REPRODUCTIVE BIOLOGY OF A TROPICAL PROCYONID, THE WHITE-NOSED COATI
GERALD ALLEN BINCZIK
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I tarried in producing this document, and as a result two of the people I most wanted to share it
with did not live to see its completion. I offer my dissertation in their memory...
My mother, Rosemary Binczik
Professor John F. Eisenberg
Front and center in my thanks are of course my advisor, Dr. Louis J. Guillette, Jr., the rest
of my academic committee, Dr. Joel H. Brendemuhl, Dr. H. Jane Brockmann, Dr. Colin A.
Chapman and Dr. Melvin E. Sunquist, and Dr. John F. Eisenberg, who was unfortunately unable
to complete his service on my committee. Having always previously worked independently
(some might have instead said I was hard-headed), I learned as much as anything else in the
course of my Ph.D. program to make better use of the expertise and life knowledge I am offered,
and I am grateful to them for all that they shared (and tried to share) with me. Thanks go out, too,
to Dr. Rebecca T. Kimball for helping out on short notice at the end of my program.
The Lincoln Park Zoo Scott Neotropic Fund provided essential funding and the Florida
Museum of Natural History and University of Florida lent necessary equipment and laboratory
access. The field work in Tikal National Park, Guatemala, was permitted by the Consejo
Nacional de Areas Protegidas (CONAP) and the Instituto de Antropologfa e Historia. I am
particularly grateful to Lic. Oscar Lara of CONAP for his efforts over and above the call of his
office. Lorena Calvo and Carlota Monroy were also important allies in getting through the
bureaucratic tangles associated with starting a field project from scratch in a remote part of a
My time in the field was made much more fun by regular assistants Prudencio Tobar,
Fermin Lima and Don Luis Savala. The Universidad de San Carlos kindly loaned the very
capable Carla Ramirez as a six-month student intern. During nesting periods Benjamin Gonzalez
and Jos6 Tobar climbed trees that I as a big, clumsy gringo could not. Jen Boyd, Jason Brueck,
Aaron King, Bob Schuyler and Josh Vinlove were all crazy enough to volunteer for various
periods of time, as well. All of these people deserve much of the credit for the fieldwork's
successful completion. Dr. David Whitacre kindly provided housing to a number of people who
assisted in fieldwork, and he, Mario Jol6n and Haroldo Garcia just as importantly served as
sympathetic ears and apt problem-solvers in helping me to overcome the various obstacles that
inevitably arise in a lengthy field project. Dr. Denise Gross similarly helped by teaching me a
simple and inexpensive method for making blow-darting equipment, a skill which I am sure the
family dog was glad I did not possess in boyhood but which proved essential in the field. Maria
Jos6 Gonzilez was thoughtful enough to drop what she was doing in order to bring me a freshly
road-killed coati, in the process turning what would have been just another senseless animal death
on the national park road into a learning experience and scientific contribution.
Barbara Booth, Randall Harmon and Dan lisser not only provided much-needed visits
from home, but also took upon themselves the not-so-small task of bringing needed equipment
and supplies when I could not leave the field. The occasional postcards and letters that made it
through to me from my family, Rosemary, Anne Marie, David, Joan, Jason and Elizabeth
Binczik, also helped keep me going.
Mae Chisholm, Satish Degala, Jenny Gates, Patricia A. Lewis, Deidre Meiggs, Dr. Don
A. Samuelson, Doris Sartain and Dr. Samuel K. Wasser all provided essential assistance to my
efforts in the laboratory. Of special note in this regard is also Dr. Edward F. Orlando, a fellow
graduate student at the time who refreshed my memory on the ins and outs of steroid
radioimmunoassay, and much more importantly helped keep me sane during some tough times by
offering his friendship and collaboration; both were and are most welcome. Dr. Timothy A.
Pearce kindly applied his malacological expertise to snail identifications.
Last and of course most, my darling wife, Dr. Susan D. Booth-Binczik, not only stood by
me through much more than would the typical devoted spouse, but was also the best part of why
and how this study was completed. I can never fully repay my debt to her nor even fully express
my thanks, but I look forward to trying! I also look forward to finally putting the adjective
"professional" alongside the "collaborators" we have long been.
TABLE OF CONTENTS
ACKNOW LEDGEM ENTS ..................................................................................................... iii
ABSTRACT ............................................... ........................................................................... vi
1 BACKGROUND ................................................................................................................... 1
2 REPRODUCTIVE SEASONALITY ......................................... ................ ......................... 4
Introduction ................................................................................................................... 4
M ethods......................................... ................................................................................ 8
Results ........ ........................................................................................................... ........ 15
Discussion .................................................................................................................. 38
3 BIRTH SYNCHRONY ....................................................................................................... 51
Introduction ..................................................................................................................... 51
M ethods........................................................................................................................... 53
Results ............................................................................................................................. 56
Discussion ................................................................................................................ 58
4 RUT.......................................................................................................................................... 64
Introduction ..................................................................................................................... 64
M ethods........................................................................................................................... 67
Results ............................................................................................................................. 74
Discussion ....................................................................................................................... 81
5 SUM M ARY ............................................................................................................................. 96
APPENDIX RAW TESTIS VALUES ................................................................................. 102
BIOGRAPHICAL SKETCH ................................................................................................. 120
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
REPRODUCTIVE BIOLOGY OF A TROPICAL PROCYONID, THE WHITE-NOSED COATI
Gerald Allen Binczik
Chair: Louis J. Guillette, Jr.
Major Department: Zoology
Some 80% of the world's mammals reside in the tropics, but few-and almost no long-
lived-species there have been subjects of reproductive research. This study contributes detailed
information on the reproduction of a long-lived procyonid carnivore, the white-nosed coati
(Nasua narica), in Tikal National Park, Guatemala.
Whereas most tropical mammals display broad reproductive seasonality (e.g., producing
offspring within a six-month period) or reproduce year-round, the coati exhibits extraordinarily
tight timing in reproductive events. In Tikal mating took place within about two weeks in the
middle of the dry season, births occurred in a comparably short period at that season's end, and
young emerged from nests and began foraging alongside adults early in the wet season. When
the timing of parturition was examined in more detail, the interquartile range (i.e., middle 50%)
for females from three social groups spanned only five and six days in consecutive years, and the
mean birthdates in those years differed by only six days. The male coati has responded to reliable
clustering of female receptivity by evolving a rut more akin to that of many ungulate species than
to the pattern of any other carnivore species yet examined.
Coati reproduction is timed such that the young are weaned over an extended period
coincident with the season of greatest food availability. Leaf litter invertebrates and fruits figure
prominently in the coati diet, but the former are evidently most important. Notable among these
are insects, and in Tikal especially the scarab beetle Enema endymion; adults and larvae of this
one species accounted for 8.6% of all invertebrates consumed by coatis over the course of this
study. Communal care and predator swamping were explored as alternative explanations for the
coati's remarkable reproductive pattern, but these hypotheses were rejected.
Exactly how the coati achieves such tight reproductive seasonality remains unknown, but
a prediction based on the hypothesis that sociality plays a role was upheld. Social cues
exchanged among associates are presumably used by the coati to augment the weak
environmental cues (e.g., photoperiod) upon which seasonally reproducing tropical species must
Although the white-nosed coati (Nasua narica) is a wide-ranging, locally abundant and
readily habituated diurnal carnivore, the most detailed studies to date have been of a single island
population. The majority of what is known and believed about coati ecology and reproduction is
based upon extensive behavioral observations by Kaufmann (1962), Russell (1979) and most
recently Gompper (1994) of a small group of animals on Barro Colorado Island (BCI), Panama.
The species is primarily insectivorous/frugivorous (Kaufmann, 1962), with some small
vertebrates taken opportunistically. Adult females are highly social, living with their immature
offspring in cohesive bands most of the year (except for a brief nesting period encompassing birth
and early lactation), and adult males are chiefly solitary (Kaufmann, 1962; Booth-Binczik, 2001).
Russell (1982; 1983) reviewed various hypotheses put forth to explain female sociality, and
asserted that it functions to protect juveniles from predation, most notably by cannibalistic adult
males (see also Russell, 1981), and ectoparasitism. Gompper (1996), however, concluded that
group living is a way for females to increase their foraging success on patchy resources, whereas
several studies at other sites have provided evidence that group living in coatis is an anti-predator
adaptation (Burger and Gochfeld, 1992; Booth-Binczik, 2001; Hass and Valenzuela, 2002).
The coati has received considerable attention for its unusual social system, but it is even
more notable among tropical mammals for its remarkable degree of reproductive seasonality.
Kaufmann (1962) and Russell (1982) addressed the timing of coati reproduction on BCI. The
species displays a strongly seasonal pattern at this locale, with apparently only one finely timed
birth period per year. The proportion of adult females reproducing can vary dramatically from
year to year; they are capable of delaying their first reproduction (which usually occurs at two
years of age) by one or two years, perhaps due to food availability, but generally reproduce every
year thereafter. Males mature later than females, and most probably do not reproduce until one or
two years after their first opportunity. It is unclear whether reproductive senescence occurs, but
Kaufmann (1962) observed one and Russell (1982) two older females (estimated to be at least
seven years of age) that did not produce young. Animals have lived to be more than 17 years old
in captivity; Poglayen-Neuwall, 1990.
A few reports (e.g., Gilbert, 1973; Hass, 2002 for Arizona; Valenzuela and Ceballos,
2000 for Jalisco, Mexico) suggest that coati reproduction may be patterned similarly at other
locations as well, but these have not been confirmed by thorough study. Moreover, even the
abundant behavioral/spatial ecological data for BCI fall short of satisfying the lack of basic
knowledge about coati reproductive biology. Many questions remain: Does the portrait
described for BCI apply elsewhere in the species' range? Where it is seasonal, is this largely
tropical mammal pursuing an obligate or facultative (i.e., opportunistic) strategy? What
evolutionary pressures drive seasonality, and what environmental cues are used in timing it?
This study addresses these questions by extending beyond prior research in four
important ways. It is the first intensive study of reproductive patterns in any coati population,
focusing on animals at a site that is distant from but ecologically similar to BCI. It is also the first
to examine male reproductive biology at any level. A much greater level of detail is presented on
certain environmental stresses confronting the coati with an eye toward resolving specific issues
left ambiguous by previous workers. Perhaps most important, this research provides the first
information on the reproductive physiology of the species.
The focus of the study is the free-ranging coatis in Tikal National Park, Guatemala. The
Park protects approximately 600 km2 of tropical moist forest in the seasonally dry "Mayan" forest
of northeastern Pet6n, and constitutes the centerpiece of Central America's largest remaining tract
of lowland rainforest (Nations et al., 1988). Unlike BCI (Willis, 1974; Glanz, 1982; Robinson,
1999) where ecological conditions could be considered somewhat artificial (Willis, 1974;
Eisenberg et al., 1979; Terborgh and Winter, 1980), Tikal retains a full complement of large
predatory species. The disagreement among studies regarding the adaptive value of coati group
living provides an example of the risks of extrapolating from data obtained on BCI to the species
as a whole. Studies at sites such as Tikal should be more useful for identifying selective
pressures important in coati evolution.
Mammals display two basic patterns in the timing of their reproduction (Bronson, 1989):
opportunism, in which they reproduce as often as conditions permit, and obligate seasonality, in
which one or more predictable environmental factors in combination with other life history
aspects compel them to follow a seasonal reproductive schedule. Evolutionarily, these are better
thought of as tendencies than rigid choices, and a single species may employ both strategies under
different conditions (e.g., the purple-faced langur, Presbytis senex; Rudran, 1973).
Obligate seasonality appears to be the strategy adopted by most long-lived mammals (i.e.,
those with lifespans of more than one year) inhabiting seasonal environments, especially species
with a substantial investment per offspring, specialized diet or extreme seasonal survival
mechanism such as migration or hibernation (Bronson, 1989). Ovarian function in such species is
regulated by the appearance of reliable proximate cues-especially photoperiodic changes-thereby
setting a schedule for sexual receptivity, ovulation and/or embryonic development. The most
environmentally sensitive phase of the species' reproductive pattern is timed to coincide with the
most favorable conditions for offspring survival (Lancaster and Lee, 1965; Sadleir, 1969).
Ultimate factors postulated to account for mammalian reproductive seasonality
(summarized in Sadleir, 1969; Bronson, 1989; Di Bitetti and Janson, 2000) include food
availability (both in terms of energy and essential nutrients), rainfall, ambient temperature,
predation pressure and competition. Of these, food availability (most notably energy intake) is
widely believed to be the strongest determinant of seasonality; other factors may modify a pattern
but are considered unlikely to override it (Bronson and Heideman, 1994).
There is an obvious logic to seasonal reproductive patterns among mammals at temperate
latitudes, where annual fluctuations in insolation profoundly affect temperature and the
availability of free water, factors critical at the base of essentially all food chains. But strong
seasonal weather patterns, particularly in rainfall, can also promote reproductive seasonality in
the tropics (Bronson, 1985). Tropical rainfall patterns have pronounced effects on the abundance
of all manner of terrestrial and arboreal food items, and climatic fluctuations could conceivably
also affect competition, predation and/or parasitism pressures on a tropical species.
Unfortunately, despite the fact that some 80% of all mammalian species reside in the tropics,
relatively few species and very few locations in the lower latitudes have thus far been examined
(O'Brien, 1993; Bronson and Heideman, 1994; Dubost et al., 2005).
In this respect, the lowlands of the Panama Canal Zone may have received more attention
than any other tropical site. In his review of the literature, Fleming (1973) found that half of the
species studied at this locality (23 of 45) were known to be seasonal breeders, with a strong
tendency toward birth or weaning occurring at the beginning of the wet season. This pattern is
exemplified by the region's best-studied carnivore, the white-nosed coati (Nasua narica).
Observations by Kaufmann (1962) and Russell (1979; 1982) indicate that coati
reproduction on Barro Colorado Island (BCI), Panama, is strongly seasonal: mating activity there
appears to take place early in the four-month dry season, litters are born late in that same season,
and young emerge from the nest and are gradually weaned beginning early in the eight-month wet
season. Scattered observations suggest reproductive events may also be seasonal at other sites
(e.g., southern Arizona: Gilbert, 1973; Hass and Roback, 2000; Costa Rica: Sienz, 1994).
Fruit and leaf litter invertebrates are the principal foods of coatis on BCI (Kaufmann,
1962; Russell, 1982; Gompper, 1996) and elsewhere (Delibes et al., 1989; Sienz, 1994). Smythe
(1970) examined fruit-fall on BCI in correlation with Kaufmann's (1962) data on coati life
history stages and concluded that the period encompassing parturition and first emergence of
young from their nests matches the time of greatest fruit supply. Russell (1982) compared coati
intake rates of fruit and animals, and concluded that reproduction is timed to allow juveniles to
begin foraging at the onset of a long-lasting peak in the availability of invertebrate prey; he
believed the concurrent availability of a major fruit crop was "perhaps fortuitous" (p. 428).
This discrepancy stems from the fact that on BCI the seasonal peak in size and abundance
of forest floor arthropods (at the onset of the wet season; Levings and Windsor, 1982) overlaps
substantially with one of two annual peaks in fruit availability (Foster, 1982b). Fruit availability
in neotropical forests often peaks at the onset of the wet season (van Schaik et al., 1993; Sakai,
2001). As arthropod abundance in tropical forests commonly varies positively with moisture
levels (Levings and Windsor, 1984; Burgess et al., 1999), such overlap is likely widespread.
Tikal National Park, Guatemala, is similar to BCI in that it is characterized by fairly
uniform temperatures and strongly seasonal rainfall (Fig. 2-1); there is a marked four-month dry
season from January through April and a six-month wet season from June through November,
with May and December being transitional periods. In contrast to BCI, however, the forests of
northern Guatemala are believed to have been significantly altered by the ancient Maya (Lundell,
1937; Puleston, 1982), whose horticultural practices likely favored species which produce
plentiful, nutritious fruit and tend to do so at what would otherwise be times of scarcity. For
example, the highly nutritious Brosimum alicastrum (ram6n, a.k.a. breadnut), a tree which has
been reported to undergo three heavy, evenly-spaced fruit sets per year (February-March, June-
July and October-November; Puleston, 1968; in Coelho et al., 1976), is unusually dense
throughout the region (Lundell, 1937; Puleston, 1982).
Given the above, it was anticipated that coati reproduction in Tikal would be highly
seasonal as it appears to be elsewhere, but that Mayan efforts there might have resulted in
sufficient temporal separation between peaks in fruit and invertebrate abundance to aid in
determining the relative importance of these foods to the timing of coati reproduction. Even if
peak overlap was found to occur in Tikal, it was believed that a more rigorous exploration of
Dry Season Wet Season
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 2-1. Mean ( SEM) monthly rainfall and maximum (filled squares) and minimum
(open circles) temperature in Tikal National Park, Guatemala, January 1984 through
April 1998. Number of years contributing to each mean ranges from 10 to 13 depending
on availability of complete month's data from the Instituto de Sismologia, Vulcanologia,
Meteorologia e Hidrologia (Guatemala City, Guatemala).
coati foraging habits accompanied by consideration of nutritional matters might clarify this issue.
The purpose of the present study was therefore twofold:
1. Coati reproductive phenological events were monitored year-round at Tikal to
document-and provide greater detail on-the species' reproductive seasonality at this new tropical
2. The abundance and quality of potential coati food items and also coati foraging
behavior and body condition were monitored year-round to clarify the relationship between the
species' reproductive pattern and food availability.
Free-ranging white-nosed coatis were studied from June 1994 into November 1996 over
an area of approximately 20 km2 at the center of Tikal National Park (17N, 89W). The Park
protects nearly 600 km2 of forest that has at various times been identified as subtropical moist or
tropical semi-deciduous (Schulze and Whitacre, 1999; based on the classification schemes of
Holdridge et al., 1971, and Pennington and Sarukhan, 1968, respectively), tropical dry (Walker
and Cant, 1977; based on Holdridge, 1957), and quasi-rainforest (Lundell, 1937). Total annual
rainfall averages 1285 80 mm (based on the eight complete years of data between 1984 and
1998 available from the Instituto de Sismologia, Vulcanologfa, Meteorologia e Hidrologia,
Guatemala City, Guatemala). Located in the "Mayan" (i.e., ancient Maya-modified; J.F.
Eisenberg, pers. comm.) forest of northeastern Peten, Tikal constitutes the centerpiece of Central
America's largest remaining tract of lowland rainforest and retains a full complement of the
region's wildlife species (Nations et al., 1988).
Rainfall in the park during the study was recorded at a uniform time daily via a
government-operated weather station. From these data monthly totals were calculated for
comparison to various phenological data.
Fifty-five adult female coatis resident to four bands (females of this species live in groups
with their maturing young except to nest, whereas males are largely solitary; Kaufmann, 1962)
were utilized in this study. Animals were initially captured by live-trap or (more often) blowgun
and chemically immobilized by intramuscular injection of approximately 7 mg/kg Telazol (Fort
Dodge Laboratories Inc., Fort Dodge, IA). Anesthetized animals were tattooed and eartagged for
identification, weighed, visually examined to determine general physical condition and measured
as described below. Forty-three of the coatis were additionally fitted with 90 g motion-sensitive
radiocollars (Advanced Telemetry Systems, Isanti, MN) to enable subsequent location and
recapture, and to confirm death if such occurred. Individual females were recaptured by blowgun
(never more often than monthly) for resampling; the average number of immobilizations per
female was 2.6 + 0.2. All animal handling procedures were conducted as per protocol #4084
approved by the University of Florida's Institutional Animal Care and Use Committee.
In order to monitor reproductive state, size indices for the vulva and right anteriormost
teat were obtained during each female's immobilization. For each body part, the longest
dimension (L), the widest perpendicular dimension (W) and the highest dimension above the
surrounding body surface (H) were measured to the nearest 0.1 mm via calipers and then
combined in the formula V= L x W x H to produce an approximation of volume. One person was
responsible for taking these measurements throughout the study. Indications of pregnancy and
lactation were also noted. (Originally serum and fecal reproductive steroids were to be monitored
as well, but samples obtained for this purpose were rendered useless by a catastrophic freezer
failure after fieldwork was completed and before radioimmunoassays could be conducted.)
As a measure of body condition, an index of each coati's body fat level was also
determined at the time of immobilization. Following Hossler et al. (1994), a fold of the animal's
skin (i.e., excluding muscle) was pinched at the back of each rear leg midway between knee and
hip, and calipers were used to measure the thickness of this fold to the nearest 0.1 mm.
Measurements for the left and right thighs were averaged to produce the index reported for that
coati/immobilization. One person took all such measurements throughout the study.
The animals in three coati bands were habituated to the close presence of investigators,
who simply visited the bands frequently over several weeks and briefly baited the coatis at the
beginning of each encounter until the animals allowed themselves to be accompanied (without
additional feeding) for several hours thereafter. By the end of the habituation period it was
possible to follow animals so closely that at times investigators had to be careful not to kick or
step on them, and sometimes the coatis even had to be gently shooed a short distance away to
allow investigators sufficient room to process immobilized animals during captures.
Thereafter, the reproductive activities of habituated females were monitored ad libitum
(Altmann, 1974) one to three times per week (as often as their bands were visited) during the
period of detailed behavioral observations (see below) or as close to daily as possible during
periods of particular interest. Data on six reproductive phenomena were recorded.
Copulation was either directly observed or presumed based on the presence of mating
wounds (see Results for description of latter).
Pregnancy was generally identified by the midpoint of gestation, when an increase in a
female's girth became visually apparent.
Nesting was defined as beginning when a female departed from its band prior to giving
birth, and ending when the infants first accompanied the female down from the nest.
Parturition was determined to have occurred when a female showed a sudden
pronounced reduction in girth followed by continued nesting behavior, and was ultimately
confirmed by observation of young still in the nest or shortly after their emergence from the nest.
Reaggregation was operationally defined as the rejoining of a postpartum female and its
young with at least three other such females and their young.
Nursing was either directly observed or inferred from evidence of lactation noted during
a female's immobilization.
During nesting, 20 nest trees were climbed and nests examined to verify the presence and
count the number of young. Ages at which young were examined varied from approximately one
to five weeks after birth, depending on when nests were initially located and how soon thereafter
opportunities occurred to examine the nests without the females nearby.
Systematic observations of female coati foraging behavior were conducted from August
1995 through September 1996. Observation sessions were scheduled two to three times per
week, and in each case began as soon after first light as the target animals could be located
(typically by 08:30) and continued for several (typically four) hours thereafter. All-day (up to 10
hr) observation sessions were conducted once each month to confirm that activity patterns did not
change such that mornings were no longer appropriate for monitoring normal foraging behavior.
In each session, one 5-minute block of focal animal sampling (Altmann, 1974) was conducted
each quarter hour, rotating through focal animals opportunistically selected for that session;
during the nesting period each session instead focused on a single female (reducing sample sizes
during this period). In the selection of target animals for systematic behavioral observations,
priority was placed on distributing observation effort across females without respect to their band
membership, such that no individual coati served as a focal animal in more than one block per
hour (except in the nesting period), nor in more than one session per week.
During each sampling block, the amounts of time spent by the focal animal foraging in,
under and away from potentially relevant fruiting trees were recorded. Also recorded were the
type and number of food items consumed in each of these three areas, as well as the time spent
consuming the items; foraging time and time of food consumption were mutually exclusive
categories. The identities of animal prey were determined to the lowest taxonomic levels
possible, and fruits eaten were identified to species.
As bases for calculating monthly rates of foraging success, the three foraging areas were
used thus: The leaf litter invertebrate zone was the combination of under and away from fruiting
trees. The fruit zone was the combination of in and under fruiting trees. A given female's
monthly foraging success rates were then calculated as both number of items consumed per
minute foraging and number of seconds spent consuming the items per minute foraging, using
monthly totals for each zone. To eliminate evident outliers caused by undersampling of specific
individuals in specific months, females with fewer than 300 seconds spent foraging in a leaf litter
invertebrate zone or 20 seconds in a fruit zone in a given month were not included in analyses;
the monthly rates for the remaining females were averaged to produce the reported values.
Leaf litter and associated invertebrates were sampled on approximately the 5", 15" and
25h of each month from September 1995 through September 1996. On each collection date 10
samples were obtained. Each sample was collected by placing a bottomless tub, the area of which
was 1 m2, on the forest floor and quickly scooping the litter and invertebrates therein into a bag;
the soil thus cleared was then dug to a depth of three to four inches (as the substrate allowed) and
any additional animals found thereby were added to the bag. Exact sample locations and times
were determined by accompanying a foraging band of coatis for two to four morning hours and
placing the tub at the front edge of the band at random intervals (skipping any preselected times
at which the animals were not foraging).
Each sample bag was weighed fresh and then hand-sifted to harvest visible invertebrates..
Animals thus obtained were identified to the lowest practical taxonomic level and weighed to the
nearest 0.1 g; animals individually weighing less than 0.1 g were grouped by type and the total
weight for the group was used to calculate an average weight per individual. Voucher specimens
were occasionally preserved to aid species identifications, but otherwise invertebrates were
bagged by sample date and then frozen in liquid nitrogen and stored in an ultracold freezer for
subsequent nutritional analyses (see below). Finally, each sample's leaf litter was dried in the sun
for a minimum of two days (or more, as weather dictated) and again weighed to obtain dry litter
mass and litter moisture content. Monthly means were calculated for total invertebrate mass,
litter mass and litter moisture content. The individual contributions of various invertebrate
classes to monthly totals were examined by calculating their monthly numbers and masses on a
Fruiting tree phenology was monitored from June 1995 through September 1996. Ten 10
m x 50 m plots were established at random intervals along 1 km of (and also at random directions
and distances up to 50 m from) each of three forest paths traversing the entire home range of one
coati band as well as parts of the ranges of three other bands. In each plot, all trees measuring at
least 10 cm diameter at breast height at the onset of data collection were tagged and identified to
species. Every month, each plot was randomly assigned to a group of 10 which was then
surveyed on approximately the 5h, 15th or 25th of that month. During surveys, tagged plants were
checked for the presence and percentage of immature and ripe fruit, and when fruit was present in
a tree it was quantified via the visual count method (Chapman et al., 1992). Samples of ripe
fruits were also collected to aid species identifications and frozen/stored as described for
invertebrates to provide material for later nutritional analyses (see below). One person was
responsible for determining fruit counts and percent ripe fruit in almost all months; a second
person trained under and performed calibration runs with the first person repeatedly throughout
the study before replacing the first person in the final month.
Using the above methodology, data were obtained on many more tree species than
appeared in the coati diet. Only those species which were ultimately seen being eaten by coatis
and/or appeared in their scat on at least 15 separate dates were selected for analysis and
presentation in this paper. The percentage of trees with fruit and the mean number of fruits and
ripe fruits per tree were calculated on a monthly basis for each of these species.
After field work was completed, proximate nutritional analyses were conducted on
monthly (i.e., mixed) leaf litter invertebrate collections and also on individual invertebrate and
fruit species deemed of particular relevance to the coati diet (as described above). Each sample,
consisting of either whole invertebrates or the edible portions of fruits, was first thawed, weighed,
dried for two days at 600C, allowed to equilibrate to room temperature overnight before being
weighed again, then finely ground and homogenized by use of a Wiley Mini-Mill equipped with a
20-mesh Monel screen (both Thomas Scientific, Swedesboro, NJ) prior to additional processing.
Crude fat and energy content were determined in nutrition laboratories of the University
of Florida's Animal Science Department following procedures outlined in Helrich (1990). To
obtain crude fat, samples were weighed, dried 3 hr at 1020C, equilibrated to room temperature
under desiccation and then extracted four times in ethyl ether using a Soxhlet apparatus.
Extracted samples were subsequently dried 12 hr at 1020C before being equilibrated under
desiccation and weighed once more; the amount of fat was calculated as the difference in mass
between pre- and post-extraction samples. Energy content was obtained by bomb calorimetry
using a 1261 Isoperol Calorimeter (Parr Instrument Co., Moline, IL) after samples had been
weighed into crucibles, dried at 650C for 24 hr, equilibrated under desiccation and weighed again.
Crude protein, ash-free neutral detergent fiber (NDFaf) and ash content were determined
in the Forage Evaluation Support Laboratory of the University of Florida's Agronomy
Department. Samples were first weighed, dried for 15 hr at 1050C and weighed again, and then
organic matter was determined by ashing for at least 4 hr at 5000C. Total nitrogen (N), and from
it crude protein (as N x 6.25), was obtained by a modification of the standard Kjeldahl procedure
(Helrich, 1990). Samples were digested using a modification of the aluminum block digestion
procedure by Gallaher et al. (1975); sample weight was 0.25 g, the catalyst used was 1.5 g of 9:1
K2SO4:CuSO4, and digestion was conducted for at least 4 hr at 375C using 6 ml of H2SO4 and 2
ml H202. Digestate nitrogen was then determined by semiautomated colorimetry (Hambleton,
1977) via a Technicon Autoanalyzer (Technicon Instruments Corp., Tarrytown, NY). Neutral
detergent fiber (NDF) was determined by the procedure of Golding et al. (1985), consisting of
boiling 1 g of sample in a neutral detergent solution for I hr, filtering, extracting with acetone and
then drying; NDFaf was obtained by correcting NDF for the amount of ash in the sample.
The moisture content of an initial sample was calculated by combining the results of the
relevant drying steps outlined above. Crude protein, crude fat, NDFaf, ash and energy content
were calculated on a dry matter basis. (It must be noted that the invertebrate and fruit samples
were subjected to the same storage freezer failure as were frozen serum and fecal samples,
although unlike the latter samples they appeared to take little or no harm from it. This topic is
addressed further in the Discussion.)
Sokal and Rohlf (1981) guided all statistical efforts, and SPSS 8.0 (SPSS, Inc., Chicago,
IL) aided computations. Correlational analyses were conducted to examine the relationships
between environmental variables, resource availability and foraging behavior. Pearson
correlations (r) were tested where scatterplots of the data appeared consistent with bivariate
normality, and Spearman rank correlations (p) were instead tested where this assumption
appeared to be violated or where samples were deemed too small for proper visual evaluation.
Because multiple comparisons were involved and there are numerous arguments against the more
elaborate methods that are often employed to adjust for such comparisons (Moran, 2003), a
conservative P, < 0.01 was adopted as the level of significance.
Ordinarily each female underwent only one complete reproductive cycle (i.e., succeeding
at least to the point of rejoining a band with recently born young in tow) each year, but rarely a
female underwent a second cycle if pregnancy, nesting or early emergence was unsuccessful.
Accordingly and unless otherwise indicated, the following data pertain only to females' lone/first
Part or all of 11 copulatory events were observed; three of these (in two coati bands)
occurred from 13 to 16 February 1995, and eight (in four bands) from 14 to 28 February 1996.
Another 40 females (in four bands) were presumed to have copulated from 13 to 27 February
1996 based on the appearance of fresh mating wounds (see below).
In each of the three copulatory events observed in entirety, the female initiated mating by
leaving its band foraging on the ground, approaching one of the males perched in the tree canopy
above the foraging band (see Booth-Binczik et al., 2004, for a detailed description of the coati
mating system in Tikal) and standing beside the male with hindquarters oriented toward the male
and tail bent upward and to the side. The male briefly sniffed at and then mounted the female,
clasped its forelegs around the female's midbody and presumably intromitted intromissionn could
not be visually confirmed). In all observed copulatory events, coitus included bouts of rapid,
quivery thrusting alternating with bouts of apparent stillness by the male, accompanied
throughout by soft grunting or panting. It often appeared as if the female was attempting to leave
during the longer moments of copulatory stillness but was held fast by the male. Typically the
female ultimately broke free from the grip of the male's claws during such a bout-leaving long,
deep parallel scratches on the sides of the female's body in the process-then turned on the male
with loud squealing vocalizations and feinted or actual bites and paw swipes (events occurred too
rapidly to confirm the extent of physical contact) before fleeing down or even falling from the
tree and racing back to its band. On one occasion the female was instead simply released by the
male, walked away and then lay down and rested on the tree limb about a meter from where the
male remained perched. Complete copulations lasted 16, 19 and 48 minutes from intromission
until separation, although it should be noted that the first of these was concluded by the male in
coitus being attacked by another male (Booth-Binczik et al., 2004). The eight copulations
already underway when first observed continued for 2-29 minutes thereafter.
Eleven pregnant females (from two bands) were confirmed to have established nests in
1995, and 28 (from four bands) in 1996. In 1995, the first of these females that was observed to
depart its band to begin nesting did so on 12 April and the last did so on 24 April. In 1996 the
onset of nesting among females under observation ranged from 12 April to 30 April.
In almost every case a nest was occupied by a solitary female, but on three occasions it
appeared that pairs of females shared a nest at least temporarily. Nests were usually situated in
large, living trees, either in their crotches or in tangles of lianas hanging free from or lying against
their trunks, at a measured height of 9.8-30.2 m above the ground. They were constructed
primarily of interwoven slender, leafy branches and vines bitten off of living plants. Nests varied
in size and shape; most often they were approximately 0.5-0.7 m-diameter hollow spheres with
one or two entrances just large enough to accommodate a female, but occasionally they were
much more simply constructed open bowls/platforms up to about 1.5 m in diameter. In any event
the nest cavities/floors were lined with approximately 5-20 cm-deep carpets of small loose leaves,
tiny branches and bromeliad tufts into which the infants burrowed.
Females often stayed out of sight in their nests for a few to several days surrounding
parturition, making precise dates of birth seldom identifiable and in turn likely resulting in an
overestimate of birth period duration. Bearing that in mind, the earliest and latest that parturition
could have occurred among nine nesting females (from two bands) in 1995 were 16 April and 29
April, respectively, and 23 females (from two bands) gave birth in 1996 from 22 April to 9 May.
Despite the fact that a substantial effort was made to pinpoint the timing of reproductive events,
copulation and parturition could both be narrowed down to reasonably tight spans of time for
only one female; it gave birth 71 to 74 days after mating. For 18 litters examined while still in
the nest, average litter size was 4.0 0.3.
Late in the nesting period it was common for a female to move its young from one nest to
another, but the infants were not considered to have emerged (and the female to have concluded
nesting) until they followed their dam down to the ground and began foraging; thereafter the
female was rarely seen on foraging excursions without its young alongside. Eight females (from
two bands) brought their young down from the nest 27 May to 4 June 1995, and 15 (from four
bands) did so from 23 May to 19 June 1996. The process of reaggregration began so soon after
emergence that it was possible on only three occasions to determine with certainty how many
young individual females emerged with; these females (the nests of which had not been
examined) brought down four, five and six young.
Complete band reintegration was in some instances a lengthy process even though
females with infants reaggregated into small groups rapidly after nesting. The earliest and latest
points at which any of 14 females (from two bands) reaggregated in 1995 were 5 June and 29
June, respectively, and 20 females (from two bands) reaggregated from 31 May to 19 June 1996.
Nursing was almost never observed, presumably because it most often occurred in
various tree canopy roosts used by the coatis for taking brief siestas during the day and sleeping
at night. Incidences of nursing on the ground (in brush piles) were seen on 4 August 1995 and 27
July 1996. On 13 September 1996 a female was observed evading attempts by an infant to
suckle, and twice on 27 September 1996 (in a single band) infants were observed investigating
but not suckling from females' teats, suggesting that nursing concluded in the middle or last half
of this month.
Three females (in two bands) in 1995 and one in 1996 underwent interrupted
reproductive cycles followed by second, successful cycles; one more female in 1995 was believed
to be in the midst of a second cycle when initially encountered, and another female in 1995 died
while nesting in a second cycle. It was never possible to identify the specific causes) of
reproductive failure, but information on the point at which it was evident may be instructive.
Females returned to their bands (presumably soon) after reproductive failure, and dates on which
this phenomenon was observed were 30 April 1995 (i.e., at the end of the observed parturition
period) for three females, 23 May 1996 (late in the nesting period), and on or shortly after 5 June
1995 (soon after reaggregating with several other females; the exact date could not be identified
because it quickly became unclear which young belonged to which females).
Although much less information was obtained on second cycles, they appeared to
proceed similarly to lone/first cycles. Late mating was not observed. Pregnant females departed
their bands to nest from 30 July to 18 August 1995 and by 7 August 1996. Parturition could be
narrowly attributed for only one female, on or shortly before 12 August 1996; this animal's litter
contained four young when it was examined while still in the nest on 5 September. Females
concluded nesting and rejoined their bands from 3 September to 6 October 1995 and by 20
September 1996. The 1996 female brought to its band all four offspring that had been found in
the nest; three other second-cycling females brought one, one and two young to their bands.
Nursing by females undergoing second cycles was observed on 13 November 1995 and 24
September and 8 November 1996.
Reproductive morphometric changes exhibited by females throughout the study (Fig. 2-2)
were revealing of physiological events underlying the coati reproductive cycle despite the
unfortunate lack of hormonal data (see Methods). Sudden, pronounced vulval swelling-
presumably an estrogenic effect-was observed in immobilized females from 7 February to 19
February, i.e., beginning shortly before the mating period and ending about midway through that
period; a much smaller secondary increase in vulva size occurred in late March to early April,
within a couple of weeks before parturition and roughly coincident with the beginning of a
prolonged increase in teat size. Teat size-presumably reflecting estrogen, progesterone and
prolactin levels in turn-rose fairly rapidly from March through the parturition period to a peak in
late May to early June, then declined more slowly until it again reached baseline around the end
of September, about the same time that lactation/nursing apparently concluded. (Note: The
greatest teat size observed in the study actually occurred on 5 September, but this data point was
unique in that it belonged to a female that was in the midst of a second reproductive cycle and
therefore out of step with its cohorts.)
The amount of leaf litter varied seasonally in strong positive correlation to rainfall (r[i =
0.81, P = 0.001), as did the overall mass of litter invertebrates (Table 2-1), falling from highs at
the onset of sample collection in September 1995 to lows in February and March 1996 before
rising again to moderately elevated levels that essentially held steady from June 1996 until the
conclusion of sample collection in September 1996 (Fig. 2-3). Leaf litter moisture content (Fig.
2-3) varied less from month to month, and although it too resembled the rainfall pattern in some
ways (e.g., rising gradually from a low reached in March 1996), the correlation between these
was not significant (r[I ] = 0.41, P = 0.17). Rainfall over the course of sample collection (Fig. 2-
3) resembled the typical year (Fig. 2-1) except that rains were somewhat heavier than usual in the
middle of the 1995 wet season and unusually light in the middle of the 1996 wet season.
Dry Season Wet Season
1 B -
> 0 rM Eiln g &^iEl a l in 1:1DD ^ifi 1111l ilTTT1
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
| ~ Teat
0 'F'FZ' Ti1 "
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 2-2. Annual vulva and teat size profiles exhibited by adult female white-nosed
coatis in Tikal National Park, Guatemala, July 1994 through October 1996. Mating
period (13 Feb 28 Feb) indicated by lightly shaded background and period of parturition
(16 Apr 9 May) more darkly shaded. Based on measurements obtained during 139
immobilizations of 55 animals.
Wet Season Dry Season Wet Season
Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep
Figure 2-3. Monthly total rainfall and mean ( SEM) leaf litter characteristics in Tikal
National Park, Guatemala, September 1995 through September 1996. Thirty 1-m2
quadrats sampled each month.
Table 2-1. Correlational statistics for relationships between monthly leaf litter invertebrate data
and rainfall, dry litter weight and litter moisture content in Tikal National Park, Guatemala.
Thirteen consecutive months compared except where otherwise indicated; significant findings
indicated in bold.
Invertebrates Rainfall Litter Weight Litter Moisture
r = 0.91, P< 0.001
r = 0.82, P = 0.001
r = 0.55, P= 0.05
Snails and Slugs
P = 0.08
P = 0.05
P = 0.008
r= 0.52, P = 0.07
r= 0.14, P=0.64
r= 0.65, P= 0.02
r= 0.36, P= 0.23
r = -0.38, P= 0.20
r= 0.47, P=0.10
P = 0.94
P = 0.75
r= 0.57, P=0.04
r= 0.12, P=0.68
P = 0.01
P < 0.001
P = 0.22
P = 0.26
P = 0.09
P = 0.23
r = 0.66, P = 0.01
r= 0.06, P= 0.83
r= 0.33, P= 0.28
r= 0.12, P = 0.70
r = -0.08, P = 0.80
r= 0.81, P= 0.001
P = 0.66
P = 0.92
r= 0.59, P=0.03
r= 0.17, P=0.58
r = 0.82,
P < 0.001
P = 0.001
r = -0.05,
r = -0.03,
P = 0.29
P = 0.87
P = 0.93
r= 0.20, P= 0.50
r= 0.36, P= 0.22
r= 0.64, P=0.02
r= 0.65, P = 0.02
r=-0.25, P = 0.41
r= 0.12, P = 0.69
P = 0.20
P = 0.45
r= 0.42, P=0.16
r= 0.23, P = 0.46
r = 0.25,
P = 0.25
P = 0.42
a based on comparison of 14 consecutive months
Leaf litter invertebrate groups varied considerably from one another in their temporal
patterns of number and mass (with spiders and insects even differing within their own groups), as
reflected in their contributions to monthly totals of these measures of abundance (Figs. 2-4 to 2-
6). Examination of the relationships between invertebrate class numbers and masses (not
proportionate numbers and masses) and environmental variables (Table 2-1) reveals that insects
were principally responsible for the correlations of overall invertebrate abundance with rainfall
and litter mass.
Insects (Fig. 2-6) dominated the leaf litter fauna in overall abundance, but there were a
few monthly exceptions: Earthworms (Fig. 2-4), the second most prevalent group overall, were
slightly more numerous in October and December 1995 and weighed more in July and August
1996. Millipedes (Fig. 2-5) were slightly more numerous in February 1996. Snails and slugs
(Fig. 2-4) were most prevalent by weight in May 1996. Spiders (Fig. 2-6) were always fairly
abundant but never contributed much weight to samples, and were never predominant in either
regard. Centipedes (Fig. 2-5) were generally scarce and contributed negligible weight even when
modestly abundant in March 1996.
Within Insecta (Fig. 2-6), beetles and their grubs were notable in that they were always
among the most abundant orders and contributed substantially more weight than any other order
in every month except April 1996, when orthopterans dominated. The contributions of a single
large scarab beetle species, Enema endymion, figured most prominently. The adults and larvae of
this species were so common in the vicinity of the park that they were known by unique names
(ronrones and gallinas ciegas, respectively) by the local people, and were at certain times of the
year extraordinarily abundant atop or in the leaf litter or at shallow depths in the soil beneath it.
By weight, adults constituted 34.2% of the June 1996 invertebrate sample, and larvae amounted
to 60.4% in September, 69.9% in October and 51.3% in November 1995 samples, and 6.3% in
August and 22.0% in September 1996 samples. Adults appeared in the coati diet (more on this
below) or scat or were noted to be abundant in general observations made aside from leaf litter
invertebrate sample collections from 8 June to 16 July 1994, 15 June to 4 July 1995 and 12 May
to 17 July 1996. Larvae were similarly recorded from 17 August 1995 to 13 January 1996 and 17
July to 8 November 1996 (when field work concluded).
Six species of fruiting trees (Fig. 2-7) were selected for detailed examination based on the
coati's reliance upon them over the course of the 16-month survey of tree phenology as well as on
I I ~
So C .)
1 5 ]0
0 0 0 c
I _______ I ______ I _____4
u C x r H-77.
s~o \0 \o Q-
co '0 a
-& I rw
^ I-' g
J s^ (d
) ^> 0
^j > o
the species' abundance in forest plots. Aspects of some species' fruiting patterns appeared
seasonal as noted below, but neither of the measures of fruit production (mean number of fruits
per tree and percent of trees in fruit, which mostly but not in all ways varied together) was more
than marginally correlated with rainfall for any species (Table 2-2). Ripe fruit was not observed
in large enough numbers (presumably because coatis and other frugivores were eating it as it
became available) to warrant depicting it in species' graphs.
B. alicastrum fruit (Fig. 2-7) was present and usually abundant in virtually every month
of the survey, with a brief peak in August 1995 and a broad peak centered around February 1996,
i.e., in the middle of those years' wet and dry seasons, respectively. Ripe fruit was recorded
March-September 1996, i.e., from the last half of that year's dry season to the middle of the wet
season. B. alicastrum was seen being eaten by coatis and/or was found in their feces on 94 dates
between 7 March and 30 September 1996, demonstrating significant correlation with ripe fruit
availability (Table 2-2). (It also appeared in the coati diet outside of the survey period three times
from 30 December 1994 to 27 January 1995, and on 3 October and 8 November 1996.)
Manilkara zapota (chico zapote, a.k.a. sapodilla) fruit (Fig. 2-7) was also present and
abundant much of the year, and displayed peaks timed similarly to B. alicastrum. There were
lulls in fruit production in June and July, i.e., early in the wet season, in both 1995 and 1996.
Ripe fruit was recorded only in May 1996, coincident with the normal onset of rains, but
distinguishing ripe from unripe fruit was particularly difficult in this species due to its lack of a
pronounced color change; possibly only the most mature fruit specimens were recognized as ripe.
This likely explains why the correlation between ripe M. zapota availability and the appearance
of the fruit in the coati diet did not achieve significance (Table 2-2). This species was recorded
eaten/defecated by coatis on 52 dates between 11 January and 27 June 1996 (and also 16 dates
between 7 January and 3 May 1995).
Fruit production in Trichilia moschata (cedrillo rosa) (Fig. 2-7) was observed only
during the last four months of the survey, June to September 1996, beginning early in the wet
season and increasing as the season progressed and the survey concluded. Ripe fruit was
recorded only in the last month, September 1996. In contrast, T. moschata appeared in the coati
diet in two bouts: 14 times from 11 April to 2 July 1996, and 10 times from 22 August to 29
September 1996 (and also on 3 and 10 October 1996)-indicating that a survey sample size greater
than the 19 trees available would have been preferred for this species. Accordingly, fruit
consumption was not significantly correlated with ripe fruit availability (Table 2-2).
Blomia prisca (tzol) fruited (Fig. 2-7) twice during the survey: for an indeterminate
period including the first two wet season months examined, June and July 1995, and in April and
May 1996 when rains resumed after that year's dry season. Fruits were very low in number
during the latter of these periods, but ripe fruit was recorded in every month of fruit production
except April 1996, correlating significantly with consumption by coatis (Table 2-2). B. prisca
appeared in the coati diet six times from 2 June to 6 July 1995 and 10 times from 22 April to 18
May 1996 (and also six times from 23 April to 31 May 1995, and on 7 and 8 November 1996).
Pouteria reticulata (zapotillo hojafina) fruit (Fig. 2-7) was present for five months
beginning with the onset of rains in April 1996. Ripe fruit was noted in the last two months of
fruit production, July and August 1996, i.e., toward the middle of the wet season. This species
was recorded in the coati diet on 23 dates from 13 June to 30 July 1996. The correlation between
ripe fruit availability and fruit consumption was marginal (Table 2-2).
Pimenta dioica (pimienta, a.k.a. allspice) fruited (Fig. 2-7) in the same wet season
months as T. moschata, June to September 1996. Unlike the latter species, however, fruit
production in P. dioica peaked early in its appearance and was waning as the survey concluded.
Ripe fruit was recorded in the last two months of the survey, August and September 1996. Fruit
was similarly recorded eaten/defecated by coatis 20 times from 24 August to 29 September 1996
(and also five times from 1 to 10 October 1996), correlating significantly with ripe fruit
availability (Table 2-2).
Brosimum alicastrum (N = 69)
Manilkara zapota (N = 26)
Trichilia moschata (N = 19)
Jun Aug Oct Dec Feb Apr Jun Aug
Jul Sep Nov Jan Mar May Jul Sep
Figure 2-7. Fruiting phenology of and reliance upon tree species important to white-
nosed coatis in Tikal National Park, Guatemala, June 1995 through September 1996.
Dates fruit seen eaten and/or in feces indicated by Xs. Number of trees surveyed each
month (N) indicated parenthetically after species name.
Blomiaprisca (N= 109)
Pouteria reticulata (N = 121)
Pimenta dioica (N= 17)
Jun Aug Oct Dec Feb Apr Jun Aug
Jul Sep Nov Jan Mar May Jul Sep
Figure 2-7 (Continued). Fruiting tree phenology and coati fruit consumption.
Table 2-2. Correlational statistics for relationships between monthly fruit availability and rainfall
and between monthly ripe fruit availability and fruit consumption by white-nosed coatis in Tikal
National Park, Guatemala. Sixteen consecutive months compared; significant findings indicated
Fruiting Tree Species
number of fruits / ripe fruits per tree
percent of trees in fruit
number of fruits / ripe fruits per tree
percent of trees in fruit
number of fruits / ripe fruits per tree
percent of trees in fruit
number of fruits / ripe fruits per tree
percent of trees in fruit
number of fruits / ripe fruits per tree
percent of trees in fruit
number of fruits / ripe fruits per tree
percent of trees in fruit
p= -0.49, P = 0.05
p = -0.53, P = 0.03
p= -0.40, P= 0.12
p= -0.54, P = 0.03
p= -0.01, P = 0.96
p = -0.03, P= 0.92
p= 0.12, P = 0.65
p= 0.14, P= 0.60
p = 0.26, P = 0.34
p= 0.32, P = 0.23
p= 0.09, P = 0.74
p= 0.03, P= 0.91
p = 0.94, P < 0.001
p = 0.48, P = 0.06
p = 0.48, P = 0.06
p= 0.78, P < 0.001
p = 0.50, P = 0.05
p= 1.00, P < 0.001
Considering coati food item nutrient composition (Table 2-3), a few general patterns
Among monthly leaf litter invertebrate samples (Table 2-3), protein, fat and energy
content per gram were all highest from September to November 1995, in the latter half of that
year's wet season and during and immediately after the months of highest rainfall recorded in the
sample collection period. All three nutritional measures then dropped to low levels during the
transition into the following dry season in December 1995, rose briefly and modestly as the dry
season progressed, then declined again as rains resumed in April 1996 leading into the next wet
season. All again began rising from August to September 1996, albeit at lower levels overall than
Table 2-3. Proximate nutrient composition of white-nosed coati dietary items in Tikal National
Park, Guatemala. Nutrients other than water reported on a dry matter basis.
Sample Water Crude Crude Ash-Free Ash Energy
(%) Protein Fat (%) Neutral (%) (kcal/g)
Leaf Litter Invertebrates
August 1995 71.7 35.9 11.2 12.0 7.7 3.33
September 1995 78.3 50.4 19.1 18.6 2.1 4.61
October 1995 75.9 58.6 24.6 17.6 0.5 5.61
November 1995 77.5 61.6 14.9 19.9 1.5 5.02
December 1995 72.4 38.3 11.1 9.6 10.2 3.13
January 1996 74.4 35.3 11.2 11.2 2.5 3.37
February 1996 72.6 40.9 8.7 23.3 3.5 3.52
March 1996 70.5 39.8 12.3 11.9 1.5 3.58
April 1996 64.6 35.0 10.2 15.9 14.5 3.02
May 1996 70.0 35.9 7.1 13.3 12.7 2.89
June 1996 66.3 34.3 8.6 14.6 12.4 2.61
July 1996 74.3 33.1 6.3 10.6 12.5 2.67
August 1996 70.8 30.3 5.4 2.42
September 1996 67.4 40.2 6.6 16.3 10.7 2.57
larva 88.7 62.3 12.1 30.5 0.5 4.83
adult 62.5 67.5 20.3 51.5 4.1 6.03
Brosimum alicastrum 83.1 11.6 1.8 4.03
Manilkara zapota 80.0 4.0 5.9 61.8 2 4.80
Trichilia moschata 77.1 12.0 25.2 5.27
Pouteria reticulata 86.3 9.7 4.8 4.37
Pimenta dioica 74.6 5.1 2.2 31.6 1.1 4.07
in the previous year. Only fat, however, was significantly correlated with rainfall (Table 2-1), the
other nutrients being at most marginally so. The contributions of E. endymion beetles and
especially grubs (Table 2-3) figured as prominently in the above pattern as they did in leaf litter
invertebrate samples. Adults provided the greatest amount of protein and energy per gram of any
food item analyzed, and the third greatest amount of fat; larvae were particularly high in protein
and energy and less so in fat, and had the highest moisture content of all food items.
No sample of B. prisca fruit was analyzed, but of the other five important fruit species
(Table 2-3), T. moschata was highest in protein, energy and especially fat content; indeed, gram-
for-gram this species achieved the highest fat level of any food item analyzed. The results for
other fruit species were variable except that P. dioica was at or near the bottom of the group in
All of the invertebrate samples contained higher protein levels than any of the fruit
samples (bearing in mind that the latter consisted only of edible portions, therefore excluding
seeds) (Table 2-3). With the aforementioned notable exception of T. moschata, invertebrates also
typically provided more fat per gram. Fruits equaled or slightly bested the mixed leaf litter
invertebrate samples in terms of energy content, and usually held more moisture as well.
In terms of foraging effort, coatis spent more-generally much more-time foraging in the
leaf litter invertebrate zone than in the fruit zone at all times of year (Fig. 2-8). As noted above,
ripe fruit was essentially unavailable for the five months of September 1995-January 1996, i.e.,
the latter part of the 1995 wet season. However, even during the greatest months of overall ripe
fruit availability, April and May 1996-i.e., at the end of that year's dry season and during the
transition into the following wet season-coatis were never observed to spend more than 6.6% of
their foraging time in the canopy of fruiting trees. The animals searched for food under fruiting
trees somewhat more, spending 22.8% of their foraging time in such areas in April, 24.6% in May
and 24.5% in September 1996, i.e., the early to middle wet season of that year, but these were
areas of overlap in which not only ripe fruits but also leaf litter invertebrates could be found.
Examining invertebrate and fruit foraging success (Figs. 2-9 and 2-10, respectively) more
closely, fewer invertebrates were consumed and fewer seconds were spent consuming them per
minute relevant foraging time than was true for fruits in almost every month in which both food
types were eaten; March and April 1996 were exceptional in that invertebrates and fruits were
consumed at comparably low rates. With respect to time spent consuming items, however, it
should be noted that coatis used some undetermined portion of that time processing fruits of
Brosimum alicastrum, M. zapota, T. moschata and Blomia prisca in order to discard various
presumably inedible or distasteful portions (husks, skins and/or seeds).
Over the 14 months of systematic behavioral observations, the coatis' consumption of
leaf litter invertebrates (Fig. 2-9) displayed a clear pattern in terms of number of invertebrates
consumed and time spent consuming them. Overall, invertebrate consumption decreased from
the beginning of the study in August 1995 through April 1996, i.e., through the last half of the
1995 wet season and all of the 1996 dry season, then rose sharply in May 1996 with the onset of
that year's wet season and peaked a month or two later (depending on the measure under
consideration) before again declining. Neither measure of foraging success was significantly
correlated with total invertebrate mass in monthly samples (number consumed: rp 11 = 0.26, P =
0.39; time spent consuming: r[i ] = 0.38, P = 0.20). In accordance with previously presented data,
beetles and grubs-again, especially E. endymion-predominated. Only 19.4% of the invertebrates
coatis preyed upon during observations could be positively identified (with an obvious bias
favoring larger or more agile species); 96.2% of these were insects, and 81.9% various
Coleoptera. E. endymion alone accounted for 44.4% of identified invertebrate prey, or 8.6%
(6.5% larvae, 2.1% adults) of all invertebrates consumed by coatis over the course of the study.
Fruit consumption (Fig. 2-10) followed a similar pattern (aside from the several months
in which ripe fruit was unavailable, of course). A notable exception resulted from the animals'
early reliance upon M. zapota, the first tree species with ripe fruit in the 1996 dry season, causing
a brief spike in consumption rates in February of that year. That species and also Brosimum
alicastrum, which ripened soon after, continued to be consumed at much lower rates throughout
the remainder of the dry season. Fruit consumption increased again in May 1996 when ripe fruit
of various species appeared with the onset of wet season, as noted previously. It was not possible
to look for correlations between fruit foraging success and total fruit availability because
phenological data were lacking for several of the fruit species which were consumed by coatis.
Wet Season Dry Season Wet Season
Sep Nov Jan Mar May
Figure 2-8. Mean percentage of time foraging adult female white-nosed coatis spent in
areas where ripe fruit was present and absent in Tikal National Park, Guatemala, August
1995 through September 1996. Based on a total of 76.1 hours observation of foraging by
Wet Season Dry Season Wet Season
- 9 16
19 1 18 18
"o 7.5 -
Aug Oct Dec Feb Apr Jun Aug
Sep Nov Jan Mar May Jul
Figure 2-9. Mean ( SEM) number of invertebrates consumed and time spent
consuming them by foraging adult female white-nosed coatis in Tikal National Park,
Guatemala, August 1995 through September 1996. Number of females sampled in each
month indicated on bars. Based on a total of 93.7 hours observation of invertebrate
foraging by 32 animals and 73.0 of 29 animals, respectively.
Time Spent Consuming
Wet Season Dry Season Wet Season
Aug Oct Dec Feb Apr Jun Aug
Sep Nov Jan Mar May Jul Sep
Aug Oct Dec Feb Apr Jun Aug
Figure 2-10. Mean (+ SEM) number of fruits consumed and time spent consuming them
by foraging adult female white-nosed coatis in Tikal National Park, Guatemala, August
1995 through September 1996. Number of females sampled in each month indicated on
bars. Based on a total of 10.4 hours observation of fruit foraging by 23 animals.
Adult female coati body fat profiles (Fig. 2-11), presumably indicative of how well
energy availability met energetic demands, displayed a clear seasonal pattern. Body fat declined
rapidly from its high point at the beginning of the dry season in January to the mating period in
the last half of February, leveled off or slightly rose thereafter until just before the parturition
period began in the last half of April, then resumed its decline through the parturition period and
well into the wet season. Fat stores appeared to reach nadir in the vicinity of July to early August
(a time of sparse data), i.e., shortly before the middle of the wet season, and then rose presumably
steadily (discounting another gap in the dataset from mid-October through November) through
the remainder of the wet season and until the end of the year.
Behavioral and morphometric data reveal that coatis in Tikal National Park normally
reproduce only once per year, with relevant events occurring across the population in narrow,
consistent periods: Mating takes place over an approximately two-week period in the middle of
the dry season. Birth occurs over a comparable period at the end of the dry season, about 10
weeks after mating. Solitary nesting, which lasts 1/V2-2 months, begins shortly before birth occurs
and ends with the emergence of young at the beginning of the wet season; the band reunites
within the month following nesting. The young begin foraging alongside their dam as soon as
they are brought down to the ground, but are not fully weaned until months later, in the middle of
the wet season. A second reproductive cycle with similar timing but delayed by approximately
four months may be initiated if the first is unsuccessful, but second cycles rarely occurred in
Tikal and appeared to result in smaller litters. Taking into account seasonal differences in
different parts of the coati's range, the reproductive pattern in Tikal fits well with previously
published information on coatis at other locales.
Based primarily on observations of increased female-male affiliative behavior and male-
male agonism from late January to mid-March 1959 and throughout January 1960, Kaufmann
(1962) identified a month-long period of sexual activity among coatis on BCI; because it appears
Dry Season Wet Season
o'- o 8 o ooo
- 0 0
o o o
; 0 0 0 00
% 0 O O 0
I I I l l | l l l I I I I l
Jan Feb Mar Apr May Jun Jul
I I I I I I I I II II I I
Aug Sep Oct Nov Dec
Figure 2-11. Annual body fat profile exhibited by adult female white-nosed coatis in
Tikal National Park, Guatemala, July 1994 through October 1996. Mating period
(13 Feb 28 Feb) indicated by lightly shaded background and period of parturition
(16 Apr 9 May) more darkly shaded. Based on measurements obtained during 135
immobilizations of 54 animals.
that actual copulation was not observed (see below), however, the mating period could not be
defined. Based on similar observations and the timing of nest emergence, Russell (1979; 1982)
concluded that the coati mating period on BCI lasted one to two weeks in early February, shifting
slightly from year to year. Sunquist and Montgomery (1973) observed a pair of coatis in coitus
on the island on 5 March 1971. Gilbert (1973) believed that the coatis in southern Arizona, at the
northernmost edge of the species' range, mated in March and early April. Hass and Roback
(2000) observed two coati copulations (excluding a possible third; see below) there on 6 April 97
and 5 April 98, and also stated without providing further details that the mating period in their
population lasted from late March to mid-April. Studying captive animals in southern Arizona,
Smith (1980) reported that copulations took place on 28 and 30 March 1976.
Previous detailed descriptions of mating activity among free-ranging coatis include
mounts lasting only a few seconds (Kaufmann, 1962; Gilbert, 1973; Hass and Roback, 2000).
Kaufmann (1962) referred to these as attempted copulations, whereas the later authors considered
them to be copulations. However, Gilbert (1973) also believed that more lengthy periods of
mating activity probably occurred out of sight of observers, when a male and a female would
depart from the band together, and Hass and Roback (2000) additionally observed two
copulations lasting approximately 60 minutes each between coatis who were out of sight of other
band members. The coatis found already engaged in copulation (most similarly to copulations
observed in the current study, in the tree canopy over a foraging band) by Sunquist and
Montgomery (1973) continued for seven minutes after being discovered. Smith described two
copulations among captive coatis: one lasted one minute in total and another continued for 23
minutes after being discovered in progress. The numerous observations of prolonged copulatory
events in the present and earlier studies suggest that Kaufmann (1962) was correct in describing
brief mounts as merely attempted copulations.
Prolonged copulation could be interpreted as an indication that coatis are induced
ovulators, as has been suggested by Hass and Roback (2000), but it may also be a form of mate
guarding by males. Litters examined in Tikal exhibited a high rate of multiple paternity (Booth-
Binczik, 2001), indicating that female coatis commonly mate with more than one male.
Observations in the current study and by Smith (1980) of females attempting to break away from
but being restrained by males suggest that prolonged copulation is more advantageous to the male
than to the female. The closely related raccoon (Procyon lotor) was demonstrated by Sanderson
(1961) to ovulate spontaneously (after initially being reported to be an induced ovulator by
Llewellyn and Enders, 1954-an erroneous finding that unfortunately persists in the literature), but
the review of carnivore ovulation by Lariviere and Ferguson (2003) demonstrates that it is unsafe
to base assumptions in this regard on phylogenetic relationships.
Coatis were not radiocollared for Kaufmann's (1962) BCI study, so he was able to keep
track of only one female during the 1960 nesting period; that animal left its band on 1 April, gave
birth on 10 April and brought its young out of the nest and joined other females with young on 20
May. Without providing any foundation, Russell (1979) stated that coati births occurred on the
island during the second week of April in 1977 and the fourth week of April in 1978, and
(Russell, 1982) that the nesting period there took place in April and May. Russell (1982) also
said that coati bands on BCI reaggregated in late May. Based on animal movements monitored
by radiotelemetry, SAenz (1994) reported that female coatis in the dry forest of Santa Rosa
National Park, Costa Rica, left their bands to nest in April. Also based on telemetry data,
Ratnayeke et al. (1994) reported that females in southern Arizona were solitary for at least a
month before parturition in early July. Gilbert (1973) stated that females there disbanded in the
third week of June, and estimated that all females in the band he was observing gave birth
between 16 and 23 June. Hass (2002) said that females in Arizona left their bands in mid-June
and reaggregated in early August when the young were five to six weeks old.
There are as yet few data available on the length of coati gestation against which to
compare the 71- to 74-day pregnancy seen in a single female in the current study. The most
concrete of these come from Smith's (1980) Arizona captive animal study, in which one female
gave birth 71 days and another 76 days after being observed copulating. Average gestation
lengths can be approximated for the two coati bands from which most data derive for the 1996
mating and parturition periods of the present study, bearing in mind that parturition periods
(which are presented and discussed in greater detail in Chapter 3) were likely overestimated; the
differences between the midpoints in the mating and parturition periods in these bands are 66 and
68 days. Kaufmann (1962) reported that three coati litters were born approximately 70 days after
the period in which he observed heightened sexual activity on BCI. The relative consistency of
these data across divergent locales-as well as the tightness and consistency of the mating period
at the start of the coati reproductive cycle-suggest this species undergoes direct development.
Understandably little information is available concerning coati lactation. Kaufmann
(1962) reported that in 1958 nursing was seen on BCI as late as 16 September. Inexplicably,
Russell (1983) put forth that young there were nearly weaned by the time of reaggregation,
months earlier. Smith (1980) noted that young born to a captive female first began eating solid
food at 45 days of age. Smith also reported that the female began moving away at the approach
of its young when the latter reached 74 days of age, and that the female's teats appeared greatly
reduced in size at that time; she observed no nursing as of 80 days postpartum.
Prior to this study, nesting behavior and parturition outside of the usual narrowly defined
coati nesting season had been observed only in Santa Rosa, where Sienz (1994) found that
females quite commonly underwent second reproductive cycles after losing their first litters to
nest predation by white-faced capuchins (Cebus capuchinus); second litters were born there in
late July or early August. Second litters have also been documented in the procyonid P. lotor in
south Texas after failed reproductive attempts were made during the normal breeding season
(Gehrt and Fritzell, 1996), and the phenomenon has been seen in many other mammals as well.
The data on food availability provide a ready ultimate explanation for coati reproductive
seasonality in Tikal, but leave some details yet to be clarified. In the present study, infants
emerged from their nests at the start of the wet season when (in at least one year) both leaf litter
invertebrates and fruit were increasing in abundance. Are these patterns in food availability
consistent, though? Taken by itself, this ecologically brief study suggests that the invertebrate
pattern may be consistent but the fruit pattern is not; placing the current results into the context of
the literature provides much keener insight into the question of pattern inconsistencyc.
Populations of leaf litter invertebrates in the tropics are strongly dependent on moisture
levels (Levings and Windsor, 1982), as is evidenced by the high correlation between invertebrate
abundance and rainfall in Tikal. As occurred in the present study, overall litter invertebrate
abundance on BCI was lowest during the dry season and highest in the early wet season (Levings
and Windsor, 1982). Nine of thirteen arthropod taxa examined there individually showed
population maxima in the early wet season (Levings and Windsor, 1985). E. endymion adults
congregate around lights at night in huge numbers (pers. obs.), and should therefore be
considered night-flying insects as well as litter insects. Night-flying insect abundance (and large
beetle abundance in particular) also peaked in the early wet season on BCI (Smythe, 1982), and
night-flying insects were most abundant in the late dry season and early wet season at the Cocha
Cashu Biological Station in Peru's Manu National Park (Terborgh et al., 1986; cited in Goldizen
et al., 1988). Moreover, differences in rainfall patterns among years affect litter arthropod
population levels but do not obscure seasonal trends in abundance (Levings and Windsor, 1985).
In contrast, seasonal patterns of tropical fruit production vary greatly from year to year
not just in Tikal but at other sites as well (Foster, 1982a; Bullock and Solis-Magallanes, 1990;
van Schaik et al., 1993; Chapman et al., 2005). None of the tree species that were most important
in the coati diet in Tikal showed the same pattern of fruit availability in June-September 1995 as
in June-September 1996 (at the beginning and end of the phenological survey). Rains were heavy
at the height of the 1995 wet season and even more unusually light in the middle of the 1996 wet
season; leaf litter fall and invertebrate and fruit availability all showed associated patterns. The
climatic fluctuations may have been due to the occurrence of an El Nifio event in 1994 (based on
information from the U.S. National Oceanic and Atmospheric Administration, Washington, DC);
the El Nifio Southern Oscillation tends to create a pattern in which dry sunny years alternate with
wet cloudy years in Central America (Wright et al., 1999). Unusually rainy dry seasons,
particularly following an El Niiio year (Wright et al., 1999), apparently result in failure of the
fruit crop in many species that normally produce fruit in the wet season, including the ordinarily
prolific B. alicastrum (Foster, 1982a). Even in the absence of obvious rainfall anomalies,
however, many tropical tree species exhibit non-annual rhythms in fruit production and/or high
intraspecific variability in phenology (Milton, 1991; Newstrom et al., 1994; Hemingway and
Overdorff, 1999). It has been reported that B. alicastrum trees in Tikal produce fruit three times
per year: February-March, June-July, and October-November (Coelho et al., 1976). The
phenology data in the present study do not support this contention, but do indicate non-
synchronous fruit production by this species, as fruit availability continued and even increased
during a lengthy period of heavy consumption by coatis and other frugivores (pers. obs.).
The data are scant but suggest consistent seasonal patterns exist in food nutrient content
as well as food availability. In the present study virtually all leaf litter invertebrate nutrient levels
rose and fell in parallel to rainfall, although only fat content correlated significantly. Fruit
nutrient levels were examined herein at only a single point in time per species, but Schaefer and
Schmidt (2002) reported that fleshy fruits in Venezuela rose steadily in caloric value throughout
the dry season, and Worman and Chapman (2005) found that the fat content of ripe fruits of a
single tropical species varied tremendously within a year, being much higher during periods of
high rainfall. This is clearly an area of tropical ecology warranting further study.
There is also the matter of the freezer failure and its potential effects on the present
study's nutritional analyses to consider. Few nutritional data are available in the literature that
are useful for comparison, but those are encouraging. Analysis of an unknown species of dung
beetle from Tanzania (Pennino et al., 1991) produced values remarkably similar to those found in
the present study for adult E. endymion for water content (58.3%), fat content (20.6%), NDF
(51.6%) and ash (7.4%), but completely different for protein (7.8% versus 67.5% in the current
study). Previously published values (Coelho et al., 1976) for B. alicastrum fruit (84% water,
15.6% protein, 3.1% fat, 3.50 kcal/g) and M. zapota fruit (75% water, 2.0% protein, 4.4% fat,
3.76 kcal/g) are likewise similar to those obtained in this study.
The relative importance of leaf litter invertebrate versus fruit availability to the timing of
coati reproduction remains uncertain, but the findings of the present study appear to support
Russell's (1982) assertion of the primacy of invertebrates. Tikal's coatis clearly focused their
foraging efforts on invertebrates more than on fruit at all times, though seasonal patterns are
somewhat difficult to assess due to the high degree of interannual variation in fruit availability
during this study. For example, the animals ate no fruit during the second half of the wet season
in 1995 because no fruit was then available, but data from September and anecdotal observations
made in October and November 1996 suggest that foraging patterns late in that year's wet season
might have looked quite different. Coatis spent the highest percentage of their foraging time in
fruit zones at the end of the dry season, but it was still much less than the time they spent foraging
for invertebrates; the time they spent consuming invertebrates was actually higher then than at
any other time, too, suggesting that both fruit and invertebrates were abundant at that time. An
evolutionary argument also exists for invertebrates being more important than fruits as
determinants of coati reproductive timing, namely that consistent annual peaks in invertebrate
availability are much more capable of driving seasonal adaptation than are profoundly less
consistent peaks in fruit availability.
In any event, the present study adds to the growing body of literature demonstrating the
importance of leaf litter invertebrates and their phenology to coatis. Kaufmann (1962) reported
that BCI's coatis spent more than 95% of their active time foraging for invertebrates, and that
fruit was eaten more during the dry season than the wet season. Russell (1982) estimated that
coatis on BCI spent 89% of their foraging time in the wet season and 54% in the dry season
foraging for invertebrates. He also reported that scats he collected consisted mostly of beetle
parts. Contrasting somewhat to the patterns in Tikal and BCI, Sienz (1994) found approximately
equal frequencies of invertebrates and fruit in the scats he collected throughout the year in Costa
Rica; as in Tikal, though, Coleoptera was the dominant invertebrate group in the diet (followed
closely by Orthoptera). A heavy reliance upon beetle prey is clearly a common facet of coati
ecology at a variety of sites, but no previous study has found as strong a relationship between this
procyonid and a single prey species as was evident in Tikal between it and E. endymion.
Coati reproductive seasonality in Tikal appears to be driven not by restrictive effects of
food availability on female physiology (reproductive or otherwise), but rather by its direct effects
on the survival of offspring. Specifically, reproduction there is timed such that gradually weaning
young (arguably the most vulnerable life stage; Russell, 1982) can capitalize on a surfeit of food
resources which first appears at the beginning of the wet season and continues through the first
half of that season. As others have pointed out, predatory species are unlikely to suffer shortages
of particular nutrients (Stephens and Krebs, 1986; Bronson and Heideman, 1994; Galef, 1996), so
when such species are timing reproduction to match food availability they are likely doing so
with regard to more purely energetic constraints. Female coatis do lose energetic stores (i.e., fat)
during nesting and early weaning (the most energetically expensive part of reproduction for many
mammals; Rutberg, 1987) even as males are rapidly rebuilding theirs after the mating period (see
Chapter 4) by relying upon the same foods (Booth-Binczik, 2001), but females apparently have
sufficient reserves to get them through. Unlike on BCI where deferment of reproduction was
common (Kaufmann, 1962; Russell, 1982), virtually all mature females reproduced each year in
Tikal (unpublished data). The fact that some females in Tikal were able to successfully complete
second reproductive cycles after failing in their first provides additional evidence of ample
energetic reserves in these animals. Reviewing reproductive seasonality among neotropical
primates, DiBitetti and Janson (2000) concluded that very small species time reproduction so that
peak food availability coincides with lactation, whereas species large enough to store sufficient
energy reserves-i.e., those similar in size to the coati-time reproduction so that peak food
availability coincides with weaning.
To more fully assess the importance of birth season to offspring survival, it would be
useful to compare survivorship among young born during second cycles to that among young
born at the normal time of year. However, the few litters that were produced during second
cycles in this study do not provide adequate data for such a comparison. One infant that was born
in August 1995 and the four that were born in August 1996 were known to survive to the end of
the study; the fates of the other out-of-season infants were unknown (as is the cause of whatever
difference in litter size exists between first and second litters).
Beyond energy availability, a few related ultimate factors might contribute to the
observed pattern of seasonal reproduction. For example, protein content of invertebrates peaked
in the middle of the wet season, so birth might be timed such that infants are fully weaned when
the most protein is available to support their continued growth. Also, the fact that birth occurs at
the end of the dry season means that juveniles have as much time as possible to grow, learn and
build energetic reserves before they have to deal with the dry-season low in invertebrate
availability (Russell, 1982). The only other factor that appears at all likely to promote strong
coati reproductive seasonality in Tikal is water availability. The region's pronounced dry season
and lack of natural permanent surface water (due to underlying permeable limestone) create the
possibility that lactating females might be restricted in milk production by a shortage of water.
However, this is not likely to be a factor in most parts of the species' range, and the fact that
females in Tikal give birth before the dry season ends suggests that water stress is not relevant to
the timing of their reproduction.
Reproductive seasonality such as the coati displays would not be remarkable in a long-
lived mammal inhabiting the higher latitudes; indeed, such patterns are the norm there (Sadleir,
1969; Bronson, 1989). Despite the paucity of studies on long-lived tropical species, reproductive
seasonality is increasingly known from the lower latitudes, too. The coati is extraordinary among
that latter group, however, in the narrow, consistent timing of its reproductive events. Looking
past species which appear to display no seasonality whatsoever, such as the sun bear (Helarctos
malayanus) in Borneo (Schwarzenberger et al., 2004), seasonal reproductive events among
tropical mammals are better thought of as indistinct statistical phenomena rather than clearly
demarcated ecological events. Spotted hyena (Crocuta crocuta) populations distributed across
very low latitudes reproduce throughout the year, but often with modest population-specific
seasonal peaks (Holekamp and Smale, 2000). Acouchies (Myoprocta exilis), agoutis (Dasyprocta
leporina) and pacas (Agouti paca) at 6N latitude similarly give birth throughout the year, albeit
with apparent seasonal peaks (Dubost et al., 2005). White-lipped peccaries (Tayassu pecari) at
8N latitude and saddle-back tamarins (Saguinusfuscicollis) at 120S latitude both display birth
peaks coincident with greatest food availability in the first half of the wet season, as does the
coati, but the peccary's mating period has been documented to last at least eight months
(Altrichter et al., 2001) and the tamarin's parturition period spans a similar length of time
(Goldizen et al., 1988). A population of the Ethiopian wolf (Canis simensis) times reproduction
somewhat more finely at 7N latitude, displaying a parturition period of about four months
(Sillero-Zubiri et al., 1998). Extremely few studies have as yet been conducted on or in the very
near vicinity of the equator, but at least one species there has already been found to display
seasonal reproduction: mandrills (Mandrillus sphinx) exhibit a 6-month mating period (Abernethy
et al., 2002).
Wherever it is found, reproductive seasonality for most long-lived mammals appears to
be a strategy of optimization. As such, the degree of variation seen in the timing of specific
reproductive events should reflect the selective pressures that favor those events taking place at
the most advantageous times. So why is the same adaptation so often expressed so differently in
the tropics than it is at the higher latitudes?
Two related arguments that have been posited are that optimal conditions occur over a
longer time period or selective pressures favoring seasonality are otherwise weaker in the tropics
than at higher latitudes because of the less variable/more benign climates generally thought to
exist in the former region. However, many (perhaps even most) tropical locales fail to live up to
such rosy expectations (Foster, 1982a; Bronson and Heideman, 1994). Even setting that fact
aside, selective pressures weak or strong are still selective pressures. Given an evolutionary
timeframe, species-or more accurately, their populations-should respond to those pressures as
best they are able.
Bronson and Heideman (1994) have argued persuasively that reproductive seasonality is
generally less well-defined in the tropics because animals residing there have less reliable or
distinct proximate cues upon which to rely. Photoperiod, the most universally relied upon
zeitgeber for long-lived mammalian reproduction (Sadleir, 1969; Bronson, 1989), obviously
diminishes in degree of both annual and daily change as one proceeds from the poles to the
equator. Tropical rains and their innumerable effects on food availability, although often
apparently consistent enough over history to promote the evolution of reproductive seasonality,
generally vary too much from year to year to serve as predictive cues for timing specific
reproductive events. Other possible cues appear even less reliable.
It is unknown at what point photoperiodic changes become too slight for animals to use
them to time or fine-tune the timing of their reproduction (Bronson and Heideman, 1994), but of
course such timing depends on the extent of both the photoperiodic changes themselves and the
genetic variability of the populations subjected to them. One prediction based on this premise in
combination with the assumption that there are no latitudinally based differences in population
genetic variability is that long-lived seasonal species (or even closely related groups of species)
should display a broadening and perhaps ultimately a disappearance of peaks in reproductive
events in populations at increasingly low latitudes, and this has been amply demonstrated to occur
(Bronson, 1985; Di Bitetti and Janson, 2000). Bronson (1989) also suggested that social species
should be able to time their reproduction more finely than would otherwise be possible at low
latitudes by using interactive cues to augment environmental cues, i.e., the reproductive responses
of the most environmentally sensitive individuals in a population could socially trigger like
responses in the population's remainder. In accordance with this prediction, the tropical
mammals that have thus far been found to exhibit relatively narrow reproductive seasonality are
indeed social species (e.g., ring-tailed lemurs, Lemur catta; Jolly, 1967; impalas, Aepyceros
melampus: Murray, 1982; greater spear-nosed bats, Phyllostomus hastatus: Porter and Wilkinson,
2001; red ruffed lemurs, Varecia rubra: Vasey, 2005). The coati may simply be one of the most
extreme examples of such found to date. (This subject is examined in greater detail in Chapter 3).
The discussion does not end there, however. As O'Brien (1993) emphasized, lamentably
few long-lived tropical mammals have so far been examined for the possibility-let alone the
extent-of reproductive photoresponsiveness and, with the exception of Nile grass rats
(Arvicanthis niloticus: Sicard et al., 1992), all of the short-lived tropical species that have so far
demonstrated any photoresponsiveness whatsoever in the laboratory have displayed continuous or
opportunistic reproduction (i.e., free from predictive cues) in the wild. Looking farther abroad
taxonomically, spotted antbirds (Hylophylax n. naevioides) at 90N latitude in Panama show
reproductive physiological responsiveness to the very slight natural changes in daylength found at
that latitude (Hau et al., 1998). If tropical mammals possess similar capabilities, that of course
merely begs the question of why they do not utilize them more often. To the delight of the
curious, scientific endeavors invariably raise more questions than they answer.
In the absence of extraordinary selective pressures, reproductive seasonality is a strategy
of optimization. As such, it often allows for considerable variation in the timing of specific
reproductive events, even within a single population of an obligate seasonal species. This is
especially true in the tropics, where annual variation in photoperiod is slight and other potential
environmental predictors may be patchy or may shift somewhat from year to year in response to
climatic conditions (e.g., rainfall; Rand and Rand, 1982). For example, as Russell (1982) pointed
out, birth peaks among Panama's seasonally reproducing mammals are largely probabilistic
events; birth periods for most species span at least a few months (summarized in Fleming, 1973).
The white-nosed coati is unusual in this regard, with all parturition on Barro Colorado
Island (BCI), Panama-the only population previously well-studied-seemingly occurring within a
1- to 2-week period each year (largely inferred from observations of first emergence of young
from nests; Kaufmann, 1962; Russell, 1979; 1982). This pattern prompts consideration of
ecological pressures favoring a high degree of birth synchrony in the species. [Note: The terms
reproductive seasonality and synchrony have sometimes been used interchangeably (e.g., Boinski,
1987), leading to confusion about the nature and causes of these phenomena. The author has
adopted Ims' (1990b) definition of reproductive synchrony: "a phenomenon caused by biological
interactions operating to produce a tighter clustering of reproductive events than would have been
imposed by environmental seasonality alone" (p. 135).]
Adult female coatis are also highly social, living in cohesive bands with their young most
of the year and becoming solitary only to nest (Kaufmann, 1962). Russell (1982; 1983)
emphasized the importance of cooperation among adult females to minimize losses of young to
predation and parasitism, and (1982) suggested that a high degree of both intra- and interband
birth synchrony exists to facilitate this strategy. Other possible explanations more generally
offered for birth synchrony include predator swamping (Pianka, 1976), dependence upon an
ephemeral food or other resource (e.g., the need of some desert anurans for breeding ponds), or
utilization of social mechanisms to supplement inadequate environmental predictors (e.g.,
photoperiod at low latitudes) employed in a seasonal reproductive strategy (Bronson, 1989).
None of these explanations has yet been rigorously examined for the coati, nor has birth
synchrony even been established as a general phenomenon in the species (although a handful of
reports suggest that coati reproduction may be patterned similarly at other locations, e.g., in
southern Arizona: Gilbert, 1973; in Costa Rica: Sienz, 1994; in Jalisco, Mexico: Valenzuela and
Ceballos, 2000). It seems reasonable to rule out transitory requirements as an explanation for
birth synchrony in this long-lived, tropical forest-dwelling mammal, however, and no such
requirements were discovered during the extensive observations of BCI's coatis by Kaufmann
(1962), Russell (1979) and Gompper (1994).
Accordingly, this study examines data on the timing of coati reproductive events at Tikal
National Park, Guatemala, to provide details on birth synchrony at this new locale and to test
specific predictions about the phenomenon based on the following hypotheses:
HI: Coati birth synchrony minimizes juvenile mortality by facilitating communal care of
The five- to six-week nesting period (Chapter 2) is the least social time in the lives of
female and juvenile coatis, so communal care cannot become a relevant factor until the animals
have reaggregated into bands after the young have emerged from the nest. Birth synchrony
would be relatively useless for facilitating communal care if subsequent reaggregation is too
asynchronous. It was therefore predicted that within a band of coatis, the variance in the timing
of reaggregation is no greater than in that of parturition.
H2: Birth synchrony minimizes juvenile mortality by predator swamping.
Even the largest bands of coatis do not include more than about a score of reproductive
females in a given year (Booth-Binczik, 2001; Chapter 2), and predators are capable of
consuming an entire litter of newborn coatis in short order (e.g., Rose, 1997). There is also much
home range overlap among neighboring coati bands (Kaufmann, 1962; Gompper, 1997). To
swamp predators rather than merely prompt them to shift their focus from band to band, then,
birth synchrony would have to exist across as well as within bands. Hence it was predicted that
the variance in the timing of parturition between bands is equal to that within bands.
H3: Birth synchrony results from the coati's reliance upon social mechanisms to enhance
If coatis are relying upon social mechanisms to augment weak environmental cues for
timing reproduction, it is reasonable to assume that this process will result in a greater degree of
birth synchrony within than between bands. In opposition to the prediction of H2, therefore, this
hypothesis generated a prediction that the variance in the timing of parturition between bands is
greater than that within bands.
It should be noted that the aforementioned three hypotheses are in fact independent rather
than mutually exclusive alternatives; more than one may pertain. Further, whereas the first two
hypotheses directly address possible ultimate causes of birth synchrony, the third instead
addresses a proximate mechanism by which the ultimate factor of seasonal food availability (see
Chapter 2) may contribute to the coati's tight clustering of birth.
From June 1994 through October 1996, various data were collected on a population of
free-ranging white-nosed coatis in Tikal National Park, Guatemala (17N, 890W) as part of a
broad study of the species' reproductive biology. (An overview of the coati's reproductive
pattern at this locale appears in Fig. 3-1.) This chapter specifically examines data on the timing
of parturition in 1995 and 1996, and the reaggregation of parous females after nesting in 1996.
Dry Season Wet Season
Figure 3-1. Timeline of white-nosed coati reproductive phenological events in Tikal
National Park, Guatemala (see Chapter 2).
In all, reproductive events occurring among 28 females from three bands contributed to
the study. Each female was initially captured by live-trap or (more often) blowgun and
chemically immobilized by intramuscular injection of approximately 7 mg/kg Telazol (Fort
Dodge Laboratories Inc., Fort Dodge, IA). Anesthetized animals were tattooed and eartagged for
identification and fitted with 90 g motion-sensitive radiocollars (Advanced Telemetry Systems,
Isanti, MN) to enable monitoring thereafter. The animals were subsequently habituated to the
close presence of observers (as described in Chapter 2), and the status of each female was
checked daily (as possible) during the time periods when parturition and (in 1996) reaggregation
took place. All animal handling procedures were conducted as per protocol #4084 approved by
the University of Florida's Institutional Animal Care and Use Committee.
Parturition was determined to have occurred when a female showed a sudden
pronounced reduction in girth followed by continued nesting behavior, and was ultimately
confirmed by observation of young still in the nest or shortly after emergence from the nest.
Parturition dates could not usually be pinpointed because the females tended to give birth while
remaining out of sight within their nests for days at a time. In this circumstance the scoring of a
female's parturition was distributed over an appropriate span of time (e.g., one-third of the
female's parturition was ascribed to each of three possible days). Data on females of
indeterminate status for two weeks or longer were not utilized.
For the purpose of this study, reaggregation was deemed to have occurred when a
postpartum female and her young were observed to have reunited with at least three other such
females. In most cases reaggregation dates could not be pinpointed and were scored as described
above for uncertain parturition dates.
Sokal and Rohlf (1981) guided all statistical efforts, and a Pa < 0.05 level of significance
was utilized in all tests. In order to assess whether the variance in the timing of parturition was
greater than or equal to that of reaggregation, i.e., to test the communal care prediction, a one-
tailed test for equality of two variances was conducted for each of the two coati bands that
contributed almost all of these data in 1996; data on a single female from a third band were
included in the descriptive statistics presented for this year. An ANOVA was conducted to
compare the variances in the timing of parturition between and within the two primary bands in
1996, i.e., to test the predictions for predator swamping and social facilitation of reproductive
seasonality. Insufficient data were obtained for the 1995 parturition period to include in such
analyses, and these data were therefore used only for comparison with the 1996 parturition period
via descriptive statistics. To aid in comparing the annual birth pulse of the coati in Tikal to that
seen at other locales and/or in other species, the interquartile range (i.e., middle 50%) was also
calculated for each year's data.
Coati parturition was both narrowly timed within each year and consistent between the
years of the study (Table 3-1). The nine births recorded in 1995 occurred in a slightly flattened
but normally distributed peak of 14 days duration, and the 24 births recorded in 1996 occurred in
a similar peak lasting 18 days. These peaks were centered six days apart late in the dry seasons/at
the beginning of the wet seasons (see Chapter 2) of their respective years.
There was a discernible difference in the timing of parturition between the two primary
bands under observation in 1996 (Fig. 3-2), with Band 1 females giving birth over a 14-day span
that began four days after the start of the 12-day span in which Band 2 females did so. Mean
birthdates in the two normally distributed birth peaks were 2 May and 27 April, respectively. The
variance in the timing of birth between Bands 1 and 2 was significantly greater than that within
the bands (F[1.21 = 24.21, P < 0.001).
Reaggregation in 1996 occurred relatively quickly after females first brought their young
down from their nests between the early and middle wet season (see Chapter 2). Eleven females
with young from Band I found at least three other such families within a 20-day span beginning
31 May, and nine females with young from Band 2 accomplished it in 16 days beginning 4 June.
Comparing parturition to reaggregation within each band: The variance in the timing of
n nn F
22 23 24 25
26 27 28
m= Band 1 (N= 12 females)
hI Band 2 (N = 11 females)
SL n H n r
29 30 1 2 3 4
7 8 9
Figure 3-2. Timing of births in neighboring white-nosed coati bands in Tikal National
Park, Guatemala, in 1996. Scoring of a female's parturition distributed over appropriate
span of time when date not pinpointed (e.g. 1/3 of female's parturition ascribed to each of
three possible days); mean ( SEM) span for all females 3.9 + 0.6 days, range 1-10 days.
S. 1 1...
reaggregation was significantly greater than in that of birth in Band 2 (F[8,0o] = 8.77, P = 0.001).
Band 1 displayed the same tendency, but it was not significant (F[lo. l = 2.28, P = 0.10).
The communal care hypothesis for coati birth synchrony assumes that the presence of
multiple females at the same reproductive stage is important to the survival of young. Certainly,
evidence suggests that coati sociality is employed in group defense, including of offspring.
Russell (1983) found that BCI's coati bands adopted a formation that was optimal for protection
of young rather than foraging efficiency when moving through the forest. He further reported
that juveniles spent more time vigilant when unaccompanied by adults, and adults were more
vigilant in smaller groups. This latter finding was also confirmed by Burger and Gochfeld (1992)
in Costa Rica. However, the presence of the band may actually be more important to the survival
of adult females than of juveniles, as mortality rates of females are higher when they are alone
than when they are in groups (Booth-Binczik, 2001; Hass and Valenzuela, 2002) and female
mortality is lower in larger groups, whereas juvenile mortality is not (Hass and Valenzuela,
2002). There are no published data on relative mortality rates of juveniles before and after band
reaggregation, and direct observations of predation are few. During the course of the present
study a pair of crested eagles, Morphnus guianensis, was observed taking five coati infants to its
Table 3-1. Descriptive statistics for observed white-nosed coati birth periods in two years in
Tikal National Park, Guatemala.
Statistic 1995 1996
Number of Females / Bands 9 / 2 24 / 3
Parturition Period 16 29 April 22 April 9 Maya
Mean Birthdate 24 April 29 Aprila
Interquartile Range (days) 5 6
Degree of Kurtosis (- 3) -0.62 -0.69
a calendar dates not strictly comparable because 1996 was a leap year
nest in a couple of weeks' time at about the end of the coati nesting period in Tikal, and other
raptor species were occasionally observed preying upon nestling coatis as well (D.F. Whitacre,
pers. comm.). Rose (1997) documented heavy predation on nestling coatis-principally within
their first few weeks of life-by white-faced capuchins (Cebus capuchinus) in Santa Rosa National
Park, Costa Rica, and Sienz (1994) reported that this resulted in as much as 100% nestling
mortality in some bands. Russell (1982) reported heavy juvenile mortality on BCI in the first
three months following reaggregation, and Hass and Valenzuela (2002) estimated 43-75%
mortality in the first 3-5 months of life in southern Arizona and Jalisco, Mexico. It appears likely
that the risks faced by a young coati do indeed diminish in the company of its band, but the anti-
predation benefits of group living do not depend on females being at the same reproductive stage.
Certain aspects of coati biology and two historical observations of coati behavior (albeit
perhaps of a single animal) prompt special consideration of the adult male as a potential predator
of young. The subject of infanticide by male coatis has been dealt with more fully by Booth-
Binczik (2001), but brief mention in the course of this discussion of birth synchrony and
communal care seems appropriate. Female coatis have been documented to undergo successful
second reproductive cycles shortly after losing litters born at the normal time (Sienz, 1994;
Chapter 2), making sexually selected infanticide a possibility (Bertram, 1975; Hrdy, 1979). The
fact that males are predominantly solitary has also been interpreted as a result of exclusion by
females for the protection of their young (Russell, 1981). In support of his interpretation, Russell
(1981) reported an apparently secondhand observation of an adult male coati on BCI consuming
an infant near a biological laboratory animal feeding station, and another of an adult male
(possibly the same animal) attacking and carrying an infant away from the same site on the
following day. However, in Tikal most coati litters are multiply sired (Booth-Binczik, 2001),
resulting in paternity confusion which would discourage sexually selected infanticide (Wolff and
Macdonald, 2004). Male coatis are also not excluded from bands in Tikal, and in fact they
frequently associate with bands there (Booth-Binczik, 2001) as well as in Arizona (Gilbert, 1973)
and Costa Rica (Sienz, 1994). Moreover, despite several hundred-possibly thousands of-hours
of field observations by several biologists at a few sites, no one other than Russell (1981) has
uncovered any evidence of infanticide in the species. To the contrary, in Tikal adult male coatis
were often seen in the near vicinity of young of all ages with neither the young nor their dams
displaying any alarm or other defensive behavior (Booth-Binczik, 2001).
In addition to cooperative defense, other forms of communal care of offspring have been
documented in the coati. Nest sharing has been observed both in the wild (Russell, 1979; 1983;
Chapter 2) and in captivity (Smith, 1980). In some instances it involved a female that had lost a
litter helping to care for another female's litter (Russell, 1979; Smith, 1980), and in one case two
females pooled their litters and both nursed all infants (Russell, 1979). There have also been
frequent observations of allogrooming by non-parents once bands have reunited after the nesting
period (Russell, 1983; pers. obs.).
As important as communal care of whatever form may ultimately be to coati offspring
survival, however, it does not appear to explain the species' remarkable birth synchrony, as a
rather basic prediction arising from the communal care hypothesis failed testing in the current
study. It is suspected but as yet unknown whether coatis undergo direct development (Chapter 2),
but females are presumably able to exert at least as much control over the timing of reaggregation
as parturition. Nonetheless, births were clustered significantly more than were reaggregation
events. Selective pressures for being in a group are evidently not the driving forces behind coati
Considering the aforementioned nest raiding by predators on coatis, the predator
swamping hypothesis might at first look somewhat more attractive. The effectiveness of this
anti-predation strategy, however, is determined not only by the degree of synchronization of a
vulnerable prey stage but also by the nature of the predators and their methods. That is, if
neighboring groups of prey differ somewhat in their timing and their predators are sufficiently
mobile, the predators will simply respond by shifting their focus from group to group accordingly
(Ims, 1990a). Considering the mobility of coati nestling predators (especially raptors) and given
the significantly greater variance in the timing of births between than within bands found in the
present study, shifting predators appear to pose a real threat to coati young. The predator
swamping hypothesis for birth synchrony is unsupported.
At least until new hypotheses and/or evidence are examined to account for coati birth
synchrony, then, the most likely explanation appears to be that this phenomenon is merely an
extreme example of a tropical animal using social cues to supplement the relatively weak
environmental predictors it relies upon for timing reproduction. Some support for this argument
is found in the results of the comparison of intra- and inter-band synchrony mentioned above.
Additional support can be found in a closer consideration of the coati's annual birth
pulse. As mentioned, both the 1995 and 1996 overall peaks in parturition were slightly
platykurtic, perhaps only differing from completely normal curves due to the method used for
estimating uncertain birthdates, i.e., attributing portions of each of these births to more than a
single day. The coati has ample cause in terms of food availability to have developed the
adaptation of reproductive seasonality (Russell, 1982; Chapter 2), which-barring severe climatic
curtailment (e.g., at very high latitudes) or protraction by variable response to weak
environmental cues (e.g., at low latitudes)-should produce a normally distributed biological
response. That is, if the coati's annual birth pulse is driven purely by seasonal considerations, it
should be normally distributed or flattened depending on the environmental predictors available
to the species and its ability to respond to them. Accordingly, most seasonally reproducing
tropical mammals studied to date display long, modestly peaked birth pulses (Fleming, 1973;
Bronson, 1989; Di Bitetti and Janson, 2000; Chapter 2), and the notable exceptions are highly
social species (Chapter 2). In contrast, the evolutionary forces that promote synchrony do so by
acting against either tail of a response curve (e.g., Findlay and Cooke, 1982), creating a
leptokurtic distribution. The coati's normally distributed parturition peaks suggest that birth
synchrony in and of itself is not adaptive in this species.
Given the likelihood of direct development in this species (Chapter 2), it is also possible
that the coati's unusually tight clustering of births is simply the result of selection for mating
synchrony. Synchrony of estrus among females should make it less possible for one male to
monopolize mating access to a group of females (Emlen and Oring, 1977; Gehrt and Fritzell,
1999a). It could therefore be advantageous for females to synchronize their receptivity if it gives
them access to more than a single male, enabling mate choice. However, some studies have
failed to find a relationship between degree of synchrony in female receptivity and degree of
monopolization by males (Poston et al., 1999; Widdig et al., 2004), and it has also been
suggested that synchronized receptivity would limit rather than enhance females' opportunity for
mate choice (Pereira, 1991; Schank, 2001). The nature of the relationship between mating
synchrony and mate monopolization/mate choice undoubtedly depends upon the degree of
overlap among females' periods of receptivity, which is not yet known for the coati.
Of course, it could be argued that to be most useful, the term "birth synchrony" should be
reserved solely for situations in which selective pressures are acting directly on the clustering of
births. From this viewpoint, the tight clustering of births exhibited by the coati would not qualify
as birth synchrony per se if it results from such phenomena as reproductive seasonality and
Left wholly unexplored is the question of precisely how the coati achieves such
extraordinarily tight reproductive seasonality. Assuming social facilitation of environmental cues
as appears to be the case, numerous possibilities exist. Among Mammalia social mechanisms for
timing reproduction have primarily been investigated in rodents. Although in most cases the
presence of chemical cues from reproductively active females inhibits ovarian cycling in others
(reviewed in Bronson, 1989), in some situations it has instead been found to trigger (Drickamer
and Hoover, 1979) or synchronize (McClintock, 1978) such cycling. Males have also been
shown to use both chemical and tactile means of initiating and/or synchronizing female
reproductive cycling (Whitten, 1956; Marsden and Bronson, 1964; Bronson and Maruniak, 1975;
Carter et al., 1986), and in the case of red deer (Cervus elaphus), even to use auditory signals
(McComb, 1987). On that last bit of information, it is perhaps worth noting that the coati is the
first carnivore to be identified in which males undergo a rut (Chapter 4), and that a hallmark
feature of this male adaptation is an advertising call. With opportunity, this will be a focus of
Long mistaken for a separate species due to its larger size and markedly different habits
(e.g., Sanderson, 1941), the male white-nosed coati (Nasua narica) has received far less attention
than has the female from researchers. This disparity in scientific effort is perhaps a result of the
chiefly solitary nature of adult males (Kaufmann, 1962; Booth-Binczik, 2001) and the problems it
poses to one wishing to work with these diurnal, semi-arboreal, medium-sized carnivores in their
primarily dense, tropical forest habitat (general ecology reviewed by Gompper, 1995). Previous
discussions of coati reproductive ecology addressed only the timing of reproductive events and
possible benefits of seasonality to females and young (Kaufmann, 1962; Smythe, 1970; Russell,
1982). This paper focuses instead on reproductive patterns displayed by the male of the species.
Environmental influences act on females in determining a species' timing of birth, and
males can be expected to respond to female imperatives or (in some instances) the environment
which shaped them with reproductive adaptations of their own. Males of many species are
opportunists, maintaining full reproductive readiness at all times for whatever mating chances
occur, but many other species have evolved temporal variation in male reproductive capabilities
in ways that reflect patterns of female receptivity. This variation seems designed both to
maximize the males' ability to capitalize on mating opportunities when they exist and to minimize
the costs associated with that ability when they do not. One such temporal pattern seen among
male mammals involves rut.
The term "rut" has often been used to describe a seasonal, principally or wholly
behavioral phenomenon displayed by males of some species or populations (e.g., the "fall rut" of
several high latitude North American ungulates; Spomer, 1996). Sometimes it has even been
used simply as shorthand for the mating period of any seasonally reproducing species (e.g.,
Genovesi et al., 1997). From an evolutionary perspective, however, rut might more properly be
thought of as an attribute of individual males rather than of seasons, species or populations as a
whole. Accordingly, the author defines rut as a brief annual period of dramatically enhanced
male reproductive capability that likely arose in response to tight female reproductive seasonality
in combination with the intense male-male competition such seasonality sometimes fosters, and
that therefore tends toward synchrony within a population. It differs from other forms of male
reproductive seasonality not only in the brevity and magnitude of the changes rutting males
undergo, but also in the nature of those changes. Specifically, males in rut display a fugue of
behavioral, physiological and anatomical characteristics that-despite some seeming maladaptive
at first glance-act in concert to temporarily maximize competitive ability and thus likely
Besides being interesting in their own right, male reproductive adaptations can also
provide insight into a species' overall reproductive strategy. One example is presented in the case
of mammals in which the males undergo seasonal aspermatogenesis. Of interest in studies of
species displaying seasonal reproduction is whether the pattern is a (relatively) rigid evolutionary
result of historical climatic rhythms or merely an opportunistic response to prevailing conditions,
i.e., whether it is an obligate or facultative strategy (Negus and Berger, 1972; Bronson, 1989).
Because spermatogenesis takes longer to complete than does ovulation after a period of
reproductive quiescence, and because males are evolutionarily impelled to have mature sperm
available whenever females might be ready to receive such, healthy adult male mammals have
generally evolved to produce mature sperm over time periods that more than encompass periods
of female receptivity (Bronson, 1989). Seasonal aspermatogenesis is thus primafacie evidence
that a species has adopted obligate reproductive seasonality, as such an extreme male adaptation
would be expected to arise only where females have a highly predictable pattern of receptivity,
especially where the potential costs of male reproductive capability are great.
The coati certainly fulfills the first part of this evolutionary scenario, being exceptional
among tropical mammals (Fleming, 1973) for its high degree of reproductive seasonality at the
few locales thus far examined (Kaufmann, 1962; Russell, 1982; Chapter 2). The pattern seen in
Tikal National Park, Guatemala, may be typical in this respect. There, virtually all copulation
occurs within an approximately two-week period near the middle of the January-April dry season
and parturition occurs in a similarly brief period beginning shortly before the onset of the rains in
May (Chapter 2).
There is also reason to hypothesize high costs of reproduction for male coatis. Unlike
males, female coatis are highly social, living in cohesive bands with their young most of the year
and becoming solitary only to nest and give birth (Kaufmann, 1962; Chapter 2). Combined with
the species' brief mating period, the result is a temporal and spatial clustering of receptive females
that could be expected to foster intense competition among males. Indeed, anecdotal
observations of male coatis by previous researchers suggest such competition exists. Chapman
(1938) and Kaufmann (1962) noted marked seasonal changes in male body condition. They and
Gilbert (1973) also attributed a seasonal proliferation of serious injuries among males to
intrasexual agonism during the mating period.
This study was undertaken to determine the extent and results of male reproductive
competition in the coati and to evaluate the hypothesis that male coatis have evolved adaptations
to maximize their competitive ability and minimize their associated costs. Specifically, year-
round data on male behavior, physiology and anatomy were collected to test the prediction that
male coatis exhibit periods of enhanced and reduced reproductive capability in a pattern which
reflects that of female receptivity.
As part of a broader study of white-nosed coati reproductive biology in Tikal National
Park, Guatemala (17N, 89W), behavioral, morphometric and endocrinologic data were
collected on 24 free-ranging adult male coatis (plus one found freshly road-killed) from June
1994 through October 1996 (except as noted below). A male was considered adult once it
dispersed from its natal band, which typically took place 1-2 months prior to the mating period in
its second year of life (i.e., at 20-21 months of age).
Each male was initially captured by live-trap or (more often) blowgun and chemically
immobilized by intramuscular injection of approximately 7 mg/kg Telazol (Fort Dodge
Laboratories Inc., Fort Dodge, IA). Anesthetized animals were tattooed and eartagged for
identification, weighed, measured and subjected to blood collection. Each male was also visually
examined to determine general physical condition, giving particular attention to the nature and
abundance of any injuries. Identified injuries were fresh scars, open wounds and missing body
parts. Eighteen of the coatis were additionally fitted with 90 g motion-sensitive radiocollars
(Advanced Telemetry Systems, Isanti, MN) to enable subsequent location and recapture, and to
confirm death if such occurred. Individual males were recaptured by blowgun (never more often
than monthly) for resampling. The research was conducted under Institutional Animal Care and
Use Committee protocol #4084 from the University of Florida. All animal handling procedures
were conducted in accordance with National Institutes of Health standards.
Four radiocollared males were habituated to the close presence of observers (see Chapter
2) and then utilized for behavioral data collection from September 1995 through August 1996.
Observation sessions were scheduled twice monthly for each available male (although frequent
radiotransmitter failures often resulted in fewer males being available at any given time), and in
each case began as soon as possible after locating the focal animal in the morning (typically
06:00-09:00) and continued for the next 4-6 h thereafter. Two observers were always present,
one of whom was responsible for monitoring and recording behavior and the other for estimating
the distance traveled by the focal animal via a tally counter as well as assisting in keeping the
animal in sight. The same people performed the same observational tasks throughout the study,
except that on eight occasions substitutes filled in for the second observer after cross-training and
calibration sessions were conducted to ensure uniformity in coati travel distance estimation.
During each observation session, the amount of time spent by the focal male with a coati
band versus alone was recorded, and the nature and duration of all interactions involving the male
were recorded. A male was considered to be with a band if the observer following him could see
any band member, typically less than 30 m distant. The frequency and/or duration of other
behaviors were recorded during the first 5 min of every quarter hour. The following states and
behaviors were utilized in the analyses: Active doing anything other than sitting or lying down.
Interaction two or more animals apparently altering their behavior in response to one another.
Agonism aggressive or defensive interaction. Fight agonistic interaction involving chasing,
knocking each other out of trees, or actual or attempted biting and/or clawing. Approach one
individual orienting toward and moving to within touching distance of another. Scentmark -
briefly (approximately 1 sec) rubbing the genital region against some substrate. Forage bringing
the nose into close proximity or manipulating various substrates in apparent search of
invertebrates or fallen fruit, or similarly investigating fruit-bearing portions of plants. Groom -
manipulating one's own body parts with mouth or paws. Travel moving from one place to
another with the head not oriented toward any substrate, i.e., not apparently foraging. Rut call -
producing a rhythmic, staccato, low-pitched vocalization which appeared to fall into Kaufmann's
(1962) "grunting" (p. 123) classification; distinguishable from similar vocalizations (e.g.,
grunting in alarm) by the combination of its slower cadence, greater loudness and much longer
duration (typically several minutes at a time) as well as by the situation in which it was produced.
Each male's teeth were visually examined for damage (such as broken tips) during each
immobilization, and intact lower canines were measured from gum line to tip with calipers. If
both teeth were undamaged, lengths (to the nearest 0. 1mm) were averaged; otherwise only the
length of the intact tooth was obtained. For comparative purposes, these same data were obtained
for the teeth of females immobilized during a companion study.
An index of each male's body fat level was determined at the time of immobilization.
This was accomplished in the manner of Hossler et al. (1994) by pinching a fold of the animal's
skin at the back of each rear leg midway between knee and hip, and using calipers to measure the
thickness of this fold to the nearest 0.1 mm. Measurements for the left and right thighs were
averaged. One person was responsible for taking these measurements throughout the study.
Testis size was also measured during each immobilization, and from the measurements
an index of testis size was determined. The scrotal skin was pulled taut around each testis and
three axes were measured with calipers to the nearest 0.1 mm: the longest dimension (L), the
widest perpendicular dimension at midlength (Wi), and the midlength dimension perpendicular to
both of these (W2). One person took all measurements. Opportunistic dissections of two freshly
road-killed coatis (one radiocollared, the other previously uncaptured) revealed a layer of fat
surrounding the testes ranging from scarcely visible on 7 June 1995 to as much as 5 mm thick on
9 January 1996. To compensate for this in approximating an animal's actual testis size, a
correction factor Y = 0.77X 2.23 was subtracted from each of the aforementioned dimensions,
with X being the fat level index of the animal at the time of examination. (For this purpose, the
fat level index of one coati for which leg pinch measurements were lacking was approximated
from a regression of all available fat data on time of year.) The above equation represents the line
connecting two points which were obtained by assigning the minimum and maximum fat level
indices recorded in the course of the study Ys of 0 and 10 (i.e., representing fat layers of 0 and 5
mm thick present on each side of a measured testis), respectively. This likely resulted in a
conservative correction of testicular dimensions, as 5 mm may have been a low approximation of
the maximum possible scrotal fat layer. The adjusted testicular dimensions were then plugged
into the formula for calculating the volume of an ellipsoid, V = L x W, x W2 x n/6, and the left
and right testis volumes so obtained were averaged to produce the testis size index reported. (The
raw data on coati testis size are presented in the Appendix.)
Five to 10 ml of venous blood were drawn from a foreleg of each immobilized coati.
Blood was kept in an ice bath from the time of collection until the serum could be separated by
centrifuge 2-4 h later and then frozen. Fecal samples were also collected whenever possible
during immobilizations, and additional samples were obtained opportunistically from males under
observation. In either case, an entire fecal deposit was collected within 5 min of defecation and
maintained at ambient temperature for 2-6 h, after which it was homogenized and a sample of up
to 10 ml was frozen. All serum and fecal samples were first held in liquid nitrogen and then at
the completion of the fieldwork were transferred to a -700C laboratory ultracold freezer for
storage until they were analyzed en masse by testosterone radioimmunoassay (RIA).
Prior to analysis, a malfunction of the ultracold freezer resulted in all serum and fecal
samples thawing for a period estimated to have been 1-2/2 weeks. The effect of this thawing on
the androgen content of the samples is not known. The endocrinologic data presented herein
should therefore be considered representative of the pattern but not necessarily of the actual
testosterone levels circulating in or excreted by coatis.
Serum Testosterone RIA. Serum testosterone concentrations were determined via an
RIA procedure (Guillette et al., 1996) utilized previously for a wide variety of species and
validated for coati serum. Duplicate 50 ll aliquots of serum were extracted twice with 2.5 ml -
300C ethyl ether, dried under filtered air and reconstituted in 100 tl borate buffer (0.5 M, pH
8.0). To these were added 100 pl bovine serum albumen solution (0.75% BSA in borate buffer),
200 pC testosterone antiserum (T3-125 at a final concentration of 1:18,750; Endocrine Sciences,
Calabasas Hills, CA) and 100 pl [1,2,6,7-3H]-testosterone radiolabel (TRK-921 diluted in borate
buffer to approximately 13,000 cpm/100 pl; Amersham Life Science, Inc., Arlington Heights,
IL). Assay tubes were then vortexed briefly and incubated overnight at 40C. Bound-free
separation was achieved by adding 500 pl 5% charcoal/0.5% dextran in phosphate-buffered saline
and then centrifuging the tubes for 30 min at 2000 g, 4C. The supernatant was decanted, diluted
with 5 ml scintillation cocktail (Scintiverse BD, Fisher Scientific, Fair Lawn, NJ) and counted on
a scintillation counter (Beckman LS 5801, Beckman Coulter, Inc., Fullerton, CA).
Counts-per-minute were averaged for each pair of duplicates, and these were subjected to
Rodbard and Lewald's (1970) logit transformation. Testosterone concentrations were determined
by comparison of transformed data to a logit-log plot of 10 standards, ranging from 3.125 pg to
1600 pg hormone per assay tube, which were also run in duplicate in each assay. Naturally low,
medium and high testosterone concentration serum pools were also run in duplicate in each assay
as a check on inter-assay variance. Final testosterone concentrations were calculated as ng/ml
serum, and are reported without correction for assay accuracy.
Two assays were conducted to analyze samples, and three more to complete validation.
For all sample pairs, the mean ( SEM) duplicate coefficient of variance (CV) was 3.5 0.4%,
and r > 0.99 for the standard curve fit in all assays. Non-specific binding averaged 2.6 0.6%.
Inter-assay CVs in low, medium and high pools of serum were 17%, 7% and 12%, respectively.
The minimum detectable concentration of testosterone, based on the mean 2SD percentage of
bound radiolabel observed in five pairs of charcoal-stripped serum, was 0.04 ng/ml. A standard
curve prepared by spiking charcoal-stripped serum (four pairs at each standard concentration) was
parallel to that made using borate buffer (test for homogeneity of slopes Ff1,6, = 0.14, P > 0.75);
assay accuracy as determined by spike-recovery was 102%. Serial dilutions (one pair each at 30
pl, 50 pl and 70 gll) of low, medium and high pools produced flat lines with CVs of 13%, 8% and
4%, respectively. Intra-assay CVs in low, medium and high pools (five pairs of each pool) were
6%, 6% and 7%, respectively. Cross-reactivities greater than 1% reported by the antiserum's
manufacturer were: dihydrotestosterone 44.0%, 8-1-testosterone 41.0%, 6-1-dihydrotestosterone
18.0%, 5a-androstan-3P,17p-diol 3.0%, 4-androsten-3, 17p-diol 2.5%, 8-4-androstenedione
2.0% and 50-androstan-3p,17p-diol 1.5%.
Fecal Testosterone RIA. Fecal testosterone concentrations were determined much as
described for serum, except fecal steroid extraction was accomplished via a slightly modified
version of the "short method" described in Wasser et al. (1994), and this procedure as well as the
subsequent assay were validated for coati feces. Fecal samples were lyophilized at -550C under
vacuum exceeding 100 x 103 mBar and then ground to fine powder (removing discernible
extraneous material such as seeds, insect parts, etc.). For each sample, 0.1 g of powder was
lightly boiled for 20 min first in 5 ml and subsequently in 2.5 ml of 90% ethanol, and the
supernatant was dried tnder filtered air and reconstituted in 1 ml absolute methanol. The
methanol suspension was diluted 100 pl to 7.4 ml borate buffer, and 100 gl of the resulting
solution were pipetted into duplicate tubes and treated as per reconstituted serum extract in the
RIA described above. The 10 standards utilized in the fecal RIA ranged from 1.5625 pg to 800
pg hormone per assay tube. Final concentrations were calculated as ng/g dried feces, and are
reported without correction for assay accuracy.
Samples were analyzed in two assays, and validation was completed in another four. The
mean CV for sample duplicates was 1.8 0.2%, and r > 0.99 for the standard curve fit in all
assays. Mean non-specific binding was 0.6 0.1%. Inter-assay CVs in low, medium and high
fecal pools were 7%, 9% and 8%, respectively. The minimum detectable concentration of
testosterone, based on the mean 2SD percentage of bound radiolabel observed in six pairs of a
naturally very low testosterone fecal pool, was 17 ng/g. A standard curve prepared by spiking the
very low pool (four pairs at each standard concentration) did not differ statistically from that
made using borate buffer (test for homogeneity of slopes F[,.16] = 0.58, P > 0.5); spike-recovery
accuracy was 111%. Serial dilutions of a medium pool (one pair each at 50 ptl, 75 pll, 100 pl, 125
pl and 150 pl methanol suspension diluted in borate buffer to a total of 7.5 ml solution) produced
a flat line with a CV of 8%. Intra-assay CVs in low, medium and high pools (five pairs of each
pool) were 3%, 6% and 1%, respectively.
Observations obtained in companion studies of female coatis conducted over the entire 2
'/2-year sojourn in Tikal (Booth-Binczik et al., 2004; Chapter 2) were used to delineate three
distinct annual periods to aid in data presentation and analysis:
The non-rutperiod comprised the nearly 11 months of the year that occurred outside of
the pre-mating and mating periods, from 29 February or 1 March (depending on year) through 22
The pre-mating period comprised the three weeks leading up to the mating period
during which multiple male coatis (as opposed to one or none) were regularly detected
accompanying a given band at a given time, from 23 January through 12 February.
The mating period comprised the 16 days during which copulations were noted, from
13 February through 28 February.
In some analyses data for the pre-mating and mating periods were combined into a single
rut period which was then compared to the non-rut period.
Whenever possible, each male was sampled repeatedly within each period every year, but
was in general represented by only a single data point per period in a given statistical test. This
was accomplished by averaging a male's subsamples where collection effort was equal (e.g., in
the case of multiple immobilizations) and adding subsamples where it was not (e.g., in the case of
multiple observation days of varying lengths). The exception was in evaluating the correlation
between serum and fecal testosterone concentrations, in which all sample pairs (i.e., serum and
feces collected from the same animal at the same time) were treated as independent; individual
contributions to this dataset ranged from one to five sample pairs per represented male.
Sokal and Rohlf (1981) guided all statistical efforts, and STATGRAPHICS Plus 5.1
software (Manugistics, Inc., Rockville, MD) aided computations. Datasets for comparing the
three periods first underwent a battery of tests for homoscedasticity and normality (using a
conservative Pa < 0.10 level of rejection), and were subjected to standard transformations if such
were indicated. When raw or transformed data met parametric assumptions, two-way ANOVAs
were conducted treating period and male as main effects (male effects were never found to be
significant and are not discussed further); otherwise the data were analyzed via Kruskal-Wallis
tests. Fisher's Least Significant Difference procedure was used to distinguish among periods
when overall significance was detected. Two-tailed t-tests were conducted instead of ANOVAs
in the few cases where only two groups were compared. A P,, < 0.05 level of significance was
utilized in all reported statistical tests.
Males interacted on 26 (90%) of the 29 occasions when they came within sight of one
another during scheduled observations. All male-male interactions quickly escalated into
relatively brief but intense physical altercations with few or no apparent preliminaries. Male
coatis fought sporadically throughout the year, but the frequency of such contests increased
during the pre-mating period and peaked during the mating period (Fig. 4-1). The frequency of
male-male agonism in the three defined periods of non-rut, pre-mating and mating could not be
statistically distinguished (H14.4.3 = 4.66, P = 0.097). However, focusing instead simply on a
comparison of non-rut to rut (the latter being pre-mating and mating combined) revealed fighting
to be significantly more common during rut (0.41 0.18 fights/hr versus 0.04 0.01 fights/hr
during non-rut; paired tt31 = 3.63, P = 0.036).
0 '-- T '---------
Non-Rut Pre-Mating Mating
Figure 4-1. Mean ( SEM) frequency of intrasexual agonistic interactions and injuries
observed among adult male white-nosed coatis in Tikal National Park, Guatemala,
September 1995 September 1996 and June 1994 October 1996, respectively.
Number of males sampled in each of the three defined periods (non-rut: 29 Feb/1 Mar -
22 Jan; pre-mating: 23 Jan 12 Feb; mating: 13 Feb 28 Feb) indicated on bars.
The incidence of injuries noted during immobilizations rose sharply during the mating
period (Fig. 4-1; H18,15,6J = 8.74, P= 0.013). Injuries most often affected the head, chest or
forelimbs and were frequently severe, including badly broken or missing teeth (especially
canines) and missing claws or entire toes, as well as a variety of deep flesh wounds. Remarkably,
none of the examined animals died as a direct consequence of such injuries, but two of the three
natural deaths recorded among radiocollared males during the course of the study occurred in
February 1995, i.e., during rut of that year; the third death occurred in August 1996.
Unfortunately it was not possible to reach the carcasses quickly enough to determine the exact
timing or causes) of death.
A comparison of male and female coati dentition revealed the males' lower canine teeth
to be substantially longer (15.5 0.4 mm for 22 males versus 8.7 0.2 mm for 49 females; t691 =
18.40, P< 0.001).
Males spent significantly more time with bands during the mating period than at other
times of the year (Fig. 4-2; F12.5] = 6.16, P = 0.045). Periods did not differ in terms of males
approaching females either quantitatively (Fig. 4-2; F[2.5] = 0.57, P = 0.618) or qualitatively;
regardless of time of year such approaches were most often rebuffed-sometimes violently.
Occasionally females allowed brief perineal sniffing by males, however, or less often investigated
the males in a similar fashion themselves. Females also sometimes initiated or reciprocated
allogrooming with males. These affiliative behaviors were infrequent and sporadically
distributed throughout the year.
Males scentmarked at a high rate throughout the year against the base of a living or fallen
tree or on a log or some other relatively low-lying feature of the terrain being traversed; they also
occasionally climbed into and traveled through the forest canopy seemingly just to repeatedly
scentmark the large, lower branches along some distance before returning to the ground.
Although scentmarking was highest during the mating period, no time of year could be
statistically distinguished from another (Fig. 4-3; F[2.5] = 2.60, P = 0.168).
Non-Rut Pre-Mating Mating
Figure 4-2. Mean ( SEM) percentage of time adult male white-nosed coatis spent with
bands and frequency with which males approached females in Tikal National Park,
Guatemala, September 1995 September 1996. Number of males sampled in each of the
three defined periods (non-rut: 29 Feb/1 Mar 22 Jan; pre-mating: 23 Jan 12 Feb;
mating: 13 Feb 28 Feb) indicated on bars.
a 40- 3
S 6 -
Non-Rut Pre-Mating Mating
Figure 4-3. Mean ( SEM) frequency of scentmarking and percentage of active time
spent calling by adult male white-nosed coatis in Tikal National Park, Guatemala,
September 1995 September 1996. Number of males sampled in each of the three
defined periods (non-rut: 29 Feb/i Mar 22 Jan; pre-mating: 23 Jan 12 Feb; mating:
13 Feb 28 Feb) indicated on bars.
The rut call was heard only during the pre-mating and mating periods (especially the
latter; Fig. 4-3; H[4.4.3] = 6.14, P = 0.046) and elicited close investigation by both females and
males. The area occupied by Tikal's large bands (up to 162 females and maturing young; Booth-
Binczik, 2001)-and thus also the males accompanying them-as the animals moved through the
forest precluded two human observers quantifying many aspects of the mating pattern,
particularly when those observers were concentrating primarily on the behavior of a single focal
animal. However, it was evident that up to several males (at least four could be conclusively
identified in one instance) were calling at the same time from individual perches in the lower
canopy directly or peripherally over a given band foraging on the ground below, changing their
perches as necessary to keep up with the movements of the band. On at least six occasions
individual females were seen leaving their bands and going to calling males, and at least three
times the females stayed with the males to copulate. Calling males usually appeared to ignore
one another, but on at least two occasions individual males were also seen briefly leaving their
calling stations to chase or fight with others calling nearby; most recorded fights occurred when
males were not calling.
Males in Tikal spent significantly less time grooming during the mating period than at
any other time of year (Fig. 4-4; H[4,4,3] = 6.66, P = 0.036). Periods did not differ appreciably in
terms of time males spent foraging or distances males traveled (Fig. 4-4; for foraging H[4.4,3] =
2.64, P = 0.268; for traveling F[2,5] = 3.36, P = 0.119).
Body fat levels measured in immobilized males rose steadily throughout most of the year
but fell sharply at the beginning of the pre-mating period, ultimately reaching a low at the end of
the mating period that was only 20-25% of the peak achieved at the end of the non-rut period
(Fig. 4-5). Fat levels were significantly lower during mating (3.8 0.6 mm versus 7.8 + 0.7 mm
and 7.7 0.8 mm during non-rut and pre-mating, respectively; H[18,15,61 = 11.27, P = 0.004).
Figure 4-4. Mean ( SEM) percentage of active time adult male white-nosed coatis spent
grooming themselves and foraging, and their rate of travel in Tikal National Park,
Guatemala, September 1995 September 1996. Number of males sampled in each of the
three defined periods (non-rut: 29 Feb/I Mar 22 Jan; pre-mating: 23 Jan 12 Feb;
mating: 13 Feb 28 Feb) indicated on bars.
The testes of immobilized males increased rapidly in volume with the beginning of the
pre-mating period, peaked late in pre-mating or early in the mating period, decreased nearly as
rapidly throughout the remainder of mating and then continued decreasing more slowly well into
the non-rut period (Fig. 4-6). The males' testes were significantly smaller during non-rut (1.8 +
0.2 x 103 mm3 versus 2.9 0.3 x 103 mm3 and 3.3 0.2 x 103 mm3 during pre-mating and
mating, respectively; F[2.13] = 8.85, P = 0.004). Histological examination of testicular
parenchyma obtained from road-killed males on 7 June 1995 and 9 January 1996 revealed the
presence of all cell types-including mature spermatids-in the seminiferous tubules at both times
Serum and fecal testosterone concentrations were highly correlated (r = 0.86 for 28
sample pairs from 16 males; t[261 = 8.42, P < 0.001) and displayed a clear circannual pattern (Fig.
4-7). Specifically, testosterone levels rose gradually beginning several months prior to the pre-
mating period, spiked suddenly to a peak near the end of pre-mating and then fell even more
suddenly back down to baseline by the end of the mating period. The three periods of the year
differed significantly whether serum and fecal testosterone values were treated separately or
combined (for combined data in which serum values (TI) were converted to fecal equivalents (Ty)
via the regression formula Ty= 52.97T, + 51.07, non-rut was 172 19 ng/g, pre-mating 549 63
ng/g and mating 358 106 ng/g; H[19,16,8] = 16.66, P < 0.001).
Lacking the showy weaponry seen in many rutting species (such as antlers, horns and
tusks), the male coati nonetheless appears well-equipped for intrasexual combat. Both sexes rely
on their teeth as well as their powerful forelegs and long, sharp claws when fighting, but the
male's dentition may be specially designed for this purpose given the degree of lower canine
tooth sexual dimorphism observed in the present study.
Wet Season Dry Season
S 12 o o
pi oo o
S- 0 0o 0 0
0. 4 o o g) o o
Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun
Figure 4-5. Annual body fat profile exhibited by adult male white-nosed coatis in Tikal
National Park, Guatemala, July 1994 through October 1996. Based on hind leg pinch
thickness measured in 42 immobilizations of 18 animals during non-rut period unshadedd
background; 29 Feb/I Mar 22 Jan), 16 of 15 animals during pre-mating (lightly shaded;
23 Jan 12 Feb) and 6 of 6 animals during mating (more darkly shaded; 13 Feb 28 Feb).
Wet Season Dry Season
0 0 00
o o o o o
S000 0 O O
0 GD CP
Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun
Figure 4-6. Annual profile of testicle size exhibited by adult male white-nosed coatis in
Tikal National Park, Guatemala, June 1994 through October 1996. Individual testis
measurements were corrected for the animal's estimated testicular fat layer (see Materials
and Methods) prior to ellipsoid volume calculation. Profile based on 45 immobilizations
of 19 animals during non-rut period unshadedd background; 29 Feb/1 Mar 22 Jan), 16 of
15 animals during pre-mating (lightly shaded; 23 Jan 12 Feb) and 6 of 6 animals during
mating (more darkly shaded; 13 Feb 28 Feb).
S Wet Season Dry Season
Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun
Figure 4-7. Annual serum and fecal testosterone (T) profiles exhibited by adult male
white-nosed coatis in Tikal National Park, Guatemala, July 1994 through October 1996.
Non-rut period (29 Feb/I Mar 22 Jan) indicated by unshaded background, pre-mating
(23 Jan 12 Feb) lightly shaded and mating (13 Feb 28 Feb) more darkly shaded. Serum
profile based on 42 samples from 18 animals during non-rut, 16 from 15 animals during
pre-mating and 6 from 6 animals during mating; fecal profile based on 70 samples from
13 animals, 8 from 8 animals and 8 from 5 animals, respectively. Twenty-eight paired
samples (i.e. serum and feces collected from same animal at same time) from 16 animals
represented by solid symbols. Curve visually fitted to all combined and smoothed data.
Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun
Other than their potential usefulness in combat, no explanation for the male coati's
enlarged canines is apparent. Although Smythe (1970) speculated that males on Barro Colorado
Island (BCI), Panama, might pursue larger prey than females, this idea was not borne out in
subsequent studies at that locale (Gompper, 1996) nor did it appear to pertain in Costa Rica
(SAenz, 1994) or Tikal (Booth-Binczik, 2001). In a comparative analysis of sexual dimorphism in
carnivore dentition, Gittleman and Van Valkenburgh (1997) likewise found that dietary
differences between the sexes were reflected in the size of the carnassials rather than the canines.
Instead, Gittleman and Van Valkenburgh (1997) determined that carnivore canine tooth
dimorphism correlated with mating system, with more polygynous species exhibiting greater
dimorphism. Such species likely experience particularly intense competition among males for
mates (Emlen and Oring, 1977), suggesting the enlarged canines' principal role lies therein. It
may be relevant in this respect to note that the ratio of male:female lower canine size found
among coatis in the present study (1.78) exceeded that displayed by any of the 45 carnivore
species Gittleman and Van Valkenburgh (1997) examined, and that the highest ratio they
discovered (1.62) belonged to Nasua nasua, a South American congener of the white-nosed coati
which has been much less studied but is believed to be similar in many respects (Emmons, 1990).
Based on the frequency with which Tikal's coati males were observed fighting and the
frequency and severity of the injuries they sustained presumably as a result, it appears that males
of this species do indeed face considerable and costly intrasexual competition. All male-male
interactions observed in the present study were agonistic. Most male-male agonism occurs during
the rut and, judging from the data on injuries, is most violent during the mating period.
Kaufmann (1962) reported that all male-male coati encounters he saw while conducting
observations of females on BCI were similarly agonistic, but stated that actual physical contact
was uncommon outside of the mating period, when most fights occurred. Chapman (1938) and
Kaufmann (1962) attributed a seasonal proliferation of injuries among BCI's male coatis to
intrasexual agonism associated with mating, and Gilbert (1973) reported a similar phenomenon
among male coatis in Arizona. Chapman (1938) and Kaufmann (1962) also detailed the severity
of the animals' injuries, noting their prevalence on the heads and forelimbs, and remarked upon
the animals' exceptional recuperative powers. Such healing ability notwithstanding, Chapman
(1938) reported the permanent partial blindness of a semi-tame male under his casual observation
after one such season, and believed the animal's ultimate disappearance two years later may have
been the result of it succumbing to injuries obtained in a subsequent mating period.
Male-male combat is also common-and often results in the injury or death of
combatants-in rutting species. Indeed, it might even be considered a hallmark of this
reproductive pattern. As in the present study, fighting among males of such species generally
spikes shortly before and during their brief annual mating periods (Wilkinson and Shank, 1976;
Clutton-Brock et al., 1982; Komers et al., 1994), and in some cases appears to cause a substantial
proportion of adult male mortality (Leslie and Jenkins, 1985; Hall-Martin, 1987; Poole, 1989). In
contrast, male-male agonism in carnivore species other than the coati tends to be centered on
territory establishment (e.g., tigers, Panthera tigris, Smith, 1993) or acquiring control of a social
group (e.g., lions, Panthera leo, Bygott et al., 1979; Packer and Pusey, 1982) before mating
opportunities arise, and conflicts during the mating period itself are less common.
Bands of female coatis were typically the focal points of male-male encounters in Tikal,
and to some extent the greater incidence of fighting among males during the mating period might
simply reflect their increased opportunity to engage in such activity at that time. The present
study revealed that males spend more time in the company of bands during the mating period than
at other times of the year, and the number of males accompanying a given band at a given time
varies accordingly; whereas bands were usually seen with only one (but not necessarily the same
one) or no male in attendance throughout most of the year, during the mating period as many as
nine males have been simultaneously observed with a single band (Booth-Binczik et al., 2004).
Kaufmann (1962) also noted an increase in male-band spatial affinity at about the time of
reproduction among coatis on BCI, but with some differences. He believed that males there only
rarely associated with bands outside of what he considered to be the coatis' month-long mating
period (essentially a combination of the pre-mating and mating periods of the present study), and
that one particular male remained with each band almost constantly throughout that time, perhaps
copulating with all females therein. (It is unclear whether he ever observed successful
copulation; Chapter 2.) Kaufmann (1962) acknowledged, however, that unobtrusive males might
have come and gone undetected because only the bands were habituated to his presence and
served as the foci of his observations. That admonition was later borne out by Gompper et al.
(1997), who found that the males most commonly seen with bands on BCI during the mating
period nonetheless did not sire most of the young subsequently produced by females in those
bands. Booth-Binczik (2001) further demonstrated multiple paternity not only within bands but
also within individual females' litters in Tikal.
One might expect that male coatis would capitalize on their closer association with bands
during the mating period by directly initiating interaction with the females therein more
frequently, but this appeared not to be the case in Tikal. There, males approached females least
during the mating period, and their approaches were generally rebuffed then just as they were at
other times of the year.
This finding contrasts somewhat with Kaufmann's (1962) report that males in association
with bands during the coati mating period on BCI spent much of their time pursuing females, and
were more often amicably received by females then compared to other times of year. He
described almost all close male-female encounters occurring outside of the mating period as
hostile. Intersexual perineal sniffing and allogrooming were reportedly especially common at the
beginning of the mating period. Russell (1981) observed less agonism and more allogrooming
between the sexes not only during BCI's mating period but also for two months prior, which he
called a period of courtship, and reported friendly male-female encounters at other times of year
as well. In Arizona, Gilbert (1973) believed that males required acceptance by bands in a social
context prior to the mating period, simply approached and mounted receptive females during that
period (though it is unclear whether any successful copulations were actually observed; Chapter
2), and remained with bands for months thereafter.
It is well established that males of rutting species spend more time with females while
rutting than at other times of year ungulatess reviewed by Main et al., 1996; Ruckstuhl and
Neuhaus, 2000; African elephants, Loxodonta africana, by Poole, 1987). They are quite variable,
however, in the strategies they employ to secure access to receptive females. Some maintain
harems (Struhsaker, 1967; Clutton-Brock et al., 1982; Lovari and Locati, 1991), others follow
and defend individual females (Lent, 1965; Lott, 1981; Lovari and Appollonio, 1994), and others
defend territories and wait for females to enter (Jarman, 1979; Miura, 1984; Schuster, 1976).
At least in Tikal, the male coati appears to depend more on advertising than on directly
pursuing receptive females. Whether its scentmarking behavior exemplifies this strategy remains
as yet unclear. Although the frequency of scentmarking behavior did not vary over the course of
the year, the true significance of this behavior may lie in its quality rather than quantity. An
unusually strong odor was incidentally noted during some male immobilizations and observations
during the mating period. The author therefore believes that scentmarking may yet be found to be
an important component of the male coati's rut.
Lending some support to this contention, Kaufmann (1962) remarked upon male coatis
on BCI "urine-rubbing" (p. 130) at the base of or in trees, especially as the males approached
bands of females during the mating period. Technically speaking, though, it should be noted that
it has not yet been determined whether males of this species employ urine, some other products)
or a combination thereof in scentmarking. Indeed, the procyonids have been studied far less than
some other carnivore families in this regard, but males of the coati's congener N. nasua reportedly
behave similarly (B. Hirsch, pers. comm.) and have sebaceous preputial glands that are believed
capable of producing copious secretary material (Shannon et al., 1995).
Scentmarking is considered ubiquitous among terrestrial mammals (Gosling and Roberts,
2001), but an accentuated form/degree of it may be another hallmark of rut. Typically,
specialized glands suddenly proliferate or activate in the rutting male (e.g., the poll glands of the
camel, Camelus dromedarius, Rai et al., 1996; temporal glands of the Asian elephant, Elephas
maximus, Eisenberg et al., 1971; a variety of glands in cervids, reviewed by Miiller-Schwarze,
1987), and the animal anoints itself and/or various natural signposts with the glands' redolent
products. Urine is similarly used by rutting males of many species (Poole and Moss, 1981;
Clutton-Brock et al., 1982; Miura, 1984), and at least in the case of the fallow deer (Dama dama),
its particularly strong odor appears to be due to the contribution of preputial glands that activate
at that time (Kennaugh et al., 1977).
More readily detected by an observer in Tikal than any coati odor is the male's rut call.
This behavior features prominently in the species' mating pattern there, which has been described
as a kind of mobile lek (Booth-Binczik et al., 2004). Nothing resembling such has yet been
reported for coatis outside of Tikal. However, Kaufmann (1962) noted that males interacting
during BCI's mating period often emitted a "loud, rapid grunting" (p. 125), and on one occasion
during that period he witnessed a male following and similarly vocalizing from a variety of
perches near a band. Hass and Roback (2000) also reported that the male of a copulating pair of
coatis in Arizona made what they interpreted as "alarm grunts" (p. 329) toward an approaching
female. These observations notwithstanding, it seems unlikely that such a flamboyant
phenomenon would have thus far been overlooked, particularly on BCI where so much time has
historically been invested in studying this and other species' behavior. It may be that lekking by
male coatis in Tikal represents a behavioral extreme compared to other populations which have
Plasticity in mating patterns may be the rule among ungulates that lek. Fallow deer, for
example, utilize at least seven different mating systems under different ecological conditions
(Langbein and Thirgood, 1989), and topi (Damaliscus korrigum; Gosling, 1991) and blackbuck
(Antilope cervicapra; Isvaran, 2005) show almost equal flexibility.
Advertising calls may be the single characteristic most strongly identified with rutting
mammals. In fact, the word "rut" originates in the Latin rugire, meaning "to roar". Accordingly,
such calls have been widely reported and are evident in all types of mating systems: territoriality
(Estes, 1969; Miura, 1984), harem-holding (Struhsaker, 1967; Lovari and Locati, 1991), and
following (Poole, 1987; Berger and Cunningham, 1991).
It is a matter of debate to what extent either olfactory or auditory signals are directed at
potential mates versus rival males. Most researchers have concluded that both rut calls and
scentmarks serve to notify rivals of a male's physical condition and the attendant risks of
challenging him (e.g., Clutton-Brock and Albon, 1979; Bowyer and Kitchen, 1987; Berger and
Cunningham, 1991). Conversely, Hurst and Rich (1999) argued that the primary function of
scentmarking is to advertise quality to potential mates. There is evidence supporting both points
of view. McElligott and Hayden (1999) found that the vocalization rate of rutting fallow deer
was affected by whether other males were nearby, but not by whether the calling male was about
to copulate or had just done so. In studies of elephants (which display an asynchronous rut
known as musth), however, female African elephants preferred to mate with musth males (Moss,
1983) and apparently used both urine trails and musth rumbles to locate them (Poole and Moss,
1989), and female Asian elephants were able to identify males in musth from the males' urine
(Schulte and Rasmussen, 1999), and the age and musth phase of males from the males' temporal
gland secretions (Greenwood et al., 2005). Additionally, both olfactory (Coblentz, 1976) and
auditory (McComb, 1987) signals from rutting males have been shown to bring females of some
species into estrus. Of course, whether the proximate function is to attract receptive females or
repel competing males (or both), male advertising ultimately functions to increase mating
success; for example, of the many factors considered by McElligott et al. (1999), the amount of
time rutting fallow deer spent groaning was most highly correlated with mating success.
In addition to the risk of injury and death from combat, the male coati may pay a variety
of less apparent costs to support its activities during rut. Coati males in Tikal spend less time
grooming themselves during the mating period than at other times of year. The other relevant
activities monitored in the present study (foraging, traveling) did not differ with time of year, but
year-round male body fat profiles demonstrate a clear negative shift in the animals' energy
balance during the rut. Female coatis display a similar shift at that time (Chapter 2), however,
and the respective roles of reproductive activities and resource availability remain unclear.
No relevant data are available on male coatis at other locales. Profound variation in body
fat or weight is seen in the coati's northern temperate relative, the raccoon (Procyon lotor), but
tends to be more gradual; it is unknown as yet what if any influence mating period may have on
the male raccoon's circannual pattern (Mech et al., 1968; Dunn and Chapman, 1983; Moore and
Kennedy, 1985; Davison, 1993; Gehrt and Fritzell, 1999b).
The indirect costs of rut are well documented in many ungulate species (see the partial
albeit lengthy list of examples provided in Mysterud et al., 2004). Looking at such costs
individually, the effects of rut per se on grooming behavior have not previously been examined,
but Mooring et al. (1996) found that territorial male impalas (Aepyceros melampus) groomed less
while rutting than did females or males that were not holding territories, and the territorial males
hosted more ticks as a result. Typically, rutting males sharply decrease or eliminate foraging
(Mitchell et al., 1976; Miquelle, 1990; Perez-Barberia et al., 1998; Pelletier, 2005) while
increasing their general activity levels (Clutton-Brock et al., 1982; Fint and Krzywinski, 1997).
As a result they experience a sudden decline in body weight and fat levels (Mitchell et al., 1976;
Dunham and Murray, 1982), in some cases even as a season of food scarcity approaches
(Miquelle, 1990; Kolle et al., 1993).
The male coati is obviously engaging in a seasonal pattern of reproduction
complementary to that of the female, but how committed is it to that strategy? Changes in
gonadal tissue are clearly involved, as demonstrated by the circannual pattern males displayed in
testis size. The finding that mature sperm are being produced both soon before and several
months after the mating period, however, suggests that at least some level of fertility persists over
a prolonged period if not year-round. Male reproductive capability beyond rut was also
demonstrated by the fact that at least five females that lost their litters not long after parturition
were apparently fertilized again more than three months after rut's conclusion, resulting in them
giving birth to second litters (Chapter 2). Testicular data are lacking for coatis outside of Tikal.
Circannual cycles of testicular regression and recrudescence are normally observed in
rutting species (included in mammalian review by Lincoln, 1989). In most species thus far
examined these cycles include a period of cessation in spermatogenesis and a brief period of
maximal sperm production coincident with rut (Robinson et al., 1965; Lincoln, 1971; Chaplin and
White, 1972; West and Nordan, 1976; Marchlewska-Koj and Kruczek, 1988; Reyes et al., 1997).
In other (typically lower latitude) species the males remain more or less competent year-round
(Abdel-Raouf et al., 1975; Brown et al., 1991; Monfort et al., 1993; Willard and Randel, 2002).
In either case it appears that the rapid testicular proliferation associated with rut serves
mainly to increase androgenesis (see Brown et al., 1991; Reyes et al., 1997), prompting rut
behavior and the development or activation of various structures in its support. The role
androgens play in the social aggression of rutting males has received the most attention
(Jainudeen et al., 1972; Lincoln et al., 1972; Yagil and Etzion, 1980; Mossing and Damber, 1981;
Bouisson, 1983; Poole et al., 1984; Pelletier et al., 2003; Mooring et al., 2004), but these
hormones have also been linked to cutaneous glandular secretion and scentmarking (Jainudeen et
al., 1972; Yagil and Etzion, 1980; Mossing and Damber, 1981; Poole et al., 1984), calling
(Clutton-Brock and Albon, 1979; Yagil and Etzion, 1980; Li et al., 2001) and diminished
foraging (McMillin et al., 1980; Newman et al., 1998) and grooming (Kakuma et al., 2003).
Many if not all of the phenomena displayed by the rutting coati may be under androgenic
control as well. The patterns observed in coati testis size and testosterone level are similar to
those reported for numerous northern temperate ungulate species (Short and Mann, 1966;
Lincoln, 1971; McMillin et al. 1974; West and Nordan, 1976; Schams and Barth, 1982; Asher et
al., 1989; Asher and Peterson, 1991; Ditchkoff et al., 2001; Mooring et al., 2004; Martinez-Pastor
et al., 2005).
In the two procyonids previously studied (also northern temperate species), the raccoon
and ringtail (Bassariscus astutus), testis size also peaked just prior to or early in the species'
annual mating periods, but the processes of recrudescence and regression were more gradual and
(near) maximum size was maintained for several months (Sanderson and Nalbandov, 1973; Dunn
and Chapman, 1983; Poglayen-Neuwall and Shively, 1991; Davison, 1993). Both of these
species also displayed a four- to five-month testicular nadir involving a marked reduction or
cessation in sperm production (Sanderson and Nalbandov, 1973; Dunn and Chapman, 1983;
Poglayen-Neuwall and Shively, 1991). Surprisingly, only scant and unclear endocrinologic data
exist for the male raccoon. Davison (1993) reported only a broad, low peak in testosterone (the
maximum serum concentration recorded was 3.69 ng/ml) roughly centered around the species'
late winter mating period in the southeastern United States, and Kaneko et al. (2005) reported that
testosterone concentrations were highest in the fall (achieving in the vicinity of 80 ng/ml) in feral
animals in Japan; both studies were hampered by small, sporadically obtained samples.
The reproductive pattern displayed by the male coati at Tikal is unusual in two regards:
First, the male coati's rut is more akin to that of many ungulate species than to the pattern
of any other carnivore species examined to date. The highly social nature and strong reproductive
seasonality of female coatis foster intense competition among males. Male coatis have