Habitat and Home Range Selection of Reintroduced Cheetah ...



Habitat Preference and Home Range of Reintroduced Cheetahs (Acinonyx jubatus) in Mkhuze Game Reserve, South Africa.

Rachel Orser

Mount Allison University, Canada

Supervisors:

Dr. Ron Aiken, Mount Allison University

Dr. Mike Perrin, Operation Wallacea Senior Scientist

Dr. Bella Davies, Operation Wallacea GIS supervisor (Week 2)

Dr. Victoria Hobson, Operation Wallacea GIS supervisor (Weeks 3-6)

Abstract

Protected areas are established to give endangered species a region to survive, but it is important to assess these areas to determine if they are providing adequate resources for these organisms (Marker et al., 2008). Habitat selection and home range of a reintroduced population of cheetahs, Acinonyx jubatus, were studied in the Mkhuze Game Reserve, South Africa from December 2006 to May 2008. It was expected that cheetahs would have similar habitat preferences to each other and that they would avoid areas of human habitation, while preferring open areas. Home range was not expected to differ according to climatic changes because they are able to live adapt to both wet and arid regions (Caro, 1994; Hunter & Hamman, 2003; Skinner & Chimimba, 2005). Temperature was expected to be relatively constant therefore it should not have influenced home range size. It was found that habitat avoidance was common for all cheetahs, but selection varied from open to closed habitat types. As anticipated, females had larger home ranges than males. Climatic analyses showed such things as temperature and rainfall could influence that home range. These results support previous works, which suggest that the cheetah is a more adaptive species than the scientific world originally thought (Bissett & Bernard, 2007).

Key Words

Cheetah (Acinonyx jubatus), Coalition, Home range, Habitat selection, Mkhuze.

TABLE OF CONTENTS

PAGE

Abstract i

Table of Contents ii

List of Figures iii

Acknowledgements v

Introduction 1

Materials and Methods 7

Results 12

Discussion 24

Literature Cited 29

Appendix 32

LIST OF FIGURES

Page

Figure 1 ArcMap 9.1 projection of the Mkhuze Game Reserve, South Africa 5

Figure 2 Map of South Africa (Soweto Tour) 6

Figure 3a Marakele point counts on Mkhuze map 14

Figure 3b Habitat preference of Marakele based selection ratio 14

Figure 4a Mahlabeni point counts on Mkhuze map 15

Figure 4b Habitat preference of Mahlabeni based selection ratio 15

Figure 5a Massina point counts on Mkhuze map 16

Figure 5b Habitat preference of Massina based selection ratio 16

Figure 6a Mavabaza point counts on Mkhuze map 17

Figure 6b Habitat preference of Mavabaza based selection ratio 17

Figure 7a Coalition point counts on Mkhuze map 18

Figure 7b Habitat preference of Coalition based selection ratio 18

Figure 8a Marakele Minimum Convex Polygon 19

Figure 8b Marakele Fixed Kernel Density Estimator 19

Figure 9a Mahlabeni Minimum Convex Polygon 19

Figure 9b Mahlabeni Fixed Kernel Density Estimator 19

Figure 10a Massina Minimum Convex Polygon 20

Figure 10b Massina Fixed Kernel Density Estimator 20

Figure 11a Mavabaza Minimum Convex Polygon 20

Figure 11b Mavabaza Fixed Kernel Density Estimator 20

Figure 12a Coalition Minimum Convex Polygon 21

Figure 12b Coalition Fixed Kernel Density Estimator 21

Figure 13 Rainfall and Home Range Analyses for Coalition and Marakele 22

Figure 14 Temperature and Home Range Analyses for Coalition and Marakele 23

Acknowledgements

All of the companies, institutions and individuals that supported my expedition to South Africa. Special thanks to Leadership Mount Allison, Mount Allison University SAC Enrichment Fund and the Explores Club Grant Program.

Amanda MacKenzie, my statistical goddess.

Beka Nxele, Mkhuze Game Reserve: Regional Ecologist

Chris Kelly, Priority Species Project Manger and Field Coordinator

Dr. Mike Perrin, Operation Wallacea Supervisor Scientist of the University of KwaZulu-Natal

Dr. Bella Davies, Operation Wallacea GIS supervisor (Week 2), Oxford Brooks University (UK)

Dr. Victoria Hobson, Operation Wallacea GIS supervisor (Weeks 3-6), Swansea University (UK)

Dr. Ron Aiken, Mount Allison University Special Topics Supervisor

Dr. David Lieske, Mount Allison University, GIS

Invertebrate lab members at Mount Allison University. Barnacles and Slugs 2009, and cheetahs too.

Janet Edwards, Priority Species Project Field Assistant (temporary)

Norman, Venita Orser and family: for understanding why I had to do this.

Operation Wallacea, South Africa division

Paul Haveman, Mkhuze Game Reserve Conservation Manager

Richard and Caz, Game Reserve Guides from Sontuli

Thami Mthembu, Priority Species Project Field Assistant

Mkhuze Game Reserve Education Camp Staff and its volunteers

Xander Combrink, EKZN Biodiversity Project Manager

Zakele Mkhwanazi, Primary Safety Officer (EKZN /MGR)

Rodney and Sipo, Safety Officers (Wilderness Research School)

Introduction

Protected areas are established to give endangered species a region to survive, but it is important to assess these areas to determine if they are providing adequate resources for these species (Marker et al., 2008). Protected areas can be defined by a number of different physical boundaries, but the most commonly used is the fence. These fences are not only free-standing structures but also include guard dogs, noise barriers and even biofences using scents from territorial species. The primary function of these protected areas is to defend biodiversity by eliminating various threats, such as humans and other competing species. They not only protect the flora and fauna of the region but also the ever-encroaching human population from potentially harmful wildlife (Hayward & Kerley, 2009).

Large mammalian predators require large areas of land and the cheetah is no exception (Burdett, 2007; Kelly et al., 1998). These large predators can be a threat to humans and their livestock but this relationship goes both ways. Cheetahs in particular are threatened by humans for the purposes of illegal trade, unregulated captive breeding, and human persecution due to stock-raiding (Hayward & Kerley, 2009; Marker-Kraus, 1997a; Marker-Kraus & Kraus, 1997b). Stock-raiding by cheetahs is a very large concern in north-central Namibia as the preferred refuge for them is commercial farmland with its large prey population and lack of large competitors (Marker, et al., 2008). The KwaZulu-Natal region of South Africa has attempted to resolve these problems with the introduction of enclosed fencing around their protected area to help such animals as the cheetah.

Cheetahs, Acinonyx jubatus, were added to the IUCN Red list first in 1986 and are currently listed as vulnerable (IUCN, 2007). The largest cheetah population is in Namibia with an estimated population size of 2500-3000 adults (Morsbach 1987). Cheetahs were once distributed widely in non-forest areas of Africa, the Arabian Peninsula, the Middle East, Pakistan, India and the southern regions of Russia (Guggisberg, 1975; Caro, 1994; Nowell & Jackson 1996; Marker 1998; Sunquist & Sunquist, 2002; Hunter & Hamman, 2003) and now only exist in small pocket populations throughout Africa and a small group of 40-50 Asian cheetah in Iran, according to a 2004 census (Hunter & Hamman, 2003). Cheetahs were eliminated around 1930 in KwaZulu-Natal but various attempts of reintroduction to the area have occurred, including one in the Mkhuze Game Reserve in 1965 (Hunter & Hamman, 2003) (Figure 1). Until now, the only successful South African reintroduction of the cheetah was in the Phinda Resource Reserve in 1994 (Hunter, 1998; Hunter & Hamman, 2003) (Figure 2).

One way to measure the new resource availability is through home range. The first definition of home range was given by Burt (1943) as the area traversed by an individual during its normal activities of food gathering, mating and caring for young. There are various methods used to determine an individual’s home range. This study focused on the kernel method (Worton, 1989) and the minimum convex polygon method (MCP) (Mohr, 1947), which is the oldest and most commonly used estimator of home range. The kernel method, a utilization distribution (UD) technique, has two subdivisions, fixed and adaptive (Worton, 1989). Utilization distribution techniques allow for habitat utilization to be investigated by determining the frequency of use, as opposed to assuming equal use of the individual’s home range, as is the case for MCP (Katajisto & Moilanen, 2006). For the purpose of this study the fixed kernel density estimator (FKDE) was used to determine home range of each cheetah since his or her re-introduction into the reserve. Although minimum convex polygon method is more commonly used to determine home range, the fixed kernel density estimator is known for being less affected by outliers and includes less unused space (Harris et al. 1990). Both methods were used in this study as the MCP gave a potential area of home range and the FKDE indicated the key habitats essential for protected based on habitat usage.

Bissett and Bernard (2007) found that cheetahs of the Kwandwe Private Game Reserve in the Eastern Cape Province of South Africa occupied different size home ranges based on their sex and their status. These home ranges varied from 32.7 km2 for mature males to 93.9 km2 for independent cubs. It is expected that similar results will be found in the Mkhuze Game Reserve.

A concern for the cheetah population is the number of population bottlenecks the species has experienced. The first of these dates back to nearly 10,000 BC and resulted in genetic disease, deformed jaws and a lack of genetic diversity within the species (Menotti-Raymond & O’Brien, 1993). Recent fragmentation of natural habitat over the last thirty years has introduced more issues, such as population fragmentation, and has been a large concern for the survival of the natural population (Bissett & Bernard, 2007; Marker et al., 2008). Attempts at captive breeding and release have had mixed results but the main issue appears to be human interference (Jule et al, 2008) and low litter survival (Marker, 2008). However, transfer programs, which involve the movement of one individual to another geographic location, are another option, which was exercised in selecting the specimens used in this study.

Biotelemetry, which can be in the form of radio telemetry, acoustic telemetry, or satellite telemetry, is an increasingly common method used by conservation scientists to assist in the understanding of threats and causes of population decline in the assessment of endangered status of species, such as the cheetah (Cooke, 2008; Hulbert & French, 2001; Millspaugh & Marzluff, 2001). This method is known for its ability to avoid problems with observation, such as transportation restraints and animal avoidance (Cooke, 2008). These observation problems result in data that are full of errors and bias, which needs to be avoided in order to show an accurate representation of habit and habitat preference (Cooke, 2008; Hulbert & French, 2001). This method was used not only for it’s convince and ease in data collection, but it has been shown that radiocollars do not influence survival rate or individual fitness of the animals studied (Golabek et al. 2008). The purpose of using biotelemetry is to collect data on an animal to illustrate such things as habitat preference, migration routes and home range that are unattainable with other research methods and to not cause excessive harm to the animal (Cooke, 2008; Hulbert & French, 2001; Millspaugh & Marzluff, 2001).

Most of the data on cheetah’s habitat preference and home range have not taken climate changes into account. However, habitat preference has been found to be different in different parts of Africa (Bissett & Bernard, 2007; Caro, 1994; Marker et al., 2008; Kelley et al., 1998). These studies also have demonstrated together that the cheetah is adaptive to the location that it is in. This study looked to take this information one step further.

The aim of this study was to investigate the habitat selection preference of six introduced cheetahs based on rainfall, temperature and home range. The objectives were threefold: (1) to investigate habitat selection for each cheetah based on point counts in polygons; (2) to determine home range for each cheetah using minimum convex polygons and fixed kernel density estimator; (3) to investigate the affects of rainfall and temperature on the habitat selection of each cheetah.

[pic]

Figure 1 – Radio tracking study area and habitat map of the Mkhuze Game Reserve, South Africa. Park measures 35,740 hectors in area and is part of the KwaZulu-Natal chain of game reserves (Latitude: -27, Longitude: 32).

[pic]

Figure 2 – Map of South Africa. ()

Methods

Study Subjects

Beginning in early October 2006, six cheetahs, two males and four females, were released into the Mkhuze Game Reserve in KwaZulu-Natal of South Africa after spending five weeks in the boma (located just before Mkumbi; Figure 1) and being fitted with radio collars. Two males from the Develdt cheetah-breeding center in the Northern Province of South Africa were released in October, and two female cheetahs known as Mavabaza and Mahlabeni from the neighboring park of Phinda were brought in and released from the boma in December. The final two females, Marakele and Massina, who were named after the two private parks that they were bought from, were released on the 15th of January 2007. The males were not named until June 2008; male cheetah two was named Wallace but male cheetah one was never named as he was killed by hyena on May 11th, 2008. On May 12th 2008, the female cheetah known as Massina was also found dead due to a hyena attack after she had fallen into a hole. Before her death she had successfully raised two male cheetahs from the previous year’s litter. Data were first collected for the young cubs on May 10th 2008, and one was fitted with a collar on July 27th 2008. As the brothers are a coalition, only one needed to be collared, as they will stay together until one dies. Data collected on these males was not used in this analysis. Mahlabeni gave birth to four cubs in February 2008; they will be collared if they reach adult size.

Study Site/Habitats

The Mkhuze Game Reserve, South Africa, is a 35,740 hectors government park. The habitat types, as identified for this study, were classified by Van Rooyen (2004). These types include: Riverine Forest (RF), Acacia nigrescens Tall Open Woodland (ANTOW), Euclea divinorum Low Thicket (LT), Spirostachys africana Dry Closed Woodland (DCW), Lebombo Open Woodland (LOW), Terminalia sericea Savanna (TSS), Acacia tortilis Low Open Woodland (ATLOW), Acacia nilotica Low Closed Woodland (ANLCW), Palm Savanna (PS), Lebombo Wooded Grassland (LWG), River Floodplains and Seasonal Streams (RFSS), Fresh water lakes and pans (FWLP), Human habitation (HH). (Appendix)

Equipment/Procedures

Six cheetahs were fitted with radio transmitter collars (Sirtraks) when they were re-introduced to the Mkhuze Game Reserve from October 2006-January 2007. The cheetahs were tracked once daily from December 2006 to end of July 2008 and GPS records of their location were recorded on both a hard copy and through the GPS device. Due to lack of data software, the information collected from June and July 2008 was not used in this analysis. Data for each cheetah were not completed daily as the cheetahs were not always found; they could have been in a valley or cave which would not allow a signal to transmit. Sightings of cheetah were described as A) actual visual, B) incomplete visual or C) triangulation. An incomplete visual meant that the researchers knew where the cheetah was but it was not seen due to tall grass or thick vegetation. When an actual visual was completed the researchers were to make notes of the cheetah’s activity; however, there were several datum entries without this information due to various persons collecting data (C.Kelly, 2008).

Before a cheetah was collared, the managers of the park were notified and the section manager was present. Cheetahs were darted by a Daniject dart gun and dart to a muscled part of the body, either the rump or shoulder, as to prevent excessive harm to the animal. The drugs used to dart the cheetah were Zoletil and Ketamine. The drug used depended on whether or not a veterinarian was present. Ketamine is known to be abused and thus may only be administered (8-10mg/kg for an adult cheetah) by a professionally trained veterinarian. The dose for Zoletil was 3-4 mg/kg for an adult cheetah, administered only by an authorized and proficient game ranger. This drug wears off after approximately two hours and a veterinarian is not required to be present (C.Kelly, 2008). Once a cheetah was darted it was also given a vitamin cocktail (⅓ Biosolamine, ⅓ Vitamin B and ⅓ Vitamin C) and a 3.5ml dose of Pedistrep.

Each cheetah was fitted with a Sirtaks collar complete with battery and micro-transmitter. The battery had a life of two to three years after which the cheetah was to be recaptured to either remove the collar or to get a replacement. The aerial of the transmitter sent out a pulse signal, ranging from 148-152 MHz, which was picked up by a receiving aerial. There were two receiving aerials used for this project, a Sirtraks model (, New Zealand) and a Telonics model (Telonics, Mesa, AZ, USA); both had a line of sight range of 15km. There were no activity signals, such as movement, heart rate, or body temperature, recorded for the cheetahs, but if a cheetah was inactive for twelve hours or more then the transmitter sent out a mortality signal. The aerial was attached to a frequency receiver,which was programmed to identify each of the cheetahs’ signals independently. Once a frequency had been selected at a given site, generally in the area of where the cheetah was last spotted, the aerial was held high above one’s head and turned vertically to scan a 360° area. In order to increase the likelihood of receiving a signal, the tracker stood on the roof of the tracking vehicle to get as high as possible. If a signal was found then the aerial would be turned horizontally to more accurately distinguish where the signal was coming from. The team then followed this signal in an attempt to get a visual sighting of the cheetah. If a cheetah was not seen from the road then the research team searched for the cheetah by foot to get an A or B sighting. However, if the cheetah was in a location where it was unsafe to walk then a triangulation would be taken (C.Kelly, 2008).

The GPS (Garmin E-trex; Garmin Ltd, Olathe, KS, USA) for this project was accurate to a 5m range. When a cheetah was located, a GPS reading was taken and marked on the GPS. Each cheetah was given a unique symbol on the GPS to keep recordings accurate; for example, the cheetah known as Mahlabeni was the car symbol. For a triangulation, two readings from different locations were taken in order to locate where the cheetah was based on the location of where the reading lines intersected. When the researcher with the frequency receiver located the strongest signal they indicated this to a second researcher who took the bearing point and a GPS reading of longitude and latitude. To get an accurate bearing point the researcher held the GPS unit in front of them, arm fully extended and walked towards a designated landmark, and this was repeated until the same bearing point was taken three times. If the cheetah was moving then a triangulation was not taken, as it would not be accurate (C.Kelly, 2008).

Data Analysis

Triangulations were made using LOAS version 4.0.3.1 Ecological Software Solutions (). A limitation of this procedure was movement of the cheetah between triangulation bearing measurements but this was unavoidable. The bearing and GPS locations were put into the LOAS software and a location of the cheetah was given based on the intersection of these two bearings. In order for a triangulation to be accurate the team had to be within a 2km radius of the collared cheetah. A final triangulation location was precise to within 200m (C.Kelly, 2008).

ArcMap GIS (version 9.1, ESRI, Redlands California, USA) was used to determine the home range based on the FKDE (Bissett & Bernard 2007; Worton 1989) and MCP (Marker et al. 2008) methods. This program was also used to determine habitat preference for each cheetah based on the Mkhuze Game Reserve habitat map. This map was made and updated in 2005 by Dr. Bob Smith from Kent University, England. The map was projected to Africa Albers Equal Area Conic so that proper calculations of area and perimeter could be completed. Habitat preference was determined for each cheetah, by completing point counts in polygons. MCPs were used to determine monthly home range area of each cheetah so that it could be compared with climate data and both MCP and FKDE determined home range. This allowed for potential area usage and actually usage to be compared.

Data on rainfall and temperature were collected by the Mkhuze Game Reserve and were used to test for correlations between rainfall and area used, and temperature and area used. Habitat preference was determine by the selection ratio:

SR = proportion used (u) / proportion available

Statistical computations were completed using SPSS 16.0.

Results

i) Habitat Preference Analysis

Habitat preference from reintroduction to May 2008 was determined by a selection ratio in which the proportion of the park used was compared to the proportion of the park that was available. If this ratio was greater than 1 then the cheetah was selecting that habitat, if it was equal to 1 then selection was random, and if ratio was less than 1 then the cheetah had avoided this habitat. Habitat preference was not consistent across the five cheetah as some regions that were preferred by one individual were avoided by another. Preferred habitats included lebombo wooded grassland, lebombo open woodland, low thicket, palm savanna, acacia nilotica low closed woodland, and acacia nigrescens tall open woodland. All cheetahs avoided human habitation, riverine forest, and fresh water lakes and pans. (Figures 3a, 3b, 4a, 4b, 5a, 5b, 6a, 6b, 7a, and 7b)

ii) Home Range Analysis

There were two methods used to determine home range in radio-telemetry studies, minimum convex polygons (MCP) and the fixed kernel density estimator (FKDE). For each of the five collared cheetahs these analyses were done for the complete data set to gain perspective on how much habitat cheetahs use over their lifetimes. The MCP has been concluded in previous studies to be a less accurate estimation as it connects the outer limits of the data and thus includes areas that may not have been used. For this reason, the FKDE was used as a second, and potentially more accurate, method of home range analysis. The FKDEs were completed at a 50%, 90% and 95% level, with the internal sections for each of the cheetahs’ analysis being 50% and then outer most section represents the 95% level. A paired samples t-test showed that there was a significant difference between MCP and FKDE for the cheetahs, with the MCP encapsulating a greater area. (Figure 8a, 8b, 9a, 9b, 10a, 10b, 11a, 11b, 12a, and 12b).

iii) Climate Analysis

Rainfall and temperature were both statistically different between wet (October to March) and dry (April to September) seasons. As these were both statistically different it was then possible to compare home range with these two climate conditions. MCP home range analyses for all females were compared for significant differences using a one-way ANOVA and were found to be not significantly different (p=0.111). This allowed for Marakele to be representative of the females. The MCP home ranges were found to differ significantly depending on both climate conditions for both sexes (p 1, random =1, avoidance < 1) with standard error bars. Habitats labeled according to description of the habitat types as described by Van Rooyen (2004).

(a)[pic]

Figure 4a – ArcMap 9.1 projection of Mahlabeni’s point counts from December 2006-May 2008 on the Mkhuze Habitat map.

(b)[pic]

Figure 4b – Selection ratio of Mahlabeni’s habitat preference (selection > 1, random =1, avoidance < 1) with standard error bars. Habitats labeled according to description of the habitat types as described by Van Rooyen (2004).

(a)[pic]

Figure 5a – ArcMap 9.1 projection of Massina’s point counts from January 2007 - May 2008 on the Mkhuze Habitat map.

(b)[pic]

Figure 5b – Selection ratio of Massina’s habitat preference (selection > 1, random =1, avoidance < 1) with standard error bars. Habitats labeled according to description of the habitat types as described by Van Rooyen (2004).

(a)[pic]

Figure 6a – ArcMap 9.1 projection of Mavabaza’s point counts from December 2006 – May 2008 on the Mkhuze Habitat map.

(b)[pic]

Figure 6b – Selection ratio of Mavabaza’s habitat preference (selection > 1, random =1, avoidance < 1) with standard error bars. Habitats labeled according to description of the habitat types as described by Van Rooyen (2004).

(a)[pic]

Figure 7a – ArcMap 9.1 projection of the coalition’s point counts December 2006 – May 2008 on the Mkhuze Habitat map.

(b)[pic]

Figure 7b – Selection ratio of coalition’s habitat preference (selection > 1, random =1, avoidance < 1) with standard error bars. Habitats labeled according to description of the habitat types as described by Van Rooyen (2004).

(a)[pic](b)[pic]

Figure 8 (a) ArcMap 9.1MCP for Marakele, set at 95% UD (Area =225.870 km²).

(b) ArcMap 9.1 FKDE for Marakele, set at 50%, 90%, 95% UD (95% Area = 168.161 km²).

(a)[pic](b)[pic]

Figure 9a – ArcMap 9.1MCP for Mahlabeni, set at 95% UD (Area: 329.675 km²).

Figure 9b – ArcMap 9.1 FKDE for Mahlabeni, set at 50%, 90%, 95% UD (95% Area = 215.888 km²).

(a)[pic](b)[pic]

Figure 10a – ArcMap 9.1 MCP for Massina, set at 95% UD (area = 230.178 km²).

Figure 10b – ArcMap 9.1 FKDE for Massina, set at 50%, 90%, 95% UD (95% Area = 55.748 km²)

(a)[pic](b)[pic]

Figure 11a – ArcMap 9.1 MCP for Mavabaza, set at 95% UD (area = 169.481 km²)

Figure 11b – ArcMap 9.1 FKDE for Mavabaza, set at 50%, 90%, 95% UD (95% Area = 109.646 km²)

(a)[pic](b)[pic]

Figure 12a – ArcMap 9.1 MCP for Coalition, set at 95% UD (area = 113.741 km²)

Figure 12b – ArcMap 9.1 FKDE for Mavabaza, set at 50%, 90%, 95% UD (95% Area = 54.197 km²)

[pic]

Figure 13 - Bar graph representing rainfall values for the duration of the study with standard error bars and line graph representing the home range values of Wallace/Coalition and Marakele.

A paired t-test found that rainfall was significantly different between the wet and dry season (p ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download