Female tannin avoidance: a possible explanation for habitat and dietary segregation of gira¡es (Gira¡a camelopardalis peralta)inNiger

Lauren E. Caister1,William M. Shields2 and Allen Gosser3 1SUNY College of Environmental Science & Forestry,133 IllickHall,1Fo restry Dr, Syracuse, NY13210, USA, 2SUNY College of Environmental Science & Forestry,1Forestry Drive, Syracuse, NY13210, USA, 3USDA, APHIS,WS,1930 Rout 9, Castleton, NY 12033, USA

les re¨ gions qu'elle occupe actuellement. L'habitat actuel Abstract de saison se© che comprend deux zones de ve¨ ge¨ tation dis- is host to the last free-roaming herd of G.c. peralta tinctes. Dans les deux zones, les girafes doivent cohabiter (Gira¡a camelopardalis peralta).We examined the foraging avec les gens de la re¨ gion. Les girafes pre¨ sentent une se¨ g- preferences of these gira¡e in their dry-season habitats, re¨ gation sexuelle tant dans la se¨ lection de l' habitat de with the goal of preserving the herd in the regions that saison se© che que dans les choix alimentaires. Les femelles they currently inhabit. The current dry-season habitat allaitantes manifestent une forte pre¨ fe¨ rence pour la zone comprises two distinct vegetation zones. In both of these interme¨ diaire (IM). Les maª les et les femelles pleines mon- zones the gira¡e must exist alongside the people of this trent une pre¨ fe¨ rence pour le Dallol Bosso (DB). Les region. The gira¡es exhibit a sexual segregation in their femelles qui s'occupent de jeunes e¨ vitent visiblement les dry-season habitat selection and forage choices. The tanins. Les maª les et les femelles sans jeune pre¨ fe© rent un females show a strong preference for the intermediate fourrage riche en prote¨ ines et en graisses, tandis que les zone (IM) when lactating. The males and pregnant subadultes manifestent une forte pre¨ fe¨ rence pour un females show a preference for the Dallol Bosso (DB). Nur- contenu e¨ leve¨ en prote¨ ines et en hydrates de carbone et sing cows exhibit an avoidance of tannins. Bulls and des taux de tanins mode¨ re¨ s. glutinosum est non-nursingcows prefer high proteinand high fat forage, l'espe© ce pre¨ fe¨ re¨ e des adultes des deux sexes dans la IM. while subadults showa strong preference forhighprotein Les maª les et les femelles pre¨ fe© rent nettement Acacia nilo- and carbohydrate contents and moderate tannin levels. tica et Acacia seyal dans le DB. Les subadultes des deux Combretum glutinosum is the preferred species for adults sexes pre¨ fe© rent nettement Prosopis africana dans la IM. of both sexes in the IM. Males and females have strong Contrairement aux femelles, les maª les conservent leur preferences for both Acacia nilotica and Acacia seyal in pre¨ fe¨ rence pourA. nilotica lorsqu'ils sont dans la IM. the DB. Sub-adults of both sexes strongly prefer Prosopis africana in the IM. Unlike females, males retain their pre- ference forA. nilotica when in the IM. Introduction Keywords: foraging preference, gira¡e, G.c. peralta, Gira¡a Niger is home to the last free-roaming herd of gira¡es camelopardalis, Niger, west African browsers. (Gira¡a camelopardalis peralta) in west and the last free-roaming herd of the race peralta. The herd in Niger Re¨ sume¨ decreased throughout the 1960s and 1970s. The severe Le Niger abrite la dernie© re population sauvage de Gira¡a droughts of the 1970s and 1980s in Niger helped to camelopardalis peralta. Nous avons e¨ tudie¨ les pre¨ fe¨ rences further degrade the landscape and decimate wildlife alimentaires de ces girafes dans leurs habitats de saison populations. Prior to the droughts, the gira¡es were se© che, dans le but de maintenir cette population dans widespread and abundant in Niger (Ciofolo, 1995). After the droughts, the gira¡es were forced into a small area surrounding the Koure¨ region and the Dallol Bosso, Correspondence: Dr William Shields. E-mail: [email protected] between Bale¨ yara and Birnin Gaoure¨ .While this de¢nes

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 201 202 Lauren E. Caister et al. the group's current range, some long distance move- Data collection ments by lone adults have been observed (Le Pendu & Ciofolo,1999). Every gira¡e was identi¢ed through a detailed photo log. The gira¡es now inhabit the Harikanassou region, The gira¡es were photographed from both sides and where they were rarely seen before, and are now inter- assigned identi¢cation numbers at ¢rst sighting. When acting regularly with humans. They are protected from possible, the entire body,and in every instance the entire hunting in Niger,where killing a gira¡e is now a criminal neck, was photographed. The original photo record was o¡ence.The region has no natural gira¡e predators other initiated by Peace Corps Volunteer Paul Peterson in than man, and only domestic herbivores as potential 1995, revamped by PURNKO ^ a Dutch conservation competitors. The occupancy of a new range, combined organization ^ in 1996 and continued by our team. We with the government'sinterest in developing ecotourism, also recorded pertinent information about the gira¡es inspired us to undertake a study of gira¡es in the region. (i.e.sex,injuries,o¡spring,maternities,agesanddeaths). A dietary assessment of the herd was requested by Peace The photographic record allowed us to determine the Corps Niger on behalf of the government of Niger. The age of each gira¡e for the duration of the study. Gira¡es preliminary questions we hoped to explore included: (i) were designated as adults (>4 years old), subadults Where were the gira¡es living and what were they eating (between 18 months and 4 years old), juveniles (6^ in their new range? (ii) Why had the gira¡es settled in 18 months old), and newborns (<6 months old). Their this region, and (iii) What, if anything, could be done to ages were determined based on known dates of birth or help ensure their continued presence? on the age of the individual at the ¢rst recorded sighting and/or photograph (some of which dated back to 1990). Methods Foster (1966) ¢rst developed the concept of the photo- record. As he noted, coat patterns on gira¡es do not change with age, and are unique to each individual. Study site The gira¡es were tracked visually while driving The gira¡es inhabit three vegetation zones in their core Yamaha 125 cc dirt bikes. We selected the closest gira¡e range. The core range is subdivided into dry season and for focal animal observation, but tried not to sample the wet season habitats. The dry season habitat is in the same animal more frequently than others in the herd. Koure¨ ^Harikanassou region (termed the intermediate Each location where the gira¡es were ¢rst encountered zone) and the Dallol Bosso. The average annual rainfall was recorded using a Grumman hand-held geographical of between 400 mm and 500 mm de¢nes this region as positioning systems (GPS). Most of our observations part of the . The herd is there from October to May, were made during the early morning (07.00^11.00 h). after which they congregate in the Koure¨ region follow- We did not observe the gira¡es eating after noon. They ing the ¢rst . The herd is widespread in the dry sea- spent most of their time resting in the shade until early son habitat.We worked in the Harikanassou area, about evening, presumably due to the high air temperatures, 85km south-east of Niamey,fromJanuary1997to Febru- which often reached 49 8C. This was not true during the ary1999. cold-season (late December^February), when the gira¡es The Dallol Bosso (DB) is an ancient riverbed with a could be observed foraging until late afternoon. During shallow water table, as shallow as 0.5 m in the valleys each focal sample we recorded: (1) the species of each and hollows. Many large, permanent waterholes and food item, (2) the height of each feeding bout, (3) the ponds also characterize this region. These ponds are duration of each feeding bout in seconds, (4) which part often shallow (<1.5m deep) and marshy. This zone is of the was being taken, (5) the individual focal ani- marked by comparatively fertile and an abundance mal,and(6) the identityofallof the othergira¡espresent of and species around the wetlands. The inter- (modi¢ed fromYoung & Isbell,1991). mediate zone (IM) is more typical of the Sahel. It has To document the available vegetation, 20 plots were sandy soils, scattered trees and low scrub brush. Most of randomly selected from the entire study area. An addi- the region is farmed in the rainy season and left fallow tional ¢ve plots were each randomly selected from the during the dry season. This area has a deep water table, Dallol Bosso (DB) and Intermediate Zone (IM).We selec- often being as much as 50^60 m down. ted an extra two plots in the IM which were frequented

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 Female tannin avoidance in giraffes 203

Table1 Known chemical compositions of in the G.c. peraltas'dry-season habitat in Niger, from dry weight samples

%Crude %Crude Tannin Protein % Fats Fibre % Ash % Moisture % NFE Sourcey Content

Acacia species A. nilotica 12.9 5.1 13.7 1.1 52.0 32.2 (1) Pods 30% (2)ÃÃ 20.0% (2) A. 28.6 3.0 16.2 0.4 61.0 19.8 (1) A. seyal 18.5 ^ ^ ^ ^ ^ (2) Pods 20% (2) Bark 20% (2), 18^30%(5) senegalensis 8.2 ^ ^ ^ 10.5 ^ (3) Balanites aegyptica 14.25 0.82 2.91 11.72 4.16 66.14 (4) Combretum glutinosum 10.0 1.02 1.82 14.15 5.04 67.97 (4) Faibherbia albida 13.63 0.77 2.15 12.75 3.88 66.82 (4) Bark 20^28% (2) Guiera senegalensis 13.13 0.18 1.55 11.46 4.06 69.62 (4) Piliostigma reticulatum 9.38 0.63 2.89 10.12 5.05 71.93 (4) Bark 18% (2) Prosopis africana 10.2 6.8 ^ ^ 7.9 49.46 (3) Bark 18% (2) () 16.0^25.74 1.7^20.0 7.6^15.76 2.94^3.4 ^ 35.56^78.5 (5)

ÃAll values apply to dry matter weight of , unless otherwise noted. ^Denotes no available data for these values. ySource citations: (1) Sauer (1983), (2) von Maydell (1983), (3) Irvine (1961), (4) Mohamed (1995), (5) Booth & Wickens (1988). by the gira¡es, resulting in a total of 32 sample plots. sample. A positive rank indicated a selective preference Each plot was 25 m in diameter (44.3 m2) (Barbour, Burk and a negative rank a selective avoidance of the species in & Pitts, 1980). We recorded the frequency of each tree question. Finally, the total amount of time the gira¡es and species within the plots (Dagg & Foster,1976; spent eating the various plant parts of each tree species Pellew,1984). was divided by the total amount of time spent by the gir- a¡es eating that tree species.This represented the propor- tional dietary importance of each tree part to the gira¡es. Data analysis Although sparse, information has been published on We de¢ned the relative frequency of each tree or shrub the chemical composition, including protein, ash, fat, species as the total number of stems per species divided ¢bre, moisture, carbohydrates (NFE) and tannin content bythetotalnumberofstemsofallspeciesdetectedin of many species in these and similar habitats (Table1). the surveys. In addition to the raw stem counts, we also This table contains all the available information on the calculated a weighted count for each species. This was species eaten by gira¡es in Niger. intended to estimate the total value, i.e. the number and/or volume of leaves, of the sampled plants. We esti- Results mated the relative foraging e¡ort by taking the total sec- onds that all gira¡es were observed foraging at a tree Herd composition and zone occurrence species and dividing it by the total time all gira¡es were observed foraging at all tree species. By the end of the study period the herd comprised18 adult The relative vegetation frequencies and the foraging males, 26 adult females, 12 subadult males, 6 subadult data were used to establish a strength-of-preference vari- females, 4 juvenile males,7 juvenile females, 4 newborn able (preference index). The preference index equalled males,1newborn female and 3 unsexed newborns, for a the relative frequency of a plant species'occurrence sub- total of 81individuals in the Niger herd. The numbers tracted from the gira¡e's relative foraging e¡ort (fre- shifted during the study due to individuals ageing, and quency) on that species. Thus, a rank of zero on the births and deaths. During the study, 3 adult males, 1 preference index scale indicated that a tree species was juvenile male and 2 newborn males died. There were 32 eaten at the same frequency with which it occurred in the births during the same period.

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 204 Lauren E. Caister et al.

We observed signi¢cant spatial sexual segregation accompanied by a juvenile o¡spring. Only 11 % of the within the herd. In the IM, 75% of all adults observed females in the IM were pregnant with juvenile o¡spring feeding (n ˆ 36) were females. In contrast, in the DB, (n ˆ 71, Fisher's exact test, P < 0.0001). In contrast, 42% 70% of all observed feeding adults (n ˆ 47) were males of observed females in the IM were accompanied by new- (Fisher's exact test, P < 0.0001). Con¢rming this bias, born o¡spring and females with newborns were never 65% of the total bouts of female feeding (n ˆ 92) occurred observed in the DB (Fisher's exact test, P < 0.001). Preg- in the IM, while 80% of adult male feeding bouts nant females without o¡spring made up 18% of those in (n ˆ130) occurred in the DB (Fisher's exact test, the IM and 12% of those in the DB. These percentages P < 0.0001). were identical for adult females accompanied by juvenile Pregnancy and reproductive stage were signi¢cantly o¡spring and who were not pregnant. Ten percent of the associated with habitat type. Sixty-four percent of the adult females in the IM, and12% in the DB, were neither females seen in the DB (n ˆ 25) were pregnant and pregnant nor had juvenile or newborn o¡spring.

Table 2 Data from vegetation surveys Weighted True True stem True stem conducted in the dry-season habitat of Species abundance abundance abundance abundance G.c.peralta in Niger name of stems of stems in IM zone in DB zone

A. digitata 9321 A. indica 21^ 1 A. macrostycha 211^ A. nilotica 43^ 3 A. senegal ^^^^ A. senegalensis 155 155 89 66 A. seyal 1212^ 12 A. siebriana ^^^ 0 B. aegyptica 168 128 83 45 B. parkii ^^^^ B. rufescens 1311 9 2 C. aculeatum 1919 1 18 C. africana ^^^^ C. glutinosum 50 36 30 6 C. nigricans ^^^^ C. occidentalis ^^^^ C. procera 1514 8 6 C. zambisicus ^^^^ D. mespiliformis 31^ 1 F.albida 141 105 17 88 F.gnaphalocarpa ^^^^ F.platyphylla 31^ 1 G. senegalensis 641591548 43 Gardenia sp. ^ ^ ^ ^ H. thebaica 104 104 ^ 104 M. crassifolia 11^1 M. inermis 137^ 7 P.africana 16642 P. j u l i a f l o r a ^^^^ P.macrophylla 29 17 7 10 P.reticulatum 52 52 35 17 S. birrea 16826 T. avicennioides 522^ T. indica 126^ 6 Z. mauritiana 1412 1 11

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 Female tannin avoidance in giraffes 205

Table 3 Feeding preference indices for G.c. peralta in the dry-season Vegetation surveys habitat of Niger

The weighted versus unweighted (simple stem counts) Preference index ranks in vegetation estimates yielded few di¡erences in the rela- Plant species dry-season habitat for G.c. peralta tive abundance of the plant species encountered in the Acacia nilotica 0.2407 surveys (Table 2), so we used the simple stem counts for Acacia seyal 0.2072 our analyses.We recorded a total of 16 species in the IM, Combretum glutinosum 0.1329 with Guiera senegalensis being by far the dominant spe- Acacia senegal 0.0506 cies (Table 2). In contrast, more species (23) were Faidherbia albida 0.0389 recorded in the DB, but their abundance was more Acacia siebriana 0.032 Prosopis africana 0.0269 evenly distributed (Table 2). Ficus gnaphalocarpa 0.0174 Maerua crassifolia 0.0163 Feeding preferences Ficus platyphylla 0.0077 Ziziphus mauritiana 0.0071 When sex, age and zone were ignored, gira¡es showed Gardenia sp. 0.0051 the strongest preference (i.e. fed the longest relative to Prosopis juliaflora 0.0027 stem abundance) for Acacia nilotica, Acacia seyal and Cassia occidentalis 0.0018 Butyrospermum parkii 0.0013 Combretum glutinosum, and the strongest avoidance of Commiphora africana 0.0006 G. senegalensis, Annona senegalensis and the- Croton zambisicus 0.0005 baica (Table 3). A strong preference index for C. glutino- Combretum nigricans 0.0004 sum was apparent in the IM (Table 4). It was followed by Diospyros mespiliformis 0.0002 relatively lower strengths of preference for Faidherbia Azadirachta indica À0.0008 albida and Acacia senegal. When each sex's preferences Adansonia digitata À0.001 are examined, this trend disappears (Table 4). Both sexes Terminalia avicennioides À0.0015 showed a strong preference for C. glutinosum, but the Tamarindus indica À0.0046 Mitragyna inermis À0.0054 similarity ends there. The males' secondary preferences Sclerocarya birrea À0.0062 for Prosopis africana and A. niloticawere relatively strong, Buhinia rufescens À0.0084 while the females'overwhelming preference for C. gluti- Calotropis procera À0.0108 nosum was followed by weaker secondary preferences Parinari macrophylla À0.0115 (Table 4). Both subadult males (61% of their total diet) Combretum aculeatum À0.0147 and subadult females (58% of their total diet) spent an Balanites aegyptica À0.0346 inordinate proportion of time eating P. a f r i c a n a in the Piliostigma reticulatum À0.0349 Hyphaene thebaica À0.0802 IM. The relative strengths of avoidance changed little Annona senegalensis À0.119 across sex or age. Guiera senegalensis À0.4559 In the DB, the herd showed overwhelming preference indices forA. nilotica and A. seyal (Table 5). The strongest avoidance scores in this zone included H. thebaica and plants, only the £owers were used, for others multiple A. senegalensis. Both Balanites aegyptica and F. albida plant parts were eaten (Table 6). This non- forage were avoided, which contrasts with the preference base included species with both high and low preference shown for these species in the IM. Separately,both males indices. and females still showed a strong preference for A. nilo- tica and A. seyal (Table 5), with males having a slightly Other important plant species greater preference for A. seyal than females. These two species of tree provided the majority of the gira¡es'diet The most important shade tree used by gira¡es was Pari- in this region, despite their low occurrence rates in the nari macrophylla,with57%ofthe65eventsobserved.F. landscape (Table 2). The preference and avoidance albida was used for shade during 9 (14%) observations. indices did not change with age. Gira¡es used plant parts Other shade trees used included P. africana, F. platy- other than leaves from 11 di¡erent species. For some phylla, C. glutinosum, A. seyal, B. aegyptica and Diospyros

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 206 Lauren E. Caister et al.

Table 4 Feeding preference indices for G.c. Preference index Preference index Preference index peralta in the intermediate zone in Niger for G.c. peralta in for females in for males in the Species the intermediate zone the intermediate zone intermediate zone

C. glutinosum 0.4243 0.4578 0.3238 F.albida 0.0902 0.0956 0.0738 A. senegalensis 0.0886 À0.1061 À0.099 P.africana 0.07310.0473 0.1505 M. crassifolia 0.0518 0.069 0 Z. mauritiana 0.0483 0.0477 0.0503 A. senegal 0.0329 0.1101 0.0241 A. siebriana 0.0232 0 0 Gardenia sp. 0.0033 0 0.0133 C. nigricans 0.0013 0.0018 0 C. occidentalis 0.0002 0 0.001 A. indica 000 B. aegyptica 0 À0.0113 0.0288 B. rufescens 0 À0.0107 À0.0107 C. zambisicus 000 D. mespiliformis 000 F.gnaphalocarpa 000 F.platyphylla 000 H. thebaica 000 M. inermis 000 P. j u l i a f l o r a 000 T. indica 000 A. nilotica À0.0012 0.0022 0.1248 B. parkii À0.0012 0 0 C. africana À0.0012 0 0 A. digitata À0.0024 À0.0024 À0.0024 S. birrea À0.0024 À0.0024 À0.0024 T. avicennioides À0.0024 À0.0024 À0.0024 P.macrophylla À0.0083 À0.0083 À0.0083 C. procera À0.0095 À0.0095 À0.0095 C. aculeatum À0.0107 À0.0012 À0.0012 P.reticulatum À0.0408 À0.0404 À0.0417 A. seyal À0.1043 0.0171 0.0417 G. senegalensis À0.6528 À0.6527 À0.6532 mespiliformis. Of 15 observations of gira¡es `scratching' ing foraging competition with their mates and o¡spring. themselves on trees,8 (53%) used P.macrophylla. We can reject this hypothesis in the case of the Niger population, because the males actually seemed to be on Discussion the better range. The browse in the DB tends to be higher in fats, moisture and carbohydrates, while being lower in ash content than the browse in the IM (Table1). This Sexual segregation range also contains the only open water in the region, Habitat segregation.The majorityof hypotheses on sexual forcing the females to make a daily or twice-daily trek segregation in ungulates apply to species that have from the IM to the DB to drink. de¢ned rutting seasons and thus temporary segregation The second hypothesis cited by Main & Coblentz periods, or they aim to explain the departure of males to (1990) was male predator avoidance owing to the stres- poorer quality ranges. Main & Coblentz (1990) suggested ses of reproduction. A third posits that segregation may that males altruistically leave superior ranges, minimiz- minimize sexual competition and aggressive behaviour

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 Female tannin avoidance in giraffes 207

Table 5 Feeding preference indices for G.c. Preference index Preference index Preference index peralta in the Dollo Bosso in Niger for G.c. peralta for females for males Species in the Dollol Bosso in the Dollol Bosso in the Dollol Bosso

A. senegal 0.3402 0.0314 0.0319 A. siebriana 0.2857 0.0047 0.0601 B. aegyptica 0.0477 À0.0207 À0.0596 A. senegalensis 0.0318 À0.1444 À0.1444 F. platyphylla 0.0261 À0.0022 0.0141 G. senegalensis 0.0105 À0.0941 À0.0941 Gardenia sp. 0.0059 0.0115 0.0044 P. africana 0.0042 À0.0044 0.0067 P.juliaflora 0.00410 0.0052 C. procera 0.0025 À0.0131 À0.0131 B. rufescens 0.002 À0.0044 À0.0042 C. glutinosum 0.0009 0.0405 À0.0119 D. mespiliformis 0.0007 0.0043 À0.0022 A. nilotica 00.35180.3368 C. occidentalis 000.0032 T. avicennioides 000 A. digitata À0.0003 À0.0022 0.0003 C. nigricans À0.0003 0 0 F.albida À0.0007 À0.0976 À0.0595 A. indica À0.0022 À0.0022 À0.0022 M. crassifolia À0.0022 À0.0022 À0.0022 C. aculeatum À0.0042 À0.0394 À0.0394 C. zambisicus À0.0131 0 0.001 S. birrea À0.0131 À0.0131 À0.0131 T. in dica À0.0131 À0.0131 À0.0131 M. inermis À0.0153 À0.0153 À0.0153 P. macrophylla À0.0194 À0.0204 À0.0191 Z. mauritiana À0.0241 À0.0241 À0.0241 P. re t i c u l a t u m À0.0299 À0.0189 À0.0331 C. africana À0.0394 0 0.0011 B. parkii À0.0509 0 0.0026 F.gnaphalocarpa À0.0679 0.0717 0.013 A. seyal À0.14440.24340.2977 H. thebaica À0.2276 À0.2276 À0.2276 in the nonbreeding seasons.We can reject both, because Main & Coblentz (1990) cited a ¢nal hypothesis that the Niger gira¡es have no de¢ned breeding season and males may optimize forage resources and females trade male^male competition occurs year round. In addition, o¡ foraging bene¢ts for habitats that are more suitable segregation in our population is not complete. for raising young.We believe that this best explains our A fourth hypothesis (Main & Coblentz,1990) was that data. We propose that the avoidance of the DB by cows males may depart to more open ranges in order to avoid with newborns serves to protect their young in two damage to their antlers, while still maintaining domi- ways:(i)itavoidsfoodinimicaltonursingtheiryoung nance hierarchies. In contrast, in gira¡es both sexes (see dietary segregation), and (ii) it minimizes contact have horns and our males occur in more densely forested with potential predators when accompanied by more areas than the bulk of the females. Foster (1966), Ginnett vulnerable newborns. & Dennent (1997), Pellew (1984), and Young & Isbell The entire range of the G. c. peralta is free from all pre- (1991) have documented males in denser cover than dators, with the exception of humans. The IM has a low females in several herds of gira¡es in east Africa. human population, with few villages and almost zero

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 208 Lauren E. Caister et al.

Table 6 Percentage of browse intake for New different parts of the tree species when Tree species Leaves Bark shoots Seeds averaged over the entire herd of G.c. peralta A. senegal 92% ^ ^ 8% ^ in their dry-season habitat in Niger A. sieberiana 75% ^ 25% ^ A. nilotica 83% 12% 2% ^ 3% B. aegyptica 92% 8% ^ ^ ^ C. glutinosum 93% ^ 2% 5% C. occidentalis 50% 50% ^ ^ ^ F.albida 95 % 2 % ^ 3% ^ F.platyafolia 21% 79% ^ ^ ^ G. senegalensis ^ ^ 100% ^ ^ M. crassifolia ^ ^ 100% ^ ^ P.africana 70% 30% ^ ^ ^

ÃIf trees are not listed then leaves comprised100% of observed forage.

farming activity during the dry season. Thus, cows with Thus, females may be supplementing their diets with A. newborns can easily avoid human contact by staying in senegal in order to make up for this de¢ciency. the IM. The DB, on the other hand, is densely populated. If this is true, why are females selecting C. glutinosum Farming in this region occurs year round with unavoid- when in the IM? Since both Acacia species are higher in able human contact. We did not have one observation fats, protein and moisture and much lower in ash than period in the DB without seeing other people. Addition- C. glutinosum they should be preferred. The reverse pre- ally, many dunes and depressions in the landscape ference may be attributed to C. glutinosum's relative reduce the long distance visibility that gira¡es rely upon abundance, high carbohydrate content or potentially for protection from predators (Pellew, 1984; Young & low tannin content. We can eliminate the abundance Isbell,1991). hypothesis because G. senegalensis, B. aegyptica and A. senegalensis are all much more abundant in the IM. Both Dietary segregation. The preference for P. africana in the G. senegalensis and B. aegyptica are comparable to C. glu- IM can be attributed entirely to subadults. This strong tinosum in the composition of their leaves in every cate- selection by subadults for P. africana (61% of males' and gory except the fat content for G. senegalensis, which is 58% of females' observed diets) may be due to the high very low (0.18%).This fact allows us to eliminate the pos- fat and low ash content (Table1). The bark, which makes sibility that females are simply seeking browse with a up 30% of the intake from P.africana, contains moderate high carbohydrate content. Pellew (1984) suggested that levels of tannin relative to the adult's preferred, but tan- females seek a high-energy browse and avoid browse nin rich, A. nilotica (Table 5). Males showed a strong pre- witha high ¢bre content inthe dryseason. ference for A. nilotica and A. seyal in the DB, both of This leaves us with the hypothesis that the females which are high in protein (Table1). They exhibit a prefer- in the IM are avoiding tannins. The pods of A. nilotica ence for C. glutinosum in the IM, much as the females contain about 30% tannin and those of A. seyal contain do, but not as strongly. Unlike the females, males retain about 20%.While the speci¢c tannin levels in C. glutino- their preference for A. nilotica in the IM, although this sum and A. senegal are unknown, Acaciasingeneral species is exceedingly scarce there. For the females, this have higher tannin levels than (Sauer, secondary preference rank is ¢lled byA. senegal. A. sene- 1983).A.senegaldoes not have high enough tannin levels gal is no more abundant in the IM than is A. nilotica, but to be used as commercially as A. nilotica and A. seyal it is much higher in protein (28.6%) and moisture (61%), are. This may explain the selection di¡erences between both of which are in low supply in their most preferred the males and females in the IM. Although there is no species, C. glutinosum (Table1).C.glutinosumactually information on the tannin content of C. glutinosum, we has the lowest protein and moisture contents of all the can provide circumstantial evidence to support this preferred species (10% protein and 5.04% moisture). hypothesis.

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 Female tannin avoidance in giraffes 209

Females in the DB, like males, exhibit strong prefer- The third strongest avoidance di¡ers between males ences for both A. nilotica and A. seyal. However, females and females, with males avoiding G. senegalensis,and in the IM show weak preferences for both of these spe- females avoiding F.albida. However, the rank of F.albida cies, while the males still seek out A. nilotica. Young & (À0.0976) is almost tied with G. senegalensis (À0.0941) Isbell (1991) also noted this shift in female feeding and for the females. Possibly the slight di¡erence in ranks is the consistency of the males'preferences when in di¡er- caused by the large stature (often >20 m tall) of F.albida. ent vegetation zones in . They concluded that this The lowest branches of these trees were often above the was because of a di¡erence in habitat preference.We dis- feeding height of the females in the DB. In the IM it is agree that this was caused solely by habitat selection. consumed much more frequently by both sexes. This When we take into account the fact that nursing cows may be due to the lack of other Mimosaceae species in this never come into the DB, and that the majority of cows zone, or because the F. albida in the IM are generally that do come in are pregnant, this trend begins to make much smaller. Although gira¡es in neither zone display sense in terms of dietary and not just habitat preference. a preference for this species, it is actually utilized fre- Nursing cows may be avoiding tannins because they are quently, and it is the abundance of the tree that drives not palatable to their young when passed on through the preference index ranks down. This avoidance rank the milk. Pregnant cows, having high energy require- should not be confused with a total avoidance of this spe- ments, would bene¢t from the higher quality forage of cies. On the contrary, F. albida comprised over 60% of theAcacias available in the DB. A. nilotica is higher in fats the observed diet of juvenile females inthe IM. and proteins than C. glutinosum (Table1).Acaciaspecies Although similar to C. glutinosum in composition, we are also lower in ash content than are the Combretum only observed the very abundant G. senegalensis being species (Sauer,1983). eaten once, and even then, only the £owers for 13 s. The The strongest avoidance (<À0.1preference index) dry weight composition data for this species tells us little by both sexes was identical in rank for both zones. about why it might be avoided. Some of the traditional This avoidance pattern indicates that these plants uses of the plant may shed light on the matter, as multi- make poor quality forage. In the DB, H. thebaica was ple medicinal uses of the leaves of G. senegalensis are well the most strongly avoided species. We saw no instan- known in Niger. In powder form the leaves are used to ces of gira¡es browsing on H. thebaica, even though it is treat everything from severe cuts and burns to internal the most abundant species in this region. Hyphaene disorders, and to stimulate lactation in both humans spp., being , are extremely ¢brous and and livestock. Whatever chemicals are at play here may therefore most likely unpalatable to browsers. This spe- cause avoidance by the gira¡es. cies is also avoided by the domestic livestock of the Ngog Nej (1984) and Mohamed (1995) provide the only region, further supporting its' being unpalatable to prior studies on gira¡e diet in west Africa. Both list B. ungulates. aegyptica among the most preferred tree species by gir- The second most avoided species in both the DB and a¡es. Our results di¡er. Although the gira¡es consumed the IM was A. senegalensis.This is a shrubby plant, rarely this species, they demonstrate an avoidance of it in the reaching more than 1m high, which may contribute to DB and a weak preference in the IM. Even in the IM, its avoidance rank. There must be more involved in this where B. aegyptica was consumed regularly,it comprised avoidance than size alone, since gira¡es have been less then 20% of the adult's observed diets and less than reported to eat low lying bean plants. Thus, if this plant 10% for subadults and juveniles. Foster (1966) suggested was attractive enough, the size and trouble involved in that gira¡es in the Serengeti National Park pass up Bala- eating it would not be enough to deter them. As nites species for more preferred food species. Obviously, McNaughton & Georgiadis (1986) suggested, there may this is also the case in Niger. B. aegyptica is one of the be nutritional imbalances in some plants that act as a most abundant species in both regions. It is much more defence against herbivory, with no deleterious e¡ects to abundant than any of the preferred species in either the plants themselves. It is quite possible that an imbal- region. The composition of this plant is higher in protein ance of this type occurs here. A further discussion of this (14.24%) than any of the preferred species (with the avoidance would be speculation, as there are no data exception of A. senegal), however, it is low in fats and rela- available on the plant's composition. tively low in moisture (0.82% fats and 4.16% moisture).

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210 210 Lauren E. Caister et al.

One could speculate that this is a case of the natural Booth,F.E.M.&Wickens, G.E. (1988) Non-Timber Uses of Selected defences of this species at work. B. aegyptica has very Arid ZoneTrees and in Africa. FAO of the UN, Rome, Italy. large spines on the branches (up to 8 cm long). Neverthe- Ciofolo, I. (1995) West Africa's last gira¡es: the con£ict between development and conservation. J.Trop. Ecol.11, 577^588. less, after many observations of the gira¡es eating both Dagg,A.I.&Foster, J. B. (1976) The Gira¡e: Its Biology, Behavior,and the leaves and the thorns, we would have to reject this Ecology. Robert E. Krieger Publications Co, Florida, USA. possibility. Knowing that the G.c. peralta in Waza Foster, J.B. (1966) The gira¡e of Nairobi National Park: home National Park, , do indeed show a preference range, sex ratios, the herd, and food. E. Afr.Wildl. J. 4,139^148. for this species (Ngog Nej,1984), and are moderate con- Ginnett,T.F.& Dennent,M.W.(1997)Sexdi¡erencesingira¡e sumers in Niger, we can eliminate the possibility that foraging behavior at two spatial scales. Oecologia 110, 291^300. this tree has a strong chemical deterrent, as may be the Irvine,F.R.(1961)Woody Plants of . Oxford University Press, case with G. senegalensis. Thus, we must assume that gir- London, UK. Le Pendu,Y.& Ciofolo, I. (1999) Seasonal movements of gira¡es in a¡es can di¡erentiate between the seemingly slight dif- Niger. J. Trop. Ecol.15,341^253. ferences in composition between this species and C. Main,M.B.&Coblentz, B.E. (1990) Sexual segregation among glutinosum. ungulates: a critique.Wildl. Soc. Bull.18,204^210. von Maydell,H.J.(1983)Arbes et Arbustes Du Sahel. Leurs Caracteristiques et Leurs Utilisations.GTZ,Eschborn. Conclusion McCullough,D.R.,Hirth,D.H.&Newhouse, S.J. (1989) Resource Unlike the ¢ndings of Bleich, Bowyer & Wehausen (1997) partitioning between sexes in white-tailed deer. J.Wildl. Manage. 53,277^283. and McCullough, Hirth & Newhouse (1989) for other McNaughton,S.J.&Georgiadis, N. (1986) Ecology of African ungulates, the results of this study indicate that the habi- grazing and browsing mammals. Annu. Rev. Ecol. Syst.17, tat choice exhibited by female gira¡es in Niger may be 39^65. driven more by nutritional demands than by an avoid- Mohamed, M. (1995) Ecological considerations for the ance of predators or resource partitioning. The apparent management of gira¡es and its implications for tourism: a case avoidance of high levels of tannins by nursing females study of Koure-Boboye area in Niger Republic. Unpublished requires them to pass up the otherwise high quality for- Thesis, University of Ibadan, Ibadan, . Ngog Nej, J. (1984) Regime alimentaire de la girafe au Parc age preferred by the males in the DB and only return to National de Waza, Cameroun. Mammalia 48,173^184. it when they have ¢nished lactating. Pellew, R.A. (1984) The feeding ecology of a selective browser, the gira¡e (Gira¡a CamelopardalisTippelskirchi). J. Zool. Lond.202, References 57^81. Sauer, J.J.C. (1983) A comparison betweenAcacia and Combretum Barbour, M.G., Burk,J.H.&Pitts,W.D.(1980)Terrestrial Plant leaves utilized by gira¡e. S. Afr. J. Anim Sci. 13,43^44. Ecology. Benjamin/Cummings Publications, California, USA. Young,T.P.& Isbell, L.A. (1991) Sex di¡erences in gira¡e feeding Bleich,V.C., Bowyer,R.T.&Wehausen,J.D.(1997)Sexual ecology: energetic and social constraints. Ethology 87,79^89. segregation in Mountain Sheep: Resources or predation? Wildl. Monogr.134,1^50. (Manuscript accepted 23 July 2002)

# 2003 African Journal of Ecology, Afr. J. Ecol., 41,201^210