8/6/2019 Population Count Direct&IndirectPaper

1/9

l~ : A week with elephants. Eds.,.H~Datye, Bombay Natural Historyand Oxford University Press, New339. (1995)

J.e. Daniel andSociety, BombayDelhi. Pp.331-

DIRECT AND INDIRECT METHODS OF COUNTINGELEPHANTS: A COMPARISON OF RESULTS FROM

MUDUMALAISANCTUARY

K . SURENDRA VARMAN UMA RAMAKRISIINAN AND R. SIlK I JM AR I(With (j text-figurer

A study comparing dcu- ..ty estimates of elephants from direct ami indirect(duug) counts was carried .vut ill Mudum ulai W ildlife S anctuary during 1~1J1-1992. Line transect mcthc,' as outlined by Burnham 1 : / al. (1980) for directcounts, and the dropping count method developed by Barnes and Jensen (1IJR7)for indirect count were followed. The sanctunry was divided into several ZOIl\':Sbased on location and habitat type. l-our permanent transects of 3-4 kill eachwere cut Illl" the direct count and walked twice each m onth . For the indirectcount, transects of 1-4 km each were cut and enumerated each of the three (dry,I wet, II wet) seasons. Experiments were set up each season to estimate dungdecay rates, while defecation rates were obtained from Watve (1992). Resultsobtained by the two methods arc compared. The estimate ofmean density [rolllthe direct count was higher (3.09 clcphants/sq kill, 95% CI = 1.40-4.711) thanthat obtained by the indirect COUlIt(1.S1 clcphants/sq kill, 95% CI = 1.01-2.10).The strengths and weaknesses of both methods arc discussed.

iNTRODUCTION

8/6/2019 Population Count Direct&IndirectPaper

2/9

3 3 2 A WEEK WITH ELEPHANTS

STIJDY AREA

The study was conducted in Mudumalai Wildlife Sanctuary,Tamilnadu, south India (11 32' to 1143' N, 76"22' to 7645' E) (Fig, 1).Three seasons can be defmed: dry (Ja nu ary -Ap ril), I wet (May-August)and n wet (September-December). A distinct rainfall gradient extendsfrom west to east, with a corresponding change in vegetation type frommoist deciduous forest through dry deciduous forest to dry thorn forest(Sukumar et al. 1992). The forests of MudumaIai Wildlife Sanctuary (321sq km), Bandipur National Park (874 sq Ian), Wynad Sanctuary (251 sqkm), Sigur plateau and the northern and eastern slopes of the Nilgiris (700sq km) fonn a contiguous elephant habitat which supports one of thelargest elephant populations in Asia.

METHODS

The basic line transect method as outlined by Burnham et al. (1980)was used. The study area was divided into 7 zones based on topographyand vegetation type, For the line transect directcount, four permanenttransects were cut (Fig. 1), which ran through different habitat types. Thetotal length of the transects in each of the habitat types was approximatelyproportional to the area of that habitat type. Each transect was walkedtwice a month, once between 7.00 hr. and 9.00 hr and egain between

8/6/2019 Population Count Direct&IndirectPaper

3/9

MEI7l0DS OF COUNTING ELEPHANTS

o -0::: . . ,-tZ:> -c : : EC I C::J !II:t- ee . . .u IIIZ ~eI/) CD :1 :UJIL. * o f: : : Ja...J Ni: 3 0' "%::J.C::J%

333

8/6/2019 Population Count Direct&IndirectPaper

4/9

314 A WEEK Willi ELEPHANTS

transect in a particular habitat type was roughly proportional to it s areaThe count was largely carried out towards the end of each season, so asto minimize the possibility of enumerating dung piles from the previousseason. On sighting a dung pile, the perpendicular distance of th e dungpile to the transect was measured. The stage of the dung pile was noted,using the categories of decomposition foUowed by Barnes and Jensen(1987). The density of dung piles was estimated using TRANSECT(While 1987) by Fourier analysis.

Defecation rates of the wild elephant population were obtained fromWatve (1992), who used focal animal sampling. His observations totalled88 hours of sampling during the day.For estimating dung decay rate, at the beginning of each season,about 50 fresh dung piles were marked. Care w as taken to see that allhabitat types were covered. The stage of each dung pile w as noted everyweek. These were followed until the dung pile decomposed to stage E(dung pile no longer visible). Thus decay rate was calculated separatelyfor each season and for each habitat type.The elephant density was calculated using the formula:

E = (Y x r) I DwhereE = elephant density/sq km ; Y = density of dung piles (/sq km); r = dungdecay rate per day; D = Defecation rate per day.

8/6/2019 Population Count Direct&IndirectPaper

5/9

MEfflODS OF COUNnNG ELEPHANTS 335

TABLEIESTIMA TE OF ELEPHANT DENSITY BY DIRECT COUNT

Pe:nod ID I A l distance No. o f Mean group Group Elephantcovered (km) sighnngs size (SE) density/km" density/km!

(SE) (95 % CI)Sep'9\ to Aug'92 3 5 2 28 4.9 (0.97) 0.63 (0.12) 3 .\ ( 1 .4 " " "4 .8 )Fcb'91 to Dec'92 675 68 5.6 (0.57) 0.71 (0.08) 4.0 (2.75.2)

When data for two years (1991 and 1992) are combined, the meandensity works out to 4.0 elephants/sq km (SE=0.64, 95% CI = 2.75 to5.27).For the indirect count, the dung density estimate for the whole yearwas found to be 2561 piles/sq km (SE = 113.9). The Defecation rate atwaterholes/saltlicks was found to be 1.15thour, while i t was only O.66/hrin the forest (Watve 1992).

Assuming that elephants spend about one hour per day on averageat waterholes, the Defecation rate was taken to be 16.33/day (we do nothave a precise estimate of standard error; this has been assumed to be 0.8for the purpose of our calculations based on Coe 1972).

The overall rate of dung decay per day in all seasons and all habitatscombined was found to be 0.0097 (SE=0.002) (Table 2). The elephantdensity worked out to l.54 elepbants/sq km (SE=0.27, 95% Cl=l.Ol-

8/6/2019 Population Count Direct&IndirectPaper

6/9

336 A WEEK WIlli kLEPHANTS

Effort (in terms of time spent) is summarized in table 4 .. The totaltime spent on the direct count was 182 hours (352 km of transect covered)for one year (1991). For the indirect count., the total time spent was 376hOUTS (129 hours to cover 64 km of transect lines, 159 hours formonitoring dung decay and 88 hours for direct observations on wildelephant Defecation).

MethodTABLE 4

EFFORT IN TERMS OF TIME SPENT FOR DIRECT AND INDIRECT COUNTEffort (hrs.) Goettitlent of vanauonD IRECT C oON ' (Scp'91 10 Aug'92)

DIRECT COUtJT (Fcb'91 to Dcc'92)INDIRECT COUNT (Scp'91 to Aug'92)a, Decay rate experimentsb. Observations on Defecation ratec. Estimating dung densityTotal (for indirect count)

182.0316.0

28 %16 ~.

158.987.9129.2376.0

DISCtlSSION

The estimate of mean elephant density obtained by the direct countmethod is almost double that obtained by the indirect count. However, thetwo means arc not statistically significantly different for the same period(z=1.73, p > 0.05). Even then the two-fold difference in density estimatecalls for an explanation if such a difference were to be consistently true

8/6/2019 Population Count Direct&IndirectPaper

7/9

METHODS OF COUNTING ELEPHANTS 337

gives an estimate of 1,9 elephant/sq km for 1991 and 2,8 elephantlsq kmfor 1991-92 combined. '

Another factor that might bias the results towards an overestimatemight be the permanent transect lines used in the direct count. There is apossibility that elephants might be attracted towards a cleared line andviolate the assumptions of a random distribution. The frequencydistribution of distance classes of sightings showed that a very largeproportion (57% in 1991 and 55% for 1991-92 combined) of sightingswere in the 0 to 20 m class interval, indicating that such a bias might beoccurnng.

The indirect count too faces several problems, The estimate ofelephant density is based on three (rather than two as in direct count)variables and 1SthUS subject to greater overall uncertainty, The estimateof mean dung density, however, is a much more precise estimate than isthe estimate of mean group density because of the much arger samplesizes (of dung) recorded. Sin . me transect lines are not permanent thereis also no chance of any bias through elephants being attracted towardsth e line. There is some uncertainty however in determining whatconstitutes one defecation (Jachmann and BeU 1984),

The highest uncertainty seems to be in determining the mean dungdecay rate, Decay is influenced by several factors including temperature,rainfall, humidity, fire (which can bum dry piles), animal action (dung

8/6/2019 Population Count Direct&IndirectPaper

8/9

JJ8 A WEEK WITH ELEPHANTS

Defecation rates may also vary seasonally due to shift in diet (Sukumar(989) and this calls for more study.

A comparison of actual time spent on the two methods shows that,in our study, effort spent on the direct count was less than half that in theindirect count during a year. The results obtained by the indirect countwere found to be more precise with a co-efficient of variation of 22% asagainst 28% in the direct count. However , with an increase in effort forthe direct count to a level roughly equal to that for the indirect count (byusing two years' data on direct count), the coefficient of variationdecreases to 16% for the direct count.

Thus, our results indicate that in MudumaJai,a high-density elephantarea, the line transect direct count method would give a more preciseestimate of elephant density than would the indirect method for equaleffort. if all the baseline variables for the indirect count have to beestablished afresh. However, once the Defecation rates and dung decayrates have been firmly established (and thus effort is not wasted on these)the indirect count would only involve estimating duog density. In suchcase the indirect count would give a more precise estimate per unit effortthan would the direct count. The indirect method might thus be moreappropriate for long-term monitoring of a population in a given area

There still remains the problem of which method is more accurate (asopposed to more precise), i.e. gives an estimate of elephant density that

8/6/2019 Population Count Direct&IndirectPaper

9/9

METHODS OF COUNTING ELEPHANTS 339

DAWSON, S. &. A.1.F.M. DHKKHR(1991): Methods for counting Asian elephants in forests:a field mchniqucB manual.FAO. B&ngk:ok.

JACHMANN, H. &. R. H. V. Bm.L( 1984): The IISC of elephant droppings in assessing num bers,occupation and age structure : a refinement of the method. A I r , 1. Ecol . 22: 127-141.

KARANTH , K.U. " M.E. SUNQU lST (1992 ) : Population structure, density and biomass oflarge herbivores in the tropical forest of Nagaraole, India. Journal of Tropical Ecology8: 2135.

RAMAKRISHNAN . U., J.A. SANTOSH&' R. SUXUMAR (Eds) 1991: Censusing elephants inforests: Proceedings of an international workshop. Asian Elephant Conservation Centre.Indian Instinne of Science, Bangalore.

SALB,1.B., A.I.T. JOHNSrNGH &. S. DAWSON (1990): Preliminary trials with an indirectmethod of estimating Asian elephant numbers, A report prepared for the IUCN/SSCAsian Elephant Specialist Group.

SUK1.JMAR,R. (1989): The Asianolephant: ecology and management. Cambridge UniversityPress, Cambridge. U.K.

SUXUMAR. R . H,S. DATTARAJA, H.S. SURBSH. r. RADHAKRISHAN. R.VASUDEVA. S.NIRMALA &, N.V. 105m (1987): Long-term monitoring of vegetation in tropicaldeciduous forest in M'udumaJ.ai, southern India. CurrnU Science 62: 608616.

WATVB, M. (1992): Ecology of host-parasite interactions in a wild mammalian hostcommunity in Mudumalai . southern India. Ph.D. thesis. Indian Institute of Science,Bangalore.

WmTH , O.C. (1987): Program TRANSECT Line transect Data Analysis Program.Colorado State University, Fort Collins.

WINO L.D." 1.0, Buss (1970): Elephants and forests. Wildlife Monographs 19; 1-92 pp.