + 8.9 3.6 + 1.5 <0.001
Huddling (trio) 29.4 + 8.4 0.0 <0.009

Statistical results based on independent samples t-tests.


1.8 -
1.7 -
1.6 -

1.4 -

0.3- t J
0.1 -

1 2 3 4 5 6 7 8 9 10


Figure 1. Composite measures of aggressive behavior for
prairie vole and montane vole groups during a 10-day
observation period in the seminatural apparatus.

One of the key differences highlighted in these results

is the degree of social contact. Figures 2 and 3 graphically

display species differences in huddling. Prairie voles spent

an appreciable amount of each observation period (600 sec) in

social contact with another animal (Figure 2). Prairie voles

spent a significantly greater amount of time huddling and/or

nesting together as male/female pairs than did montane voles,

and they also spent a greater amount of time huddling/nesting

in contact as trios than did montane voles which were never

observed to do so (Figure 3).

In addition to spending a greater amount of time

huddling in contact, prairie vole males spent that time

preferentially with one of the females in the pair. Prairie

voles spent an average of 28% of the total observation time

huddling (Table 1) and in 10 of the 15 groups of prairie

voles, males spent at least 75% of this huddling time with

one particular female. Moreover, this relationship was

relatively stable over the testing period. Montane voles, on

the other hand, spent only 2% of their total observation time

huddling in male/female pairs.

During the observation period copulation was observed in

5 of the 15 prairie vole groups and in 6 of the 11 montane

vole groups. All five of the observed prairie vole

copulations occurred after day 8; one copulation was observed

in montane voles on the day the male was introduced, day 4;

the remaining copulations occurred after day 8.

1 2



3 4 5 6 7 8


9 10


2. Mean huddling duration for prairie vole groups
a 10-min observation period.

500 -

400 -^

300 -

200 -

100 -1



- ---

600 -

500 -

400 -

300 -

I 1 I I I I I I
1 2 3 4 5 6 7 8 9 10




Figure 3. Mean huddling duration for montane voles groups
during a 10-min observation period.

p,=, 1 7-



The species differences that emerged from this study are

consistent with results from previous laboratory and field

studies. In general, montane voles appear to be more

aggressive and less social than prairie voles. Of interest

is the fact that the species do not differ if only

female/female interactions are considered. That there was no

appreciable aggression between female prairie voles was

surprising in light of the findings of Carter et al. (1986)

who reported that mated prairie vole pairs directed

aggression toward strangers. One might expect aggression to

have increased following introduction of the male and the

formation of male/female pairs. This was not the case.

However, in the Carter et al. study aggression was directed

toward unfamiliar animals; in the present situation the

females had been housed together in the enclosure for days

prior to the introduction of the male and were thus familiar

to each other in advance of any pair formation. The amount

of time all three prairie voles spent huddling/nesting

together following pair formation (Figure 2, days 8, 9, and

10) is therefore most likely a result of prior association of

females as well as a species-typical propensity for social


More indicative than female/female interactions in

determining social organization may be the interactions

between males and females, and in this regard there was a

clear-cut species difference. When male/female interactions

are considered, montane voles differ from prairie voles in

chases, rough and tumble fights, and boxing. Montane voles

displayed the greatest amount of aggression on day 4 (Figure

1), the day that the male was introduced. Prairie vole

levels of aggression, on the other hand, were affected little

by the introduction of the male. This finding is in

disagreement with Kleiman (1977) who concluded that female

mammals of monogamous species exhibit higher levels of both

inter and intrasexual aggression than their non-monogamous

counterparts. It has previously been noted that female

montane voles are much more aggressive than are prairie voles

to the initial approaches of the male (Gray, Kenney, &

Dewsbury, 1977). Indeed, stand-up bouts of boxing between

males and females normally occur before each mount during

copulatory bouts in montane voles whereas this behavior

rarely if ever proceeds mounting in prairie voles. Whether

or not the male, the female, or both are initiating this

aggression is unclear.

Unfortunately the present data do not permit a

comparison of intrasexual levels of aggression within and

between the species. However, previous studies have reported

that male montane voles display higher levels of aggression

than prairie vole males (Colvin, 1973b; Dewsbury, 1983).

From these data it would appear that levels of male/male and

male/female aggression are higher in montane voles whereas

levels of female/female aggression are similar in both

species. Dewsbury (1983) found that species differences in

levels of aggression did not correlate with mating systems in

six species of muroid rodents. Nevertheless, they may

provide insight into the proximate mechanisms which determine

spacing or territorial relationships (see General


For instance, Colvin (1973a) found montane vole males to

be more aggressive than meadow vole males. Consistent with

this laboratory finding is the fact that although both

species display polygamous mating systems in the field, the

territories of male montane voles do not overlap whereas

those of meadow vole do (Madison, 1980a).

That prairie voles spent more time in contact with each

other and that males preferentially paired with one of the

two females is consistent with results from the field (e.g.,

Getz et al., 1981; Getz & Hofman, 1986; Jannett, 1980, 1982)

and the laboratory (e.g., Carter et al., 1986; Fuentes &

Dewsbury, 1984; Thomas & Birney, 1979) demonstrating a

propensity toward monogamy in prairie voles. Montane voles

spent only 2% of their total time huddling in contact.

Recall that there was actually a slightly higher percentage

of copulations observed in this species than for prairie

voles. The observed lack of pair formation and social

contact cannot, therefore, be attributed to a general absence

of sexual receptivity on the part of female montane voles.

In the next experiment, vaginal cytology is examined in

an effort to correlate species differences in cell types

with differences in social organization.

Experiment 2: Vaginal Smears


In the previous experiment male and female prairie

voles displayed a greater amount of social contact and pair

formation and were less aggressive in their overall

interactions than were montane voles. Both of these

behavioral traits could serve as "mechanisms" in maintaining

differences in their social organization. Mechanisms other

than behavior, however, can be important in predisposing a

population in favor of one mating system rather than

another. For instance, relationships between social

environment and ovarian function can be an important

determinant of the operational sex ratio at any particular

time and this can be an important factor underlying the type

of mating system which a population displays (Emlen & Oring,

1977). In rodents, it is known that olfactory pheromones

from both males and females can have dramatic effects on

estrous cyclicity. Male pheromones are known to reinstate

ovarian cyclicity in a group of acyclic females (Whitten,

1956), to block pregnancies pre- and postimplantation

(Bruce, 1959, 1960; Labov, 1981, and references therein;

Stehn & Richmond, 1975), as well as to accelerate the onset

of puberty in young females (e.g., Marchlewska-Koj, 1977;

Vandenbergh, 1973). Female pheromones can cause the

suppression of cyclicity in mice (van der Lee & Boot,

1956) and the opposite, enhancement and synchronization

of cyclicity in rats (McClintock, 1983a) and humans

(Graham & McGrew, 1980; McClintock, 1971). As

McClintock (1983a) points out, the interaction between

male and female behavior and ovarian cycle components can

be radically different in different species. The purpose

of this experiment is to examine vaginal smears within

the social context of the seminatural enclosure and

evaluate the possible functional significance of any

observed species differences in cell patterns.

Prairie and montane voles are considered induced

ovulators. That is, ovulation in these species is

contingent upon stimulus input. In both species either

copulatory stimulation or direct interaction with a male

can trigger ovulation (see Sawrey & Dewsbury, 1985, for

review). Female montane voles separated from a male by a

hardware cloth barrier or exposed to male bedding failed

to ovulate (Davis, Gray, Zerylnick, & Dewsbury, 1974;

Gray, Davis, Zerylnick, & Dewsbury, 1974). Likewise no

prairie voles females housed across form a male (0/7) or

exposed to male bedding showed histological signs of

ovulation (Richmond & Conaway, 1969a; Gray et al., 1974).

In addition to the above similarities, the two species

also show interesting differences. One difference has to do

with the amount of copulatory stimulation necessary to

produce functional corpora lutea. In montane voles, the

percentage of animals producing functional corpora lutea was

found to increase with the amount of copulatory stimulation

they received (Davis et al., 1974). Davis et al. found that

25% of the females that received only one ejaculatory series

ovulated, 75% of those that received two ejaculatory series

ovulated, and all of those receiving four or more did so.

In contrast, 90% of the prairie vole females receiving just

one ejaculatory series ovulated and produced functional

corpora lutea (Gray et al., 1974).

Of particular relevance for the present discussion is

that in addition to differences in the amount of copulatory

stimulation required to induce ovulation and/or a functional

luteal phase, prairie and montane vole females housed in

individual cages under identical conditions in the

laboratory display a curious difference in the appearance of

their vaginal cytology. This is important because vaginal

cytology generally reflects and is correlated with both the

endocrine activity of the ovaries and behavioral receptivity

(Sawrey & Dewsbury, 1985). Prairie voles caged in isolation

display smears dominated almost entirely by leukocytes

(Richmond & Conway, 1969a, 1969b); montane voles housed

under these same conditions display smears dominated by

cornified cells (Sawrey, unpublished data).

That females of these two species display differences in

vaginal cell patterns when they are housed in isolation

raises the possibility that they will also display

differences in the presence of other females and males and

that these differences will have functional significance.


Animals. The same group of animals described in the

previous experiment were utilized for the recording of

vaginal smears.

Procedure. Estrous stages were monitored by vaginal

smears taken once each day during the testing period

outlined in the previous experiment. The smears were

assessed for percentages and types of cells present.

Samples were taken for 3 days prior to entry in the

seminatural enclosure, while females were individually

housed, for 3 days following entry, when the two females

were together in the seminatural apparatus, and for 7 days

following the introduction of the male into the apparatus.

Thus, vaginal cytology was recorded for each female in the

seminatural study over a 13-day period which covered three

social conditions: 1) isolate, 2) female, and 3) trio. Care

was taken to disturb the animals as little as possible

during the smear taking process.

The procedure for obtaining a smear is relatively

simple and requires very little manipulation of the animals.

The technique consists of inserting a small wire loop into

the vagina and microscopically examining the cells obtained.

The correlation of vaginal cytology with

reproductive/endocrine state is based on the well studied

rat model. In the rat, the ovarian cycle is rhythmic and

can be described by a sine wave with a period of roughly 4-5

days. The sequence of events is follicular development,

ovulation, and, often, the formation of corpora lutea. Each

stage in this series of events is correlated with a

different vaginal smear pattern (after McClintock, 1983b,

1984). The cycles begin with smears containing primarily

nucleated epithelial cells (proestrus); this is followed by

smears in which cornified epithelial cells dominate

(estrus); the third phase consists of smears containing both

cornified epithelial cells and leukocytes (diestrus-1); in

the final stage of the cycle, smears consist primarily of

leukocytes (diestrus-2). Induced ovulators, such as prairie

and montane voles, do not, in general, cycle with regularity

as do rats and mice. However, what is important for this

study is not cyclicity but the consistent correlation of

particular cell types in the vaginal smear with

corresponding endocrine/behavior stages. That is, the

consistent correlation between, for instance, cornified

cells and behavioral estrus. In this regard, rats and mice

appear similar to voles of the genus Microtus (e.g., Breed,

1967, and references therein; Chitty & Austin, 1957).


Figure 4 reveals that the two species displayed

significant differences in their vaginal cytology when

housed in isolation. During the first 3 days of smears when

the females were still housed in their home cages, 18 of the

22 (82%) montane voles displayed at least one day of vaginal

proestrus or estrus. In contrast, of the 30 prairie vole

females observed, only 6 (20%) showed at least one estrous

smear during the first 3 days, (1l, N = 42) = 10.24, p <


Days 4-6 reflect the period when female dyads were in

the enclosure. During this period, 4 of the 30 (13%)

prairie vole females displayed either a proestrous or

estrous smear (i.e., were possibly receptive behaviorally);

13 of 22 (59%) montane vole females showed an estrous or

proestrous smear, (1, N = 42) = 12.07, p < .001.

On day 7, the male was introduced into the apparatus;

days 7-13 thus reflect the period in which

one-male/two-female trios were housed together. The number

of females in estrus rose to a high for both species on day

8 and then tapered off until day 13 (Figure 4). In 60%

(9/15) of the prairie vole groups, only one female of the

pair was ever observed to display vaginal estrus after the

introduction of the male. That is, the other female of the

trio remained in diestrus for the entire 10 day period in

the seminatural enclosure. In six of these nine groups, the

female that displayed vaginal estrus was the one which was

designated as preferred according to the criterion discussed

in Experiment 1 (i.e., a female was designated as preferred

if the male of the trio spent at least 75% of his total

huddling time with that female). In contrast, in all of the

11 montane vole groups both females displayed at least one

vaginal estrus following the introduction of the male and in

all but one group both females cycled through both diestrus

and estrus at least once.

Figures 5 and 6 depict the percentage of female pairs

displaying estrous synchrony, i.e., pairs in which both

females displayed cornified smears, on each day of the test.

The percentage of prairie vole pairs in synchrony (Figure 5)

reaches a peak during days 7 (24 h after introduction of

male) through 9. However, the absolute percentage was still

lower relative to montane voles (13, 20, and 13% for prairie

voles vs. 36, 36, and 27% for montane voles). On days 10-13

the percentage of pairs in estrous synchrony decreased for

both species dropping to 0% on the final day of the

observation. While the trend for estrous synchrony was

analogous for both species as the test period ends, a

comparison of Figures 5 and 6 reveals an interesting

difference in the percentage of pairs which are

asynchronous, that is, pairs in which one of the females was

in either proestrus/estrus and the other was in diestrus.

The percentage of montane vole pairs which were asynchronous

increased from a low of 18% on day 10 to a high of 73% on

day 13. Prairie voles showed a slight trend in the opposite

direction; the percentage of pairs in asynchrony went from

33% on day 10 to 7% on day 11 and remained at 20% for days

12 and 13.


The above results indicate that female prairie and

montane voles display different patterns of vaginal cytology

both in isolation and in social contexts. During days 1-3

the smears of individually housed montane voles were

dominated by cornified cells whereas the smears of prairie

voles were dominated by leukocytes. These data are

consistent with previous finding reported by Sawrey and

Dewsbury (1985, and references therein). The important

question raised by these observations is whether or not

individually housed montane voles are in a state of constant

behavioral receptivity and prairie voles not. While it is

tempting to draw inferences from cytology to behavior, the

relationship between vaginal cytology and behavioral

receptivity has unfortunately not been conclusively


60 -

50 -



20 -


0 I -- -- -- I"-- i-----i1i -- i -

1 2 3 4 5 6 7 8 9 10 11 12 13

Figure 4. Percentage of female prairie and montane voles
displaying vaginal estrus (i.e., smears dominated by cornified
cells) for the entire 13-day observation period. During days
1-3 females were individually housed; during days 4-7 they
were in female/female pairs in the seminatural enclosure;
during days 7-13 they were in two-female/one-male trios.

- -- #- -


\\- _

-\ \\ \\\ r
-\ \ ., \ ,N
\- .:
-': \\ d
-\\ \,,,

- \ _-. _. .",

1 2 3 4 5 6 7 8 9 10

11 12 13



Figure 5. Percentage of female/female prairie vole pairs
displaying synchrony or asynchrony in their vaginal cell types
for entire 13-day observation period. During days 1-3 females
were individually housed; during days 4-7 they were in
female/female pairs in the seminatural enclosure; during days
7-13 they were in two-female/one-male trios.












1 2


3 4 5 6 7


8 9 10 11


Figure 6. Percentage of female/female montane vole pairs
displaying synchrony or asynchrony in their vaginal cell types
for entire 13-day observation period. During days 1-3 females
were individually housed; during days 4-7 they were in
female/female pairs in the seminatural enclosure; during days
7-13 they were in two-female/one-male trios.











-X ol

-/ # -#

-\ \ ,, .\ ,.\ .\ ,

\"\ \ \ "\ "\ i
\ \ -\ \ -\ .,

., ..
"-: \ \'"\ N 1

demonstrated for any species of induced ovulators (Sawery &

Dewsbury, 1985).

Nevertheless, the existing evidence seems to indicate

that these two phenomena are in fact correlated for prairie

and montane voles. First, being housed adjacent to males

resulted in vaginal cornification for both species (Gray et

al., 1974; Richmond & Conaway, 1969a). Richmond and Conaway

(1969b) also reported that females with at least 50%

cornified smears were almost always found to be receptive to

a male. The data from the present study also support the

view that cornified cells are correlated with behavioral

receptivity. There were a total of six occasions on which

sperm was found in a smear and in all cases it was

associated with smears predominated by cornified cells.

Whether cornified cells are a necessary or sufficient

condition remains to be established.

What may be more important than cell types present at

any one time, are the temporal changes of cell types.

McClintock (1983a) suggests that in spontaneously cycling

species, like laboratory rats, ovulation may be represented

not simply by the types of cells in a smear but by the

transition from leukocytes to epithelial cells in the smear

pattern. In addition, McClintock (1984) reported that

during social isolation in rats, the lordosis component of

ovarian cyclicity became disassociated from vaginal

cornification (with which it is normally correlated) and

could be elicited on any day even though vaginal

cornification was not evident. Whether a pattern of

constant cornified smears is correlated with a constant

state of behavioral estrus remains to be determined.

Following 3 days of smears taken in isolation, the

females were placed in the seminatural enclosure for 3 days.

Relative to the first 3 days in isolation, there appears to

be a weak trend toward the suppression of vaginal estrus.

It appears that in both species there was a slight decrease

in the number of females in estrus on days 4-6 when two

females were in the enclosure in the absence of the male.

Indeed, on day 5 both prairie voles and montane voles

experienced the lowest numbers of females displaying

cornified smears. These data are consistent with those of

Getz et al. (1983) who found that female prairie vole

females may actively suppress ovarian cyclicity in each

other. Carter et al. (1986) also observed some degree of

female suppression in one-male/two-female trios. Although

they found some mating activity in all four of the six trios

they observed, they noted that the presence of two females

"disrupts the behavior of one and perhaps both females in

the trio" (p. 135). Nevertheless, it is difficult to

determine conclusively whether or not suppression was taking

place in the present experiment since the females in this

study may not have been housed together long enough for

these effects to manifest.

There are other complications as well which preclude

any unequivocal interpretations of these data. First, the

effect can vary with species (van der Lee & Boot, 1956).

McClintock (1983a, 1984) has studied the phenomenon of

female suppression in rats and has shown that the situation

is quite different than it is for mice. In rats, grouped

females seem to shorten (i.e., increase the frequencies)

rather than suppress their ovarian cycles. The situation is

even more complex in that there seems to be one pheromone

(during ovulation) that phase delays the cycle and another

(a preovulatory pheromone) that phase advances or enhances

the cycle.

In the present study, following the introduction of the

male, the number of females in estrus rose slightly for both

species and then levelled off at the end of the test period.

In 60% of the prairie vole pairs in the present study only

one of the females ever displayed a cornified smear. Put

another way, in 60% of the prairie vole pairs one female

remained in diestrus throughout the entire observation

period. Interestingly, in four of the five pairs in which

both females displayed estrous smears the females came into

"vaginal estrus" either simultaneously or within 1 day of

each other. However, as mentioned above, female/female

suppression may have precluded any possibility of observing

estrous synchrony within a test paradigm such as the one

used in this experiment. Moreover, the necessary use of

only one male may have also affected this outcome.

Montane voles, which never huddled as trios, did

display a high degree of asynchrony. During the last 3 days

of the test period montane vole trios displayed a distinct

trend toward asynchrony with the highest value (80%)

occurring on the final day. And it is important to note

that the asynchrony in montane voles was not the result of

one female cycling and the other remaining in diestrus but

rather the alternation of the two females. Of interest is

that in 100% of the montane vole trios both females

experienced at least one complete cycle. Thus, although

induced ovulators are not generally considered to be cyclic,

in the presence of other males and/or females they may well


Had the test period extended for a longer period of

time species differences in vaginal cytology may have been

more revealing. As with other differences displayed by

these species, an understanding of the contrasting patterns

of vaginal cytology in both proximate and ultimate terms is

essential and is fertile ground for future research.


The results of the previous section highlight a species

difference in the propensity for social contact between

males and females. This particular behavioral trait,

perhaps more than any other, may be the key difference that

distinguishes a monogamous and/or highly social species from

polygamous and/or solitary-living species. There are many

questions one might ask concerning the evolution and

ontogeny of this trait. For instance, to what degree is

contact proneness in adult animals the result of the

treatment that they received as infants? In species which

form pairbonds, are both the male and the female more social

than their counterparts in species that do not form

pairbonds? What are the physiological mechanisms which

underlie differences in this behavioral propensity? The

following series of experiments focuses on this particular

trait and attempts to answer the above questions. The

overall research strategy was to utilize huddling or resting

in contact as a dependent variable and to then examine the

effects of various factors on this behavioral propensity.

Experiment 3: Test-Cage Baseline


Experiment 3 was the first in a series of experiments

designed to gain a better understanding of why it is that

these two species differ in such a fundamental way as their

propensity for social contact. The specific goal of this

experiment was to establish a baseline, or control,



Animals. Thirty female and 30 male prairie voles and

30 female and 30 male montane voles were used in this study.

All animals were laboratory bred and at least 60 days of age

at the time of testing. Animals were individually housed

for at least 1 week before testing. Animals were housed in

clear polycarbonate cages measuring 29 x 19 x 13 cm with

wood shavings used as bedding. They had continuous access

to Purina rabbit chow and water. There was a 16L:8D

photoperiod of white flourescent light with light onset at

2,000 h; dim red lights were on at all times. All animals

were sexually inexperienced at the time of testing.

Apparatus. The test-cage used in all of the

experiments in Phase II was a clear polycarbonate cage

measuring 48 x 27 x 13 cm. Wood shavings were used as

bedding; the test-cage bedding was changed before each test.

A wire mesh top was placed on the cage during testing and

animals were provided with rabbit chow and water during


Procedure. Male/female pairs were chosen arbitrarily

at the beginning of each test. The criteria were that the

animals were not littermates, were unfamiliar, and had been

used in no previous tests of any kind. To begin the test a

male and a female of the same species were placed in the

test-cage and allowed to freely interact for 2 h. At the

end of this time the female was removed from the test cage

and placed in her home cage (29 x 19 x 13 cm) for 5 min.

Following this 5 minute period, the female was replaced in

the test cage. The behavior of the pair was then videotaped

for the next 60 min. It was during this third hour that the

time spent huddling or resting in contact was recorded. To

summarize, the procedure which was followed in this and in

all of the experiments of Section III was as follows: First,

2 h of familiarity, then a 5 min removal of the female, and,

finally, a third and final hour during which huddling in

contact was scored. All tests were begun at the onset of

the dark phase of the cycle, at 1200 h.

The specific protocol utilized in this section was

arrived at after various pilot tests were run. The results

of these preliminary tests were as follows. When animals

were scored for huddling immediately after they were placed

in the test-cage (i.e., during the first or second hour of

familiarity), there was no huddling in montane vole pairs

and "some" huddling in prairie voles. On the other hand,

when pairs were left overnight and scored the following

morning for huddling, there was almost universal huddling in

prairie voles and some huddling in montane voles. As a

result of this temporal titration process, it was found that

2 h of familiarity followed by recording during the third

hour was the particular combination that produced the most

robust species difference in huddling duration; hence it was

chosen. It is likely that both the short and long

pre-scoring time periods would have produced statistically

significant differences between the species. However, the

goal of this series of experiments was not simply to

demonstrate a species difference. It was, rather, to

produce the most marked species difference so that any

effects of independent variable manipulation (i.e.,

Experiments 4 through 6) would be more easily detected

should they arise.

The process of removing the female for 5 min following

the 2 h familiarity period was included so that Experiments

3, 4, and 5 would be consistent with Experiment 6, in which

it was necessary to inject both animals with drugs prior to

behavioral scoring during the third hour.

In all of the experiments in Section II, the dependent

variable recorded was the time spent huddling. Animals were

defined as huddling when they were sitting or lying in

contact for at least 5 sec. While huddling the animals

might be active (i.e., grooming, sniffing, etc.) or

inactive; the defining criteria were simply that they be

stationary and in contact with one another.

In addition to huddling duration, a subset of the

animals in Experiment 3 and in Experiment 6 was scored for

activity levels during the non-huddling time period of the

test. The activity of individual males and females was

allocated to one of the following two categories.

(1) Stationary/alone: This category included

non-contact sitting or lying in place, self-grooming, etc.

(2) Locomotion: This category included walking,

digging, or jumping.


There was a significant species difference in the

amount of time animals huddled or sat in contact (Figure 7).

Prairie vole pairs huddled for a mean duration of 31.22 min;

montane vole pairs huddled for a mean duration of only 1.29

min, t(28) = 11.78, p < .001.

In the subset of pairs analyzed for activity levels, a

two-way ANOVA (Species x Sex) with repeated measures over

the second factor, revealed a species difference in the mean

time spent stationary/alone (Figure 8). For prairie voles

the mean duration in this category was 9.62 min; for montane

voles the mean time spent stationary/alone was 30.53 min, F

(1,9) = 21.42, g < .001. There was also no significant

difference between males and females either between or

within the species, F(1,9) = 0.06, R > .005. There was no

statistical difference in locomotion between prairie voles

(M = 14.76) and montane voles (M = 25.43), F(1,9) =

2.75, p > .05. Again there was no difference between males

and females either between or within species, F(1,9) =

1.02, p > .05.

The difference in huddling duration for this subset of

pairs was similar to the larger sample. Mean huddling

duration for prairie voles was 35.28 min and was 2.13 min

for montane voles p < .001, Mann-Whitney U test.


The difference in social contact evident in the

seminatural apparatus was the focus of this experiment.

Evidently the "reinforcing value" of body contact between

male/female pairs is markedly different in these two species

and is consistent with the mating systems that they display

in the wild. Although Wilson (1982a) did not directly

compare prairie voles with montane voles, she did find that

when juvenile prairie voles were compared to meadow voles in

dyadic encounters prairie voles showed more sitting in

contact than did meadow voles. Her results are also

consistent with the mating systems of those two species









05 I I/////A

Figure 7. Baseline comparison of huddling duration for prairie
vole and montane vole male/female pairs during 1 hour test.




16 -

10 -








Figure 8. Baseline activity patterns for a subset of prairie
and montane vole male/female pairs during 1 hour test.


since meadow voles, like montane voles are reported to have

a promiscuous mating system with females actively defending

territories and not co-nesting with males (Madison, 1980a).

Carter et al. (1986) measured the strength of

pairbonding in terms of aggression toward unfamiliar males

or females following copulation. It is of interest that

none of the animals used in the above tests were observed to

mate. Therefore, the pairing of male and female prairie

voles should be considered a robust species-typical

propensity for social contact and not simply a result of

copulation per se as observed by Getz et al. (1981) and

Shapiro et al. (1986).

Although prairie voles had a slightly higher locomotion

score, the results of the subset analysis demonstrate that

species difference did not reach statistical significance.

These results are consistent with those of Wilson, Vacek,

Lanier, and Dewsbury (1976) who compared these two species

in the open-field. Wilson et al. (1976) found that montane

voles entered more squares during an hour test than did

prairie voles but, as in this study, the differences were

not significant. In the present study, what differed

between the species was the manner in which they spent their

stationary time (Figure 8). Prairie voles spent their time

in contact; montane voles did not. These results provide

further evidence that the phenomenon we are observing in

this study reflects a genuine species difference in contact

proneness rather than simply an indirect result of differing

activity levels.

Experiment 4: The Effects of Cross-pairing


An interesting question one might ask concerning a

behavioral tendency toward social contact versus a tendency

toward a solitary existence is whether or not the phenomenon

is gender-specific or whether it is shared equally by both

sexes. Carter et al. (1986) suggested that it may be

gender-specific. Progesterone has been shown to inhibit

sexual receptivity in rabbits and in brown lemmings (Dluzen

& Carter, 1979) whereas in species with spontaneous

ovulation it facilitates receptivity (Carter et al., 1986).

Carter et al. suggest that in induced ovulators, such as

prairie and montane voles, progesterone, while not necessary

to facilitate receptivity, may function to inhibit female

sexual receptivity and thus terminate male-female

encounters. Therefore, species differences in female

receptivity may underlie species differences in social


This is difficult to determine empirically. Behavioral

observations may not reveal any overt aggression and, even

if they do, it is often difficult to determine whether it is

the male or female that is initiating/terminating them.

Although less than ideal, one possible experimental approach

is to cross-pair the genders of two closely related species,

one that pairs and one that does not. At least in the

laboratory, different species of Microtus have shown

interspecific copulation. Gray, Kenney, and Dewsbury (1977)

observed interspecific copulations in laboratory tests

between meadow vole males and prairie and montane vole

females. More to the point, cross-species copulation has

been observed for prairie voles and montane voles (Shapiro,

unpublished observations). Given that interspecific

copulations do take place in these two species it is not

entirely unreasonable to assume that gender-related

differences in contact proneness might be revealed.

That is, if prairie voles males are paired with montane

vole females (pvm X mvf), and conversely, if montane voles

males are paired with prairie vole females (pvf X mvm), will

pairing be exhibited during the third hour of observation,

and if so for which combination? The following experiment

utilized this paradigm in an attempt to infer

gender-specific tendencies for social contact.


Animals. Nineteen pairs of prairie vole males and

montane vole females (pvm X mvf) and 19 pairs of montane

vole males and prairie vole females (pvf X mvm) were

utilized in this experiment. None of the animals used in

this experiment had previously been used in any other

experiment (see Experiment 3).


The procedure followed in this experiment was identical

to that followed in Experiment 3 above, with the exception

that the pairs of animals placed in the test-cage were

interspecifically crossed.

This experiment had two parts. In the first, 7 pvm X

mvf crosses and 7 pvf X mvm were scored during the third

hour. In part 2, the procedure was slightly different in

that 24 h before the beginning of testing animals of each

species were taken from their home cages and placed in the

home cage of a like-sex heterospecific. For example,

montane vole females were placed in the home cages of

prairie vole females and prairie vole females were removed

from their home cages and placed in home cages of montane

vole females where they remained overnight. This extra step

was added in order to habituate the subjects to the

olfactory stimuli of the other species prior to testing.


In part 1 of this experiment--i.e., 7 pairs of each

cross with no heterospecific cage habituation--there was no

pairing observed in any of the 14 tests. Following

habitation in the home cage of a heterospecific, the pvm X

mvf group differed significantly from the pvf X mvm group

(M = 20.01 min vs. M = 4.28), t(20) = 2.21, p < .05.


The most salient result of this study was that in the

initial cross-pairing neither group displayed a significant

amount of huddling. None of the 7 pvm X mvf pairs huddled

during the third hour and only 1 of the 7 pvf X mvm pairs

huddled and that was for 11 min, well below the average for

the baseline group of Experiment 3. This may have been the

result of interspecies factors other than contact proneness.

Part 2 of the above experiment was an effort to eliminate

such potential confounding variables as species-typical

differences in olfactory cues. In other words, the

manipulation was an attempt to give animals a chance to

become familiar with or habituate to the olfactory stimuli

associated with the other species. A comparison of the

groups that were exposed to this treatment displayed a

significant difference--male prairie voles with female

montane voles huddled significantly more than did prairie

vole females paired with male montane voles.

First, from Experiment 3 it is apparent that both

prairie vole males and females have a propensity for social

contact with each other--i.e., pairing or resting in contact

must involve both members of the pair. Second, either one

or both members of montane vole pairs might be responsible

for the lack of huddling in the baseline study. (Indeed,

the question of gender-specific aggression only becomes

relevant in reference to a species which does not form

pairs.) Third, in the second part of the present experiment

female montane voles did not forcibly resist or prohibit

male prairie voles from resting in contact with them (pvm X

mvf). If, for the moment, certain obvious species confounds

that might have interfered with these results are ignored,

it would superficially appear that montane vole males and

not females are the causal factor for that species' lack of

huddling in Experiment 3.

One is thus tempted to infer from these results that

montane vole males rather than females were the "cause" of

the lack of huddling displayed by this species in Experiment

3. Indeed, this conclusion is consistent with the results

from Experiment 1 in which there were no differences between

the species in levels of female/female aggression, yet

male/female aggression was higher in montane voles.

Unfortunately, it is just as likely in the above experiment

that prairie vole females were resistant to male montane

voles or male montane voles simply found female prairie

voles unattractive (pvf X mvm). These are possibilities

which cannot be addressed by the present data. It would

have been helpful to have gathered data on the number of

approaches, chases, etc., in order to gain further insight

into gender-specific differences.

As mentioned in the introduction, Carter et al. (1986)

hypothesized that post-copulatory levels of progesterone

might terminate sexual behavior and increase inter-animal

distances. It should be remembered that no mating took

place in the present study. Therefore, if copulation were a

prerequisite for aggression in female montane voles, then

the fact that no pairing occurred either in the montane vole

control test (Experiment 3) or in the second part of this

study (pvf X mvm) could be explained by a high level of

baseline agonism in montane vole males relative to prairie

vole males and females (e.g., Dewsbury, 1983; Shapiro &

Dewsbury, 1986) as well as to females of their own species.

If Carter et al. (1986) are correct regarding their

hypothesized progesterone/aggression relationship, then

based on the results of Experiment 1 in which no pairing was

observed in montane voles following mating, we would expect

either higher levels of post-mating progesterone in montane

voles versus prairie voles or an increased sensitivity to

the hormone. In fact, when Gray, Kenney, and Dewsbury

(1976), measured plasma levels of progesterone in female

montane voles following mating, they observed a significant

increase 1 h following mating. In contrast, when Carter et

al. (1985) measured plasma progesterone in female prairie

voles following mating they found no difference either 1 or

24 h following mating.

Experiment 5: The Effects of Cross-Fostering


The nature/nurture issue has always been one of the

most important and troublesome questions confronting

students of animal behavior (see Dewsbury, 1978, p. 154).

Although most contemporary workers recognize that behavioral

ontogeny is a complex mutually interdependent interaction

between genes and environment, the issues seem no more

resolved today than they were in the fourth century B. C.

when Aristotle discussed, in insightful and distinctly

sociobiological terms, his view of this very issue (see

Baumrin, 1975). This study was designed to determine

whether or not the species-typical difference in the

reinforcing value of social contact delineated in Experiment

3 is a result of experiential or genetic differences.

There have been numerous studies which have correlated

differences in the amount of social contact received by

infants and the contact proneness of adults. Happold (1976)

found a correlation between the amount of social contact

received by nestlings and the later contact proneness of

different species of rodents. Hartung and Dewsbury (1979)

compared the parental behavior of prairie voles and montane

voles and found that prairie voles displayed a greater

amount of maternal and paternal behavior than did montane

voles. Wilson (1982a) compared prairie and meadow voles in

the laboratory and found that prairie vole young received a

greater amount of parental care than did meadow vole young.

In two studies comparing the maternal and paternal behavior

of meadow voles, pine voles, and prairie voles in a

seminatural enclosure, McGuire and Novak (1984, 1986) found

that the monogamous pine and prairie vole parents displayed

more care than did the promiscuous meadow voles.

All of these studies demonstrate that the young of

contact prone species receive more parental care than do

young of less social, polygamous species. The above studies

demonstrate a correlation between patent-infant contact and

adult social behavior but not a causal relationship.

Nevertheless, the general consensus appears to be that early

social experience in fact causes the patterns and degree of

sociality displayed by adults (e.g., Scott, 1962). For

instance, Barnett (1963) suggested that the continuous

mother-young contact experienced by the infant rat was a

significant influence on huddling behavior in the adult rat.

Wilson (1982a) concluded that the species differences in

body contact experienced by prairie voles and meadow voles

"may be all that is necessary for a major species difference

in social structure in natural conditions, i.e., pairbonding

in prairie voles vs. independent nests and home ranges of

mated meadow voles" (p. 305). In addition, Reite and Short

(1986) assert that "Part of the explanation for the

differences in activity between bonnet and pigtail macaques

is reflected in the mother-infant relationship. Bonnets

exhibited a greater degree of distancing from the mother

throughout development" (p. 574).

Even when reared in the confines of the laboratory,

young of these species can still experience many important

differences in the nature of their parent-infant

interactions (e.g., Hartung & Dewsbury, 1979; Wilson, 1982a;

McGuire & Novak (1984, 1986). Variations in parental care

can translate into early variations in such things as

warmth, tactile and vestibular stimulation, amount of

nursing and milk etc., and these, in turn, can translate

into variations in adult social behavior.

It is the purpose of this experiment to test the

supposition that early differences in experience cause adult

differences in the propensity for social contact. One of

the most useful and successful methods of exploring the

nature/nurture issue is the technique of cross-fostering.

Newborn litters from two strains or species which differ

along one or more behavioral dimensions are switched shortly

after birth so that the mother of one species rears the

young of the other species and vice versa. This

experimental paradigm will be utilized in the following



Animals. Sixteen prairie vole and 16 montane vole

pups from a total of 5 litters of each species were

cross-fostered. The pups were taken from established

breeding pairs which were in the breeding colony for 1-6

months prior to the beginning of this study. The 5 prairie

vole litters were taken from 5 different breeding pairs; the

5 montane vole litters were from 4 different breeding pairs.

Animal care and maintenance was the same as that described

in Experiment 3.

Apparatus. The apparatus used in this experiment was

the same as that described in Experiment 3.

Procedure. Litters of pups were exchanged on either

day 1 or 2 postpartum. This procedure required that

simultaneous litters be born in both colonies within 1 or 2

days of each other. When this criterion was met, pups were

removed from their respective breeding cages and exchanged

between species. The litter with the smallest number of

pups was utilized as a standard and pups from the larger of

the two litters were sacrificed until the two litters were

of equivalent size. Thus equal numbers of pups were always


The pups remained with their foster parents until they

were 21 days of age. At that time they were placed with

their littermates in a clear polycarbonate cage measuring 48

x 27 x 13 cm where they remained until they were at least 60

days of age. The pups were then separated from their

littermates and individually housed in cages measuring 29 x

19 x 13 cm for at least 1 week before testing. Foster

montane vole females were always tested with foster montane

voles males which were from a different litter and no animal

was ever used in more than one test. The procedure for

testing was the same as that described in Experiment 3.


All 5 of the montane vole litters were successfully

reared to weaning at 21 days of age. None of the prairie

vole pups were alive at weaning. Seven days was the longest

a litter of prairie voles remained alive with their foster

montane vole parents. The prairie vole pups were not

immediately rejected by their foster parents. That is, all

fostered litters were observed to be nursing by the end of

the first day of the exchange. However, none of the litters

survived past day 7 with their foster parents. A total of

six tests were run with montane vole males paired with

montane vole females that were from different litters. The

number of tests run was limited by the number of male

montane voles available; of the 16 pups only 6 were males.

Therefore, 4 females remained untested. The mean huddling

time for the 6 cross-fostered pairs was 0.71 min (Figure 9).

A one-way ANOVA comparing the mean of the cross-fostered

group with the means from both groups run in Experiment 3

revealed that the cross-fostered montane voles did not

differ from the control montane voles and that both groups

differed from the control prairie voles (Scheffe comparison

of means, F(2,33) = 5.39, P < .001).


A major disappointment of this study was the failure to

rear cross-fostered prairie vole pups. The reason for the

failure is unknown. Laboratory rats will readily accept and

foster other strains of rats or even mice (Denenberg,

Paschke, Zarrow, & Rosenberg, 1969). Indeed, montane voles

successfully fostered pups from gray-tailed voles, Microtus

canicaudus, (McDonald & Forslund, 1978). However, these

two species are so close morphologically that they were

until recently considered one species; in the laboratory

they will interbreed with 10% of the young surviving to

weaning (McDonald & Forslund, 1978). In our laboratory we

have found that prairie and montane voles will intermate but

will not interbreed.

As mentioned above, it did not appear that montane vole

females rejected the pups outright as if they recognized

them as foreign. In all cases the prairie vole pups were

observed to be nursing within several hours of their

transfer. There are several possible explanations for this

phenomenon. First, it is known that these two species




O 2.5-

0 2

D 1.5-



Figure 9. Huddling duration for montane vole male/female
pairs: Baseline versus cross-fostered conditions.

display differences in the amount of time males and females

spend with their young (e.g., Hartung & Dewsbury, 1979; see

also Experiment 9 below). Perhaps prairie vole pups require

amounts of body contact or nursing which are in excess of

that required by montane vole young and these requirements

were not met by their foster parents.

A second, and more likely, possibility is that the pups

were, in fact, rejected by their foster parents but not

until sometime after the first few days. Of interest is the

fact that prairie vole pups and montane vole pups suckle

with markedly different intensities. That is, nursing

prairie voles attach to the nipple of the dam making it

difficult to remove them by hand without using extreme

force. This is not the case for montane vole pups which

easily slip off of the nipple when handled. This rather

striking species difference in neonatal behavior has not

been examined and thus nothing is known concerning either

its proximate mechanisms or adaptive significance. In any

case, it is conceivable that this propensity of prairie vole

pups might have so physically irritated montane vole dams

that they significantly reduced the time they spent nursing

or even refused to do so entirely.

Whatever the reason, only montane voles could be tested

for the effects of cross-fostering. The results of the

montane vole tests indicate that being reared by parents of

a contact-prone species has no effect on adult huddling

behavior, i.e., they displayed no differences from the

control group in Experiment 3. Although the behavior of

prairie vole foster parents was not quantified, daily

observations indicated that their maternal and paternal

behavior was normal throughout the 20-22 day period from

fostering to weaning. Thus montane vole pups likely

experienced higher than normal levels of body contact (i.e.,

sniffing, huddling, nursing, grooming, etc.) during their

sensitive postnatal period.

Developmental psychobiologists have convincingly

demonstrated that variations in these aspects of the

mother-infant interaction can have pronounced effects on

later behavior. For instance, it is well known that the

degree of social isolation which young animals experience is

directly correlated with the intraspecific aggression they

display as adults (e.g., Daly, 1973; Taylor, 1980).

Moreover, many of the specific mechanisms involved in

producing these effects have been delineated in numerous

carefully controlled laboratory studies. Mason and Berkson

(1975) found that the amount of vestibular stimulation

experienced by young rhesus monkeys during ontogeny plays an

important role in determining the level of aggression they

displayed as adults. Other factors that might affect adult

social behavior are such things as the activity levels of

the parents (Hudgens et al., 1967) and the amount of time

spent in direct physical contact with parents (e.g.,

Denenberg et al., 1969; Taylor, 1980). It has even been

suggested that endorphin levels in the milk may vary between

species and that this might produce differing degrees of

adult sociality (Hofer, 1981). Nevertheless, the huddling

behavior of fostered montane voles was unaffected by their


Perhaps, differences experienced in the field are of a

greater magnitude than those experienced in the laboratory.

Recall that, in the wild, prairie vole males and females

nest together and that animals from older litters have been

observed to huddle with the young of the first litter and to

retrieve those that wander out of the nest (Getz & Carter,

1980). This is to be contrasted with montane voles that

experience no paternal care, and are in many instances

abandoned by their mother at 15 days of age. Hofer (1981)

observes that for the mammalian newborn the mother is the

environment. Based on field data, prairie vole and montane

vole young must experience quite different post-natal

environments. In spite of these differences, the results of

the present study, although inconclusive, indicate that the

lack of social contact seen in montane vole adults is a

preprogrammed, species-typical trait.

Experiment 6: The Effects of Morphine and Naloxone


The intent of this section is to focus on the striking

difference between these two species in their propensities

for social contact. This predisposition, perhaps more than

any other, may determine the differences in social structure

which they display in the wild. It has been suggested above

that, even within the restricted context of the laboratory,

"inherent" differences in parental care could vary enough

between the species to cause, not just correlate with, this

difference in adult social behavior. The results of

Experiment 5 did not support this hypothesis. Rather, the

contrasting propensity for social contact and pairbonding

evident in these species appears to be an inherited rather

than an acquired trait. This is an important point because

it means that, as a result of differences in past selection

pressures, underlying neural mechanisms must also exist that

are, in turn, species-typical. The following experiment (as

well as the rest of this dissertation) is a first attempt to

explore the nature of these hypothesized neural differences.

There have been numerous attempts to explain the

phenomenon of social bonding from a behavioral perspective

(e.g., Ainsworth, 1967; Lamb, Thompson, Gardner, Charnov, &

Estes, 1984; Wickler, 1976) but only recently has "social

motivation" been studied as a discrete neural system in its

own right (Herman & Panksepp, 1977). One reason for this is

that well-studied behavioral systems, such as feeding or

sex, are thought to involve "primary" physiological

mechanisms (Pankesepp et al., 1980). Affiliative behavior,

on the other hand, is generally thought to involve secondary

mechanisms, i.e., to be an indirect function of the primary

motivational needs of an organism. According to this logic,

the reinforcing value of social contact with another

organism is an indirect result of that organism fulfilling a

primary need such as food or shelter.

Rejecting this conclusion, several workers have argued

that ample evidence now exists for considering social affect

to be a primary physiological system. Panksepp (e.g.,

Panksepp, Herman, Vilberg, & Bishop, 1980; Panksepp, 1982)

and MacLean (1986, 1985) have independently noted the

similarities between withdrawal from narcotic addiction and

the separation induced distress experienced by animals when

socially isolated. Both phenomena are characterized by

strong emotional attachments and both result in severe

distress when the objects of attachment are removed or

withdrawn. In addition, the behavioral and physiological

manifestations of separation distress are remarkably

consistent across a wide variety of species (Panksepp et

al., 1980). The two processes may thus involve a common

neural substrate--the endogenous opioid system.

If separation distress is a behavioral manifestation of

an endogenous endorphin withdrawal process whose "ultimate"

function is the mediation of social bonding, then systemic

injections of morphine agonists and antagonists, such as

naloxone, should selectively affect separation distress

and/or social attachment. As predicted, exogenous morphine

reduces and naloxone increases the separation induced

distress vocalizations (DV's) displayed by the young of a

wide variety of species when they are separated from their

mother (e.g., Benton, 1985; Newman, Murphey, & Harbaugh,

1982; Panksepp et al., 1980).

In the following experiment we analyze the effects of

systemic injections of morphine sulphate and naloxone

hydrochloride on huddling behavior in pairs of prairie and

montane voles. Naloxone hydrochloride, is a widely used

psychopharamcological tool for exploring endorphin/behavior

relationships (Herman & Panksepp, 1977). It is a highly

specific morphine antagonist which has no adverse effects

itself and binds exclusively with opiate receptors (Snyder,

1975). Based on results obtained in other species, we

hypothesize that morphine will reduce and naloxone will

increase huddling duration.


Animals. A total of 43 male and 43 female prairie

voles and 25 male and 25 female montane voles were used in

this experiment. None of the animals in this experiment had

been used previously in any other experiment.

Apparatus. The apparatus used in this experiment was

the same as that described in Experiement 3.

Procedure. The identical procedure followed in the

previous experiments of Section II was followed in the

present experiment with one major alteration. During the 5

min period when the females were removed from the test cage,

each animal was injected with either morphine sulphate (5 or

10 mg/kg), naloxone (5 or 10 mg/kg), or 0.9% saline. All

injections were given sub-cuntaneously in the nape of the

neck in a volume of 10 ml/kg body weight using 0.9% saline

as the drug vehicle (e.g., Herman & Panksepp, 1977; Panksepp

et al., 1980; Vickers & Patterson, 1983; Zagon & McLaughlin,

1985). The doses of drugs used were selected following

pilot studies in which doses of 1 and 2 mg/kg of morphine

sulphate or naloxone produced no effects on huddling in

either species.


Figure 10 shows the results of morphine sulphate and

naloxone hydrochloride on the huddling duration of prairie

vole pairs. A one-way ANOVA testing the effects of naloxone

in prairie voles revealed that neither the 5 mg/kg, the 10

mg/kg, nor the saline injected group displayed huddling

durations which differed from the baseline values obtained

in Experiment 3, F(3,32) = 0.23, p > .05. Morphine

sulphate, however, did produce an effect in prairie voles at

a dose of 10 mg/kg. At this dose, the mean huddling

duration of 5.98 min was significantly less than the other

three conditions, F(3,39) = 10.72, p < .001.

For montane voles a one-way ANOVA revealed no effects

of either drug. For naloxone, there were no significant

differences between the baseline condition, the saline

group, the 5 mg/kg, or the 10 mg/kg group, F(3,26) = 0.74,

p > .05. These same negative results were obtained for

morphine, F (3,26) = 0.73, p > 0.05.

The motor activity patterns of a subset of prairie vole

pairs that received the 10 mg/kg dose of morphine were

analyzed (Figure 11). In addition to huddling or sitting in

contact, the behavior of individual males and females was

allocated to one of two behavioral categories: 1)

stationary/isolate, and 2) locomotion (see Experiment 3 for

descriptions of these categories).

A two-way ANOVA (Species x Sex) with repeated measures

over the second factor was used to compare the 10 mg/kg

morphine sub-group with the baseline group (Experiment 3).

There were no significant differences between males and

females at either dose, within or between conditions.

However, there were significant differences between the

conditions in the amount of time spent stationary/isolate

between the baseline group (M = 9.67) vs. the 10 mg/kg

morphine group (M = 54.22), F(1,8) = 268.30, p < .001.

There was also a significant difference between the two

prairie vole groups in the amount of time they spent in

active locomotion (M = 14.75 baseline vs. 3.71 10 mg/kg),

F(l,8) = 6.73, p < .05. As there were no differences

between the sexes in either species, the combined average

activity measures for both males and females are plotted in

figure 11.


Morphine sulphate. The working hypothesis of this

experiment was that if endogenous opiates mediate social

bonding then exogenous activation of these systems should

lower the reinforcing value of social contact and thus

reduce huddling duration. Conversely, opioid blockade with

naloxone should simulate separation distress, lower the

reinforcing value normally obtained from social contact and,

thus, potentiate or increase the tendency to huddle. In

analogous tests, Herman and Panksepp (1977) and Panksepp, et

al. (1980) examined the effects of sub-cutaneous injections

of morphine sulphate and on measures of proximity in guinea

pigs and rats. They found that low doses of morphine (i.e.,

1 mg/kg and less) decreased the amount of time socially

housed animals of these two species spent in proximity to

each other. In the present study the only effects we

40 -4

30 -

20 -


5 mg/kqg


10 rmg/kg


Figure 10. Huddling in prairie voles as a function of the dose
of morphine sulphate and naloxone hydrochloride.



Figure 11. Activity levels in prairie voles following a
10 mg/kg injection of morphine sulphate.

observed were in prairie voles at the 10 mg/kg dose of

morphine sulphate. At this relatively high dose the time

males and females spent huddling was markedly reduced.

It is essential to determine whether the observed

reduction in huddling duration was the result of a reduced

"need" for social contact or a more general effect of

morphine sulphate on overall activity levels (Stolerman,

1985). An increase in general activity levels would

indirectly lower the duration of huddling while having no

effect on social motivation per se. The relationship

between exogenous morphine and activity levels is not clear

and there are differences both between species and doses.

For example, low to moderate doses of morphine sulphate of

less than 20 mg/kg produce excitation in mice, cats, and

rabbits, whereas dogs and rats show depression at the same

doses (Ayhan & Randrup, 1973).

If morphine sulphate were selectively affecting social

motivation one would expect no differences in overall

activity levels. Activity levels were evaluated in a subset

of animals in an attempt to distinguish between these two

alternative explanations. As Figure 8 shows, when prairie

voles and montane voles were compared in Experiment 3 there

were no differences in levels of locomotion but there were

differences in how the two species spent their stationary

time. It was concluded at that time that a species

difference in contact proneness, rather than activity

levels, was responsible for the results. On the other hand,

the present comparison (Figure 11) shows that when baseline

activity in prairie voles was compared to the morphine

sulphate 10 mg/kg group there were differences in both the

amount of time males and females spent in locomotion and

stationary/alone. In the 10 mg/kg group both the time spent

huddling and the time spent locomoting were reduced, while

time spent stationary/isolate increased. In other words,

there was a significant effect of this dose on general

activity levels.

It should not be concluded, however, that the observed

effects were solely the result of a reduction in activity

levels. Major non-opiate tranquilizers and behavioral

depressants such as sodium pentobarbitol depressed activity

levels but did not decrease distress vocalizations in

certain species (Herman & Panksepp, 1977). And in the

present experiment, although motor activity was

significantly reduced, the animals in the 10 mk/kg morphine

sulphate group did not appear to be generally sedated.

Indeed, although the animals were stationary, there appeared

to be an increase in the amount of self-grooming and bodily

movements in this group relative to the controls (personal

observation). Therefore, whether or not morphine at this

dose is affecting the dependent measure indirectly or

directly is still unclear. A more precise analysis of the

behavioral effects of morphine sulphate at this and at other

doses is necessary.

Panksepp et al. (1980) reported that doses of morphine

sulphate needed to reduce DV's are much higher in adult

animals than in young animals (i.e., 10 20 mg/kg vs. 0.5 -

2.5 mg/kg). Moreover, these same workers state that the

efficacy of morphine sulphate in reducing DV's in various

species is "attenuated" if animals had been isolated before

testing. It should be noted that in this study animals were

1) adults not neonates or juveniles and 2) had been socially

isolated for at least a week prior to testing. Both of

these factors could have rendered lower doses of morphine

sulphate (5 mg/kg and below) ineffectual in reducing


There were no effects of morphine sulphate at any

dose on huddling in montane voles. Hypothetically,

exogenous morphine should act to reduce the reinforcing

value of social contact by replacing the function normally

mediated by endorphins in this process (Herman & Panksepp,

1977). If these mechanisms are not normally functional in

montane voles, then the lack of an effect represents nothing

more than a basement effect, i.e., there is no huddling

duration to be lowered.

Naloxone hydorochloride. There were no effects of

naloxone in prairie voles. If a negatively reinforcing

withdrawal state is in fact mediating social attachment,

then the simulation of this process with exogenous naloxone

should have increased the huddling duration in prairie

voles. The results of this study do not support that

hypothesis. The lack of an increase in huddling duration in

the 10 mg/kg group might be attributed to a ceiling effect

biochemically. However, it seems likely that an induced

state of withdrawal that was not alleviated by social

contact would have significantly increased huddling duration

in this group.

Again, the lack of an effect in montane voles was not

surprising. Whatever the neural mechanisms which mediate

the reinforcing value of social contact in highly social

species appear to be absent in this species, thus nullifying

the effects of these pharmacological agents. Social

isolation has been shown to increase the number of opiate

receptors in the mouse brain (Bonnet, Miller, & Simon,

1976). Perhaps the number and/or distribution of opiate

receptors differs between these species. An understanding

of the naturally selected neural mechanisms which underlie

these behavioral differences between prairie and montane

voles remain to be determined.

The experimental paradigm followed in the present study

is admittedly weak in several respects. The lack of

behavioral specificity of morphine agonists is well

known--they are known to affect pain, aggression, feeding,

activity, reward systems, etc. (Steinberg & Sykes, 1985).

That morphine receptors were present in the nervous system

was virtually unknown until 1973 and much work remains to be

done on the relationship between this system and behavior.

Nevertheless, understanding the neural mechanisms that

underlie complex social/emotional behavior is of such

importance that methodological difficulties should not

preclude our utilizing any approach which, however

inconclusive, might shed some much needed light on this



The rationale behind Section IV of this dissertation

is that prairie and montane voles represent a unique model

system for elucidating general neurophysiological

principles underlying behavior (Ewert, 1984). A wealth of

field and laboratory data have now been compiled indicating

many striking differences in the social and reproductive

behavior between these closely related species. The focus

of comparative analyses will now shift from behavior to the

identification of neuroanatomical/neurochemical differences

that may be related to the behavioral differences which

these two species display. The strength of this research

strategy is that it is not limited to simply identifying a

brain region responsible for a particular behavior. It

also has the potential for determining exactly the specific

features of neuronal organization in a specific region

(e.g., receptor distribution, synaptic relationships, etc.)

associated with the behavioral function of that region.

The following three studies are the first in series of more

detailed experiments designed to identify brain/behavior

relationships in prairie and montane voles.

Experiment 7: Species Differences in Brain Size


When data on the relationship between brain weight was

first plotted against body weight for various mammalian

species, the slope of the line connecting the species'

averages was 2/3. The resulting "mouse to elephant" curve

has become the basis for one of the best established

allometric relationships--that between brain weight and body

weight (Clutton-Brock & Harvey, 1980). The accepted

explanation for this relationship, though never supported by

any empirical evidence (Szarski, 1980), is as follows. In

mammals, body surface area scales to the 2/3 power of body

weight and the brain must integrate information from surface

receptors and control surface effectors. Therefore, the size

of the brain should expand at the same rate as surface area

(Armstrong, 1982).

Recently, however, as the data base has expanded the

slope of the regression of brain weight versus body weight

has approached a value closer to 3/4. While an exponent of

2/3 suggested body size requirements, several workers have

attempted to explain the new value in terms of energetic

requirements or basal metabolic needs, (e.g., Armstrong,

1982; Harvey & Bennett, 1983; Hofman, 1983; Martin, 1981).

This issue is still much debated and it now appears certain

that no single interaction underlies this relationship. As

Mann, Glickman, & Towe (1986) conclude, small animals have

large brains and large animals have large brains, but it is

not clear why this is so.

Arguing against the existence of a simple physiological

relationship underlying brain size differences is the

variability in the brain sizes of closely related species of

equivalent body size (Mann et al., 1986). With body size

and phylogeny ruled out, differences in brain size are most

likely caused by ecological factors (Harvey, Clutton-Brock,

& Mace, 1980; Hofman, 1983). Indeed, recent work has

correlated variation in brain size with such factors as

habitat (e.g., Mace, Harvey, & Clutton-Brock, 1981), mating

system (e.g., Gittleman, 1986), foraging strategy

(Clutton-Brock & Harvey, 1980; Eisenberg & Wilson, 1981;

Pirlot & Stephan, 1970), home-range size (Clutton-Brock &

Harvey, 1980) and certain life history variables associated

with K-selection (Eisenberg, 1975; Eisenberg & Wilson, 1981;

Mace & Eisenberg, 1982).

To compare the brain sizes of two species one must

first eliminate any differences due to body size, i.e., one

must compare relative brain sizes. This is normally done

utilizing an allometric equation of the following form:

log y = a (log x) + log b

This equation plots a straight line relating brain weight

(y) and body weight (x) with a as the intercept and b as the

slope of the curve. As mentioned above, values for various

taxa were plotted and standard values were established for a

and b. The amount of departure from this line displayed

by a species' average is taken as a measure of relative

brain size or encephalization (Jerison, 1973). For

comparative purposes, an encephalization quotient (EQ) is

determined by dividing a species' brain weight by the brain

weight predicted using the above equation.

There are many problems inherent in this method (e.g.,

Gould, 1977; Lande, 1979; Mace et al., 1981; Szarski, 1980).

One of the major difficulties in this calculation is that

the relative brain size for species of one taxonomic group

are based on data from another perhaps distantly related

group. This is important because it is now known that the

slope of the curve varies with the taxonomic level (Mace et

al., 1981). Although brain size increases to the 3/4 power

of body weight across mammals as a whole, as one looks at

lower taxonomic levels the exponent becomes more shallow.

Thus, while the standard mouse to elephant curve has an

exponent of 3/4, intrageneric comparisons have slopes of 0.2

- 0.4 (Lande, 1979), and when one compares adult members of

the same species there may be no relationship at all between

brain weight and body weight (Szarski, 1980). One reason

for this is that the body weight of adult mammals varies

with such factors as state of health, and amount of body

fat--all factors which do not affect brain weight (Sholl,

1948; Szarski, 1980).

In the following study we compare the brain sizes of

two closely related species of voles which differ in their

ecology and their behavior. Comparing the relative brain

size of closely related species like prairie and montane

voles alleviates many of the problems associated with

brain/body relationships at different taxonomic levels

(Mace & Eisenberg, 1982). Moreover, since much behavioral,

ecological, and life history data exist for these two

species functional explanations may be possible for any

differences that might be found to exist.


Animals. Eight male and 9 female prairie voles and

13 male and 11 female montane voles were used in this

study. All animals were laboratory bred and between 90 and

120 days of age at the time of testing. Animals were

selected for this study following their use as test

subjects in Experiments 3, 4, and 5. They were arbitrarily

chosen from the pool of animals completing the above


Procedure. Animals were anesthetized with ether

and perfused intracardially with 0.9% saline solution and

10% formalin. Brains were surgically removed from the

skull and immediately weighed to the nearest 0.001 g.


Figure 12 represents a scatterplot of the values for

the 24 montane voles and 17 prairie voles. Brain weight

and body weight were not correlated for either montane

voles (r = .13) or for prairie voles (r = .17). A

two-way ANOVA (Species x Sex) revealed no difference in

body weight between prairie voles (M = 39.45 g) and

montane voles (M = 42.37 g), F (1,37) = 1.66, p > .05.

There was, however, a significant difference in body weight

between the sexes with males being heavier than females, F

(1,37) = 14.32, p < .05. This sex difference was more

pronounced for montane voles (M = 43.45 g males vs. M =

35.45 g females) than for prairie voles (M = 43.53 g

males vs. M = 41.2 g females).

The average prairie vole brain was 12% heavier than

the average montane vole brain (M = 0.566 g prairie voles

vs. M = 0.494 g montane voles), F(l,37) = 48.62, p <

.001. There was no difference in brain weight between the

sexes in either species, F(1,37) = 1.08, p > .05.

Least squares regression lines were plotted for log

brain weight on log body weight for each species. The

regression equation for prairie voles is of the following


y = 0.05 (x) 0.526

and for montane voles:

y = 0.04 (x) 0.37

A test of slope revealed that neither slope differed

significantly from 0: for prairie voles, t(15) = 0.69,

p > .05; for montane voles t(22) = 0.64, p > .05. These

results are consistent with the above findings that there

was no relationship within either species between body

weight and brain weight.


For both prairie voles and montane voles brain weight

varied independently of body weight and body weight did not

differ between the species. There was, however, a highly

significant species difference in brain weight. The

knowledge that two species differ in relative brain size

tells us nothing about which areas of the brain are

enlarged relative to others. However, a relative size

difference is of biological importance in its own right as

it allows for the assessment of those ecological and life

history variables that correlated with it (Clutton-Brock &

Harvey, 1980; Eisenberg, 1981).

Table 2 summarizes several ecological and behavioral

adaptations which have been correlated with variability in

brain size in various mammalian species.

1.) Food distribution-- As Mace et al. (1981) point

out, there is a consistent correlation between species

which forage for food that is distributed unevenly in space

or time and large relative brain size (e.g., granivorous

and insectivorous small mammals (Mace et al., 1981),

frugivorous bats (Eisenberg & Wilson, 1978), and primates

(Clutton-Brock & Harvey, 1980). Differences in the

foraging strategies for the above-mentioned species all


+ + 4- 4-

+ + + a +

+ C+I

0 0o


-0.14 -
-0.16 -

-0.2 -


-0.24 -



-0.3 -

-0.32 -


-0.38 -


-0.42 -

-0.44 -

Figure 12. Log brain weight versus log body weight for prairie
and montane voles.



Table 2

Ecological and life-history correlates
of brain size differences

Small Brain Size

Evenly disperesed
food source

Small home-range

Low species diversity
and competition

Passive antipredator

Specialization in
one or two sensory

Locomotion in
two dimensions

Low level of social

Maximum preprogramming
of information

r-selected traits

Large Brain Size

Rich but patchy
food source

Large home-range

High species diversity

Active antipredator

Multimodal sensory input
visual system highly

Locomotion in
three dimensions

Complex social organization

Information storage and
retrieval based on
individual experience

K-selected traits

Modified From Eisenberg (1982, p. 98).

involve fundamental differences in food type. Granivorous

mammals must be more selective in what they eat and may

utilize more complex behavioral patterns than folivorous

species (Mace et al., 1981). In contrast to this, both

prairie voles and montane voles feed predominantly on the

same food source, herbaceous grasses and sedges (Getz,

1985). Moreover, each of the two species occurs in a

variety of habitats from dense to sparsely covered and from

xeric to wet. It is difficult, therefore, to imagine a

difference in food distribution selecting for brain size

differences in these two species.

2) Home-range size--Although home range size was

positively correlated to brain size in certain primate

groups, Clutton-Brock and Harvey (1980) suggest that this

correlation actually reflects the complexity of food

distribution. As mentioned above, the dietary adaptations

are so similar for these two species it is difficult to

conceive of selection favoring a larger brain size based on

a difference in home-range size even if this were an

established statistic.

3) Competitor diversity-- Mace and Eisenberg (1982)

note that larger brain size is correlated with competitor

rich habitats in which efficient niche exploitation would

be favored by selection. M. ochrogaster which is sympatric

with M. pennsylvanicus also competes with two other

herbivorous small mammals, (Synaptomys cooper and

Sigmodon hispidus). Montane voles must also experience

equivalent degrees of competition as they overlap with six

congeneric species in the Northern Cascades (Rose & Birney,

1985). As with the previous two categories, this one must

also be considered relatively unimportant as a serious

selection pressure.

4) Antipredator Strategy--Passive antipredator

strategies (e.g., rolling into a ball) are generally

correlated with small brains whereas active strategies

(e.g., flight or counterattack) are generally correlated

with large brain size (Eisneberg, 1981). Deemphasizing any

strong selection for antipredator adaptations in this group

of rodents, Pearson (1985) suggests that most of the

adaptations of Microtus sprcies seem to equip them for life

in the grasslands rather than for defense against

predators. In fact, their main antipredator strategy may be

a high rate of reproduction (Pearson, 1985). Thus, as with

competition, it is difficult to see how differences in

predation pressures could have selected for differences in

brain size.

5) Sensory modalities--Based on all that is known of

their ecology, it is probably safe to assume similar if not

identical development of sensory systems in voles. Pearson

(1985) considers Microtus in general to have less developed

sensory systems than even most other small mammals

occupying similar habitats.

6) Locomotion in two vs. three dimensions--Complex

locomotor patterns, as in aquatic or terrestrial species,

are generally correlated with larger brains. Again, it

would be difficult to categorize these species as being

different in this regard. Both species occupy similar

niche dimensions and appear primarily to make extensive use

of surface runway and tunnel systems (e.g., Getz, 1985;

Pearson, 1985).

7) Social organization and mating systems--In certain

orders of primates polygynous species had significantly

larger brain sizes than those that were monogamous

(Clutton-Brock & Harvey, 1980). Gittleman (1986), in a

study comparing different families within the order

Carnivora, reported no differences in relative brain size

between monogamous and polygynous species or between

different types of parental care systems. However, he did

find that single-male species have smaller relative brain

sizes than do multi-male species. This would support the

proposition that more complex social systems such as

multi-male groups (in which there would also be strong

sperm competition) select for larger brain size. If this

is true, than in primates large brain size should be

positively correlated with large testes size, not the

reverse as one might imagine.

In fact, Dewsbury et al. (in preparation) report that

prairie voles do have larger testes than do montane voles.

However, it would be difficult to categorize the mating

system of prairie voles as more complex than that of

montane voles.

8) Information storage ability--The functional

criteria used by Eisenberg (1981) to distinguish those

species which are more flexible behaviorally from those

which are less adaptable to changing environmental

circumstances were probably not meant to distinguish

species as closely related as montane and prairie voles.

However, in the laboratory, prairie voles have displayed a

certain degree of sensory discrimination based on previous

experience which has not been shown by montane voles tested

under identical conditions (e.g., Colvin, 1973b; Ferguson,

Fuentes, Sawrey, & Dewsbury, 1986; Shapiro & Dewsbury,

1986; Shapiro et al., 1986). Not displaying a preference is

not necessarily evidence that an individual animal does not

discriminate. Nevertheless, the above findings suggest the

possibility that prairie voles possess a higher level of

sensory complexity than do montane voles.

9) r and K strategists--In the terminology of growth

equations the symbol "r" represents the intrinsic rate of

increase of a population and the symbol "K" represents the

carrying capacity or equilibrium density of the population.

An r-strategist is an opportunistic species displaying

rapid population growth, large litter sizes (with their

concomitant genetic diversity) short life spans, low

parental care, rapid development, etc. A K-strategist, on

the other hand, is a species displaying slow growth rates,

small clutch or litter sizes, longer life spans, and higher

parental investment in young, etc. (see Pianka, 1970). An

r-selected species is adapted for and tends to predominate

in the colonization of early successional, or unstable,

uncrowded environments, whereas K-selected species are

better adapted and predominate during later stages of

ecological succession where densities have reached

equilibrium and where there is intense interspecific

competition (MacArthur & Wilson, 1967).

Of particular relevance to the present discussion is

the fact that Eisenberg (1975) and Eisenberg and Wilson

(1981) in comparative studies of didelphid marsupials have

found that K-selected species have larger brains relative

to r-selected species of similar body size. Moreover,

Nadeau (1985) suggested that within the genus Microtus

there may also have been enough divergence in ontogenetic

and reproductive strategies to warrant classifying

different species along an r-K continuum. Nadeau (1985)

compared prairie voles and montane voles in prenatal

mortality, neonatal weight, litter size, and rate of

postnatal development. The only species difference he

reported was relative to litter size with prairie voles

classified as K-selected and montane voles as r-selected.

When we computed the weighted averages of litter sizes from

Keller (1985, p.754-756) prairie voles had a mean of 3.68

pups whereas montane voles had a mean of 5.1 pups, which is

consistent with Nadeau's classification. As Mace and

Eisenberg (1982) note, litter size may not be a reliable

measure of reproductive rate. A better measure, mean

annual production of young (Mace & Eisneberg, 1982), is not

presently available for prairie and montane voles.

Many of the behavioral and life-history differences

between prairie and montane voles parallel those displayed

by pine and meadow voles. In the field, meadow voles

display a polygamous mating system whereas pine voles

appear to be monogamous (Madison, 1980a; McGuire & Novak,

1984). Meadow vole young are also weaned much earlier than

pine voles (13 vs. 21 days), a situation which mirrors that

in prairie versus montane voles. Prohazka et al. (1986)

recently reported that meadow voles displayed a pattern of

rapid neuromuscular development relative to pine voles. In

their discussion of this finding, Prohazka et al. relate

this contrast to an overall difference between meadow and

pine voles on the r-K continuum.

Of interest to the present research, is the fact that

Prohazka et al. (1986) also report mean brain and body

weights for their experimental groups (p. 529). Consistent

with the present findings, there was considerable sexual

dimprophism in body weight for meadow voles but not pine

voles whereas there was no sexual dimorphism in either

species for brain weight. An examination of their data

revealed that r-selected meadow voles had an average brain

weight which was roughly 28% higher than that of K-selected

pine voles. Recall that, in the present study, K-selected

prairie voles had a 12% higher average brain weight than

r-selected montane voles. In the present research, however,

there was no difference in body weight to confound the

comparison. In the study of Prohazka et al. (1986) meadow

voles were roughly 46% heavier than pine voles. Therefore,

it is not possible to accurately determine relative brain

size differences between these two species from their data.

While it is tempting to compute brain weight/body weight

ratios, it is first necessary to determine the relationship

between body weight and brain weight within this genus.

Interspecific exponents relating log brain weight to log

body weight are generally in the range of 0.66 (Szarski,

1980). However, Mace and Eisenberg (1982) calculated a

relatively high value of 0.71 for species within the genus

Peromyscus. Until such information is available for

Microtus, it will not be possible to determine whether or

not these correlations hold.