Variation in Physiological Health of Diademed Sifakas Across Intact and Fragmented Forest at Tsinjoarivo, Eastern Madagascar

American Journal of Primatology 72:1013–1025 (2010)

RESEARCH ARTICLE Variation in Physiological Health of Diademed Sifakas Across Intact and Fragmented Forest at Tsinjoarivo, Eastern Madagascar

1Ã 2 3 1 MITCHELL T. IRWIN , RANDALL E. JUNGE , JEAN-LUC RAHARISON , AND KAREN E. SAMONDS 1Redpath Museum, McGill University, Montreal, Quebec, Canada 2St. Louis Zoological Park, St. Louis, Missouri 3Department of Animal Biology, University of Antananarivo, Antananarivo, Madagascar As undisturbed habitat becomes increasingly rare, managers charged with ensuring the survival of endangered primate species must increasingly utilize disturbed and degraded habitats in species survival plans. Yet we have an imperfect understanding of the true long-term viability of primate populations in disturbed habitat, and census data can be misleading because density is not necessarily correlated with habitat quality and population viability in predictable ways. Here we present clinical laboratory data on hematology, serum biochemistry, fat-soluble vitamins, minerals, iron analytes, viral serology, and parasitology of diademed sifaka (Propithecus diadema), derived from the capture of 26 individuals spanning eight groups and two habitats (undisturbed vs. disturbed and fragmented) at Tsinjoarivo, Madagascar. Blood from fragment individuals had significantly lower values for several factors: white blood cell counts, bilirubin, total protein, albumin, calcium, sodium, chloride, manganese, zinc, iron and total iron-binding capacity. Several biochemical variables were higher in immature individuals, probably due to active growth. The large number of interhabitat differences suggests that habitat disturbance has an impact on physiological health within this population, perhaps reflecting dietary stress and/or immunosuppression. These results, combined with previous data showing altered diet, slower juvenile growth, and reduced activity in disturbed forest fragments, suggest that fragment sifakas may be less healthy than continuous forest groups. Finally, Tsinjoarivo sifakas have extremely low blood urea nitrogen (perhaps reflecting protein limitation) and selenium levels relative to other lemurs. Despite their survival and reproduction in the short term in fragments, these sifakas may represent a riskier conservation investment than conspecifics in undisturbed forest, and may be more susceptible to environmental stressors. However, more data on the fitness consequences of these biochemical differences are needed for a better interpretation of their impacts on long-term viability prospects. Am. J. Primatol. 72:1013–1025, 2010. r 2010 Wiley-Liss, Inc.

Key words: Madagascar; Propithecus diadema; physiology; conservation; habitat fragmentation

INTRODUCTION advantage of disturbed areas in regional conservation plans. It is therefore critical to know the conservation Habitat fragmentation and degradation threaten value of altered habitats: ﬁrst, how quickly they can biodiversity and ecosystem integrity worldwide regenerate into communities and ecosystems that can [Chapman & Peres, 2001; Laurance et al., 2000]; function to meet management needs, and second, indeed, fragmented and degraded habitats constitute under what conditions they can sustain viable a large proportion of many species’ remaining populations of animal species of interest. habitat. Using satellite technology it is relatively easy to monitor the extent of habitat loss and fragmentation [Harper et al., 2007; Jorge & Garcia, Contract grant sponsors: National Geographic Society Committee 1997], and to a lesser extent, habitat disturbance for Research and Exploration; St. Louis Zoo Field Research for [Ingram et al., 2004]. However, relationships between Conservation Program; Margot Marsh Biodiversity Foundation; habitat changes and species extinctions are complex NSERC. and poorly understood, largely due to the confound- ÃCorrespondence to: Mitchell T. Irwin, Redpath Museum, McGill University, 859 Sherbrooke St. W, Montreal, Quebec, Canada ing effects of fragmentation, degradation, and direct H3A 2K6. E-mail: [email protected] anthropogenic impacts such as hunting [Fahrig, Received 27 November 2009; revised 27 April 2010; revision 2003; Redford, 1992]. Given the increasing forest accepted 27 April 2010 loss, fragmentation, and degradation throughout the DOI 10.1002/ajp.20847 world, truly pristine habitat is becoming rare for Published online 1 June 2010 in Wiley Online Library (wiley many species, and planners are increasingly taking onlinelibrary.com). r 2010 Wiley-Liss, Inc. 1014 / Irwin et al.

Biologists have documented differences in popu- expected reference values for a species, (2) qualita- lation density across habitat disturbance and frag- tive and quantitative data for population viability mentation gradients in many study systems analysis programs, (3) comparison with the same [Laurance et al., 2002]. Many authors have demon- population at a future date to determine the effects strated marked interspecific variation in responses to of disturbance (i.e., logging, weather extremes, fragmentation and disturbance (with some species habitat loss, ecotourism), and (4) comparison among increasing and others decreasing) among primates populations, including translocated or captive ani- [Chiarello & de Melo, 2001; Ganzhorn et al., 2003; mals, to understand the relative quality of different Lehman et al., 2006a,b; Onderdonk & Chapman, habitats, judge various management strategies, help 2000] and other mammals [Goodman & Rakoton- understand the etiologies of captivity-related dis- dravony, 2000; Nupp & Swihart, 2000]. However, eases, and aid in risk assessment of reintroduction densities may be misleading for primates, which are programs. long-lived and relatively mobile. Primate populations Here we present biomedical health assessment in degraded habitats may (1) demonstrate time lags data for 26 diademed sifaka (Propithecus diadema) between disturbance or fragmentation and ultimate in eight groups, across intact and degraded/fragmen- extinction, and (2) represent sink populations in ted habitat at Tsinjoarivo, Madagascar. We present which groups persist only through immigration from data on hematology, plasma total protein, serum better-quality habitats [North & Ovaskainen, 2007; chemistry, fat-soluble vitamins, trace minerals, Pulliam, 1988]. In either case, degraded habitat may measures of iron metabolism, and serological evi- contain primates in the short term (sometimes at dence of infectious agents. We ask the following: (1) increased density), but might be of low conservation what are typical baseline values for these parameters value for particular species in the long term. for P. diadema at Tsinjoarivo? and (2) do values An ideal way of assessing population viability in differ between habitats (intact vs. fragmented) and disturbed habitats is to assess vital demographic age classes (immature vs. adult)? statistics (natality, mortality, and migration rates), which can be used in population viability analyses. However, for primates (and other animals with slow METHODS life histories), the time required to collect these data All animal capture and medical evaluation is long, which can be problematic when management protocols were approved by McGill University’s decisions must be taken quickly. At the other Animal Care Committee and St. Louis Zoo’s IACUC, extreme, censuses are rapid and useful in document- and adhered to the legal requirements of Madagascar. ing population density differences among habitats This research adhered to the American Society of [Lehman et al., 2006b], but unless repeated over time Primatologists’ principles for the ethical treatment they yield few data that can be used to infer of nonhuman primates. population viability. An intermediate solution is to directly compare animals’ health and behavior in altered and intact habitats, to identify behavioral Study Site and Study Species and physiological responses to habitat changes. Tsinjoarivo forest (191410S, 471480E; Fig. 1) is Primatologists have already shown that some species located southeast of Ambatolampy and atop the have altered behavior in degraded habitats [Estrada escarpment dividing Madagascar’s central plateau et al., 1999; Irwin, 2008a; Menon & Poirier, 1996; from the eastern coastal lowlands. This region Onderdonk & Chapman, 2000], but the fitness contains a unique, under-explored block of central effects of these changes remain relatively unexplored domain, mid-altitude rainforest; its eastern half [but see Chapman et al., 2006]. Primates are remains relatively intact and undisturbed while its behaviorally flexible; documenting behavioral shifts western half has been fragmented and degraded by is an important first step, but determining how those human settlers. Although Tsinjoarivo is being con- shifts affect fitness is critical for conservation. It is sidered for protected status, its current status is therefore vital, when possible, to measure fitness and ‘‘Classified Forest’’ and there is little protection on health more directly in primate populations in the ground. Forest loss and disturbance (e.g., disturbed habitat. selective extraction of hardwood trees) continue in An increasing number of health assessments some areas. Tsinjoarivo contains a unique P. diadema have been published for free-ranging primate spe- population that has lower body mass than other cies, including lemurs [Dutton et al., 2003, 2008; P. diadema populations. Although previous genetic Junge & Garell, 1995; Junge & Louis, 2002, 2005a,b, studies did not assign it to a new subspecies [Mayor 2007; Junge et al., 2008]; however, data are limited et al., 2004], more recent natural history reviews to relatively few species, often examine a small treat it as distinct [Mittermeier et al., 2006]. This number of parameters, and typically sample only sifaka is most likely limited to the small forest block one habitat. This is unfortunate, since these studies between the Mangoro and Onive rivers, a very small have great potential to provide (1) ‘‘normal’’ or part of P. diadema’s overall range (o2,000 km2).

Am. J. Primatol. Sifaka Health in Fragmented Forest / 1015

Fig. 1. Location of continuous (CONT) and fragmented (FRAG) forest study groups. Madagascar map shows approximate extent of rainforest; regional map shows rainforest cover based on a 2001 IKONOS satellite image (GeoEye, Dulles, VA). Home ranges are based on ranging data except CONT3 (estimated home range illustrated because this group was not followed systematically). CONT5 was not sampled for this study.

The Tsinjoarivo sifakas have been the subject of 5 male) and 13 were aged 1–5 years; 6 of the adult behavioral and ecological research since 2002 [Irwin, females had infants (o2 months) and 2 were 2008a,b], centered around three camps. This study pregnant (near term). Groups were pre-habituated examines four ‘‘continuous forest’’ groups in intact, and all captures were performed at close range using undisturbed forest (groups CONT1, CONT2, CONT3 a blowgun loaded with Pneu-dart 9 mm disposable at Vatateza: 19143.250S, 47151.410E; 1,396 m; group nonbarbed darts. Sifakas were immobilized with CONT4 at Ankadivory: 19142.980S, 47149.2930E; tiletamine/zolazepam (Telazols, Fort Dodge Animal 1,345 m) and four ‘‘fragment’’ groups in disturbed, Health, Overland Park, Kansas 66225; 25 mg/kg). fragmented forest (groups FRAG2, FRAG4, FRAG5, Once anesthetized, we moved animals to the nearest FRAG6 at Mahatsinjo: 19140.940S, 47145.460E; camp for processing. Each animal was given a 1,590 m). Forest composition differs between sites, complete physical examination and monitored by largely due to past disturbance; CONT habitats have assessing heart rate, respiratory rate, and body higher species richness, higher canopy, fewer (but temperature. A balanced electrolyte solution equiva- larger) trees, and higher overall basal area per lent to the blood volume collected was administered hectare [Irwin, 2006a]. Sifakas’ diet composition in subcutaneously. Animals were allowed to recover in terms of plant parts is relatively consistent across burlap sacks before re-release at the original capture the study population [53% of feeding time on foliage, site or near group-mates. 24% on fruits, 7% on seeds, and 15% on ﬂowers; Medical evaluations followed the standard pro- Irwin, 2008a], but species composition of the diet tocol used elsewhere in Madagascar by the Prosimian varies greatly, with FRAG groups relying heavily on Biomedical Survey Project (PBSP) Team. This mistletoes (a fallback food). Groups are capable of protocol has been used since 2000, on over 570 wild persisting in fragments over the short term (frag- lemurs from 31 species across 16 sites [Dutton et al., ment groups have survived and reproduced since at 2003, 2008; Junge & Louis, 2002, 2005a,b, 2007; least 2000), but indirect signs of stress are evident: a Junge et al., 2008]. Blood samples were collected not dietary shift with increased reliance on mistletoe exceeding 1% of body weight (1 mL/100 g body [Irwin, 2008a], reduced body mass, especially for weight). Whole blood (0.5 mL) was placed into EDTA juveniles [Irwin, 2006a; Irwin et al., 2007], and anticoagulant and the remaining volume into non- altered activity patterns including reduction of anticoagulant tubes and allowed to clot. Serum tubes energetically costly activities such as play and were centrifuged within 8 hr of collection, and serum ranging [Irwin, 2006a]. pipetted into plastic tubes and frozen in liquid nitrogen for transport to the Saint Louis Zoological Park (St. Louis, MO) and stored at 701C until Capture and Data Collection Methodology analysis. From 8 to 22 July 2008, we captured 26 sifakas Within 2 hr of collection, two blood smear slides (CONT1: 4, CONT2: 4, CONT3: 1, CONT4: 3, were made from each anticoagulant sample; smears FRAG2: 4, FRAG4: 5, FRAG5: 2, FRAG6: 3). Of were ﬁxed and stained. A total white blood cell these animals, 13 were resident adults (8 female and (WBC) count was done within 4 hr of collection

Am. J. Primatol. 1016 / Irwin et al.

(Unopette System, Becton Dickinson & Co., Franklin factors were investigated: habitat (fragmented for- Lakes, NJ), and stained smears examined micro- est, at the Mahatsinjo site, vs. continuous forest, scopically for differential blood cell count and incorporating Vatateza and Ankadivory sites), and hemoparasite examinations. Serum was submitted age class (breeding adults of minimum 5 years old vs. to the following laboratories for analysis: serum all other immatures, aged 1–4 years and residing in biochemical profile (AVL Laboratories, St. Louis, natal groups). We did not explore sex as a fixed factor MO); 25-hydroxycholecalciferol (vitamin D) and due to small sample size; this would have reduced the trace mineral analysis (Animal Disease Diagnostic sample within each age/sex/habitat combination to o5. Laboratory); fat soluble vitamin analysis (University We identified outliers within age/sex combinations of Illinois Nutrition Laboratory); iron metabolism using Dixon’s test [Sokal & Rohlf, 1995] and analysis (Kansas State University); viral serology for repeated some analyses with outliers removed. adenovirus group-specific IgG antibody, herpes virus We did not apply experiment-wide Bonferroni (SA8), influenza A antibody, rotavirus (SA11) group- corrections due to the problems applying these specific antibody, reovirus, hepatitis A antibody, and methods to field ecological studies [Moran, 2003]. West Nile virus IgG and IgM antibody (for n 5 1 Instead, we report P-values for all tests, and indicate CONT animals and n 5 11 FRAG animal except West for each analysis the number of significant relation- Nile virus, n 5 6 FRAG animals; Exoterix, San ships expected by chance. Levene’s test indicated Antonio TX); viral serology for herpes simplex virus heterogeneity of variance for several of the ANOVAs. (HSV), simian retroviruses (SRV1, SRV2, SRV5), Unfortunately, this was not improved by square root simian T-lymphotropic virus (STLV), simian immu- or logarithmic transformations. In each case when a nodeficiency virus (SIV), simian foamy virus (SFV), factor was statistically significant within an ANOVA and measles virus (n 5 5 CONT animals and n 5 9 with a positive Levene’s test, the difference among FRAG animals except HSV, n 5 3 CONT animals and habitats or age classes was confirmed using a simple, n 5 9 FRAG animals; Diagnostic Laboratory, Wa- nonparametric alternative (Mann–Whitney U-test). shington National Primate Research Center); serol- However, we cannot exclude the possibility that ogy for Toxoplasma gondii IgG (n 5 4 CONT animals heterogeneity of variances contributed to the non- and n 5 12 FRAG animals; University of Tennessee significant P-values for some factors; thus, signifi- Comparative Parasitology Service). Pathogens se- cant factors may be under-reported. lected are either human pathogens known to exist in Madagascar, therefore having anthropozoonotic potential (adenovirus, herpesviruses, influenza, rota- RESULTS virus, reovirus, hepatitis, West Nile virus, measles) All lemurs appeared clinically normal (Table I). or primate viruses (SRV, SIV, STLV, SFV) that have One female (CONT1: BR) showed evidence of past not yet been identified in lemurs, but could have eye trauma (left eye with corneal scarring and significant health implications if present. Fecal smaller globe size). All individuals had mites similar samples were collected from freshly voided feces to Liponysella madagascariensis (H. Klompen, when possible and placed into 10% formalin for personal communication). Results of hematology, endoparasite assessment; results are not presented serum biochemistry, serum vitamin, and serum here as they are part of a longitudinal assessment of mineral analyses are shown in Tables II–V. known individuals (Raharison and Irwin, unpub- Hematology results were in some cases driven by lished data). Ectoparasites were removed with a outliers. For total WBC counts (Table II and Fig. 2), cotton swab or forceps and placed in 95% ethanol. the removal of three unusually high outliers (MAHA4 JUV: 8030 WBC/mcl; MAHA6 PG: 7260 WBC/mcl; and CONT4 RAD: 16960 wbc/mcl; Analysis Dixon’s test: Po0.01) resulted in a significant Univariate general linear models (GLM) were ANOVA model including a significant effect of site used to explore interindividual variation. Two fixed (P 5 0.002), a near-significant effect of age class

TABLE I. Weight and Vital Signs of Captured Sifakas Adults 1–4 yr old

CONT FRAG TOTAL CONT FRAG TOTAL

N 6 7 13 6 7 13 Mass (g) 5,1427387 5,0187199 5,0757294 4,4317538 3,8987980 4,1447823 Temperature (1C) 35.470.4 36.871.1 36.171.1 35.970.7 36.571.0 36.270.9 Pulse (min1) 108720 104713 106716 111713 104714 107713 Respirations (min1) 41.0715.8 34.3713.3 37.4714.3 47.0715.4 46.07 11.7 46.5712.9

Am. J. Primatol. Sifaka Health in Fragmented Forest / 1017 0.724 0.653 0.05). 4 0.376 0.248 0.100 0.207 0.766 0.819 0.488 0.187 0.547 0.475 0.200 0.066 0.004 P 5.0 17.6 1,926 0.80 0 (12) – – – 0 (12) – – – 17.6 0 (12) – – – 7 7 7 7 7 7 7 7 (12) (12) (11) (12) (12) 45.4 52.0 0.50 47.5 3,944 riances (Levene’s test, 4.3 2,050 14.0 0(6) 0 0.82 0(6) 0 14.6 0(6) 0 7 7 7 7 7 7 7 7 (6) (7) (6) (6) (6) 42.3 44.3 0.33 55.3 3,418 3.7 1,480 18.5 0(6) 0 0.82 0(6) 0 17.8 0(6) 0 7 7 7 7 7 7 7 7 (6) (4) (6) (6) (6) Fig. 2. WBC counts of adult and immature (IMM) sifakas in 48.5 59.7 0.67 39.7 continuous (CONT) and fragmented (FRAG) habitats. 4,865

(P 5 0.088), and a signiﬁcant interaction (P 5 0.025). 3,979 4.9 0 (13) 0 12.4 0 (13) 0 0.38 0 (13) 0 13.0 In general, counts tended to be low for both age 7 7 7 7 7 7 7 7 (13) (13) (13) (13) (13) classes in FRAG groups, CONT adults were inter- 45.8 45.2 0.15 55.0

4,469 mediate, and CONT immatures were highest. For this ANOVA, variances were heterogeneous (Levene’s test: Po0.05), but the site difference 15.3 1,732 0(7) 0 3.7 0(7) 0 16.4 0.38 0(7) 0 7 7 7 7 7 7 7 7 persisted in a Mann–Whitney U-test either with (7) (7) (7) (7) (7) outliers removed (Z 5 2.596, P 5 0.007, N 5 24) or 51.7 43.7 47.4 0.14 3,512

-Values) included (Z 5 2.021, P 5 0.042, N 5 21). P No positive results were detected for any of the 9.6 5,624 0(6) 0 5.2 0(6) 0 5.8 0(6) 0 0.41 15 viral serology assays, and toxoplasmosis titers 7 7 7 7 7 7 7 7 were negative for all animals tested. (6) (6) (6) (6) (6) 58.8 48.3 42.7 0.17 5,586 DISCUSSION 3,154 15.5 0 (25) 0 0 (25) 0 4.8 0 (25) 0 15.2 0.63 Effects of Habitat and Age 7 7 7 7 7 7 7 7 (24) (25) (25) (25) (25) For hematology variables, packed cell volume 51.4 0.32 45.6 48.5

4,229 (PCV) was signiﬁcantly higher in CONT animals, and WBC counts were elevated in CONT animals,

0.05). especially for immatures (Fig. 2 and Table II). 1,824 14.4 0 (13) 0 0.60 0 (13) 0 3.9 0 (13) 0 14.8 o P 7 7 7 7 7 7 7 7 Although we cannot exclude the possibility that (14) (13) (13) (13) (13) differing PCV reﬂects underlying physiological dif- 53.4 0.23 43.1 46.0 All animals Adults 1–4 yr old 3,465 ferences, it is more likely due to the fact that FRAG animals were captured ﬁrst (when the centrifuge battery was fully charged) and CONT animals were 4,294 0 (12) 0 16.9 0 (12) 0 0.67 0 (12) 0 4.3 15.8 12 14 26 6 7 13 6 7 13 Site Age Interaction 7 7 7 7 7 7 7 7 captured second (when the battery was less fully (10) (12) (12) (12) (12) CONT FRAG TOTAL CONT FRAG TOTAL CONT FRAG TOTAL Effect of

49.3 charged and the centrifuge ran more slowly). Thus, values for CONT animals may be more clinically relevant. For WBC counts, statistically signiﬁcant effects of site and age class emerged only after outliers were removed; a nonparametric test suggests an effect of site with or without outliers. We believe that the

neutrophils (%) removal of outliers is justiﬁed by the complicated WBC/mcl 5,298 Eosinophyls (%) 0 Segmented Basophils (%) 0 Monophyls (%) 0.42 Band neutrophils (%) 0 TABLE II. Hematology Results, With Results of GLM Analyses ( Because of missing values for some variables, sample size for these factors is indicated in parentheses. No factors demonstrated heterogeneity of va Bold indicates a signiﬁcant factor in the GLM ( N Packed cell volume (%) 48.4 Lymphocytes (%) 51.2 time scale on which WBC counts respond to external

Am. J. Primatol. m .Primatol. J. Am. al. et Irwin / 1018

TABLE III. Serum Biochemical Proﬁles, With Results of GLM Analyses (P-values)

All animals Adults 1–4 yr old

CONT FRAG TOTAL CONT FRAG TOTAL CONT FRAG TOTAL Effect of

N 12 14 26 6 7 13 6 7 13 Site Age Interaction

Aspartate transaminase 29.42714.21 25.29714.07 27.19714.01 25.0079.86 17.29713.02 20.85711.89 33.83717.33 33.29710.48 33.54713.42 0.425 0.023 0.488 (AST) (IU/L) Alanine 43.7574.27 40.36713.61 41.92710.36 43.1774.36 33.14714.71 37.77711.97 44.3374.50 47.5777.98 46.0876.56 0.361 0.043a 0.082Ã aminotransferase (ALT) (IU/L) Total bilirubin (mg/dL) 0.71770.248 0.50070.196 0.60070.243 0.61770.117 0.50070.100 0.55470.120 0.81770.313 0.50070.271 0.64670.323 0.020 0.259 0.259 b Ã Alkaline phosphatase 262.3798.2 257.97153.2 259.97128.3 196.0755.9 157.0720.4 175.0743.8 328.5787.0 358.97163.3 344.87129.4 0.913 o0.001 0.383 (IU/L) Gamma glutamyl 13.3374.58 14.6474.73 14.0474.62 15.3372.58 16.1472.54 15.7772.49 11.3375.47 13.1476.07 12.3175.63 0.466 0.060 0.779 transferase (GGT) (IU/L) Total protein (g/dL) 7.4470.45 7.0370.47 7.2270.50 7.5870.51 7.2070.48 7.3870.51 7.3070.37 6.8670.43 7.0670.45 0.030 0.093 0.869 Albumin (g/dL) 5.4170.27 4.9770.29 5.1770.36 5.5270.23 4.9470.31 5.2170.40 5.3070.28 5.0070.29 5.1470.32 0.001 0.482 0.232 Globulin (g/dL) 2.0370.28 2.0670.35 2.0570.32 2.0770.36 2.2670.36 2.1770.36 2.0070.21 1.8670.21 1.9270.22 0.840 0.058 0.166 Blood urea nitrogen (mg/ 3.9271.44 4.8671.51 4.4271.53 4.1771.17 5.0071.73 4.6271.50 3.6771.75 4.7171.38 4.2371.59 0.133 0.521 0.860 dL) Creatinine (mg/dL) 0.55870.162 0.59370.114 0.57770.137 0.61770.098 0.62970.076 0.62370.083 0.50070.200 0.55770.140 0.53170.165 0.522 0.090 0.674Ã Phosphorus (mg/dL) 3.7071.09 4.1971.18 3.9771.15 3.3070.84 3.8470.75 3.5970.81 4.1071.23 4.5471.48 4.3471.34 0.276 0.103 0.911 Calcium (mg/dL) 10.3870.43 9.9770.58 10.1670.55 10.1870.35 9.7170.52 9.9370.50 10.5870.44 10.2370.56 10.3970.52 0.041 0.025 0.766 Glucose (mg/dL) 107.5714.3 109.6723.0 108.7719.2 104.8717.4 113.1726.6 109.3722.4 110.2711.3 106.1720.3 108.0716.2 0.789 0.917 0.444 Sodium (mmol/L) 145.572.1 139.975.1 142.574.9 146.572.1 139.476.4 142.776.0 144.571.8 140.373.7 142.273.6 0.002c 0.726 0.384Ã Potassium (mmol/L) 4.4770.51 4.0870.54 4.2670.55 4.4770.53 3.9370.51 4.1870.57 4.4770.53 4.2370.56 4.3470.54 0.079 0.483 0.483 Chloride (mmol/L) 116.872.1 112.974.5 114.774.1 117.571.5 113.075.6 115.174.7 116.072.5 112.973.7 114.373.5 0.017d 0.585 0.652Ã Creatinine 390.67301.7 364.67316.9 376.67304.0 333.57234.8 135.3769.0 226.87189.6 447.77370.5 594.07300.1 526.57328.6 0.805 0.011e 0.111Ã phosphokinase (IU/L) Ã 2.55 significant P-values would be expected by chance in this analysis; thus, some of the 11 significant factors reported may be due to chance. Levene’s test is significant (Po0.05), indicating heterogeneity of variances. aSignificant Levene’s test indicates heterogeneity of variances, but significant effect of age confirmed by Mann–Whitney U-test directly comparing the two age classes (Z 5 2.057, P 5 0.04). bSignificant Levene’s test indicates heterogeneity of variances, but significant effect of age confirmed by Mann–Whitney U-test directly comparing the two age classes (Z 5 3.949, Po0.001). cSignificant Levene’s test indicates heterogeneity of variances, but significant effect of site confirmed by Mann–Whitney U-test directly comparing the two habitats (Z 5 3.135, P 5 0.001). dSignificant Levene’s test indicates heterogeneity of variances, but significant effect of site confirmed by Mann–Whitney U-test directly comparing the two habitats (Z 5 2.303, P 5 0.02). eSignificant Levene’s test indicates heterogeneity of variances, but significant effect of age confirmed by Mann–Whitney U-test directly comparing the two age classes (Z 5 2.487, P 5 0.01). Bold indicates a significant factor in the GLM (Po0.05). Sifaka Health in Fragmented Forest / 1019

Ã stimuli. Long-term changes in baseline values (as well as habitat differences) likely reﬂect underlying health or infection prevalence, while short-term spikes reﬂect recent infection. It is possible that the three individuals with high values had had recent

0.340 0.857 infections, and that ANOVA results with these individuals removed more closely reflect underlying health differences. However, the WBC distributions 0.007 were not altered in a way consistent with recent infection (specifically, increased segmented and band 0.459 0.762 0.874 0.874 49.4 0.054 0.764 0.437 1.42 0.493 0.370 0.355 1.65 0.114 1.000 0.460 174 0.206 0.453 0.903 7.5 0.638 0.577 0.328 7.7 0.119 0.072 0.671 0.678 neutrophils). There was a nonsignificant trend 7 7 7 7 7 7 7 7 toward neutrophilia and lymphopenia in FRAG 0.05), indicating heterogeneity of variances.

o groups, which on the surface seems consistent with P increased stress; however, the neutrophilia and lymphopenia were not associated with WBC counts 0.238 0.338 59.1 69.6 0.79 1.51 1.80 2.24 207 763 4.1 38.2 9.8 22.2 0.695 0.662 (data not shown). Choosing which outliers to remove 7 7 7 7 7 7 7 7 is especially problematic since two of the values (MAHA4 JUV: 8030 wbc/mcl, MAHA6 PG: 7260 wbc/ mcl) were statistical outliers, yet within the ‘‘normal’’ reference range clinically (for captive 0.659 0.300 18.3 90.6 1.89 1.07 1.59 2.46 129 804 10.6 37.2 3.7 24.4 0.532 0.943

Levene’s test is signiﬁcant ( lemurs). These two ‘‘borderline’’ values may still 7 7 7 7 7 7 7 7 Ã represent mild elevations due to past illness; con- versely, ‘‘normal’’ ranges may better represent captive animals but may be too high for stressed wild animals. In summary, the data suggest higher 0.411 0.383 35.6 45.2 1.46 2.02 1.05 1.98 203 716 13.9 39.5 4.1 19.5 0.567 0.333 -Values) 7 7 7 7 7 7 7 7 WBC in CONT groups, but more investigation is P necessary. The fact that FRAG animals had the lowest WBC is interesting, as lower body mass and reduced activity also suggest that this population is more 0.098 0.369 41.8 63.8 0.97 2.08 0.99 2.27 147 821 13.0 36.1 4.0 17.5 0.594 0.454 7 7 7 7 7 7 7 7 stressed [Irwin, 2006a; Irwin et al., 2007]. If infection prevalence is high, one might expect many individuals to have elevated WBC due to recent infections. An alternative hypothesis is that (notwithstanding 257 870 0.627 0.357 26.1 73.0 2.00 2.14 0.66 2.86 15.3 39.1 4.0 18.9 0.204 0.771 the outliers) low WBC counts in FRAG groups reﬂect 7 7 7 7 7 7 7 7 a weakened immune system due to underlying stress [including dietary differences among populations; Irwin, 2008a]. Several studies have shown that undernutrition causes reduced immune activity in 188 764 0.427 0.383 42.3 53.0 1.44 2.00 1.36 1.58 11.0 32.5 6.5 16.0 0.621 0.083 mammals, including calorie-restricted captive rats 7 7 7 7 7 7 7 7 [Cunha et al., 2003] and fasted captive American mink [Mustonen et al., 2005]. Undernutrition can also lead to increased infection prevalence and intensity [Coop & Kyriazakis, 1999]: for example, 0.05). 176 792 0.177 0.354 50.0 66.7 1.01 1.79 1.41 2.25 9.3 37.1 7.7 19.8 0.627 0.558

o increasing protein intake decreases infection inten- 7 7 7 7 7 7 7 7 P sity of a helminth parasite in lactating ewes [Houdijk

All animals Adultset al., 2009], 1–4 yr old and food supplementation decreases infection intensity of a helminth parasite in wild showshoe hares [Murray et al., 1998]. 195 837 0.613 0.329 21.9 81.8 1.85 1.61 1.18 2.66 13.1 38.1 4.1 21.6 0.406 0.857 7 7 7 7 7 7 7 7 12 14 26 6 7Several site 13 and agedifferences 6 were 7 also apparent 13 Site Age Interaction

CONT FRAG TOTAL CONT FRAG TOTAL CONTin serum FRAG biochemistry TOTAL Effect of Site, Age (Table III). FRAG individuals 49.1 2.01 17.8 0.208 were lower in total bilirubin, total protein, albumin, calcium, sodium, and chloride than CONT individuals. Bilirubin is a breakdown product of heme catabolism. g/dL) 740 -values would be expected by chance in this analysis; thus, the single signiﬁcant factor reported may be due to chance. m g/dL) 0.383 P

m Clinically, elevated bilirubin indicates impairment of g/dL) 36.0

m normal liver function or an increased breakdown of zeaxanthin

1 erythrocytes. Thus, the observed differences may g/dL) g/dL) g/dL) m m m ( ( Vitamin D (nmol/L) ( reflect depressed liver function in CONT groups, or -tocopherol ( -carotene ( a b Gamma tocopherol Lutein Retinyl palmitate 1.78 Retinol ( 25-Hydroxy N Bold indicates a significant factor in the GLM ( TABLE IV. Serum Fat-Soluble Vitamin Analysis, With Results of GLM Analyses ( Retinyl stearate 1.2 significant slower erythrocyte turnover in FRAG groups. Total

Am. J. Primatol. m .Primatol. J. Am. al. et Irwin / 1020

TABLE V. Serum Mineral and Iron Analyte Analysis, With Results of GLM Analyses (P-Values)

All animals Adults 1–4 yr old

CONT FRAG TOTAL CONT FRAG TOTAL CONT FRAG TOTAL Effect of

N 12 14 26 6 7 13 6 7 13 Site Age Interaction

Co (ng/mL) 3.9572.66 3.0071.75 3.4472.22 4.4973.26 2.4971.46 3.4272.56 3.4172.07 3.5171.98 3.4771.93 0.295 0.972 0.245 Cu (mg/mL) 1.0370.18 1.0170.20 1.0270.19 1.0970.13 0.9970.26 1.0470.21 0.9670.20 1.0370.14 1.0070.17 0.871 0.539 0.252 Mn (ng/mL) 5.1872.08 3.6071.03 4.3371.76 4.7272.02 3.6671.34 4.1571.71 5.6372.21 3.5470.71 4.5171.86 0.023 0.539 0.431 Mo (ng/mL) 3.1573.47 1.9071.76 2.4872.71 3.0373.95 1.8671.50 2.4072.83 3.2773.30 1.9472.11 2.5572.69 0.269 0.886 0.947 Se (ng/mL) 19.079.0 19.873.1 19.476.4 16.574.5 18.973.9 17.874.2 21.5711.9 20.771.7 21.177.8 0.760 0.191 0.543Ã Zn (mg/mL) 0.85070.139 0.63370.097 0.72870.159 0.81470.152 0.61970.121 0.70070.163 0.88070.134 0.64770.073 0.75570.157 o0.001 0.340 0.703 (11) (25) (5) (12) Iron (mg/dL) 3097137 192769 2437118 3477191 154743 2357155 277775 229773 251775 0.008a 0.954 0.092Ã (11) (25) (5) (12) TIBC (mg/dL) 5897114 469742 5227100 6057170 465748 5237130 575749 473738 520767 0.002b 0.760 0.581Ã (11) (25) (5) (12) Ferritin (ng/mL) 81.4743.6 120.9779.4 103.5767.9 70.2729.1 132.9798.2 106.8781.3 90.7753.9 108.9760.8 100.5756.1 0.158 0.950 0.429 (11) (25) (5) (12) Transferrin 51.0713.3 41.3714.6 45.5714.6 54.7716.3 33.278.9 42.2716.2 47.8710.8 49.3715.2 48.6712.8 0.069 0.389 0.038 saturation (%) (11) (25) (5) (12) Ã 1.5 significant P-values would be expected by chance in this analysis; thus, some of the three significant factors reported may be due to chance. Levene’s test is significant (Po0.05), indicating heterogeneity of variances. aSignificant Levene’s test indicates heterogeneity of variances, but significant effect of site confirmed by Mann–Whitney U-test directly comparing the two habitats (Z 5 2.409, P 5 0.02). bSignificant Levene’s test indicates heterogeneity of variances, but significant effect of site confirmed by Mann–Whitney U-test directly comparing the two habitats (Z 5 2.985, P 5 0.02). Bold indicates a significant factor in the GLM (Po0.05). Sifaka Health in Fragmented Forest / 1021

protein is the sum of albumin (which is produced in the involves several compartments and movement be- liver and reflects nutritional status) and globulin tween the compartments depends on health status, (which is produced by the immune system and reflects nutritional status, and physiological function. Recent disease exposure). The combination of lower total research has clarified the significance of iron in protein and albumin (but not globulin) in FRAG disease of captive lemurs and evaluated monitoring groups therefore most likely reflects lower protein mechanisms [Glenn et al., 2006; Williams et al., intake; dietary differences have been documented 2006, 2008]. Dietary iron is absorbed from the [Irwin, 2008a] and nutritional differences are currently intestinal tract by the transfer molecule transferrin, under investigation. The lower levels of some and is stored in the body as soluble ferritin or serum electrolytes (calcium, sodium, and chloride) insoluble hemosiderin. Under normal physiological observed in FRAG groups arealsorelatedtointake conditions serum iron remains stable, but serum and may reflect further nutritional differences between ferritin levels vary with iron load (ferritin production the two habitats. increases with iron availability), while transferrin Adults were lower in aspartate transaminase saturation reflects dietary intake (high dietary iron (AST), alanine aminotransferase (ALT), alkaline causes increased transferrin saturation) [Smith, phosphatase, calcium, and creatine phosphokinase 1997]. The combination of lower serum iron, lower (CPK). This is expected, as these parameters are TIBC (which reflects fewer transferrin molecules in typically higher in subadult individuals due to active the blood), and lower transferrin saturation (which growth (e.g., alkaline phosphatase is a byproduct of approaches significance) in FRAG animals is consistent osteoblast activity and is elevated when bone growth with lower dietary iron intake. Low serum iron reflects is occurring). low intake directly, and the magnitude of the difference Among fat-soluble vitamins, only one significant (38% reduction in FRAG animals) suggests a rather factor was found: fragment individuals had higher large dietary difference. Low transferrin saturation levels of retinyl stearate (a form of vitamin A). Again, indicates that the transfer protein (transferrin) is not this is related entirely to diet, as serum concentrations being utilized to the maximum capacity (transferrin depend on intake. However, the fact that only one of 24 saturation increases when dietary iron increases as significance tests yielded Po0.05 raises the possibility more is absorbed and bound, and decreases when that this significant finding may be due to chance. dietary iron decreases), and lower TIBC may reflect a Among minerals and iron analytes, FRAG reducedneedforthistransferprotein. individuals had lower levels of manganese, zinc, and iron, and lower total iron-binding capacity (TIBC). There was a significant age–habitat interaction for Comparisons With Other Lemur Populations transferrin saturation: immatures had similar values Utilizing the Prosimian Biomedical Survey across habitats, while CONT adults had higher values Project Database, comparisons to other lemur species and FRAG adults had lower values. The mineral can be made. Interspecific comparisons of serum differences (manganese, zinc, and iron) are due to biochemistry reveal large differences in blood urea intake differences, and iron analyte differences may nitrogen (BUN). Tsinjoarivo P. diadema have the also reflect iron intake. Iron is an essential element lowest mean value of any lemur species sampled, at and involved in many biological systems; its metabolism 4.42 mg/dL (Table VI). Other lemur genera and

TABLE VI. Variance in BUN Across Lemur Species in Madagascar

BUN (mg/dL)

Species Site Mean Range N Reference

Propithecus diadema Tsinjoarivo 4.4271.53 2–7 26 This study Propithecus diadema Mangerivola 5.070 5–5 2 PBSP Propithecus diadema Mantadia 6.6771.86 5–8 6 PBSP Propithecus tattersalli Daraina 26.30719.17 4–85 25 PBSP Propithecus tattersalli Daraina 16.275.02 n/a 30 Garell and Meyers [1995] Propithecus deckeni Tsiombikibo 50.6718.1 28–90 25 Junge and Louis [2005b] Indri indri Various 10.8273.33 6–17 35 Junge and Louis [2002]; PBSP Eulemur albifrons Various 10.6379.58 0–50 52 Junge et al. [2008] Eulemur macaco Lokobe 17.4075.70 8–29 23 Junge and Louis [2007] Eulemur rufus Tsiombikibo 22.80 715.30 7–54 15 Junge and Louis [2005b] Lemur catta Tsimanampetsotsa 13.3074.50 5–20 20 Dutton et al. [2003] Varecia variegata Several 9.2676.40 2–19 60 Junge and Louis [2005a] Variegata rubra Masoala 8.5874.49 3–19 24 Dutton et al. [2008]

PBSP, Prosimian Biomedical Survey Project (R. Junge, unpublished data); BUN, blood urea nitrogen.

Am. J. Primatol. 1022 / Irwin et al.

species display a wide range of values, up to ten times high BUN values are related to hydration state. Most higher (51 mg/dL). Urea synthesis provides a me- notably, the P. deckeni population was sampled in a chanism for ammonium excretion, and protein is the dry habitat in the dry season (July); thus, hydration major source of ammonium for urea synthesis. status may have contributed to elevated BUN levels. Changes in BUN can reflect altered renal function More research is necessary to determine whether (as urea is excreted by the kidneys), altered protein BUN can be a reliable indicator of protein intake. metabolism, or altered state of hydration [Finco, In terms of iron analytes (iron, ferritin, TIBC, 1997]. Since there is no concurrent evidence of renal and transferrin saturation), interspecific compari- compromise, variability in BUN across lemurs most sons suggest that iron may not be a limiting factor likely reflects protein intake or state of hydration. for the Tsinjoarivo sifakas (despite the difference BUN is therefore a potentially useful indicator of between CONT and FRAG individuals). Tsinjoarivo dietary protein (in contrast, total protein reflects sifakas’ average levels of iron, TIBC, and transferrin physiologic and immunologic proteins and is often saturation are higher than those of nine captive not affected by intake). lemur species [Williams et al., 2006], and wild Lemur However, BUN does not sort easily with diet catta [Dutton et al., 2003], and their iron and TIBC type. Strict frugivores (Varecia) have low values, the levels are higher than wild Propithecus deckeni, more omnivorous Eulemur and Lemur are inter- Eulemur rufus, Eulemur albifrons, E. macaco, and mediate, and the mostly folivorous Propithecus Varecia rubra [Dutton et al., 2003, 2008; Junge & species range from very low (Tsinjoarivo and other Louis, 2007; Junge et al., 2008]. Thus, although little P. diadema populations) to intermediate (P. tatter- is known about species-specific requirements or salli) to very high (P. deckeni). This wide intragene- reference ranges, comparative data suggest that ric variation may be related to habitat: P. diadema Tsinjoarivo sifakas might not be iron-stressed, and live in eastern humid rainforests, P. deckeni lives in the difference between CONT and FRAG animals western dry, deciduous forest, and P. tattersalli lives may not be clinically relevant. in intermediate habitat. Leaves in drier, more Finally, Tsinjoarivo sifakas have extremely low seasonally deciduous habitats typically have higher serumselenium(19.476.4 ng/mL). This is consider- turnover, higher metabolic rates, and are typically of ably lower than other lemur species sampled to date: higher quality (more protein, less fiber) than leaves P. diadema at other sites (56.0071.41, n 5 2), in less seasonal habitats [van Schaik et al., 2005]. E. albifrons from various sites (92.05753.31, n 5 50), Thus, Propithecus species in highly deciduous wes- and Varecia variegata from various sites tern forest (such as P. deckeni at Tsiombikibo) may (254.57747.56, n 5 28) [Junge et al., 2008; R. Junge, have much higher available protein intake than do Prosimian Biomedical Survey Project, unpublished species in evergreen rainforest (such as P. diadema), data]. Selenium values in plants typically reflect soil which could converge upon the low-protein diets of concentrations, which are known to vary substantially strict frugivores. Interestingly, although the sample depending on underlying geology [Lenz & Lens, 2009]. is small, BUN seems to increase with seasonality. In areas with low naturally occurring selenium, This may also be related to Propithecus population deficiencies cause health problems in humans, wild density [Irwin, 2006b], which is lowest in rainforest mammals and birds, and livestock; in extreme cases species (1–10 individuals/km2), highest in dry forest muscle membranes rupture and leak cellular contents, species (37–500 individuals/km2), and intermediate resulting in nonfunctional muscle [‘‘white muscle in P. tattersalli (60–70 individuals/km2). If BUN disease’’; Robbins, 2001]. Values for domestic livestock indeed reflects dietary protein, this suggests that range between 80 and 500 ng/mL [Radostits et al., protein intake of P. diadema at Tsinjoarivo is 2007] and are considered deficient at values less than relatively low [specifically, the intake of available 25–60 ng/mL. Carnivore values for serum selenium are protein; Rothman et al., 2008], with little difference in the 200–300 ng/mL range, with values less than between continuous and fragmented habitat. Pre- 120 ng/mL considered deficient. However, minimum liminary analyses suggest low dietary protein for requirements in primates have not been well defined. Tsinjoarivo sifakas (average available protein con- A range of 10–150 ng/mL has been reported for rhesus tent of analyzed foods, dry matter basis 5 8.7%; macaques, and it has been suggested that relatively Irwin, unpub. data), as well as for P. diadema at low levels in primates suggest they may be ‘‘resistant Mantadia [6.8%; Powzyk & Mowry, 2003] and a close to selenium deficiency’’ [Butler et al., 1982]. relative, P. edwardsi at Ranomafana [5–6%; Arrigo- Nelson, 2006]. As these levels are close to or below measured protein requirements for similar-sized Management Implications primates [National Research Council, 2003; Oftedal, The data presented here have direct relevance to 1991], protein should be considered as a potential managing sifaka populations in captivity and in the factor limiting population density and individual wild. Previous studies have shown that sifakas growth for rainforest sifakas. However, one alter- persist in small, degraded fragments at Tsinjoarivo native explanation worth investigating is that some over the short term, but various lines of evidence

Am. J. Primatol. Sifaka Health in Fragmented Forest / 1023

suggest that fragment-living groups may have a could become an important part of a toolkit for rapid lower-quality diet and reduced energetic status, thus assessment of the viability of wild primate popula- their long-term survival prospects remain uncertain. tions in degraded habitats. The data presented here directly reflect physiological health of sifaka groups in habitats with varying degrees of degradation, and suggest that degradation ACKNOWLEDGMENTS may have impacts on the physiology, and possibly the We thank the Government of Madagascar, CAFF/ health, of sifakas. It is especially striking that 12 of CORE and the Ministry of Environment, Forests and 40 variables derived from blood analyses indicated a Tourism for research authorization. For research significant effect of habitat (11 if packed cell volume facilitation we thank P. Wright, B. Andriamihaja, the is removed); this is much more than expected by Malagasy Institute for the Conservation of Tropical chance (approximately 2 of 40). Environments (MICET), and the Institute for the This study also provides important data that will Conservation of Tropical Environments (ICTE). be useful for maintaining sifakas in captivity. This We are grateful for funding from National Geographic study greatly increases the sample size of ‘‘baseline’’ Society Committee for Research and Exploration, health data for this genus and species, and also St. Louis Zoo Field Research for Conservation Pro- reveals a range of variation for unstressed and gram, Margot Marsh Biodiversity Foundation, and the stressed sifakas in the wild, providing a more support of an NSERC postdoctoral fellowship to meaningful context for judging if captive animals M. T. I. We thank L. Jones-Engel and L. Kuller are physiologically stressed. Rainforest sifakas have (Washington National Primate Research Center) for typically been difficult to keep in captivity, with only serological testing. For invaluable capture and data one living specimen worldwide; if captive breeding collection assistance, we thank research assistants becomes a component of sifaka conservation strate- E. Razanadrakoto, H. Rakotoarimanana, E. Ranaivoson, gies, understanding their dietary needs will be J. Rakotofanala, H. Rakotondratsimba, N. Rakotoniaina, crucial for success. A. Rahajasoa, J.-P. Rakotorahalahy, N. Razanadrakoto, Future study is necessary before we can make C. Razanadrakoto, L. Danielson, and R. Nelson. We are generalizations about how lemurs and other pri- grateful to K. Glander for training in animal capture. mates respond to habitat disturbance. This study C. Chapman, J. Rothman, and two anonymous reviewers focuses on a relatively large and mostly folivorous provided useful comments on an earlier draft of this lemur; the proximate pressures of habitat distur- manuscript. bance likely differ among species, and these pressures might be reflected in different baseline physiological parameters. Previous studies relating REFERENCES a species’ ecology to its tolerance of disturbance and Arrigo-Nelson SJ. 2006. The impact of habitat disturbance on fragmentation have been lacking since they have the feeding ecology of the Milne-Edwards’ Sifaka (Propithecus relied mainly on observational data, population edwardsi) in Ranomafana National Park, Madagascar [PhD]. density, and biomass [Irwin, 2008b; Johns & Stony Brook, NY: Stony Brook University. 181p. Skorupa, 1987; van Horne, 1983]; unless data reflect Butler JA, Whanger PD, Tripp MJ. 1982. Blood selenium and glutathione peroxidase activity in pregnant women: com- long-term population trends [Chapman et al., 2000], parative assays in primates and other animals. American there is a danger of inferring a population is healthy Journal of Clinical Nutrition 36:15–23. when it is in fact artificially crowded, stressed, and Chapman CA, Peres CA. 2001. Primate conservation in the failing to reproduce. Considering physiological data new millenium: the role of scientists. Evolutionary Anthro- that directly reflect health may: (1) reveal important pology 10:16–33. Chapman CA, Balcomb SR, Gillespie TR, Skorupa JP, differences among populations, which might impact Struhsaker TR. 2000. Long-term effects of logging on population viability, (2) aid in the identification of African primate communities: a 28-year comparison from macro- and micronutrients that may limit growth Kibale National Park, Uganda. Conservation Biology and reproduction, and (3) contribute to a broader 14:207–217. Chapman CA, Wasserman MD, Gillespie TR, Speirs ML, understanding of how interspecific ecological differ- Lawes MJ, Saj TL, Ziegler TE. 2006. Do food availability, ences cause different likelihoods of extirpation in parasitism and stress have synergistic effects on red colobus disturbed habitats. populations living in forest fragments? American Journal of Finally, it is important to remember that the Physical Anthropology 131:525–534. links between habitat disturbance, physiology, and Chiarello AG, de Melo FR. 2001. Primate population densities and sizes in Atlantic forest remnants of Northern Espı`rito population health remain poorly known. Demon- Santo, Brazil. International Journal of Primatology strating physiological differences between habitats is 22:379–396. an important first step, but a great deal of work Coop RL, Kyriazakis I. 1999. Nutrition-parasite interaction. will be necessary to determine whether these Veterinary Parasitology 84:187–204. Cunha WDS, Friedler G, Vaisberg M, Egami MI, Rosa L. differences have any consequences for population 2003. Immunosuppression in undernourished rats: the health. However, if these links can be found, effect of glutamine supplementation. Clinical Nutrition physiological health assessments such as this one 22:453–457.

Am. J. Primatol. 1024 / Irwin et al.

Dutton CJ, Junge RE, Louis EE. 2003. Biomedical evaluation information systems (GIS). Forest Ecology and Manage- of free-ranging ring-tailed lemurs (Lemur catta) in Tsima- ment 98:35–47. nampetsotsa Strict Nature Reserve, Madagascar. Journal of Junge RE, Garell D. 1995. Veterinary examination of ruffed Zoo and Wildlife Medicine 34:16–24. lemurs (Varecia variegata) in Madagascar. Primate Con- Dutton CJ, Junge RE, Louis EE. 2008. Biomedical evaluation servation 16:44–46. of free-ranging red ruffed lemurs (Varecia rubra) within the Junge RE, Louis EE. 2002. Medical evaluation of free-ranging Masoala National Park, Madagascar. Journal of Zoo and primates in Betampona Reserve, Madagascar. Lemur News Wildlife Medicine 39:76–85. 7:23–25. Estrada A, Juan-Solano S, Martıńez TO, Coates-Estrada R. Junge RE, Louis EE. 2005a. Preliminary biomedical evalua- 1999. Feeding and general activity patterns of a howler tion of wild ruffed lemurs (Varecia variegata and V. rubra). monkey (Alouatta palliata) troop living in a forest fragment American Journal of Primatology 66:85–94. at Los Tuxtlas, Mexico. American Journal of Primatology Junge RE, Louis EE. 2005b. Biomedical evaluation of two 48:167–183. sympatric lemur species (Propithecus verreauxi deckeni and Fahrig L. 2003. Effects of habitat fragmentation on biodiver- Eulemur fulvus rufus) in Tsiombokibo Classified Forest, sity. Annual Review of Ecology, Evolution and Systematics Madagascar. Journal of Zoo and Wildlife Medicine 34:487–515. 36:581–589. Finco DR. 1997. Kidney function. In: Keneko JJ, Harvey JW, Junge RE, Louis EE. 2007. Biomedical evaluation of black Bruss ML, editors. Clinical biochemistry of domestic lemurs (Eulemur macaco macaco) in Lokobe Reserve, animals, 5th ed. San Diego: Academic Press. p 441–484. Madagascar. Journal of Zoo and Wildlife Medicine 38:67–76. Ganzhorn JU, Goodman SM, Dehgan A. 2003. Effects of forest frag- Junge RE, Dutton CJ, Knightly F, Williams CV, Rasambainarivo mentation on small mammals and lemurs. In: Goodman SM, FT, Louis EE. 2008. Comparison of biomedical evaluation for Benstead JP, editors. The natural history of Madagascar. white-fronted brown lemurs (Eulemur fulvus albifrons)from Chicago: University of Chicago Press. p 1228–1234. four sites in Madagascar. Journal of Zoo and Wildlife Medicine Garell DM, Meyers DM. 1995. Hematology and serum 39:567–575. chemistry values for free-ranging golden crowned sifaka Laurance WF, Vasconcelos HL, Lovejoy TE. 2000. Forest loss (Propithecus tattersalli). Journal of Zoo and Wildlife and fragmentation in the Amazon: implications for wildlife Medicine 26:382–386. conservation. Oryx 34:39–45. Glenn KM, Campbell JL, Rotstein D, Williams CV. 2006. Laurance WF, Lovejoy TE, Vasconcelos HL, Bruna EM, Retrospective evaluation of the incidence and severity of Didham RK, Stouffer PC, Gascon C, Bierregaard Jr RO, hemosiderosis in a large captive lemur population. Laurance SG, Sampaio E. 2002. Ecosystem decay of American Journal of Primatology 68:369–381. Amazonian forest fragments: a 22-year investigation. Con- Goodman SM, Rakotondravony D. 2000. The effects of forest servation Biology 16:605–618. fragmentation and isolation on insectivorous small mam- Lehman SM, Rajaonson A, Day S. 2006a. Lemur responses to mals (Lipotyphla) on the Central High Plateau of Mada- edge effects in the Vohibola III Classified Forest, Madagascar. gascar. Journal of Zoology (London) 250:193–200. American Journal of Primatology 68:293–299. Harper GJ, Steininger MK, Tucker CJ, Juhn D, Hawkins F. Lehman SM, Rajaonson A, Day S. 2006b. Edge effects and 2007. Fifty years of deforestation and forest fragmentation their influence on lemur density and distribution in south- in Madagascar. Environmental Conservation 34:325–333. east Madagascar. American Journal of Physical Anthropo- Houdijk JGM, Jackson F, Kyriazakis I. 2009. Nutritional logy 129:232–241. sensitivity of resistance to Trichostrongylus colubriformis in Lenz M, Lens PNL. 2009. The essential toxin: the changing lactating ewes. Veterinary Parasitology 160:258–266. perception of Selenium in environmental sciences. Science Ingram JC, Dawson TP, Whittaker RJ. 2004. Mapping tropical of the Total Environment 407:3620–3633. forest structure in southeastern Madagascar using remote Mayor MI, Sommer JA, Houck ML, Zaonarivelo JR, Wright sensing and artificial neural networks. Remote Sensing of PC, Ingram C, Engel SR, Louis Jr EE. 2004. Specific status Environment 94:491–507. of Propithecus spp. International Journal of Primatology Irwin MT. 2006a. Ecological impacts of forest fragmentation 25:875–900. on diademed sifakas (Propithecus diadema) at Tsinjoarivo, Menon S, Poirier FE. 1996. Lion-tailed macaques (Macaca eastern Madagascar: implications for conservation in frag- silenus) in a disturbed forest fragment: activity patterns and mented landscapes [PhD]. Stony Brook, NY: Stony Brook time budget. International Journal of Primatology 17:969–985. University. 410p. Mittermeier RA, Konstant WR, Hawkins F, Louis EE, Irwin MT. 2006b. Ecologically enigmatic lemurs: the sifakas of Langrand O, Ratsimbazafy J, Rasoloarison R, Ganzhorn JU, the eastern forests (Propithecus candidus, P. diadema, Rajaobelina S, Tattersall I, Meyers DM. 2006. Lemurs of P. edwardsi, P. perrieri, and P. tattersalli). In: Gould L, Madagascar. Washington, DC: Conservation International. Sauther ML, editors. Lemurs: ecology and adaptation. New 520p. York: Springer. p 305–326. Moran MD. 2003. Arguments for rejecting the sequential Irwin MT. 2008a. Feeding ecology of diademed sifakas Bonferroni in ecological studies. Oikos 100:403–405. (Propithecus diadema) in forest fragments and continuous Murray DL, Keith LB, Cary JR. 1998. Do parasitism and forest. International Journal of Primatology 29:95–115. nutritional status interact to affect production in snowshoe Irwin MT. 2008b. Diademed sifaka (Propithecus diadema) hares? Ecology 79:1209–1222. ranging and habitat use in continuous and fragmented Mustonen AM, Pyykonen T, Paakkonen T, Ryokkynen A, forest: higher density but lower viability in fragments? Asikainen J, Aho J, Mononen J, Nieminen P. 2005. Biotropica 40:231–240. Adaptations to fasting in the American mink (Mustela Irwin MT, Glander KE, Raharison J-L, Samonds KE. 2007. vison): carbohydrate and lipid metabolism. Comparative Effect of habitat and sex on body mass and morphometrics Biochemistry and Physiology A-Molecular & Integrative of diademed sifakas (Propithecus diadema). American Physiology 140:195–202. Journal of Physical Anthropology 132:134. National Research Council. 2003. Nutrient requirements of Johns AD, Skorupa JP. 1987. Responses of rain-forest nonhuman primates. Washington, DC: National Academies primates to habitat disturbance: a review. International Press. 286p. Journal of Primatology 8:157–191. North A, Ovaskainen O. 2007. Interactions between dispersal, Jorge LAB, Garcia GJ. 1997. A study of habitat fragmentation competition, and landscape heterogeneity. Oikos 116: in Southeastern Brazil using remote sensing and geographic 1106–1119.

Am. J. Primatol. Sifaka Health in Fragmented Forest / 1025

Nupp TE, Swihart RK. 2000. Landscape-level correlates of protein intake of primates. American Journal of Primato- small-mammal assemblages in forest fragments of farm- logy 70:690–694. land. Journal of Mammalogy 81:512–526. Smith JE. 1997. Iron metabolism and its disorders. In: Oftedal OT. 1991. The nutritional consequences of foraging in Keneko JJ, Harvey JW, Bruss ML, editors. Clinical biochem- primates: the relationship of nutrient intakes to nutrient istry of domestic animals, 5th ed. San Diego: Academic requirements. Philosophical Transactions of the Press. p 223–239. Royal Society of London Series B-Biological Sciences Sokal RR, Rohlf FJ. 1995. Biometry. New York: W. H. 334:161–170. Freeman. 887p. Onderdonk DA, Chapman CA. 2000. Coping with forest van Horne B. 1983. Density as a misleading indicator of habitat fragmentation: the primates of Kibale National Park, quality. Journal of Wildlife Management 47:893–901. Uganda. International Journal of Primatology 21:587–611. van Schaik CP, Madden R, Ganzhorn JU. 2005. Seasonality Powzyk JA, Mowry CB. 2003. Dietary and feeding differences and primate communities. In: Brockman DK, van Schaik CP, between sympatric Propithecus diadema diadema and Indri editors. Seasonality in primates: studies of living indri. International Journal of Primatology 24:1143–1162. and extinct human and non-human primates. Cambridge: Pulliam HR. 1988. Sources, sinks and population regulation. Cambridge University Press. p 445–463. American Naturalist 132:652–661. Williams CV, Campbell J, Glenn KM. 2006. Comparison of Radostits OM, Gay CC, Hinchcliff KW, Constable PD. 2007. serum iron, total iron binding capacity, ferritin, and percent Veterinary medicine: a textbook of the diseases of cattle, transferrin saturation in nine species of apparently healthy horses, sheep, pigs and goats. Philadelphia, PA: Saunders captive lemurs. American Journal of Primatology 68: Ltd. 2065p. 477–489. Redford KH. 1992. The empty forest. BioScience 42:412–422. Williams CV, Junge RE, Stalis IH. 2008. Evaluation of iron Robbins CT. 2001. Wildlife feeding and nutrition. New York: status in lemurs by analysis of serum iron and ferritin Academic Press. 352p. concentrations, total iron-binding capacity, and transferrin Rothman JM, Chapman CA, Pell AN. 2008. Fiber-bound saturation. Journal of the American Veterinary Medical nitrogen in gorilla diets: implications for estimating dietary Association 232:578–585.

Am. J. Primatol.