A POPULATION SURVEY OF FOREST-DWELLING RHESUS MONKEYS ( PRIMATES : CERCOPITHECIDAE ) IN NORTH INDIA

The status of rhesus monkey ( M acaca mulatta Zimmerman 1780) populations is better known than that of most other primates as a result of surveys undertaken in north India at periodic intervals since 1959 (Southwick et al., 1961a, b; Southwick and Siddiqi, 1966, 1968; Southwick et al., 1964 ; Mukherjee and Mukherjee, 1972 ). These surveys sampled rhesus monkey populations in various habitats; however, figures for forest-dwelling groups were based on survey of an area of only approximately 26 kms. We present here the results of more extensive surveys conducted during 1964 and 1965 in forested areas of the north Indian states of Uttar Pradesh, Punjab, and Himachal Pradesh. Although it is expected that the numbers of monkeys in the sampled areas have changed in the years since our study was made, our results represent an important baseline for comparison with other surveys-especially the nation wide survey India is now undertaking.


INTRODUCTION
The status of rhesus monkey ( M acaca mulatta Zimmerman 1780) populations is better known than that of most other primates as a result of surveys undertaken in north India at periodic intervals since 1959 (Southwick et al., 1961a, b; Southwick andSiddiqi, 1966, 1968;Southwick et al., 1964 ;Mukherjee and Mukherjee, 1972 ).These surveys sampled rhesus monkey populations in various habitats; however, figures for forest-dwelling groups were based on survey of an area of only approximately 26 km s .We present here the results of more extensive surveys conducted during 1964 and 1965 in forested areas of the north Indian states of Uttar Pradesh, Punjab, and Himachal Pradesh.Although it is expected that the numbers of monkeys in the sampled areas have changed in the years since our study was made, our results represent an important baseline for comparison with other surveys-especially the nation wide survey India is now undertaking.
The forests of present-day India represent a small but significant percentage of the total rhesus monkey habitat, and for many years the monkeys in these areas were relatively inaccessible to trapping for biomedical use.Forest groups have been a major source of rhesus monkeys for only the last 5 to 7 years (Southwick, personal communication).
Because there have been extensive surveys of rhesus monkeys living in rural and urban areas, the major goal of our work was to assess the abundance of those in forest areas.Of the more than 9,400 km surveyed, 1582 were along forest roads in portions of the Himalayan foothills.

Recorda of the Zoological Survey of 1 'naia
We include data on surveys conducted in non-forest regions-areas designated below as rural and urban-to draw comparisons between our findings and those reported by Southwick and his colleagues for independent surveys they carried out in Uttar Pradesh during approximately the same period of time.

METHODS
Our survey took place between October, 1964 andJune, 1965.Methods for locating groups in forest and rural habitats were basically the same as those used by Southwick et al., (1961b) in their survey of transportation routes, i. e., travel over passable roadways in a vehicle from which the authors scanned adjacent areas.Travel was at slow speeds, rarely exceeding 30 km per hour.We regularly surveyed throughout the day, including portions of the interval from 1100 to 1500 hours, regarded by Southwick and colleagues as potentially less productive in locating groups.In urban areas, we confined our attention to those groups encountered in traveling through the towns and cities along our roadside survey routes.In determining incidence of groups per km of travel, we have excluded urban groups and urban travel from the data.
As each group was sighted it was counted, and, only then, efforts were made to entice it with peanuts into open areas where counting was easier.(The one exception was in towns and cities where feeding tended to attract large numbers of curious humans.)Where observation conditions permitted, composition data were obtained using the age and sex categories defined by Southwick et al., (1961a).In addition we noted general activity of groups at time of sighting, evidence of mating, impressions of health, and such habitat features as elevation, topography, type of vegetation, water source, human habitations, and associated fauna.
Our classification of rhesus monkey habitats into forest, rural, and urban differs substantially from the categories used by Southwick et al., (1961aSouthwick et al., ( , 1961b)).The authors have commented on the arbitrary nature of certain habitat designations.For example, groups found .nearvillages which are located along roadsides could be placed in either a roadside or village category.They resolved this problem "by determining which locale represented the usual and most frequent habitation of the groups" (1961b, p. 701).We found that such determinations were often difficult to make on the basis of the limited amount of time it was possible to spend with each group.Furthermore, it is our contention that the mo~t important habitat distinctions are those reflecting the amoqnt of each group's contact with the human populace.From this standpoint, forest areas constitute a fairly homogeneous portion of the total range of habitats.Non-forest groups were divided into urban and rural, the distinguishing factor being presumed access of groups designated as urban to prominent bazaar areas.Usually, groups frequenting bazaars, whether located in cities, villages, or temples, move freely through dense concentrations of humans and are least dependent on natural foods and foraging for subsistence.Our rural category includes groups intermediate between forest and urban in amount of human contact.While frequenting roadside trees and adjacent fields certain of these groups regularly visited small mud villages, or tea shops at rural crossroads.
For purposes of data analysis, we assigned group counts to one of three categories: "complete"," incomplete", or "no count attempted."Groups which fled at contact or ,vhich were seen on the rooftops of a distant village are examples of circumstances under whieh it was impossible to obtain a count.For the remainder, case of observation was the primary criterion for judging completeness of counts.For example, groups seen in nearby open fields or groups which responded to feeding were obviously much easier to count than those which remained partially concealed in tall grass -or roadside trees.In a few instances group composition was used as an additional indication of completeness of counts, as, for example, a tally shoWing substantially more infants than adult females was considered incomplete.
Most of our survey in forest areas coincided with the beginning the 1965 birth season and, as a result, we sighted nearly one hundred newborn infants.Because other habitat types were not surveyed at this time of year, newborn infants have been excluded from the totals.

RESULTS
Groups in rural habitats were much more easily counted that those in forests or urban areas.Counts of 48.4% of the groups sighted in 15.80/0 for urban habitats (table 1).These figures reflect in a general way differences in shyness of groups and differences in habitat features which obstructed our view.That rural groups were least shy is further supported by the fact that 54 0 10 of feeding attempts in this habitat were successful, compared to only 11 % for forest groups.Despite their greater shyness, on several occassions forest groups could be counted as we came upon them resting in the open or drinking at a stream.
Validity of the distinction between complete and partial group counts was assessed by computing separate means for group size in each habitat category.The discrepancy is greatest for forest and urban habitats, where partially counted groups are smaller by an average of 10.4 and 8.8 individuals per group respectively.The distinction seems least valid for rural habitats, where the difference was only 0.6 individuals per group.

Incidence of groups:
The total number of groups sighted in all habitats and under both survey and non-survey conditions is given in table 1. Incidence of groups is determined, however, by using only those sighted under survey conditions (242 groups in rural habitats, and 134 in forests).Ou;r systematic survey covered a total of 9,510 original km (km which were not resurveyed) 16. 6% of which were in forest areas (table 2).Groups were thus located at a rate of one for each 11.6 km of forest road, compared with one for each 32•8 km of rural travel (P 0.025, Xi).A breakdown of survey routes by state reveals that there wer~ fewer rhesus groups in rural habitats in Punjab and Himachal Pradesh than in Uttar Pradesh (table 2).The incidence in forest areas for these two 'states approaches that for Uttar Pradesh, but is based on only lZ9 km of forest travel.
. ' Although sampling urban groups was incidental to forest and rural surveys, it is interesting that only 16 urban groups were seen in all •of Punjab, three in the hill city of Simla, and the remainder within a radius -X ."0 ." 0 U\ 10 V'\ 0 '" --Nrot,., ~~.lin"" Group Size (S.D. 10.96)  DOLHtNow & LINDBURG: Population studies of Rhesus monkeys 213 ted previously (Southwick, Beg and Siddiqi. 1961b).The ranges in group sizes for rural, forest, and urban habitats are shown in figures 1, 2, and 3.

Group compositio'lf,:
The number of adult males per group was fairly similar in all three habitat types despite differences in mean group size (table 3).All remaining age and sex categories averaged fewer individuals in rural than in forest or urban groups, however.The greatest difference was in number of juveniles, with forest groups averaging 7.7, while rural groups had only 1.6 juveniles per group.Urban groups had intermediate values for females and juveniles, but had the highest incidence of infants.It is possible that the relatively low number of infants in forest groups was due in part to errors in classification of immatures as discussed below.
Habitat usage: In forest areas we found a higher percentage of groups in relatively pure Shore a robusta (sal) forests than in any other type of vegetation (table 4), These findings reflect to some extent the  (Nath, 1963).In a study of forest groups near Dehra Dun, Lindburg (1971Lindburg ( , 1976) ) found that sal was utilized as food by rhesus monkeys more than any other single species, but that food preferences varied greatly on a seasonal basis.Additional confirmation of this point is found in our observations of groups concentrated in sheesham (Dalbergia sissoo) forest in Corbett Park during in the winter season, whereas Southwick's survey of the same area in the dry season indicated a preference for mixed deciduous forest (Southwick 1961b).The type of vegetation in which a group is seen in a brief encounter may not be preferred by that group on a year round basis.
For rural groups located along roadsides, although several species were often present at each sighting, the mango tree (Mangifera indica) occurred in 69% of the locations for which we collected this kind of Records of the Zoological Survey of 1 'fIilia information (N = 20 4 ).The prevalence of other tree species at group locations (table 5) probably reflects incidence of tree types along roadsides more than preferences by the monkeys.In comparing results of our surveys with those of Southwick and his colleagues, we have used data from their 1964-1965 survey of rural and urban habitats (Southwick andSiddiqi, 1966, 1968).The only data for forest groups available for comparison are from their 1961 report.
Comparisons are based on their forest, roadside, and town habitats, which we believe are the nearest equivalents to our three habitat categories.As a matter of convenience we shall hereafter refer to these earlier reports as the Southwick surveys.
Incidence of groups: Our results indicate an incidence of one group for each 32.8 km (20.3 mi) of rural survey, compared to one each 25;1 km (15.6) in Southwick's report.However, it should be noted that our effort included nearly 2,500 km of travel in Punjab and Himachal Pradesh, for which the incidence of groups per km was very low (table 2).When we limit the comparison to Uttar Pradesh, the incidence from our survey is one group each 25.3 km (15.7 mi) of travel, indicating remarkable close agreement with Southwick's findings.
Our survey of forest roads yielded an incidence of one group each 11.6 km of travel, which is more than twice as frequent as in rural areas of Uttar Pradesh.Southwick, and Siddiqi (1966) have questioned thereliability of roadside survey in detecting forest groups, noting that they are not attracted to roadsides for food or sleeping trees as are groups living in rural areas.While we are in essential agreement with this observation, we believe forest roads may be attractive as open areas for sunning and grooming at certain times of day.Sightings in other instances were undoubtedly fortuitous.In any case, the results suggest that an incidence of 0.7 groups per square mile of forest, as reported by Southwick, et ale (1961b), may be a conservative figure .Group size and composition: The number of groups in each habitat category for which comparisons are made are indicated in table 6.For all age/sex categories in groups from rural areas, Southwick's figures are slightly lower than ours, resulting in a difference in mean size of 3.7 individuals per group.There is very close agreement on the incidence ,of juveniles per group.The overall difference in results for the two surveys could easily be a consequence of somewhat different classification -of habitats.Our larger mean size is most likely a result of including in our data certain village groups which Southwick treated separately, and for which he reported a mean size of 24.4.
For urban habitats, although our sample is samaller,-there is good agreement in results of the two surveys.The greatest difference is in approximately three more juveniles per group in Southwick's survey.Both surveys indicate a higher incidence of juveniles than was found in rural areas, but lower than that for forest groups.
The greatest discrepancy in results is for forest groups.We found the mean size of forest groups to be larger than in either rural or urban habitats, but only 28.2 individuals per group, compared to a mean of 49.8 reported by Southwick.This is the only habitat for which the difference in results from the two studies is statistically significant (p 0.025, Chi square).Average numbers of individuals in all age/sex categories were higher in Southwick' s sUIV~y, with the greatest difference being in numbers of adult females.These differences are not surprising, given the difference in sample size.(See table 6.) Population composition: Although we found forest groups to be significantly smaller than Southwick has reported, there is remarkably close agreement between surveys in percent of individuals in the different age/sex categories (table 7).The composition of the rural population shows good agreement as well.In both surveys, percentages of adult males and infants in the three habitats were lowest for forest groups, whereas the forest population had the highest percentage of juveniles.We found the percentage of adult females to be nearly identical in all three habitats.
With respect to juveniles, the differences are undoubtedly a consequence of the amount of trapping for export which occurred in each habitat.The reasons for the low number of infants relative to number of sexually mature females in forests are not obvious.Given the generally better health and less frequent human contact for forest JIlonkeys, we would expect conditions in thi~ habitat to be the most It is also possible that, with a larger juvenile population in forests, classification of a number of larger juveniles as sexually mature females was a second source of error.
Abundance oj forest population: In their 1960 survey of forest habitats in Uttar Pradesh, Southwick et al. {1961b) estimated an incidence of 0.7 groups in each square mile of forest.This figure was based on their success in detecting groups during foot travel through blocks of forest one square mile in size.By pausing frequently to listen for vocalizations or sounds of movement, they estimated that in the dry season animals the size of rhesus monkeys "could be heard for a distance of approximately 200 yards" (p.700).For vehicular survey over forest roads we estimated a maximum detection range of 50 metres on either side.On this basis, conversion of linear distance to area equivalents (Southwick and Cadigan, 1972) yields an incidence of 0.91 groups/km2.Following the formula of Southwick et aZ.(1961b), and using the revised mean for group size from our surveys, we obtain an estimate of 200,000 rhesus monkeys in the forests of Uttar Pradesh for 1964-65.This figure compares with 100,000 estimated by Southwick et al. (1961b) for 1959-60.It is unlikely that forest population were experiencing a rate of growth which would account for an increase of this magnitude in only five years, particularly in view of the documented decline of the population in other habitats (Southwick andSiddiqi, 1966, 1968).It is more likely that abundance in 1959-60 was underestimated.
If incidence and size of groups in Himachal Pradesh, with 12,517 km s of forest (Government of India 1968), are the same as in Uttar Pradesh, as our limited sample indicates, then the total forest population for that state would fall between 60,000 and 70,000 monkeys.This figure is probably too high, however, since 66 0 / 0 of the forest in Himachal Pradesh is coniferous, compared to 27% for Uttar Pradesh (Government of India, 1968), and a very few groups in our total forest sample were found in coniferous forests (Table 4).
We have no census data for the approximately 18,000 km 2 of forest in Punjab, but were repeatedly informed by people in the country-side that rhesus monkeys survived in this state only in the foothill forests.Given their geographical proximity to broadly similar forests in Himachal Pradesh and the foothills of Uttar Pradesh, and a percentage of ~8 Records of the ZoologicaZ Survey 01 India non-coniferous vegetation identical to that of of the latter state, it is possible that another 60,000 rhesus monkeys resided in the forests of Punjab.
Size (S.D. 12.71) Text-fig.3. Number of groups and group sizes in urban habitats.

TABLE 1 .
Rhesus monkey groups encountered in various habitats under both survey a.nd non-survey conditions.
210Record8 of the Zoological Survey of I ruJAa rural habitats were judged complete, compared to 22. 6 0 / 0 for forest, and

TABLE 2 .
Incidence of gr.oups of rhesus monkeys in forest and rural surveys in Uttar Pradesh, Punjab, and Himachal Pradesh.

•
Text-fig.!.Group size of rhesus monkey groups living in rural habitats in India.In pair-wise comparisons differences in mean size of groups for our three habitats (table 3) were insignificant (Chi square test).
of 100 km from New Delhi.Apart from the groups in this region of Punjab, and an occasional group in the Himalayan foothills, it would appear that there are few if any rhesus monkeys in most areas of the state of Punjab.Records of the ZoologicaZSurvey oj 1 ""di,a, Group size: Z Group Size (S.D. 17.06) Text-fig.2. Number of groups and group size in forest habitats.

TABLE 3 .
Average group size and composition for rhesus monkeys in forest, ruml and urban habitats.
While forest-dwelling groups were larger, on the average, than groups in other habitats, they were smaller by nearly 50% than has been repor-, ..

TABLE
Per cent of group sightings in different types of forest vegetation.
prevalence of different types of vegetation in northern India.In the Dehra Dun Forest Division of Uttar Pradesh, for example, 79% of the forest is classified as sal

TABLE 5 .
Prevalence of different species of trees in roadside habitats of rhesus monkeys.

TABLE 6 .
Average group compositions compared with compositions for equivalent habitats in Southwick's survey.Column A=this survey; column B=Southwick survey.

TABLE 7 .
Population composition compared with c~mposition in equivalent habitats from Southwick's survey.Column A = this survey; column B=Southwick survey.It seems likely, therefore, that the low incidence of infants is a result of errors in classification.Forest surveys in both studies were conducted primarily in the dry season when infants of the previous birth season were approaching yearling size, possibly resulting in a number of them being erroneously counted as juveniles.