The costs of dominance: testosterone, cortisol and intestinal parasites in wild male chimpanzees

Ngogo is in Kibale National Park in western Uganda. The park is located between 0°41'N, 30°19'E and 0°13'N, 30°32'E, with a total area of approximately 750 km². The Ngogo study area is about 25 km² and contains old growth, regenerating, and swamp forest, Acanthus scrub, and other vegetation types [60]. The field site is devoid of domestic herbivores and pets. Human observation of the chimpanzees is restricted to researchers and Ugandan field assistants, and latrines and garbage pits are used for disposal of human waste and refuse at the research camp. The chimpanzees do not enter camp, nor do they enter fields outside of the park boundaries, limiting potential contact of chimpanzees with human or domestic animal feces.

The Ngogo chimpanzee community (Figure 1) was originally studied by Ghiglieri in the late 1970's and early 1980's [61]. Research and habituation efforts resumed at Ngogo in 1991, and have been continuous since 1995. All adult and adolescent male chimpanzees are well habituated and are observable within 5-10 m on the ground. At the time of this study, the Ngogo community had 24 adult males, 14 adolescent males and a total of approximately 150 members.

Exact ages of Ngogo community members are unknown. Adult animals were assigned to the following age categories, based on physical characteristics (noticeable teeth wear, thinning of hair, loss of muscle mass) and on the history of observations: 1 = old, 2 = prime old, 3 = prime, 4 = young prime, 5 = young. Adolescent males were classified on a different scale: 1 = closest to young adulthood (oldest adolescents), 6 = closed to juvenile stage (youngest adolescents). Age is controlled for in all statistically analyses because hormones and dominance rank usually covary with age. However, age was unassociated with intestinal parasite richness in these animals [62].

67 fecal samples were collected opportunistically from 22 adult male chimpanzees at Ngogo, between July and September 2002. A mean of 3.32 samples was collected per individual (range = 1 to 5). Samples from the same individual were collected on non-consecutive days (spacing between the consecutive samples ranged between 2 and 21 days). Samples were collected immediately following defecation, thus insuring positive matching of the individual with the sample. Portions of samples that might have been contaminated by soil or pooled water were not collected, nor was diarrhea. Blood and mucous were not observed in any fecal mass collected, nor did color or consistency differ significantly between masses.

Most samples were collected before 11 AM. Diurnal effects on parasite output in chimpanzees are unknown. Diurnal variation in steroid hormone secretion can be a significant concern, particularly in smaller-bodied primates such as common marmosets (Callithrix jacchus) [63] and tufted Capuchins (Cebus apella nigritus) [20]. However, diurnal variation in fecal hormone levels is probably of less concern in larger-bodied animals, such as chimpanzees (Pan troglodytes), due to longer gut retention time [64, 65]. Fecal steroids represent long-term baseline levels with little susceptibility to minor rapid fluctuations in the hypothalamic-pituitary-gonadal axis.

A portion of each sample was preserved using Para-Pak plastic transport vials (Meridian Diagnostics, Cincinnati, OH, USA) pre-aliquoted with 10% neutral buffered formalin. A separate portion of each sample was dehydrated on an aluminum dish for approximately 2 hours at 100°C in a portable Coleman oven placed atop a kerosene stove. Following desiccation, each sample was individually packaged with silica gel and shipped back to the USA using a CDC import permit.

For each extraction of testosterone and cortisol, a 0.3 gm sample of feces was homogenized in 4 ml methanol:acetone (8:2, v/v) and filtered with a 0.2 μm nylon centrifuge filter (Centrex MF; Scheicher & Schuell, Keene, NH, USA). The filtrate was extracted on Sep-Pak VAC C18 columns (500 mg) (Water Corp., Milford, MA, USA). An equal volume of water was added to dilute the sample, which was then layered onto a column primed according to manufacturer's instructions. The column was washed with 5 ml water, and the steroid fraction eluted with 3 ml methanol. Extraction recovery, measured by the addition of I¹²⁵ labeled steroid to fecal samples prior to extraction, averaged 65% for testosterone and 72% for cortisol.

The testosterone assay used reagents from the Equate Testosterone RIA kit (Binax, South Portland, ME, USA). An aliquot of each extract was reconstituted in working buffer (0.1% gelatin phosphate buffered saline) at a 1:5 dilution. ¹²⁵I testosterone tracer (50 μl) and 100 μl antiserum (diluted 1:2) were added to 100 μl aliquots of standards (diluted 1:10 to give concentrations of 1-100 ng/dL), samples, and controls (diluted 1:10). After vortexing and overnight incubation at room temperature, 500 μl second antibody (PEG goat anti-rabbit antibody solution diluted 1:2) was added. After 20 min incubation at room temperature, incubates were centrifuged at 1500 rpm × gm for 60 min at 4°C. The supernatant was decanted and the radioactivity in the precipitant was determined by 5 min counts in a gamma counter. Sensitivity was 6 ng/dL. Cross-reactivity was 1.7% for dihydrotestosterone and less than 0.1% for all other steroids. Accuracy was tested by the addition of steroid standards to a chimpanzee extract. The mean percentage of observed concentration to expected values in the Equate testosterone assay was 91.4 ± 5.0% (n = 6).

Internal controls were run in every assay and consisted of human serum controls (male and female) provided with the Equate RIA kit along with clinical serum standards (BioRad, Hercules, CA, USA). Intra-assay variation was assessed using the mean coefficient of variation of duplicates of male controls (n = 5) and chimpanzee fecal extracts (n = 19). The mean intra-assay coefficients of variation for duplicates of male serum control (55.0 ng/dL) was 2.7%; that for fecal extract duplicates was 4.4%. Inter-assay variation for serum controls was assessed using the coefficient of variation of male and female and BioRad controls from five separate assays. The inter-assay coefficient of variation of samples was assessed using the mean of coefficients of variation for three chimpanzee samples analyzed in two separate assays. Inter-assay coefficients of variation were 4.2% for the Equate female serum control (4.8 ng/dL), 4.6% for the Equate male serum control (55.0 ng/dL), 4.2% and 7.2% for BioRad controls 1 (4.7 ng/dL) and 2 (58.2 ng/dL), and 10.1% for the three chimpanzee samples.

The cortisol assay used reagents from the Diagnostics Products Corporation Double Antibody ¹²⁵I cortisol kit (DPC KCOD, Los Angeles, CA, USA) for serum determinations. Working buffer was distilled water. A tracer-antiserum solution was prepared by mixing equal parts of ¹²⁵I cortisol and cortisol antiserum, and 50 μl added to 25 μl aliquots of the standards (diluted 1:10 to give concentrations of 1-50 ng/ml), samples (concentrated 10:1), and controls (diluted 1:10). Each was vortexed and incubated at 37°C. After 45 min, 250 μl of cold precipitating solution was added, and the incubates were vortexed, incubated an additional 5 min at room temperature, and centrifuged at 3000 rpm × gram for 15 min at room temperature. Following decanting of the supernatant, the radioactivity of the precipitate was determined by 5 min counts in a gamma counter. Sensitivity of the assay was 2.2 ng/ml. Cross-reactivities were 3.9% for cortisone, 3.6% for 6β-hydroxycortisol, 1.1% for corticosterone, and less than 1% for all other steroids. Accuracy, tested by the addition of cortisol standards to a chimpanzee extract, averaged 96.8 ± 2.6% (n = 5).

Internal controls were run in every cortisol assay and consisted of clinical serum standards (Bio-Rad 1, 2, 3). Intra-assay variation was assessed using the mean coefficient of variation of duplicates of BioRad controls (n = 4) and chimpanzee fecal extracts (n = 12). Mean intra-assay coefficient of variation for duplicates of BioRad controls was 2.2%. Mean intra-assay coefficient of variation for sample duplicates was 11.6%. Inter-assay variation for serum controls was assessed using the coefficient of variation of BioRad controls from five separate assays. Mean inter-assay coefficient of variation for BioRad serum control #1 (low) was 11.5%; mean inter-assay coefficient of variation for BioRad serum control #2 (medium) was 8.7%; mean inter-assay coefficient of variation for BioRad serum control #3 (high) was 5.3%.

The formalin-fixed samples were examined using the formalin-ethyl acetate sedimentation technique [66]. Stool samples were emulsified and filtered through two layers of wet gauze into a plastic cup. The stool was washed with saline solution, placed into a 15 ml conical-bottom centrifuge tube, and centrifuged at 500 rpm × gram for 3 minutes. The supernatant was discarded, and the sediment was re-suspended in 10 ml of 10% formalin. 3 ml of ethyl acetate was added to separate the fat in the sample, and the suspension was shaken vigorously for 30 seconds. The specimen was re-centrifuged, the fat/debris plug was removed with an applicator stick, and the supernatant discarded. The remaining pellet was re-suspended using a drop of Lugol's iodine solution, and the entire pellet was examined at 10× and 40×.

Intestinal parasite infection status is often measured as parasite richness (the number of species recovered from hosts' fecal samples) or parasite intensity (the number of eggs/cysts/larva per unit mass of feces). Parasite excretion can vary dramatically within and between individuals, and parasite egg/cyst/larvae abundance in any given fecal sample may not directly correlate with the number of parasites in the chimpanzee host at any given time. Parasite excretion may not reflect the immune status of a host, although disagreement exists on this point [67]. Parasite richness probably at least reflects the ability of the host to control infections with multiple parasites at any given time. Therefore we used data only on parasite richness because we consider this a more robust measure than intensity.

Data were entered into an Access database that was imported into SAS/STAT software (SAS Institute Inc., Cary, NC, USA). Mixed modeling (PROC MIXED) was used to examine relationships between intestinal parasite richness, dominance rank and hormone levels. Mixed modeling allowed the use of all data points, including individuals with missing observations, and avoided the need for averaging testosterone levels and parasite measures for individuals and sampling intervals. It also allowed examination of within-subject effects of continuous variables and control of within-subject covariates (age). A time-series covariance structure that did not assume equal spacing of sample intervals was used in addition to a compound symmetry covariance structure that assumed correlations remained constant. This is a reasonable assumption given the short 3-month sampling period and stability of the dominance hierarchy (see below). A time-series covariance structure accounts for unequal time periods between sequential samples as well as differences in the number of samples collected for each animal. Sampling frequency did not vary consistently with dominance rank (i.e., higher ranking animals were not sampled more frequently). However, parasite species richness significantly increased for every sequential sample taken (up to four samples) from those adult, adolescent and juvenile males sampled within the Ngogo community [62]. The average time between consecutive samples collected was 7.74 days.

Partial Spearman correlations (controlling for age) were used in addition to the mixed models. A negative correlation coefficient indicates a positive association because the highest ranking animal was ranked number one whereas the lowest ranking animal was ranked number twenty two. Level of significance was always set at 0.05.

Susceptibility to infection differs among individuals, and neuroendocrine mechanisms may account for these differences. The mammalian immune responses to gastrointestinal infection are typified by a combination of phagocytosis [69, 70], activation of the complement cascade and antibody responses to block cellular invasion [71], and Th-1 and Th-2 cytokine release, which facilitates gut inflammation [71‐74]. Nematode infections are typically controlled via the Th-2 response with eosinophilia, goblet cell hyperplasia, mucin production, and intestinal mastocytosis [75‐79]. Both testosterone and cortisol may affect these responses.

Testosterone's immunomodulatory actions appear to be primarily suppressive, increasing suppressor T cell populations, reducing T-helper cell function, inhibiting cytokine and antibody production, and impairing natural killer cell and macrophage activity [80‐90]. By favoring the development of a CD4+ type-1 phenotype of peripheral lymphocytes and cytokines [91‐93], elevated testosterone levels may increase susceptibility to infections that are normally cleared via the Th-2 response, like gastrointestinal infections. Not surprisingly, testosterone administration to female soft-furred rats results in reduced expulsion of the nematode Nippostrongylus brasilnesis [94]. Testosterone treatment in mice is also associated with increased tapeworm egg production [33] and increased susceptibility of females to Strongyloides ratti infection [95]. Saino and Moller [96] also identified a negative association between testosterone and parasite load in barn swallows (Hirundo rustica). It may be that testosterone-mediated suppression of Th-2 anti-inflammatory cytokines diminishes allergic responses that are needed to clear intestinal helminth infections. This may explain why those chimpanzees with higher testosterone level exhibit increased helminth burden.

Cortisol's immunomodulatory actions are also primarily inhibitory [[97], and see above]. In addition to chimpanzees at Ngogo, inverse associations between cortisol and immune measures have been identified in wild baboons and red colobus monkeys. In female baboons, cortisol was inversely associated with total lymphocyte levels [98]. In male baboons, cortisol was inversely associated with insulin-like growth factor I [99]. Chapman and others [100] have identified a direct association between fecal cortisol levels and nematode infection in wild red colobus monkeys (Procolobus rufomitratus tephrosceles) of Kibale National Park, Uganda. In these and other cases, elevated or otherwise dysregulated glucocorticoid responses to behavioral or physical stressors could result in various physical impairments, including altered lipid profiles and other cardiovascular system changes [101]. In the present study, cortisol was not significantly associated with dominance rank, but was associated with intestinal parasite richness.

Relationships between dominance rank and immune measures have been identified in several species. In some, immunocompetence is lower in high status animals. For example, intestinal infections with the trematode Genitocotyle mediterranea were greater in dominant male European wrasse (Symphodus ocellatus) than in smaller, subordinate males [102]. High rank was associated with lower spleen mass and antibody levels in response to human IgG in Brandt's voles (Lasiopodomys brandtii) [103]. In contrast, high status individuals of other species frequently exhibit elevated immune responses relative to their subordinate counterparts. High ranking male greenfinches (Carduelis chloris) clear infections from Sindbis virus more quickly than lower ranking animals [104]. High ranking female dairy goats had fewer gastrointestinal parasite eggs in their feces than medium and low ranking individuals [105]. Dominant pigs exhibited higher lymphocyte proliferation to Aujeszky disease virus and the mitogens concanavalin A and phytohemagglutinin than subordinate animals [106, 107].

In contrast, dominance status was unrelated to both testosterone and fecal parasite egg counts in male Alpine ibex (Capra ibex) [68]. Dominance status was also unrelated to anti-Seoul virus IgG responses in inoculated male Norway rats (Rattus norvegicus) [108].

Studies of nonhuman primates also provide mixed results. In three small, mixed sex, captive groups of chimpanzees, dominance rank was significantly negatively correlated with immunoglobulin (IgG and IgM) levels [109]. Dominant male longtailed macques (Macaca fascicularis) exhibited lower primary antibody responses to tetanus toxoid [110]. High ranking male yellow baboons at Amboseli, Kenya, had greater intestinal helminth infections than low ranking males [111]. In contrast, Muller-Graf and others [112] found no association between helminth infection and dominance rank in olive baboons (Papio cynocephalus anubis). Social subordinance is associated with increased louse prevalence, lower insulin-like growth factor I, and fewer circulating lymphocytes in olive baboons (Papio anubis) [99, 101, 113]. Low ranking female rhesus macaques (Macaca mulatta) had lower CD4+ and CD8+ lymphocyte counts than higher ranking females [114], and low ranking male longtailed macaques (Macaca fasicularis) were at greater risk of adenovirus infection than high ranking animals [115]. The present study suggests that high ranking adult male chimpanzees have increased helminth burden compared to low ranking males.

However, just as the relationships between dominance rank and circulating hormone levels within a species may depend on many factors, including access to social support, stability in the dominance hierarchy, and individual personality [116‐121], so too should the relationships between dominance and immune status depend on several factors. Because high ranking males typically have more mating opportunities, they may be at increased risk of acquiring directly-transmitted infections [122]. Likelihood of exposure may vary with home range size, daily travel distance, and variation in social networks. Increased grooming opportunities can decrease the risk of arthropod-born diseases. Receiving grooming, in particular, may protect individuals against deleterious effects of chronic stress responses [121] and thereby promote immune status. High ranking males at Ngogo are attractive grooming partners [123] and the amount of grooming received was positively associated with rank during our study period. Increased social support can lead to decreased cortisol and catecholamine levels independently of rank [e.g., baboons: [118‐120]], although that high ranking males at Ngogo received more grooming and more coalitionary support than low ranking males might help to explain why cortisol levels were not significantly correlated with rank. Personality factors, particularly sociability, can mediate disease outcomes [124, 125]. Also, nutritional status is a major determinant of disease susceptibility and immunocompetence [126]. In so far as high rank confers greater access to nutritional resources, high ranking individuals should be able to bolster immune responses and withstand greater infection loads than subordinates.