<%BANNER%>

Patterns and Determinants of Human-Carnivore Conflicts in the Tropical Lowlands of Guatemala

Permanent Link: http://ufdc.ufl.edu/UFE0022317/00001

Material Information

Title: Patterns and Determinants of Human-Carnivore Conflicts in the Tropical Lowlands of Guatemala
Physical Description: 1 online resource (48 p.)
Language: english
Creator: Soto, Jose
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Human-carnivore conflicts have been identified as a major cause of large carnivore population declines and can have negative economic impacts on local people. Although Human-carnivore conflicts (HCC?s) have been studied intensively around the world, very little information on HCCs in Mesoamerica is available. I applied a survey questionnaire to cattle ranchers to examine patterns of livestock depredation and estimate total economic loss due to large predators in the tropical lowlands of Guatemala. Furthermore, I compared ranches with and without attacks using logistic regression and Aikaike's Information Criterion (AIC) to determine whether ranch characteristics (e.g., size, and number of cattle), livestock husbandry practices, and landscape structure in and around ranches explained the probability of occurrence of HCCs. Understanding patterns of livestock depredation and the influence of livestock husbandry practices and landscape features on HCCs is important for site-specific conflict mitigation strategies and identification of areas and ranches most prone to HCCs. The jaguar was the carnivore most accused of livestock attacks, followed by the puma and coyote in much smaller percentages; the type of livestock most attacked was cattle, followed by goats, with cattle taken falling between the ages of 2 days and 12 months, while weights of cattle attacked ranged from 11.4 to 431.82 kg. I detected a preference towards male cattle, and, similar to other sites, most attacks occurred at night, and were more common during the rainy season, reaching a peak during the wettest months. A major difference detected between other sites previously studied in the neotropics (e.g., South America) were possible livestock attacks by coyotes in parts of the study site. Economic losses from carnivore attacks on livestock were minimal, yet, because small cattle ranches are predominant in the area, the economic impact of each loss may be perceived as significant by the ranches impacted by these losses. Finally, landscape variables (e.g., forest cover area, distance to forest cover, and distance to rivers) were the best predictors of HCCs. Even though ranches did not employ practices considered beneficial to reducing carnivore attacks on livestock, ranch characteristics and livestock husbandry practices were not important predictors of livestock attacks in this study because they were similar among most ranches.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jose Soto.
Thesis: Thesis (M.S.)--University of Florida, 2008.
Local: Adviser: Giuliano, William M.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022317:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022317/00001

Material Information

Title: Patterns and Determinants of Human-Carnivore Conflicts in the Tropical Lowlands of Guatemala
Physical Description: 1 online resource (48 p.)
Language: english
Creator: Soto, Jose
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Human-carnivore conflicts have been identified as a major cause of large carnivore population declines and can have negative economic impacts on local people. Although Human-carnivore conflicts (HCC?s) have been studied intensively around the world, very little information on HCCs in Mesoamerica is available. I applied a survey questionnaire to cattle ranchers to examine patterns of livestock depredation and estimate total economic loss due to large predators in the tropical lowlands of Guatemala. Furthermore, I compared ranches with and without attacks using logistic regression and Aikaike's Information Criterion (AIC) to determine whether ranch characteristics (e.g., size, and number of cattle), livestock husbandry practices, and landscape structure in and around ranches explained the probability of occurrence of HCCs. Understanding patterns of livestock depredation and the influence of livestock husbandry practices and landscape features on HCCs is important for site-specific conflict mitigation strategies and identification of areas and ranches most prone to HCCs. The jaguar was the carnivore most accused of livestock attacks, followed by the puma and coyote in much smaller percentages; the type of livestock most attacked was cattle, followed by goats, with cattle taken falling between the ages of 2 days and 12 months, while weights of cattle attacked ranged from 11.4 to 431.82 kg. I detected a preference towards male cattle, and, similar to other sites, most attacks occurred at night, and were more common during the rainy season, reaching a peak during the wettest months. A major difference detected between other sites previously studied in the neotropics (e.g., South America) were possible livestock attacks by coyotes in parts of the study site. Economic losses from carnivore attacks on livestock were minimal, yet, because small cattle ranches are predominant in the area, the economic impact of each loss may be perceived as significant by the ranches impacted by these losses. Finally, landscape variables (e.g., forest cover area, distance to forest cover, and distance to rivers) were the best predictors of HCCs. Even though ranches did not employ practices considered beneficial to reducing carnivore attacks on livestock, ranch characteristics and livestock husbandry practices were not important predictors of livestock attacks in this study because they were similar among most ranches.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jose Soto.
Thesis: Thesis (M.S.)--University of Florida, 2008.
Local: Adviser: Giuliano, William M.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022317:00001


This item has the following downloads:


Full Text

PAGE 1

1 PATTERNS AND DETERMINANTS OF HUM AN-CARNIVORE CONFLICTS IN THE TROPICAL LOWLANDS OF GUATEMALA By JOSE R. SOTO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Jose R. Soto

PAGE 3

3 To God, who made a perfect world for us to live in and has never abandoned us. To my parents, Maria and Neftali (deceased ), for all their sacrifices and support. To Nancy and Danny, for bringing so much joy into my world and always reminding me of what truly matters in life. To the Peten: magical, mysterious land of beauty and life.

PAGE 4

4 ACKNOWLEDGEMENTS I have to thank m any people and organizations for their support. Funding for field work was provided by the United States Department of State, The University of Florida through a Jennings Scholarship, the Chester Zoo, and the Jaguar Conservation Prog ram of the Wildlife Conservation Society. My studies at the University of Florida were funded by a Fulbright/Organization of American States Ecology Initiati ve Fellowship and the Center for Latin American Studies of the University of Florida. I would like to thank the people at LASPAU, especially Renee Hahn, who administered my fellowship very efficiently and helped me obtain an extension on this fellowship. I am grateful to my advisor, Bill Giuliano for sharing his knowledge on all aspects of carrying out research projects and for all the help (academic and non-academic) he gave me while at Gainesville. Also to the members of my committee, Melvin Sunquist and Howard Quigley, who provided valuable comments a nd suggestions to improve this work. My field biologists Tomas Dubon, Jose Lop ez, and Ambrocio Marin always worked enthusiastically and provided tremendous suppor t in the field, as well as very insightful suggestions as to how to improve the project. Upon returning to Gainesville after completing my field work, Guillermo Lopez, Willy Raxon, and Me lvin Merida took over the project and have done a great job of keeping it a live. Guillermo Lopez has been instrumental for the continued success of this project, overseeing the logi stics, budgets, equipment, etc. I am also grateful to all the cattle ranchers and local people who willingly participated in this study. They opened their doors to us and always made us feel welcome.

PAGE 5

5 The WCS program for Guatemala provided lo gistical assistance and office space. I especially would like to thank Roan McNa bb and the Jaguar Conservation Program for supporting my work from the beginning. I thank CEMEC for providing all the GIS layers and data on precipitation. John Tirpak and Victor Hugo Ramos came to my rescue while I was struggling with some of the GIS analyses. Rob Fletcher, Meghan Brennan and Alan Agresti pr ovided very useful guidance in the statistical analyses. My parents, brother and sister, nieces, nephews and cousins haves always been there for me and they provide me with constant encouragement and love. Last, but definitely not least, all my l ove and gratitude go to Nancy and Danny. The sacrifices they have made si nce I started working in the fi eld are too numerous to mention. Danny has never questioned my long absences, busy schedule and all the moving around, and has always accepted our strange lifestyle w ith a happy and loving disposition. Nancy gave up a good job, her plans and two years of her life to support me while I obt ained my Degree. At home, she was my rock; if ever there was a s uperwoman, it would be the working wife of a graduate student. She took care of our son, me, the house, the car, the bills, etc., while she worked and applied to Graduate School, and al ways with a smile on her face. How she managed to pull that off, I will never know, but I respec t and love her with all my heart for it.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGEMENTS.............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT.....................................................................................................................................9 CHAP TER 1 INTRODUCTION..................................................................................................................11 Conservation Issue..................................................................................................................11 Determinants of Human-Carnivore Conflicts......................................................................... 12 Human-Carnivore Confli cts in Mesoam erica......................................................................... 13 2 METHODS.............................................................................................................................15 Study Area..............................................................................................................................15 Reports of Livestock Depredation by Carnivores .................................................................. 16 Patterns of Livestock Depredation.......................................................................................... 17 Livestock Husbandry Practices............................................................................................... 18 Landscape Metrics.............................................................................................................. ....19 Variable Selection, Model Building and Evaluation .............................................................. 20 3 RESULTS...............................................................................................................................28 Patterns of Livestock Depredation by Carnivores..................................................................28 Determinants of Human-Carnivore Conflicts......................................................................... 29 4 DISCUSSION.........................................................................................................................35 Patterns of Livestock Depredation.......................................................................................... 35 Determinants of Human-Carnivore Conflicts......................................................................... 38 Conservation and Management Implications......................................................................... 39 LIST OF REFERENCES...............................................................................................................43 BIOGRAPHICAL SKETCH.........................................................................................................48

PAGE 7

7 LIST OF TABLES Table page 2-1 Landscape characteristics and livestock husbandry practices obtained for each ranch in the Peten District, Guatem ala, 2007.............................................................................. 25 2-2 Results of the Principal Com ponents Analysis Corre lation Matrix, Peten, Guatemala, 2007....................................................................................................................................26 2-3 Results of the Principal Components Analysis for continuous landscape and lives tock husbandry practice variables meas ured at each ranch in Peten, Guatemala, 2007....................................................................................................................................26 2-4 Candidate models evaluated through logistic regression, Peten, Guatem ala, 2007........... 26 2-5 Ninety-five percent Wald confidence limits of the odds ratios obtained for the global model, Peten, Guatem ala................................................................................................... 27 3-4 Comparison of five candidate models in the Peten District, Guatem ala, 2007................. 34

PAGE 8

8 LIST OF FIGURES Figure page 2-1 Map of the Peten District, 2007......................................................................................... 24 2-2 Frequency of responses for categorical variables obtained fro m each ranch surveyed in the Peten District, Guat emala (2007). Graph shows hi gh outcomes of one response type for all variables. Tabl e 2-1 gives variable names.......................................................27 3-1 Map of study area showing ranches survey ed and landscape m etrics measured in the Peten District, Guatemala, 2007........................................................................................ 30 3-2 Sizes of cattle ranches surveyed in P eten, Guatemala, 2007. Where small ranches: 11.2 to 500 ha; medium ranches: 500 to 1,000 ha; and large ranches: >1,000 ha. Sizes were obtained directly from ranch owner or administrator...................................... 31 3-3 Livestock depredation incidents by jagua rs, pumas and coyotes in Peten, Guatem ala from 2003 to 2007.............................................................................................................. 31 3-4 Livestock depredation incidents by carnivores per m onth and monthly rainfall (mm). Rainfall data obtained from CEMEC in th e Peten District, Guatemala, 2003 2007....... 32 3-5 Livestock depredation inci dents by carnivores per year a nd total annu al rainfall (mm) in the Peten District, Gu atemala from 2003 to 2007. Rainfall data obtained from CEMEC.......................................................................................................................... ....32 3-6 Cattle depredation incidents by weight cla sses of animals attacked in the Peten District, Guatemala from 2003 to 2007............................................................................. 33 3-7 Cattle depredation incidents according to age (months) of animals attacked in the Peten District, Guatem ala from 2003-2007....................................................................... 33

PAGE 9

9 Abstract Presented to the Gradua te School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science PATTERNS AND DETERMINANTS OF HUM AN-CARNIVORE CONFLICTS IN THE TROPICAL LOW LANDS OF GUATEMALA By Jose Roberto Soto December 2008 Chair: William Giuliano Major: Wildlife Ecology and Conservation Human-carnivore conflicts have been identified as a major cause of large carnivore population declines and can have negative economic impacts on local people. Although Humancarnivore conflicts (HCCs) ha ve been studied intensively around the world, very little information on HCCs in Mesoamerica is available. I applied a survey questionnaire to cattle ranchers to examine patterns of livestock depredation and estimate total ec onomic loss due to large predator s in the tropical lowlands of Guatemala. Furthermore, I compared ranches with and without attacks using logistic regression and Aikaikes Information Criterion (AIC) to dete rmine whether ranch characteristics (e.g., size, and number of cattle), livestock husbandry practices, and landscape structure in and around ranches explained the probability of occurrence of HCCs. Unde rstanding patterns of livestock depredation and the influence of livestock husba ndry practices and landscape features on HCCs is important for site-specific conflict mitigation st rategies and identification of areas and ranches most prone to HCCs. The jaguar was the carnivore most accused of livestock attacks, followed by the puma and coyote in much smaller percentages; the type of livestock most attacked was cattle, followed by

PAGE 10

10 goats, with cattle taken falling between the ages of 2 days and 12 months, while weights of cattle attacked ranged from 11.4 to 431.82 kg. I detected a preference towards male cattle, and, similar to other sites, most attacks occurred at night, and were more comm on during the rainy season, reaching a peak during the wettest months. A ma jor difference detected between other sites previously studied in the neot ropics (e.g., South America) were possible livestock attacks by coyotes in parts of the study site. Economic lo sses from carnivore atta cks on livestock were minimal, yet, because small cattle ranches are predominant in the area, the economic impact of each loss may be perceived as significant by the ranches impacted by these losses. Finally, landscape variables (e.g., forest cover area, distance to forest cover, and distance to rivers) were the best predictors of HCC s. Even though ranches did not employ practices considered beneficial to reducing carnivore attacks on livestock, ranc h characteristics and livestock husbandry practices were not important predictors of livestock attacks in this study because they were similar among most ranches.

PAGE 11

11 CHAPTER 1 INTRODUCTION Conservation Issue Hum an-carnivore conflicts (HCC) in the neotropics arise when large carnivores such as jaguars ( Panthera onca ), and pumas ( Puma concolor ) attack livestock. Conflicts with humans have been identified as the most important cause of adult mortality in large carnivore populations and could lead to their local extinction, even within protected areas (Woodroffe and Ginsberg, 1998). In most cases, livestock de predation involves an economic lo ss for cattle ranchers and the culling of the carnivore believed responsible for the attacks (Crawshaw, 2004; Zimmerman et al., 2005). Nevertheless, the impacts caused by carnivores on livestock is often overestimated. For instance, researchers have found th at jaguars are usually responsi ble for a small percentage of cattle losses (Hoogesteijn et al., 1993; Mazolli et al., 2002; Polisar et al., 2003; Michalski et al., 2006; Cascelli, 2008; Palmeira et al., 2008); similarl y Graham et al. (2004) determined that large predators are responsible for only 2 to 3% of all domestic animal mortality. However, because of a lack of strict cont rol and supervision of livestock, mortality incidents due to other causes are often attributed to large predators because their presence coincides with these losses (Gos line, 2004; Graham et al., 2004). Furthermore, when carnivores such as jaguars are hunted down, they are either killed or wounded; th e latter exacerbates the problem when wounded individual s are forced to look for eas y prey, such as livestock (Rabinowitz, 1986; Hoogesteijn et al ., 1993; Polisar et al., 2003). Consequently, efforts to identify and implement human-carnivore conflict mitigation strategies are urgent, especia lly on reserve borders and buffer zones where contact between humans and carnivores is more likely (W oodroffe and Ginsber g, 1998; Sunquist, 2002;

PAGE 12

12 Crawshaw, 2004). However, before taking such ac tions, it is important to examine spatial and temporal patterns of such conf licts in order to propose viab le and effective site-specific interventions (Treves et al., 2006). Determinants of Human-Carnivore Conflicts Hum an-carnivore conflicts may be influenced by prey depletion, lands cape structure, and livestock husbandry practices. HCCs generally occur in areas with a large forest-human interface, where human settlements are found adjacent to protected areas or large tracts of forest. Nyhus and Tilson (2004) found that human-tiger c onflicts occurred more often in forests with intermediate levels of disturba nce (e.g., multiple use forests) than in protected areas. The notion that livestock depredation by car nivores is more common in areas with low prey abundances dates as far back as Theodore Roosevelt (Rabinowitz, 2005) and ha s been discussed and studied in detail since then (Hoogesteijn, et al., 1993; Tr eves, et al., 2004; Siller o-Zubiri and Laurenson, 2001; Polisar et al., 2003; Bagchi and Mishra 200 6). Nevertheless, some studies suggest the opposite trend, a positive relationshi p between depredation rates and wild prey availability, due to higher predator densities in re sponse to an increase in prey densities (Stahl et al., 2002). With regards to landscape stru cture, livestock depredation incidents by jaguars and pumas have been found to be positively correlated with forest area and distance to human settlements, but negatively correlated with proximity to fo rest cover and water s ources (Saenz and Carrillo, 2002; Michalski et al., 2006; Cascelli and Murr ay, 2007; Palmeira et al ., 2008). Other studies on HCCs report a negative association of livestock depredation by carnivores and density of human roads and settlements (Treves et al., 2004). In most tropical forests, rudi mentary and non-techni cal livestock management practices are considered one of the principal factors leading to depredation incidents (Quigley and Crawshaw, 1992; Hoogesteijn et al., 1993; Polisar et al., 2003; Rabinow itz, 2005). Livestock in the

PAGE 13

13 neotropics is not adequately managed because most ranches are small and operate at very low costs (Saenz and Carrillo, 2002). Examples of rudimentary husbandry practices that increase the probability of livestock depredation incidents include k eeping free ranging and unattended cattle and other livestock near prime carnivore habitat (e.g., adequate forest c over and water sources). Additionally, adult males may be the only type of cattle that exhibit de fensive behaviors agains t predators (Sunquist and Sunquist, 1989), yet because most cattle ranches keep a higher proportion of females and calves, they inadvertently maintain higher numbers of mo re vulnerable cattle classes. Also, ranchers are not accustomed to regulating calving seasons; as a result, the most vulnerable cattle age class is available for depredation year round. Keeping livestock surrounded by low quality fencing and enclosures that can easily be penetrated by ca rnivores is also a principal factor favoring depredation incidents (Rabinowitz, 1986; Hoogest eijn et al., 1993; Hoogesteijn, 2001; Saenz and Carrillo, 2002; Polisar et al., 2003; Sognamillo et al., 2003; Conforti and Cascelli, 2003; Crawshaw, 2004; Graham et al., 2004; Zimmerma n et al., 2005; Michalski et al., 2006). Furthermore, studies examining livestock depr edation by jaguars and pumas have detected a positive correlation between carnivore attacks a nd bovine herd size (Michalski et al., 2006), and have found higher depredation rates during the peak calving seasons (Sognamillo et al., 2003; Polisar et al., 2003; Michalski et al ., 2006; Palmeira et al., 2008). Human-Carnivore Conflicts in Mesoamerica Most research on HCC s in the neotropics ha s been conducted in South America (Quigley and Crawshaw, 1992; Hoogesteijn 2001; Polisar et al., 2003; Conforti and Cascelli, 2003; Crawshaw, 2004; Graham et al., 2004; Zimmerma n et al., 2005; Michalski et al., 2006); in contrast, very few studies have examined H CCs in Mesoamerica (Rabinowitz, 1986; Saenz and Carrillo, 2002). As a result, lit tle is known about HCCs in Me soamerica and it is not clear

PAGE 14

14 whether livestock depredation is similar to other areas or if husbandry practices and landscape features affect probability of occurrence of human-carnivore conflicts in the same manner throughout their enti re distribution. The northern half of the Peten District, in Guatemala, along with adjacent protected areas in Belize and Mexico, forms part of the larges t continuous forest in Mesoamerica (Grunberg, 2000). Despite this, forest fragmentation in this district is increasing due to current migration trends (Grunberg, 2000; Hayes et al., 2002). Although cattle ranching is one of the principal forms of livelihood (Grunberg, 2000) and is very common in this area (Instituto Nacional de Estadistica, 2003), it has not been studied in deta il. There is no information on whether ranches and their livestock husbandry practices are similar throughout, or if they va ry according to size and capacity, and if the ranches apply measures designed to protect their livestock from wild carnivores. Therefore, it is not clear if a difference in landscape structure due to fragmentation, and livestock husbandry practices of ranches could be a principal factor affecting the occurrence of conflicts with carnivores in the area. I applied a survey questionnaire to cattle ranche rs in the Peten Distri ct. My objectives were to quantify and examine livestock depredation incidents by carni vores and test the hypothesis that the occurrence of HCCs in the study area was influenced by landscape variables, ranch characteristics, and livestock husbandry pr actices. To examine depredation incidents I determined type, age, weight and sex of livesto ck most frequently attacked, as well as the temporal patterns of the attacks and the tota l and average monetary loss due to depredation. Finally, I compared landscape variables, ranch ch aracteristics, and husbandry practices between ranches with and without carnivore attacks on livestock to identify the best predictors of HCC occurrence.

PAGE 15

15 CHAPTER 2 METHODS Study Area This study was carried out in the Peten District, Guatem ala, the largest and northernmost district of Guatemala. It cove rs an area of 36,000 square kilomete rs (Hayes et al., 2002) with a population of 366,735 people (Instituto N acional de Estadistica, 2003). Altitudes in Peten range from 100 to 300 mete rs above mean sea level (Grunberg, 2000), annual precipitation varies from 1,300 to 2,500 mm. (McNabb and Polisar, 2002), and mean temperatures range from 22 C to 34 C (Novack, 2003). The vegetation in Peten consists primarily of high canopy tropical lowland forest s, seasonally flooded lowland forests with a dense understory of shrubs and small trees, wetla nds along rivers and lakes, and flat savannahlike grasslands (Grunberg, 2000). Within the Peten District, we find severa l protected areas that are surrounded by human communities whose princi pal livelihood is cattle ranching (Figure 2.1). A survey carried out in 2003 estimated that the number of cattle ranches in the Peten District was between 5 and 7 thousand (Instituto Nacional de Estadistica, 2003). The largest protected area of Guatemala, the Ma ya Biosphere Reserve (MBR) is situated in the northern half of the Peten District. The M BR covers an area of over 2.1 million hectares (McNabb and Polisar, 2002) and is located within the broadlea f sub-tropical forest of the Atlantic lowlands, known as the Selva Maya (Maya Forest). The Maya Forest represents the largest area of subtropical forest in Mesoameric a, extending over 3 million hectares in Belize, Guatemala, and Mexico (Consejo Na cional de Areas Protegidas, 2001). Contained within the MBR is a combination of land use designations designed to provide for ecological protection and sustained human development. Protected areas of the MBR cover a noncontiguous area of 767,000 ha. The MBR also encompasses a multiple use zone of 848,440

PAGE 16

16 ha; the multiple use zone is an area divided into a series of comm unity managed forest concessions, where the sustainable extraction of timber and non-timber natural resources is permitted to local communities living within this zone. There is also a buffer zone of 497,500 ha, which is a 15 km wide strip of land situated al ong the southern edge of the MBR where land use is limited to activities compatible with the biod iversity objectives of the MBR (Consejo Nacional de Areas Protegidas, 2001). Another important series of pr otected areas is the Southern Protected Areas Complex of Peten. This complex is comprised of 11 protecte d areas which cover an area of 180,881 ha, plus a series of buffer zones surrounding each protected area, which total 270,011 ha. (AHT/PROSELVA, 2000). Reports of Livestock Dep redation by Carnivores To record o n livestock attacks, I surveyed cattle ranchers; I focuse d on attacks within the past 5 years (2003 to 2007). In addition, I gather ed information on livestock husbandry practices in ranches with and without reports of carnivor e attacks. Through office visits and workshops, I first surveyed Wildlife Service government offi cials and NGO personnel to identify ranches that had reported livestock depredation by carnivores. I then surveyed these ranches. After surveying these ranches, I applied the same questionnaire in as many adjacent ranches as possible. I increased my sample size by using a snowball tec hnique, in which I asked the surveyed cattle ranchers to provide information a bout other attacks of which they might be aware in the area (or other areas within the Peten Distri ct). I tried to survey an equal number of ranches with reports of attacks and ranches without attacks. At each ranch surveyed, I documented the ranch geographically through a GPS unit and followed the same process of inquiry. I asked if there had been any depredation incidents within the time frame of interest and, when an incident was reported, I asked a suite of questions about

PAGE 17

17 each incident that would allow for the evaluation of the reports reliabil ity and characterize the attack. Only first-hand reports of attacks were accepted and the reliability of each attack was assessed based on the following indications of an attack (Hoogesteijn et al., 1993; Hoogesteijn et al., 2002): Signs of struggle: blood on animal or surroundings or trampled vegetation. Visible wounds on animal: claw and bite marks, location of wounds (back of neck, skull or on throat), or animal found with neck twis ted, and certain parts of animal consumed. Description of site where animal was found: animal being dragged away from ranch, or hidden beneath leaves. For analyses, all second-hand and unreliable repo rts were discarded. In addition, all attacks were pooled regardless of the car nivore (jaguar, puma, or coyote) believed responsible for the attack, because it is difficult to distinguish between attacks by these predators (Palmeira et al. 2008). Patterns of Livestock Depredation For each rep orted attack, I determined the ca rnivore believed responsible for the attack, type of livestock attacked (e.g., cattle, goat, horse, pig, and dog), and time of day, month, and year of incident, when known. Hour of day wa s categorized as night (6:00 p.m to 6:00 a.m), morning (6:00 a.m to 12:00 p.m), and afternoon (12:00 p.m to 6:00 p.m), because in many cases the exact hour was not known. Fu rthermore, monthly data were categorized as occurring in the wet (June to January) or dry season (February to May) to test for seasonality. Chi-square ( X2) Goodness of Fit Tests were used to test the hypotheses that species of attacker (e.g., jaguar, puma or coyote), type of animal attacked (e.g., cattle, goats, horses, pigs, and dogs), time of day, and seasonality of attack s were equal in number for all categories; Fishers Exact tests were used when frequencies of responses were small (Zar, 1999).

PAGE 18

18 I performed Spearman Rank Correlations (rs) to examine associations between frequency of attacks per year and annual total precipitati on (mm) and between frequency of attacks per month and monthly rainfall. Rainfall data were obtained through CEMEC (Center for Ecological Evaluation and Monitoring), which is the Geogra phic Information Systems department of the National Council of Protected Areas (CONAP) in Guatemala. I also performed similar and more detailed analyses for cattle only, because cattle is the most abundant and important type of livestock in the area. Therefore, I recorded sex, age (in months), and weight (kilograms) of each animal attacked. Cattle weights were categorized into 9 different weight cla sses (11.4 to 450 kg.). Chi-square ( X2) Goodness of Fit Tests were used to test the hypothesis that attacks occurred on both sexes equally. Spearman rank correlations (rs) were used to test the relationship between number of attacks and age, and weight of animals attacked. Analyses for temporal patterns (e.g., hour and seas onality) of attacks on cattle were similar to those carried out for all attacks. A total and average annual monetary loss during the study period was obtained for all attacks pooled and for cattle separately by asking ranchers to estimate cost of animal attacked and converting their estimates to US dollars at an exchange rate of local currency of Q7.60 for US$1.00. All analyses were performed in SAS 9.1 (S AS Institute, 2003). Statistical significance was measured at P < 0.05 for all analyses. Livestock Husbandry Practices I recorded if ranches applied livestock husbandry practices that researchers (Hoogesteijn et al., 1993; Hoogesteijn, 2001; Polisar et al., 2003) have recommended to reduce probability of livestock depredation incide nts by large felids. These husbandry practices are:

PAGE 19

19 1. Secure enclosures. I measured height of fences where most cattle were kept, number of rows of fences, average spacing between rows and hei ght of last fence row. This last measure was based on the observation that th e last fence row presents a re latively larger separation from the ground than the other fence rows do between each other. All fences were made of the same material, wooden posts and barbed wire, th erefore type of fence or materials were not evaluated. Only one ranch had electric fenci ng, and this practice was not used in all the pastures of this ranch. 2. Intensity of guarding. This was measured by the presence of guard dogs, number of people in ranch and number of hours cat tle are watched per day. 3. Presence of maternity pastures at a distance from forest cover and mixing adult males in with calves and females. Both these practices help protect the most vulnerable age class of cattle (i.e., calves). Because adult males are larger and may defend themselves against predators (Sunquist and Sunquist, 1989), it is assumed their presence near calves would help protect them. I also evaluated ranch characteristics, and type of livestock control and care for each ranch surveyed. I assumed that larger ranches with high er number of livestock that applied a suite of livestock husbandry practices recommended by researchers (Quigley and Crawshaw, 1992; Hoogesteijn et al., 1993; Hoogesteijn, 2001; Polisar et al., 2003) would pr ovide better care and protection for their cattle. Descriptive characteristics evaluated included ranch size, number of cattle, and number of pastures in ranch. The following practices were also examined for each ranch: frequency of cattle rotation between pastures, frequenc y of veterinary care and whethe r ranches kept updated cattle records, and regulated calving seasons (this pr actice would imply a more technical control of breeding and the economic capacity for artificial insemination). The ranchers were given complete freedom as to which questions to answer; this limits the number of responses and da ta available per variable. Landscape Metrics The landscape attribu tes at and around each ranch (Table 2.1) were measured using Arc Toolbox in ARCGIS 9.0; all GIS layers where provided by CEMEC. I used vector layers for

PAGE 20

20 human settlements, rivers, bodies of water (i.e., lake s, lagoons, etc.), roads, and a raster layer for forest cover obtained from Land Sat images for the year 2003. I created a layer of the GPS coordinates of all the cattle ranches surveyed (with and without attacks). To calculate distance from each ranch to settlem ents, rivers, roads, and bodies of water, I joined the cattle ranch layer to the corresponding layer of which I wanted to obtain the metric. This function joins each ranch point to th e nearest attribute in th e layer being analyzed and calculates distance to that feature. To calculate forest cover and distance to fore st, I used Spatial Anal yst. I created a 5 km buffer around each ranch point and used the tabulat e area function to estim ate forest cover in each 5 km buffer. To estimate distance to forest c over, I used the Euclidean distance function. Variable Selection, Mode l Buildin g and Evaluation I performed a logistic regression analysis on a priori models based on three categories (application of husbandry practi ces designed to prevent large carnivore attacks, ranch characteristics and type of livesto ck control and care, and landscap e variables) of explanatory or predictor variables and used Aikaikes Inform ation Criterion (AIC) to compare models and identify the model or models that received the most support for explaining occurrence of livestock depredation by carnivores in cattle ranches. All anal yses were carried out in SAS 9.1 (SAS Institute, 2003). Although prey depletion is regarded as an important factor in fluencing livestock depredation, I did not consider it in my research because several studies (Polisar et al., 1998; Baur, 1998; Carrillo et al., 2000; Novack et al., 2005) have found a marked decrease in prey abundances around human settlements due to su bsistence hunting. Furthermore, Soto (2006) reports a perceived decrease of carnivore prey species around cattle ranc hes by local people in the Peten District. Because the ranches included in this study are all near human settlements

PAGE 21

21 (Figure 3-1), I assumed similar prey richness and abundances around them due to intense hunting pressure and therefore, little or no effect of this variable on lives tock depredation by carnivores. The models evaluated were the following: 1. Global model (combination of all ranch characte ristics, livestock husbandry practices, and landscape variables) 2. Ranch characteristics and type of livestock control and care combined with husbandry practices that protect against predators. 3. All husbandry practices related directly to protection of cattle from carnivore attacks (i.e., secure enclosures, intens ity of guarding, etc.). 4. Ranch characteristics and husbandr y practices indirectly related to livestock protection from predators, but explaining ranch type and leve l of cattle care and control (i.e., ranch size, number of cattle, veterinary care pr ovided, updated cattle records, etc.). 5. All landscape variables. I removed the variable presence of guard dogs from the analyses because only one ranch reported the use of guard dogs, and these guard do gs were not present during the study period. Models with many parameters are usually not well supported, unless sample size or effect size are large or if the residua l variance is small (Anderson, 2001). Therefore, I first conducted a Principal Components Analysis (PCA) followi ng McGarigal et al. (2000) on all continuous variables as a variable reducti on technique. The PCA allowed me to reduce correlated variables and retain variables that c ontribute more to the varia tion within my dataset. I used a criterion of r 0.5 to determine if there was a significant corr elation between covariates (Table 2.2). Five pairs of explanatory variable s were highly correlated ( r2 0.5) (Table 2.2). As a result I eliminated one variab le from each pair based on my criteria of which variable would contribute more to the analyses. I then analyzed the component loadings from each of the first five components selected by the scree plot. From this analysis, I retained 5 continuous variables. Thre e of the final variables

PAGE 22

22 selected were landscape metrics and the remainde r related to husbandry pr actices (Table 2.3). To these, I added the 3 categorical variables whic h could not be analyzed by a PCA (McGarigal et al., 2000). Consequently, I used 8 explanatory variables (2 continuous and 3 categorical husbandry practice variables, and 3 landscape variab les) to construct the final models (Table 2.4). I used Proc Logistic (SAS Institute, 2003) to evaluate the models using presence or absence of attacks in each ranch as my de pendent variable for the 5 different models. I used a Hosmer-Lemeshow Goodness of Fit test to examine the model fit ( 0.05) for each model proposed (Hosmer and Lemeshow, 2000). Models were then compared using Aikaikes Information Criter ion (AIC) (Burnham and Anders on, 2002). I transformed the AIC estimate obtained for each model into a second-order quasi-l ikelihood AIC (QAICc) following Burnham and Anderson (2002) because of a small sample size and possible over dispersion of the categorical variables. Over dispersion occurred because the categorical variables contained sparse data. In other words each variable had mostly outcomes of one t ype (Agresti, 2007). Variab les with sparse data can be identified by their 95% confidence intervals, which are either very large or infinite (i.e., CI = <0.001; >999.999). All categorical variables were identified as containing sparse data (Table 2.5, Figure 2.2) I also tested for over dispersion by dividing the deviance of the global model (55.92) by its degrees of freedom (41) using PROC GENMOD (SAS Institute, 2003). The value obtained from this analysis (1.36) gave me further reason to suspect over dispersion of these variables.

PAGE 23

23 Finally, I estimated the QAICc differences ( i) and the Aikaike weights ( wi) according to Burnham and Anderson (2002) to determine the best supported models: i = QAICi QAICmin (2-1) wi= R r r 1 i) 2 1 exp( ) 2 1 exp( (2-2)

PAGE 24

24 Figure 2-1. Map of the Peten District, 2007.

PAGE 25

25 Table 2-1. Landscape characteristics and livestock hus bandry practices obtained for each ranch in the Peten District, Guatemala, 2007. Variable Description Codes/ valuesVariable type Name 1 Attacks in ranch 0=No / 1=Yes Categorical ATT 2 Size of ranch hectares Continuous SR 3 No. of cattle per ranch Number Continuous HCAT 4 Fence height Meters Continuous FH 5 No. of fence rows Number Continuous FR 6 Average spacing between rows Centimeters Continuous FSR 7 Height of last row in fence Centimeters Continuous FSLR 8 No. of caretakers in ranch Number Continuous PP 9 No. of hours per day livestock is watched Hours per day Continuous HD 10 Guard dogs present in ranch 0=No / 1=Yes Categorical DG 11 No. of pastures per ranch Number Continuous PST 12 Frequency of cattle rotation Days Continuous FCR 13 How often do you update your cattle records? Months Continuous URC 14 How often do your cattle receive veterinarian care? Months Continuous FVC 15 Do you regulate calving season? 0=No / 1=Yes Categorical RCS 16 Do you have maternity pastures? 0=No / 1=Yes Categorical MP 17 Are males mixed in with females and calves? 0=No / 1=Yes Categorical MFC 18 Forest cover in 5 km2 buffer around ranch Squared meters Continuous FC 19 Distance to forest cover Meters Continuous DFC 20 Distance to rivers Meters Continuous DR 21 Distance to bodies of wa ter Meters Continuous DBW 22 Distance to human settlements Meters Continuous DHS 23 Distance to roads Meters Continuous DRD

PAGE 26

26 Table 2-2. Results of the Principal Components Analysis Corre lation Matrix, Peten, Guatemala, 2007. Correlated explanatory variab lesCorrelation coefficient Variable eliminated FR + FSR -0.77753 FSR HCAT + PST 0.68296 PST FSR + FSLR -0.59328 FSLR HCAT + FCR 0.68296 FCR DR + DRD 0.83281 DRD Only pairs of variables with a significant correlation ( r2 0.5) are shown. From each pair of correlated variables, one was kept for the final logistic regression analysis. Refer to table 2-1 for variable names. Table 2-3. Results of the Principal Co mponents Analysis for continuous landscape and livestock husbandry practice variables measured at each ranch in Peten, Guatemala, 2007. Component and 1 2 3 4 5 variance explained by each factor 3.4851 2.9779 2.1597 1.601 1.3278 Variable Component Loading SR -0.304 0.2198 0.147 0.079 0.63058 HCAT -0.2473 0.7944 0.1238 0.1115 -0.1655 FC 0.7424 0.0387 0.0478 -0.1918 0.1032 DFC -0.2375 0.1283 0.1728 0.6829 -0.3725 DR 0.4753 0.5749 0.5502 0.1322 -0.0206 Only variables selected from the factor pattern ou tput for the first five factors retained from the PCA are shown. The variable with the highest component loading wa s retained from each factor. Refer to table 2-1 for variable names. Table 2-4. Candidate models evaluated through logist ic regression, Peten, Guatemala, 2007. Model Number Model description Variables included in model 1 Global model SR HCAT RCS MP MFC FC DFC DR 2 All livestock husbandry practices SR HCAT RCS MP MFC 3 Livestock husbandry practices directly related to protection of livestock from carnivores MP MFC 4 Livestock husbandry practi ces indirectly related to depredation incidents (ranch characteristics and type of livestock control and care) SR HCAT RCS 5 Landscape variables FC DFC DR Refer to table 2-1 for variable names.

PAGE 27

27 0 5 10 15 20 25 30 35 40 45 50 R C SM PM F C Categorical variablesFrequency of response No YesTable 2-5. Ninety-five percent Wald confidence limits of the odds ratios obtained for the global model, Peten, Guatemala. Variable Confidence limits SR 0.998 1.002 HCAT 0.996 1.017 RCS* 0.093 114.554 MP* <0.001 >999.999 MFC* 0.003 34.401 FC 0.997 1 DFC 1 1.001 DR 1 1.001 *Large confidence limits indicate over dispersion of categorical variables. Refer to table 2-1 for variable names. Figure 2-2. Frequency of responses for categorical va riables obtained from each ranch surveyed in the Peten District, Guat emala (2007). Graph shows hi gh outcomes of one response type for all variables. Table 2-1 gives variable names.

PAGE 28

28 CHAPTER 3 RESULTS I surveyed a total of 83 ranc hes throughout the Peten Distri ct (Figure 3.1). Of these, 32 ranches reported 104 reliable livest ock depredation incidents within the tim e frame of interest. Size of ranches varied (mean SE = 175.9 41.62 ha, range = 11.2 2,688 ha, mode = 44.8 ha, n = 70) with a high percentage of small ra nches (11.2 to 500 ha) and very few mediumsized (500 to 1,000 ha) and large (>1,000 ha) ranches (Figure 3.2). Th e mean number of cattle per ranch was 120.51 17.57 ( SE), range = 5 750 cattle per ranch, and mode = 100 ( n = 69). Patterns of Livestock Depredation by Carnivores The jaguar was the carnivore accu sed of m ost attacks on livest ock (78.85% of all attacks, P < 0.0001, n = 104), while the puma and coyote ( Canis latrans) were accused of 15.38% and 5.77% of the attacks, respectivel y (Figure 3.3). Cattle was the type of livestock most attacked, followed by goats (Figure 3.3). Furthermore, most cattle were attacked by jaguars, while jaguars and pumas attacked an equal amount of goats and coyotes only attacked goats. (Figure 3.5) Most attacks occurred at night (P < 0.0001, n = 100) and during the wet season (P = 0.0004, n = 99). Attacks were correlated to monthly rainfall (rs = 0.590; P = 0.044) (Figure 3.4) but not to yearly rainfall (rs= -0.30; P = 0.624) (Figure 3.5). For cattle, most attacks occu rred at night (P < 0.0001, n = 55) and during the rainy season (P = 0.0003, n = 57). Attacks were not correlated to average monthly rainfall (rs =0.527, P = 0.079,) nor to average annual rainfall (rs = 0.10, P = 0.873). Attacks were negatively correlated to weight classes (rs = -0.815, P = 0.0074, 135.06 11.5 kg, range = 11.4 431.82 kg., n = 56) but not to age (rs = 0.114, P = 0.698, 6.59 0.597 months, range = 0.067 (1 day) 12 months, n = 46) (Figures 3.6 and 3.7, respectively). Most attacks were concentrated on cattle between the weights of 11.36 and 227.27 kg, with a high number of attacks occurring on cattle in the weight

PAGE 29

29 classes of 11.36 to 50 kg and 151 to 200 kg (28.57% and 23.21% respectively). The age class most killed was 9 to 10 months (45.65%), fo llowed by 1 day to 4 months old (30.43%), and finally 5 to 8 months (19.57%). Sex of cattle attacked was significant for males (P = 0.0009, n = 44). The total economical loss due to livestock de predation by large carni vores during the study period was estimated at $17,401, with an annual average estimated loss of $3,480 $1,526. Whereas for cattle, total reported loss was $14,736, while annual average loss was $2,947 $1,491. A total of 0.70% of all cattle reported in the study site was lost to carnivores in the five year span of the study. Determinants of Human-Carnivore Conflicts I used 44 ranches in the final analyses of suitable models due to incom plete data for livestock husbandry practices. Eighteen of these 44 ranches analyzed had reports of attacks. The only supported model was made up of landscape variables; the Akaike weight ( wi= 0.997, Table 3.4) for this model was almost comple te; furthermore, the model fit was accepted (P = 0.9484). This suggests that landscape variables (a mount of forest cover around ranch, distance to forest cover, and distance to rivers) are the only predictors of depred ation incidents, whereas ranch characteristics and livestock husbandry practices are not us eful for predicting occurrence of human-carnivore conflicts in the study area.

PAGE 30

30 Figure 3-1. Map of study area showing ranches surveyed and landscape metrics measured in the Peten District, Guatemala, 2007.

PAGE 31

31 Small ranches Medium ranches Large ranches 0 10 20 30 40 50 60Number of attacks Cattle GoatsHorses Dogs Animals attacked Jaguar Puma Coyote Figure 3-2. Sizes of cattle ranches surveyed in Pe ten, Guatemala, 2007. Where small ranches: 11.2 to 500 ha; medium ranches: 500 to 1,000 ha; and large ranches: >1,000 ha. Sizes were obtained directly from ranch owner or administrator. Figure 3-3. Livestock depredation incidents by jaguars pumas and coyotes in Peten, Guatemala from 2003 to 2007.

PAGE 32

32 0 2 4 6 8 10 12 14 16Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DecMonthsNumber of attacks0.00 50.00 100.00 150.00 200.00 250.00 300.00Average monthly rainfall Number of attacks Average rainfall 0 500 1000 1500 2000 2500 3000 20032004200520062007 YearTotal annual rainfall0 5 10 15 20 25 30 35 40 45 50Number of attacks Total yearly rainfall Number of attacks per year Figure 3-4. Livestock depredati on incidents by carnivores per m onth and monthly rainfall (mm). Rainfall data obtained from CEMEC in th e Peten District, Guatemala, 2003 2007. Figure 3-5. Livestock depredation incident s by carnivores per year and total annual rainfall (mm) in the Peten District, Gu atemala from 2003 to 2007. Rainfall data obtained from CEMEC.

PAGE 33

33 0 2 4 6 8 10 12 14 16Frequency of attacks 11.36-50 51-100 101-150 151-200 201-250 251-300 301-350 351-400 401-450Weight classes (Kgs.) 0 2 4 6 8 10 12 14 16 18Number of attacks 0.070.511.332456788.591012 Age (months) Figure 3-6. Cattle depredation incidents by weight classes of animals attacked in the Peten District, Guatemala from 2003 to 2007. Figure 3-7. Cattle depredation incidents according to age (months) of animals attacked in the Peten District, Guatemala from 2003-2007.

PAGE 34

34 Table 3-4. Comparison of five candidate mode ls in the Peten District, Guatemala, 2007. Model K Hosmer-Lemoshow model fit QAICc i wi 5 6 0.9484 73.62 0 0.997 1 11 0.836 85.22 11.59 0.003 4 6 0.9074 104.07 30.44 2.5 x 10-7 2 8 0.7362 111.24 37.62 6.7 x 10-9 3 5 1 111.89 38.26 4.9 x 10-9 K = number of parameters in mode l (includes an intercept term = 0, and the variance term = 2). The P value for the Hosmer-Lemoshow Goodness-of-Fit Test is evaluated at P < 0.05. QAICc = the AIC estimate corrected for small sample size and over dispersion. i= the difference for the QAICc values for the most supported model and the given model. wi = Akaike weight for each model. Refer to Table 2-4 for model descriptions.

PAGE 35

35 CHAPTER 4 DISCUSSION Patterns of Livestock Depredation As in m ost other areas, cattle loss due to ca rnivores was small (0.7% of all cattle reported in ranches), and may be negligible compared to losses due to other causes (Hoogesteijn et al., 1993; Mazolli et al., 2002; Polis ar et al., 2003; Michalski et al., 2006; Cascelli and Murray, 2007; Cascelli, 2008; Palmeira et al., 2008). Neve rtheless, when a cattle rancher owns a small number of head of cattle or other livestock, any loss is signi ficant (Saenz and Carrillo, 2002), forcing the ranchers to take re taliation measures and exacerbati ng a negative attitude towards carnivores. Therefore, even though the overall eff ect of carnivores on livestock appears minimal, it should be considered a problem that can lead to negative impacts on both local people and carnivore populations in areas with small cattle ranches, such as the Peten District. The jaguar was blamed for a high percentage of the attacks in this study (78.85%), followed by the puma and coyote; nevertheless, this should be interpreted with caution, because culprit of attack was not syst ematically identified in most cases and local knowledge and perception of these carnivores may have influenc ed these results. Althou gh coyote presence has not been verified in the area, there are anecdot al accounts of its expansio n into Peten, which may imply an increase in conflicts with this carnivore Furthermore, attacks could also be the act of feral or domestic dogs (Sillero and Laurenson, 20 01). Consequently, it is important to educate ranchers in predator damage identification and to establish unequivocally which carnivore species are truly responsible for the attacks in order to propose species-specific mitigation strategies and understand the true impact of jaguars on livestock. Cattle were reported as the type of livestoc k most attacked, although goat farming is only now becoming popular in the area (pers. obs.) an d a small number of ranches owned goats (n =

PAGE 36

36 5), in contrast to cattle being present in all the ranches; therefore, patterns of type of livestock attacked may reflect availabili ty. Mazolli et al. (2002) found a higher rate of depredation on goats and sheep than on cattle by pumas. Goats are an easy prey item that may provide suitable energy intake for carnivores, at least for pumas, as they have been known to take smaller prey than jaguars (Scognamillo et al., 2003; Palmeira et al., 2008; Cascelli, 2008). The coyote is also known for taking smaller livestock such as sheep and goats (Knowlton et al., 1999). As a result, where goats are present and not adequately protected, they may be targeted by carnivores more frequently, hence an increase in goat farming in th e region may lead to an increase in HCCs. All carnivore attacks on cattle we re concentrated on animals th at weighed between 11.4 to 431.82 kg., although 94.64% of all attacks occurred on cattle that weighed between 11.4 and 227 kg. Cattle attacked were between th e ages of 2 days to 1 year old, with a peak at 10 months old (37% for this age class). These results are c onsistent with studies that have found higher depredation rates on younger age classes of cattle and smaller lives tock (Hoogesteijn et al., 1993; Polisar et al., 2003; Michalski et al., 2006; Palmeira et al., 2008). For example, Palmeira et al. (2008) found higher depredation rates on newborn cattle and very few attacks on cattle older than 8 months old. Similarly, Michalski et al. (2006) found a higher percentage of attacks on cattle between 0 to 5 months old. Nevertheless, both studies report a sm all number of attacks on cattle older than 12 months, whereas, the highest age class attacked I report is 12 months (only two attacks on this age class). Ja guars and pumas may take larger cattle in these study sites, because these felids have been f ound to be larger in South Amer ica (McNab, 1971; Iriarte et al., 1990) than in Central America. These results impl y that some livestock depredation patterns will vary according to site. Nevertheless, these result s should also be interpre ted with caution because ranchers do not keep records of mortality inci dents and the weights a nd ages reported depend on

PAGE 37

37 ranchers memory and estimates. This may explain the discrepancy between the analyses of weight and age classes attacked (a negative corr elation between weight classes and number of attacks and no correlation between age classes and number of attacks). Furthermore, I also detected a significantly higher number of atta cks on male calves. Palmeira et al. (2008) also found this pattern and consider this a possible effect of the male calves being more independent from their mothers. With regards to temporal patterns of attacks, most attacks in my study site occurred at night and during the rainy season. Researchers speculate that attacks may be higher during wetter months because weather conditi ons are adverse for human activity (Mazolli et al., 2002), limiting the amount of time humans are present near thei r cattle. Furthermore, carnivores may be harder to detect during high rainfall periods by both humans and livestock. In addition, jaguars and pumas have been found to be more active during wetter months when temperatures are lower and prey are not concentrated around permanent water sources (Scognamillo et al., 2003). Consequently, the increase in de predation incidents during the rainy season ma y be due to higher activity patterns of carnivores, a more disper sed distribution of prey and lower detection probabilities of carnivores by humans and livestock. An increase in livestock depredation incidents during wetter months ha s been reported in other studies (Saenz and Carrillo, 2002; Mazolli et al., 2002). Palmeira et al. (2008) also found a higher depredation rate during the wetter months, yet th ey relate this to the majority of calf births occurring during this period. Michalski et al. (2006) found th e opposite, an increase in attacks during drier months, mostly re lated to calving months. Other studies have also found a correlation between depredation and calving peaks (Sognamillo et al., 2003; Polisar et al., 2003). Since most ranches in our study site do not re gulate birth seasons, making calves available

PAGE 38

38 almost all year round for carnivore s, it is not likely that attack peaks ar e related to calving season, suggesting that they may be more relate d to weather conditions and seasonal differences in carnivore ecology. Nevertheless, birth rates should be monitored in conjunction with livestock depredation, as well as seasonal variations in carnivore ecology, and calf and prey abundance and distribution to determine whic h factors may contribute more to the observed pa ttern of more attacks during the wetter months. Determinants of Human-Carnivore Conflicts Results from this study indicate that lands cape variab les were the only predictors of livestock depredation inci dents and I did not detect an infl uence of ranch characteristics and livestock husbandry practices on HCCs. The landscape variables measured that had an effect on occurrence of depredation incidents were amount of forest cover surrounding each ranch, and distance to forest cover and rivers. As a result, these variables could be used to construct predictive models to identify sites and ranche s most prone to depredation incidents. These findings are consistent with other studies (Michalski et al., 2006; Cascelli and Murray, 2007; Palmeira et al., 2008) that have also identified landscape variables such as proximity to forests and water sources and fore st area surrounding ranche s as major predictors for conflicts with carnivores. Th ese variables were found to be im portant predictors because they also describe prime carnivore habitat and sites where there is a high probability of carnivore presence. Although my study found that livestock husbandry practices do not pr edict occurrence of HCCs, they should not be discarded as importa nt determinants of liv estock depredation by carnivores. Low variance and unde r dispersion of ranch variables indicated that most ranches were similar and applied the same husbandry prac tices. Consequently, thes e variables were not important predictors of livestock depredation, not because they do not influence HCCs, but

PAGE 39

39 because low variability influenced the outcome of the statistical analyses. These results imply that small ranches that apply the same management regimes to their livestock, which may predispose livestock to depredation by carnivo res, predominate in the areas surveyed. More important is the conclusion that mo st ranches did not implement practices recommended by researchers to prevent carnivor e attacks, although lives tock depredation has been recorded. This may occur because ranchers do not have the knowledge or capacity to modify their practices, or because of a negati ve attitude towards carnivores and conservation. Several authors report th at local people are reluctant to modify their husbandry practices to prevent livestock depredation (Oli et al., 1994; Weber and Rabinowitz, 1996; Mazolli et al., 2002). Additionally, livestock depredation is not a regularly recurring problem, which does not warrant a significant investment to modify thei r traditional practices. Consequently, eliminating the animal believed responsible for the attacks ma y be a more effective and economic measure in the ranchers opinion. Finally, I make the caveat that results of this study should be interpreted with caution due to the small sample size and the predominance of small cattl e ranches in the study site. Nevertheless, my results coincide with observations from previous researchers (Quigley and Crawshaw, 1992; Hoogesteijn et al., 1993; Saenz and Carrillo, 2002; Polisar et al., 2003; Rabinowitz, 2005; Michalski et al., 2006) that state that most cat tle ranches in the neotropics apply rudimentary livestock husbandry practices that predispose cattle to depredation by carnivores and that landscape variables such as pr oximity to forest cover are important factors for predicting occurrence of human-carnivore conflicts. Conservation and Management Implications Exam ining livestock depredation patterns a nd the influence of la ndscape structure and livestock husbandry practices on HCCs will help us propose conflict mitigation strategies for

PAGE 40

40 the area. Determining livestock most vulnerable to carnivore attacks and s easonality of attacks, among other depredation patterns, will guide the implementation of conflict prevention measures. Furthermore, by understanding which factors predict occu rrence of livestock depredation by carnivores, we can identify in which sites and ra nches to concentrate conflict mitigation strategies. HCCs in the neotropics have been intens ively studied in South America and mostly limited to conflicts with jaguars and pumas. Fe w studies that examine spatial and temporal patterns of HCCs in Mesoamerica exist. Informa tion for this area is needed because of socio economical and ecological differences with sites previously studied. As an example of this, values for ranch size and number of cattle per ranch in this study were much smaller than ranches studied in Brazil and Venezuela (Polisar et al., 2003; Michalski et al., 2006; Palmeira et al., 2008). As a result, livestock husbandry prac tices may vary, with some South American ranchers having the capacity to implement recomm ended practices such as artificial insemination to control birth periods (Palmeira et al., 2008) an d introducing water buffaloes into their herds (Hoogesteijn and Hoogesteijn, 2007), whereas it is hi ghly unlikely that in Ce ntral America this is the case. This study gives us a first detailed insight in to human-carnivore conflicts in Mesoamerica. Similarities and differences were found with st udies from SA; the differences found imply that conflict-mitigation strategies recommended for sites previously studied need to be altered in order to be effective in Mesoamerica. The trend of smaller ranches with low tec hnical capacity was noted by Saenz and Carrillo (2002) for Costa Rica, so this s hould be considered an expected result. Therefore, large, well managed ranches are an exception for most parts of Central America. This suggests that the

PAGE 41

41 capacity of ranches to improve their operations and implement practices that will prevent livestock depredation is very limited and s hould be considered when proposing management modifications. Furthermore, I found that very fe w ranches implement practices that may prevent carnivore attacks. A considerable challenge will be to work with a large number of small cattle ranchers in each site, as opposed to working with a few, large ranc hes with better economic capacity. My results indicate that human-carnivore c onflict mitigation strategies should be implemented in ranches near forest cover and wa ter sources. Livestock prot ection efforts in these ranches should concentrate on cattle younger than 1 year that weigh less than 431.82 kg., especially during the wetter months. This can be done through regulati ng birth seasons when possible and the practice of maternity pastures located at a distance from forest cover, near human presence and with better designed fences (Mazolli et al., 2002; Palmeira et al., 2008). Finally, an important difference found between livestock depredation in Mesoamerica and South America is the possibility of conflicts with coyotes. Because the large forest tract of the Darien in Panama may be acting as a barrier for coyotes (Hidalgo-Miha rt et al., 2004), this carnivore has not invaded South America, where most research on neotropical human-carnivore conflicts have been carried out. Most studies fr om South America deal with depredation caused by jaguars and pumas and not with coyotes. Theref ore, the presence of conflicts with jaguars, pumas and coyotes occurs only in Mexico and Ce ntral America where the distributions of these three carnivores overlap. Coyotes are very intense livestock predators (Knowlton et al., 1999; Sillero-Zubiri and Laurenson, 2001) and the cont rol and prevention of coyote depredation on livestock requires species-speci fic measures (Knowlton et al., 1999). These measures have not been included in researchers recommendations to mitigate human-carnivore conflicts in the

PAGE 42

42 neotropics; therefore, it is cruc ial to find depredation control met hods that would be effective for preventing livestock attacks by jaguars, pumas and coyotes.

PAGE 43

43 LIST OF REFERENCES Agresti, A., 2007. An Introduction to Categorical Data Analysis, 2nd. edition. John Wiley & Sons, Inc., New York, NY. AHT/PROSELVA, 2000. Diseo de un Sistema de Monitoreo y Evaluacin de Indicadores Biolgicos para las reas Protegidas del Sur de Petn. Sant a Elena, Peten. Unpublished report to CONAP. Anderson, D.R., Link, W.A., Johnson, D.H., Burnha m, K.P., 2001. Suggestions for presenting the results of data analyses. Journal of Wildlife Management 65, 373 378. Bagchi, S., Mishra, C., 2006. Living with large car nivores: predation on livestock by the snow leopard ( Uncia uncia ). Journal of Zoology 268, 217. Baur, E.H., 1998. Final Report: study of subsiste nce hunting in the forestry concession of Carmelita, San Andres, Peten. Unpublished report to Propeten/Conservation International. Burnham, K. P., Anderson, D. R., 2002. Model se lection and multimodel inference: A practical information-theoretic approach, 2nd. edition. Springer-Verlag, New York, NY. Carrillo, E., Wong, G., Cuaron, A.D., 2000. Monitoring mammal populations in Costa Rican protected areas under different hunting re strictions. Conservation Biology 14, 1580-1591. Cascelli, F.C., Murray, D.L., 2007. Evaluation of factors predispos ing livestock to predation by jaguars. Journal of Wildlife Management 71, 2379 2386. Cascelli, F.C., 2008. Food habits and livestock depredation of sy mpatric jaguars and pumas in the Iguacu National Park Area, Sout h Brazil. Biotropica, doi:10.1111/j.17447429.2008.00404.x. Conforti V. A., Cascelli, F., 2003. Local perceptions of jaguars ( Panthera onca) and pumas ( Puma concolor) in the Iguacu National Park area, south Brazil. Biological Conservation 111, 215-221. Consejo Nacional de reas Protegidas CONAP 2001. Plan maestro de la Reserva de la Biosfera Maya 2001-2006. Serie: Co-edici ones tcnicas No. 3 0. CONAP Guatemala. Crawshaw, P., 2004. Depredation of domestic animals by large cats in Brazil. Human Dimensions of Wildlife 9, 329-330. Gosline, A., 2004. Crying Wolf over pred ator attacks. New Scientist 183, 10. Graham, K., Beckerman, A.P., Thirgood, S., 2004. Human-predator-prey conflicts: ecological correlates, prey losses an d patterns of management. Biological Conservation 122, 159171.

PAGE 44

44 Grunberg, W., 2000. Modelling deforest ation risk in the Maya Bios phere Reserve, Guatemala. Masters thesis. University of Arizona. Hayes, D.J., Sader, S.A., Schwartz, N.M., 2002. An alyzing a forest conver sion history database to explore the spatial and temporal characte ristics of land cover change in Guatemalas Maya Biospehre Reserve. Landscape Ecology 17, 299-314. Hidalgo-Mihart, M.G., Cantu-Sa lazar, L., Gonzalez-Romero, A., Lopez-Gonzalez, C.A., 2004. Historical and present distribution of coyote ( Canis latrans) in Mexico and Central America. Journal of Biogeography 31, 2025-2038. Hoogesteijn, R., Hoogesteijn, A., Mondolfi, E., 1993. Jaguar predation and conservation: cattle mortality caused by felines on three ranches in the Venezuelan llanos, in: Dunstone, N. & Gorman, M.L., (Eds.), Mammals as Predator s, Symposium of Zoological Society London 65, 391-407. Hoogesteijn, R., 2001. Manual on the problem of depredation caused by jaguars and pumas on cattle ranches. Unpublished report to th e Wildlife Conservation Society. Hoogesteijn, R., Boede, E.O., Mondolfi, E., 2002. Observaciones de la depredacin de bovinos por jaguares en Venezuela y los programas gubernamentales de control, in: Medelln, R.A., Chetkiewicz, C.L.B., Rabinowitz, A., Redford, K.H., Robinson, J.G., Sanderson, E.W., Taber, A.B. (Eds), El jaguar en el nuevo milenio: una evaluacion de su estado, deteccion de prioridades y recomendaciones para la conservacion de los jaguares en America. Universidad Nacional Autonoma de Mexico/Wildlife Conservation Society, Mexico. Hoogesteijn, R., Hoogesteijn, A., 2007. Ca n water buffalo presen ce facilitate jaguar conservation in the Neotropics? Cat News 46. Hosmer Jr., D. W., Lemeshow, S., 2000. Applied Logistic Regression, 2nd. edition. John Wiley and Sons, New York, NY. Instituto Nacional de Estads tica, 2003. Technical report: Censo Agropecuario Nacional, Guatemala. Irriarte, J.A., Franklin, W.L., Johnson, W.E., Redford, K.H., 1990. Bioge ographic variation of food habits and body size of the Am erican puma. Oecologia 85, 185-190. Knowlton, F.F., Gese, E.M., Jaeger, M.M., 1999. Coyote depredation control: an interface between Biology and Management. Journa l of Range Management 52, 398-412. Mazolli, M., Graipel, M.E., Dunstone, N., 2002. Mountain lion depredation in southern Brazil. Biological Conservation 105, 43-51.

PAGE 45

45 McGarigal, K., Cushman, S., Stafford, S., 2000. Multivariate Statistics for Wildlife and Ecology Research. Springer-Verlag, New York, Inc. McNab, B.K. 1971. On the ecological signifi cance of Bergmanns rule. Ecology 52, 845-854. McNab, R., Polisar, J., 2002. Una metodologa part icipativa para una estimacin rpida de la distribucin del Jaguar ( Panthera onca ) en Guatemala, in: Medelln, R.A., Chetkiewicz, C.L.B., Rabinowitz, A., Redford, K.H., Robinson, J.G., Sanderson, E.W., Taber, A.B. (Eds), El jaguar en el nuevo milenio: una evaluacion de su estado, deteccion de prioridades y recomendaciones para la c onservacion de los jaguares en America. Universidad Nacional Autonoma de Mexico/W ildlife Conservation Society, Mexico. Michalski, F., Boulhosa, R.L.P., Faria, A ., Peres, C.A., 2006. Human-wildlife conflicts in a fragmented Amazonian forest landscape: de terminants of large felid depredation on livestock. Animal Conservation 9, 179-188. Novack, A., 2003. Impacts of subsistence hunting on the foraging ecology of jaguar ( Panthera onca ) and puma (Puma concolor ) in the Maya Biosphere Rese rve, Guatemala. Masters Degree Thesis, University of Florida. Novack, A.J., Main, M.B., Sunquist, M.E., Labisky, R.F., 2005. Foragin ecology of jaguar ( Panthera onca ) and puma (Puma concolor ) in hunted and non-hunted sites within the Maya Biosphere Reserve, Guatemala. Journal of Zoology 267, 167-178. Nyhus, P., Tilson, R., 2004. Characterizing human-tiger conflict in Sumatra, Indonesia: implications for conservation. Oryx 38, 68-74. Oli, M.K., Taylor, I.R., Rogers, M.E., 1994. Snow Leopard ( Panthera uncia) predation on livestock: an assessment of local perceptions in the Annapurna Conservation Area, Nepal. Biological Conservation 68, 63-68. Palmeira, F.B.L., Crawshaw,P.G., Haddad, C.M., Ferraz, K.M., Verdade, L.M., 2008. Cattle depredation by puma ( Puma concolor ) and jaguar (Panthera onca) in central-western Brazil. Biological Conservation 141, 118-125. Polisar, J., McNab, R.B., Quigley, H., Gonzal ez, M.J., Cabrera, M. 1998. A preliminary assessment of the effects of subsistence hunti ng in the Maya Biosphe re Reserve: Part 1. Progress report: Game populations inTikal National Park and Uaxactun. Unpublished report to the Wildlife Conservation Society. Polisar, J., Maxit, I., Scognamillo, D., Farell, L., Sunquist,M.E., Eisenberg, J.F., 2003. Jaguars, pumas, their prey base, and cat tle ranching: ecological inte rpretations of a management problem. Biological Conservation 109, 297-310. Quigley, H. B., Crawshaw Jr., P. G., 1992. A Conservation Plan for the Jaguar Panthera onca in the Pantanal Region of Brazil. Biological Conservation 61, 149-157.

PAGE 46

46 Rabinowitz, A. 1986. Jaguar predation on domestic li vestock in Belize. Wildlife Society Bulletin 14, 170-174. Rabinowitz, A.R., Nottingham, Jr., B.G., 1986. Ecology and Behaviour of the jaguar ( Panthera onca ) in Belize, Central America. Journal of Zoology 210, 149-159. Rabinowitz, A. 2005. Jaguars and livestock: liv ing with the worlds th ird largest cat, in: Woodroffe, R., S. Thirgood, & A. Rabinowitz (E ds.), People and wildlife, Conflict or coexistence. Cambridge University Press. Saenz, J., Carrillo, E. 2002. Jaguares depredad ores de ganado en Costa Rica: un problema sin solucin?, in: Medelln, R.A ., Chetkiewicz, C.L.B., Rabinowitz, A., Redford, K.H., Robinson, J.G., Sanderson, E.W., Taber, A.B. (E ds), El jaguar en el nuevo milenio: una evaluacion de su estado, deteccion de prioridades y recomendaciones para la conservacion de los jaguares en Ameri ca. Universidad Nacional Autonoma de Mexico/Wildlife Conservati on Society, Mexico. SAS Institute, 2003. SAS for Windows, Version 9.1. SAS Institute, Inc., Cary, NC. Scognamillo, D., Maxit, I.E., Sunquist, M., Polisar, J., 2003. Coexistence of jaguar ( Panthera onca ) and puma (Puma concolor ) in a mosaic landscape in the Venezuelan llanos. Journal of Zoology 259, 269-279. Sillero-Zubiri, C., Laurenson, M.K.., 2001. In teractions between carnivores and local communities: conflict or co-e xistence?, in: Gittleman, J. L., Funk, M.S., Macdonald, D., Wayne, R.K. (Eds.), Carnivore Conservation, Conservation Biology Series 5. Cambridge University Press, Cambridge. Soto, J.R., 2006. Final Report: Preliminary assessm ent of the impacts of jaguars and pumas on livestock in the Maya Biosphere Reserve, Peten, Guatemala. Unpublished report to the Wildlife Conservation Society. Stahl, P., Vandel, J.M., Ruette, S., Coat, L., Coat, Y., Balestra, L., 2002. Factors affecting lynx predation on sheep in the French Ju ra. Journal of Applied Ecology 39, 204-216. Sunquist, M.E., Sunquist, F.C., 1989. Ecological constraints on predation by large felids, in: Gittleman, J.L. (Ed.), Carnivore Behavior, Ecology and Evolution. Cornell University Press, New York. Sunquist,, M., 2002. Historia de la investigacin sobre el jaguar en el continente americano, in: Medelln, R.A., Chetkiewicz, C.L.B., Ra binowitz, A., Redford, K.H., Robinson, J.G., Sanderson, E.W., Taber, A.B. (Eds), El jaguar en el nuevo milenio: una evaluacion de su estado, deteccion de prioridades y recomendacione s para la conservaci on de los jaguares en America. Universidad Nacional Autonoma de Mexico/Wildlife Conservation Society, Mexico.

PAGE 47

47 Treves, A., Naughton-Treves, L., Harper, E.K., Mladenoff, D.J., Rose, R.A., Sickely, T.A., Wydeven, A.P., 2004. Predicting Human-carnivor e conflict: a spatial model derived from 25 years of data on wolf predation on livestock. Conservation Biology 18, 114-125. Treves, A., Wallace, R.B., Naughton-Treves, L., Morales, A., 2006. Co-managing humanwildlife conflicts: a review. Human Dimensions of Wildlife 11, 383-396. Weber, M., Rabinowitz, A.R., 1996. Global persp ectivas on large carnivore conservation. Conservation Biology 10, 1047-1054. Woodroffe, R., Ginsberg, J.R., 1998. Edge effect s and the extinction of populations inside protected areas. Science 280, 2126-21-28. Zar, J.H. 1999. Biostatistical analysis. 4th ed ition. Prentice Hall, U pper Saddle River, NJ. Zimmerman, A., Walpole, M.J., Leader-Willia ms, N., 2005. Cattle ranchers attitudes to conflicts with jaguar ( Panthera onca ) in the Pantanal of Brazil. Oryx 39, 406-412.

PAGE 48

48 BIOGRAPHICAL SKETCH Jose was born in the sm all town of Chimaltenango, approximately 54 kms from Guatemala City. When he was 8 years old, he and his fa mily moved to Boston, Massachusetts, where he lived and attended elementary school for 5 years. Shortly after gr aduating high school in Guatemala, he began to teach English, an occupa tion that he practiced for 15 years. Having been raised most of his life in the city; it was only duri ng a trip to the ruins of Tikal in Peten that he realized how rich and abundant th e natural resources of his country were. He was fascinated with the idea that he was walking through a jungle wh ere such majestic anim als as jaguars roamed. This trip was pivotal in his de cision to pursue a degree in wild life biology. The next year he enrolled in the San Carlos University of Guatem ala, where he obtained hi s bachelors degree in biology. This is where he met his beautiful wi fe, Nancy, with whom he has a wonderful boy named Danny. Jose began working in wildlife res earch projects with th e Wildlife Conservation Society, Program for Guatemala in Peten in 2001. Since then he has worked in research and outreach projects involving endangere d neotropical vertebrates. The focus of his work has been on jaguars and subsistence hunting species, exam ining the effects of hunting on their abundance through line-transect sampling and camera tr apping censuses and analyzing human-jaguar conflicts. Jose plans to follow on with his doc toral degree in the Wildlife Ecology Department, and study neotropical carnivore ecology in Guatemala in more detail, and in the future he plans to return to his home country to apply and sh are the knowledge and experience he has obtained throughout his graduate st udies with the local conservation community.