Title: Jaguars, pumas, their prey base, and cattle ranching
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00100748/00001
 Material Information
Title: Jaguars, pumas, their prey base, and cattle ranching ecological perspectives of a management issue
Physical Description: ix, 239 leaves : ill. (some col.), maps (some col.) ; 29 cm.
Language: English
Creator: Polisar, John R
Publication Date: 2000
Copyright Date: 2000
Subject: Wildlife depredation -- Venezuela   ( lcsh )
Jaguar -- Venezuela   ( lcsh )
Puma -- Venezuela   ( lcsh )
Wildlife Ecology and Conservation thesis Ph. D   ( lcsh )
Dissertations, Academic -- Wildlife Ecology and Conservation -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )
Summary: ABSTRACT: Jaguar and puma depredation on livestock may be influenced by 1) innate and learned behavior; 2) health and status of individual cats; 3) division of space and resources among jaguar and puma; 4) cattle husbandry practices; 5) abundance and distribution of natural prey. Our study in Los Llanos Altos of Venezuela aimed to establish how all these inter-related elements were related to cattle being lost to cat depredation. Linear foot transects, vehicle transects, point counts, incidental observations, camera trapping, net, hoop trap, funnel trap, haul seine, box trap, and noose captures, and detailed vegetation sampling and mapping were employed to understand the patterns of prey distribution by species and available biomass. Prey distribution was influenced by forest composition, topographical characteristics, and degree of habitat interspersion. Climate, topography, and soils interact to define variation in primary productivity, dictating prey distributions, and large cats use space accordingly. The few preferred prey species were both large and productive. Large reptiles were used less than their high biomass would suggest, presumably a result of access and risk. The biomass of natural prey in the study area was adequate to support the resident large cats without a subsidy of domestic livestock. Selective rather than opportunistic hunting by the cats reinforced that conclusion. However, the distribution of natural prey was far from uniform. Puma were responsible for more attacks on livestock than jaguar, frequently in maternity pastures set in upland areas of relatively low prey availability. The mammalian biomass in the study area rivaled that of the most productive savanna/forest mosaics of the Old World. Up to 97% of that high biomass was represented by grazers introduced from the Old World, the majority being bovid livestock apparently occupying niches left vacant since the megafauna extinctions of the Pleistocene. The closing discussion of management recommendations focuses on Los Llanos Altos of Venezuela, but contains elements applicable to all the savanna/forest mosaics of South America where similar issues may arise.
Thesis: Thesis (Ph. D.)--University of Florida, 2000.
Bibliography: Includes bibliographical references (leaves 226-238).
Additional Physical Form: Also available on the World Wide Web; Adobe Acrobat Reader required to view and print PDF file.
Statement of Responsibility: by John Polisar.
General Note: Printout.
General Note: Vita.
 Record Information
Bibliographic ID: UF00100748
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 47941624
alephbibnum - 002675537
notis - ANE2744


This item has the following downloads:

polisar_final ( PDF )

Full Text







Copyright 2000


John Polisar


Grants supporting prey component field work came from The National

Geographic Society, Wildlife Conservation International (WCS), the British Embassy to

Venezuela's Cooperation Fund, and the Lincoln Park Zoo Scott Neotropic Fund.

Katharine B. Ordway Chair endowment funds made available by Dr. John F. Eisenberg

also supported the work. Eisenberg's support was manifold. The predator component of

this project generously shared a donation from the Cat Specialist Group of the Species

Survival Commission (IUCN-World Conservation Union). The WCS MesoAmerican

and Caribbean Program donated camera trap units. Faunal aspects of the study were

assisted in the field by Victor Juan Meires, Orlando Ramirez, Gilson Rivas Fuenmayor,

Diego Giraldo, Sandra Melman, Emiliana Isase, Telva Carantona, and Marcus Trepte.

Floral aspects were assisted by Rafael Ortiz and Dr. Francisco Delascio. Important field

observations were provided by Daniel Scognamillo, Ines Maxit, Laura Farrell, Roy

McBride, Rafael Hoogesteijn, Rocky McBride, Tibisay Escalona, Bruno Pampour, the

llaneros of Pifiero, and students and staff of Dr. Juhani Ojasti's wildlife management

class (UNELLEZ at Guanare). Francisco Bisbal and his staff at Rancho Grande provided

logistical support, equipment loans, and humor. Edgardo Mondolfi served as mediator

during storms. Don Antonio Julio Branger provided a large private area with primitive

roads that was mostly devoid of hunting, allowing excellent opportunities for animal

observations. Various staff at Pifiero provided humour and facilitated our endeavors.

Joanna Persis Hemmat and the U.S. Embassy to Venezuela also provided logistical

assistance and a stable operating base in Caracas.

Fortunately the field work provided abundant data. Many thousands of data

points needed to be condensed. All the people who helped process the data in Gainesville

did so in good humour and I hoped we converted tedium into adventure. Abundant

thanks go to all of the following. Matt Burgess was absolutely invaluable loading and

sorting transect data and assisted with a number of other files. His buddy Kyle Harris

also helped as I thought of additional contortions we could put the big files through.

Stephen Taranto helped with preparing vegetation data for analyses, efficiently coping

with some large files. Richard Owens also helped meld big vegetation files. Tom Hoctor

was invaluable as a liaison with the UF Department of Urban and Regional Planning's

GEOPLAN Center. He introduced me to Maria Fernanda Zermoglia and Rosanna

Rivero, both of whom worked hard in the final stages of converting the draft vegetation

map into an ARCVIEW reality, allowing spatial analyses. Financial support for

assistants at the University came from the WCS jaguar conservation fund, the U.S.

Federal Work-Study Program, and Katharine B. Ordway Chair endowment funds made

available by Dr. John Eisenberg. During all this my wife, Joanna Persis Hemmat,

covered for me. Many home tasks, whether banal, arcane, distasteful, tedious or

otherwise, she took care of, allowing me to focus on the dissertation. She deserves

substantial recognition for the many ways that she facilitated the project, during the

fieldwork, and during the analyses. My older sister, Lisa Polisar, also deserves

recognition, for taking care of some serious Polisar business during my graduate studies,

which allowed me to indulge these studies in distant places. Lisa took care of some deep

family responsibilities during my various projects abroad.

Thanks to the members of my doctoral committee who reviewed the dissertation.

They were John F. Eisenberg, Melvin Sunquist, F. Wayne King, Jim Nichols, and Fred

Thompson. Thanks also to numerous personnel at the Florida Museum of Natural

History for company and support.

When things got rough down in Venezuela, contemporary electric blues

paraphrased the state of affairs, and inspired me to sharpen my machete and cut trail to a

brighter day. The Pifiero project was sometimes one hell of a rodeo. It chafed at times,

and there was some hollering, but we stayed on top, rode it down, and took it to the finish




A C K N O W L E D G M E N T S ........................................................................ .....................iii

A B S T R A C T ...................................................................ii


INTRODUCTION ............................... ...................... ..........


In tro d u ctio n ................................................................................................ 1 1
M ethods.......................................... .............. 14
R e su lts ...................................... ........................................ ............... 1 7
Forest Types ................................................................ ... ..... ......... 17
F forest P henology ........................................... ..... .................... 20
Savannas .......... ......... ................................................................................. 2 1
V vegetation M ap ...... ...... .......................... ...................................... .... 23
D isc u ssio n .............................................................................................. ............... ..... 2 4

AND PUM A PREY ....................................................... ........... .. ............ 63

Introduction............................... ........... .......... 63
M eth o d s .................................................................................... . 6 4
Linear Foot Transects .................. ............................. .. ...... .. .......... .. 64
V vehicle Transects .................. ....................................... .. .......... 65
Capybara and Caim an Counts................................. .. .. ..... .................. 66
Capture-Mark-Release-Recapture: Tortoises, Turtles, Anacondas........................... 68
A additional Brief A ssessm ents ........................................................ ......... ..... 70
C am era T raps ........................................ 70
Results ................... .................... ........ ......... ...................71
Distribution of Animals According to Encounter Rates ....................................... 71
Vehicle transects .......................... ........................... ........ 71
Foot Transects and Cam era Trapping................................................................ 72
Seasonal Changes in Distributions, Densities, and Group Sizes of Prey .............. 79
A bundance and Biom ass ............................................................ ......... ...... 83

DISTANCE Density Estimates, DensityValidations, Distribution Validations ....... 90
Sorted E encounter R ates .................................................................................... ... 98
Standing Crop Biomass Estimates From Distance Density Estimates.................. 99
D discussion .............................. ........... ........... 102


Introduction ............................ ............... ..... 14 1
Study Area..................................... ..... .............. 145
M eth o d s ............... ................................................................................. ............... ..... 14 6
R e su lts ................................47.............................
Discussion....................... ............... ..... .............. 149

GENERAL AND HATO PINERO IN PARTICULAR.................................................175



SORTED BY: 1) FORM; 2) FAMILY; 3) GENUS AND SPECIES ............................198

C LIST OF FISH SAMPLED IN HATO PINERO. ................. .....................206

D INDIVIDUAL PASTURES OF HATO PINERO, ............................................. 207

C A L C U L A T IO N S ................................................................................................ 2 09

F FOREST CLASSIFICATIONS .................... ....... ...............213

LITERATURE CITED ................ .. ........ ................215

BIOGRAPHICAL SKETCH........................ ................228

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



John Polisar

December, 2000

Chairman: Dr. John F. Eisenberg
Major Department: Wildlife Ecology and Conservation

Jaguar and puma depredation on livestock may be influenced by 1) innate and

learned behavior; 2) health and status of individual cats; 3) division of space and

resources among jaguar and puma; 4) cattle husbandry practices; 5) abundance and

distribution of natural prey. Our study in Los Llanos Altos of Venezuela aimed to

establish how all these inter-related elements were related to cattle being lost to cat

depredation. Linear foot transects, vehicle transects, point counts, incidental

observations, camera trapping, net, hoop trap, funnel trap, haul seine, box trap, and noose

captures, and detailed vegetation sampling and mapping were employed to understand the

patterns of prey distribution by species and available biomass. Prey distribution was

influenced by forest composition, topographical characteristics, and degree of habitat

interspersion. Climate, topography, and soils interact to define variation in primary

productivity, dictating prey distributions, and large cats use space accordingly. The few

preferred prey species were both large and productive. Large reptiles were used less than

their high biomass would suggest, presumably a result of access and risk. The biomass of

natural prey in the study area was adequate to support the resident large cats without a

subsidy of domestic livestock. Selective rather than opportunistic hunting by the cats

reinforced that conclusion. However, the distribution of natural prey was far from

uniform. Puma were responsible for more attacks on livestock than jaguar, frequently in

maternity pastures set in upland areas of relatively low prey availability. The mammalian

biomass in the study area rivaled that of the most productive savanna/forest mosaics of

the Old World. Up to 97% of that high biomass was represented by grazers introduced

from the Old World, the majority being bovid livestock apparently occupying niches left

vacant since the megafauna extinctions of the Pleistocene. The closing discussion of

management recommendations focuses on Los Llanos Altos of Venezuela, but contains

elements applicable to all the savanna/forest mosaics of South America where similar

issues may arise.


Cattle production is a profitable and relatively non-destructive land use system in

the seasonally flooded lowland savannas of Venezuela, Colombia, Brazil, Bolivia, and

Guyana. On many large ranches, habitat modification is minimal and wildlife values are

high. However, a major source of mortality for jaguars (Panthera onca) and puma

(Puma concolor) is persecution by cattle ranchers who attribute livestock losses to the

cats. In some cases these losses to cats are very real (Hoogesteijn et al. 1993). In some

situations the problem has been exacerbated by poor herd management (Hoogesteijn et al.

1993), or indiscriminant shooting, which can result in disabled incipient "problem cats"

(Rabinowitz 1986). Cattle mortality due to flooding, disease, parasites, and malnutrition

may be high, but with adequate records rare, that issue is often obscured (Mondolfi &

Hoogesteijn 1986). In some areas, pumas may play a larger role in livestock losses than

jaguar (Farrell 1999; Gonzalez-Fernandez In Press; Scognamillo et al. In Press).

Rabinowitz (1986) found healthy adult jaguars reluctant to enter open pastures,

despite the presence of potential domestic prey. Domestic animals left untended in forest

were quickly dispatched by the same jaguars that avoided human habitations. In the

Pantanal, many cattle killed by cats were very lightly managed. In habitat use and

behavior they resembled wild prey, and during high water, were forced to use elevated

forest areas (Quigley & Crawshaw 1992; Schaller & Crawshaw 1980). Half the cattle

killed by cats in the same area were calves (Quigley & Crawshaw 1992). Given the

jaguar's propensity for closed forested habitats and areas near water, restricting calving to

open areas with few permanent streams seemed one means for decreasing depredation

(Quigley & Crawshaw 1992). The same applied for adult cows; keeping all cattle out of

forested habitats should reduce depredation (Quigley & Crawshaw 1992). An analysis of

cattle management methods and depredation problems on three Venezuelan ranches by

Hoogesteijn et al. (1993) suggested much of the same. Lower depredation rates were

hypothesized to result from: 1) exclusion of cattle from gallery forest; 2) adequate

distance between calving grounds, young calves, and forested areas; 3) pasturing problem

areas with bulls over a year old; 4) maintaining adequate populations of wild prey.

Shaw (1977) hypothesized that the number of cattle taken by puma in Arizona

was inversely proportional to the size of the deer herd. Mondolfi and Hoogesteijn (1986)

hypothesized a similar relationship for jaguar and puma in Venezuela, where the large

cats exploit a more diverse prey base. These speculations were vastly pre-dated by the

observation by Roosevelt (1914) that ranches in Brazil that possessed abundant native

prey experienced fewer jaguar problems. Eighty years later these ideas had yet to be

tested with data.

This dissertation represents part of one team's effort to examine the overlapping

variables contributing to the conflicts between jaguar, puma, their prey base, and cattle

production in tropical America. Throughout the bio-region in which our study was

located there are problems with attacks on livestock by both jaguar and puma (Gonzalez-

Fernandez In Press). Jaguar predation on cattle stranded on forest islands amidst flooded

savannas, a common situation in the Pantanal of Brazil, was rare in our particular study

area (Hato Pifiero). This was more a function of sound cattle management than evasion

of jaguar attacks. Adult cattle can forage in water but need a dry place to rest at night.

Calves cannot forage in water. Calf survivorship (and hence production and profits) are

increased when calving occurs in uplands during wet months. However, every action has

a reaction, and throughout the Llanos Altos, and in our study area in particular, the

greatest losses to predators were young calves lost to puma in maternity pastures, during

the rainy season.

Losses usually were not high. In the region, 70% of the ranches lost less than

0.4% of the herd annually (Gonzalez-Femandez In Press). Between 1991-1997 in Hato

Pifiero, 13.3% of all calf losses were attributable to jaguar (1.8%) and puma (11.5%)

(Scognamillo et al. In Press). In other ranches in the region, losses were not as skewed

(43% problems due to jaguar and 57% due to puma) but overall, the highest losses were

still calves taken by puma (Gonzalez-Femandez In Press). Ranches set in the evergreen

moist forests south of the Orinoco River experienced more problems with jaguars (Juan

La Vieri pers. comm.). Although the number of cattle lost are usually low, some cats

become habituated on cattle and local losses then become unacceptably high to ranchers.

In the region, a few ranches had lost between 2.5% and 5% of all calves born to felids

(Gonzalez-Femandez In Press). Such outbursts of chronic losses usually result in dead

cats and our study hoped to suggest means for reducing the problem. The study began

with the management issue of cat and cattle co-existence, but also addresses questions of

broader interest relating to resource competition, foraging theory, and the ecology of the

prey species themselves.

The range of the puma spans 110 in latitude (Culver et al. 2000). The species

essentially covered the Americas (now extirpated in some regions), from far north to

extreme south, but it has yet to appear in southern Alaska. In North America, large

ungulates, primarily deer, constitute 68% of the puma's diet by frequency of occurrence;

large rodents, lagomorphs, small carnivores, and where present, armadillos, constitute the

remainder (Iriarte et al. 1990). Evidence from numerous North American locales

(Anderson 1983; Dalrymple & Bass 1996; Hornocker 1970; Kunkel et al. 1999; Logan et

al. 1996; Maehr 1997a; Murphy 1998; Shaw 1977) convinced many workers that the

puma specializes on deer. The breadth of puma diet appears to have an inverse

relationship with latitude due to greater diversity in the tropics. While the food habits of

puma have been well-studied in North America and in Chile (Iriarte et al. 1990), in the

tropics the subject began as a minor component in jaguar studies (Emmons 1987; Quigley

& Crawshaw 1992; Rabinowitz & Nottingham 1986). Very recently, more data have

become available. Aranda and Sanchez-Cordero (1996) made observations on jaguar and

puma coexistence in southern Mexico based on 37 jaguar scats and 15 puma scats. Taber

et al.(1997) examined the same issue in the Paraguayan Chaco based on 106 jaguar scats

and 95 puma scats. Nufiez et al. (In Press-a) employed a more comprehensive

methodology, using telemetry to study jaguar and puma in west-central Mexico. Their

conclusions on food habits were based on 50 jaguar scats and 19 jaguar kills and 65 puma

scats and 26 puma kills (Nufiez et al. In Press-b).

In the Peruvian Amazon, puma appeared to use more smaller prey than jaguar

(Emmons 1987). The same held true in intact habitats in the northern Paraguayan Chaco

(Taber et al. 1997) and in dry forests in western Mexico (Nufiez et al. In Press-b). In the

Pantanal of Brazil, puma attacks on cattle was more skewed towards calves than those by

jaguar (Crawshaw & Quigley In Press). The puma is a highly adaptable species whose

size and ecology can vary among habitats and latitudes.

Unlike the puma, which is found across a vast stretch of habitats and latitudes, the

jaguar is restricted to the subtropics and tropics. In the northwestern United States, where

the jaguar is not present, puma are larger than in the tropics and regularly prey on young

elk (Cervus elaphus) and adult mule deer (Odocoileus hemionus) (Hornocker 1970;

Iriarte et al. 1990; Logan et al. 1996; Murphy 1998). Mule deer weights are double those

of capybara and tropical white-tailed deer and triple those of white-lipped peccary

(Anderson & Wallmo 1984). In Alberta, Canada, male puma, averaging 75 kg (larger

than many tropical forest-dwelling jaguars), take moose (Alces alces), including 250 kg

yearlings (Ross & Jalkotzy 1996). As mule deer populations dropped in New Mexico

during the late 90s, puma turned to bighorn sheep (Ovis canadensis) (Eric Rominger

pers.comm.). Chilean puma are also larger than tropical puma, and pursue some large

prey, though less than the northern puma (Iriarte et al. 1990). The puma's adaptation to a

staggering variety of habitats across latitudes has demanded plasticity in diet. Size

differences between leopards and tigers (weight ratio 1:4) are more extreme than between

puma and jaguar (Seidensticker 1976), but echoes of their relationship may be found

where the New World large cats co-occur. Perhaps puma take a higher frequency of

smaller prey when in the company of the larger bodied jaguar? Even this postulate must

account for observed variation. In Hato Pifiero, the average jaguar weight was 70 kg and

that of puma 41 kg. On average Pifiero jaguar were 1.7 times the size of Pifiero puma. In

west-central Mexico, where perhaps jaguar were "making do" in puma habitat, jaguar

were 35-55 kg and puma 25-50 kg, approximately the same size (Nufiez et al. In Press-a;

In Press-b).

Jaguar appear to use waterside habitats when they are available more than puma

(Emmons 1987). In such habitats, the jaguar may eat freshwater turtles, caiman, fish, and

even sea turtles (Carrillo et al. 1994; Carrillo & Saenz In Press; Emmons 1987; 1989;

1991; Hoogesteijn & Mondolfi 1993). Large aquatic and terrestrial reptiles figured

heavily in the diet of Emmon's study animals, leading her to speculate that the large head

and great bite force of the jaguar could be an adaptation for breaking through the hard

integuements of large reptiles (Emmons 1987; 1989). Aranda (1994) also speculated on

the origins of the powerful physique and bite force of jaguars. However, his study area

lacked large areas of surface water, and thus large reptiles, and the jaguar's morphology

seemed well-suited to killing peccaries, which, though not large (23-35 kg) do have fairly

formidable defenses. The current and Recent ranges of Tayassu tajacu and Panthera

onca do coincide (Aranda 1994).

In the Peruvian Amazon, puma did not prey on turtles and crocodilians (Emmons

1987; 1989). In habitats of the Florida Everglades too low in elevation to be prime puma

habitat, 11.1% of florida panther (Puma concolor coryi) kills were alligators (Alligator

mississippiensis) (Dalrymple & Bass 1996). In the upland habitats that the panther

prefers (Maehr 1997a), and which are more productive for ungulates and panthers, feral

hogs (Sus scrofa) and white-tailed deer were more dominant food items. Alligator

consumption became negligible (Maehr 1997a).

In the xeric habitats of western Mexico and the Paraguayan Chaco, dietary

differences between jaguar and puma were subtle (Nufiez et al. In Press-b; Taber et al.

1997). Yet, in western Mexico, the jaguar did show slightly more preference for collared

peccary than did puma (Taber et al. 1997). In the Yucatan Peninsula of Mexico collared

peccary were the most important prey for jaguar, while brocket deer (Mazama

americana) and white-tailed deer (Odocoileus virginianus) were the most important prey

for puma (Aranda & Sanchez-Cordero 1996). In the Peruvian Amazon, jaguars took

peccaries more than expected from known densities (Emmons 1987). In the Pantanal of

Brazil, jaguar slightly preferred peccary over capybara, while puma took more capybara

than any other natural prey (Crawshaw & Quigley In Press). In perhaps the most

comprehensive study of jaguar and puma coexistence to-date, Nufiez (In Press-b). found

7 prey items in 50 jaguar scats and 11 prey items in 65 puma scats. The habitat in their

west central Mexico study area (90% low deciduous dry forest) may be better suited to

puma than jaguar in the same way that the lower areas of the Brazilian Pantanal might be

better suited for jaguars than pumas (Crawshaw & Quigley In Press). Including kill data,

Nufiez (In Press-b) found more diversity in puma diet than in jaguar diet. Although both

species may be flexible (Rabinowitz & Nottingham 1986), the puma would be expected

to be more of a generalist, particularly in the presence of a social dominant (Seidensticker


In the rain forest of Belize, the three most important prey species for jaguar

(ranked in descending order by frequency of occurrence) were armadillo (Dasypus

novemcinctus), paca (Agouti paca), and collared anteater (Tamandua mexicana)

(Rabinowitz & Nottingham 1986). The armadillo, a relatively small animal, represented

54% of all prey identified. There were a total of 16 prey species from 228 samples.

Emmons (1987) reported similar breadth in the taxon and size of jaguar prey in the

western Amazon. There were 40 prey items in 25 jaguar feces and 12 in 7 puma feces,

item/sample ratios of 1.6 to 1.7 respectively, suggesting both cats have a diverse diet, and

that more feces would have yielded yet more prey species. Both studies suggest that

jaguar hunt opportunistically in densely forested habitats. In Peru, jaguars even ate snails

and skinks (Emmons 1991).

A caution in interpreting jaguar and puma diets across their entire low latitude

distributions is that prey diversity varies among study areas. The sub-tropical moist

forests of southern Belize have relatively high prey diversity (Rabinowitz & Nottingham

1989). The upper Amazon in Peru is one of the most biologically diverse areas in the

world (139 mammal species (Voss & Emmons 1996). In contrast, the llanos of

Venezuela are less diverse (75-80 mammal species (Eisenberg & Polisar 1999). The

llanos prey base is more diverse than that of temperate study areas (Dalrymple & Bass

1996; Hornocker 1970; Logan et al. 1996; Maehr 1997a; Maehr 1997b; Murphy 1998),

but its forest-dwelling prey are a subset of the assemblage native to more constantly

humid forests (August 1983; Eisenberg & Polisar 1999; Eisenberg & Redford 1979), that

penetrate the savanna via the mesic and structurally complex forest. This is a situation

similar to other South American savanna-forest mosaics (Mares et al. 1985; Redford &

Fonseca 1986), and typically involves very few endemic mammals (Eisenberg & Redford


The diverse prey in rainforests may be relatively evenly distributed in comparison

to savanna-forest mosaics. Fauna and flora cannot truly be evenly distributed, but

contrasts can be expected. In large blocks of high-alpha diversity tropical moist forest,

edges are softer and much production (primary, secondary, and tertiary) arboreal

(Eisenberg 1980). In the llanos, edges are abrupt, much production is at ground-level,

and oscillating aquatic habitats facilitate an impressive amphibious prey base (capybara,

caiman, turtles). Not only is terrestrial production high in the llanos (Eisenberg 1980;

Eisenberg et al. 1979), and mean weight of prey high (Hoogesteijn & Mondolfi 1996),

the physiognomy of the region with its horizontal beta-diversity should result in a more

patchy distribution of resources, even for secondary consumers such as jaguar and puma.

These predicted contrasts in the patchiness of prey distributions and subsequently,

concentrations of cat food, would in turn predict that the hunting patterns of llanos cats

would be less opportunistic and more selective than those of rain forest cats.

The aggregate of management and ecological questions that this component of

Hato Pifiero project addressed follows. Do jaguar and puma have adequate natural prey

to survive without a subsidy from livestock? How much natural prey biomass is

available to the large cats? Are prey distributions even or patchy? Do prey distributions

vary seasonally? Do habitat characteristics affect prey distributions? If so, what habitat

characteristics are important? Do the large cats hunt opportunistically, taking prey in

relation to abundance? If diets demonstrate selective hunting, which prey are preferred by

jaguar and which by puma? How do the diet of jaguar and puma differ in terms of prey

body sizes and overall diet breadth? What are the relationships between felid attacks on

livestock and the seasonal distribution of livestock? Are attacks on livestock related to

local prey availability? What can be done to reduce the frequency of large cat attacks on


Chapter 2 introduces the study area, describing it in some depth. It also presents

summaries of vegetation analyses, concentrating on forest floristics and physiognomy.

Since the most important prey of the large cats were likely to be herbivores, it was

important to clarify forest types and evaluate the relative ability of those types to sustain

prey. Chapter 3 focuses on the distribution, abundance, and biomass of potential prey.

While Chapter 3 contains many subtopics of interest, it also serves to set the stage for the

section that follows it. Chapter 4 is a broad, yet condensed synthesis of the availability of

prey, the selective patterns of the large cats, and how those both relate to cat attacks on

livestock. Ecological theory and Paleontological perspectives assist the interpretations.

Chapter 5 contains comments and recommendations, both general and specific, relating

to ways in which conflicts between large cats and cattle ranchers can be reduced. These

recommendations address the immediate region in which we conducted this study, under

the premise that large-scale gains in conservation are constructed of, and constrained by,

the sum of local advances. Despite that perspective and presentation, elements of the

recommendations are quite applicable to the complete range of savanna/forest mosaics

present in South America.



Hato Pifiero is a working 80,000 ha cattle ranch/wildlife preserve located between

8 40' and 9 00 N and 68 00 and 68 18 W(Miller 1992) in the southeast comer of Estado

Cojedes in north-central Venezuela. The northern boundary of Pifiero lies among hills

that rise to 396 m above sea level (Farrell 1999). The western boundary is formed by the

Cojedes and Portuguesa rivers, the southern and eastern boundaries by the Chirgua and

Pao rivers (Fig. 1). Smaller streams (carios) run through this basin. The lowest elevations

are approximately 65 m above sea level in the open esteros in the southern part of the

ranch. The landscape can be characterized as a complex mosaic of interdigitated forests

and open areas with vegetation types based on interactions of elevation, substrate, and

hydrology. The ratio of open to forested areas is roughly 50:50 depending on

interpretation (Fig. 2). Many of Pifiero's forests follow stream beds, but relatively large

blocks of semi-deciduous forest not adjacent to stream beds are a characteristic of this

region, termed Los Llanos Boscosos (the forested plains) (Velasco & Ayarzaguena

1995). Only in the far southern reaches of Pifiero is broad savanna laced with narrow

strips of gallery forest, a landscape characteristic of areas further south in Los Llanos

Bajos. The estimated 407 ha of open water in the study area during the late dry season is

in stark contrast with the almost 80 % of the ranch shallowly inundated in the middle of

the rainy season.

The hyper-seasonal environment (Thorbjarnarson 1991 a) of the llanos is a result

of interactions between climate and soils. Pifiero receives an annual average of 1468.8

mm of precipitation, with the majority falling between the beginning of May and the end

of November. The dry season runs from December 1 through April. For faunal analyses,

I designated the wet and dry seasons as May 16-November 30 and December 1-May 15,

respectively. This was based on 1996 and 1997 rainfall patterns (Fig. 3), patterns of leaf

flush and fall, and seasonal shifts in animal distributions. The clay soils in the plains

cause surface water to accumulate starting in June and reaching peak levels in July and

August (Fig. 4). This flooding is relatively shallow (Fig. 5). The deepest water

encountered on foot transects was about Im (waistline), and the deepest ever encountered

about 1.5 m (belly deep for horses). Although up to 80-90% of the surface area of some

forests are shallowly flooded following strong July rainfalls, the average is less (Fig. 4)

and contrasts in micro-elevations important. Islands of damp land remain in most low

elevation forests, even when surrounded by a shallow sea of flooded savanna.

Many savannas retain water through January, as does the evergreen forest (Fig.

4). As the dry season progresses, mid-day temperatures in the sun may exceed 40 C.

Vast areas of surface water contract, forcing impressive concentrations of caiman, turtles,

and fish in shrinking aquatic habitats. The same savannas that are 100% inundated

during the rainy season are burnt by the ranch workers (llaneros) during March to

improve forage. Forests occupied by fish and frogs in July are bone-dry by March, the

hard clay soil covered with leaf-litter from the bare trees above. The shallow, phosphorus

and calcium poor, granite derived soils and gravels in the hills are highly permeable and

dry rapidly (Ramia 1993). These edaphic characteristics create extreme drought in the

dry season, and the vegetation in the high hills is clearly fire-adapted.

The predominant races of cattle among Pifiero's 14,000 head were cebu (nelore,

brahma, guzerat, gir) (Bos indicus). Other races were criollo (Bos taurus), senepol,

romosinuano, and various hybrids. A herd of approximately 150 water buffalo (Bubalus

bubalis) was maintained in the southern savannas. Approximately 420 horses, mules,

and burros fulfilled working and breeding needs. Most cattle were moved from flooded

areas to higher drier pastures during the wet season. Artificial insemination resulted in

calving peaks from July through September (Femando Corrales pers.comm.).

Cattle ranching is the dominant land use in the llanos. Land is far from equitably

distributed among the populace. Powerful landowners hire men to manage cattle and

horses, and women to cook. Wages hover around the minimum required. Despite any

ramifications this has for social mobility among rural folk, the landscape is relatively

intact considering the profits generated, and the llaneros proud of their heritage and skills.

Hunting has been banned in Pifiero. As a result, most wildlife is common and visible,

giving support to a profitable on-premises ecotourism lodge. Poaching does occur on

ranch perimeters, and to a lesser degree, within the ranch in several less frequented areas.

Sampling investments were lower in such areas to avoid potential disruptions in study

design and execution.

On two sides of Pifiero, neighboring large ranches also contain good wildlife

habitat and reduced levels of hunting (Hato Socorro to the northwest, Hato Corralito to

the northeast). Cafio Caujaral enters Pifiero through a valley bisecting the northern hills.

Visits to the cailo where it ran through Hato Mata Clara (north of Pifiero), ultralight

aircraft reconnaissance, and satellite imagery all testified to the dense strip of forest

surrounding the cario as it extended north of the ranch. These factors of connectivity are

what has allowed jaguars, white-lipped peccaries, and tapir to persist in the agricultural

matrix of los llanos altos.

Pifiero possessed a high degree of horizontal habitat heterogeneity. This was

predicted to influence prey distributions, and correspondingly, large felid intra-specific

and inter-specific partitioning of space and resources. A classification of forest types was

critical for mapping purposes and to evaluate the relative utility of habitat types for

potential prey. The remainder of this chapter describes 1) quantitative and qualitative

vegetation sampling in forests, savannas, and pastures; 2) forest classification; 3) some

aspects of forest phenology; 4) mapping efforts and spatial analyses of the study area; 5)

plant resources available to prey in different habitat types; and 6) all of the above in

relation to work conducted by colleagues in the same region.


The 26 transects established to evaluate faunal distributions and abundance were

also used to record quantitative floristic and physiognomic data. Reconnaissance trips by

vehicle, foot, and ultralight aircraft were used in planning transect design. A vegetation

map, created by Dr. Francisco Delascio's (Venezuela National Herbarium & Botanical

Garden) interpretation of 1989 Landsat Thematic Mapper imagery assisted planning,

although reconnaissance also indicated that it needed additional detail. The numbers of

transect lines and their length were as follows: dry hill forests [4, 9.8 km]; semi-

deciduous forest not adjacent to permanent water [3, 5.5]; semi-deciduous forest near

permanent cario or prestamo [4, 7.51]; interspersions of semi-deciduous forest and

savanna, ecotone crossed perpendicularly [2, 4.5]; semi-deciduous forest edge, parallel to

pasture [1, 2]; small flooding savannas and managed pastures near forest [5, 5.65]; large

flooding savanna, close to forest [2, 4]; large open flooding savanna away from forest [2,

4.7]; higher elevation, non-flooding pastures [3, 3.55] (Fig. 2). This design took in a

representative cross-section of habitat types, and facilitated evaluations of the effects of

proximity to water, habitat interspersion, and distance to cover on animal distributions.

Structural characteristics of transects were recorded at 100 m intervals. Variables

recorded in forests were: substrate; canopy height;canopy closure; mid-story closure;

shrub-level closure; two densiometer readings; distance to and DBH of nearest overstory

trees in four quarters; distance to permanent water; distance to ecotone; and habitat and

elevation categorizations. The variables recorded in savannas, pastures, and calcettas

(pockets of savanna in forest) were: distance in quarters to nearest tree; height of

preceding; distance to ecotone; distance to permanent water; and habitat and elevation

categorizations. GPS locations recorded on transects or at nearest open area facilitated


During the first week of every month, for twelve months, phenology observations

of marked individuals of 29 tree species important as food for prey were recorded along

3-5 trails in forests. This was complemented by published phenology data on 14 tree

species from nearby Cafio Benito (Ortiz 1990), observations of animals feeding in our

study area, and data presented in (Monasterio & Sarmiento 1976) and (Robinson 1986).

With help from Venezuelan National Botanical Garden/Herbarium staff (Dr.

Francisco Delascio and Rafael Ortiz) forest composition was sampled along an altitudinal

cross section of Pifiero's forests. All trees and vines over 4 m in height were identified

and enumerated in 10 x 50 m plots. Understory species were assigned a cover

classification in each plot. The 35,000 m2 sampled ranged from hilltop chaparral to

seasonally flooded evergreen forest. The majority of plots were placed at 100 m intervals

along four transects with rich data on animal observations. Eleven plots not along the

foot transects represented potential intermediate types. These plots lacked the

physiognomic data collected along transects. Qualitative observations of relative species

composition were recorded for several types of savannas and pastures and some

additional forest areas of interest.

The existing vegetation map was improved through the following: 1) GPS

locations along transects and roads and areas in question; 2) overlaying the map and my

modifications on topographic maps and a plotted February 27, 1990 (dry season)Landsat

TM Scene classified in Bands 3, 4, and 5; 3) finalizing classifications (pooling for

simplicity in spatial analyses) and drawing polygons; 4) scanning into a TIF file; 5)

digitizing in ESRI's ARCVIEW 3.1 to obtain habitat-specific area estimates and a

product for presentation.

Forest plots were classified using hierarchical cluster analyses using furthest

neighbor linkages and squared euclidean distances (SPSS 1999). These were based on

tree species frequencies in each plot. Analyses were run with 1) all plots and all species;

2) 59/70 plots with rare trees removed (occurring in 2.857%< of plots) and structural

parameters added (mean height of canopy trees, mean DBH of canopy trees substrate and

elevation classifications); 3) all plots, no structural data and rare and common species

(occurring in 4.28%< or >50% of the plots) removed. Following classification, data from

clustered plots were pooled for composition comparisons and assessment of relative food

availability. The list of plants used as food for the more important prey came from

observations presented in (Barreto et al. 1997; Brokx 1972; Danields 1991; Delascio-

Chitty & Branger 1996; Delascio-Chitty & Stergios 1996; Moskovits & Bjorndal 1990;

Robinson 1986; Robinson & Eisenberg 1985). In addition, we substantially

supplemented the literature using observations made by project personnel throughout the

study period, and complemented by personal observations supplied by llaneros.


Forest Types

The most definitive results from cluster analysis came from transect plots, in

which floristic differences were complemented by physiogonomy, and rare species were

screened out (Fig.6). Evergreen Forest/Bosque Siempre Verde (BSV), a small forest

patch on loam soils in a valley subjected to prolonged flooding was unique due to the

dominance (over 56% of all trees) of Vochsyia venezuelana Vochysiaceae (Table 1).

Mixed Dry Forest/Bosque Seca Mezcla (BSM), so-named because it was a hillside (50%

high hillsides, 50% low hillsides) mix of hill and lowlands elements, was, despite

relatively high diversity, dominated alternatively by Protium heptaphyllum Burseraceae

and Erythroxylum orinocense Erythroxylaceae (both over 25%)(Table 2).

Species area accumulation curves run on subsets of the Semi-Deciduous Forest

Type 1/Bosque Semi-Deciduo Type (BSDT1) type (Fig. 7) indicate that, in relative

homogeneous forest (La Candelaria) one might expect a leveling out at around 6 plots

(3000 m2). The Caujaral Norte transect (also BSDT1) differed. Located parallel to the

Cafio but at varying distances, mild topographical irregularities due to past and present

seasonal feeder streams resulted in higher diversity. Although the BSV and BSM types

were based on only three plots each, these small forest patches were clearly distinct.

(Table 1, Table 2, Fig. 6). The Atypical Bosque Seca Mezcla (ABSM) was located in the

shaded fold of a high hill, and was sometimes used by Cebus olivaceous, uncommon for

that elevation. Topographical irregularities and subsequent variation in exposures

allowed elements from lower elevations to penetrate these higher sites. BSM plots were

50% on high hillsides, 50% on low hillsides, and the substrates were 75% rocky and 25%


Dry Forest/Bosque Seca (BS) could be divided into two sub-types based on

elevation (Table 3, Table 4). Dry Hill Forest/Bosque Seca de Cerros (BSC) occurred at

lightly higher elevations (40% high hillsides, 60% hill bases, substrates 60:40

rocky/gravel (Table 3). Hill Base Dry Forest/Bosque Seca de Falda (BSF) occurred

along the base of hills (100% hillbase, 62.5:37.5 gravel/rocky substrates (Table 4)).

Dry Savanna with Chaparral/Sababa Seca con Chaparros (SS/CH) could be

broken down to that occurring on high ridges (SS/CH Alta: 60% ridge top, 20% high

hillsides, 20 % low hillside, 60% gravel, 40% rocky (Table 5)) and low ridges (SS/CH

Bajo: 50% hillside, 50% hillbase, 50%gravel, 50% rocky (Table 6)). No trees in these

plots equaled or exceeded 4 m in height, which excluded them from cluster analyses

while establishing their uniqueness.

The greatest number of plots were in Bosque Semi-Deciduo. This apparent

sampling bias echoed the spatial dominance of BSD in Pifiero. Bosque Semi-Deciduo 1

(BSDT1) was wide-spread (Table 7). Bosque Semi-Deciduo Tipico 2 (BSDT2), more

prone to inundation, occurred relatively close to Cafio Caujaral(Table 8) All BSDT plots

were on hard clay-based soils. Other variants of BSD included: (BSDG=Galeria) directly

adjacent to Cafio Caujaral; (BSDS=Seca) typical of non-flooding plains below hills; and

(BSDQ=Quebrada) in ravines among the high hills in the northwestern part of the study

area (Tables 9,10,11). The latter three sub-types of BSD were less strongly defined .

Lacking physiognomic data such as tree height and DBH, the classifications were more

variable (Fig. 8). Nonetheless, the quebrada (BSDQ) forests, the hill base forests

(BSDS), and the gallery forests (BSDG) were set amidst other BSD plots (Fig. 8). The

CER11-13 plots, that accompanied them in the same section of the dendrogram were

BSC plots that classified poorly until they received the added variables of tree height and

DBH. The gallery forest directly adjacent to Cafio Caujaral did differ in composition

from that further from the stream (Tables 7, 8, 9). Along the west shore of Rio Pao

standard BSD continued up to the edge of the steep river banks.

The physiognomic characteristics of 8 forest types are presented in Table 12.

Tree height and girth increased with decreasing elevation and increasing moisture as did

canopy cover, mid-story and understory. Strata in physiognomic data collection were not

limited by the 4 m criteria used in floristic sampling. Tree spacing was positively

correlated with tree height and DBH (large trees cannot be clustered tightly together).

The BS forests tended to have a high density of small stems. At the far end of the

moisture gradient, DBH and tree height declines where flooding persisted. The deeply

shaded BSV forest was full of slender stems forming a dense canopy.

Vines species in forest types are presented in Tables 13 through 20. Understory

species (including the "stunted" canopy of the SS/CH types) of all forest types are

presented in Tables 5, 6 and 21 through 28. The understory was best developed in the

BSDT types, which provided food options for major felid prey at all levels (Table 12).

The entire community of the SS/CH types is compressed into Tables 5 and 6. Those

partially open habitats did have woody species, and stratification, but all were less than 4

m in height.

The relative abundance of important food items in the forests types is presented in

Tables 29 through 37 [trees], Tables 38 through 41 [vines], and Tables 5, 6 and 21

through 28 [understory], and summarized in Figs. 9 and 10. Vines of food importance

appeared scarce in BSDS, BSDT2, BSDG, and BSV. The low numbers of plots in BSDS

require caution regarding that conclusion for that habitat. SS/CH types appeared to be

poor providers, a result of their incomplete ground cover and reduced vertical

development. BSDT was rich in comparison, with abundant under story. Although the

greater number of plots in BSDST1 has to be considered, the habitat did have multiple

productive levels. Its greater area provided more food options overall. The BS forest

type was less extensive than BSDT1 and presented fewer food options (partly but not

entirely an artifact of area). BS still possessed considerable food value for prey,

confirmed by numerous animal sightings in those hilly areas. The altitude/moisture

gradient relationships of the forest types were as follows: SS/CH Alto; SS/CH Bajo;


Forest Phenology

The introduced mango (Mangifera indica) was a common and seasonally

important food (Table 42), but restricted to well-drained sites. Jobo (Spondias mombin)

fruit fall was superabundant during the rainy season in some sections of BSDS and

BSDT1. It also was relatively widespread (Tables 3, 4, 7,10,11,42). Annonaceae

species, also fruiting in the rainy season, occurred in nearly every forest type (Table 42).

Corozo palms (Acromia aculeata) occurred in better-drained sites, palma llanera

(Copernicia tectorum) in wetter sites (Table 42). Bromelia chrysanta and B. pinguin

were confined to low elevation BSDT1 & 2. The Marantaceae were also confined to

BSD (Tables 24, 25, 42). The tubers of these were used by peccaries in all seasons. The

above ground portion of the plant was luxuriant in the rainy season, but collapsed in the

dry, when peccary excavations for the tubers were most noticable. Guacimo (Guazuma

tomentosa) fruits were also superabundant when in season (January through April) and

used very heavily by peccaries. This species, common to BSD, thrives in gaps, thus

becoming most abundant along pasture edges and roadsides, rather than the forest

interior. Ficus sp. were uncommon in BSD (Table 33), but relatively common in BSV

(Table 37). Copernicia was never abundant in BSD (Tables 33, 34), completely absent in

the semi-deciduous forests near Rio Pao, and common in seasonally flooded savannas.

Lycaniapyrifolia fruited in the rainy season (Table 42). A narrow strip of forest along a

seasonal cafio in a valley in the northern hills was frequented by collared peccaries

(Tayassu tajacu) during the rainy months. Though species poor, this forest was rich in

Annonajahnii, A. purpurea and L. pyrifolia (Table 42). Another forest, in the same

valley, lining permanent canio, was considerably more diverse.


The pastures and savannas of Pifiero varied widely in terms of floristics, degree of

seasonal inundation, proximity to forest, extent of woody vegetation, and proximity to

permanent water. Measurements taken on the most wide-open savanna transect in the

south resulted in a mean distance to tree in the open area (MDTR) of 548 m, a mean

distance to ecotone (DIE) of 765 m, and a mean distance to water (DIH) of 1772 m

respectively. The average for all four transects in the large savannas in the south were:

MDTR 280 m; DIE 490 m; DIH 956 m. The average for six transects in small savannas

with forest nearby (a common arrangement in Pifiero) was 115, 116, and 437 m

respectively. The values for the savanna transect with the highest deer densities and

perhaps overall highest open area prey densities were: MDTR 93 m; DIE 87 m; DIE 147

m. All the preceding were savannas that are shallowly flooded for seven months of the

year. The average values for the four transects in pastures in higher, drier areas were:

MDTR 50 m; DIE 66 m; DIH 460 m. These pastures tended to be set in relatively

narrow valleys. The bands of BSDS and BS lining them were usually also narrow, with

SS/CH above. The values for distance to permanent water in the high dry pastures was

approximately equal to that in small seasonally flooded savannas. In the higher pastures

that water was water tanks for cattle or relatively barren prestamos. In the small

seasonally flooded savannas the water sources were occupied by caiman and frequented

by capybara.

Brachiaria humidicola, an introduced species from Africa was the most common

grass in better drained pastures. The small seasonally flooded savannas were sometimes

dominated by Thalia geniculata (Marantaceae). Other important species in those habitats

were Sporobolus jacquemontii and Panicum laxum (Graminae), Sida acuta and

Wissadula periplocifolia (Malvaceae), Eleocharis elegans (Cyperaceae), Ipomea carnea

(Convulvaceae), and Cassia reticulata (Leguminosae). Less common were Hymenache

amplexicaulis, Cynodon dactylon and Paspalum fasciculatum (Graminae), and Hydrolea

spinosa (Hydrophyllaceae). The most common grasses in SS/CH were Andropogon

angustatus and Trachypogonplumosus. Ipomea was common in the large savannas in

south-central Pifiero. The broadest savannas in the south were dominated by Paspalum

fasciculatum. The latter has reasonable protein content, but low palatability for cattle.

Buffalo (Bubalus bubalis) handle its high cellulose/lignin content well, and also use

Cyperaceae. Cattle best use P. fasiculatum sprouts following a bum.

Brachiaria humidicola is a trampling-resistant invader with slightly better forage

quality than native grasses. In general, all forage protein contents decline by 70-75%

during the dry season, but this loss can be reduced by fertilizing. The Trachypogon in the

hills is poor forage. Neither it nor the Andropogon are good ground cover. Thalia is a

poor forage invader, forming impressive and tall cover during the rainy season. When its

tall (3m) stems dry and fall during the dry season grasses can emerge from the rows of

rubble. Copernicia was common along the fringes of small and large seasonally flooded

savannas and also scattered in varying densities throughout the open areas away from

edge. Cassia reticulata fringed many of the smaller seasonally flooded savannas

(pers.obs., Rafael Hoogesteijn and Francisco Delascio pers.comm.).

Vegetation Map

The pooled vegetation types when preparing the map (Fig. 2) were as follows:

SS/CH; BS; BS/AP; BSD; BSV; M; PS; SI (Table 43). The BS classification pooled all

Bosque Seca sub-types. AP represented Agro Pecuaria (cultivated crops). The BSD

classification pooled all BSD sub-types. M represented Mangera, which is a local term

for a large grove of mango trees (abandoned orchard). PS represented Pasto Seca (high

dry pastures). SI stood for Sabana Inundable, and pooled all flooding savannas, both large

(low interspersion indices) and small (high interspersion indices). Many times they

actually were the same savanna. The interconnectivity of habitats in Pifiero was such that

a polygon of wide-open savanna in the east, curved around the far south, swept through

the southwest as even broader savanna, and also formed short narrow peninsulas of

savanna every place where it curved back into BSD. Animal densities and compositions

varied throughout, but, in some respects, it was all the same savanna. Similar situations

arose with BSD.

A simplified areal summary of habitats in the 63,227 ha study area west of Rio

Pao is as follows: seasonal flooded savanna (sabana inundable = SI) 24,699 ha; non-

flooding high pastures (pasto seco = PS) 1,806 ha; evergreen forest (bosque siempre

verde = BSV) 48 ha; low-elevation moist semi-deciduous forest (bosque semi-deciduo =

BSD) 21,434 ha; mango (Mangifer indica = M) groves > 7 ha; mid-elevation, hillside dry

semi-deciduous forest (bosque seca = BS) 4,986 ha; and on top of high hills and low

ridges, (sabana seca con chaparros) = SS/CH 9,648 ha (Table 43).


The composition of semi-deciduous forest in Hato Pifiero in Estado Cojedes

differed from that in Hato Masaguaral in adjacent Estado Guarico (Robinson 1986;

Robinson & Eisenberg 1985; Troth 1979). In Masaguaral's semi-deciduous forest,

Copernicia tectorum was the most abundant tree, Genipa americana the second most

abundant, and, collectively, Ficus sp. were quite common (Robinson 1986). All three of

these important food-producing plants were far less common in Pifiero's forests (Tables

33, 34). The nearly year-round availability of Ficus fruits alone at Masaguaral (Robinson

1986) suggests an explanation for later-discussed differences in mammal densities

between the two study areas. That these two sites, both in the north-central llanos could

be so different is hardly surprising. The variation inside Pifiero alone was impressive.

Copernicia was common in the southern stretches of BSD where increased flooding also

resulted in lower stature forests. Soil characteristics (low Magnesium content) excluded

it entirely from the forests leading to Rio Pao (pers.obs., Rafael Ortiz pers.comm.).

Barreto et al.(1997) considered its seeds in Tayassu pecari feces in that area evidence of

long distance movements. In fact, the palm was common in the savannas just west of the

Pao forest, from where we once encountered the group of peccaries returning.

The soils in Pinero's low elevation savannas (and adjacent forests) were relatively

rich (eutrophic) (Ramia 1997). Those in the high hills are poorer (oligotrophic) (Ramia

1993). In the hills, the line between savanna and forest was not dictated by soil fertility,

rather by moisture gradients, which were a result of soil texture and depth, inclination,

physiographic position, and underlying drainage patterns in bedrock (Ramia 1993).

Thus, risking simplification, savannas occur at Pifiero's highest elevations due to low soil

moisture and at Pifiero's lowest elevations because of excessive moisture (prolonged

saturation), with forests occupying the gradients between. Among forest types, those

with the greatest vertical complexity presented the greatest diversity of food types for

herbivorous prey. Most of Pifiero's forests were seasonally deciduous. A forest that is

nearly leafless for five months of the year will present a hardship to obligate arboreal

folivores, and few would be expected (August 1983). Although soft fruits were available

from both native and introduced trees, hard fruits such as the dry pods of Leguminosae

were important food for a variety of terrestrial mammals in Hato Pifiero.

- - .
-- --


70 --

Figure 1. Northwest and north-central Venezuela. Shaded areas show location of Hato
Pifiero study area in relation to locations where intensive Smithsonian research projects
were initiated in the 1970s.

? "

* .i. -' I' ,T: '.-.

r 2

-j k-v -. -

r N
-^ -

* .1

Figure 2. Color-coded vegetation map of study area.

tc" ( ." ,: ..
/ I, /^ l~









1996 11997 -----34 eamean

Figure 3. Monthly precipitation. Records for 1996-1997 from Hato Pifiero. Monthly
averages for 34 years in nearby El Baul from Sistema Nacional de Informacion

Hidrologica y Meteorologica, Caracas, Venezuela.


80 I

i /

/S I


Figure 4. Mean percent of ground surface covered by water in two forest types during
rainy season. BSD (Bosque Semi-deciduo) data collected along six foot transects (11,600
m). Maximum inundation was 90% of ground surface shallowly flooded. BSV (Bosque
Siempre Verde) data comes from one 300 m transect section at the base of a hill system.

Mean Monthly Water Depths in Flooded BSD


Figure 5. Mean depth of flooding along six transects in semi-deciduous forest.
Maximum depths were localized (such seasonal streams feeding into cario or canio
spreading beyond its banks) and approximately one meter.


-- .. I
I* I-
I' i

: --: -!


.-- Ii

Figure 6. Complete furthest neighbor cluster analysis on all plots on foot transects (59 x
500m2). Variables were tree (>4m tall) species frequencies, with added variables of
canopy height, mean DBH, substrate type, and elevation category. Of 101 tree species
identified in 70 plots total (11 without structural data) rare trees (occurring in 2.86% or
less of the plots) were removed for this analysis, leaving a total of 80 tree species. This
dendrogram omits an entire dry savanna/forest type SS/CH, which is populated by woody
tree species, but all are under 4m. Vine frequencies were recorded concommitant with
tree frequencies but not used in analyses. Understory species were recorded in coverage
categories, a variable not compatible with the counts per species made for trees.

Tree Species Accumulation Curve BSDT1 along two transects

500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500
meters squared
CaujaralN 13 lh50mplots spaced ove Canddlana 13 10x 50m plot spaced
24a00mteval, near streatbed aralN C aoves 1850m1teval
-- Caujaral N -* Candelana

Figure 7. Species area curves for low elevation semi -deciduous forest in Hato Pifiero.
Heavy line presents rate of new species encounters along a relatively homogenous stretch
of forest (13 10 x 50 m plots spaced over 1850 m interval along Candelaria transect).
Light line presents rate along a transect passing through more variation, both in micro-
elevations and in variable proximity to a permanent caio (13 10 x 50 m plots spaced over
2400 m interval along Caujaral Norte transect).

-w -


6.0 i:


I -



----- i .-..J

Figure 8. Complete furthest neighbor cluster analysis on all forest plots (70 x 500m2).
Variables were tree (>4m tall) species frequencies. Of 101 tree species identified in the
70 plots rare trees (occurring in 4.28% or less of the plots) and common trees (in > 50%
of the plots) were removed for this analysis, leaving a total of 65 tree species.
Highlighted are plots not occurring on foot transects: 10/11 fell within BSD plots on this
cluster run without structural variables. These are as follows: BSDG (gallery); BSDQ
(quebrada); BSDS (unlike BSDT1 & 2 and BSG, never inundated). The remaining plot
(CGS3) fell within the BSF type, reflecting its location at the upper end of a phenology
trail abutting the CG transect (CGT plots) which dominated the BSF type. Circles denote
three high hill forests (BSC, Fig. 6), that without physiognomic variables consistently
clustered with BSD plots.

- I

Ir i~








Figure 9. Numbers of species of plants that contribute to the diets of prey in three

categories (trees > 4 m in height, vines > 4 m in height, and "understory" being woody or

herbaceous species < 4 m in height). Potential prey considered were Tayassu tajacu,

Tayassu pecari, Dasyprocta agouti, Odocoileus virginianus, and Geochelone

carbonaria). Species in which mature individuals sometimes do not exceed 4 m in height

may be represented in both trees and understory.

Number6 of Species

Vine s
Under Story

Known to be Useful
as Food to Select
Major Prey

Type Total

SSiCH 4 m Height




Figure 10. Percent of total plant individuals in three categories by forest type
contributing to prey production. Potential prey considered were Tayassu tajacu, Tayassu
pecari, Dasyprocta agouti, Odocoileus virginianus, and Geochelone carbonaria. Tree
and vine proportions presented are percentages of all individual plants in a that category
in that vegetation type. Understory indices are cruder: the sum of the maximum
categorical values assigned by occular estimates. The understory indices can exceed 100
and are not directly comparable with the tree and vine percentages.

Table 1. Tree species inventory ofBosque Siempre-Verde

Number Relative
Species of trees density

Vochysia venezuelana 151 56.55
Protium heptaphyllum 18 6.74
Leguminosae sp. 18 6.74
Ficus maxima 16 5.99
Annonajahnii 13 4.87
Copaifera officianalis 13 4.87
Pterocarpus acapulcense 11 4.12
Ouratea grossourdii 8 3.00
Annonapurpurea 5 1.87
Cochlospermun vitifolium 3 1.12
Genipa americana var caruto 3 1.12
Hecatostemon completus 2 0.75
Randia hebecarpa 2 0.75
Astronium graveolens 1 0.37
Jacaranda obtusifolia 1 0.37
Bombacopsis quinata 1 0.37
Curatella americana 1 0.37

Table 2. Tree species inventory ofBosque Seca de Mezcla

Number Relative
Species of trees density

Protium heptaphyllum 128 26.02
E I tlih,r hi, orinocense 123 25.00
Astronium graveolens 71 14.43
Jacaranda obtusifolia 25 5.08
Vitex capitata 25 5.08
Copaifera officianalis 18 3.66
Randia hebecarpa 11 2.24
Genipa americana var caruto 10 2.03
Luehea candida 9 1.83
Tabebuia ochracea 7 1.42
Connarus venezuelanus 7 1.42
Trichilia unifoliata 7 1.42
Soracea spruce 7 1.42
Ouratea grossourdii 7 1.42
Allophyllus occidentalis 5 1.02
Annonajahnii 3 0.61
Cochlospermun vitifolium 3 0.61
Curatella americana 3 0.61
Byrsonima crassifolia 3 0.61
Bowdichia virgiloides 3 0.61
Pterocarpus acapulcense 3 0.61
Unk 3 0.61
Lonchocarpusfendleri 2 0.41
Roupala montana 2 0.41
Vochysia venezuelana 2 0.41
Tabernaemontana cymosa 1 0.20
Sciadodendron excelsom 1 0.20
Cassia moschata 1 0.20
Capparis sp. 1 0.20
Machaerium aculeatum 1 0.20


Table 3. Tree -pec ie- illnenilol of Bo-qiie Sean ce Cern'o

il i.in .-r F.-l.it --
"-:i- ;__________. t t[ ',r- -: I- n. [ _

- '. .' , :, ,: ,

.-1. v .I .- . . 1 1 -




1 1
.' I.., .. .:4


-. .. ,, -. 1 4
- -',' ., "1, -. '. .- -1"1

,,--.: '. 1 ', ,:.-' .-' :: ' ".,.,,-. 4 1 ,:

1 _
... .' .', I -

, -- 1
.: .,, ,' ', ,, '. .- .-.' ." 1
.: .-,.- .', '. ,.-.- : :- 1


Table -1. Tree specie- invlenlor) of Boqliie Secn le Faldn

.- . v i

.t 4
S .

.- ;, . 1* 1 .: _4
.- .. .. 1 1

A .- 1 1

.i ......- .,

.. .- 1 ,

.....1 1: 4
1 41
1 14

S1 14
-.1 1..14


. ., '
.-- .- '. ..',' v :, -14


Table 5. Specie- --iim in height ofSnabain Seca con C'linp)uro- Ahlo

H-i . ...: F I

-. -" . ; : .. :. i t ".I

.4' . :. 1 :

- I

- v.- . /," .,. .. 1,'

S- I '.
.. . - I,'

,' .. . .:, I '
"-- ' ,2 v. ,

.--f ,' ,. .', .,...-.' 1 (
." :..-.- : .:- .-.. . .'1 ,1
-: . ,' .-, , :: .-. 1'
,.-- '. .. .-. .'. ... -. .'. ,- : .:1 ,:
' 1,
. .-' .- f.: ,d -.:

.v. i .
.[.k ? ,. : .'.'. -.- :.l '[
: . .- .- l ' .
: -. v .- : .. ., -., .'. .'.l '[

r. 1-:-iIL e :4


Table 6. Specie- : in Ileigli! of Snbuman Seen con Clinpaluo- Bajo

H- l1. "- ,..: F : I

S- '

S* ,.

-'--I". ".. ..

S' ,I
.- .' . .'I


-- ,, ..- , I,

I_ .- ,-.- .- -


Table -. Tree lpecie- inveinollo of Bo-ilue Semni-Decidico Tilpico 1

m,, ,, I I I rI

- ., '-,i4i i

4 i


I1 I
_I.i "

1 44

1 .1 4

SI., 1 l

I .1 ,


S .' .' ." 4II i
." ." .' ,., i.' ',. ..' I



b. -- ,. ; ' ..'I
."r- , ,I-
',,', ,, '. ',, .."/ ." " ,, L II


Table ".--Continued.

1- 1 I ."- F.- l.-it L

L }.-

F..I. L 1" :'
.-.- .. .. -.-

r .,, ,

., '. .- ..- '..

-' -* *- .. ...


Table 8. Tree -pecie- iinenlnoiy of Bo-quee Semi-Dec ichio Tipico 2

l i,. n i ,- F.-l.,t L-


." . 1 -
. ,- ' -. . .', .',' : I

A: - .. H. .- 1 -:.. ._-

. .-1...' .. ., :.-' ..- 1 1 :.

_4 '. :.. *-. _'. '_". 4 1'
1.: .-. '. ', .. .' 4 4 :4
-' -. .... . 1 4 4

. .- . .- i.- . .-. . -4 1 : -

E i ,. . ..
- .. .. 1 11

- -' 1 1--. '

.4 '1 4

.-" -. : '. ,- H_
.... 11.-. .-'_4


: 1.' ,

I 1 1

Table 9. Tree -pec ite inv enllol of Bo-qiie Senii-Dec iiio Galeria

1 !-.in 1 1. i -r F.- l. L t

- ' .- .,, / .- 1_':'4 1:, .:-

-. ., : 1 -



. . 1 : 4 ]
S.. ., 1 4 1

. : 4

A. .'. 4 1

1 1. 4 1
1 41."1

1 41
1 41
,- .., '. ;, .' ; , ; , '
,- . :, ':, -- .,, .' .. . :
. ;, .- .-.' .- ',-,. -.- _i 1
E l. [ ,.;: .I'


Table 10. Tree l-pecie- ino enloln of Bo-qiie Seniii-Decidio Seca

[,.-r F.-l.it -I

-t 1 _-1.

'. .. -;, .,,'-

.. -: .-' , . -' ..-: 1 4 4
. . ', '. . .- '. .1, .-.-' ,.. 1 .

- -. -, ; .' '1 .

. .: '* ,.- ., 1 '


.- -
, .-, , .-., .,, -..,1 .
.I ;, '. ,- .-; '. ;, ',1 -
:,..E :,..-' .-: : '..-',' -" -' .'
-;, .-.,- ', .-, ;, .-.,-1;
-- .. - .- : '. .- .'


Table 11. Tree -pecie- invenl tol of Bo-qini Seinii-Decilniio Qiiebradn

1 !.1 i.- 'r F. lit L
-''. L:'I.- [ [[' '- I- '- l: [ _-LI


.,I 4-

S .,. ~ I :

', .. . 1 _
4, -.
,. ;, .- , ,,, ,- .

Table 12. Physiognomic characteristics of forest types along foot transects used for
animal observations. Types presented progress from dry fire adapted ridge top forests on
the far left to seasonally saturated valleys on the far right. Substrates and elevations are
discussed in more detail in the main text.










Canopy Height m 3.6 6.6 6.7 10.9 17.5 19.1 18.8 16

Canopy Cover (CC) 26 43.4 65 60.6 62.5 60.6 74.4 87.5

Mid Story Cover % 3 16.5 7.5 43.4 47.5 52.8 30.5 22.5

1 m Understory % 1 7.2 5 6.9 7.5 38.1 24.3 7.5

DBH Canopy Trees 14.2 14.7 17.6 24 28.2 48.2 47.1 27.3

Spacing Canopy 5.8 6.3 6.6 6.5 8.5 8.8 9.6 4.8
Trees m

Densiometer (CC) 12.2 66.4 94.5 98 95 93 98.8 100

Sample Size


Table 13. Vine ltpecie- ilnvenolo of Bo-qiie Secn de Cetn-o

= t F.-L1.it l e
. [-[t rit

4 14 .1
r, ,.- . .. ,

. v. .
r, ,

Table 14-. Vine l-pec ie illnenlolo of Boqiile Seca de Mlezcla

-'7.i IIr I I, it

....-.- .- 4


Table 15. \ ine -pecie- illneinol) ofBo-qii Seca de Falda

S. t F.-'l.it

..- :. '. :. .'. : 4 "4
.1 1 14 .:

1 4:

..- . 1 1 3:


Table 16. \ine -pecie- inlentoly of Bo-clue Senii-Decidtno Seca

= F.- .it L
S. It

Table 1~. Wine qpecie iviventoiy of Bo que Senii-Deciduo Tipico 2

m t FA.- .it -
.4 1

. l . : I

Table 1-. \iine lpecie- iIventoloy of Bo-clue Senii-Decidtlo Tipico 1

= .1 .t
' -, 1 '

-. . .'.

,%-, . , : , .,


Table 19. Vine -pec ie- in- enlol) of Bo-qiie Seiiii-Dec idio Galeria

=. t F.-I.

Table 20. \ine specie- iventllol ofBocqiell Sienipre-\ erle

i !.[n i. -r F.-l.it l

... . v-4


Table 21. Species -:1 :1 in Iltiidlr of Bo-que Seca (le Ct'eiro

S41 1 t : 4:.
.4 .. , , ,

- .- 1 t 4
.- .... .., 4

- ., ,, .- f.- ., ' ,. _1 ,

. : . .
- 1'
.: .-- ,.- .- ,.. ', ',1 ':
'- ' ', ,- -' -: ,:- ,' ,.-1 ':

: . . ,. : . ,-' '1 ':

.-",". : ', -' '


Table 22. Specie- : 4 -I1 in leidtl of Bo -qie Secn de Nlezcnl

H- l. . r . -

F : I
.-.-. I-r-

S11 t .

11 t

I I.
-. ." ..' 1 ,

I I.
-. 1 .

.4 . . . 1 .
.-' .' . .v .: .1.

4 -.- .-; '. .-; .- '. .

Table "3. Specie- : -1 n in heielr of Bo--qie Sea cle Faldn

Ii-T l'. .1. ..'.: F I

-; r. H.-.. -.- '. .1.;.I

.4 .'. -. I '0.
I I.

I I.
.". ~. I.
,.'. II

Table 21. Specie- : n in heights of Bo-qlle Semi-Decidiho Seca

1..1 ,. : F : I

-'1 [ t':- .I '''. -.,:" -
.....-...-i.- .r, ,,-,".

S: .. -:. l.:iJi I-r.. 1'i'


Table 25. Specie- :: 4 1i in height of Bo-qiie Senii--Deccliio Tipico 1

-i .. F I
-' r "' "H['.1' -[.ii"7 j~ I -

. .......

r ,







hiL "I-.

Table 26. Specie- :: 4 1 i n height of Bo-qiie Semni-Decidlio Tipico 2

H i'l.. ., .i:

F : I

-j -r Hr 1 j. r

-~~ ~ ~ *,", IIL:.

-1 1 t

*~ 1 L

.j. I-r.lii k


Table 2-. Specie- :: 4 In in hleigit of Bo-qiie Senni-Decidiio Quiebradn

H-ri'l.. . .i.

F : I

' -1 t ,

- ..-. I
.i - .' . .- 1,1

Table _8. Specie- : -1 ini in hei-lt of Bo qlle Siemlpre-\er(le

H--r .I'1. -. ..t: j : I
-'. "t:': : ' 1" ":' 'I'


. ..-
. 4-[ H' h -iiL I-r.. '

Table 29. Food trees ofBosque de Cerros (BSC)

Food Species Relative Frequency
Myrcia guianensis 7.93
Guettardia divaricata 6.91
Anona ahnii i i i 4.22
Genipa americana var cardo 2.36
Annona purpurea 2.19
Vitex capitata 1.52
Copaifera officianalis 1.35
Spondias mombin 0.51
Randia hebecarpa 0.51
Number of species that are food producers 9.00
Percent of all trees that are food-producers 27.49
Percent all species that are food producers 23.68

Table 30. Food trees ofBosque Seca de Mezcla (BSM)

Food Species Relative Frequency
Vitex cap itata 5.08
Copaifera officianalis 3.66
Randia hebecarpa 2.24
Genfpa americana var cano i i 2.03
Annona jahn/i 0.61
Capparis sp. 0.20
Cassia moschata 0.20
Number of species that are food producers 7.00
Percent of all trees that are food-producers 14.02
Percent all species that are food producers 23.33

Table 31. Food trees ofBosque Seca de Falda (BSF)

Food Soecies

Relative Frequency

Guettardia divaricata
Myyrcia gu ianensis
Annona purpurea
IGuazuma tomentosa
Spondias mombinab
Copaikfera officianalis
Genqpa americana var cando
Cassia moschata
Annona jahnii
Acromia aculeata
Vitex cap itata
Number of species that are food producers
Percent of all trees that are food-producers
Percent all species that are food producers


Table 32. Food trees ofBosque Semi-deciduo Seca (BSDS)

Food Species Relative Frequency
Guettardia divaricata 10.13
Gen/pa americana var carno 2.53
Spoendas mombin i i 1.90.
Annona purpurea 1.90
Sapindus saponaria 1.90
Myrcia guianens i i i 1.90
Copaifera officianalis 0.63
Number of species that are food producers 7.00
Percent of all trees that are food-producers 20.89
Percent all species that are food producers 24.14

Table 33. Food trees ofBosque Semi-deciduo Tipico 1 (BSDT1)

Food Species
Zizyphus cyclocardia
Caesalp ina coriaria
Capparis odoratissima
Genipa americana var carnto
Spond/as mombin
Copernicia tectorumw
Guazuma tomentosa
Guettardia divaricata
Pithecellobium tortum
Pithecellobium dulce
Pithecellobium saman
Annona purpurea
Sapindus saponaria
Pouteria glomerata
Brosimium alicastrum
Ficus maxima

Relative Frequency


Number of species that are food producers
Percent of all trees that are food-producers
Percent all species that are food producers

Table 34. Food trees ofBosque Semi-deciduo Tipico 2 (BSDST2)

Food Species
Zizyphus cyclocardia
Capparis odoratissimra
Pithecellobium tortum
Caesalpmia coriara
Pithecellobiwnum dulce
Copernicia tector.um.
Pithecell ob lu p istaciaefol /um
Guazuma tomentosa
Sapindus saponaria
Genpa americana var caruto
Brosimium al icastru
Pithecellobiwn saman
Pouteria .lomerata

Relative Frequency

Number of species that are food producers
Percent of all trees that are food-producers
Percent all species that are food producers


Table 35. Food trees ofBosque Semi-deciduo Galeria (BSDG)

Food Species Relative Frequency
Pouteria glomerata 28.63
Annona jahnii 2.07
Genpa americana var carno i i 2.07
Zizyphus cyclocardia 1.24
Copaifera officianahis 0.83
Pithecellobium ligustrium 0.41
Pithecellobium pistaciaefo i i 0.41
Pithecellobium tortum 0.41
Guazuma tomentosa 0.41
Randia hebecarpa 0.41
Myrcia guianensis 0.41
Number of species that are food producers i 11.00
Percent of all trees that are food-producers 37.34
Percent all species that are food producers 37.93

Table 36. Food trees of Bosque Semi-deciduo Quebrada (BSDQ)

Food Species Relative Frequency
Mangife ra indica 7.69
Anvnonajahnii 5.77
Guettardia divaricata 3.85
Spondias m ombainb 1.92
Copaftera officianalis 1.92
Genapa americana var carno 1.92
Vitex capitata 1.92
Number of species that are foodproducers 7.00
Percent of all trees that are food-producers 25.00
Percent all species that are food producers 31.82

Table 37. Food trees of Bosque Siempre-Verde (BSV)

Food Species Relative Frequency
Ficus maxima 5.99
Annonpa jahnii 4.87
Copaifera officianalis 4.87
AnPnona purpurea 1.87
Genipa americana var caruno 1.12
\Randia hebecarpa 10.75
Number of species that are food producers 6.00
Percent of all trees that are food-producers 19.48
Percent all species that are food producers 35.29

Table 38. Food vines of Bosque Seca de Cerros (BSC)

Food Species Relative Frequency
Comb return fnrticosum 14.81
Number of vine species that produce food 1.00
Percent of all vines that are food producers 14.81
Percent of all vine species that are food producers 16.67

Table 39. Food vines of Bosque Seca de Mezcla (BSM)

Food Species i Relative Frequency
Comb return fruticosum 66.67
Number of vine species that produce food 1.00
Percent of all vines that are food producers 66.67
Percent of all vine species that are food producers 33.33

Table 40. Food vines of Bosque Seca de Falda (BSF) .

Food Species Relative Frequency
Combretumr frticosum 4.05
Number of vine species that produce food 1.00
Percent of all vines that are food producers 4.05
Percent of all vine species that are food producers 11.11

Table 41. Food vines of Bosque Semi-deciduo Tipico 1 (BJSDT1)

Food Species Relative Frequency
Entaa polystacha 5.63
Combreturm ficos i i i 4.23
Philodendron acutatum 2.82
Number of vine species that produce food 3.00
Percent of all vines that are food producers i 12.68
Percent of all vine species that are food producers 23.07

Table 42. Fruiting phenology (and foliage phenology for selected species). Large X represents mature fruits, particularly those

available to terrestrial mammals. Large X also represents period of active growth of foliage in representatives from

Nlaranrncene, Arnceae, C'ollibretacene, and ; (Legiiiniiio-ne). Small \, relre-ent- poleilial tear-round i'e

of tubers ofMarantaceae, stems of Combretaceae and Araceae, roots of Entada and fleshy basal leaves ofBromeliaceae.

Tapocho de Monte
Tapocho de Monte
Espinito, Cabrito
Canafistolo burreto
Cauj ate

Mangifera indica
Spondias mombin
Annona purpura
Philodendron species
Acromia aculeata
Bactris guineenisls
Ccpernic2a fecfornum
Bromelia chrysanta
Bromela pinguin
Pereslna guamacho
Lycana pyr folha
Calathea latfohia
Maranta arundinacea
halia geniculata
Myrcia guianensis
Genipa americana var caruto
Guefftardia divaricata
Randia hebecarpa
Guazuma ftomentosa
iatex capitafa
Caesalpina coriana
Cassia grandis
Cassia moschata
Copalfera oiffcianals
Enftada poystacha
Enferolobium cyclocarpum
Pithecellobum saman


S afford
Moc & Sesse

(Jacq) Lood Mart
(L) E Moore
(H.B.K) Mart.
(Loefl.) Stuntz.
(Link) KI
(P.&E) Koern
(Aubl.) D.C.
(H.B.K) S chum
(HF G.B) Standl
(Jacq Willd)
(Jacq) Oriseb
(Jacq) Benth.

Anacardiaceae Tree
Anacardiaceae Tree














x x x x x x X X X X X X

X X x x x x x x x x X x

x x x x x x x x x x X
_xxx xxxxxXX

[x x x x x x x x x x x x

x x x x x x x X X X X X

Sx x x x X X

x K x K x X X X X X X

x x x x x x X X X X X X

xxx xx

x x x x x x X X X
x K K K
x x x xXXX

Table 43. Areal estimates of habitat types in Hato Pifiero. SS/CH = dry savanna with
chaparros. BS = dry forest. BS/AP = dry forest/crops. M = mangos. BSD = semi-
deciduous forest. BSV = evergreen forest. PS = dry pasture. SI = flooding savanna.
"Core" refers to areas in which hunting does not occur. "Poached" refers to areas where
it does. Open water estimate presented, but is an underestimate.


Core ha

Poached ha

Total ha

SS/CH includes 3,355.5 3,355.5 6,711
SS/CH all core 2,937 0 2,937
SS/CH Subtotal 6,292.5 3,355.5 9,648 st
BS includes poached 1,426.7 652.3 2,079
BS all core 2,907 0 2,907
BS Subtotal 4,333.7 652.3 4,986 st
BS/AP 238 238 476 st
M (Valle Hondo) 0 7.0 7 st
BSD includes 17,409.32 3,617.68 21,027
BSD all core 407 0 407
BSD Subtotal 17,816.32 3,617.68 21,434 st
BSV includes 0 11 Valle 11
poaching Hondo
BSV all core 37 0 37
BSV Subtotal 37 11 48 st
PS includes poaching 42.9 100.1 143
PS all core 1,663.0 0 1,663
PS Subtotal 1,705.9 100.1 1,806 st
SI includes poaching 21,542.4 1,314.6 _22,857
SI all core 1,842 0 1,842
SI Subtotal 23,384.4 1,314.6 24,699 st
Other side of Pao Outside of Study Area 16,743 st
Conspicuous open 123 st
Hato Pifiero D ST 79,970
Forest 26,951
Open 26,505
SS/CH 9,648
Outside of study 16,743

Inside study area




This chapter provides the foundation for discussions in subsequent chapters, as

well as context for the spatial characteristics of cat attacks on livestock. The prey base

study revolved around the following questions, which this chapter begins to address.

1) Is the natural prey base in the study area sufficient to support jaguar and puma

without a subsidy from domestic livestock? This can be addressed through the

standing crop biomass and annual gross productivity of important prey.

2) Is natural prey patchily distributed? What areas are most productive?

3) How do natural prey densities, distributions and/or vulnerability vary by

season (in response to water availability/inundation and/or plant phenology)?

4) Will jaguar and puma hunt opportunistically, taking prey in proportion to rate

of encounter, or are they selective (e.g selecting larger prey over smaller


5) How do the spatial and temporal patterns of primary productivity influence

prey distributions?

6) How does prey distribution influence jaguar and puma use of space?

7) How are 5 and 6 related to the interactions between cats and cattle?

8) If the frequency of cattle depredation has an inverse relationship with

availability and vulnerability of natural prey, is there a converse direct

relationship with availability and vulnerability of livestock?

This chapter will address questions 2 and 3 in full and questions 5, 7, and 8 in

part. The task of setting forth patterns of abundance, and biomass of the prey base was

sufficiently broad and lengthy that it required this chapter of its own. This sets the stage

for subsequent chapters that, with botanical and faunal background condensed, will be

able to examine biomass estimates and an array of questions in considerably greater

detail and freedom.


Linear Foot Transects

Reconnaissance using vehicles and ultra-light aircraft, and review of a

preliminary vegetation map, facilitated the design of a system of 26 foot transects. The

number of lines and kilometers sampled were distributed as follows: dry hill forests [4,

9.8 km]; semi-deciduous forest not adjacent to permanent water [3, 5.5]; semi-deciduous

forest near permanent cario or prestamo [4, 7.51]; interspersions of semi-deciduous forest

and savanna crossed perpendicularly [2, 4.5]; semi-deciduous forest edge, parallel to

pasture [1, 2]; small flooding savannas and managed pastures near forest [5, 5.65]; large

flooding savanna, close to forest [2, 4]; large open flooding savanna away from forest [2,

4.7]; higher elevation, non-flooding pastures [3, 3.55]. This design took in a cross-

section of habitat types, and facilitated tests of proximity to water, habitat interspersion

indices, and distance to cover. Transects actually employed each month depended upon

water levels. Restricted accessibility and laborious walking through water with tall

emergent vegetation during the rainy season, and destruction of markers by livestock

caused the savanna sample to be less complete than the forest sample.

The narrow footpaths cleared in forest with machetes were marked at 50 m

intervals. Cattle ate both plastic and metal tags in some pastures and savannas, forcing us

to resort to 1.8 m metal stakes topped with orange spray paint at 100 m intervals. Sign,

feces, and sightings of prey and predators were recorded along a systematic calendar of

morning and dusk walks. A subset of forested transects was originally walked at night

(beginning at 2200 hours) using adjustable beam Koehler Wheat lamps. Though

interesting, these walks were discontinued, as the exertion did not justify their low

productivity. Distance estimates, group number and dimensions, behavior, time, and

location were recorded for all visual observations. These data on terrestrial and arboreal

mammals, cracids, and terrestrial tortoises were used for habitat specific density

comparisons and coarser grained absolute densities via the models in DISTANCE

(Buckland et al. 1993). Analyses used perpendicular distances to centers of clusters.

Subjective degree of inundation by percentage and mean water depth was recorded every

time a transect was walked in the rainy season.

Vehicle Transects

One driver and two observers proceeded in a truck driven at 20 km/hr beginning

at 0600, 1800, and 2200 hours. During the night the two observers used spotlights

(200,000 and 400,000 cp) and headlamps. Two routes (high and low) were sampled

twice each (dry and wet season), while another route (large open savannas in the south)

was sampled once. All routes were approximately 16 km in length. Species, time, group

size, behavior, location and habitat type were recorded for each observation. Availability

of habitats was recorded for the high and low routes. The low route was extremely

heterogenous: small pastures and savannas surrounded by forest, strips and large blocks

of forest, permanent prestamos and caho, and low hills. Much of the low route flooded

during the wet season, some of the savannas retained water long after the rains ceased,

and surface water was available throughout the dry season, albeit, more widely dispersed

as the season progressed. The high route started out in similarly heterogenous mixed

forest and savanna types, but within 4 km entered the massif of El Baul. In the higher

hills, short dry forests and savannas on rocky soils with poor water retention and fertility

dominated. Pockets and strips of taller forests occurred along hills bases and valleys,

often in a dendritic pattern. The overall availability of surface water was lower on the

high route. The third route began in interdigitated savanna and forest, then skirted the

tongues of semi-deciduous forest that extended into the large savannas, and ended in an

open treeless expanse. Though these savannas flood completely, during the dry season

the only surface water available is at windmill pumps and prestamos.

Capybara and Caiman Counts

Several full-days of observation of capybara (Hydrochaeris hydrochaeris),

caiman (Caiman crocodiles), and turtles (Podocnemis voglii) at Lagunas Alta, Cerritos,

and Escorzonera during the 1996 dry season made it clear that capybara activity patterns

varied among sites. At times, capybara make heavy use of forest and shrub cover. Each

group has its own activity pattern. Consequently the timing of counts needs to be group-

specific and repeated counts often necessary to obtain confidence that entire groups have

been observed. The late dry season is optimal, as visibility and capybara concentrations

are at maximum. As water bodies gradually dry out, small groups of capybara merge into

larger groups in some areas, sometimes traveling several kilometers to do so.

During the first capybara census (April 2-23, 1997) every area on the ranch that

possibly contained capybara was visited (on foot or by boat), some up to 6-10 times.

With the more productive hours and occupied sites identified, a second census (April 13-

20, 1998) was more efficient. With the help of Diego Giraldo (Universidad Simon

Bolivar) capybara were classified as: male or female adults (>40 kg); juveniles ( 20-30

kg); infants ( 5-15 kg, born preceding October-November); newborns (2 kg ); or adults

gender undetermined. Caution was necessary to avoid excessive estimates by double

counting more mobile groups in semi-forested areas. Where poaching was taking place

late night visits were required. Caiman were sometimes counted during the same visits.

Age specific counts were used to generate biomass estimates.

In 1985, Allan Woodward and Dennis David, both competent crocodilian

specialists, assessed Pifiero's entire caiman population. In 1986, Lee Fitzgerald, also a

competent crocodilian specialist, conducted another ranch-wide survey. With their

information available, I limited my night-counts to a sample of the same areas that

Woodward, David, and Fitzgerald had sampled, using my counts as a calibration factor, if

necessary, for the counts made 10 years previous. In 1996 and 1997, we calibrated our

eyes to the size classes used by the Venezuelan government agency PROFAUNA

(Ayarzaguena 1983; Thorbjarnarson, 1991a; Thorbjarnarson & Velasco 1998; Velasco &

Ayarzaguena 1995) by estimating animal sizes at night, then capturing the animals with

nooses and measuring and releasing. Subsequently we made night counts at Cano de la

Canoa, Tapa de Los Patos, Rio Pao, Caho Caujaral, Lagunita Escorzonera, Cano de la

Iguana, Lagunas Alta & Cerritos, and Laguna Grande. Additional counts were made at

Caho Manglarito during our capybara censuses and by Juhani Ojasti's wildlife

management class. With the exception of those counts made during the April capybara

census, most caiman counts were made in the month of March. On three occasions I

conducted sighting fraction experiments, capturing caiman, then releasing them with

small cyalume lights attached to their head, counting visible lights every 15 minutes for

several hours. The ratio of average number of lights visible to the known number of

lights provided an estimate of sighting fraction. This manipulation provided an

opportunity to record weights in conjunction with the snout-vent lengths used to

determine size classes. Later, when it became apparent that a caiman's head is the only

body part that retains its original size and shape after predation by jaguar, it became

necessary to capture a few more animals to calibrate skull length with size class with

weight. Size class distributions were based on a sample of 5,998 animals. Size class-

weight relationships were based on project captures and data in (Ayarzaguena 1983) and

(Thorbjarnarson 1991a). Biomass was estimated as the product of: 1) numbers in size

classes; and 2) average weight per size class.

Capture-Mark-Release-Recapture: Tortoises, Turtles, Anacondas

Morrocoy tortoises (Geochelone carbonaria) were frequently encountered along

forested transects during moist months. Between April 20 and June 14 of 1997, we also

captured, marked, and released tortoises in a 42.5 ha square plot in forest adjacent to a

transect. Cleared and flagged trails divided the completely forested plot into 35

approximately square blocks. Block interiors were searched between 0730 and 1000

hours. Each tortoise captured was given an individual numerical marking via notched

marginal scutes, allowing a record of its capture history. Additional morrocoy

measurements and observations were recorded by Tibisay Escalona in 1996, and our

crew in 1996 and 1997. Average weights derived from 87 captures were combined with

population estimates generated using DISTANCE (Buckland et al. 1993) density

estimates and Schnabel's multiple capture-recapture model (Seber 1982) for biomass


We conducted mark-recapture exercises with galapago turtles (Podocnemis voglii)

in prestamos in two pastures, Lagunita Escorzonera, and a section of Cafio Caujaral. We

used haul seines in the prestamos, chicken wire funnel traps in the caho, and funnel traps

and short nylon hoop traps in Escorzonera. The marginal scutes of Escorzonera

galapagos were notched to identify their order of capture. White epoxy enamel marks

painted on the carapace also provided identification. Three days of basking counts were

made at both Laguna Alta and Escorzonera. Three days of head counts in the water were

made at Escorzonera. After Escorzonera, with abundant measurements in hand, prestamo

and cario galapagos were simply marked with a small hole drilled through a rear

marginal, and subsequent holes added for each recapture. The freshwater turtle capture-

recapture efforts all took place between mid-April and early May. The lowest water

levels of the year and separation of first and second samples by no more than 6 days

guaranteed that the populations were closed. Average weights derived from 181

measured turtles were combined with Chapman's adjusted Lincoln-Petersen Estimator

(Seber 1982) for biomass estimates.

Incidental to work on mata mata (Chelusfimbriatus) reproduction, we caught,

marked, and released 36 of these large turtles (up to 13 kg) during October and

November 1996 (high water) in Cafio Caujaral. Most successful for mata mata was a

long stick with a shark hook attached, basically a thin gaff used when mata mata were

breathing near the surface. This did not cause a single injury. Large mesh sinkerless flag

gill nets also worked, but caught an equal proportion of mata mata and large freshwater

stingrays (Potamotrygonidae).

Anacondas (Eunectes murinus) up to 50 kg were also captured, measured, and

released in 1996 (Rio Pao and Cafio Caujaral). In 1997, small samples were marked in

Rio Pao and Cafio de la Iguana.

Additional Brief Assessments

In February 1996, ten prestamos were haul seined and mass and species

composition of fish and turtles recorded. Hoop nets employed in the Pao River and

trammel nets employed with Cafio Caujaral provided some familiarity with the fish

fauna. Iguanas (Iguana iguana) were counted, by boat, along Cafio Caujaral, Cafio de la

Iguana, and Rio Pao in 1997. During 1997 and 1998, 90+ trap nights with large nationals

(98 cm long x 52 cm tall x 40 cm wide 50% baited with fruit, 50% with dog food and

sardines), 120 trap nights with mid-sized tomahawks (61-66 cm x 17-25 cm x 17 cm

baited with dog food and sardines), and 369+ trap nights with Shermans (23 cm x 9 cm x

8cm baited with oat mixes and queso llanero) provided some additional insights on

terrestrial vertebrates. A standardized short form was used for opportunistic observations

while driving or during other field activities, providing important additional information

on group sizes and habitat affinities.

Camera Traps

Animal tracks were abundant in moist soft substrates during the rainy season, but

only large hooved animals left good tracks in the forest during the dry season. Camera

traps were used in several areas during the dry season, to complement the visual

observations recorded on walking transects. TrailTimer infrared sensing camera trap

systems were combined with inexpensive (<$100) Canon, Kodak, and Vivitar cameras

loaded with 400 ASA Fujichrome film. During the 1997 dry season, exploratory work

took place along water holes in drying Cafio Caujaral/Roseta. During the 1998 dry

season, up to 12 units were employed along retired foot transects. For two weeks, ten

cameras were divided between two transects (low elevation semi-deciduous forest vs dry

hill forests). During that time, the remaining free cameras were employed along drying

creek beds, a jaguar kill, and prestamos. During the remainder of the dry season all units

were employed along a foot transect paralleling Cafio Caujaral through semi-deciduous

forest. As the next dry season (1998-1999) began, 12 to 14 units were employed along

the same three transects, and also along a narrow forested stream valley in steep hills. In

January 1999, all units still functional were deployed opportunistically along known

animal travel routes near drying carios and prestamos and baited (plantains, bananas,

mangos, guavas, fish, dog food, raw beef, oat-banana-peanut butter mixes, and salt) for

two weeks.


Distribution of Animals According to Encounter Rates

Vehicle transects

With the caveat that transects along dirt roads result in a bias towards edge-

preferring species, and miss details that foot travel in irregular topography can detect,

vehicle transects provided useful sketches of general patterns. During the dry season,

jaguar prey biomass was highest along the well-watered low route consisting of

interspersed forest, savannas, and low hills. Capybara, white-tailed deer (Odocoileus

virginianus), and collared peccary (Tayassu tajacu) were more numerous on this low

route than along the route dominated by hills, rocky soils, dry forest types, dry savanna,

and less abundant surface water (Table 44, Fig. 11). However, rabbits (Sylvilagus

floridanus) were more abundant on the high route and the Sylvilagus detected on the low

route were using the edges of a small hill. Crab-eating foxes (Cerdocyon thous) were

only slightly less common on the high route than the low.

Observations in the large, seasonally flooded savanna during the dry season were

dominated by deer. Capybara were scarce there because there were no large ponds on the

route (Fig. 11). The deer were detected near the savanna/forest ecotone. During the heat

of the day they were in shade, not exposed savanna. Crab-eating foxes were common in

the savanna.

These general distributional patterns continued during the rainy season, but

animals became more dispersed as surface water and green forage became more widely

available. Vehicle transect data indicated no massive exodus from low areas in response

to the shallow flooding. Instead, animals appeared to "spread out" within large annual

use areas, making use of more dispersed resources, fewer surface water constraints, and

the phenology of favored plant foods. Interesting side notes are that ocelots (Leopardus

pardalis) were using savanna (though not common, this occurred along foot transects as

well) and tayra (Eira barbara) used dry grassy hillsides on the high route during the rainy

season. The latter was presumed to be associated with nearby strips of semi-deciduous

forest in valley bottoms.

Foot Transects and Camera Trapping

The primary intent of transects was abundance estimates of select prey species,

and thereby, availability. However, contrasts in animal distributions within and among

transects illustrated the patterns of prey distributions across the landscape.

Contrasts in encounter rates over a variety of habitats are presented in Fig. 12.

Diversity (and biomass) was high on La Candelaria transect (21 sp.). In this wide tongue

of semi-deciduous forest, connected to larger blocks, strongly-forest dependent species

such as tapir (Tapirus terrestris) occurred, but the nearby edge resulted in high numbers

of white-tailed deer. Deer numbers were lower on Caujaral Norte (20 sp.), but a large

group of white-lipped peccary (Tayassu pecari) appeared to offset the drop in cervid

abundance, particularly considering that the bars exhibited are independent of group size.

Although this white-lipped herd used the area most heavily in the dry season, animals and

sign were detected during the rainy season, albeit in more scattered locations, and with

clear evidence of use "outside" the sampled area. Collared peccaries were less frequent

in Caujaral Norte than La Candelaria, and when group size is considered, even more so

than Fig. 12 suggests. Jaguar and tapir sign frequencies were highest in Caujaral Norte,

and ocelots (Leopardus pardalis) abundant, particularly near the caho. Howler monkeys

(Alouatta seniculus) and yellow-knobbed currasow (Crax daubentoni) were also more

common in Caujaral Norte. In vegetation-based cluster analyses, many forest plots from

Candelaria and Caujaral Norte fell within the same clusters. Composition of forest plots

near the stream bed of Cafio Caujaral Norte did differ from La Candelaria. The Caujaral

Norte transect was 100% forested, while La Candelaria was 86%. The remaining 14% of

La Candelaria was small pockets of savanna, locally called calcetas. Mean distance to

nearest edge recorded at 100 m intervals was 1448 m (n=27) for Caujaral Norte and 247

m (n=22) for La Candelaria.

Of the nine transects presented in Fig. 12, collared peccaries occurred on seven,

white-lipped peccaries on one. The size of the contiguous cahio-side forest in Caujaral

Norte (see above interspersion index), its connection to a long riparian strand extending

north of Pifiero, the proximity to permanent water (means: 68 m Caujaral Norte: 512 m

La Candelaria: 653 m Bosque de Los Cerritos; 279 m Lado de Cerro Guaical; 522 m

Encima de Cerro Guaical), as well as forest composition in damper areas all may have

played a factor in determining the areas where white-lipped peccaries concentrated.

Collared peccaries used forests in the hills (Bosque de Los Cerritos, Lado de

Cerro Guaical, Encima de Cerro Guaical, and Bosques de Las Penitas, Fig. 12). White-

tailed deer, Dasypus novemcinctus, Geochelone carbonaria, and other taxa also used hill


Puma sign occurred on four transects in Fig. 12, jaguar sign on three. Caujaral

Norte and La Candelaria forests (jaguar and puma presence recorded) were 100%

relatively high stature (mean canopy height and overstory tree DBH 19 m 47 cm and 18

m and 39 cm, respectively) bosque semi-deciduo (BSD). El Bosque de Los Cerritos

(only puma recorded) crossed a number of forest types as it ascended and descended a

hill. While the first 350 m ofbosque siempre verde (BSV) had a canopy height of 16 m

and overstory tree DBH of 27 cm, the bosque seca types (BS) had canopy heights of 12,

7, and 4.5 m, and the well-represented sabana seca con chaparros (SS/CH), canopy

heights under 4 m. On El Lado de Cerro Guaical (only puma recorded), also a mix of

types (81% BS 9.5% BSD 9.5% SS/CH), mean canopy height was 10 m, mean overstory

tree DBH 20 cm. Las Penitas (jaguar presence recorded) was a mix of short dry hill

forest (BSC and BSF), high dry pasture, pockets of tall semi-deciduous forest (BSDS),

and sections of sabana seca con chaparros (SS/CH). Mean canopy heights and mean

overstory tree DBH measurements were as follows: BSC 4.37 m 6.8 cm; BSDS 17.5 m

28.2cm; BSDS 13 m 21 cm; SS/CH 4.4 m 14.75 cm. While a jaguar was seen on the dry,

mostly open Las Penitas transect (Fig. 12), it was probably not focusing on SS/CH or dry

pastures, it may have been interested in the hill-base pockets of BSD or hill flank BS

forests where collared peccaries and rabbits occurred (camera trapping and foot transect

data), but it probably was enroute to a valley filled with moist seasonally flooded forest

with permanent water where it was frequently radio-located.

Diversity declined as elevations increased (Bosque de Los Cerritos, Lado de

Cerro Guaical, Encima de Cerro Guaical Bosques and Pastos de Las Penitas Fig. 12).

Collared peccaries did use hill sides covered by bosque seca types but rarely occurred in

sabana seca con chaparro types (Fig. 13). On a 2750 m transect that crossed a hill (Fig.

13, Bosque de Los Cerritos), all species of interest clustered in the BS types and were

rare in SS/CH. Cebus olivaceous sometimes used hill side forests, specifically, richer BS

mixes. Both cebids clustered at lower elevations closer to permanent water. White-tailed

deer used both BSD and BS but appeared to avoid ridge-tops (Figs. 12, 13). Puma were

recorded on this hill trail. Jaguar were not. The transect Encima de Cerro Guaical (Fig.

2) had pockets of BSD in short hill-top valleys. This sort of inter-digitation of habitats

facilitated a wide distribution of animals.

Chacalacas (Ortalis ruficauda) exhibited a slightly broader ecological amplitude

than yellow-knobbed curassows, better tolerating drier forest types, open areas, and

distance from water (Figs. 12, 13,14). Nine-banded armadillos (Dasypus novemcinctus)

used both BSD and BS (Figs. 12, 13). Dasyprocta agouti used both BSD and BS but

seemed to prefer lower elevations (Figs. 12, 13). Agoutipaca was rare on the ranch. The

few sightings that occurred (both along riparian zones) did not fall within foot transect or

camera trapping sampling efforts. Cerdocyon thous was very common in Pifiero. It

occurred in all forest types except SS/CH but was clearly a creature of edge, occurring

commonly along savannas (Figs. 12, 13, 14). Procyon cancrivorous appeared

uncommon in comparison, and used low elevation BSD and pasture edges. Leopardus

pardalis occurred in wet and dry forests, and also made use of savanna and pasture edges

(Figs. 12, 13, 14); logical given the high seasonal densities of rodents such as

Zygodontomys brevicauda in those habitats.

Myrmecophaga tridactyla and Tamandua tetradactyla occurred in both BS and

BSD, butMyrmecophaga made more use of short stature dry forests on hills than the

more arboreal Tamandua (Fig. 12). Geochelone carbonaria used both BSD and BS types

(Figs. 12, 13). Eira barbara preferred BSD and proximity to water (Fig. 12, but made

forays into more open habitats (Figs, 12, 14). T. tajacu frequently occurred along the

edges of savanna (Figs. 12, 14). Crossings were not often seen, but clearly occurred (Fig.

15). Capybara were counted directly because the transect system was not designed to

estimate their abundance. They were common in Potrero Los Venados (Fig. 12) a small

savanna dotted with prestamos and close to forest: a habitat matrix. They also occurred

in the forests near the water bodies Tapa Escorzonera and Laguna Cerritos (Figs. 12, 13).

The phenomenon of capybara in forest near water was a common one: project observers

posted at water holes in the dry season noticed that some herds spent most of the day in

forest cover (preyed upon by jaguar in same area). Capybara also occurred in more open

savannas such as Mata de Guafa 2 (Fig. 14).

Overall diversity was lower in savannas than forests. The three pasture transects

(Los Venados, Los Cerritos, Claro Cerrillos) in Fig. 12 represent a gradient from: 1) low

elevation savannas dotted with prestamos and adjacent to BSD nearby; to 2) higher

elevation pastures with scarcer surface water, poorer soils, adjacent to BS. White-tailed

deer were very common on the Los Venados transect (mean distance to BSD 116 m,

mean distance to permanent water 147 m). Densities were lower on the Los Cerritos

pasture transect, where high dry soils were ameliorated by proximity to a lake and water

tanks (mean distance to water 650 m) and annual applications of fertilizer, and proximity

(mean 86 m) to BS. Prey availability on the Claro Cerrillo transect (Fig. 12) was clearly

low. Though mean distance to ecotone was only 19 m, that species-poor narrow-strip of

forest lining a seasonal rivulet was primarily productive in the rainy season. The fruits

available then (Annonapurpurea, Annonajahnii, Lycaniapyrifolia, Table 42) were

presumably related to the concentrations of collared peccaries (groups as large as 42) in

the area that time of year. During the dry season, all animals were scarce in this area.

Mean distance to water was 800m and soils were poor.

Similar patterns of prey distribution in savannas are presented in Fig. 14. These

are all low elevation savannas, with the gradients being indices of interspersion and

access to water. Mean distance to ecotone, proceeding from Juncal Saman Gacho (rear)

to Guanabano Abierto (front) were: 84, 30, 87, 305, 352, and 765 m respectively. Mean

distance to permanent water in the same sequence was 570, 341, 486, 303, 269, and 1772

m respectively. White-tailed deer abundance was limited by distance to cover and water

during the dry season. Prey diversity and abundance in wide-open and poorly watered

savannas was very low. Conversely, prey were abundant in small (frequently not over 1

km or less in any direction) savannas surrounded by strips and blocks of semi-deciduous

forest. Caiman and turtles in the prestamos dotting such savannas added to overall prey


The Los Venados transect (Fig. 12) was an example of such a situation. Deer

densities were high, there were good numbers of capybara, caiman, and freshwater

turtles, and collared peccary were using the forest edge. Radio-locations of cats in the

forests adjacent to these small savannas attest to their importance to felids (Daniel

Scognamillo, Ines Maxit, Laura Farrell pers.comm.). Worthy of mention is that

Guazuma tomentosa, the most important mid-dry season food for peccaries, though a tree

of semi-deciduous forest, prospers at the forests' edge.

Figure 15 exhibits felid and prey distributions along a 2650 m transect crossing

sabana arbolada and BSD alternatively. The majority of animals, and hence, prey

diversity clustered in and around the forest. Deer and chacalacas used sabana arbolada

more than the other species, though anteaters also crossed it. Deer observations also

clustered around the permanent water source. Jaguar tracks were observed at the

transect's beginning and end. This area was part of a male jaguar's circuit, and tracks

were often seen on nearby dirt roads, sometimes for several continuous km.

Camera trapping data from deep forest and the cario bisecting it (Fig. 16)

demonstrated a species composition very different from those of hills (Figs. 12, 13).

Yellow-knobbed curassow were common, chacalaca absent. White-lipped peccary were

abundant, collared peccary relatively scarce. Deer were far scarcer than in more

heterogenous areas. Ocelots were common. Never common, tapir were present.

Although sample sizes were small for large cats, in these habitats, the number of jaguar

images was twice that of puma images. The frequency of crab-eating fox images was

surprisingly high given all other data demonstrating the species proclivity for edge (Fig.

16). Nearly every fox photographed was on the dry cario bed. Annuals establish

themselves rapidly as waters recede and opportunities for scavenging are high where

aquatic forms are being stranded. The drying cario bed was clearly a travel corridor for

ocelots and white-lipped peccaries. It may have constituted a rich seasonal edge for the


Giant anteaters were not infrequent in deep forest (Figs. 12, 16), thus they

occurred in nearly every terrestrial habitat in Pifiero. Wading birds and other large birds

were common in and near the cailo. Pooled camera trapping images from drying carios

(other than Caujaral) and prestamos (in or adjacent to forest) during the dry season

(n=90) were 40% wading birds, 7% crab-eating raccoons, 5% jaguars, 3% crab-eating

foxes, and 2% pumas. Of all images taken in hills (n=30), 40% were crab-eating foxes,

20% giant anteaters, 13% rabbits, 13% collared peccaries, 10% ocelots, and 3%

chacalacas. Wading birds were common in Pifiero, but neglected by most of my

methodology. Frequencies from camera trapping (Fig. 16 and preceding image data in

text) are the only quantification I have. Suffice to say, rolls of film in camera traps

placed to view the water's edge ran the risk of being filled with wading bird images.

Seasonal Changes in Distributions, Densities, and Group Sizes of Prey

Use areas of the amphibious capybaras retracted during the dry season, and

expanded during the wet. Small groups of capybara aggregated during the extreme end

of the dry season, coalescing into larger groups in excess of 100 animals. Freshwater

turtles were very concentrated (maximum estimated at 1,466/ha) in the dry season.

Marked turtles were observed 5 km from their dry season ponds during the rainy season.

The highest caiman densities I recorded were 609, 690, 653, and 958 per hectare during

the dry season. The rainy season's shallow flooding facilitated dispersal and by July and

August, caiman were encountered on foot transects that were dry land at other times of

the year. There also was overland movement of caiman in the dry season. As the quality

of smaller water bodies diminishes, some caiman move to more permanent habitats. The

shallower and hotter the water the more easily the caiman are prompted to move. Some

caiman embedded themselves in rubbish in forests along drying cahos, waiting for the

change in weather. Swimming a haul seine through a prestamo in the rainy season was

illustrative. In the dry season each haul yielded kilos of fish, turtles, and young caiman.

In late June it yielded nothing, while fish swam in an adjacent road bed. Distributions of

amphibious and aquatic prey were very strongly influenced by season.

Changes in seasonal ranges were less striking in ungulates. White-lipped

peccaries frequented mud-wallows during the dry season. Despite the species reputation

for long-distance movements, group locations were fairly predictable in the late dry

season. Heavy infestations of ticks in their intensive use areas near mud wallows

suggested that the mud was either a barrier or a relief for tick bites. All indications

(cameras, transects, opportunistic observations) were that the white-lipped peccaries used

a larger area during the wet season than the dry. Pooled observations suggest that their

dry season home range was a subset of their wet season range.

Seasonal shifts in white-tailed deer and collared peccary ranges were less obvious.

Observations of deer during the dry season frequently revolved around water (prestamos

and water tanks). Yet, deer were in the same general areas during both seasons. Group

size did not differ between seasons (Table 45). Smaller group sizes recorded on transects

than in opportunistic observations could be due to bias for larger groups in opportunistic

observations, but the lower proportion of juveniles observed on transects suggest poorer

detection was the cause. Even in open habitats, the foot traveler created more of a

disturbance than the vehicle rider, (a mild wake of disturbance through savanna grasses)

and in many cases, the observer's height was lower and thus more obstructed. The

proportions of juveniles were slightly higher during the rainy season, concurring with the

observations of Brokx (1972) that white-tailed deer breed year round, with a pulse of

parturition in the late rainy season. The proportion of juveniles was also similar to that

found by Brokx (1972). Widespread surface water in the wet season resulted in shifts,

but not large movements. As Brokx (1972) put it "the interspersion of habitats in many

areas was such that the seasons merely shifted emphasis from one part of the home range

to another".

Mean collared peccary group sizes recorded on transects were significantly larger

( p<.05, T =1.74, 42 degree of freedom) in the dry season (6.77) than wet (4.93).

Robinson and Eisenberg (1985) had observed the same phenomena on their grid system

in Hato Masaguaral (Table 46). Mean group sizes recorded in opportunistic observations

were consistently larger than observations made on transects, and dry and wet group

sizes roughly the same (Table 46). Mean group sizes recorded were larger on the few

occasions when peccaries crossed open areas on transects (Table 46). Proportion of

young observed on transects was less than the proportions recorded in opportunistic

observations. The differences between opportunistic observations and transect

observations, and between observations in the open versus in forest imply detection

factors. First-hand knowledge of rainy season under storygrowth obtained from cleaning

transects causes me to discount the statistically significant difference in group sizes

between dry and wet as an artifact of better visibility in the dry season. Reproduction of

collared peccary appeared to be year-round, and though Castellanos (1982) identified a

parturition pulse in the transition between the late dry and early wet season, I could not

attest to its validity (Table 46).

Without telemetry data or repeated observations of marked collared peccaries

throughout the year, a definitive statement on how the seasons affected home range is

impossible. Since different collared peccary groups occupied very different sorts of areas

within the single study area, a tight generalization might be difficult even with better

data. Table 47 is a list of seven areas in which, between foot transects, phenology trails,

and opportunistic observations, we recorded collared peccary in both wet and dry

seasons. The habitats that these different groups occupied ranged from very low-

elevations subject to considerable flooding to low ridges that never inundated and the hill

sides and valleys between. Collared peccary presence was noted in forests inundated up

to 70-90% with an average depth of 10 cm. Seasonal home-ranges of collared peccary in

Hato Pifiero presumably echo those in Hato Flores-Moradas (adjacent to south boundary

of Hato Masaguaral), where Castellanos (1982) found that the mean wet season home

range was three times the mean dry season range (Fig. 17). Within that average there was

considerable variation. Some seasonal ranges were roughly equal in size, but with

different foci. The distribution of water over the landscape during the rainy season would

seem to facilitate dispersal. However, rising water can also present an inconvenience as

areas elevated above widespread water contract. Foot travel on transects provided a first-

hand view of how limited the peccary hoof-marked pockets of forested land became in

lower plains. In some cases, patchy plant distributions and moisture-facilitated increases

in primary production could require less searching than in the dry season. Although it

can be presumed that most collared peccary groups were less confined by availability of

water or wallows after the arrival of the rains, the heterogeneity present in Pifiero

precluded a blanket generalization.

Abundance and Biomass

Counts revealed capybara to be less abundant than their highly visible nature (in

comparison to peccary) and proclivity for semi-open to open habitats had suggested

(water is an open habitat even when lined with forest). Capybara sometimes spend

considerable time in forest cover. Despite that, our familiarity with the study area,

activity patterns, and repetitious counts, yielded a confident estimate of 547 animals for

the entire study area (Table 48). Some examples of mean, median, max and min dry

season group sizes are 15, 15, 40, 1 (n=27) in Los Venados/Juncal, 10, 10, 23,1 (n=15) in

Escorzonera/Guaical, and 31, 16, 144, 1 (n=34) in Los Patos/Puente Benjamin. Taking

population structure into account and assigning weight estimates resulted in a biomass

estimate of 20,315 kg for the 63,227 ha study area.

April counts measure minimum biomass. Adults lose weight during the late dry

season due to reduced forage quality. Mortality is highest during the dry season. And

although some reproduction can occur year-round, with capybara there is a very clear

birth pulse between September and December, with the peak in October and November.

All those factors point towards maximum biomass at the end of the rainy season. Several

factors soften the contrast between the maximum and the minimum. Newborn weigh

approximately 1.5 kg, not a huge increment of increase. During the dry season, those

same first-year animals are growing, as are the animals approaching their second year

(maturity), even if adults can lose as much as 5% of their weight (Ojasti 1973). Taking

these overlapping factors into account results in a maximum of 22,654 kg in November.

The increase is due to five months (June through November) when high forage quality

and availability as well as the addition of new (albeit small) animals allow weight gains.

Caiman densities varied a great deal. The two larger lakes set amidst hills

appeared oligotrophic. Laguna Grande densities ranged between .875 and 1.61/ha.

Laguna Alta densities ranged between 37 and 44/ha. Surface area and total volume of

smaller water bodies were very dynamic. The drying of very small ponds and shallow

cahos and severe shrinking of permanent pools resulted in very high densities (e.g. 609-

958/ha). Caiman habitats are in a constant flux. Along cahos, stretches dry out

completely forcing caiman to walk to the next pool. Sampling in the deeper more

productive pools or sections of any caho can generate a misleading extrapolation for the

entire water body. Walking many kilometers along or in drying cahos helped generate

realistic estimates. Prestamos and cahos with abundant shelter, such as short emergent

aquatic vegetation or woody debris became havens for the year's young caiman. Many

productive deeper pools with numerous large animals and with scant hiding cover saw a

marked reduction in the proportion of the same class of young animals between the

beginning and end of the dry season. Situations where caiman biomass became very

concentrated, arose when the numerous animals inhabiting a long cano or wide savanna

during the wet season were drawn into a single pool as the dry season progressed.

A caiman biomass estimate generated from dry season counts fails to address

seasonal variation in biomass. Reduced stress during the wet season and the October-

November birth pulse might result in peak biomass in November. Unfortunately caiman

dispersal during the wet season is so extreme that numerical evaluations are impractical.

The caiman biomass estimate of 167,827 kg in the 63,227 ha study area, based on four

years of counts, an estimate of 15,408 animals for the study area, size class proportions

based on 5,998 observations, and 407 ha of late dry season water surface (Table 49) will

have to suffice.

Two groups of white-lipped peccary used Pifiero. The larger group

(approximately 107 animals) frequented the forest surrounding the northern section of

Cafio Caujaral (BSD with a high percentage of briefly deciduous species). Two transects

ran though this area and I deployed camera traps on and off of transects (Figs. 12, 16). In

the dry season, the peccaries made heavy use of mud wallows as pools sequentially dried

out. The group sometimes traveled along the dry cario bed, but more frequently wove in

and out using trails that were easily recognizable due to the volume of traffic and tracks.

Some pockets of forest in some (not all) of this area were cool and green overhead during

the dry season: a perception given validity by daytime Geochelone carbonaria activity

when the tortoises were embedded in cover elsewhere. The local name of the core area,

"La Roseta" referred to how the cailo, with a single canal downstream, branched out into

numerous active and retired stream beds (ramales) in the north. These areas were low,

and hence subject to considerable inundation. Small hills penetrating this deep green

strip where it bisected the massif of El Baul presumably relieved the peccaries during the

wettest months, and would be expected to impart dietary variety. The group was less

confined during the rainy season, though the full limits of its movements an unknown.

Despite the Candelaria transect's three years of existence, T. pecari sign was observed on

it for the first time during botanical sampling in late 1999, testimony to the importance of

long-term studies and risks of short-term evaluations. Because faunal sampling on the

transect had been retired, this one-time event is not shown in Fig. 12. Nor does the

observation detract from conclusions drawn from Fig. 12. The presence of white-lippeds

was palpable in Caujaral Norte/Rosetta. The forest was full of trails. The creek bed was

modified (Figs. 12, 16).

The smaller group of white-lipped peccaries, approximately 60 animals, used the

northeastern corner of the ranch. Observations of animals and sign ran from the BSD

laced rim of oligotrophic Laguna Grande through a large block of semi-decidous forest

between the lake and Rio Pao. This area also was subject to inundation but was sprinkled

with a few low hills. It also had an old river bed, Cafio Rosario, that though technically

dry much of the year, maintained green vegetation longer than the surrounding forest, and

was lined with a disproportionate amount of saman trees (Pithecellobium saman). This

group used mudwallows in Cafio Matajei a tributary of Rio Pao, but avoided the river,

perhaps due to the heavy traffic of poachers. Although, I was personally less familiar

with this group, it was studied by Barreto et al. (1997) and Hernandez et al. (1995).

Good group counts of white-lipped peccaries are difficult to obtain. In flight there

is chaos. Even without flight there is a good deal of coming and going, usually in fairly

dense cover with narrow openings. Group structure has rarely been reported in the

literature (Barreto & Hernandez 1988; Bodmer et al. 1997a; Fragoso 1998; Fragoso 1999;

Hernandez et al. 1995; Kiltie & Terborgh 1983; Mayer & Wetzel 1987; Sowls 1997).

Pooling averages of four methods of estimates of population structure in Pifiero yielded

71% adults and 29% juveniles. Kiltie and Terborgh (1983) reported that in 60

observations in the Peruvian Amazon the proportion of peccary youngsters was usually

less than 20%. Assigning weight to two size classes resulted in a biomass estimate of

5,005 kg for the 63,227 ha study area (Table 50).

The most common freshwater turtle in the study area, the galapago (Podocnemis

voglii) is an impressively adaptable small-bodied generalist. it occupied a variety of

habitats (prestamos, lakes, cahos, and rivers). The larger bodied and more specialized

terecay (Podocnemis unifilis) had a more restricted distribution. Basking counts along

the Rio Pao were 7.5% P. unifilis : 92.5% P. voglii. The composition of captures in the

deepest remnant pool in Cafio Caujaral at the end of the dry season were 27% P. unifilis :

73% P. voglii. During high water, the cario was connected to the rivers because all the

lower elevations on the ranch were flooded. During the dry season it simply ended in a

section of semi-deciduous forest locally referred to as Rabo de Agua (tail of the water).

The P. unifilis in Rio Pao were residents and reproduction was occurring. Those in Cafio

Caujaral were smaller individuals that had been stranded as the high water receded. The

high proportions in the pool sampled were a reflection of the species' preference for

deeper water. The terecay were forced into the pool, while presumably a number of P.

voglii remained in shallower sections of the cario and some may have simply embedded

in mud to wait out the dry season. Though a few introduced P. unifilis were also seen in

Laguna Alta, the species could be considered rare in Pifiero. P. voglii nests between

October and February. The greatest frequency of recently predated P. voglii nests was

during late December and January

Mata mata turtles (Chelusfimbriatus) were quite common in Cafio Caujaral. In

less than 200 m of Cafio Caujaral three of 38 captures in relatively high water between

October 27 and November 30, 1996 were recaptures. Individuals were sometimes found

wandering forest in the late dry season, presumably looking for the nearest pool with

water. In December of 1999 I found 5 shells of 5 adult C. fimbriatius in a circle of 4-5

meters in forest just beyond the bank in the northern part of Cafio Caujaral. The remains

were old, and no sign of the presumed predator present. Mean weight of 36 mata mata

was 6.894 kg (n=36). Maximum was 13.5 kg. Females captured October 27 and 31

contained eggs. Clutch sizes of the brittle spherical eggs obtained by induced oviposition

were 13 and 8. Mean diameter and mass of 5 eggs were 37.9 mm and 33.5 gm. Since all

females were palpated, indications were that the nesting season was largely over by

November. Mata mata were presumably in the rivers, but we had no indication that they

were in the prestamos. Kinosternon scorpiodes was encountered several times on

transects, at savanna edge during the rainy season.

In some water bodies (e.g. Lagunita Escorzonera and Cafio Manglarito) P. voglii

reaches impressive concentrations. The modified Lincoln-Petersen estimator developed

by (Chapman 1951) and the variance calculations for it from (Seber 1982), as discussed

in (Lancia et al. 1994) yielded an estimate of 1,466 galapago/ha (95% confidence interval

(CI) of 450-2,482) from a total of 188 captures in Escorzonera. When 5/188 turtles are

recaptures and the center of the lagunita is speckled with bobbing turtle heads, it is clear

that numbers are high.

Table 51 presents results of late dry season capture-mark-recaptures in two typical

prestamos and a deep pool in Cafio Caujaral. Ramo (1982) estimated 377.8 galapago/ha

(95% CI 296.2-482.9, n=526) in a cario in Hato El Frio in Estado Apure. The variety and

number of ephemeral and permanent water bodies in our study area was staggering. The

limitations of extrapolating from a sample of three typical water bodies is recognized.

Taking into account possible bias, in which the areas were chosen due to visible turtles,

and hence higher chances for capture success, and knowing that some ponds and stream

stretches are less productive, a conservative biomass estimate can still be generated.

Using the average of the lower limit of the three confidence intervals in Table 51,

multiplying by dry season water area 407 ha and mean weight of P. voglii (1.247 kg

n=181) results in a study area biomass estimate of 56,674 kg for the 63,227 ha study area.

The estimate for caiman biomass was 167,827 kg (Table 49), putting the biomass

ratio of the relatively small freshwater turtles to the much larger and more visible caiman

at .338. Since caiman are so vastly more visible, and individuals larger, this at first seems

counter-intuitive. Consider the following. The ratio of galapago/caiman biomass in well-

studied Escorzonera was .519. The estimate of 56,674 kg does not factor in the biomass

of mata mata and terecay, both much larger than galapago (terecay are roughly twice the

size). Given the uncertainty that the small number of water bodies sampled imposes, the

estimate of 56,674 kg is conservative. The biomass of freshwater turtles may be higher.

In contrast to Hato El Cedral in Apure (Mufioz & Rivas 1994), anacondas

(Eunectes murinus) were neither common nor widespread. Pifiero possesses much less

year-round water than the lower areas of Apure (Llanos Bajos). Mean weight of 17

anacondas captured in Pifiero was 17.2 kg. A Lincoln Petersen estimator run on 11

captured in the shallow mud flats of Caflo La Iguana indicated 6.7 anaconda/ha, but the

conditions were extreme drought and the real catchment area uncertain. La Iguana was a

stream running though savanna with a relatively low, even shrubby at times, riparian.

The largest specimen, 4.85 m long and 50 kg, was from Cafio Caujaral, where, for

reasons unknown, anaconda were very rare. Anacondas were relatively common in Rio

Pao, where during the dry season, they could be located, often in mating balls in riverside

holes in the steep clay banks. This was a setting where a capture-recapture effort could

have been fruitful. However, water management authorities purged excess water from an

upstream reservoir during the time set aside for the exercise, eliminating opportunities for

easy captures. Anacondas were observed in one other habitat (a permanent pool in low

elevation cailo). Although common along the banks of Rio Pao, and concentrated in the

nearly dry Cafio la Iguana, anacondas were not abundant at Pifiero.

Maximum counts of iguana (Iguana iguana) were 16/3.8 km along Cafio

Caujaral, 17/3.8 km along Rio Pao, and 31/km at Cafio la Iguana, where dry conditions

forced concentrations. As speculation: 16 iguanas per 3.8 km of stream side forest (both

sides) = 4.21 iguanas/linear km of stream side forest (both sides); rounded up for

undetected to 5/km; estimated mean weights for males 1.53 kg and for females 1.14 kg

(Dugan 1982), with 1:1 sex ratio, resulting in an average iguana = 1.33 kg; summing

total stream lengths using ARCVIEW, Mata de Guafa, La Iguana, Rio Pao, Cafio

Caujaral, and Cafio la Canoa = 100.7 kg; yields a total of 669.6 kg of iguana for the study

area. Though lacking any measure of confidence, this figure is some measure of the

magnitude of iguana biomass, an item neither readily accessible, nor important to large


DISTANCE Density Estimates, DensityValidations, Distribution Validations

The input data from transect lines were assigned to the following habitat

categories: A) BSD near Cafio; B) Standard BSD not near Cafio; C) Hill Forests-BS

mixes; D) High dry pastures-PS; E) Small flooding savanna near edge and water-SI; and

F) Wide open savanna, near and away from edge and permanent water-SI. Sabana Seca

con Chaparro (SS/CH) sections with few to no animal observations were omitted from

input, and from abundance estimations. Two transects with forested and open stretches

were sectioned according to habitat types. Total input was from up to 28 lines. Data

from each individual line were pooled over time. Analyses started with "all forest

pooled" and "all savanna pooled", and were then stratified by habitat as far as sample size

allowed. Every analysis started with half-normal, hazard rate, and uniform models and

no truncation. Preliminary results were reviewed and models that fit poorly were

eliminated. After the preliminary analyses most data sets were truncated (elimination of

the animal observations collected at the furthest distances from the transect where

detection patterns were inconsistent). Truncation usually eliminated less then 5% of the

total number of observations and rarely exceeded 10% (Buckland et al. 1993).

Deer were less common in forest (strata A, B, C, all forest pooled) than expected.

A half-normal model (Buckland et al. 1993) using observed group size 1.0513 from the

39 observations on 14 lines (after discarding 9.3% of total observations), with data

truncated at 100 m resulted in a density estimate of .01/ha, with a 95% confidence

interval of .005-.023/ha (Table 52 converted into individuals/km2). Although fond of

forest edge, and making use of fruit production in forests, deer were not common in

forest interior (see contrast between Candelaria and Caujaral Norte transects in Fig. 12).

Deer densities were very high in stratum E; small savannas with abundant edge

and permanent water. Data quality were such that truncation was not needed (no

improvement). A hazard rate model (Buckland et al. 1993) using observed group size of

1.592 froml03 observations along 6 lines and a strip width of 350 m< resulted in a

density estimate of .147/ha and a 95% confidence interval of .083-.261/ha. The number

of observations was smaller and conformation of data poorer in strata F (wide-open

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs