Population Dynamics And Elephant Movements Within The .

before 2007, minus those identified in 2007, plus the new identikits added at the end of2007. As the number of animals at risk of being identified in 2008 can not be calculatedfor reasons outlined above, survival rates for 2007 and 2008 will depend on gatheringdata in subsequent years (Table 2).Table 2. The annual estimated population size (estimate SE) of young adult (17-35years) and senior adult ( 35 years) elephant bulls as estimated by the Jolly stochasticcapture-recapture model (Jolly 1965) and the bull count recorded during the annual aerialcensus.YearModel population estimateAerialcensusYoung adultbullsYoung adultSurvival %Senior adultbullsSenior bullsSurvival %TotalbullsTotalBulls2003-86-94-1762004235 ( 53)64128 ( 11)843621742005199 ( 24)85136 ( 11)753351472006281 ( 36)66132 ( 10)574131912007201 ( 19)-80 ( 4)-2811882008-----109The aerial census is conducted in an accepted, standardized manner at a similar timeevery year (Dr. Ian Whyte's annual census reports), creating the best opportunity for aprecise count of the population. To test this assumption, the Jolly Stochastic modelprovided an independent measure of the elephant population.The Jolly stochastic model calculated a bull population which was twice that of thecensus for most years with the exception of 2007 (range 1.50 to 2.28 times) (Table 2).The total number of bulls calculated by means of the model, showed similar trends interms of increases and decreases when compared to the census results (Figure 2 (c)). Thisis encouraging as it suggests that both methods are tracking the same pattern in theelephant population despite being based on different data collected over different periodsof time and serves to corroborate the aerial census data. These findings substantiate thatmovements in and out of the APNR, and especially those of young adult bulls, criticallydetermine the fluctuations in the number of bulls from year to year.Although census results have indicated that there has been a sharp decline in the numberof bulls counted in 2008 when compared to previous years (Figure 2.(c)), both the youngand senior bull segment of the population had declined in 2007 according to the modelpredictions (Figure 2 (a) and (b)). The consistent decline in the senior bull segment of thepopulation over the past four years has however been masked by the annual aerial surveyresults because of the inability of this survey technique to distinguish between young andprime bulls from the air. The importance of the Jolly Stochastic model’s ability todistinguish changes in the number of different segments of the bull population is apparentin the survival rate estimates of both young and senior bulls. The survival rate of seniorbulls has decreased over time (r2 0.98, F 91.3, P 0.01) while the survival rate of youngbulls has fluctuated more widely (Table 2).

Number of young (a)200320042005200620072008YearNumber sightedModel predictionNumber of senior 008YearNumber of young and seniorbullsNumber sightedModel 00620072008YearNumber sightedModel predictionCensusFigure 2. Annual estimates of the number of young (a), senior (b) and total (c) number ofbulls within the APNR.

Objective 2: To determine how many breeding herds use the APNR.Twenty four different breeding herds have been identified within the APNR. In total 385different cows have been identified while 61 cow identikits could not be grouped intoany particular herd and have been classified as ‘uncategorised’. Identikits which couldnot be classified were collected during sightings were visibility was poor due to the densityof the vegetation or in areas where breeding herds where skittish which only permitted one ortwo unknown individuals to be sighted. The re-sighting rate of cows is 88% (Table 3)Table 3. The proportion of breeding herds that have been re-sighted of the total numberof herds identified within the APNR.Total number ofNumber ofNumber ofRe-sightingbreeding herdsbreeding herds rebreeding herdspercentageidentifiedsightedsighted only oncesince May 2003since May 2003since May 20032421388Multiple sightings of the same herd were pooled within a month to obtain a monthly resighting rate. A third order polynomial model was used to describe the relationshipbetween the accumulative number of new sightings over time and the rate at whichbreeding herds were re-sighted. The regression model adequately explained the variationin new sightings and re-sightings over time (Figure 3, r2 0.99 and r2 0.99 respectively).By May 2004 the accumulative re-sighting rate of breeding herds had exceeded theaccumulative sighting of new herds so that re-sightings of identified herds were morefrequent than new sightings of herds. Similarly to senior bulls, breeding herds form thestable components of the population as very few new sightings of herds have been madeover time.Accumulative number ofsightingsBreeding ebJulyDecMayOctMayAprJanJunNovTime (months)Figure 3. The accumulative number of re-sightings and new sightings of individualbreeding herds

Objective 3: To identify the big tuskers that frequent the APNR.Elephant tusks grow at a constant rate in length throughout life with the tusks of bullsincreasing on average, at 11cm per year while female tusks increase at 8.5 cm per year.In elephants the rate at which the tusks increase in weight accelerates as the pulp cavityfills and this does not occur until old age in bulls. Although tusks increase in weight at arate of 2g a day (Spinage 1994), the almost exponential increase in male tusks weightwith age is not reflected in this daily rate of tusk growth. Laws (1966) found that in bullsthe rate of tusk growth increases progressively throughout life to at least 240 lbs (109 kg)at 60 5 years. Growth in tusks does however depend on genetics, the geology of theplace that the elephant occupies as well as the frequency at which the tusks are used. Ifthere was no wear and tear on a bull elephant’s tusks he could acquire tusks as long as550cm in his lifetime however, bull’s tusks generally reach 250cm in length upon old age(Spinage 1994).Large tusked bulls are under-represented in most African elephant populations (DouglasHamilton 1997). This is the legacy of a history of uncontrolled hunting and poaching onthis continent. Although, trophy hunting is now regulated in a few countries it stillcontinues to focus on the few large tusked individuals remaining in the population. Thereis concern that long-term selective off takes will ultimately depress the quality of trophies(Stalmans et al. 2002).Bulls with tusks estimated to be in excess of 50lbs were specifically recorded as largetusked individuals to determine how many large tusked bulls frequent the APNR. SinceMay 2003 we have recorded 26 large tusked individuals. For large tusked bulls, theregression model adequately explained the variation in new sightings and re-sightingsover time (Figure 4, r2 0.99). In July 2004 the accumulative re-sighting rate exceededthe new sightings over time. Furthermore, the accumulative new sighting rate hadreached an asymptote by January 2005 which means that very few new large tuskedindividuals are entering the APNR. Hence we recommend that bulls with tusks in excessof 50lbs should be afforded protection throughout the APNR in terms of their aesthetic,social and genetic importance.Accumulative number ofsightingsLarge tusked FebJulyDecMayOctMayAprJanJunNovTime (months)Figure 4. The accumulative number of re-sightings and new sightings of prime bullswith tusks in excess of 50lbs.

Objective 4: To determine the movement of elephants between the APNR andadjacent areas.Objective 6: To establish the extent to which elephants frequent different parts ofthe APNR and KNP.These two objectives focus on the distribution of elephants, differing primarily in theirspatial and temporal scale. The first (Objective 4) considers distribution betweenProtected Areas and the second patterns (Objective 6) within these.Between May 2004 and December STE-SA fitted GPS-telemetry collars to 28 elephantswithin the APNR, 21 bulls and seven cows. The distribution data derived from these areused to determine movement patterns (see section 5 for information on the deployment ofcollars, study animals and the quality of the data).Elephants collared within the central APNR move freely between these private reserves,the KNP and to a lesser degree the other protected areas contiguous with the westernborder of KNP. On average, just over 72% of the location records lay within the APNR,20% occurred within KNP, less than 2.5% in other conservation areas and almost 5%outside protected areas. Movements outside protected areas are primarily north of theOlifants River between the APNR and Phalaborwa. There are differences in distributionaccording to sex, on average 84.7% of the cows location records are located within theAPNR, compared with 68.8% of the senior adult bull's; 67% of the adult bull's and 68.7%of the sub-adult bulls' data points (Figure 5), however these are not statisticallysignificant (Friedman's ANOVA). It is possible that individual variability within thesegroups is masking the biological significance of the differences between them.Percentage of location records40.035.030.025.020.015.010.05.00.0Herds (cows)BaluleBulls – senior adultKlaserieTimbavatiUmbabatBulls – adultKNPManyeletiSabi SandsBulls – sub-adultLetabaoutsideFigure 5. The proportional abundance of elephant distribution data points withindifferent conservation areas.The veterinary fence that defines the western boundary of the study represents anartificial and abrupt barrier to elephant movements, for this reason the recently developedLocal Convex Hull (LoCoH) method of home range determination (Getz & Wilmers2004) is appropriate. Getz et al. (2007) found the adaptive LoCoH (a-LoCoH) to besuperior to others. They recommend, as a rule of thumb, using greatest distance betweenany two points as the estimate for the a value. Given the eccentric shape of many of thehome ranges within the study we refined this rule of thumb to be the average of the

primary axis (greatest distance between any two data points) and the secondary axis(greatest distance between two points perpendicular to the 10 axis). This is intended toreduce the difference in the a value between ranges of a similar area but differingeccentricity. The annual a-LoCoH plots for each study animal are presented in Figure 6.Table 4 includes the mean annual home range size for each of the study animals forwhich there are sufficient data ( 1 year October – September). The average breedingherd home range, as derived from the a-LoCoH plot is 490 km2. Bull’s annual homeranges are substantially larger with the sub-adult bulls being on average 1 599 km2, theadult bulls 809 km2 for and the senior adult bulls 805 km2.Table 4. Telemetry summary and range statistics for elephant collared by STE-SA since2002. Age classes of bulls cover the following periods: sub-adult 25 years old; adult 2535 years old; senior adult 35 years old. All cows are adults. Collared refers to themonth and year the animal was first tagged and so indicates the period for whichtelemetry data are available. Sample size (n) includes data points up the end ofSeptember 2008. MCP (minimum convex polygon) is the home range estimate in km2.Adaptive local convex hull (a-LoCoH) is the area of the 100% isopleth in aughleyEverestGowerProudWessaMellowCapt MandyUmbabatLapajumaKeoleriaSummerAge classsenior adultadultadultadultsub-adultsenior adultsub-adultsub-adultadultadultsenior adultadultadultsenior adultadultadultsenior adultsub-adultsenior adultsenior adultadultCollaredMay 04Nov 04Nov 04May 05Oct 05Oct 05June 06Sept 06Sept 06Sept 06Oct 06Nov 06Nov 06April 07April 07June 07June 07June 07Nov 07April 08July 08May 04Nov 04May 05Oct 05Oct 06July 07July 53791526a-LoCoH Notes7501327data stops July 200512971298711data stops March 20077502353563396data stops July 20088634501200897431data stops January 2008data stops June 20081997data stops July 2008Nov 2005 – collared by TEMBONov 2005 – collared by TEMBO435451640457260696Nov 2005 – collared by TEMBO

Figure 6. Annual adaptive local convex hull home-range plots (100% isopleth) ofelephants collared within the APNR. Annual data starts from the beginning of October(typically the dry-wet transition month when the first substantial rains fall) and end inSeptember (typically the end of the dry season).

Figure 6 cont. Annual adaptive local convex hull home-range plots (100% isopleth) ofelephants collared within the APNR. Annual data starts from the beginning of October(typically the dry-wet transition month when the first substantial rains fall) and end inSeptember (typically the end of the dry season).

Figure 7. Adaptive local convex hull home-range plots (100% isopleth) of elephantscollared in the eastern Kruger NP. Data from December 2006 to September 2008 areincluded in the range calculation.

Two other STE-SA projects are also relevant to Objective 4: the deployment of eightcollars on bulls in the east of KNP in December 2006, and the fitting of collars toelephants in the far north of the KNP. The first project aims to explore movementsbetween KNP and Limpopo NP in Mozambique and the second movements betweenKNP and Gonarezhou NP in Zimbabwe.Of the seven new elephants collared south of Shingwedzi (the last collar was fitted toMac TKM05) two (28.6%) moved into LNP within seven weeks and have not returned intwo years following (Figure 7). The results of this study are currently being written upfor publication.The first of 12 collars to d be deployed in the northern KNP was fitted to a young bullelephant in October 2008. We plan to deploy another 8 or more this winter. STE-SA isalso exploring the possibility of collaboration with wildlife biologists in Zimbabwe tohave collars fitted in Gonarezhou NP.Objective 5: To determine the changes in the density of elephants within the APNRand how this changes over time and whether these changes are through births,deaths or elephant movements to and from the KNP.Census resultsIn 1992 the fence between the Timbavati-Umbabat and Klaserie PNRs was taken down,creating the consolidated Associated Private Nature Reserves (APNR). A year later thefence separating the APNR and Kruger NP was completely removed. These actionsradically altered the landscape and dynamics of the elephants within the Private Reservesand created a situation where management decisions have to take into consideration thecircumstances and actions of the other properties as well. Fortunately a repeated annualaerial census across the three reserves was initiated in 1992 and so we have comparabledata reflecting dry season population estimates over the past 16 years. However, animalmovements and environmental conditions such as rainfall and woody canopy cover, aswell as observer bias change from year to year giving aerial counts an inherent variabilityover and above real changes in the animal's demographics. For this reason it is preferableto track trends in the population rather than focus on the absolute value in any one year.To reduce the erroneous variability in census results, a weighted average (two-pointweighted interpolation - Owen-Smith, Mason & Ogutu 2005) was calculated and thesmoothed plot is also presented in Figure 8. This smoothing function relies on the data ofthe year in question as well as the next year's data hence there is no estimate for 2008.Balule PNR recently joined the APNR (in 2005) which meant that including these data inthe evaluation would render comparisons over longer time periods invalid. Furthermore,no data is currently available on the break down of the number of bulls, breeding herdsand calves within Balule PNR. Consequently the subsequent evaluation focused onelephant demographics within the Timbavati, Klaserie and Umbabat PNRs.In general, the 16 years since 1992 has seen an increase in elephant numbers within theAPNR from ca. 500 to ca. 1 300, reflecting a mean annual population growth of just lessthan 6.1%. This is less than the inferred 6.7% growth in Kruger NP during the cullingperiod (1973-1994, Whyte et al. 1998) when the population was at a lower density andbelow the 7% physiological limit suggested by Hanks & McIntosh (1973), Calef (1988)and Owen-Smith (1988). A growth rate of 6.1% is nevertheless larger than growth rates1

typically reported elsewhere (2-4%) for elephant populations under favourableenvironmental conditions, in the absence of external influences such as poaching,demographic distortion or range compression (Armbruster & Landel 1993; Moss 2001).Growth rates greater than 10% have recently been reported from other areas in southernAfrica (van Jaarsveld et al. 1999; Mackey et al. 2006), however these are derived fromsmall populations subject to very artificial conditions. ‘Natural’ populations (i.e. largewith relatively undistorted demographics) have experienced growth rates of 9.6 - 11.2%(Chobe NP 1973 to 1993, Junker et al 2008) and 17.7% (Hwange 1986 to 1990,Chamaillé-Jammes et al 2008) but these have been largely ascribed to immigration fromsurrounding areas and in both cases the population appears to have subsequentlystabilized.Figure 8. Consolidated elephant census data from the Klaserie, Timbavati and UmbabatPNRs2

Figure 9. The annual elephant population (observed census data) and the hypotheticalpopulation based on a 6.1% annual growth rate (hypothetical). Black bars show yearswhen the census counts were larger than the hypothetical population and light grey barsindicate when it was less. The annual growth rate is the percentage change in theobserved population from year to year (thick red line).The above may suggest that the increase in APNR elephant numbers is due to theinherent growth expected of a population under favourable habitat conditions. However,when one looks at change in the APNR population from year to year substantialfluctuations are obvious, even in the smoothed data (Figure 8). Annual growth ratesbetween 1992 and 2008 vary between -34% and 120% and in most years the observedpopulation was below that predicted by a 6.1% growth rate (Figure 9).Two explanations may account for this:1)erratic counting from year to year and2)movement of elephants between the APNR and other protected areas within theGreat Limpopo Transfrontier Park.When considering how each of the various segments of the population (breeding herds,bulls and calves) contribute towards the overall trends in density, the total elephantpopulation density in the APNR appears to be largely determined by the density ofbreeding units. The trends in the total population density follow those of breeding unitsmore closely than any other component of the population (Figure 10 (a) and (b)).Breeding units also make up the largest proportion of the total population when comparedto either bulls or calves (81% as opposed to 15% and 4% on average since 1992). Since2003, breeding unit numbers within the APNR have fluctuated above and below thegeneral trend in numbers found within the KNP (Figure 10(b)). The density of bulls3

(a)Breeding herd density(number of elephants2)per kmDensity (number of2)elephants per km1.21.

adjoining(APNR) the Kruger National Park (KNP) 2. Senior Researcher and co-workers Henley, M.D. Transboundary Elephant Research Programme (Save the Elephants) Henley, S.R. Transboundary Elephant Research Programme (Save the Elephants) D