Sei sulla pagina 1di 13

Herpetologists' League

Diet and Food Preferences of the Tortoises Geochelone carbonaria and G. denticulata in
Northwestern Brazil
Author(s): Debra K. Moskovits and Karen A. Bjorndal
Source: Herpetologica, Vol. 46, No. 2 (Jun., 1990), pp. 207-218
Published by: Herpetologists' League
Stable URL: http://www.jstor.org/stable/3892906
Accessed: 27/08/2010 21:42

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=herpetologists.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Herpetologists' League is collaborating with JSTOR to digitize, preserve and extend access to Herpetologica.

http://www.jstor.org
Herpetologica,46(2), 1990, 207-218
? 1990 by The Herpetologists'League, Inc.

DIET AND FOOD PREFERENCES OF THE TORTOISES


GEOCHELONE CARBONARIA AND G. DENTICULATA IN
NORTHWESTERN BRAZIL
DEBRA K. MOSKOVITS"3AND KAREN A. BJORNDAL2
'Department of Biology, University of Chicago, Chicago, IL 60637, USA
2Center for Sea Turtle Research, Department of Zoology,
University of Florida, Gainesville, FL 32611, USA

ABSTRACT: We present the diet of the two Amazonian tortoises, Geochelone carbonaria and
G. denticulata, based on feeding observations and scat examinations of free-ranging individuals in
the rainforest of Maraci, northwestern Brazil. Fruits were prominent in the diet throughout the
year, but especially in the wet season; flowers were prominent in the dry season. Tortoises consumed
live and dead foliage, stems, fungi, soil, sand, pebbles, and animal matter throughout the year.
To examine the characters potentially associated with observed food preferences, we analyzed
samples of several food and non-food items for nutrient concentrations and fermentability. Typically
eaten foods had high relative abundance and calcium concentrations, while preferred foods had
high fermentabilities, high concentrations of total minerals, nitrogen and phosphorus, and low
calcium to phosphorus ratios. Nutrient profiles for preferred and rejected items were similar, and
nearly converse that of regularly eaten items. Concentrations of the nutrients analyzed cannot be
used to predict diet choice.
Key words: Amazonian rainforest; Diet; Food preference; Geochelone carbonaria; Geochelone
denticulata; Minerals; Nutrient; Tortoise

A NUMBER of studies have attempted to reportedly differ in food preferences (Cas-


identify factors important in the selection taiio-Mora and Lugo-Rugeles, 1981; Fre-
of diets (reviewed in Pyke, 1984). For her- tey, 1977), although these differences may
bivores, it is increasingly recognized that be due to individual variation.
concentrations of specific nutrients in po- In this paper, we present data on the
tential foods may play a more significant diet of G. carbonaria and G. denticulata
role than energy content in determining in the rainforest of Marac'a, in Roraima,
diet choice (Belovsky, 1981; Bryant and northwestern Brazil, based on feeding ob-
Kuropat, 1980; Owen-Smith and Novellie, servations and scat analysis. Nutrient com-
1982). Generalistherbivoresmust select an position of some of the items regularly eat-
appropriately balanced diet from a wide en, preferred, or rejected by the tortoises
array of possible food items. was determined. We examine seasonal
The foraging behavior, ecology, and changes in diet and food availability and
population sizes of the two tortoises, Geo- discuss potential characters associated with
chelone carbonaria (Spix)and G. denticu- the observed food preferences.
lata (Linne)-large, ectothermic herbi-
vores living in the relatively benign MATERIALS AND METHODS
environment of the tropical rainforest- Field Work
might be considerably affected by the Field work was carried out at Ilha de
availability of key nutrients. Yet the diets Maraca, a 110,000 ha island reserve ad-
of the two tortoises are poorly known. In- ministered by SEMA (Secretaria Especial
formation on free-ranging individuals is do Meio Ambiente), situated in the rain-
limited to brief notes in the literature, forest-savanna ecotone of northern Rorai-
which were summarized by Pritchardand
ma, Brazil. The patchy habitat composi-
Trebbau (1984). In captivity, the species tion of the 570 ha study site is roughly 85%
forest and 15% open swamps or savanna
3 Present Address: Division of Birds, Field Museum (Moskovits, 1988). Both species of Geo-
of Natural History, Chicago, IL 60605, USA. chelone used all habitat types. Between
207
208 HERPETOLOGICA [Vol. 46, No. 2

March 1981 and November 1982, 227 in- cional de Pesquisas Amazonicas (INPA),
dividuals (285 captures) were marked and in Manaus, for identification. The presence
measured. Of these, 56 (20 male, 25 female of flowers along these trails was noted, but
G. carbonaria; eight male, three female G. quantities were not estimated, and trees
denticulata) were radiotracked or trailed were not measured.
for a total of 1442 tortoise-days (Moskovits
and Kiester, 1987). Location and activity Enclosure Observations
of the tortoises were checked at least once Observations were made on eight tor-
daily (0600-1830 h), and active tortoises toises (three male, three female G. car-
were followed until lost from sight. Indi- bonaria; one male, one female G. denticu-
rect evidence was obtained from trailed lata) held captive for up to 20 days during
routes (e.g., partially eaten fruit, dug-up both wet and dry seasons. The 400 m2 out-
soil). Feeding observations (direct and in- door enclosure had red clay soil and was
direct) were recorded on 132 occasions (95 at the savanna-forest ecotone. Captive tor-
for G. carbonaria, 37 for G. denticulata): toises were fed large quantities of the fruits
110 in the dry (Aug-Mar) and 22 in the and flowers available in the field, and fresh
wet (Apr-Jul) season. Samples of foods water was supplied daily.
were collected for identification, and, when Items consistently refused by the tor-
possible, larger samples were collected for toises and not observed in the diet of free-
nutrient analyses (see below). ranging individuals were considered "non-
Tortoise scats found along routes of foods." Preference for different foods was
trailed individuals or next to tortoises en- easier to distinguish in the enclosure, where
countered, or collected while keeping tor- tortoises chose certain foods and ate them
toises overnight for processing, were in- first or exclusively. Preference levels were
dividually picked and searched for subjectively categorized as: items refused
recognizable items. Most seeds recovered or parts of foods not eaten (non-foods);
from the 43 scats examined were matched items eaten often or in moderation (i.e., in
with those of fruits in the forest at the time. amounts smaller or proportional to their
Other seeds and animal parts were iden- abundance in the study area); and items
tified in museums and herbaria. Due to highly preferred (i.e., eaten immediately
wide variation in degree of food digestion, when available or presented in captivity,
retrieved items were not quantified. The seen eaten simultaneously by several tor-
few items digested beyond recognition toises in the field, or found in higher con-
were not further analyzed. Several of the centration in tortoise scats than in the for-
seeds recovered were planted to test for est).
viability.
Analyses of Nutrient Composition
Estimates of Fruit Availability and Fermentability
Fruit availability was estimated by the Thirty-three food and non-food items
spatial and temporal frequency with which were collected in sufficient quantity to be
fruits were encountered along two census analyzed for nutrient composition and fer-
trails (2 km each). In the 25 samples taken mentability. These samples were sun-dried
over 18 mo in the field (one sample per 3 in the field, and were redried in the lab-
wk), 5 m wide strips were surveyed to the oratory at 60 C to constant mass (approx-
left or right of the designated trails (chosen imately 24 h). Dried samples were ground
at random with respect to habitat or mi- through a 1 mm screen in a Wiley mill. A
crohabitat types). The side surveyed was portion of each sample was dried at 105
switched every 50 m. Each tree encoun- C to determine percent dry matter and
tered in fruit was recorded, and the total then ashed in a muffle furnace for 3 h at
number of fruits on the tree and the pro- 500 C to determine percent organic matter
portion apparently ripe were estimated. and ash (an estimate of total mineral con-
Specimens of plants that we collected were tent). In vitro organic matter digestibility,
sent to the herbarium at the Instituto Na- or fermentability, was determined using
June 1990] HERPETOLOGICA 209

the Tilley and Terry (1963) method as TABLE 1.-Items recorded in the diet of Geochelone
modified by Moore and Mott (1974). This carbonariaand G. denticulata.
analysis consists of a 48 h incubation under Fruits.-Spondias lutea (Anacardiaceae), Anacardium gigantea (An-
CO2 at 39 C with an inoculant of rumen acardiaceae), Annona sp. 1 (Annonaceae), Annona sp. 2 (Annona-
ceae), Duguetia surinamensis (Annonaceae), Philodendron sp. (Ara-
fluid (from a fistulated steer) followed by ceae), Bromeliad sp. (Bromeliaceae), Trattinnickia ravifolia
(Burseraceae), Licania kunthiana (Chrysobalanaceae), sp. (Lecythi-
a 48 h acid-pepsin treatment to remove daceae), Myriaspora egenensis (Melastomataceae), Bagassa guianen-
sis (Moraceae), Brosimum potabile (Moraceae), Ficus sp. (Moraceae),
undigested microbes. The percent of or- Passiflora coccinea (Passifloraceae), Passiflora vespertitio (Passiflora-
ceae), Mauritia flexuosa (Palmae), Desrnoncus polyacanthus (Palmae),
ganic matter that disappears during the 96 Duroia eriopila (Rubiaceae), Genipa americana (Rubiaceae), Guet-
tarda argentea (Rubiaceae), Posoqueria sp. (Rubiaceae), Pradosia sp.
h is the fermentability of the sample. (Sapotaceae), Richardella (surinamensis) (Sapotaceae), Pouteria hirta
Percentage of cell walls (cellulose, hemi- (Sapotaceae), Ecclinusa bacuri (Sapotaceae), Clavija sp. (Theophras-
taceae).
cellulose, lignin and cutin, or neutral de- Flowers.-Jacaranda copaia (Bignoniaceae), Cochlospermum orino-
cense (Cochlospermaceae), Mauritia flexuosa (Palmae).
tergent fiber ash-free) was measured by Miscellaneous.-mushrooms (several species); Maximilliana (palm) frond
the Van Soest technique (Goering and Van base; unidentified grasses, live leaves, vine stems, roots; sand, soil,
pebbles; tortoise scat.
Soest, 1970) with decalin and sodium sul- Animal matter.-snails (at least 2 species); ants (several species); ter-
mites; beetles (assorted body parts recovered); butterflies (wings re-
fite omitted (Golding et al., 1985). Amylase covered); Euglossine bees; snakes (ventral scales recovered); lizards;
was used in the determination of cell walls birds (leg of Psarocolius viridis recovered); agouti; peccary; deer
carcasses.
when necessary (Robertson and Van Soest,
1977). Percentage of cell contents (starch,
soluble sugars, protein) was determined by Too few scats of G. denticulata were
subtracting percent cell walls from percent collected for statistical comparisons be-
organic matter. Analyses for percentages tween species or sexes, and seasonal com-
of acid detergent fiber (a sub-fraction of parisons were only possible for G. carbo-
cell walls which is approximately equal to naria. Juveniles were not included in any
the cellulose and lignin content), cellulose, of the analyses.
potassium permanganate lignin, and cutin
followed Goering and Van Soest (1970). RESULTS
Percent concentrations of total (Kjeldahl)
nitrogen, phosphorus, and calcium were Seasonal Diet and Food Preferences
measured with a block digester (Gallaher The diverse diet of Geochelone carbo-
et al., 1975) and an automated Technicon naria and G. denticulata at Maraca' con-
analyzer (Hambleton, 1977). Replicates of sisted of various vegetative and reproduc-
all analyses were acceptable within 1%rel- tive plant parts (grasses, leaves, vines, roots,
ative error. bark, fruits, and flowers), fungi (several
gilled and woody mushrooms), animal
Statistical Analyses matter (vertebrate carrion, insects, snails),
Diet components were not distributed soil, sand, and pebbles (Table 1). The pro-
normally in either the foraging observa- portions of these various food items in the
tions (FO) or the scat samples (SC). There- diet differed markedly when based on for-
fore, comparison tests were performed on aging observations (FO) or on scat exam-
ranks (Wilcoxon scores, equivalent to the ination (SC) (Table 2). Although fruit
Wilcoxon rank-sum test for two levels and ranked highest in both categories, flowers,
the Kruskal-Wallis test for n levels; SAS which ranked second highest in foraging
Institute Inc., 1987). Nutrient composi- observations, were never detected in the
tions of the different preference levels were visual examination of scats. In contrast, an-
also compared with non-parametric tests imal matter (both vertebrate and inver-
(Wilcoxon and median scores; SAS Insti- tebrate) was recovered from nearly half of
tute Inc., 1987). Nutrient profiles for the the scats examined, but was observed eaten
different classes are presented as propor- only once in the field (one male G. car-
tional departures from the overall sample bonaria feeding on an agouti carcass).
median. Spearman rank correlation (r,; Vegetative plant parts (grasses, leaves, leaf
SAS Institute Inc., 1987) were used to ana- litter, bark) also were detected more easily
lyze the relationships among the different in scats than in field observations.
nutrient concentrations measured. Diet composition did not differ signifi-
TABLE 2.-Food categories in the diet of Geochelone carbonaria (C) and G. denticulata (D), grouped by: (a) sex
(b) season (D = dry, W = wet). For each food category, percent of foraging observations (FO) is presented first, fo
(in parentheses). n = total number of observations or scats (pooled over all individuals). Note that scats contained
up to 100. The two highest entries in each row are boldfaced for FO and underlined for SC. For the pooled da
significantly for all food categories (Wilcoxon scores range from 5.3-48.8, df = 1, 0.0001 < P < 0.02). For G.
significantly (W scores from 15.6-20.1, df = 1, P < 0.001).
(a) Grouped by species (SP) and sex (SX)
SP SX n Fruit Flower LV' DV2 Fungi Soil."

C M 22 (16) 68.2 (81.3) 13.6 (0) 9.1 (81.3) 0.0 (43.7) 4.6 (18.7) 0.0 (43.7)
C F 70 (17) 40.0 (70.6) 27.1 (0) 21.4 (64.7) 2.9 (35.3) 4.3 (17.7) 4.3 (41.2)
C J 3 (5) 66.7 (60.0) 0.0 (0) 0.0 (60.0) 33.3 (60.0) 0.0 (0.0) 0.0 (100.0)
C 95 (38) 47.4 (73.7) 23.2 (0) 17.7 (71.1) 3.2 (42.1) 4.2 (15.8) 3.2 (50.0)
D 37 (5) 46.0 (60.0) 29.7 (0) 8.1 (80.0) 5.4 (80.0) 0.0(40.0) 10.8 (20.0)
(b) Grouped by species (SP) and season (SN)
SP SN n Fruit Flower LV, DV2 Fungi Soil,

C D 76 (27) 40.8 (63.0) 29.0 (0) 15.8 (66.7) 4.0 (44.4) 5.2 (11.1) 2.6 (51.9)
C W 20 (11) 70.0 (100.0) 0.0 (0) 25.0 (81.8) 0.0 (36.4) 0.0 (27.3) 5.0 (45.5)
D D 36 (3) 41.7 (33.3) 30.6 (0) 8.3 (100.0) 5.6 (66.6) 0.0 (33.3) 11.1 (0.0)
D W 2 (2) 100.0 (100.0) 0.0 (0) 0.0 (50.0) 0.0 (100.0) 0.0 (50.0) 0.0 (50.0)
LV = live vegetative plant parts: leaves, stems, roots.
2DV = dead leaves (leaf litter) and bark.
Soil = soil + sand + pebbles.
VERT = vertebrates.
INV = invertebrates.
ANIM = any animal part (vertebrate or invertebrate).
June 1990] HERPETOLOGICA 211

a. Fruit: 1 = eaten
=
3
CI 30
2 possibly eaten
co 3 = not eaten
._ Flower:
0 20

Z 10

~b.
30
-a

4.0
co 20
E

0.O
M J J A S O N D J F M A M J J A S 0
1981 -- 1982
WET DRY WET DRY
FIG. L-(a) Phenology of fruiting and flowering trees. Fruits are grouped as eaten (1), possibly eaten (2),
or not eaten (3) by Geochelone. (b) Estimated number of ripe fruits (total over all trees censused) present
per sample.

cantly between species (Table 2; P > 0.1 of the multi-seeded Duguetia were re-
for the nine food categories in FO; SC not covered simultaneously. Of six species re-
tested). Within species, the only sex dif- trieved from scats and planted, four ger-
ference detected was the higher consump- minated successfully.
tion of fruit by male than by female G. In the rainyseason,when fruitingpeaked
carbonaria (significant for FO only, x2- in both volume and number of species (Fig.
5.5, df = 1, P = 0.02). la,b), fruits made up over 70%of the feed-
Seasonal differences, testable for G. car- ing observationsand were present in 100%
bonaria only, were pronounced in the con- of the scats examined. During this season,
sumption of fruit and flowers. Fruit con- large fruits (such as Duguetia surinamen-
sumption was considerably higher in the sis, Richardella surinamensis?, Pouteria
wet than in the dry season (x2 = 5.4, df- hirta) or smallerfruitsin large-canopytrees
1, P = 0.02 for both FO and SC), while (Spondias lutea, Pradosia sp.) were widely
flowers were observed eaten only in the available (present in 43% of the 40, 100-
dry season. m blocks censused; Moskovits, 1985).
Fruits (27 identified and at least four The rich, oily fruit of the palm Mauritia
unidentified species) were the major item flexuosa was available in the forest and was
consumed. They comprised nearly half of consumed by the tortoises throughout the
the feeding observations for both species year (Fig. 2). Tortoisesoften ate the fruits
and were recovered from over 60% of all of two species, Bagassaguianensis(Mora-
scats examined (Table 2). Up to five dif- ceae) and Myriaspora egenensis (Melas-
ferent species of fruit were retrieved from tomataceae), which were rare in the study
a single fecal sample, and up to 34 seeds area. Only one tree of Myriaspora and
of the single-seeded Spondias lutea (xt = three trees of Bagassa were found in the
8.6 g, single fruit wet weight) and 40 seeds study track, and neither was present along
212 HERPETOLOGICA [Vol. 46, No. 2

MONTHSIN FRUIT (OR FLOWER) 1


FAMILY GENUS SPECIES -wet season f----dry season -wet season-Fdry season-
__________________________________________ May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct
FRUITS
Anacardiaceae Spondias lutea
Annona sp*1
Annonaceae Annona sp.2
Duguetia surinamensis
Araceae Philodendron sp.
Bromeliaceae - sp.
Burseraceae Trattinnickia ravifolia

Chrysobalanaceae Licania kunthiana


Lecythidaceae _ sp.
Melastomataceae Myriaspora egenensis
( Bagassa guianensis
Moraceae Brosimum potabile
Ficus sp.
Passiflora coccinea
Passif loraceae Passiflora vespertilio

Palnae Mauritia flexuosa


Duroia eriopila
Rubiaceae Genipa americana
Guettarda argentea
(Pradosia sp.
Richardella (surinamensis)
Sapotaceae Pouteria hirta

Ecclinusa bacuri
Theophrastaceae Clavija sp.

FLOWERS
Bignoniaceae Jacaranda copaia
Cochlospermaceae Cochlospermum orinocense
Palmae Mauritia flexuosa J ,_ ,

FIG. 2. -Phenology of the important species of fruits and flowers eaten by Geochelone. Hatched lines
represent fruits that were probably unripe.

the fruit census trails. Groups of 2-5 in- bright yellow Cochlospermum orinocense
dividuals fed simultaneously under trees made up 30% of all flowers observed eaten.
(Bagassa) bearing ripe fruit, even though Yet the visually very similar Allamanda
individuals of Geochelone were generally cathartica (Apocynaceae) was never seen
solitary animals. Tortoises also aggregated eaten in the field and was rejected in cap-
under Genipa americana bearing ripe tivity (Table 3). The small and fragrant,
fruit. Fruits not seen eaten by tortoises, and orange-yellow flowers of Mauritia palms
considered unlikely items in their diet (type were available over much of the year (Fig.
"3" in Fig. 1), were generally non-fleshy 2), but were observed eaten only during
and wind-dispersed or too large and hard the dry season.
for tortoises to consume. Several species of mushrooms were eat-
Flowers constituted close to 30% of the en readily by the tortoises. Often the fungi
feeding observations during the dry season were recovered seemingly intact from tor-
for both species of Geochelone. The flow- toise scats. Mushrooms were considerably
ering peak at Marac'a occurred in the mid- more conspicuous in the forest during the
dle to late dry season (Fig. 1), when the rainy season, but tortoises were observed
fruits in the forest were generally small or eating them only during the dry season
present singly or in small clusters. Three (Table 2). No seasonal difference was ev-
species (Table 1) comprised over 72% of ident from the scat data. Also consumed
the flower feeding observations, and the year-round were live grasses, leaves, stems,
June 1990] HERPETOLOGICA 213

vines, and roots,and dead leaves, bark,soil, eral concentrations. They were relatively
sand, and pebbles. Scat reingestion was ob- abundant and high in Ca concentration
served once in the enclosure. with high Ca:P ratios. In contrast, pre-
ferred foods had low Ca concentrations
Feeding Behavior and Ca:P ratios, low cell wall concentra-
When feeding in patches of dense fruit tions, high fermentability, and high N, P,
or flowers, tortoises did not systematically and total mineral concentrations (Fig. 3).
eat the firstitem encountered.Instead,they Surprisingly, items preferred and rejected
zig-zagged back and forth, holding their were similar in nutrient composition. This
heads low and sniffing loudly. They often may be, in part, because regularly eaten
took only a few bites of a fruit or flower foods analyzed were mostly fruits (16 of
that they later returned to eat whole, and 18 items), while those preferred or rejected
when eating leaves, they generally bit and were mostly flowers (two of three and nine
rejected several before swallowing one. of 12, respectively).
This same pick-and-choosebehaviorwas
DISCUSSION
displayed in the enclosure, even when the
tortoiseswere presented non-naturalfoods Land turtles are primarily opportunistic
(e.g., banana, papaya). Captive tortoises feeders, their diets often reflecting season-
often ate two or more items simultaneous- al and regional availabilities of food (e.g.,
ly, taking a few bites from each and Hansen et al., 1976; Klimstra and New-
switching frequently. Individual prefer- some, 1960; Luckenbach, 1982). However,
ences varied among the tortoises tested, turtles are capable of discriminating colors
but all ate meat first when presented with and of feeding selectively in their envi-
a variety of items. Large live grubs were ronment (e.g. Macdonald and Mushinsky,
consumed immediately when presented to 1988; Quaranta and Evans, 1949; Swing-
one captive individual. land and Frazier, 1979). In a study in Flor-
The tortoises swallowed most of their ida, Gopherus polyphemus were consid-
foods whole in the field, only occasionally ered intermediate between foraging
using their forelimbs to help tear larger specialists and generalists (Macdonald and
items. However, in captivity, they manip- Mushinsky, 1988).
ulated the foods in their mouths until seeds In Marac'a, the two species of Geoche-
or hard items were ejected. They did not lone ate primarily fruits throughout the
swallow the seeds or bones as they did in year. They heavily supplemented their diet
the field. Both species showed keen dis- with flowers in the dry season, and with
criminating capabilities in captivity, com- animal matter, live and dead vegetative
ing directly to a favored item from as far plant parts (leaves, vines, roots, leaf litter),
as 20 m away and learning to avoid per- fungi, soil, sand, and pebbles year-round.
manent obstacles(such as a dividing fence) Species and sex differences were not de-
along the way. tected, except for a higher overall con-
sumption of fruit by male than by female
Nutrient Analyses G. carbonaria (significant in FO only). Sea-
Nutrient analyseswere performed on 12 sonal differences, analyzable only for G.
rejected, 18 regularlyeaten, and three pre- carbonaria, were pronounced in fruit and
ferred items. Of the variables considered flower consumption, and reflected overall
(Table 3), median values differed signifi- availability in the forest. The pronounced
cantly for abundance, fermentability, cell differences between foraging observations
wall (cellulose, hemicellulose and lignin), and scat examinations suggest that the diet
nitrogen (N) concentration,phosphorus(P) information presented here is preliminary.
concentration, and calcium (Ca):P ratios Given both the complexity of the rainfor-
(Fig. 3). The foods regularly eaten seem est and its ample food abundance, and the
"poor"in quality, with high cell wall con- broad diet of the Geochelone, additional
centrations (especially cutin), low fer- food items should be identified with fur-
mentability, and low N, P, and total min- ther study.
TABLE 3.-Nutrient composition (%) of items rejected (R), accepted (A), and preferred (P). Abundance categories a
and 2 = abundant. Fermentability is expressed as % organic matter; the other variables are % dry matter. ADF

Prefer- Abun- Ferment- Organic Cell


ence dance ability matter contents Cell walls ADF Cellulose Lignin Cutin Ni

Fruit (unripe)
Aegiphila membranacea R 1 46.6 91.9 43.5 48.3 37.3 22.1 10.8 4.7
Duguetia surinamensis R 1 27.4 96.9 22.7 74.3 55.2 33.3 15.7 6.3
Gustavia augusta R 0 46.7 95.5 44.4 51.4 34.8 21.6 10.1 3.1
Flower
Allamanda cathartica R 2 63.3 93.9 66.0 29.8 20.7 13.6 4.8 1.2
Genipa americana R 1 56.0 94.7 56.0 40.0 33.0 15.3 15.5 3.3
UT 1 R 1 63.3 94.6 60.3 34.9 24.7 11.7 11.7 1.9
UT 2 R 1 85.8 94.9 78.0 17.6 10.8 6.8 2.7 0.9
UT 3 R 1 61.3 92.1 65.7 29.5 21.1 13.0 4.8 1.0
Ul 4 R 1 72.4 94.7 74.4 20.9 13.9 9.0 2.2 2.4
Ul 5 R 1 31.8 95.9 52.7 44.9 28.5 8.5 14.5 6.4
UT 6 R 1 63.9 96.0 76.7 19.9 13.2 6.8 3.3 3.4
UT 7 R 1 - 97.1 67.2 30.5 17.5 9.5 5.9 2.4
Stem
Ul vine A 2 55.7 78.4 30.7 56.9 45.6 22.1 14.0 4.0
TABLE 3.-Continued.
Prefer- Abun- Ferment- Organic Cell
ence dance ability matter contents Cell walls ADF Cellulose Lignin Cutin Nit

Fruit (ripe)
Bagassa guianensis A 0 51.9 91.2 47.9 50.9 43.7 13.1 11.2 14.3
Duguetia surinamensis A 1 31.0 96.4 24.6 72.3 53.4 30.4 15.6 7.4
Ecclinusa guyanensis A 2 12.2 97.9 19.1 81.4 73.5 20.7 6.8 45.2 0
Genipa americana A 2 52.7 95.3 38.5 57.0 44.3 29.8 11.1 3.3
Genipa a. (rind) A 2 75.4 95.9 67.6 28.3 24.2 16.5 6.3 1.2 0
Geophila gracilis A 2 50.9 90.9 38.3 53.5 43.0 3.4 7.7 11.2 2
Guettarda argentea A 0 36.9 94.8 37.9 58.6 46.6 21.4 10.2 14.7 0
Ficus sp. A 1 42.0 95.4 34.7 61.5 42.3 18.3 11.7 12.1
Mauritia flexuosa A 2 20.7 94.9 37.7 59.3 46.1 21.8 8.6 13.4 0
Mauritia f. (seed coat) A 2 65.4 94.8 41.1 54.8 38.5 31.4 6.4 1.3 0
Myriaspora egenensis A 0 66.9 96.2 66.5 30.2 25.9 7.3 7.6 11.0 1
Passiflora sp. A 2 43.3 94.1 30.2 64.1 47.3 27.3 13.5 6.6 2
Philodendron sp. A 1 76.5 93.7 71.7 22.1 15.4 6.9 6.1 2.3 1
Spondias lutea A 2 20.2 95.6 21.1 75.6 64.9 40.4 12.7 11.2 0
UT 1 A 0 35.1 96.6 37.7 60.3 53.7 16.9 9.5 26.9 1
UT 2 A 1 29.3 97.6 32.1 66.4 54.5 20.8 10.6 22.7 1
Flower
Passiflora sp. A 2 56.9 92.4 53.2 41.6 19.0 9.1 5.2 4.7 3
Flower
Jacaranda copaia P 1 65.4 96.1 60.4 37.0 24.9 9.0 6.4 8.7 2
Cochiospermum orino- P 1 62.1 91.8 58.2 36.0 26.9 13.4 9.6 3.2 2
cense
Fungus
Ul mushroom P 1 58.7 80.0 45.8 42.4 25.8 11.5 6.2 1.9 5
216 HERPETOLOGICA [Vol. 46, No. 2

ITEMS
Eaten regularly Preferred Rejected

++ Abundance
++ Fermentability
* % Organic matter-
% Cell contents -
** ++ % Cell wall -
* ++ % Cellulose -
% Lignin -
** + % Cutin -

** + % Nitrogen -
% Minerals -
** ++ % Phosphorus -
% Calcium
** + CA:P

-1 0 1 -1 0 1 -1 0 1

Departure from overall sample median (%)


FIG. 3.-Nutrient composition profile. Levels of significance in comparison of medians: Wilcoxon scores
(rank sums) * = P < 0.05, ** P < 0.01; median scores (number of points above the median) + = P <
0.05, ++ = P< 0.01.

Despite the seasonal changes observed, treefalls regularly (ideal environments for
the diet of the Geochelone did not solely the growth and establishment of seed-
reflect abundance in the forest. The tor- lings), they are probablyeffective seed dis-
toises frequently ate at least two fruits that persers. This also has been suggested for
were rare in the study area. They rejected Gopherus polyphemus (Macdonald and
the flowers of Allamanda cathartica, which Mushinsky, 1988) and Testudo graeca
were available simultaneously with, and (Cobo and Andreu, 1988).
closely resembled, the favored flowers of A large portion of the diet of the tor-
Cochlospermum orinocense. Further- toises was high in cell wall concentration
more, they consumed flowers of Mauritia (especially cutin), was relatively low in fer-
flexuosa only in the late dry season, al- mentability and cell contents (starch, sol-
though the flowers were available through uble sugars, protein) and was low in total
much of the year (Fig. 2). Both species of mineral concentration. However, these
Geochelone also showed keen discrimi- foods were relatively abundant in the for-
nating capabilities in the enclosure, often est and were high in Ca concentrations.In
coming directly to a favored food item. contrast,the few preferred items were low
In the enclosure, where soil was harder in cell wall contents, high in cell contents
clay and sand was not available, tortoises and fermentability, and high in overall
did not swallow their foods whole as they minerals, P, and total N concentrations.
did in the forest. As suggested by many Items rejected were very similar to the
(e.g., Luckenbach, 1982; Marlow and Tol- preferred foods in nutrient composition,
lestrup, 1982; Sokol, 1971), sand may be although their total mineral, P, and N con-
important as an abrasive agent enhancing centrations were not as high (Fig. 3). This
digestion, and its absence in the enclosure similarity may have been due, in part, to
may have contributed to the observed shift the high proportionof flowers represented
in behavior. in both the non-food and the preferred
Given this tendency of the Geochelone samples. The observed tendency for tor-
to swallow foods whole, and to frequent toises to eat several foods simultaneously
June 1990] HERPETOLOGICA 217

in captivity, their seemingly haphazard project. Finally, we are deeply grateful to R. Mos-
behavior in the field, and the diversity of kovits for her immeasurable assistance in the field.
items recovered from single scat samples LITERATURE CITED
may indicate that tortoises reach an ade-
quate balance of nutrients by ingesting a BELOVSKY,G. E. 1981. Foodplantselectionby a
generalistherbivore:
Themoose.Ecology62:1020-
wide variety of foods. 1030.
Based on the nutrient profiles obtained BJORNDAL, K. A. 1989. Flexibility of digestive re-
in this study, concentrations of the nu- sponsesin two generalistherbivores,the tortoises
trients analyzed cannot be used to predict Geochelonecarbonariaand Geochelonedenticu-
lata. Oecologia 78:317-321.
diet choice. Yet three points should be con- BRYANT, J. P., AND P. J. KUROPAT. 1980. Selection
sidered. First, toxic and secondary com- of winter forage by subarctic browsing vertebrates:
pounds were not examined in this study, The role of plant chemistry. Ann. Rev. Ecol. Syst.
and they might be important in determin- 11:261-285.
ing the diet of the tortoises. It is likely, CASTA&O-MORA, 0. V., AND M. LUGO-RUGELES.
1981. Estudio comparativo del comportamiento
however, that fruiting trees have not de dos especies de morrocoy: Geochelone carbo-
evolved specific noxious compounds to de- naria y Geochelone denticulata y aspectos com-
ter the tortoises, which are potential seed parables de su morfologia externa. Cespedesia 10:
dispersers. Second, fermentability, used 55-122.
COBO, M., AND A. C. ANDREU. 1988. Seed con-
here as a relative measure of microbial sumption and dispersal by the spur-thighed tortoise
degradability of the foods, may not reflect Testudo graeca. Oikos 51:267-273.
digestibility in the tortoises' guts. How- FRETEY, J. 1977. Les Ch6loniens de Guyane Fran-
ever, both species of Geochelone rely on gaise. 1. Etude Preliminaire. Thesis, University of
a hindgut microbial fermentation, similar Paris, Paris.
GALLAHER, R. N., C. 0. WELDON, AND J. G. FUTRAL.
to the rumen fermentation, to degrade the 1975. An aluminum block digester for plant and
cell wall fraction of their diet (Bjorndal, soil analysis. Soil Sci. Soc. Am. Proc. 39:803-806.
1989). Third, the relative concentrations GOERING, H. K., AND P. J. VAN SOEST. 1970. Forage
of a nutrient in a series of food items do Fiber Analyses (Apparatus, Reagents, Procedures,
not necessarily reflect the relative avail- and Some Applications). Agric. Handbook 379,
United States Department of Agriculture, Wash-
abilities of that nutrient in the food items. ington, D.C.
Minerals,especially, vary greatly in avail- GOLDING,E. J., M. F. CARTER, AND J. E. MOORE.
ability, and their absorption in the gut is 1985. Modification of the neutral detergent fiber
a complex interaction of the pH, the pres- procedure for hays. J. Dairy Sci. 68:2732-2736.
HAMBLETON, L. G. 1977. Semiautomatedmethod
ence or absence of other minerals, and the for simultaneous determination of phosphorus, cal-
chemical form of the mineral (Robbins, cium and crude protein in animal feeds. J. Assoc.
1983; Van Soest, 1982). Appropriate data Off. Agric. Chem. 60:845-852.
on relative availabilities, along with nu- HANSEN, R. M., M. K. JOHNSON, AND T. R. VAN
trient analysesof a broadersample of items DEVENDER. 1976. Foods of the desert tortoise,
Gopherus agasizii, in Arizona and Utah. Herpe-
regularly eaten, preferred, and rejected, tologica 32:247-251.
should help clarify the nutritionalbasis for KLIMSTRA,W. D., AND F. NEWSOME. 1960. Some
food preferences. observations on the food coactions of the common
box turtle (Terrapene c. carolina). Ecology 41:637-
647.
Acknowledgments-We thank Dr. P. Nogueira LUCKENBACH, R. A. 1982. Ecology and manage-
Neto for SEMA facilities; G. M. de Oliveira, L. Pes- ment of the desert tortoises (Gopherus agassizii) in
tana, J. and E. Lima da Silva, and Amazonas for help California. Pp. 1-37. In R. B. Bury (Ed.), North
in the field; and Dr. P. E. Vanzolini for invaluable American Tortoises: Conservation and Ecology. U.S.
support at every stage. For specimen identifications, Fish and Wildlife Service, Wildlife Research Re-
we thank Dr. M. F. da Silva, INPA (plants); Dr. R. port 12, Washington, D.C.
B. Foster, FMNH (seeds); and Dr. J. W. Fitzpatrick, MACDONALD, L. A., AND H. R. MUSHINSKY. 1988.
FMNH (bird parts). J. Moore generously provided Foraging ecology of the gopher tortoise, Gopherus
use of his laboratory. A. Bolten, D. F. Stotz, R. D. polyphemus, in a sandhill habitat. Herpetologica
Semlitsch, and four anonymous reviewers made valu- 44:345-353.
able suggestions on earlier versions of the manuscript. MARLOW, R. W., AND K. TOLLESTRUP. 1982. Min-
Grants to DKM from the Chicago Zoological Society, ing and exploitation of natural mineral deposits by
the Hinds Fund of the University of Chicago, and the desert tortoise, Gopherus agassizii. Anim. Be-
the National Academy of Science helped fund the hav. 30:475-478.
218 HERPETOLOGICA [Vol. 46, No. 2

MOORE, J. E., AND G. 0. MOTT. 1974. Recovery of visual learning of Testudo vicina. Anat. Rec. 105:
residual organic matter from in vitro digestion of 580.
forages. J. Dairy Sci. 57:1258-1259. ROBBINS,C. T. 1983. Wildlife Feeding and Nutri-
MOSKOVITs, D. K. 1985. The Behavior and Ecology tion. Academic Press, New York.
of the Two Amazonian Tortoises, Geochelone car- ROBERTSON, J. B., AND P. J. VAN SOEST. 1977. Di-
bonaria and Geochelone denticulata, in North- etary fiber estimation in concentrate feedstuffs. J.
western Brazil. Ph.D. Dissertation, University of Anim. Sci. 45(Suppl. 1):254.
Chicago, Chicago. SAS INSTITUTE INC. 1987. SAS/STATT" Guide for
. 1988. Sexual dimorphism and population Personal Computers, Version 6 ed. SAS Institute
estimates of the two Amazonian tortoises (Geoche- Inc., Cary, North Carolina.
lone carbonaria and G. denticulata) in northwest- SOKOL, 0. 1971. Lithophagy and geophagy in rep-
ern Brazil. Herpetologica 44:209-217. tiles. J. Herpetol. 5:69-70.
MOSKOVITS, D. K., AND A. R. KIESTER. 1987. Ac- SWINGLAND, I. R., AND J. G. FRAZIER. 1979. The
tivity levels and ranging behaviour of the two Am- conflict between feeding and overheating in the
azonian tortoises, Geochelone carbonaria and Geo- Aldabran giant tortoises. Pp. 611-615. In C. J. W.
chelone denticulata, in north-western Brazil. Amlaner and D. W. W. Macdonald (Eds.), A Hand-
Functional Ecol. 1:203-214. book in Biotelemetry and Radio Tracking. Perga-
OWEN-SMITH, N., AND P. NOVELLIE. 1982. What mon Press, London.
should a clever ungulate eat? Am. Nat. 119:151- TILLEY, J. M. A., AND R. A. TERRY. 1963. A two-
178. stage technique for the in vitro digestion of forage
PRITCHARD, P. C. H., AND P. TREBBAU. 1984. The crops. J. Brit. Grassl. Soc. 18:104-111.
turtles of Venezuela. Soc. Stud. Amphib. Rept., VAN SOEST, P. J. 1982. Nutritional Ecology of the
Contrib. Herpetol. 2:1-403. Ruminant. 0 & B Books, Corvallis, Oregon.
PYKE, G. H. 1984. Optimal foraging theory: A crit-
ical review. Ann. Rev. Ecol. Syst. 15:523-575. Accepted: 28 June 1989
QUARANTA, J. V., AND L. T. EVANS. 1949. The Associate Editor: Raymond Semlitsch

Potrebbero piacerti anche