Excerpted from

©2004 by the Regents of the University of California. All rights reserved.May not be copied or reused without express written permission of the publisher.

click here toBUY THIS BOOK

Given the precarious status of twoof the three subspecies of the San Salvador

iguana (Cyclura rileyi) in the wild (Hayes et al.,this volume), consideration should be given totreating these three subspecies as separatemanagement units. The phenetic analysis bySchwartz and Carey (1977) provided the currenttaxonomy of Cyclura, which recognizes the sub-species of C. rileyi. Their conclusions were de-rived largely from assessment of morphologicalcharacters, primarily scale counts. More recently,Hollingsworth (1998) and Malone et al. (2000)proposed interspecific relationships based onmodern phylogenetic analyses of morphologicaland mitochondrial DNA (mtDNA) variation, re-spectively, and both studies found support forthe specific status of C. rileyi. However, becauseof the lack of mtDNA differentiation withinC. rileyi (Malone et al., 2000), Malone and Davis(this volume) question the need for managing thethree subspecies as separate entities. Althoughmore detailed genetic analyses in our lab remainuncompleted, we have collected morphologicaldata that, in addition to our ecological data (Hayes

et al., this volume), may shed some light on thisimportant question.

In this chapter, we address two major areasthat relate to the conservation of C. rileyi. First,we present findings on morphological variationbetween and within the three subspecies. Weconsider several aspects of morphology, includ-ing body size, frequency of injuries, femoralpore counts, and head scalation. Although sub-ject to phylogenetic constraints, population dif-ferences in body size and injury rates often re-flect differences in, and sometimes adaptationto, local ecology. Within a population, temporalvariation in these measures can be expected totrack environmental change. Thus, morpho-logical study may offer insights on the relativehealth of a population. Differences in femoralpore and head-scale counts, in contrast, gener-ally reflect phylogeny (although sometimesontogeny or sexual dimorphism) rather thanecology, and therefore are useful for taxonomicpurposes (e.g., Snell et al., 1984; Hollingsworth,1998). Using these characters, we employ dis-criminant analyses to evaluate the possible taxo-

2 5 8

18

Conservation of an EndangeredBahamian Rock Iguana, II

MORPHOLOGICAL VARIATION AND CONSERVATION PRIORITIES

Ronald L. Carter and William K. Hayes

nomic distinctiveness of the three subspecies.Second, we offer specific conservation recom-mendations that take into consideration our ma-jor findings from both our population assess-ments and behavioral ecology studies (Hayes etal., this volume), in addition to the morphologi-cal data presented here.

MORPHOLOGICAL VARIATION

METHODS

CAPTURE AND DATA COLLECTION

Between 1993 and 1999, we captured, marked,and measured snout-vent length (SVL), headlength, and tail length of 484 iguanas (75 C. r.cristata, 198 C. r. nuchalis, 211 C. r. rileyi ) as de-scribed in Hayes et al. (this volume). We noteddamage to toes and tail, measured tail regener-ation if present, and counted the number offemoral pores on both legs. Close-up photosof the head (35 mm color slides), both dorsal andlateral angles, were obtained for most lizards.Although taken primarily for identification pur-poses, these photos were later used for scalecounts. Even though coloration differences areconspicuous between some populations, wedid not undertake a formal study of this trait.Several ecological features of individual popula-tions were considered in the analyses of bodysize and injuries. These included size of cay,population size, iguana density, number of plantspecies, and presence of rats (see Hayes et al.,this volume).

DATA TREATMENT AND ANALYSES

Statistical tests were conducted using SPSS forWindows (release 8.0, 1997; SPSS Inc., Chi-cago, Illinois), with α = 0.05. Both parametricand nonparametric tests were conducted de-pending on the data properties. For some data,log transformations were required prior toanalyses. In some cases, we report parametrictests of data that failed to meet parametric as-sumptions, but when possible, we used non-parametric alternatives to confirm suitabilityof the parametric tests. For multivariate tests,

effect sizes (proportion of variance explained byan independent variable) are indicated in somecases by η2.

RESULTS AND DISCUSSION

BODY SIZE VARIATION

Our analyses of the relationship between bodymass and SVL revealed several patterns of vari-ation. The strong correlation between body massand SVL was best explained by a power regres-sion equation (mass = 0.049 SVL2.908; r2 =0.94). An analysis of covariance (ANCOVA)model (log mass = sex × subspecies × log SVL)revealed that the relationship between massand SVL (a measure of body condition) was sim-ilar for males and females (P = 0.77, η2 = 0.00)but differed significantly among the three sub-species (F2,433 = 18.47, P < 0.001, η2 = 0.08):C. r. nuchalis was leaner than the other two sub-species. When the ANCOVA was restricted onlyto adult iguanas (≥20 cm SVL), the same pat-terns emerged, again showing C. r. nuchalis tobe leaner than the other subspecies (F2,366 =22.71, P < 0.001, η2 = 0.11; figure 18.1). AnotherANCOVA model (log mass = sex × season × logSVL), restricted to captures of adult C. r. rileyiin March, May–June, and November 1995 (n =36, 32, and 10, respectively), revealed a signifi-cant interaction between sex and season (F2,71 =4.87, P = 0.01, η2 = 0.12), arising because maleswere heavier than females in March, whereasfemales were heavier than males in May–June(November samples were too small for compar-ison). The significant main effect of season(F2,71 = 8.70, P = 0.016, η2 = 0.05) indicated thatiguanas, regardless of sex, were heaviest in Marchand leanest in November (figure 18.1).

These data are consistent with the conclu-sions of Iverson (1979) and Auffenberg (1982a)—based on analyses of fat bodies and diet, respec-tively, for C. carinata—that iguanas feed mostheavily in spring and less heavily in the summer(when time and energy are devoted to reproduc-tion) and winter (when food resources are leastavailable). However, the extent to which relativehydration influences body condition remains to

C O N S E R V AT I O N O F A N E N D A N G E R E D R O C K I G U A N A , I I 2 5 9

be investigated and represents a potentially con-founding factor.

Among the eleven populations comprisingthe three subspecies, there were significant dif-ferences in mean body mass (one-way ANOVA:F10,445 = 67.14, P < 0.001, η2 = 0.60) and meanSVL (one-way ANOVA: F10,470 = 26.23, P < 0.001,η2 = 0.36), with the largest iguanas occurringon Low Cay and in the translocated population,and the smallest occurring on Manhead andWhite Cays (table 18.1). Iguanas from the trans-located population attained significantly largerbody size than those from their source popula-tion on Fish Cay (Scheffe post-hoc comparison,P < 0.001). A similar phenomenon was reported

by Knapp (2001a) for a translocated populationof C. cychlura inornata in the Exumas and mayresult from reduced competition for food andthermoregulatory sites. Iverson (2001) and Iver-son et al. (this volume) noted that individualsof C. c. inornata on sparsely populated AllensCay in the northern Exumas were substantiallylarger than those on two adjacent, densely pop-ulated cays. He proposed that the extraordinarygrowth may be a result of reduced competition,the presence of unique food plants (e.g., morn-ing glory [Ipomoea pes-caprae] ), or the possibleinclusion of significant animal protein (nestingAudubon’s shearwaters [Puffinus lherminieri] ) intheir diet. We additionally suggest that moreenergy can be allocated to growth when socialinteractions are fewer (compare Christian et al.,1986; Wikelski et al., 2001) on sparsely popu-lated islands.

Spearman correlation analyses among theeleven populations indicate that both maximumSVL and mean body mass were independent ofcay size, population size, iguana density, andnumber of plant species (all P > 0.20). The rea-sons for population differences in body size re-main unclear to us. However, we suspect thatvariation in quantity and quality of food has asubstantial influence on growth rate, maximumbody size, and population density (e.g., Iverson,1979, this volume; Knapp, 2001a; Tracy, this vol-ume; Wikelski and Carbone, this volume). Moredetailed analyses of vegetation on these caysshould be informative.

Body size dimorphism is fairly distinct inC. rileyi. As in most iguanas, males attain a largersize (25.4 cm SVL and 683 g; n ≥ 245 for eachmean) than females (22.5 cm SVL and 474 g;n ≥ 198 for each mean), with females averagingonly 89% of the SVL and 69% of the body massof males. An ANCOVA model (head length =sex × subspecies × SVL) showed that males alsohave significantly larger heads than do females(F1,458 = 19.30, P < 0.001, η2 = 0.04). No differ-ences in head size dimorphism were apparentamong the subspecies (P = 0.16, η2 = 0.01).Sexual dimorphism in head size exists inmost iguanas studied to date and may arise from

2 6 0 R O N A L D L . C A R T E R A N D W I L L I A M K . H A Y E S

FIGURE 18.1. Relationships between log body mass (g) andlog snout-vent length (SVL, cm) in (A) adults of threesubspecies of Cyclura rileyi and (B) adults of C. r. rileyicaptured in March, May–June, and November 1995.

TAB

LE 1

8.1

Bod

y Si

ze a

nd E

colo

gica

l Var

iabl

es fo

r Igu

anas

Sam

pled

from

All

Kno

wn

Popu

latio

ns o

f Cyc

lura

rile

yi c

rist

ata,

C. r

. nuc

halis

,and

C. r

. rile

yi

mas

s (k

g)

svl

(cm

)ec

olo

gic

al. v

aria

bles

area

pop

size

den

sity

plan

tta

xon

cay

nm

ean

±se

ran

ge

mea

n ±

sera

ng

e(h

a)(N

)(N

/ha)

spec

ies

rats

C. r

. cri

stat

aW

hit

e75

0.37

1 ±

0.02

50.

037–

0.76

020

.1 ±

0.5

10.2

–28.

014

.913

69.

120

+P

rese

nt

C. r

. nuc

halis

Fish

570.

465

± 0.

026

0.02

0–0.

830

23.8

± 0

.79.

0–31

.473

.994

8412

8.3

50+

Abs

ent

Nor

th87

0.37

8 ±

0.01

40.

046–

0.66

022

.4 ±

0.4

11.1

–28.

051

.730

3658

.750

+A

bsen

tTr

ansl

oc p

opn

541.

097

± 0.

048

0.03

0–1.

650

29.8

± 0

.69.

2–36

.03.

331

495

.247

Abs

ent

C. r

. rile

yiG

aulin

30.

650

± 0.

029

0.60

0–0.

700

24.4

± 0

.922

.8–2

5.8

1.6

Rec

entl

y—

10A

bsen

tex

tirp

ated

Gou

ldin

g35

0.55

7 ±

0.04

60.

033–

1.10

023

.1 ±

0.9

9.3–

31.0

2.9

116

40.0

16+

Abs

ent

Gre

en86

0.58

9 ±

0.02

00.

100–

0.90

023

.4 ±

0.5

8.0–

31.1

5.1

130

25.5

10A

bsen

tG

uan

a15

0.66

5 ±

0.13

90.

050–

1.20

026

.2 ±

1.8

12.1

–35.

21.

630

18.8

42P

rese

nt

Low

161.

481

± 0.

098

0.75

0–2.

300

34.2

± 0

.926

.3–3

9.5

10.8

423.

931

Pre

sen

tM

anh

ead

250.

292

± 0.

028

0.10

9–0.

500

21.3

± 0

.716

.0–2

7.0

3.3

3811

.515

Abs

ent

Pig

eon

310.

603

± 0.

042

0.10

0–1.

050

24.7

± 0

.712

.5–3

1.0

7.8

709.

07

Pre

sen

t

Sour

ces:

Ecol

ogic

al v

aria

bles

from

Hay

es e

t al.

(thi

s vo

lum

e).

Not

es:E

xclu

des

a ve

ry s

mal

l pop

ulat

ion

on th

e m

ain

isla

nd o

f San

Sal

vado

r th

at c

ould

not

be

sam

pled

. Dat

a ar

e fr

om 1

993

to 1

998.

Diff

eren

ces

in lo

wer

bou

nd o

f bod

y si

ze r

ange

gen

eral

ly r

efle

ctsu

cces

s in

the

capt

ure

of y

oung

igua

nas

and

time

of y

ear

of s

ampl

ing

(som

e po

pula

tions

wer

e no

t vis

ited

duri

ng fa

ll ha

tchl

ing

seas

on).

intrasexual selection (e.g., male-male inter-actions), intersexual selection (via female choice),differential allocation of energy for reproduc-tion, or resource partitioning (Hayes et al., thisvolume). At present, we lack data to identify thecause(s) of head size dimorphism in Cyclura.

INJURIES

We observed a range of injuries, including miss-ing toes, tail fractures, loss of a portion of thetail, frequent penetration by cactus spines, miss-ing nuchal spines, missing feet (n = 2), missingportions of the snout (n = 2), and damage to orloss of an eye (n = 2). One iguana showed frac-tures of the ribs and spine that had healed, butthis resilient individual was handicapped bylower limb paralysis. Injuries to toes and tailswere most frequent (table 18.2). We restrictedour analyses to counts of missing digits (fre-quency and number of digits lost), missing por-tions of the tail (frequency, either missing orinferred from regeneration), and length (cm) oftail regeneration.

For toe injuries (table 18.2), there were sig-nificant differences between the three sub-species (χ2

2 = 28.47, P < 0.001) and among theeleven populations (χ2

10 = 53.1, P = 0.001), butecological explanations for the differences wereunclear. The proportion of all iguanas withmissing digits was 24.9%, and all five popula-tions above this value were of C. r. rileyi. Becausehigh rates of toe injury were evident on cayscomposed largely of both sandy (Pigeon) androcky habitats (Green, Goulding), substrate dif-ferences probably do not account for populationdifferences. Although hermit crabs (Coenobitaclypeatus), which are particularly dense on SanSalvador’s cays (W. Hayes and R. Carter, unpubl.data), might contribute to the high rate of digitloss in C. r. rileyi, Pigeon Cay has few (if any) her-mit crabs. The high proportion of Low Cay igua-nas with missing digits (75%) may be related tothe skin disease detected by Auffenberg (1982b),although the mean number of lost toes is highernow than was observed in 1982 (3.1, n = 16; and1.2, n = 13, respectively). Excluding Gaulin Caybecause of small sample size, Spearman corre-

lation analyses revealed that the frequency oftoe loss among populations was independentof cay size, population size, and iguana density(table 18.2; all P > 0.054). However, there was aweak and possibly spurious relationship betweentoe loss and number of plant species (rs = –0.70,P = 0.024). Frequency of toe loss was similar forrat-infested (n = 4) and rat-free (n = 6, excludingGaulin) cays (table 18.2; Mann-Whitney U test,exact one-tailed P = 0.31).

Among all iguanas, males (30.5% of 262iguanas) were more likely to have missing toesthan were females (19.4% of 201 iguanas; χ1

2 =7.38, P = 0.007), presumably as a consequenceof more frequent or more intense agonistic be-havior. The proportion of iguanas with missingtoes increased with age class (5.0%, 10.4%, and28.5% for 20 juveniles, 67 subadults, and 390adults, respectively; χ2

2 = 14.35, P = 0.001), sug-gesting that toe loss accumulates gradually asiguanas age. The average number of toes lost(table 18.2) shows patterns similar to the fre-quency data, except that there was no relation-ship with number of plant species (P = 0.062).

For tail injuries (table 18.2), there were nodifferences between the subspecies (P = 0.36),but there were significant differences amongthe populations (χ2

10 = 34.3, P < 0.001). The pro-portion of all iguanas suffering a tail injury was35.8%, similar to that reported in some Cyclura(e.g., C. carinata carinata, C. cychlura cychlura,C. pinguis, C. cornuta stejnegeri), but greater thanthat of others (e.g., <15% in C. cychlura figginsi,C. cychlura inornata), as reviewed by Iverson etal. (this volume). Populations exceeding thisvalue included three C. r. rileyi populations(Guana, Low, and Pigeon Cays), the translocatedpopulation of C. r. nuchalis, and the single C. r.cristata population. All of these populations co-exist with rats except for the translocated popu-lation, although we suspect rats may be presentthere as well. Consequently, iguanas on con-firmed rat-infested cays had a higher frequencyof tail breaks than those on rat-free cays (49.3%and 31.8%, respectively; Mann-Whitney U = 3.0,exact one-tailed P = 0.034). Although rats mightbe able to sever the tail of juvenile iguanas, we

2 6 2 R O N A L D L . C A R T E R A N D W I L L I A M K . H A Y E S

TAB

LE 1

8.2

Des

crip

tive

Stat

istic

s fo

r Igu

anas

Sam

pled

Sho

win

g In

jurie

s

mis

sin

g d

igit

(s)

inju

red

tai

lta

il r

egen

erat

ion

nu

mbe

rpo

rtio

npe

rcen

tag

ere

gro

wth

(cm

)ta

xon

cay

npe

rcen

tag

e(m

ean

±s.

e.)

nm

issi

ng

(%)

of

inju

red

(mea

n ±

s.e.

)ra

ts

C. r

. cri

stat

aW

hit

e75

16.0

0.23

±0.

0775

42.7

90.6

5.9

±0.

8P

rese

nt1

C. r

. nuc

halis

Fish

5712

.30.

16 ±

0.06

5724

.685

.76.

8 ±

1.1

Abs

ent

Nor

th87

14.9

0.26

±0.

0886

32.6

96.4

7.9

±0.

5A

bsen

tTr

ansl

oc p

opn

5320

.40.

37 ±

0.12

5451

.989

.37.

4 ±

1.0

Abs

ent

C. r

. rile

yiG

aulin

333

.30.

33 ±

0.33

30.

0—

—A

bsen

tG

ould

ing

3534

.30.

74 ±

0.21

3522

.987

.58.

0 ±

2.0

Abs

ent

Gre

en86

33.7

0.76

±0.

1684

22.6

63.2

6.4

±1.

2A

bsen

tG

uan

a15

13.3

0.20

±0.

1415

40.0

83.3

6.3

±1.

9P

rese

nt

Low

1675

.03.

06 ±

0.69

1650

.037

.54.

2 ±

1.2

Pre

sen

t1

Man

hea

d19

21.1

0.32

±0.

1722

36.4

62.5

5.5

±1.

5A

bsen

tP

igeo

n31

51.6

0.94

±0.

2431

64.5

80.0

7.4

±0.

8P

rese

nt

Tota

ls47

824

.90.

52 ±

0.06

478

35.8

82.5

6.9

±0.

3

1R

ecen

tly e

radi

cate

d.

doubt they could do much damage to adult tailsand concur with others (e.g., Vitt et al., 1974;Iverson, 1979; Jaksic and Greene, 1984; Iversonet al., this volume) that explanations other thanpredation (i.e., social interactions) should be con-sidered for causes of tail loss. Spearman corre-lation analyses revealed that the frequency of tailloss among populations (excluding Gaulin Cay)was independent of cay size, population size,iguana density, and number of plant species (allP > 0.15).

Males (38.3% of 262 iguanas) and females(31.2% of 202 iguanas) were equally likely tosuffer tail injury (P = 0.11). Tail breaks were re-ported with equal frequency in males and fe-males of most Cyclura studied to date (C. c. car-inata, C. c. cychlura, C. c. figginsi, C. c. inornata),but males experienced disproportionately highbreakage in C. c. stejnegeri and C. pinguis (re-viewed by Iverson et al., this volume). In ouranalyses, the frequency of tail injury was similarfor all size classes (35.0%, 37.9%, and 35.5% for20 juveniles, 66 subadults, and 391 adults, re-spectively; P = 0.93). This finding implies thatthe majority of tail losses occur when iguanasare young. If adults were equally likely to losetheir tails, then adults would have a greater cu-mulative frequency of tail loss than younger ageclasses, as the primary evidence of such injuries(regeneration) persists through life. However,there are survival and social costs associated withtail loss (e.g., Iverson, 1979; Fox et al., 1990; Wil-son, 1992) that are not reflected in our data. Ifyoung with lost tails have lower survivorship,then tail loss in adult iguanas could be morefrequent than suggested by our data.

For those iguanas that have lost a portion oftheir tail (n = 141), 82.5% showed regeneration(table 18.2). This proportion was consistentamong the populations, sexes, and size classes.The mean length of regenerated tissue was6.9 ± 0.3 cm (n = 141), and likewise was similaramong the populations, sexes, and size classes.The maximum regeneration length was 17.3 cm.Iverson (1979) concluded that tail regenerationappears to be rapid in both juvenile and adultC. c. carinata. Healing of the tail sometimes re-

sulted in distinct bifurcation (n = 5) or trifurca-tion (n = 1) of regenerated portions, as noted forC. c. carinata, and is presumably the result ofincompletely severed tails (Iverson, 1979).

FEMORAL PORE COUNTS

The number of femoral pores differed signifi-cantly among the three taxa (one-way ANOVA,F2,467 = 113.4, P < 0.001, η2 = 0.33; table 18.3).Scheffé post-hoc comparisons indicate thateach taxon differed significantly from the others.Within C. r. nuchalis, the three populations dif-fered significantly (one-way ANOVA, F2,186 =6.81, P = 0.001, η2 = 0.07), with counts onNorth Cay higher than those on Fish Cay or inthe translocated population. The founders ofthe translocated population originated from FishCay (Hayes et al., this volume), and the meancounts from these cays were indistinguishable,although variance in the translocated populationwas lower. Within C. r. rileyi, the six populations(excluding that on Gaulin) also varied signifi-cantly (one-way ANOVA, F2,200 = 2.53, P = 0.03,η2 = 0.06), but Scheffé post-hoc comparisonsdetected no pairwise differences among cays (allP > 0.085). For the entire data set, the count offemoral pores was similar for males and females(P = 0.43).

HEAD SCALATION CHARACTERS

Undergraduate student Melissa Andres con-ducted head scale counts based on careful reviewof 35 mm color slides taken of 223 specimenscaptured in the field. The photos were close-upshots of iguana heads (usually one dorsal viewand one lateral view) intended originally for iden-tification purposes. A summary of these counts,based on scale names in Smith (1995), is pre-sented in table 18.3. Unfortunately, completescale counts were not available for most iguanasbecause of limited photo quality (lighting, angle,and focus), skin conditions that obscured somefeatures (ecdysis remnants, soil or food residues,and scarring), and the application of conserva-tive criteria to data collection. Tests for sexualdimorphism within individual populations hav-ing sufficient samples (White Cay for C. r. cristata;

2 6 4 R O N A L D L . C A R T E R A N D W I L L I A M K . H A Y E S

all three populations of C. r. nuchalis; Green Cayfor C. r. rileyi) showed only one character to bedimorphic (RN [defined in table 18.3] for twopopulations of C. r. nuchalis; Mann-Whitney two-tailed P < 0.05 for both), but with opposite pat-terns on each cay.

One-way ANOVAs followed by Scheffé post-hoc comparisons were used to separately com-pare variance between the three subspecies, andvariance within C. r. nuchalis and C. r. rileyi, inscalation features. Although much of the datafailed to meet the assumptions of parametrictests, nonparametric Kruskal-Wallis ANOVAsconducted on the same data sets yielded identi-cal results. The analyses (table 18.3) showed thatmeans for twelve of the thirteen scalation char-acters differed significantly among the threesubspecies. Seven of these characters differedsignificantly between taxa but not across popu-lations within taxa. Thus, a combination of thesecharacters might lead to reliable diagnosis. Onecharacter, number of infralabials to eye center(IL), revealed the close affinity of the translo-cated C. r. nuchalis population (7.1 ± 0.1) to itssource population on Fish Cay (7.1 ± 0.1). Bothof these populations differed significantly fromNorth Cay (6.4 ± 0.1; Scheffé P < 0.001).

DISCRIMINANT ANALYSES

Stepwise discriminant function analysis (DFA),which is robust to violations of parametric as-sumptions (McGarigal et al., 2000), was con-ducted to further evaluate distinctiveness of thethree taxa using counts of femoral pores andselect head scales. Differences in these charac-ters are generally regarded as nonadaptive and,therefore, are suitable for phylogenetic evalu-ation (e.g., Snell et al., 1984; Hollingsworth,1998). The predictor variables included numberof femoral pores (FP), in addition to most headscalation characters (BF and PF in table 18.3were excluded due to small samples). Because ofmissing data for many characters, only a subsetof iguanas was available for the analysis. The fi-nal model, which included sixty-eight iguanasand six predictors (FP, BP, NA, SP, ST, BL; seetable 18.3 for their definitions and values), re-

veals a high degree of distinctiveness betweenthe three subspecies (Wilks’s λ = 0.173, χ2

12 =100.8, P < 0.001). Cyclura r. cristata (n = 11) wascorrectly classified 100% of the time. C. r. nuchalis(n = 26) was correctly assigned 84.6% of thetime (twenty-two cases), and was incorrectlyclassified as C. r. rileyi in four cases (15.4%). C. r.rileyi (n = 31) had the lowest percentage of cor-rect predictions (twenty-five cases; 80.6%) andwas incorrectly classified as C. r. nuchalis in fourcases (12.9%) and as C. r. cristata in two cases(6.5%). Characters FP and SP provided thegreatest discrimination. A canonical plot ofthe discriminant function (DF) scores for indi-vidual iguanas shows clustering (figure 18.2).Whereas C. r. cristata and C. r. nuchalis areclearly distinct from each other, C. r. rileyi showsintermediate characters.

Principal components analysis was alsoconducted on the correlation matrix of the sixcharacters identified by the final DFA model.Three principal components (PC) were extracted(using Varimax rotation), with factor loadingsshown in table 18.4. PC1 is composed largely ofcharacters BP, NA, and SP; PC2 is composedlargely of characters FP and ST; and PC3 islargely made up of character BL. None of thesecomponents showed sexual dimorphism. Thefactor scores of each individual iguana areplotted in figure 18.3, where clustering of thethree subspecies is evident, with C. r. rileyi ex-hibiting intermediate scores. DFA of the factorscores again revealed a high degree of distinc-tiveness among the three taxa (Wilks’s λ = 0.194,χ6

2 = 104.8, P < 0.001). However, the proportionof iguanas correctly assigned was somewhatlower than that achieved with DFA of originaldata. C. r. cristata (n = 11) was correctly classified90.9% of the time (ten cases) and was incor-rectly grouped with C. r. rileyi in one case. C. r.nuchalis (n = 26) was correctly assigned 84.6%of the time (twenty-two cases) and was incor-rectly classified as C. r. rileyi in four cases (15.4%).C. r. rileyi (n = 31) was correctly classified 77.4%of the time (twenty-four cases) and was incor-rectly classified as C. r. nuchalis in five cases(16.1%) and as C. r. cristata in two cases (6.5%).

C O N S E R V AT I O N O F A N E N D A N G E R E D R O C K I G U A N A , I I 2 6 5

TAB

LE 1

8.3

Com

paris

ons

of F

emor

al P

ore

Cou

nts

and

Scal

atio

n D

iffer

ence

s fo

r Pop

ulat

ions

of C

yclu

ra r

ileyi

cri

stat

a,C

. r. n

ucha

lis, a

nd C

. r. r

ileyi

C. r

. cri

stat

aC

. r. n

ucha

lisC

. r. r

ileyi

p

char

acte

rn

mea

ns.

e.ra

ng

en

mea

ns.

e.ra

ng

en

mea

ns.

e.ra

ng

eth

ree

ssp.

Crn

Crr

FP

: Tot

al n

um

ber

of

7242

.6a

0.3

35–4

918

947

.1b

0.2

41–5

520

943

.6c

0.2

35–5

10.

000

0.00

10.

030

fem

oral

por

es

BF

: Sca

les

bord

erin

g 5

6.4

0.4

5–7

256.

40.

24–

828

6.5

0.1

5–8

0.80

70.

884

0.74

1fr

onta

l

BP

: Sca

les

betw

een

14

1.9a

0.1

1–2

470.

6b0.

10–

154

1.1c

0.1

0–2

0.00

00.

415

0.00

5pr

efro

nta

ls (m

inim

um

)

CS:

Ch

in s

hie

lds

to

186.

6a0.

25–

811

56.

0b0.

15–

774

6.1b

0.1

5–8

0.00

40.

958

0.34

8ey

ecen

ter

IC: S

cale

row

s be

twee

n

192.

6a0.

12–

311

72.

0b~0

1–3

822.

1c0.

11–

30.

000

0.15

00.

000

infr

alab

ials

an

d ch

in

shie

lds

belo

w e

yece

nte

r

IL: I

nfr

alab

ials

to e

yece

nte

r19

6.7

0.2

5–8

110

6.9

0.1

5–8

766.

70.

15–

80.

044

0.00

00.

316

NA

: Sca

le r

ows

betw

een

14

0.9a

0.1

0–1

431.

6b0.

10–

461

0.9a

0.1

0–3

0.00

00.

288

0.08

4n

asal

s an

d fi

rst a

zygo

us

scal

e

PF

: Sca

le r

ows

betw

een

2

5.5a

0.5

5–6

254.

1b0.

13–

512

4.3b

0.1

4–5

0.00

00.

851

0.58

8pr

efro

nta

ls a

nd

fron

tal

(min

imu

m)

RN

: Sca

les

touc

hin

g ro

stra

l18

2.0a

0.0

2–2

991.

4b0.

11–

270

1.7c

0.1

1–2

0.00

00.

345

0.00

0be

twee

n n

asal

s an

d su

pral

abia

ls

SL: S

upr

alab

ials

to

197.

4a0.

26–

910

46.

9b0.

16–

868

6.7c

0.1

6–8

0.00

00.

446

0.35

6ey

ecen

ter

SN: T

otal

nu

mbe

r of

20

5.0ab

0.3

4–7

108

4.8a

0.1

4–6

715.

5b0.

14–

80.

000

0.06

90.

573

supr

anas

als

and

post

nas

als

SP: S

cale

s be

twee

n

152.

1a0.

11–

368

1.0b

~01–

269

1.3c

0.1

1–2

0.00

00.

382

0.73

4su

pran

asal

s an

d pr

efro

nta

ls (m

inim

um

)

ST: S

upr

alab

ials

to

167.

2a0.

16–

880

7.0ab

0.1

6–8

566.

7b0.

16–

100.

013

0.50

80.

382

tria

ngu

lar

scal

e at

m

outh

cor

ner

BL:

Su

blab

ial i

sola

ted

19P

rese

nt:

1 (5

%)

114

Pre

sen

t: 45

(39%

)80

Pre

sen

t: 25

(31%

)0.

0121

——

ante

rior

ly f

rom

oth

er

Abs

ent:

18 (9

5%)

Abs

ent:

69 (6

1%)

Abs

ent:

55 (6

9%)

subl

abia

ls (p

rese

nce

or

abse

nce

in in

divi

dual

s)

Not

es:P

roba

bilit

y le

vels

(P,

base

d on

one

-way

AN

OVA

s un

less

oth

erw

ise

indi

cate

d) a

re s

how

n fo

r su

bspe

cific

diff

eren

ces

(com

pari

son

of th

ree

taxa

) an

d fo

r po

pula

tion

diff

eren

ces

with

in C

. r.

nuch

alis

(thr

ee p

opul

atio

ns)

and

C. r

. rile

yi(u

p to

six

pop

ulat

ions

). F

or c

hara

cter

s ha

ving

a s

igni

fican

t mai

n ef

fect

, mea

ns th

at d

iffer

sig

nific

antly

from

thos

e of

oth

er s

ubsp

ecie

s ar

e in

dica

ted

bydi

ffer

ent s

uper

scri

pts

(a, b

, or

c; S

chef

fé p

ost-

hoc

cont

rast

s).

1χ2

test

.

Of the three components, PC1 provided the great-est discrimination.

These analyses suggest subspecific and pop-ulation differences that warrant a formal studyinvolving more complete scale counts. The ad-dition of other scalation characters and molec-

ular data to the DF would increase the likeli-hood of complete diagnosis of the three taxa.Schwartz and Carey (1977), who provided thecurrent taxonomy of this group, thought thatC. r. cristata might be sufficiently distinct fromthe other two forms to warrant specific status.Our data here lend some support to this idea.Schwartz and Carey (1977) also remarked thatC. r. cristata has characters intermediate betweenC. cychlura, which likewise occurs on the GreatBahama Bank, and the other forms of C. rileyithat are isolated on separate banks.

CONSERVATION PRIORITIES

A unifying goal in conservation biology is toidentify and conserve genetically important, nat-urally occurring populations, thus allowing thedynamic process of evolution to continue un-affected by human factors as much as possible.Unfortunately, present-day anthropogenic pres-sures make this goal virtually unattainable, es-pecially for populations endemic to insular eco-systems, which are particularly vulnerable toinvasive species and habitat degradation. A morepractical goal is to preserve and manage eco-systems. To achieve this, conservation programs

2 6 8 R O N A L D L . C A R T E R A N D W I L L I A M K . H A Y E S

FIGURE 18.2. Canonical plot of the discriminant function(DF) scores for each individual of the three subspecies ofCyclura rileyi ( , C. r. cristata, n = 11; +, C. r. nuchalis, n = 26;♦, C. r. rileyi, n = 31). Open circles denote group centroidsfor the three taxa.

TABLE 18.4Factor Loadings of Each Character

for the Three Principal Components

factor loadings

character pc1 pc2 pc3

FP −0.149 0.687 0.404

BP 0.829 0.040 0.084

NA −0.669 0.052 −0.076

SP 0.773 0.046 −0.249

ST 0.125 0.837 −0.201

BL 0.030 −0.001 0.902

Variance 29.5 19.7 18.2explained (%)

Notes: Principal Components (PC) extracted (with Varimaxrotation) from the correlation matrix of scalation characterstaken from adult Cyclura rileyi. For each PC, the characterswith the highest factor loadings (>0.60) are in bold. Totalvariance extracted was 67.4%. Characters are defined intable 18.3.

FIGURE 18.3. Three-dimensional plot of the principalcomponent (PC) scores for each individual of the threesubspecies of Cyclura rileyi ( , C. r. cristata, n = 11; +, C. r.nuchalis, n = 26; ♦, C. r. rileyi, n = 31). Each of the threeprincipal components represents one or more scalationcharacters, as summarized in table 18.4 and defined in table 18.3.

are often directed toward umbrella or keystonespecies, because the preservation of an individ-ual species requires protection of its habitat andother aspects of ecosystem function. We believethat many species of Cyclura qualify as umbrellaspecies; additional ecological studies will revealtheir potential as a keystone species.

Here we recommend the implementation ofa series of conservation initiatives and new re-search foci that ideally will move C. rileyi fromcrisis management to full remediation and re-covery from endangered status. This task will re-quire carefully prioritized planning, the cooper-ative efforts of many private and governmentalentities, and must be data driven. We suggest thata research-based approach to conservation ismost beneficial in the long term and that C. rileyican serve as a model species for developing con-servation plans for other West Indian iguanas.

CLARIFICATION OF

SYSTEMATIC RELATIONSHIPS

Advances in DNA sequencing, informatics, andfamily tree analysis have helped to answer thecontentious question of what units of life to pre-serve. Divergent and often conflicting defini-tions of species and subspecies and the unequalapplication of these concepts across taxa havecomplicated conservation efforts. The evolution-arily significant units (ESU) category was cre-ated to identify units in nature that, if saved,would help preserve biodiversity, evolutionaryprocesses, and ecosystems (Ryder, 1986; Moritz,1994, 1999). ESU criteria were developed tohelp prioritize the use of limited conservationresources. Criteria for ESU designation are de-bated and range from absence of gene flow tothe more commonly accepted criterion of mtDNAsequence-based reciprocal monophyly. For taxathat do not meet the stringent conditions ofstrong phylogenetic branches, Moritz (1999) hassuggested a lesser category: a management unit(MU). Management units can be distinguishedby different alleles being more common in oneMU than in another.

Malone et al. (2000) recently examined thephylogeographic relationships of Cyclura using

a mtDNA ND4 to a leucine transfer RNA (tRNA)sequence and suggested conservation prioritiesamong the species based largely on ESU criteria,with the goal of preserving as much biodiversityas possible. Their results support a southeast-to-northwest pattern of island divergence for thegenus, the phylogenetic basal status of C. pinguisand C. collei, and the more recent establishmentof C. rileyi in the Bahamas. These findings haveprofound implications for establishing conser-vation priorities among West Indian iguana taxaand support the ranking of C. pinguis and C. colleias priority species because of their ancient di-vergence and imperiled status (Alberts, 2000).Malone et al. (2000) further questioned the ap-propriateness of subspecific recognition withinC. rileyi, based on their findings of no haplotypicdifferences among the subspecies.

Recognition of each of the three subspeciesof C. rileyi as ESUs may be debatable, depend-ing on the criteria used. However, the sub-species certainly do not meet the stringent ESUcriterion of reciprocal monophyly for mtDNAsequences (Moritz, 1999). The second ESU cri-terion, often used for populations that are sepa-rated by a significant genetic distance, has notbeen formally evaluated at the population level.We concur with Malone et al. (2000) that finalresolution of the taxonomic status of C. rileyishould be determined by analysis of more rap-idly diverging nuclear markers, such as micro-satellites. In our lab, we have used random am-plified polymorphic DNA (RAPD) as a first-levelsurvey tool and have initiated microsatellitestudies. Even though these assays are incom-plete, preliminary results from RAPD bands in-dicate the absence of gene flow between islandpopulations and band-sharing differences that,if confirmed by microsatellites, will inform thediscussion. Discriminant analyses of our mor-phological data, although based on a limitednumber of characters, suggest that a suite ofmorphological features exist that, with addi-tional data and further analysis, will allow forcomplete diagnosis of one or more of the threesubspecies and possible elevation to a higherconservation priority.

C O N S E R V AT I O N O F A N E N D A N G E R E D R O C K I G U A N A , I I 2 6 9

Given the historic and present isolation ofthe subspecies (each more than 100 km apart andpossibly on their own evolutionary trajectory)and the apparent morphological differences, webelieve these populations to be at least impor-tant conservation MUs, if not emerging ESUs.Although we advocate ESU priority-settingcriteria for determining the use of limited con-servation resources, we believe it would be pre-mature and unfortunate to alter conservationplanning by declaring C. rileyi as a single ESUbased solely on mtDNA and limited microsatel-lite samples. Clearly, further genetic analysis isneeded to understand more fully this taxon’sevolutionary past, current levels of diversity,and potential for adaptive change in the future.

Apart from questions of a population’s ESUstatus and the amount of genetic diversity be-tween populations (MU status), there are addi-tional reasons to direct conservation activitiestoward C. rileyi populations. By doing so, we em-phasize the need to save and manage ecosys-tems rather than species. These iguanas are ide-ally suited as umbrella species, and conservationefforts to save them will provide protection formany additional species of plants and animals.For example, several cays that harbor iguanashost some of the largest and most diverse sea-bird communities in the Bahamas, and theseurgently need protection. Furthermore, we be-lieve that iguana conservation in general isheuristically and esthetically valuable at local,national, and regional levels, and needs the fullsupport of all stakeholders.

FORMAL PROTECTION OF

EXISTING POPULATIONS

The greatest immediate threat to C. rileyi is thatexisting populations or habitats will become fur-ther diminished in size or quality. At present,only the translocated population in the northernExumas is afforded formal protection within theNational Park system managed by the BahamasNational Trust (BNT). Although some of thecays supporting iguanas are privately owned(see Hayes et al., this volume), others are CrownLand (public lands owned by the nation and not

presently granted to anyone) that deserve con-sideration for inclusion within the National Parksystem.

We urgently recommend protection of C. ri-leyi populations by the creation of three nationalpark units. First, and perhaps most important,the addition of White Cay as a satellite unit to theexisting Exumas Land and Sea Park (or anotherrecently proposed Exumas park) would benefitC. r. cristata. Second, at least two disjunct re-gions of San Salvador Island should be desig-nated as a new national park to protect C. r. rileyi,including the northern cays of Grahams Harborthat host iguanas (Gaulin and Green Cays) anddense colonies of seabirds (Gaulin, Catto, andWhite Cays), and the interior cays within thesouthern portion of Great Lake that harbor igua-nas (Pigeon Cay) and dense rookeries of heronsand cormorants (Cormorant Cay and adjacentislets). This new park would protect not onlyiguanas, but also the most diverse and possiblythe largest remaining seabird and waterbirdcolonies in the entire archipelago (W. Hayes,unpubl. data). The remaining cays that supportC. r. rileyi are privately owned, but their purchaseby BNT warrants consideration. Third, Fishand North Cays in the Acklins Bight should beincorporated within a new national park for thebenefit of C. r. nuchalis and other sensitivefauna, including conch, sharks, bonefish, andthe flock of flamingos that regularly forage there.This new park should include other nearby caysthat appear suitable for translocation of newiguana populations. In collaboration with theIUCN Iguana Specialist Group, we have pre-sented these ideas to the Bahamian govern-ment and the BNT (Carey et al., 2001), and urgetimely enactment of these recommendations.The desire and commitment of the BNT toprotect the iguanas and their habitats was ex-pressed as early as 1983 in a document sub-mitted to the Bahamas government (S. Buckner,pers. comm.).

To enforce protection of these populations, werecommend instituting a system of wardens, asimplemented for other national parks, and plac-ing informational signs on protected cays similar

2 7 0 R O N A L D L . C A R T E R A N D W I L L I A M K . H A Y E S

to those already posted on Green Cay by theInternational Iguana Society. The signs shouldserve notice that the iguanas are protected bynational and international law and warn againstbringing animals (e.g., dogs, cats) to the cays.Seasonal restrictions on visitation of the caysshould also be considered, to protect iguana andseabird nests, which are especially vulnerable todisturbance (e.g., trampling, overheating). Aneducational program, as discussed below, wouldincrease vigilance by local residents.

CONTINUED RESEARCH

Conservation efforts have become increasinglyfocused and publicly recognized for a number oftaxa, including West Indian iguanas. The effec-tiveness of individual programs depends onmany factors, including the relative rarity of thespecies, the quality and quantity of remaininghabitat, the levels of funding and interventionrequired to preserve a species, and political is-sues that may benefit or impede the program.The basic biology of the organism—arguably themost important factor—is often the least appre-ciated and most neglected facet of recovery pro-grams. Although our research has provide-important and useful data on the biology ofC. rileyi, we are far from fully understandinghow iguanas interact with their environment. Asa result, we cannot provide reliable estimates ofthe minimum viable population and area neededto sustain the species. Because C. rileyi popula-tions vary considerably in demography, habitatdiversity, behavioral ecology, and levels of threatfrom invasive species and habitat degradation,this species is ideally suited for research on therelationships between behavioral ecology, lifehistory, and local adaptation.

INVASIVE SPECIES CONTROL, HABITAT

RESTORATION, AND MONITORING

In addition to habitat protection, conservationmanagement often involves habitat restorationand enrichment. Further study is needed tobetter understand the nutritional and nestingrequirements of iguanas and the direct or indi-rect impact of rats and other invasive species. Al-

though we have undertaken considerable efforton several cays to control invasive species and re-store nesting habitat (Hayes et al., this volume),continued monitoring will be essential to evalu-ate the success of our efforts and detect newthreats. The methods of rodenticide deliveryappear to work, and should be applied to otherpopulations that still coexist with rats. Althoughthere is no available means to control the Cacto-blastis moths that are decimating Opuntia caction San Salvador’s offshore cays, the loss of veg-etation important for food and cover could beameliorated by the planting of alternative plantspecies. The removal of invasive Australian pine(Casuarina) is also needed on several cays. Weanticipate that complete removal of Casuarinafrom White Cay would greatly reduce the num-ber of falcons that visit the island and use thetrees as hunting perches during migration andwinter. Our methods of estimating populationsize will allow us to continue accurate monitor-ing and should be suitable for surveys of otheriguana populations.

ESTABLISHMENT OF NEW POPULATIONS

Reintroductions of either wild-captured orcaptive-headstarted iguanas may be useful forsupplementing existing populations (whenoutbreeding depression can be avoided), creat-ing new populations, and selectively augment-ing genetic diversity within existing or new pop-ulations (e.g., Knapp and Hudson, this volume;Welch et al., this volume). We urgently recom-mend the initiation of translocation projects tobenefit all three subspecies of C. rileyi andsuggest the following three approaches. First,because C. r. cristata is presently confined to asingle cay, one or more additional cays mustbe found that are (or can be rendered) suitablefor a translocated population. Nearby Leaf Cayappears to be ideal, except that the dense ratpopulation must first be extirpated and permis-sion granted by the owner. Other cays in thesouthern Exumas should be considered as well.Because of the great difficulty in locating andcapturing adult females, we suggest that hatch-lings (or older juveniles) be translocated unless

C O N S E R V AT I O N O F A N E N D A N G E R E D R O C K I G U A N A , I I 2 7 1

a captive headstarting program can be estab-lished as an intermediate step. Second, for C. r.rileyi, we recommend the restocking of recentlyextirpated cays with individuals from nearby cays(e.g., Green to Gaulin, Low to High, Guana toBarn, and Manhead to Cut). Furthermore, werecommend establishment of a new populationon the main island via translocation of indi-viduals from Goulding Cay to the relativelyundisturbed area east of Storrs Lake. To maxi-mize genetic diversity, individuals from otherpopulations could be released as well (but seeWelch et al., this volume). Radiotracking trans-located individuals will be essential to determinehow far they might wander from the relativesafety of this remote area. As a third approach,we recommend the translocation of C. r. nuchalisfrom North Cay and Fish Cay to six unoccupied,smaller, nearby cays in the Acklins Bight. Usingthese islands, we hope to investigate experimen-tally the relative success of translocation at dif-ferent times (before versus after the matingseason) and under different conditions (intooccupied versus unoccupied habitats).

As a final consideration, all of the cays in theAcklins Bight are low elevation (see Hayes et al.,this volume) and therefore subject to inundationby storm surges and rising sea levels. Obviously,long-range considerations are important for con-servation planning. We suspect that the historicloss of populations on the six smaller cays re-sulted from previous storms, and we should notassume that North Cay and Fish Cay are invul-nerable to similar events in the future. Thus,although the small population introduced tothe northern Exumas occupies an island withgreater relief (see Hayes et al., this volume), weneed to search for yet another safe haven withinthe Acklins-Crooked-Long Cay island group.

ESTABLISHMENT OF HEADSTART PROGRAMS

For political and logistical reasons, no headstartprograms exist for any of the Bahamian iguanas.Because of the success of headstart programswith other iguanas (Hudson, 2000a; Alberts etal., this volume; Wilson et al., this volume), westrongly recommend that such programs be ini-

tiated for the Bahamian species. Specifically, werecommend establishing captive rearing facili-ties for C. r. rileyi at the Gerace Research Center(formerly Bahamian Field Station) on San Sal-vador Island and for C. r. cristata at a yet to beidentified facility on Great Exuma. To safeguardagainst the introduction of exotic disease and re-duce pressure on natural populations from com-mercial herpetoculturists, we urge that captivepropagation and/or rearing of C. rileyi be con-ducted solely in the Bahamas.

DEVELOPMENT AND IMPLEMENTATION

OF EDUCATIONAL PROGRAMS

Education will be central to the full recovery ofC. rileyi. As our work with C. rileyi progresses,we plan to devote a greater proportion of ourtime to conservation education. Although wehave engaged college students in our researchactivities and have sponsored one Bahamianstudent for Ph.D. studies at Loma Linda Uni-versity, California, general interest in careers inconservation appears limited by few employmentopportunities in the Caribbean region. However,working with younger students has been moreencouraging. Our recent work with San Salvadorschool children through the local Boy Scoutschapter and their involvement in habitat restora-tion (Hayes et al., this volume) was especiallywell received. Currently we are working with theDisney Corporation through their conservationgrants program to develop hands-on learningmodules that will engage preteens and teens inconservation problem-solving activities. Thesemodules will include practical activities thatdemonstrate, among other things, the tools ecol-ogists use to study animals (e.g., radiotelemetry)and the impact humans have on iguana andseabird populations (e.g., loss of eggs due totrampling and overheating). These projects arein addition to the general lectures we offer thatheighten awareness and appreciation of iguanasand the fragile ecosystems of the Bahamas.Although our efforts are focused locally on theisland groups where we work, we recognize theneed for a broader emphasis on conservation ed-ucation throughout the islands. Success will be

2 7 2 R O N A L D L . C A R T E R A N D W I L L I A M K . H A Y E S

measured ultimately by the degree to which theBahamian people become invested in conser-vation activities, taking full ownership of theprocess of protecting and managing their na-tional treasures.

ACKNOWLEDGMENTS

As in the preceding chapter, this work benefitedsubstantially from the assistance of numerousvolunteers. In addition to the individuals men-tioned there, we thank Melissa Andres for hercareful evaluation of the close-up photos thatled to our discriminant analyses. We have alsobenefited from local support provided by the fol-lowing friends: Kenneth Buchan, Don Geraceand Kathy Gerace, and Dan Suchy of the GeraceResearch Center (San Salvador); Basil Minns,Raymond Sears, and Bailey Smith (Great Exuma);

Cindy Bates, David Meadows and Nell Meadows,and Elwood Gibson (Crooked Island); and SandraBuckner and Paul Harding (Nassau). Eric Careyand Maurice Isaacs, of the Bahamas Departmentof Agriculture, enthusiastically provided researchpermits and were always a delight to interactwith. Allison Alberts, Sandra Buckner, DonGerace and Kathy Gerace, John Iverson, CharlesRadcliffe, and Rich Reading offered informativediscussion, encouragement, and support. Fund-ing was provided by out-of-pocket contribu-tions, David Winters (donation of a boat motor),Southern Adventist University, Loma Linda Uni-versity, the Denver Zoological Society, the Chi-cago Zoological Society, the Bahamas Departmentof Agriculture, and the Disney Foundation’sconservation program.

C O N S E R V AT I O N O F A N E N D A N G E R E D R O C K I G U A N A , I I 2 7 3