Variation in the Resumption of Cycling and Conception by Fecal Androgen and Estradiol Levels in Female Northern Muriquis (Brachyteles hypoxanthus)

American Journal of Primatology 67:69–81 (2005)
RESEARCH ARTICLE
Variation in the Resumption of Cycling and Conception by
Fecal Androgen and Estradiol Levels in Female Northern
Muriquis (Brachyteles hypoxanthus)
KAREN B. STRIER1n and TONI E. ZIEGLER2
1Department of Anthropology, University of Wisconsin–Madison, Madison, Wisconsin
2National Primate Research Center, University of Wisconsin–Madison, Madison,
Wisconsin
We measured fecal androgen (T+DHT) and estradiol (E2) levels in female
northern muriquis (Brachyteles hypoxanthus) at the Estac¸a
˜o Biolo
´gica de
Caratinga/RPPN Feliciano Miguel Abdala, Minas Gerais, Brazil, to
evaluate the hormonal bases underlying individual variation in the
resumption of cycling and conception. We found that androgen levels
were significantly lower in females than in males, and that there were no
consistent patterns in female androgen levels across precycling or cycling
conditions. Females that resumed cycling earlier in the study (weeks 4–8)
had higher precycling E2 levels and correspondingly lower precycling
androgen/E2 ratios than females that resumed cycling later (weeks 12–
16). There were no differences in female precycling androgen levels, but
cycling females that conceived during or immediately after the study
period had lower androgen levels and threefold higher E2 peaks than the
one cycling female that failed to conceive. These results suggest that
minimum E2 thresholds are necessary for both the resumption of ovarian
cycling and conception. Individual variation in these components of
fertility may be regulated by differences in E2 levels, which affect
androgen/E2 ratios, rather than by androgen levels per se. Further
research into the relative concentrations of T vs. DHT will be necessary
to fully evaluate whether androgens affect cycling and conception in this
species. Am. J. Primatol. 67:69–81, 2005.
r 2005 Wiley-Liss, Inc.
Key words: female androgens; estradiol; northern muriqui; Brachyteles
hypoxanthus; fertility; DHT
Contract grant sponsor: Liz Claiborne and Art Ortenberg Foundation; Contract grant sponsor:
Margot Marsh Biodiversity Foundation; Contract grant sponsor: Graduate School, University of
Wisconsin–Madison; Contract grant sponsor: National Primate Research Center, University of
Wisconsin–Madison.
nCorrespondence to: Karen B. Strier, Department of Anthropology, University of Wisconsin–
Madison, 1180 Observatory Drive, Madison, WI 53706. E-mail: kbstrier@wisc.edu
Received 14 September 2004; revised 10 December 2004; revision accepted 18 December 2004
DOI 10.1002/ajp.20170
Published online in Wiley InterScience (www.interscience.wiley.com).
r 2005 Wiley-Liss, Inc.

---

70 / Strier and Ziegler
INTRODUCTION
Interest in androgen levels in female mammals has focused on species such as
spotted hyenas (Crocuta crocuta) [Dloniak et al., 2004; Drea et al., 2002; Goymann
et al., 2001; Place et al., 2002], ring-tailed lemurs (Lemur catta) [von Engelhardt
et al., 2000], and redfronted lemurs (Eulemur fulvus rufus) [Ostner et al., 2003] in
which adult females exhibit masculinized genitalia, large body size, and social
dominance over males. The masculinized morphological and behavioral traits of
females in these species have been attributed to their exposure to elevated
androgen levels during fetal development. As adults, females in these species have
lower testosterone (T) and dihydrotestosterone (DHT) concentrations than males
[Dloniak et al., 2004; Goymann et al., 2001; von Engelhardt et al., 2000].
Experimental manipulations of androgen antagonists in pregnant female
rhesus macaques (Macaca mulatta) have shown that some prenatal exposure to
androgens is necessary for normal sexual differentiation in the genital
morphology and behavior of females [Herman et al., 2000; Tomaszycki et al.,
2001]. In redfronted lemurs, high prenatal androgen/estrogen ratios, rather than
absolute androgen levels per se, appear to be responsible for female masculiniza-
tion [Ostner et al., 2003]. Excessive exposure to prenatal androgens has also been
linked to androgen-related infertility in adult females. In women with polycystic
ovary syndrome (PCOS), hypersecretion of ovarian androgens during or before
puberty can result in anovulation and subfertility [Abbott et al., 2002; Barnett &
Abbott, 2003]. Similarly, female rhesus macaques exposed to high levels of
androgens during development exhibit genital and behavioral masculinization of
various degrees [Goy et al. 1988], and ovarian and adrenal hyperandrogen
production as adults [Eisner et al., 2002].
In female primates, T is produced by both the adrenal glands and the ovaries,
where it is either metabolized into DHT or aromatized into estradiol (E2).
Increases in androgen production commonly occur during the follicular phase of
the ovarian cycle, when surges in luteinizing hormone (LH) trigger the
production of T by the ovaries. Although T and E2 levels often rise during the
follicular phase of ovarian cycling [Dixson, 1998], lower androgen levels have been
found in female ring-tailed lemurs with more advanced follicular development,
and their twofold androgen elevations during the breeding season appear to be
due to increased levels of aggression rather than ovarian steroidogenesis [von
Engelhardt et al., 2000]. These findings suggest that changes in androgen levels
or androgen/E2 ratios may be related to the seasonal regulation of reproduction
and behavior in other female primates.
We examined the relationship between fecal androgen and E2 levels during
the onset of the mating and conception seasons in wild female northern muriquis
(Brachyteles hypoxanthus). Like spotted hyenas and lemurs, female northern
muriquis exhibit masculinized morphological and behavioral traits, including a
hypertrophied clitoris and large body size [Hill, 1962], and behavioral
codominance with males [Strier, 1992]. Northern muriqui females also exhibit
high interindividual variation in the timing of postpartum resumption of cycling,
which can differ by months among females whose prior parturitions occurred only
days apart. In addition, parous females that conceive during the same breeding
season differ in the number of cycles (two to seven) to conception [Strier &
Ziegler, 1997; Strier et al., 2003]. The onset of the muriquis’ annual mating
season is triggered by the onset of ovarian cycling [Strier & Ziegler, 1994; Strier
et al., 2003], but male androgen levels do not fluctuate in response to
environmental variables or reproductive opportunities [Strier et al., 1999,

---

Muriqui Androgen and Estradiol Levels / 71
2003]. In this study we examine the relationship between female androgen and E2
levels in their mediation of ovarian cycling and conception to better understand
the hormonal regulation of fertility in this critically-endangered species.
Like masculinized females in other species, we expected that the levels of
androgens excreted by adult female muriquis would be lower than those of adult
males. We also tested the prediction that female androgen and E2 levels covary
with reproductive condition and with individual differences in the timing of
cycling and conception. In both male and female muriquis, rates of aggression are
low throughout the year [Strier, 1992]. Therefore, in contrast to female ring-
tailed lemurs [von Engelhardt et al., 2000], any fluctuations observed in the
androgen levels of female muriquis are expected to correspond with reproductive
condition alone.
MATERIALS AND METHODS
Study Site and Subjects
Behavioral and hormonal data were collected between 1 August and 22
December 1998 from members of one northern muriqui group that has been the
subject of long-term research at the Estac¸a
˜o Biolo
´gica de Caratinga (EBC)/RPPN
Feliciano Miguel Abdala, Minas Gerais, Brazil. Details on the study site and group
during this study period were previously described [Strier et al., 2003].
Five of the 19 adult females in the study group were selected for hormonal
sampling based on their prior reproductive histories (Table I). All five of these
females had previously given birth to surviving infants in 1996, and could
therefore be expected to resume cycling and conceive during the 1998–1999
mating and conception seasons [Strier, 1996; Strier et al., 2003]. Three of the
target females (DD, NY, and RO) were already carrying infants when systematic
observations on the study group were initiated in June 1982, and were therefore
older than the other two females (BR and PL), which gave birth to their first
infants in 1989 and 1992, respectively.
Data Collection
All copulations involving members of the study group were recorded
whenever they were observed. The five target females resumed copulating
between 25 August and 14 November, coincidentally with the resumption of
their ovarian cycles, as determined from analyses of fecal progesterone and E2
Table I. Female Study Subjects
Previous
Resumed
N nonconceptive
N fecal
Female
parturition
cycling
cycles this study
Conceptiona
samples (N weeks)
PL
17–18 June 96 20 Sep 98
5
28–31 Dec 98
61 (20)
BR
4–5 July 96
12 Sept 98
3
23–26 Oct 98
64 (19)
NY
24 July 96
25 Aug 98
6
24–27 Dec 98
76 (21)
RO
1–7 Oct 96
14 Nov 98
2
13–16 Dec 98
71 (21)
DD
30–31 Oct 96
27 Oct 98
3
None this year
63 (21)
aConception dates for BR and RO were determined by sustained elevations in E2 and progesterone [Strier and
Ziegler, 1997; Strier et al., 2003], whereas those for PL and NY were estimated by subtracting the mean7SD
gestation (215–218 days; [Strier and Ziegler, 1997]) from the dates on which their new infants were first sighted in
1999.

---

72 / Strier and Ziegler
levels [Strier & Ziegler, 1994, 1997; Strier et al., 2003; Ziegler et al., 1997]. The
females experienced from two to six cycles during the present study period, and
all but one female conceived during or within 2 weeks after the study period
(Table I).
A total of 335 fresh fecal samples were collected from the five target females
over the 144-day study period. Fecal samples were collected at 1–2-day intervals
except when individuals could not be located, in which case the collection regime
resumed at the first opportunity. An average of 6776.28 samples (median=64)
were collected from each of the target females (Table I).
Trained observers followed the females until they defecated in a discrete
location. Fresh feces were scooped into clean 50-ml polypropylene tubes labeled
as to date, time, and individual, and stored on ice in insulated pouches
until late afternoon when they were transferred to the freezer (–201C) at the
field station.
In previous studies we found that the time of sample collection did not affect
fecal E2, progesterone, or cortisol levels in female muriquis [Strier & Ziegler,
1994; Strier et al., 2003] or fecal T or cortisol levels in males [Strier et al., 1999].
To evaluate whether female fecal androgen levels were affected by the time of
day, we compared 27 pairs of samples collected from the same individuals on
sequential days before and after 1100 hr (n=27, three to eight pairs of days per
female). There were no significant differences between the log-transformed
androgen levels of morning (before 1100 hr) and afternoon samples (paired t-test,
t(26)=0.390, P40.65).
Hormone Assays
The collected feces were mixed in the field, divided into 0.1 g samples, and
extracted into an ethanol-water (50:50) solution, as described by Strier and
Ziegler [1997]. The extracted steroids were transported to the Wisconsin National
Primate Research Center for analysis. The samples were analyzed for E2 by
radioimmunoassay (RIA), as previously reported [Strier & Ziegler, 1997; Strier
et al., 2003]. The intra- and interassay coefficients of variation (CVs) for the
muriqui fecal pool were 3.58 and 15.68, respectively. For the androgen assay of
the female samples we used the same technique described in Strier et al. [1999]
for male samples. This technique uses solvolysis to remove the conjugates, as
described in Ziegler et al. [1996], and dried samples are resuspended in 500 ml of
ethanol. The androgen antibody cross-reacts mainly with T (100%) and DHT
(57.37%), and only minimally with androstenedione (0.27%) or other androgens
(o0.04%). As described in Strier et al. [1999], high-performance liquid
chromatography (HPLC) separation with RIA of two male samples yielded
T/DHT ratios of 2.60 and 4.56, respectively. By contrast, celite separation and RIA
of five samples from one female in the present study yielded T/DHT ratios of 0.44–
0.89, indicating that DHT accounts for a higher proportion of the ratio in females
than in males. We used the same volume (50 ml) for both females and males
[Strier et al., 1999]. The intra- and interassay CVs for the female muriqui
pool were respectively 9.89 and 13.26 for the low pool, and 3.17 and 6.72 for
the high pool.
Data Analyses
We analyzed each fecal sample for androgen and E2 levels, and calculated the
androgen/E2 ratios for each sample. The mean (7SE) of each female’s weekly

---

Muriqui Androgen and Estradiol Levels / 73
steroid levels were used in comparisons across females (n=19–21 weeks per
female; Table I), and averaged across females for comparisons with mean fecal
T levels measured from six adult males during the same weeks of this study
[Strier et al., 2003].
As described in Strier et al. [2003], we divided the weekly mean steroid levels
for each female into their respective precycling (or anovulatory), cycling, and
pregnant weeks. We also used each female’s reproductive condition to evaluate
whether her steroid levels or androgen/E2 ratios changed across her respective
precycling, cycling, or pregnant conditions.
The females resumed cycling 4–16 weeks into the study period. Two females
conceived during the study period, two conceived within 2 weeks after the last
sample was taken, and one failed to conceive during the 1998–1999 mating season
(Table I). We examined individual androgen, E2, and androgen/E2 profiles across
the 3–15 precycling weeks, and from all samples obtained during the two to six
successive nonconceptive periovulatory periods of each female during this study
(Table I). We also compared female androgen levels during their periovulatory
periods and the intervals between periovulatory periods. Periovulatory periods
were defined as the estimated day of ovulation73 days based on the first sustained
elevation in progesterone [Strier & Ziegler, 1994, 1997; Ziegler et al., 1997].
We present descriptive statistics on the measured levels of each hormone;
however, we conducted all statistical analyses on log-transformed values because
the measured steriods were not normally distributed in all cases. We used paired
t-tests to compare the weekly mean androgen levels of females and males. We also
compared each female’s mean androgen levels during and in between her
periovulatory periods. General linear model (GLM) repeated-measures analyses
of variance (ANOVAs) with Tukey post hoc tests were used to evaluate differences
in hormone levels across female reproductive conditions, and individual variation
during the periovulatory periods of conceptive vs. nonconceptive females. We
considered Po0.05 to be statistically significant, but we looked for patterns in
the data whenever P40.05 because the small number of females in our sample
gave us a low power to detect statistical differences. It is important to note that
we deliberately opted to sample fewer females intensively, instead of more
females sporadically, to increase our ability to detect patterns in continuously
fluctuating hormone levels.
RESULTS
Sex Differences in Androgen Levels
Consistent with our expectation, the androgen levels were lower in females
than in males across the 21-week study period (t(20)=–8.43, Po0.0001; Fig 1).
The mean (7SE) androgen levels in females prior to cycling (43.0471.96 ng/g;
median=41.01 ng/g, n=5), during cycling (38.8271.76 ng/g; median=39.55 ng/g,
n=5), and during the early weeks of pregnancy (41.5875.18 ng/g; median=41.58
ng/g, n=2) were lower than the corresponding values in males, and significantly
lower prior to the resumption of female cycling and copulating (or first confirmed
ejaculation) and after conception (or the first conception-related copulation for
males; Fig. 2).
Variation in Androgens Across Reproductive Conditions
The mean female androgen levels did not differ across reproduction
conditions (F=1.07, df=2,9, P40.38; Fig. 2), and there were no consistent

---

74 / Strier and Ziegler
90
85
80
75
70
Males (n=6)
65
60
SE) T+DHT ng/g 55
+
50
45
40
Weekly Mean (
35
Females (n=5)
30
25
∗
∗
∗
∗
∗
∗
∗
∗
∗
∗
20
8/1
8/8
9/5
8/15
8/22
8/29
9/12
9/19
9/26
10/3
11/7
12/5
10/10
10/17
10/24
10/31
11/14
11/21
11/28
12/12
12/19
Week, 1998
Fig. 1. Androgen levels in female and male northern muriquis. Male data are from Strier et al.
[2003]. Asterisks indicate weeks in which independent t-tests of the individual weekly mean
androgen levels showed significant sex differences (Po0.05).
patterns in the androgen levels of individual females across their respective
precycling, cycling, or early pregnancy conditions (Fig. 3a). By contrast, female E2
levels varied with their reproductive conditions (F=29.06, df=2, 9, Po0.0001),
with higher E2 levels occurring after they resumed cycling than before (Po0.01;
Fig. 3b). Average androgen/E2 ratios also varied significantly across reproductive
conditions (F=12.37, df=2,9, P=0.003). Tukey post hoc tests indicated that
female androgen/E2 ratios were significantly higher prior to the resumption of
cycling (Po0.05) than during cycling. The mean androgen/E2 ratios followed
fairly consistent patterns across individual females, with twofold higher mean
precycling androgen/E2 ratios in the two females (RO and DD) that resumed
cycling later during this study period compared to females that resumed cycling
earlier (Fig. 3c).
Precycling Steroid Profiles
The females resumed cycling at different times. The last female in our
sample resumed cycling 12 weeks after the first female resumed cycling, which
occurred in week 4 of this study. However, there were no obvious patterns in
female androgen levels in the weeks prior to cycling (Fig. 4a). By contrast, the two
females whose cycles resumed later (weeks 12–16) exhibited twofold lower
E2 levels than the other three females, whose cycles resumed 4–8 weeks earlier

---

Muriqui Androgen and Estradiol Levels / 75
Pre-cycling or mating with ejaculate
Cycling or mating only
65
Pregnant or post-first conception
p=0.02
60
p=0.02
55
50
45
40
35
30
T+DHT ng/g
25
20
15
10
5
0
Females
Males
Fig. 2. Androgen levels in female and male northern muriquis across reproductive conditions. There
were no significant differences across reproductive conditions in either females (see text) or males
[Strier et al., 2003]. Androgen levels were significantly lower in females (n=5) than in males (n=6)
prior to the resumption of cycling vs. the first confirmed ejaculate of males (t(9)=À2.82, P=0.02),
and after females conceived (n=2) vs. the first conception-related copulations of males (t(6)=À3.18,
P=0.02).
(Fig. 4b). None of the females resumed cycling until her mean weekly E2 levels
reached >20 ng/g, and all females showed at least a 3-week delay between
elevations in their E2 levels and the onset of their cycles (Fig. 4b).
Androgen/E2 ratios were inversely related to E2 levels (Fig. 4c). Females that
resumed cycling later had higher androgen/E2 ratios corresponding to their lower
E2 levels than females that resumed cycling earlier.
Androgen Levels During Cycling
Androgen levels (mean 7SE) did not differ during the periovulatory
(36.6672.19 ng/g, median=37.84 ng/g) or interperiovulatory (36.6672.19 ng/g,
median=37.84 ng/g) periods for any of the five females (t(4)=–0.71, P40.51). The
mean androgen levels were also similar across the two to six periovulatory periods
of the females in this study (F=1.05, df=4, 14, P40.41).
The steroid profiles of females after they resumed cycling appeared to
differ between the four conceptive females and the one nonconceptive female
(Fig. 5a–c). The nonconceptive female (DD) had significantly lower periovulatory
androgen
levels
than
the
females
that
ultimately
conceived
(F=4.99,
df=1, 17, Po0.04), although her maximimum and minimum cycling androgen
levels fell within the range of those of conceptive females (Fig. 5a). This

---

76 / Strier and Ziegler
Anovulatory
A.
Cycling
55
*
50
Pregnant
45
40
35
30
25
20
T+DHT ng/g
15
10
5
0
BR
NY
PL
RO
DD
B.
550
500
450
*
400
*
350
*
300
250
E2 ng/g 200
150
*
100
*
50
0
BR
NY
PL
RO
DD
C. 5.5
5.0
*
*
4.5
*
4.0
3.5
3.0
*
*
2.5
2.0
1.5
1.0
Androgen/E2 ng/g
0.5
0.0
BR
NY
PL
RO
DD
Fig. 3. Individual patterns in mean (+SE) steroid levels and androgen/E2 ratios across reproductive
conditions. Asterisks indicate significant differences (Pr0.02) except in the case of DD’s E2 levels,
which only approached significance (P=0.059).

---

Muriqui Androgen and Estradiol Levels / 77
BR
RO
A.
70
NY
DD
60
PL
50
40
30
T+DHT ng/g
20
10
0
B. 100
50
45
40
35
30
25
E2 ng/g
20
15
10
5
0
C. 13
12
11
10
9
8
7
6
5
Androgen/E2
4
3
2
1
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Weeks to Resumption of Cycling
Fig. 4. Individual hormonal profiles prior to the resumption of cycling. Week 1 corresponds to
the first week of this study period, and weeks continue for each female until the week prior
to the resumption of cycling. Shown are the mean weekly (A) androgen levels, (B) E2 levels,
and (C) androgen/E2 ratios. Closed symbols indicate females that resumed cycling earlier (weeks
4–8 of the study); open symbols indicate females that resumed cycling later (week 12 and 16,
respectively). Shaded area represents the minimum E2 threshold for cycling to resume, as detected
in this study.

---

78 / Strier and Ziegler
BR NY PL RO DD
A.
80
70
60
50
40
T+DHT ng/g
30
20
10
B.
650
600
550
500
450
400
350
300
E2 ng/g
250
200
150
100
50
0
C.
7
6
5
4
3
Androgen/E2
2
1
0
Nonconceptive Cycles
Fig. 5. Individual hormonal profiles during cycling. All samples are plotted for (A) androgen levels,
(B) E2 levels, and (C) androgen/E2 ratios for each female from the onset of cycling to either
conception or the end of the study period, with variable numbers of periovulatory periods (Table I).
NY and PL conceived after the end of the study period; DD (open circles) did not conceive during the
1998–1999 annual mating season. Shaded area represents the minimum peak E2 threshold for
conception, as detected in this study.

---