ROOST SELECTION BY INDIAN PEAFOWL (PAVO CRISTATUS) IN GIR FOREST, INDIA 1 Pranav Trivedi2 and A.J.T. Johnsingh3 Key Words: roost, riverine forest, preference index, selection, structure, floristics, predation A study was carried out on roost selection of Indian peafowl (Pavo cristatus) in Gir forest, Gujarat. The results revealed that all the roosts were located in the narrow riverine forest bells. Peafowl selected tall trees growing on steep river banks with thorny undergrowth and climber thickets in the canopy for roosting. Thisclearly indicatesthatroostselectionischiefly influencedbytherisk ofpredation fromnocturnal mammalian predators such as leopard. Trees ofPongamiapinnato and Holoptelia integrifolia showed more than expected use. However, it could not be confirmed whether a choice at species level does exist at all. Roost selection appeared to be a hierarchial process with structure at first and floristics at second level affecting the choice. Introduction referred to as Gir Protected Area (PA)] located in Gujarat, India. Gir PA (1412 sq. km) is the only Roost selection is a vital component of the remaining large, contiguous, forested tract in the overall habitat selection process. Therefore Saurashtra peninsula of Gujarat. The PA is covered information on roost selection by a species carries with tropical dry deciduous forests, thorn forests and immense importance for assessing its conservation riparian forests. The chief floral elements include needs. Gadgil and Ali (1975) attempted to explain Tectona grandis, Diospyros melanoxylon, Wrightia the communal roosting habits of Indian birds based tinctoria, Zizyphus mauritiana, Ficus bengalensis, on the existing hypotheses which include reduced Morinda tinctoria, Phyllant/uis emblica, Bauhinia heat loss, information sharing, assessment of racemosa, Holoptelia integrifolia, Boswellict serrata population and reduced risk of predation. Though, and Lannea coroniandelica. Indian peafowl (Pavo cristatus), a common bird of The PA is the last stronghold of the Asiatic India is known to roost in the trees at night, no lion (Panthera leo persica) and apart from lion the information exists on roost selection by the bird. In vertebrate fauna includes leopard (Pantherapardus), a strict sense, it is neither a communal nor a solitary jackal (Canis aureus), jungle cat (Felis chans) and rooster (Trivedi 1993). crested hawk eagle (Spizcietus cirrhatus) as potential This paper attempts to provide information on predators of peafowl. Checklist of mammals is roost selection by Indian peafowl in a wild landscape. available in Spillett (1968). Nearly 250 species of The following results were obtained during a study birds have been recorded. carried out from November 1992 to April 1993 The ‘Maldharis’ who are local pastoralists and (Trivedi 1993) on habitat selection by peafowl in have changed their nomadic lifestyle to a settled one, Gir forest. reside inside the Sanctuary in their settlements called ‘ness’. However, N.P. is free from all human Study Area activities. Buffalo grazing, tourism, grass harvesting, fireline burning and non-wood forest produce The study was carried out in Gir National Park collection are the chief human influences. (N.P.) and Sanctuary [(both collectively hereafter Methods 'Accepted January 1994. 2WWF-India, Ahmedabad Division Office, ‘Sundarvan’,Jodhpur Tekra, Ahmedabad 380 015. Eight different localities in three study sites ’’Wildlife Institute of India, P.O. Box 18, Dehradun-248 001. (Sasan, Chhodavdi, Dodhi) were surveyed for roost 26 JOURNAL BOMBAY NATURAL HIST. SOCIETY, Vol. 93 (1996) Table I PREFERENCE INDICES FOR SOME PHYSICAL FEATURES OF ROOST TREES USED BY PEAFOWL Slope category Distance to water (in m) Height (in in) Height of first branch (in m) Class FI Class PI Class PI Class PI Very steep 1.45 0-25 LOO 0-10 0.08 0-2 0.61 Steep 0.81 26-50 0.87 11-15 3.38 2.1-4 1.73 Gradual 0.90 5 -75 1.08 16-20 6.25 4.1-6 2.57 1 Flat 0.60 76-100 0.73 21-25 8.10 6.1-8 2.69 — - >100 0.73 >25 8.09 8.1-10 3.44 PI= Preference Index. tree use by peafowl. Both, direct and indirect areas. Roosts were found to be either continuously methods were used to locate and identify the roost spread along the riverine areas or located at the trees. The former involved walking along the confluence oftwo streams which is normally a steep riparian areas during late evening or early morning, area. to flush the roosting birds and locate the trees. The Structure: There were differences in the use latter involved searching for droppings below oftrees with and without certain structural features. potential roost trees to identify actual roost trees. There was a significant difference between the use When a roost tree was located, GBH, height, of the of trees with and without thickets ofclimbers in the first branch, tree height, slope category of the site canopy (x2= 10.62, df=l, p<0.01). Similarly, there where the tree was standing (rated qualitatively as was also a significant difference between the use of very steep, steep, gradual and flat), distance from trees with and without thorny undergrowth (x2= water (or water body), canopy and understorey 24.61 , df=1 , p< 0.001 ). In both the cases, trees with characteristics were recorded. The same data were thickets of climbers in the canopy and with thorny collected on the ten nearest trees from the roost tree undergrowth were used more often for roosting. to get availability information. In this manner, use Preference indices (PI) for various structural features and availability of the trees was determined. A are presented in Table 1. Trees on very steep and widely used method given by Neu etcil. ( 1974) was steep river banks received a higher usage followed employed for analysing the availability-use by gentler slopes. Trees growing on flat areas were information. Preference index (PI) which is least used for roosting. All tree height categories expressed as a ratio of per cent usage to per cent above 15 metres were highly used, while category availability was calculated for the structural <10 metres was used least. Nearly ninety percent of parameters of roost trees. the trees were within 75 metres from water. Trees m with 8-10 high first branch were used more and m Results the use went in a decreasing order towards 0-2 height. 034 trees were quantified as described earlier. Floristics: Twenty one plant species were 1 Of these 128 were roost trees, which reflected the identified as roosts (Appendix of which twenty 1 ) use, and the remaining gave an idea about the were trees and one was Dendrocalciums strictus (i.e. availability. All the roosts were located in riparian bamboo). Table 2 shows the availability and use of ROOSTSELECTIONBYINDIANPEAFOWL 27 Table 2 and had a higher first branch. All these features ROOST TREE PREFERENCE OF PEAFOWL indicate that while selecting a roost tree, the most [Using Neu et al. (1974) technique] important aspect is ofreducing the risk ofpredation. Tree Relative Expected Observed Confidence In Gallinaceous birds, predation is amajorpopulation species availability use use intervals regulatory mechanism (Lack 1954, Hill and Holoptelia 0.051 6.554 18 0.058-0.223** Robertson 1988) and therefore it is likely to influence integrifolia habitat selection significantly. Selection oftrees with the above mentioned features is obviously an Tectona 0.124 15.846 14 0.035-0.184 antipredatory strategy against nocturnal mammalian grandis predators such as leopard and jungle cat which can Pongamia 0.198 25.370 41 0.209-0.431** climb trees and capture peafowl. In Gir, trees with pinnata such features are available only in riverine areas and therefore these forests become crucial for peafowl. Syzygium 0.131 16.717 13 0.030-0.173 The location of roosts al the confluence of two rubicunda streams was due to the fact that this region is steep Tamarindus 0.181 23.168 6 0.000-0.097* and therefore predators would find it difficult to •• indica approach from below. Diospyros 0.033 4.224 2 0.000-0.045 It is pertinent to point out that the height of melanoxylon first branch does not seem to be of significance in roost selection. Similarly, distance from watercarries Others 0.282 36.096 34 0.161-0.371 secondary importance as all the roosts are located in the riverine area and one hardly finds a roost >100 Indicates that the species was used less than availability. m ** Indicates that the species was used more than availability. from water. However, trees growing right along Rest were used in proportion to availability. the bank with overhanging branches above the river (Z=2.6899, X2= 45.36) provide ideal roosts as birds are safe from the major roost trees by peafowl.Holoptelia integrifolia, predators due to water. The vital features, therefore Tectona grandis, Pongamia pinnata, Syzygium appear to be height ofthe tree, steepness ofthe bank/ rubicunda, Tamarindus indica and Diospyros slope on which the tree is situated and the presence melanoxylon were the commonest tree species of thickets in the undergrowth and in the canopy. It available and used as roost by peafowl. Rest of the was realized that height alone can be sufficient for species were in meagre numbers and therefore these selection ifthe tree is > 16 m. But, ifit is shorter than were clumped and collectively called ‘others' for that, the other tree features play a crucial role. In a analysis. Availability-use analysis ofthese six species semi-arid and deciduous forest system such as Gir, and others showed (Table 2) that only H. integrifolia trees hardly attain a height ofover 15 m and therefore and P. pinnata were used more than expected; T. it is the presence of thickets and steepness of the grandis, S. rubicunda, D. Melanoxylon and others slope that should be of significance in the selection were used in proportion to availability whereas T. process. indica was used less than its availability. Peafowl in semi-urban and rural landscapes often use unusual substrates as roosts, like electric Discussion pylons. Palmyra trees Borassus flabellifer) are ( commonly used in the Southern districts of Tamil The five most striking features of the roost Nadu (pers. obs.). This flexibility probably reflects trees selected by peafowl were; they had climber a synergistic effect of the absence of predation thickets in the canopy, possessed thorny pressure and a low availability of good quality undergrowth, were on steep river banks, were tall roosts. 28 JOURNAL. BOMBAYNATURAL HIST. SOCIETY. Vol. 93 (1996) Interpretation ofthe data suggests that it is the the magnitude of such potential competition might strueture whieh is the unit of seleetion at a broader not be significant. Only one roost tree of chicks seale, but at a finer scale, the selection can be for was identified which was short (c. 10 m) and had species. Any tree which satisfies the structural extensive thorny thickets wrapped around the requirements for avoiding predators should be stem. With the exception of four trees (out of 128), selected by the birds. Structure undoubtedly appears no roost tree was located close to the road to be the first step in roost selection process. It is presumably to avoid the disturbance caused by the possible that only certain tree species possess the vehicles. necessary structural features of an ideal roost tree Peafowl alone with common langur are which means the choice can be at the level ofspecies. important buffer prey which facilitate the niche The situation seems to be one of a hierarchial separation of leopard and lion in the Gir PA (Ravi selection as described by Svardson (1949), Hilden Chellam 1993). Both the prey species need to be (1965) and Wiens (1985). However, this is just a conserved. Both need roost trees in the riparian areas. logical speculation and no experimental evidence is In Gir, there is hardly any disturbance to the trees in available to test it. Peafowl (Genus Pavo) are riverine areas, but incidences of repeated fire can regarded as the terminal lineages of peacock destroy the undergrowth thickets substantially pheasants (Geist 1977). Pavo left their original rain thereby reducing the availability of good quality forest habitat and started exploiting the productive roosts. At present it is safe to conclude that the forest-water ecotone (Geist 1977). They gradually population of peafowl in Gir does not face any advanced to human dominated landscape also, but imminent danger. Our data on roost selection can be were always lied to riverine habitats. Roosting on used to predict and confirm the use ofroosts in other riverine trees might have evolved at the time oftheir such deciduous forest ecosystems in a wild dispersal from climax forests to more xeric landscape. environments, because in these habitats only riverine forests can provide good quality roosts. Acknowledgements One more important feature which influences roost selection is the occupancy of trees by other We extend our thanks to the Forest species. It was observed on ten occasions that Department, Gujarat for granting us the permission peafowl did not use particulartrees (even when these to work in Gir and for providing the necessary were ideal for roosting) because common langurs infrastructure. Qamar Qureshi and Ravi Chellam of Presbytis entellus were roosting there. This brings Wildlife Institute ofIndia (W.I.I.) commented upon ( ) in the question of competition between taxa for a the earlier draft of this paper and gave valuable crucial resource, as langurs also roost to avoid suggestions. Diwakar Sharma of W.I.I. provided We predation by leopard. Langurs too, like peafowl arc encouragement and support. thank our field distributed along the riverine areas in Gir (Joslin assistants at Gir without whose help it would not 1973) and they too roost in riverine forest. However, have been possible to obtain this data. REFE ENCES Gadgil. M. & S. Ali (1975): Communal roostinghabitsofIndian Management and Conservation. BSP Professional books, birds. J. Bombay not. Hist. Sac. 72(3): 716-727. Oxford. 281 pp. Geist, V. ( 1977): A comparison ofsocial adaptation in relation to Joslin, P. (1973): Behaviour and ecology of the Asiatic Lion ecology in gallinaceous bird and ungulate societies. Ann. Rev. (Panthera leopersica). Ph.D. Thesis, University ofEdinburgh. Ecol. Systematics 8: 193-207. Lack, D. (1954): The natural regulation of animal numbers. Hilden, O. ( 1965): Habitat selection in birds: A review. Ann. Zool. Clarendon Press, Oxford. 343 pp. Fenn. 2: 53-75. Neu, C.W.. C.R. Byers & J.M. Peek (1974): A technique for Hill. D.A. & P.A. Robertson (1988): The pheasant: Ecology, analysis ofutilization-availability data. WilclI. Manage. 38: ROOSTSELECTIONBYINDIANPEAFOWL 29 541-545. OikosL 157-174. Ravi Chellam (1993): Ecology ofthe AsiaticLion (Panthera leo Trivedi, P. (1993): Habitat selection by Indian peafowl (Pavo persica). Ph.D. thesis. Saurashtra University, Rajkot. cristatus Linn.) in Gir forest, India. M.Sc. Dissertation, Spilu-tt, J.J. (1968): A report on wildlife surveys in South and Saurashtra University, Rajkot. 78 pp. West India, November-Deceinber 1966.7. Bombaynot. Hist. Wiens, J.A. (1985): Habitat selection in variable environments: Soc. 65: 1-46. shrub steppe birds. In Cody, M.L. ed. Habitat selection in Svardson, G. (1949): Competition and habitat selection in birds. birds. Academic press, Inc., New York. 227-252 pp. APPENDIX 1 LIST OF SPECIES USED FOR ROOSTING BY PEAFOWL Ficus glomerata 1 1. Holoptelia integrifolia 1 . 12. F. bengalensis 2. Tectomi grandis 13. Miliusa tomentosa 3. Pongarnia pinnata 14. Mitragyna parviflora 4. Syzygium rubicunda 15. Garuga pinnata 5. TamarIndus indica 16. Sterculia urens 6. Diospyms melanoxylon 17. Acacia Senegal 7. Tenninalia bellerica 18. Anogeissus latifolia 8. T. tomentosa 19. Phoenix sylvestris 9. Manilkara lie.xandra 20. Dendrocalamus striclus 10. Syzygium atmini 2 Sapindus emarginatus 1.