Citation
Egg survivorship and primary sex ratio of green turtles, Chelonia mydas, at Tortuguero, Costa Rica

Material Information

Title:
Egg survivorship and primary sex ratio of green turtles, Chelonia mydas, at Tortuguero, Costa Rica
Creator:
Horikoshi, Kazuo, 1956-
Publication Date:
Language:
English
Physical Description:
vii, 158 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Green turtle -- Nests -- Costa Rica ( lcsh )
Green turtle -- Breeding -- Costa Rica ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1992.
Bibliography:
Includes bibliographical references (leaves 149-157)
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Kazuo Horikoshi.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
028623400 ( ALEPH )
27634832 ( OCLC )

Downloads

This item has the following downloads:

AA00063501_00001.pdf

00027.pdf

00028.pdf

00057.pdf

00080.pdf

00159.pdf

00050.pdf

00046.pdf

00079.pdf

00111.pdf

00143.pdf

00019.pdf

00071.pdf

00103.pdf

00101.pdf

00075.pdf

00047.pdf

00025.pdf

00141.pdf

00158.pdf

00039.pdf

00086.pdf

00001.pdf

00017.pdf

00095.pdf

00120.pdf

00044.pdf

00023.pdf

00139.pdf

00008.pdf

00077.pdf

00038.pdf

00026.pdf

00162.pdf

00054.pdf

00112.pdf

00085.pdf

00070.pdf

00135.pdf

00030.pdf

00011.pdf

00131.pdf

00082.pdf

00084.pdf

00122.pdf

00154.pdf

00110.pdf

00091.pdf

00145.pdf

00121.pdf

00004.pdf

00156.pdf

00056.pdf

00117.pdf

00144.pdf

00053.pdf

00083.pdf

00066.pdf

00073.pdf

00088.pdf

00146.pdf

00125.pdf

00018.pdf

00020.pdf

00166.pdf

00127.pdf

00142.pdf

00074.pdf

00031.pdf

00129.pdf

00092.pdf

00115.pdf

00151.pdf

00022.pdf

00147.pdf

00081.pdf

00099.pdf

00061.pdf

00069.pdf

00134.pdf

00126.pdf

00068.pdf

00140.pdf

00049.pdf

00036.pdf

00096.pdf

00002.pdf

00076.pdf

00098.pdf

00072.pdf

00065.pdf

00032.pdf

00106.pdf

00113.pdf

00024.pdf

00009.pdf

00041.pdf

00062.pdf

00012.pdf

00132.pdf

00167.pdf

00118.pdf

00124.pdf

00097.pdf

00052.pdf

00003.pdf

00114.pdf

00067.pdf

00104.pdf

00090.pdf

00059.pdf

00094.pdf

00160.pdf

00155.pdf

00087.pdf

00048.pdf

00015.pdf

00123.pdf

00040.pdf

00128.pdf

00152.pdf

00102.pdf

00005.pdf

00060.pdf

00100.pdf

00037.pdf

00007.pdf

00150.pdf

00108.pdf

00051.pdf

00148.pdf

00163.pdf

00164.pdf

00064.pdf

00119.pdf

00161.pdf

00043.pdf

00105.pdf

00010.pdf

00133.pdf

00055.pdf

00013.pdf

00137.pdf

00089.pdf

00063.pdf

00093.pdf

00130.pdf

00157.pdf

00116.pdf

00165.pdf

00138.pdf

00058.pdf

00033.pdf

00153.pdf

00021.pdf

00034.pdf

00029.pdf

00136.pdf

00042.pdf

00149.pdf

00006.pdf

00107.pdf

00109.pdf

00016.pdf

00078.pdf

00014.pdf

00045.pdf

00035.pdf


Full Text
EGG SURVIVORSHIP AND PRIMARY SEX RATIO OF GREEN TURTLES,
CHELONIA MYDAS, AT TORTUGUERO, COSTA RICA
BY
KAZUO HORIKOSHI
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA 1992




ACKNOWLEDGEMENTS
First I thank my parents Shigeru and Kiku Horikoshi, and my
wife Harumi Horikoshi, whose love and encouragement strengthened my resolve to complete this task.
I would like to thank the members of my committee: Dr. Martha Crump, Dr. Karen Bjorndal, Dr. Jack Kaufmann, Dr. Clay Montague and Dr. Jeanne Mortimer who offered continuous encouragement and advice throughout this study. I feel fortunate to have been able to work under their guidance. I also would like to thank the late Dr. Archie Carr who offered invaluable encouragement and an opportunity to undertake this study at the Tortuguero beach. Dr. Takako Oshima provided advice on statistics. I am very grateful to Dr. Louis Guillette who was most generous in allowing me use of his histology laboratory. Dr. Alan Bolten and Dr. Blair Witherington offered much appreciated advice by sharing their knowledge of sea turtles. My field work in Costa Rica greatly depended on the aid of Elver and Elvin Gutierrez, Robert Carlson, Harumi Horikoshi, and Carlos Diez. Stephen Morreale kindly recorded some weather data in 1986.
I would like to thank the Caribbean Conservation Corporation, Sigma Xi, and the Archie Carr Center for Sea Turtle Research, the Center for Latin American Studies, and the Department of Zoology, University of Florida for partial funding for this project. I would
ii




also like to thank the Costa Rican National Park System for allowing me to work in Tortuguero National Park.
III




TABLE OF CONTENTS
ACKNOW LEDGEMENTS ............................................................................................... i i
ABSTRACT .................................................................................................................... v i
CHAPTERS
1 INTRODUCTION .................................................................................................. 1
2 METHODS ............................................................................................................. 8
Study Area ................................................................................................... 8
W eather Record ........................................................................................ 9
Beach Zones ................................................................................................ 10
Nesting Census .......................................................................................... 10
Sam pling of Egg Survivorship Nests ............................................... 12
Sampling of Nests to Determine Sex Ratio ................................. 14
Histology and Sexing .............................................................................. 16
Sand Temperatures ................................................................................. 16
Statistical Analysis ............................................................................... 17
3 RESULTS ............................................................................................................ 18
Physical Environment Surrounding Nests ..................................... 18
Air Temperatures .............................................................................. 18
R a in fa ll .................................................................................................. 1 9
Vertical Position of Clutches ...................................................... 2 1
Ground W ater ....................................................................................... 22
W aves ...................................................................................................... 23
Daily Fluctuation of Sand and Nest Temperatures ............... 24
Seasonal Fluctuation of Sand Tem peratures ......................... 26
Seasonal Nest Distribution and Nest-Site Selection .............. 28
Analysis of Egg Survivorship Samples .......................................... 30
Fates of Sample Nests .................................................................... 30
Abiotic and Biotic Factors Affecting Egg Survivorship .... 33
i v




Fate of eggs from sample nests ....................................... 33
Criteria for assessing the cause of egg mortality ... 33
Abiotic factors .......................................................................... 34
Biotic factors ............................................................................. 36
Other Categories of Mortality ...................................................... 42
Analysis of Sex Ratio ............................................................................. 43
Sexed Samples ..................................................................................... 43
Sex Ratio vs Temperature .............................................................. 44
Seasonal Variation and Zone Effect in 1988 .......................... 46
Overall Sex Ratio in 1988 .............................................................. 47
Seasonal Variation and Zone Effect in 1986 .......................... 48
Overall Sex Ratio in 1986 .............................................................. 49
Prediction of Sex Ratio from Independent Variables .............. 49
Logistic Regression Model ............................................................. 50
Logistic Regression with a Single Variable .......................... 50
Nest and sand temperatures ................................................ 50
Incubation period ...................................................................... 51
R a infa ll ......................................................................................... 5 1
Bottom depth of clutch .......................................................... 52
Multiple Logistic Regression Model ........................................... 53
4 DISCUSSION .................................................................................................... 55
Nesting Density and Tem poral Nest Distribution ...................... 55
Seasonality of Environm ental Parameters ................................... 57
Spatial Distribution between the Zones ........................................ 58
Overall Reproduction Rate ................................................................... 61
Spatial Effects on Mortality Factors .............................................. 63
Seasonal Fluctuation of Em ergence Success ............................... 65
Spatial Effects on Sex Ratio ............................................................... 67
Tem poral Effects on Sex Ratio .......................................................... 69
Annual Variation in Prim ary Sex Ratio .......................................... 70
Predictability of Sex Ratio by Environmental Factors ........... 71
5 SUMMARY AND CONCLUSIONS ..................................................................... 78
LITERATURE CITED ................................................................................................ 149
BIOGRAPHICAL SKETCH ...................................................................................... 158
v




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
EGG SURVIVORSHIP AND PRIMARY SEX RATIO OF GREEN TURTLES,
CHELONIA MYDAS, AT TORTUGUERO, COSTA RICA
By
KAZUO HORIKOSHI
AUGUST, 1992
Chair: Martha L. Crump
Cochair: Karen A. Bjorndal
Major Department: Zoology
Tortuguero beach in Costa Rica is one of the last major nesting sites in the western Atlantic for green turtles. During 1986, 1988 and 1989, I studied temporal and spatial nest distribution, egg survivorship, and sex ratio of hatchlings in two thermal zones: the vegetation/border zone (within 2 m of the border of dense vegetation) and the open zone (between the vegetation/border zone and the sea). Although the distribution of nests between the two zones differed significantly from year to year, it was biased toward the vegetated area of the beach throughout each season.
vi




I analyzed the emergence success of 49 clutches in 1986, 88 in 1988, and 113 in 1989. Emergence success from these nests showed a bimodal distribution (0% and >70%) and did not differ significantly between the zones. Abiotic factors including surf, freshwater inundation, and hurricane killed approximately 20% of sample eggs. Biotic factors including predation and turtle digging killed 5% of sample eggs in the open zone and 19% of sample eggs in the vegetation/border zone. The relative importance of each factor varied from year to year. Mammals destroyed four times as many nests in the vegetation/border zone as in the open zone. My estimation of hatchling emergence success was 57.2% (1986), 45.7% (1988) and 66.6% (1989).
A sample of hatchlings was collected for direct sexing from each of 12 nests in 1986 and 56 in 1988. The sex ratios of hatchlings fluctuated intra-seasonally in 1988, but not in 1986. The estimated overall proportion of female hatchlings was 10.1% in 1986 and 40.6% in 1988. 1 used single and multiple logistic regression models to test environmental and clutch variables for the prediction of sex ratio. The pivotal temperatures were 29.400 for mean nest temperature, and 28.500 for mean sand temperature. Mean nest temperature during the middle third of development was the best fit model (pseudo R2=0.20) in the single regression analysis. Mean nest temperature, incubation period, rainfall, and nesting zone were significant parameters for multiple regression (pseudo R2=0.24). However, even the model with the best fit was still not reliable for predictive purposes.
vii




CHAPTER 1
INTRODUCTION
Recently it has become apparent that there are two types of sex determination mechanisms in reptiles: genotypic and environmental. In the latter case, sex of the offspring is decided after fertilization by environmental factors. The common form of environmental sex determination in turtles is temperaturedependent sex determination (TSD), in which incubation temperature during a critical portion of embryonic development controls gonadal differentiation. The physiological mechanism involved in this phenomenon is not understood (reviewed in Bull, 1983: Ewert and Nelson, 1991; Janzen and Paukstis, 1991).
At present, seven of eight species of sea turtles are known to show the TSD phenomenon. These species are: Caretta caretta (Yntema and Mrosovsky 1980), Chelonia mydas (Miller and Limpus 1981), Chelonia agassizi (Alvarado and Figueroa, 1989), Dermochelys coriacea (Mrosovsky et al., 1984), Eretmochelys imbricata (Dalrymple et al., 1985), Lepidochelys olivace (McCoy et al., 1983), and Lepidochelys kempi (Shaver et al., 1988).
There are three patterns of TSD in reptiles (Bull, 1983). Type A pattern results in females at lower temperatures, males at higher temperatures and both sexes at intermediate temperatures (most crocodilians and lizards); Type B pattern results in females at
1




2
warmer temperatures, males at lower temperatures and both sexes at intermediate temperatures (most turtles); Type C pattern results in females at warmer and lower temperatures and males at intermediate temperatures (three crocodile species, one lizard species and three freshwater turtle species) (see list of species in Janzen and Paukstis, 1991). All seven species of sea turtles with TSD show the Type B pattern.
At present, no species with heteromorphic sex chromosomes is known to show TSD. Karyotypes of five species of sea turtles have been examined microscopically, but no discernible heteromorphic sex chromosomes have been confirmed (Caretta caretta, Bickham, 1981; Chelonia mydas, Bickham et al., 1980; Dermochelys coriacea, Medrano et al., 1987; Eretmochelys imbricata, Kamezaki, 1990; Lepidochelys olivacea, Bhunya and Mohanty-Hejmadi, 1986). However, there is evidence that the two sexes of sea turtles are genetically different at the molecular level. In adults of Caretta caretta and Chelonia mydas, males had a higher level of H-Y antigen in their blood cells than did females (Wellins, 1987).
A series of experiments on a European freshwater turtle, Emys orbicularis (Zaborski et al., 1982, 1988), which shows Type B TSD, indicated that genetic and environmental factors can operate simultaneously. It appears that incubation temperatures at both extremes can overide genotypic influence, whereas at intermediate temperatures the genetic factor can influence the sex of turtles to some extent. There was a strong correlation between sexual phenotype of gonads and H-Y antigen phenotype of blood cells (male negative / female positive) from eggs incubated at the




3
intermediate temperature. Girondot and Pieau (1990) proposed a similar mechanism to explain the variation of the pivotal temperatures observed in loggerhead turtles (Limpus et al., 1985; Mrosovsky, 1988). Further investigation of the effect of genetic factors over TSD in species of sea turtles has not been conducted.
Tortuguero beach on the Caribbean coast of Costa Rica is one of the last major nesting sites in the western Atlantic for green turtles, Chelonia mydas. Spotila et al. (1987) found that the temperatures of the nest during the middle third of development influenced the sex ratio of Tortuguero green turtles in natural nests. Mean temperatures less than 28.50C produced mainly males, mean temperatures greater than 30.30C produced only females, and mean temperaturess between 28.5 and 30.30C produced both sexes. They also found that the different thermal zones at the beach yielded different sex ratios; nests under vegetation produced a high percentage of male hatchlings, whereas nests in the open beach produced mainly females. No seasonal trend of sand temperature was apparent during their monitoring period (24 July 22 September 1980). Assuming no seasonal fluctuation of sex ratio, and using 1977 data for the distribution of nests between the shady zone and the open sand zone at Tortuguero (Fowler, 1979), they estimated that 67% of the hatchlings were females in the 1977 season.
However, their study did not cover the full seasonal profile at Tortuguero, and the sample size was very limited (15 nests). The main incubation season of the green turtle colony at Tortuguero extends from July through November (Fowler, 1979). The rainfall at Tortuguero is extremely variable from year to year (Myers, 1981).




4
During the middle of the season, a short dry period occurs (usually in September). Carr (1979) described the June August period, the early incubation season, as "monsoonlike, with many consecutive days of heavy overcast and rain". In general, rainfall and sand temperature are inversely related on the beach. Therefore, it is very probable that seasonal fluctuation of sand temperature, and consequently seasonal fluctuation of sex ratios, can occur regularly at Tortuguero. Therefore, in addition to evaluating whether the 1980 season is typical or atypical, it is essential to examine the entire season over several years to document seasonal and yearly variation for a better estimation of the primary sex ratio at Tortuguero.
All species of sea turtles are considered to be threatened or endangered because of over-exploitation and destruction of their habitats. The green turtle, a circumtropical large species, is a valuable food source for many coastal people, and thus has been harvested for a long time. Current management practices for sea turtles focus on conservation of each population. Among the conservation practices, protection of incubating eggs in several forms of artificial hatcheries is presently the most common measure. However, little regard has been given to the incubation temperature until recently. Incubation of sea turtle eggs in styrofoam boxes subjected the eggs to a different thermal environment from that experienced on a natural beach and showed a masculinizing effect on the embryos (Mrosovsky, 1982; Dutton et al., 1985). Because of this incubation method, the early years of the Kemp's ridley head starting project unintentionally produced male




5
dominated turtles much of the time (Shaver et al., 1988). On the other hand, the hatchery on a Sarawak beach produced highly female biased green turtle hatchlings for many years (Leh et al., 1985). Location and type of artificial hatchery can easily alter the sex ratio of hatchlings produced. Currently there is no consensus on the best sex ratio to produce in hatcheries because the sex ratios of sea turtle populations and their dynamics are uncertain.
Therefore, to understand better the ecological and
conservation implications of TSD among sea turtles, it is critical to know the process of sex determination and the primary sex ratio of hatchlings on a natural beach. However, natural sex ratios of hatchling sea turtles have been studied in only a few populations, and estimation of primary sex ratio of a population has not been feasible, partially because of small sample sizes and limited seasonal coverage (Limpus et al., 1983; Maxwell et al., 1988; Rimblot-Baly et al., 1987; Spotila et al., 1987). Quantitative investigations that cover the full seasonal profile exist only for the Suriname green turtle and leatherback turtle colonies (Mrosovsky et al., 1984) and a Florida loggerhead turtle colony (Mrosovsky and Provancha, 1989; Provancha and Mrosovsky, 1992). However, these studies do not include information on egg survivorship on the beach. To estimate reliably the primary sex ratio, egg survivorship is an essential factor because survivorship might differ in different thermal zones on the beach or within seasons.
Fowler (1979) found that the nest position affected survival rate of green turtle nests at Tortuguero because of differential mammal predation and inundation rates. Nests near or in the beach




6
border vegetation suffered greater predation than did nests on the lower part of the beach. However, because introduced dogs destroyed about one fourth of the nests in the 1977 season, Fowler's results do not represent natural survivorship of the green turtle nests at Tortuguero. Recently, due to success of a dog control program by Tortuguero park guards, the beach has almost returned to the natural condition so that the coati, Nasua narica, has become the main predator. No quantitative study of egg survivorship in the green turtle colony has been conducted since the dogs have been controlled. There is a large yearly fluctuation in the number of nesting green turtle females at Tortuguero; 1977 was a year with relatively few nesting females (Bjorndal et al., 1985). Nest density might affect selection of nest sites on the beach by turtles and also influence spatial and temporal predation patterns. Therefore, a prolonged study to cover several years is needed to understand fully the reproduction of Tortuguero green turtles and to document the biotic and abiotic factors contributing to their natural mortality.
A combination of (1) distribution of nests among thermal zones; (2) egg survivorship; and (3) seasonal trend of sand temperature and its consequence the seasonal trend of sex ratio in the different thermal zones, are recognized as major factors in determining primary sex ratios.
The primary objective of this study was to estimate the
primary sex ratios of the green turtle population at the Tortuguero beach and to obtain information concerning egg survivorship and sex ratio of the nests throughout the entire season for three years. An additional objective was to accumulate information on the physical




7
characteristics of natural nests on the beach, on which the TSD phenomenon operates, and to assess the predictability of sex ratio from environmental factors.
This parameter, primary sex ratio, is not only necessary for improving conservation practices, but it also will increase our understanding of the dynamics of the most important population of green turtles in the western Atlantic Ocean. Furthermore, I believe that information on primary sex ratio will also give us a key to assess adult sex ratio of all sea turtles.




CHAPTER 2
METHODS
Study Area
My study area was located at Tortuguero, Costa Rica. The
Tortuguero beach extends 22 miles (35.4 km) from the mouth of the Rio Tortuguero south to the Rio Parismina on the Caribbean coast of Costa Rica (Figure 1). The beach is located on a long, narrow island that is separated from the mainland by a freshwater lagoon and estuaries of the rivers.
Tortuguero beach is the nesting beach for the largest surviving nesting population of green turtles in the Atlantic Ocean (Groombridge and Luxmoore, 1989). Leatherback turtles, Dermochelys coriacea, and hawksbill turtles, Eretmochelys imbricata, also nest on the same beach, but are much less abundant than green turtles.
The shoreline is composed of a continuous series of spits and guts. The shape of the shoreline shifts considerably with surf erosion and rebuilding, sometimes in a short period of time. High waves occur almost throughout the nesting and incubation season of green turtles (June to December). The beach is classified as a high energy beach because of constant high wave activity.
The beach was divided into 22 one-mile sections from the
northern end to the southern end. The central area of the beach (mile
8




9
3 to mile 18), has been protected as a part of Parque Nacional Tortuguero since 1975.
To investigate egg survivorship and primary sex ratio of the Tortuguero population of green turtles, I selected two miles of beach (mile 6 to mile 8) for my study area. This area is one of the most highly utilized areas by nesting green turtles. Approximately 17% of nests laid on the beach are deposited in this two-mile section (Carr et al., 1978). The beach shape, vegetation, and animals in the central area remain undisturbed because the area is successfully protected as part of the national park. There is no human habitation in the park area.
Railroad vine, lpomoea pes-caprae, and beach lily,
Hymenocallis littoralis, predominate on the rear of the open beach. Behind the beach, cocoplum, Chrysobalanus icaco, seagrape bushes, Cocoloba uvifera, and coconut palms, Cocos nucifera, are common. Inland, there is a well developed and largely undisturbed tropical wet forest. The dense canopy of the inland forest is high (> 8 m).
Weather Record and Ground Water
Ambient temperature and rainfall data were collected with
standard meteorological instruments. In 1986 and 1989, data were collected at the Green Turtle Research Station adjacent to the beach, located about 9 km north of the study area. During the 1988 season, data were taken at the study area from July through the middle of October, and later at the station. Rainfall and maximum




10
and minimum air temperatures were taken daily around 0900. Also at this time, height of shore waves was estimated by eye.
Seasonal fluctuations of ground water level on the beach were monitored during the incubation season from July through 10 December in 1986 and 1989. During the 1988 season, the wells were stolen several times during the study, therefore only very scattered data were taken. Plastic PVC pipes (diameter 5 cm) were placed to a depth of 140 cm in 1986 and 150 cm in 1989. Two wells were set in places of 50% shade for the border/vegetation zone and three wells in places of 0% shade for the open zone for each year. This measurement revealed the level of ground water only when it was above the depth of the well.
Beach Zones
The beach was divided into two zones (Figure 2). The
vegetation/border zone lies within 2 m of the border of dense vegetation; there is 5-100% cover, usually of sea grapes and coconut palms. The open sand zone lies below the vegetation/border zone, with <5% cover, usually of sparse herbaceous vines.
.Nestina Census
To obtain information concerning the seasonal distribution of nests within the two zones, I conducted a census within the study area throughout most of the nesting season during 1986, 1988 and 1989. 1 recorded the number and zone of all nests deposited from




11
the previous night. Crawl tracks and nest marks were used to locate nests and to identify species (Pritchard et al., 1983).
In 1986 the census was carried out from 2 July through 29
November at an interval of once every two to four days. Two surveys on 10 and 14 July were not completed because high waves and strong wind erased most of the tracks before the surveys began. Therefore, these surveys were eliminated from analysis. In 1988, the census was carried out from 16 June through 30 November. The interval between surveys was two to five days; most were two to three days. No survey was conducted during the evacuation for Hurricane Joan between 19 and 23 October. In 1989, the census was conducted from 17 June through 30 November. The interval between surveys was two to three days; most were two days.
The census data were pooled into half month periods (1st to 15th, and 16th to 30th or 31st). The daily mean number of nests deposited per period was calculated by dividing the total number of nests counted by the number of censuses during each period. The total number of nests deposited during each period was calculated by multiplying the daily mean number of nests by the number of days during the period. Total number of nests over the season was the sum of the numbers of nests during each period.
The proportion of nests deposited in the vegetation/border zone and in the open sand zone was calculated for each half month period. The proportion of nests in the two zones for the entire season was calculated from pooled numbers of nests counted during the censuses.




12
Sampling of Egg Survivorship Nests
A representative sample of nests deposited within the study area was marked and followed throughout the incubation period. As much as possible, I attempted to sample equal numbers of nests in the two zones and for each half month period of time to analyze seasonal and spatial variances of egg survivorship.
Sampling at night was conducted seven to 10 times during each half month period to locate nesting females. During 1986 and 1988, only females found during the early stage of nesting behavior (before digging the egg chamber) were sampled; for these females I counted the number of eggs during deposition. Because fewer females nested during 1989 than during 1986 and 1988, it was necessary to include nesting females that were already in the process of depositing eggs; thus during 1989 the number of eggs was not counted. Sample nests were marked with a numbered stake placed 1 m from the egg chamber. I also placed two pieces of vinyl tape in the vegetation to triangulate the location. A piece of numbered flagging tape was placed in the egg chamber to confirm the location in the event of nest destruction.
Due to the difficulty of finding nesting females in a specific zone during a specific period, the numbers of marked nests in each block are not equal. I omitted from analysis several marked nests (1 in 1986, 2 in 1988 and 1 in 1989) that were located so close to other egg chambers that I was unable to determine hatching success for the individual nests because the eggshells were inseparable.




13
Furthermore, in spite of intensive marking, some marked nests (2 in 1986, 3 in 1988 and 7 in 1989) were lost. Some of these lost nests may have been completely excavated by other nesting females.
In 1986, 49 marked nests were successfully sampled from the latter half of July through the early half of September. Although the actual sampling period was longer, only these periods contained adequate numbers of sample nests in both zones for analysis. In 1988, 88 marked nests were successfully sampled from July through September. In 1989, 113 marked nests from the latter half of July through the early half of October were sampled. The difference in sampling periods was due to the adjustment for the shift of temporal nest distribution between the two years.
Marked nests were examined during the beach census for signs of emergence, depredation, inundation, and any disturbance. When the nests were predated, the species of predator was identified by tracks. Emergence success reported in this study was the fraction of eggs that resulted in hatchlings that emerged from the sand.
After emergence of hatchlings, the marked nest was excavated. The numbers of hatched and unhatched eggs, and the numbers of dead or live hatchlings remaining underground were determined. Unhatched eggs were opened to check for development; if development had occurred, size of the embryos was recorded. Unhatched eggs lacking visible embryos or blood formation were classified as infertile eggs. Evidence of underground disturbance such as by ghost crabs and flooding, and the number of eggs affected, were determined.




14
During 1986 and 1988, clutch size was counted during
deposition, and emergence success was determined directly. In 1989, when the clutch size was not counted, clutch size was estimated by the number of remaining eggshells. In most undisturbed nests, the hatched shells remained in one piece. If the shells were fragmented, the pieces were put together to represent one shell. Fowler (1979) used this method to estimate the clutch size of green turtles in Tortuguero and found that the error was no more than 8 eggs. In the 1989 season, for nests that were partially destroyed or completely lost by any disturbance, the mean clutch size from 1989 (107.1) was used as the clutch size to calculate emergence success.
Sampling of Nests to Determine Sex Ratio
I originally planned to obtain five nests to determine sex ratio in each zone and for each half month period in the nesting season. Night sampling was conducted seven to 10 times during each half month period to locate nesting females. Only females during the early stage of nesting behavior (before digging the egg chamber) were sampled. During the deposition of eggs, a thermocouple probe was placed at the approximate center of the nest. For comparison, and to allow measurement of metabolic heat from the eggs, another thermocouple probe was placed at the same depth 1 m along the beach from the egg chamber.
To protect the temperature-monitored nests from activities of other nesting females, several logs found on the beach were




15
placed around the nest; these logs were placed in such a way as to avoid creating shade on the sand surface above the egg chamber. The temperatures of nests and the sites 1 m away from the nests were monitored once a day at an interval of two to four days in 1986 and at an interval of two days in 1988 and 1989.
During the 1986 and 1988 seasons, I collected samples of eggs to be sexed; during 1989, 1 monitored temperatures of the nests, but did not collect any eggs. After 50 60 days of incubation, I uncovered the nests and randomly selected 20 developing eggs. The distance from the bottom of the egg chamber to the sand surface was measured. The numbers of developing eggs and dead eggs were counted, and dead eggs were opened to check for any sign of development. Remaining developing eggs were reburied in the egg chamber, and eventually the day of emergence was recorded.
In 1986, 24 nests were equipped with thermocouples, but
subsamples of embryos to be sexed were collected from only 12 of these nests. In 1988, 92 nests were equipped with thermocouples, and sub-samples were collected from 42 of these nests. Loss of the sample nests was due to destruction by nesting females, drowning of whole clutches by flooding or hurricane, and depredation by animals. Nest temperature data were obtained from all the nests intended for future sexing of the hatchlings, except one nest in 1986. In 1989, 74 nests were set up only for temperature monitoring; nest temperature data were collected from 69 of these nests. Sand temperature data near each sample nest were unobtainable for some of the nest temperature monitored nests in all years: 3 nests in 1986, 8 nests in 1988, and 13 nests in 1989.




16
Histology and Sexing
Sample eggs were kept in plastic bags with holes until pipping. Hatchlings were killed, and the gonads were fixed in 10% neutral buffered formalin. The gonads were transferred to the University of Florida for histological analysis. Collection permission was issued by Servicio de Parques Nationales, Costa Rica through Mr. Fernando Cortes; the CITES permit number is 092-88 in Costa Rica, PRT725607 in United States. Transverse serial paraffin sections, 5-8 I.m thick, were prepared from the median portion of each left gonad. Harris' haematoxylin and eosin were used for staining. Criteria for sexing green turtle hatchlings were the same as those reported by Spotila et al. (1983).
Sand Temperatures
To investigate the seasonal thermal profile of sand
temperatures at the same depths as egg chambers of green turtles on the beach, temperatures at depths of 60 cm (all three years) and 80 cm (1988, 1989) in each zone were monitored along two (1986) or four (1988 and 1989) transect lines from July through early December in all three years. These transect lines included both relatively wide and narrow portions of the beach. I placed a set of thermal probes at the border (50% shade) in the vegetation/border zone and at two to three points (0% shade) in the open sand zone: 5 m from the border point and others at 5 m intervals along the transect.




17
The temperature readings were taken once every two to three days, with a Bairly BAT12 thermocouple meter. To determine the amount of diurnal fluctuation, several monitoring sessions lasting 24 h each, at 2 h intervals, were conducted during the study: 13 August and 29 September 1988, and 12 August and 14 October 1989.
Statistical Analysis
Both parametric and nonparametric statistical tests were used for data analysis. The arcsine transformation was applied to the percentage data (e.g., emergence success and sex ratios). Whenever the required assumption of homogeneity of variance was met by using the Fmax test, parametric statistical tests were utilized. If the assumption was not met, nonparametric tests were substituted. To test binomial data (e.g., nest site selection between the two zones), Chi-Square tests were utilized. The rejection level for the null hypothesis in all tests was alpha = 0.05. For those nests where complete data sets were not available, the existing data were included in analyses whenever possible. Each statistical test is mentioned in the results section.




CHAPTER 3
RESULTS
Physical Environment Surrounding Nests
Air Temperatures
Monthly minimum and maximum air temperatures from July
through November in the 1986, 1988 and 1989 seasons are shown in Figure 3. Whereas minimum temperat tIres varied little throughout the season (between 23.3 and 24.1 OC), maximum temperatures showed slight increases in August and September for the three years. The ranges of monthly maximum temperatures for each year were 28.400 (July) 29.90C (September) in 1986, 29.200 (July and November) 30.50C (September) in 1988 and 28.700 (November) 30.800 (August) in 1989.
The highest daily air temperatures occurred during September of each year: 35.000 on 1 September 1986, 34.000 on 23 September 1988 and 33.000 on 2 and 3 September 1989. The lowest daily air temperatures were recorded as 22.100 on 10 July 1986, 20.500 on 7 July 1988 and 22.500 on seven occasions throughout the 1989 season, except in September.
Yearly variation in mean temperature throughout each season among three years was very low. The mean minimum and maximum temperatures during each five-month period were 23.9 and 29.300 in
18




19
1986, 23.4 and 29.70C in 1988 and 23.7 and 29.80C in 1989, respectively. Overall mean three-year minimum and maximum temperatures during the same period were 23.7 and 29.80C, respectively.
Rainfall
Daily rainfall records from 16 June through 10 December
during 1986, 1988 and 1989 are shown in Figure 4. The sampling period covered the entire incubation season of green turtle eggs in Tortuguero. For all three years, heavy rain events sporadically occurred. Rainfall greater than 100 mm during 24 hours was recorded on seven occasions in 1986, five occasions in 1988 and five occasions in 1989. The heaviest 24-hour precipitation each year was 186 mm on 21 October in 1986, 221 mm on 6 October 1988 and 155 mm on 25 October 1989. Among those intense rainfall events, at least two events in 1986 (5 August, 141 mm/day; 6 December, 178 mm/day), one event in 1988 (6 October, 221 mm/day) and one consecutive three-day event in 1989 (30 October-1 November, total 367 mm for three days) caused freshwater flooding on the beach and consequently damaged some green turtle nests (see section on ground water).
The 1986 season overall was the wettest among the three years. The total rainfall during the nearly six month period (16 June-10 December, 178 days) was 3871 mm in 1986, 2578 mm in 1988 and 2896 mm in 1989. Mean rainfall taken at the Tortuguero park station during the same period from 1978 through 1989 (I have




20
omitted 1984 data because of an incomplete data set, Instituto Meteorogico Nacional) was 2882 mm (SE=622, n=11). The rainfall (3871 mm) in the 1986 season was the second highest on record; the greatest rainfall (3929 mm) was recorded in 1982.
Figure 5 shows seasonal variation of monthly rainfall taken at the Tortuguero National Park Station from 1978 through 1989 (Instituto Meteorogico Nacional, Costa Rica). The major incubation season of green turtles (July through November) includes several of the wettest months (July and November) during the year. September, however, which was the middle of the incubation period, had the least rain. Within the five-month comparison (July through November), the monthly rainfall during September was significantly less than that during July and November, but not significantly less than in August and October (one-way ANOVA, df=4, p=0.064; Fisher's procedure of least significant difference, alpha=0.05). These data indicate that a drier period normally occurs during the middle of the incubation season of green turtles at Tortuguero.
The monthly rainfall for the three study years and the mean rainfall from 1978 through 1989 are shown in Figure 6. During the 1986 season, the monthly rainfall during the early to middle incubation season well exceeded the mean values. The rainfall in August (846 mm) and in September (514 mm) during 1986 were the highest monthly records. As a result, the 1986 season did not experience a dry period in the middle of the season, and the monthly rainfall exceeded 500 mm throughout the entire season. On the other hand, the 1988 and 1989 seasons were much drier than the 1986 season during the middle of the incubation season. The rainfall in




21
August 1988 (256 mm) and in September 1989 (157 mm) were the lowest ones since 1978. The rainfall in October of both years exceeded the average values. Consequently the 1988 and 1989 seasons showed distinctive seasonal fluctuation in rainfall.
Vertical Position of Clutches
A total of 344 sample nests (including both egg survivorship samples and sexing/temperature monitoring samples) during the three seasons was measured for the distance between the bottom of the egg chamber and the level of the sand surface on the beach at the time of excavation. Nest mounds above the clutches made by females were generally flattened to the level of the beach surface by the time of excavation. In addition, 43 sexing/temperature monitoring sample nests in the 1988 season were measured for the height of the clutch mass (a distance between the bottom and the top of egg mass) when 20 sub-sample eggs were taken for sexing at around 50-60 days of incubation. Bottom depth of green turtle clutches during the three years averaged 76.8 cm (SE=0.8, range=52105, n=176) in the open zone and 79.0 cm (SE=0.8, range=45-120, n=168) in the vegetation/border zone. The difference between the two zones was almost significant (t=1.93, p=0.0549). Overall bottom depths of green turtle clutches (two zones combined) averaged 77.9 cm (SE=0.6, range=45-120, n=344; Figure 7). Mean height of the clutch mass was 19.8 cm (SE=0.9, range=8-33, n=43). If the center of the clutch is assumed to be located at half of the height of the clutch (about 8.9 cm from the bottom), mean vertical




22
position of the center of the clutch is 69.0 cm below the sand surface. The top of the clutch is 58.1 cm below the sand surface.
Ground Wate
Figures 8 and 9 show seasonal fluctuations of the ground
water table and rainfall during the 1986 and 1989 seasons. During the 1988 season, the wells were stolen several times during the study, therefore only very scattered data were available. The level of the water table on the beach was affected by rainfall. Continuous rainfall and occasional heavy rainfall were associated with a rise in the ground water table. During the 1986 season, which was very wet throughout, the level of the water table frequently rose close to the general depth of the clutches in both zones (overall mean bottom depth of clutches was 77.9 cm; Figure 7). On the other hand, the level of ground water during the 1989 season rose close to the depth of clutches only early and late during the incubation season when rainfall was heavy.
On several occasions, excessive rainfall caused freshwater flooding on the beach. Flooding events were confirmed by water marks that remained at the margin of a depression area on the beach that recorded the highest level of water pools during the excessive rain events. However, only one event (5 August 1986; 141 mm/day) was actually recorded on the well as a sharp increase of the water table much above the general depth of clutches on the beach. Because the measurements of the water table at the time of other excessive rain events were taken 6 to 9 hours after the rain had




23
stopped, the levels of the water table recorded were probably much lower than the actual maximum level. The water marks remaining in the depression area suggested that the water tables on 6 December 1986 (178 mm/day), and on 1 November 1989 (total 367 mm for three days) also rose above the general depth of the clutches. During the 1988 season, freshwater flooding was observed on one occasion, associated with a heavy rain event on 6 October (221 mm/day).
In addition to rainfall, high waves probably raised the water table. While the amount of rainfall was relatively small (28.5 to 37.0 mm) on 19 and 20 November 1989, the water table showed a substantial high level on 20 November. At the same time, very high waves (>2 m) were recorded. This height of waves rarely occurred in Tortuguero during the study. Except on this one occasion, waves of this height only occurred during the time that Hurricane Joan passed near Tortuguero on 21-22 October 1988. Data on the water table during the hurricane are not available.
Waves
Figure 10 shows frequency distributions of relative heights of waves throughout each incubation season in 1986, 1988 and 1989. On the Tortuguero beach, heights of waves were primarily associated with the size of swells from offshore rather than with the wind speed over the beach. Wave actions were generally high in Tortuguero during most of each season. Less than one third of the days were identified as calm throughout each season (wave height
0.5 m; 17.7% in 1986, 31.4% in 1988 and 31.5% in 1989). Wave




24
activity varied considerably with a similar seasonal trend throughout each season. Generally, there were more calm days during the middle of the season. For the three years combined, September was the month most likely to have a calm day. The amount of monthly rainfall was negatively related to the number of calm days (wave height 0.5 m) in the month (Y=17.147-1.656X, R=0.663, n=15; t=3.19, p=0.007; Figure 11). The greater the rainfall, the higher were the waves on the beach.
Continuous days of calm waves resulted in sand accretion on the lower part of the beach, and the entire beach widened gradually. On the other hand, when the wave activity was higher (> 0.5 in), beach erosion was apparent and the shape of the beach changed rapidly. Heavy beach erosion frequently constructed beach platforms (up to 1.5 m height) along part of the beach. Green turtles either gave up trying to nest in these areas, or they crawled over the platforms and laid their clutches in a site protected from waves. Very few females nested below the platforms at the time of high waves. However, nests that had been deposited before the high platforms were formed or advanced well inland were washed away or completely inundated.
Daily Fluctuations of Sand and Nest Temperatures
To determine the extent of daily fluctuation of sand and nest temperatures, temperatures along two transects (two points in the vegetation/border zone, five points in the open zone) and temperatures at the center of clutches (14 nests in each zone) were




25
monitored at two-hour intervals for 24 hours for four days during the study. Those clutches monitored averaged 23.0 days from deposition (SE=2.2, range=3-54, n=28) at the time of temperature measurement. Two of the days (13 August and 29 September 1988) had no rainfall and two days (12 August and 14 October 1989) had minor rainfall (2 mm and 25 mm, respectively) during the 24-hour monitoring period. In general, these four days are within the range of normal weather in Tortuguero.
The sand temperatures at depths of 60 cm and 80 cm were relatively stable in both zones (Table 1). Mean ranges of daily fluctuation in these four categories were from 0.410C to 0.490C. The range of individual monitoring points on the transects was from
0.10C to 0.80C. The daily fluctuations in temperatures were not significantly different between the two zones, or between depths of 60 cm and 80 cm (two-factor ANOVA; zone, p=0.3515; depth, p=0.4716).
The daily fluctuation of temperatures in the center of the clutches showed low ranges similar to the fluctuation in sand temperatures along the transects (Table 1). The range of daily fluctuation averaged 0.460C in the open zone and 0.490C in the vegetation/border zone with the same individual range (0.2-0.80C). The mean daily fluctuations of nest temperatures were not significantly different between the two zones (unpaired t-test, t=0.377, p=0.710).
Although daily temperature profiles varied with different weather, mean temperature profile among the four separate days showed a general trend of daily fluctuation (Figure 12). Sand




26
temperatures at depths of 60 cm and 80 cm and nest temperatures in both zones showed a similar trend of daily profiles. While the period of highest temperatures was not distinct, the lowest temperatures were recorded from 1000 to 1200 in the morning. Times when the temperature was closest to the 24-hour mean were around 0600 and 1500. Most of my temperature readings were conducted during the two periods between 0700 to 1000 and between 1500 and 1700. Overall, the temperatures collected for this study were probably not far from the values of 24 hour means.
Seasonal Fluctuation of Sand Temperatures
In Tortuguero, local meteorological conditions affected the fluctuations of sand temperatures. Rainy days resulted in lower sand temperature, while sunny days resulted in higher temperatures. In general, amount of rainfall was inversely related with sand temperatures. Being associated with variable rainfall (Figure 4), the sand temperatures showed considerable seasonal fluctuations for the three years (Figure 13).
Occasional heavy rain events, particularly those that caused freshwater flooding on the beach (Figure 4), resulted in rapid and substantial short-term declines of sand temperatures along the entire beach (Figure 13). In addition to such flooding rain events, overcast days with moderate rainfall sometimes resulted in a substantial decline of sand temperatures. A continuous rainy event from 21 through 25 July 1988 accumulated 165 mm of rainfall, and lowered the sand temperature up to 2.90C at 60 cm depth in the open




27
zone. During the entire five days, the sky was overcast during the daytime. Prolonged rainy days kept sand temperatures low for the duration and caused long-term fluctuation. The continuous low temperature during July through August in 1986 was one such situation.
Table 2 shows overall mean sand temperatures throughout each season at depths of 60 cm and 80 cm (only 1988, 1989) in the two zones. For all three seasons, the trends of spatial thermal profiles were identical. Sand temperatures at depths of 60 cm and 80 cm in the open zone were significantly higher than those in the vegetation/border zone. Within the same zone, sand temperatures at a depth of 60 cm were significantly higher than those at a depth of 80 cm (one-way ANOVA with repeated measure; 1986, 1988, 1989, all tests p < 0.0001; one-way ANOVA of repeated measure with Bonferroni adjustment of P for multiple comparison, overall alpha
0.05, 1988, 1989, all tests p<0.008). For overall mean sand temperatures throughout the season, the differences between the two zones ranged from 0.80C (1986) to 1.50C (1989) at a depth of 60 cm and from 1.10C (1988) to 1.30C (1989) at a depth of 80 cm. The differences between the depths of 60 cm and 80 cm ranged below
0.40C in both zones in 1988 and 1989. For overall mean sand temperatures throughout the season, the 1988 season showed the highest, the 1989 season the intermediate, and the 1986 season the lowest sand temperatures at a depth of 60 cm in both zones during the study.
Figure 14 shows seasonal fluctuations of mean monthly sand temperatures at a depth of 60 cm for the three years in both zones.




28
Mean monthly sand temperatures from July through November differed among years in both zones (one-way ANOVA for each depth and zone; same results for all six tests, df=4, p<0.0001). Fisher's procedure of least significant difference identified those months where mean temperatures are significantly different from each other (Table 3). Although the seasonal trends of monthly sand temperatures varied from year to year, September's sand temperatures in both zones resulted in the highest rank within each season for all three years. Figure 15 shows the negative relationship between the amount of monthly rainfall and mean monthly sand temperatures at a depth of 60 cm on transects along the beach during the study (the open zone, Y=30.020-0.003X, R=0.718, n=15, t=3.718, p=0.003; the vegetation/border zone, Y=28.524-0.002X, R=0.665, n=1 5, t=3.21 0 p=0.007).
Seasonal Nest Distribution and Nest-Site Selection
By the first censuses, minor nesting activity already had
occurred in all three years. At least 16 nests in 1988 and five nests in 1989 were counted at the study area before the censuses started in the middle of June. Because the census in 1986 started in July, at least a half month of nesting activity at the beginning of the season was missed. The number of unchecked nests in 1986 was not known but was probably very minor relative to the whole season.
Most nesting activity occurred from July through the first
half of October in all three years (Figure 16). However, the seasonal distribution shifted among the years. Peak nesting activity occurred




29
during the latter half of August in 1986 and the first half of September in 1988. In 1989, nesting activity reached a plateau from the latter half of August through September. The last nest in the study area was observed on 27 November 1986, 15 November 1988 and 20 November 1989. An insignificant amount of nesting activity was observed during November of all three years.
Approximately 11,724 nests in 1986, 10,509 nests in 1988, and 4,491 nests in 1989 were deposited in the two-mile study area of the beach during the census period. The density of nests along the beach was 7.3 nests/m in 1986, 6.5 nests/m in 1988, and 2.8 nests/m in 1989.
The proportion of nests deposited in the vegetation/border
zone and in the open sand zone across the season was, respectively, 51.1% and 48.9% in 1986 (n=3,543), and 58.0% and 42.0% in 1988 (n=3,521) and 63.2% and 36.8% in 1989 (n=2,041) (Figure 17). The proportions among these three years were significantly different (df=2, X2=81.8, P<0.0001). Tukey-type multiple comparison tests for proportional data revealed that the proportion of clutches deposited in the vegetation/border zone in 1989 was significantly the highest, intermediate in 1988 and the lowest in 1986 among the three years [q 0.05, -, 3 = 3.314, q=24.883 (86 vs 89), 16.226 (86 vs 88), 10.976 (88 vs 89)]. [Data were transformed by the equation: p'=1/2(arcsin/(X/(n+1 )+arcsin(X+1 )/(n+1 )), Zar (1984).]
Figure 17 also shows the seasonal fluctuation of nest site selection. The proportions of nests found in each zone were significantly different at each half-month period in each year (df=7, X2=31.7, P<0.0001 in 1986, df=8, X2=41.2, P<0.0001 in 1988 and df=8,




30
X2=19.9 P<0.01 in 1989). Tukey-type multiple comparison tests for the proportional data (Zar, 1984) identified the differences among each seasonal period. There was no common seasonal trend among the three years. In 1986, the proportion of clutches deposited in the vegetation/border zone was significantly higher early in the season than during the rest of the season, and the proportion decreased as the season progressed. In 1988 and 1989, the proportion of nests in the two zones oscillated throughout each season in a similar way. The proportions of clutches deposited in the vegetation/border zone shifted from high to low, and then again from high to low throughout each season.
Analysis of Egg Survivorship Samples
Fates of Sample Nests
The fates of 49 nests in 1986, 88 nests in 1988 and 113 nests in 1989 were determined. Mean clutch size for the sample nests was 116.9 eggs (SE=3.4, range=28-165, n=49) in 1986, 109.1 eggs (SE=2.1, range=53-148, n=88) in 1988, and 107.1 eggs (SE=2.1, range 53-152, n=89) in 1989. The frequency distributions of the emergence success rate of these nests were far from a normal curve and instead showed concave bimodal shapes in each year (Figure 18). For all three years, most of the sample nests fell into one of two extremes: highly successful emergence percentage (>70%) or entire dead clutches. No hatchlings emerged from 24.0% and 25.0% of sample nests deposited in the open zone and in the vegetation/border




31
zone respectively in 1986, 21.6% and 39.2% in 1988, 9.4% and 28.3% in 1989. Total nest loss was caused by beach erosion, inundation by excessive rain or Hurricane Joan, depredation by coatis (Nasua narica) and ghost crabs (Ocypode quadrata), and excavation by nesting female turtles (Figure 19).
For each nest, emergence success was defined as the percentage of its egg clutch that produced hatchlings that successfully emerged from the sand column. Mean emergence success throughout each season was 54.8% (SE=7.3, n=25) in 1986, 57.8% (SE=6.2, n=37) in 1988, and 74.7% (SE=4.8, n=53) in 1989 in the open zone; emergence success was 47.3% (SE=8.1, n=24) in 1986, 42.7% (SE=5.6, n=51) in 1988, 60.0% (SE=5.4, n=60) in 1989 in the vegetation/border zone. Although the mean emergence success in the open zone was slightly higher than that in the vegetation/border zone during each year, the differences between the two zones were not significant (Mann-Whitney U test, p=0.778 in 1986, p=0.069 in 1988, p=0.061 in 1989). With pooling the two zones, overall mean emergence success of sample nests throughout each season was 51.2% (SE=5.4, n=49) in 1986, 49.1% (SE=4.2, n=88) in 1988 and 66.9% (SE=3.7, n=113) in 1989. The overall mean emergence success in 1989 was significantly higher than that in either 1986 and 1988 [Kruskal-Wallis Test, df=2, H=21.372, p<0.0001; nonparametric multiple comparison test with unequal sample sizes (Zar, 1984), 1989 vs 1988, Q=4.121, 1989 vs 1986, Q=3.465, Q 0.05,3=2.394].
Emergence success exhibited a different seasonal trend for each year (Figures 20, 21 and 22). For analyzing the effects of




32
seasonal period (half-month interval) and nest location (open zone vs vegetation/border zone) on egg survivorship, nonparametric twoway factorial ANOVA with unequal replication (Zar, 1984) was applied for each year's data (Table 4). The ranks of the data, rather than the raw data, were used. The analysis revealed a significant difference in seasonal emergence success in 1988, but not in 1986 or 1989. There were no significant effects of Zone or Zone x Period interaction in any of the three years. These results agree with an analysis carried out with a parametric ANOVA in which case the data were transformed to their arcsine, but the assumption of homogeneity was not satisfied.
For the 1988 data, a nonparametric multiple comparison test with unequal sample sizes (Zar, 1984) identified a significant seasonal difference of emergence success between the clutches deposited in the latter half of July (79.7%, SE=4.6, n=16) and ones in the latter half of August (16.9%, SE=9.2, n=14)(Q=4.10 > Qo.o5,6=2.94; Figure 21, total). The other comparisons between each period were not significant. The significant seasonal difference in the 1988 season was mainly accounted for by two heavy inundations that occurred close together in October. These inundations caused by excessive rain on 6 October (221 mm/day) and by the surf of Hurricane Joan on 21-22 October caused substantial loss of nests deposited after the latter half of the three month nesting season. The sample nests deposited in the latter half of July in 1988 hatched before these inundations, and all these clutches produced hatchlings. However, the inundations caused the complete loss of seven of 14 nests deposited during the latter half of August 1988.




33
In addition, predation loss of three nests by mammals and one nest by ghost crabs further decreased the emergence success during this period.
Abiotic and Biotic Factors Affecting Eag Survivorship
Fate of eggs from sample nests
Table 5 and Figures 23-24 show the fate of green turtle eggs from sample nests in 1986, 1988 and 1989. The emergence success of eggs from the sample nests throughout each season was 54.6% (n=2915) in 1986, 59.5% (n=3899) in 1988, and 75.5% (n=5890) in 1989 in the open zone; comparable figures for the vegetation/border zone are 46.4% (n=2801) in 1986, 43.0% (n=5715) in 1988, and 58.5% (n=6216) in 1989.
Criteria for assessing the cause of egg mortait
The category of "Destroyed by mammals" includes all eggs lost because of initial damage by mammals. For example, when coatis excavated green turtle nests, they rarely consumed all the eggs. However, soon other animals such as black vultures, ghost crabs and ants often destroyed the remaining eggs in the half excavated nest. In this case, I categorized the entire clutch as lost under "Destroyed by mammals." If the partially depredated nest successfully produced hatchlings, the numbers of missing egg shells (clutch size minus counted egg shells at excavation), which were assumed to be removed from the nests by mammals, were classed under this category. The same principle was applied to the category of "Destroyed by female turtles."




34
For assessing the cause of embryonic death, when the timing
of inundation reasonably matched the stage of the arrested embryos, I assumed these eggs were killed by the inundation event. In the 1988 season, because two inundations (6 October excessive rain and 21-22 October Hurricane) occurred so close to each other, I could not determine how much of egg mortality was caused by each inundation.
Abiotic factors
Among abiotic factors, erosion and inundation by surf,
excessive rainfall resulting in inundation from ground water, and Hurricane Joan were mainly responsible for reducing egg survivorship. These abiotic extremes, as a whole, were responsible for loss of 18.5% of eggs in the open zone and 20.0% of eggs in the vegetation/border zone throughout the three years (the percentages of egg loss for the three years were calculated as the weighted mean in each zone). In addition to these abiotic extremes, artificial debris prevented a few hatchlings from emer ging.
Beach erosion washed away 5.7% (open zone) and 2.9%
(vegetation/border zone) of the eggs throughout the three years. The damage from erosion showed considerable yearly variation, and the 1986 season suffered the highest loss (15.3% of eggs in open zone,
8.7% in vegetation/border zone in the 1986 season). Surf inundation did relatively minor damage to the eggs throughout the three years (1.4% in open zone, 1.3% in vegetation/border zone).
Hurricane Joan, whose center passed about 180 km off the
coast of the Tortuguero beach in the Caribbean Sea during the night




35
of 21 through the morning of 22 October 1988, caused substantial damage to green turtle nests. When Hurricane Joan passed, the Tortuguero beach was at the outer margin of a 70 km/h wind speed zone (Instituto Meteorologico Nacional, Costa Rica). During the first beach census after the hurricane, I confirmed that the waves apparently washed up to the vegetation area along most of the beach and that all developing sample nests were covered by waves to various extents. However, beach erosion was not severe and none of the sample nests was washed away. All damage by Hurricane Joan on the sample nests was caused by surf inundation. Hurricane Joan brought only a little rainfall to the Tortuguero beach (total 25 mm during 21 to 24 October 1988). It is not known whether the ground water raised by the high waves or the surf salt water itself suffocated the eggs.
The extent of damage caused by Hurricane Joan in 1988 was not accurately assessed because the damage to several nests was inseparable from that caused by excessive rain on October 6. The minimum estimated damage by Hurricane Joan was 1.6 % egg loss in the open zone and 10.8% egg loss in the vegetation/border zone. However, if I include the undetermined dead eggs killed by either the hurricane or the excessive rain, the figures increase to 9.5% egg loss in the open zone and 13.3% egg loss in the vegetation/border zone. No other hurricane threatened the Caribbean coast of Costa Rica during the three-year study.
Freshwater inundation by ground water, mainly associated with excessive rainfall, was a major and constant abiotic factor causing mortality of eggs throughout all three years. All developing




36
stages, including emerging hatchlings, were vulnerable to suffocation. Clutches in both zones were vulnerable to flooding. Excluding eggs that died from unknown causes between the flooding and Hurricane Joan in 1988, freshwater inundation was responsible for 8.4% of egg loss in the open zone and 11.3% of egg loss in the vegetation/border zone throughout the three years. When eggs of undetermined fate are included, the figures increase to 11.0% and 12.1%, respectively. The risk of mortality by flooding between the two zones was not consistent among years. The percentages of eggs lost by freshwater inundation are as follows: 13.6% and 17.8% in 1986, 3.0-10.7% and 12.5-15.0% in 1988, 8.5% and 3.8% in 1989 for eggs deposited in the open zone and in the vegetation/border zone, respectively. In 1986 and 1988, significantly more eggs were killed by flooding in the vegetation/border zone than in the open zone (X2=18.7, p<0.0001, 1986; X2=7.3, p=0.0068, using possible closest rate in 1988), while the opposite trend occurred in 1989 (X2=116.5, p<0.0001, 1989).
Artificial debris, such as tangled fishing nets and ropes were frequently seen on the beach. On one occasion, a buried tangled rope physically prevented 10 hatchlings of a sample nest from emerging to the surface, although the remaining 119 hatchlings escaped through the rope. Fishing nets and ropes have a high potential to cause substantial damage to emerging hatchlings. Biotic factors
Predation by several types of animals and excavation by female turtles were major factors responsible for mortality of




37
green turtle eggs. Altogether, 5.4% of the eggs in the open zone and 20.5% of eggs in the vegetation/border zone were destroyed throughout the three years by these causes. Significantly more eggs in the vegetation/border zone were destroyed than in the open zone for all three years (3.7% vs 14.0% in 1986, 6.2% vs 15.9% in 1988,
6.2% vs 31.4% in 1989; X2=188.9, p<0.0001, 1986; X2=209.4, p<0.0001, 1988; X2=1246.0, p<0.0001, 1989).
Coatis, Nasua narica, were responsible for the majority of nest loss by mammal predation. However, because it was sometimes difficult to confirm the mammal species on the sample nests only by their footprints, all nests excavated by mammals were classed as "Destroyed by mammals."
Coatis are diurnal predators on green turtle nests in
Tortuguero. In the protected park area, solitary or more often a band of coatis were frequently sighted walking along the upper part of the beach during the daytime. The band membership typically consisted of two to six cubs and several adults. They generally excavated a series of adjacent nests along the border of the vegetation. Although they rarely consumed all the eggs in a nest, the rest of the eggs were soon eaten by other animals. For the sample depredated nests, 72.7% of nests depredated by mammals (n=33) did not produce any hatchlings.
Raccoons, Procyon lotor, were also responsible for destruction of nests, but to a lesser extent. The extent of damage was not accurately obtained because of the difficulties of species identification. For the depredated sample nests, only one nest in the vegetation/border zone in 1989 was confirmed to be damaged by




38
raccoons. The density of raccoons in Tortuguero was probably much lower than that of coatis, because I sighted only two adult raccoons at night and one juvenile during the daytime during the three years of study. The juvenile was seen eating several eggs of an unsampled nest under the vegetation.
In addition to the above two predator species, one skunk and a few feral dogs were sighted excavating several unsampled nests in the study area. It is not known whether some of the sample nests were depredated by these species. A skunk (unknown species) was seen excavating an unsampled nest in 1986; this was the only time a skunk predator was sighted during the study. Four unsampled nests in 1988 were predated by feral dogs. Feral dogs were rarely seen in the park area throughout the study because of a successful dog control program on the beach by personnel of Parque Tortuguero.
Predation by mammals was responsible for loss of 3.0% of the eggs in the open zone and 17.3% of the eggs in the vegetation/border zone throughout the three years. Only four of 115 (3.5%) sample nests in the open zone were partially or completely preyed upon during the three years as compared to 29 of 135 (21.5%) sample nests in the vegetation/border zone (X2=16.0, p<0.0001).
There was considerable yearly variation in predation by
mammals. In the 1989 season when the number of nests deposited was about half that of the 1986 and 1988 seasons, predation by mammals increased to the highest percentage in both zones (egg loss of 5.0% in the open zone, 30.9% in the vegetation/border zone in the 1989 season). The 30.9% egg loss by mammal predators in the




39
vegetation/border zone in the 1989 season was the highest loss factor among all the categories throughout the three years.
The mean time of predation for the sample nests for which the predation date was confirmed was 41.1 days after deposition (range=Day 0 to Day 72, SE=4.0, n=26) (Figure 25). The distribution of predation days was composed of two groups: before 11 days and after 32 days. Most (79.2%) of the egg depredation occurred between 32 and 68 days of incubation. One nest was depredated at Day 72 when the hatchlings were emerging to the surface.
For a supplemental observation on mammal predation, I
recorded the number of recently depredated nests in the study area during the beach census. For all three years, more nests depredated by mammals were observed in the vegetation/border zone than in the open zone (X2=82.9, p<0.00O1, n=233, 1986: X2=152.0, p<0.0001, n=330, 1988: X2=174.3, p<0.0001, n=456, 1989) (Figure 26). Although nest predation by mammals was observed throughout the incubation period, the seasonal trend of predation intensity varied considerably among the three years (Figure 27). The trend in 1988 showed an almost opposite trend to that in 1989. While predation intensity in the 1988 season increased to the highest rate in the early part of the season and gradually decreased as the season progressed, nest predation in the 1989 season steadily increased as the season progressed almost to the end. The peak of nesting activities occurred between the latter half of August and September for the three years (Figure 16). The seasonal fluctuation of predation intensity did not match the seasonal fluctuation of nesting act ivi t ies.




40
Although ghost crabs, Ocypode guadrata, were very common on the Tortuguero beach, predation by ghost crabs was not a serious factor affecting egg survivorship (mean egg loss of 0.8% in the open zone and 1.3% in the vegetation/border zone throughout the three years). Only three sample nests were confirmed to have been damaged by ghost crabs during the three-year study. Two of them were completely destroyed, and the other was partially destroyed by ghost crabs. For those nests that were completely destroyed, numerous crab burrows were observed around the nests during the incubation period; at the time I excavated the nests to check on the eggs, only the flagging tape marker and many scattered torn shells remained at the bottom of the nest. For the partially destroyed nest, I used a characteristic nipped hole on the egg shells for identifying the eggs damaged by ghost crabs. However, my assessment rate of ghost crab predation for this study might b conservative. In cases where only a few eggs from a clutch were dep0redated by crabs and the typical nipped holes were not clear enough to identify, torn shells were classed under the "Rotten intact or ruptured eggs" category.
Predation by termites, unknown species, was a very minor
factor causing mortality of eggs. Eight sample nests were found to be partially included in termite nests, but generally only small numbers of eggs from each clutch (mean 3.7 eggs, SE=2.1, range=1 to 16, n=7) were killed by surrounding clay materials. The insides of most eggs were stuffed by the clay material through a few tiny holes. The termite nests were typically constructed under washed-




41
out logs on the beach, and all sample nests depredated by termites were located near such logs.
An unidentified animal was responsible for minor damage on eggs (mean egg loss of 0.9% in the open zone, 0.2% in the vegetation/border zone throughout the three years). Typically, there were a few tiny holes (diameter 1-2 mm) on the egg shells, which were different from the nipped holes by ghost crabs, and the insides of the shells were often partially stuffed with dry sand. There were no visible animals in the shells at excavation. Mean number of eggs damaged by this cause was 6.2 per clutch (SE=2.1, range=1-47, n=23). I suspect that fire ants, which were very common on the Tortuguero beach, might be responsible for these damages, but I have no confirmation.
Excavation of sample nests by nesting female turtles occurred twice in the 1986 season and six times in the 1988 season, but never in the 1989 season. The extent of damage by this cause (egg loss of 1.1% in the open zone and 4.8% in the vegetation/border zone throughout the three years) was minor relative to predation by mammals. Generally the damage occurred when a female dug an egg chamber overlapping a previously laid clutch. In such a case, numerous eggs were excavated to the surface. Two of the damaged sample nests did not produce any hatchlings because mammal and other predators destroyed the rest of the eggs. Another six sample nests produced some hatchlings; those nests were well covered by sand from another or the same female turtle's activities before predators invaded the nests.




42
As a supplemental observation on excavation by female
turtles, I recorded the number of recently excavated nests by turtles in the study area during the beach census. In total, 238 nests in 1986 (48 surveys), 123 nests in 1988 (56 surveys), and 19 nests in 1989 (80 surveys) were observed to be excavated by female turtles. The percentage of the excavated nests of total nests in each season was calculated as 6.5% in 1986, 3.5% in 1988, and 0.9% in 1989, respectively. In the 1989 season when nest density was the lowest among three years, the intensity of excavation by turtles decreased to a very low level relative to the other two years (Figure 28), and none of the sample nests was excavated. The seasonal fluctuation of excavation by turtles was positively related to the extent of nesting activities (Figure 29).
Only one sample nest was damaged by plant roots. A bundle of young coconut tree roots infiltrated three eggs of a sample nest, which was located close to the trunk of the tree. Several unsampled nests were found to be partially tangled by roots of sea oats, but no apparent damage by the roots was recognized. These roots did not penetrate any egg shells, and those nests had high emergence success. Natural damage by plant roots was probably minimal on green turtle eggs at Tortuguero.
Other Categories of Mortality
Over the three years, a mean of 3.6% of eggs in the open zone and 3.1% of eggs in the vegetation/border zone ceased embryonic development from no apparent cause. Actual causes of mortality of these eggs, whether genetically inherent or environmentally induced,




43
are not known. Possibly, some previous physical disturbance, such as a minor inundation, caused the arrested development of some of these eggs at a later stage.
Over the three years, 3.3% of the eggs in the open zone and
1.9% of the eggs in the vegetation/border zone did not show any sign of visible embryos or blood formation. I did not use a white circle or patch on the egg shell as a criterion of development. Therefore, the above figures probably overestimate the percentage of infertility.
Over the three years, 5.3% of the eggs in the open zone and 4.3% of the eggs in the vegetation/border zone were rotten with intact shells or were ruptured at the time of excavation. Most late stage embryos were not included in this category because the remaining shell and bones were detectable even in a ruptured shell. However, some of the earlier stage developing eggs and infertile eggs were possibly included in this category.
Analysis of Sex Ratio
Sexed Samples
A total of 55 nests (12 in 1986 and 43 in 1988) were sampled for sexing (Table 6), and the temperatures of 52 of those nests were monitored successfully. Twenty eggs were collected as a subsample from each clutch during 50-60 days except from one sample nest in 1986. Because I missed sampling eggs in this nest before the emergence of the clutch, only the two remaining hatchlings in the




44
nest were collected for sexing. Some embryos were not fully developed and some tissues deteriorated before being fixed, thus the number of sexed gonads was reduced to an average of 17.8 gonads per clutch (SE=0.4, range=2 to 20, n=55 nests, total turtles=979). Undetermined gonads, which possessed both germinal epithelium (female component) and seminiferous tubules (male component), were found in seven of the 979 individuals (0.7%).
Sex Ratio vs Temperature
To date no laboratory experiment to determine accurately the critical period of green turtle eggs in sex determination has been conducted. Spotila et al. (1987) found that mean incubation temperatures during the middle third of development explained the observed sex ratio in green turtles on the natural beach in Tortuguero. Results of temperature shift experiments in the laboratory with loggerhead turtles, Caretta ;aretta, (Yntema and Mrosovsky, 1982) also suggested that the thermosensitive period for sex determination occurs during the middle third of development. The middle third of development is also the critical period in freshwater turtle species (e.g., Bull and Vogt, 1979). Thus, I used the middle third of the development period as a critical period in green turtles for this study.
Emergence lag (interval between pipping and emergence) in
green turtles is believed to take from 3 to 7 days (Balazs and Ross, 1974), but no quantitative data are available. Recently, Christens (1990) found the mean lag period to be 5.4 days for loggerhead




45
turtles. For this study, I assumed 5 days as the emergence lag in Tortuguero green turtles, and I subtracted 5 days from the deposition to emergence period to obtain an approximate developmental period for each clutch. Mean temperature during the middle third of the development period was calculated for each clutch. Figure 29 shows the relationship between sex ratios of sample nests and mean nest temperatures in the center of the clutch and mean sand temperature (1 m away from the clutch, at the same depth) during the critical period. The proportions of females were positively correlated with ambient temperatures. Both sexes were produced over the entire temperature range of the sample nests (mean nest temperature range=27.0-30.70C, mean sand temperature range=26.4 30.20C). Because metabolic heat is produced during the middle of development, the mean nest temperatures were slightly higher than the sand temperatures at the same nest during the critical period (mean difference=0.71C, SE=0.05, range=0.18-1.48, n=41). As a result, the regression of sex ratios and mean sand temperature was shifted down approximately 0.70C on the function of temperatures, as compared to mean nest temperature data. The pivotal temperatures (expected equal sex ratio) were calculated by logistic regressions as 29.40C for the mean nest temperatures, and 28.50C for the mean sand temperatures (see Prediction of Sex Ratio from Independent Variables for a detailed explanation of the logistic regression model). For both data sets, there was a substantial amount of variation in sex ratio over the range of incubation temperatures.




46
Seasonal Variation and Zone Effect in 1988
In 1988, the sex ratios varied from 0 to 100% female for the sample nests in the open zone and from 0 to 89.5% female for the sample nests in the vegetation/border zone (Table 6). Figure 30 shows the seasonal variation of sex ratios in the 1988 season. The extent of seasonal fluctuation was more apparent in the open zone than in the vegetation/border zone. For analyzing the effects of seasonal period (half-month interval) and nest location (the open zone vs the vegetation/border zone) on sex ratio throughout the season, two-way factorial ANOVA with unequal replication was applied (Table 7). Data were transformed (p'=1 /2(arcsin1(X/(n+1 )+arcsin (X+1 )/(n+1)) (Zar, 1984). An Fmax test showed homogeneity of variances of the transformed data. The analysis revealed significant differences in seasonal fluctuation of sex ratio and in effects of Zone, but no significance in Period x Zone interaction. Fisher's procedure of least significant difference revealed a significant seasonal difference of sex ratios in the open zone (Figure 31), but not in the vegetation/border zone. In the open zone, the sample nests deposited in the latter half of July had significantly more females than during other periods within the season except for the nests deposited in the first half of August.
From the end of August through September 1988, the sand temperatures, at least in the open zone, rose above the pivotal temperature (28.50C), associated with less rainfall (Figure 6, Figure 13). Most sample nests in the open zone deposited from the latter




47
half of July to the middle of August encountered a higher sand temperature period during the critical stage and produced femalebiased sex ratios. However, the nests in the vegetation/border zone during the same period showed only a slight increase towards a female-biased ratio. The difference in sex ratios between zones during the high sand temperature period was significant (the open zone, mean=82.1%, SE=7.9, range=25.0-100.0, n=10; the vegetation/border zone, mean=31.7%, SE=11.3, range=7.1-89.5, n=7, one-way ANOVA, df=l, F=14.24, p=0.018, sample nests deposited from the latter half of July to the first half of August). A heavy rain event on 6 October (221 mm/day) and inundation by Hurricane Joan during 21 to 22 October decreased the sand temperatures (Figure 13). In both zones, two of eight sample nests deposited during the latter half of August and all 11 sample nests in September were affected during the temperature sensitive period by this cooling. Eleven of those 13 sample nests showed male-biased sex ratios in both zones (the open zone, mean=20.9%, SE=8.5, range=0.0-60.0, n=7; the vegetation/border zone, mean=20.7%, SE=10.2, range=0.0-56.3, n=6), and there was no significant difference between the zones (one-way ANOVA, df=l, F=0.01, p=0.923).
Overall Sex Ratio in 1988
To estimate overall sex ratio, frequency of temporal nest
distribution was combined with egg survivorship and sex ratio data by half month intervals. Because a significant effect was found in the seasonal factor, but not in the zone factor on egg survivorship




48
(Table 4), the egg survivorship data in both zones were pooled. Since both the seasonal and zone factors significantly affected sex ratio (Table 8), each factor was treated separately. Because multiple comparison tests on both egg survivorship and sex ratio data for seasonal analysis failed to assign each half-month period to significantly different subsets, mean values on each bimonthly base were applied. Table 8 shows the value of each parameter applied and the process of estimation. Overall sex ratio was calculated to be 40.6% females in the 1988 season.
Seasonal Variation and Zone Effect in 1986
In 1986, numerous rainy days and sporadic heavy rains kept sand temperatures lower than the pivotal temperature for most of the season (Figure 13). Eight of 12 sample nests produced only males (Table 6). The proportion of females in sample nests varied from 0 to 47.1% in the open zone and from 0 to 68.8% in the vegetation/border zone. Because of the small sample size and sampling bias, assessing the difference between sex ratios between the zones was not possible. To analyze the seasonal factor, data in both zones were pooled by monthly intervals. Figure 32 shows the seasonal variation of sex ratios in the 1986 season. Throughout the season, sex ratios were strongly biased toward males. There was no significant difference in the sex ratios observed throughout the season (Kruskal-Wallis test, df=2, Hc=0.983, p=0.612; because of heterogeneity of variance, nonparametric analysis was applied).




49
Overall Sex Ratio in 1986
Because a significant seasonal effect was not detected (as
mentioned above), overall proportion of females was calculated as a mean value of all sexed sample nests to be 10.1% (n=11) in the 1986 season. The number of sexed sample nests was biased towards nests in the open zone (8 of 12, 66.7%) as compared to the observed nearly 1:1 distribution between the two zones (Figure 17). Because the open zone showed higher sand temperatures than the vegetation/border zone (Figure 13), the 10.1% proportion of females may be an over-estimate.
Prediction of Sex Ratio from Independent Variables
To assess predictability of sex ratios of Tortuguero green
turtle hatchlings from independent variables, simple and multiple logistic regression models were applied. Linear correlations between nest temperature, an assumed main environmental factor, and other variables were also analyzed. For the independent variables, mean nest temperatures and mean sand temperatures during both the critical period and the incubation periods, three different sets of rainfall data, bottom depth of nests, and zone of the nests were included. The data set included 52 nests consisting of 944 sexed turtles, for which nest temperatures were collected in 1986 and 1988. The data set on sand temperatures was reduced to 41 nests consisting of 728 sexed turtles from the above data set.




50
Logistic Regression Model
For dichotomous dependent variables (i.e., female or male), a logistic model (F(Z)=exp(Z)/(l+ exp(Z)), z=B1+BO*X) is an attractive alternative to the linear probability model because the logistic model satisfies 0-1 constraint on probability unlike the linear specification, and provides a smooth symmetric S-shaped curve (Aldrich and Nelson 1984). Because this model is one of the most widely used nonlinear models, the availability and flexibility of computer programs are better. The logistic models were calculated with maximum likelihood estimation, and pseudo R2 was computed as (-loglikelihood for Model) / (-loglikelihood for C Total) (JMP version 2.02, SAS Institute Inc., 1989).
Logistic Regression with a Single Variable
Nest and sand temperatures
Mean nest temperatures were highly correlated with mean sand temperatures during the critical period (mean nest temperatures=0.99 + 1.02 mean sand temperatures, R2=0.92, n=41, p<0.001). Mean nest temperature data showed the best fit (pseudo R2=0.20) to a logistic model among the variables for a single regression model (Table 9, Figure 33). However, the variability of sex ratios with mean nest temperatures was still very high. Particularly at the middle range of temperatures, residuals spread up to 0.4 probability (Figure 33). Mean sand temperature data




51
showed a slightly lower fit to the logistic model (pseudo R2=0.17, Table 9) than did mean nest temperature data. Incubation period
Incubation periods of the sexed sample nests varied from 51 to 74 days (mean=61.7, SE=0.7, n=52). Incubation periods were negatively correlated with mean nest temperatures (mean nest temperatures=38.3 0.15 incubation days, R2=0.49, n=52, p<0.001, Figure 34). Therefore, female sex ratio was also negatively correlated with incubation period (Figure 35). Only males were observed in nests that had over 68 days of incubation, whereas both sexes were found in nests that had incubation periods of between 56 and 66 days. Incubation period data were the second best fitted regression model (pseudo R2=0.18, Table 9). However, the variability of sex ratio for the middle range of incubation days was so high that the observed ratios almost covered the entire range of sex ratios. Therefore, predictability of sex ratio by incubation period was poor. Rainfall
The amount of rainfall recorded at each sample nest was calculated three ways for statistical testing: R-I=mean daily rainfall throughout the critical period, R-II=mean daily rainfall throughout the critical period plus the previous 10 days, R-Ill=mean daily rainfall from the deposition of eggs throughout the critical period.
Rainfall decreased the sand temperature at the depth of nests. All three data sets of rainfall were negatively correlated with the mean sand temperatures (R2=0.16 (R-1), 0.33 (R-11), 0.47 (R-Ill), all




52
models, n=52, p<0.05). As expected, the proportion of females decreased with an increase of mean daily rainfall (Figure 36). This trend was the most apparent for the nests in the open zone and as a whole in the data set of R-111, which showed the highest correlation with mean nest temperatures. There seemed to be an interaction between rainfall data and zones. When the mean daily rainfall was low (< 8 mm/day), the nests in the open zone showed higher female sex ratios than ones in the vegetation/border zone in general. On the other hand, as the mean daily rainfall increased, there was no apparent difference of sex ratios between the zones. R-111 pooled zone data (pseudo R2=0.17) fit a regression model almost as well as incubation period and mean sand temperature data did. If only data for the nests in the open zone were used, the pseudo R2 increased to
0.27. However, variation of the data was still high for the whole range.
Bottom depth of clutch
Bottom depth of sample nests varied from 59 to 105 cm (average=78.3, SE=1.4, n=52). In general, sand temperatures decreased with depth of the nest. However, no significant correlation between mean nest temperatures and bottom depth was observed (p=0.215, n=51). As expected, there was no apparent trend between the sex ratios and the bottom depth of the sample nests (Figure 37).




53
Multiple Logistic Regression Model
Because of a strong correlation between mean nest
temperatures and mean sand temperatures (R2=0.92), only mean nest temperature was used for the multiple regression model to avoid a problem of multi-colinearity. Mean nest temperature, incubation period, three sets of rainfall data, bottom depth of nests, and zone of the nests (the open zone=1, the vegetation/border zone=0) were entered into logistic multiple regression models. Backward elimination (Table 10) and stepwise regression procedures were used to eliminate nonsignificant variables (alpha=0.05). For both procedures, the same model was selected, in which mean nest temperatures, incubation period, mean daily rainfall from deposition through the critical period (R-Ill), and zone significantly explained some variation of sex ratios (pseudo R2=0.24, Table 11). Figure 38 shows residuals on this selected model. Although the selected model was slightly improved from the best single variable model of mean nest temperatures (Figure 33), fitness of the model was still low.
All first order interactions of the selected variables were
tested individually by a likelihood ratio test (Hosmer and Lemesho, 1989) to determine their significant contribution to the selected model. Two interactions, between incubation period and zone (G=10.42, df=2, p<0.025) and between mean daily rainfall (R-Ill) and zone (G=7.44, df=2, p<0.01), were significant. However, in both cases when each interaction was included in the selected model, the




54
variable of incubation period became insignificant. Moreover, improved fitness was negligible (pseudo R2=O.25 for both models).
Finally, two easily measured param ete rs-- mean daily rainfall (R-111) and zone--were entered into a multiple regression model to assess their predictive ability relative to that of nest temperature data. Because there seems to be an interaction between rainfall data (R-111) and zones in affecting the sex ratios (Figure 36), their interaction was also included. Mean nest temperature is biologically the primary factor for determining sex of green turtle hatchlings and the single regression model showed it to be the best single fit model among the applied independent factors. The fitness of the multiple regression model using rainfall and zone (pseudo R2=O.23) (Table 12) was slightly better than that of the single regression model using mean nest temperatures (pseudo R2=O.20).




CHAPTER 4
DISCUSSION
Nesting Density and Temporal Nest Distribution
High yearly fluctuation of nesting density is known for
Tortuguero green turtles (Bjorndal et al., 1985). The 1986 and 1988 seasons were two of the highest density years since 1956, whereas the 1989 season was in the middle of the range according to the long term monitoring project at the northernmost 8.1 km section (K. Bjorndal, pers. comm.). My study site was part of the highest nest density area along the beach (Carr et al., 1978), and this trend was observed throughout the study (Horikoshi, unpublished data). Thus, the recorded densities in 1986 (7.3 nests/m) and in 1988 (6.5 nests/m) throughout each season were probably among the highest existing at the Tortuguero beach since 1956. As far as I know, nesting densities of green turtles of this magnitude are only known on an Australian island, an Oman beach, and an Ascension beach (Groombridge and Luxmoore, 1989; Mortimer, 1981; Mortimer and Carr, 1987).
Temporal distribution of green turtle nests varied greatly in the frequency pattern and slightly in the range of season among the three years. Carr et al. (1978) noted that the main breeding activity takes place in July, August and September, although a few turtles nest throughout the year. The 1988 season in general agreed with
55




56
this trend, but substantial nesting activity was extended at least until the middle of October in the 1986 and 1989 seasons. The number of nests deposited in October (662) exceeded that in July (389) in the 1989 season. Carr et al. (1978) presented data on the mean seasonal profile of nesting arrivals between 1956-1976, in which the peak of nesting arrival occurred at the end of August. However, because green turtles deposit multiple nests (1-7) within a season (Carr et al., 1978), the pattern of nest number profile cannot be clearly estimated from their figure. So far, the only available information on the full seasonal profile of green turtle nests at Tortuguero is for the 1977 season taken by Fowler (1979). She noted that nesting began in June, peak activity occurred in early August, and only a few turtle nests were found by late November in 1977; she showed the actual nest profile only from 13 July through 14 September. The three years of this study also showed the beginning of nesting activity to be sometime in June, and minimum activity to be in November, but the timing of peak activity was different; in this study the peak was in the latter half of August in 1986, in the first half of September in 1988, with a plateau from the latter half of August through September in 1989. The principal nesting season in Tortuguero green turtles seems to be confined to June through October, with peak activity sometime between August and September, but the actual dates vary from year to year.




57
Seasonality of Environmental Parameters
Because the Tortuguero beach is located in the equatorial zone, seasonal fluctuation in air temperatures is very low relative to the high fluctuation in rookeries in the temporal zone. However, there is predictable seasonal change in several environmental factors that are important in influencing egg survivorship, sex ratio, and possibly nest site selection between different thermal zones in Tortuguero green turtles during their incubation period (June through December).
The rainfall record indicated that there are two cycles of rainy/dry seasons in Tortuguero and that a high level of rainfall fluctuation occurs at the middle of the green turtle incubation period; rainfall decreases in September. Since sand temperature was inversely related to rainfall, the warmer sand temperature on the beach was expected to occur in September relative to the rest of the season. This tendency was observed during all three years of this study.
Although seasonal trends of wave activity varied among the three study years, the wave activity was relatively low in September for each season. The positive correlation found between wave activity and rainfall suggests that fluctuation of wave activity also has a general seasonal trend; September has the most calm days at Tortuguero. At Tortuguero, the mean tidal range is 0.2 m (Limon/Bluefield, Tide Table 1986: East Coast of North and South America). Therefore, the high fluctuation of wave size (almost none to >2.0 m) has more influence on the width of the beach than the tide




58
does on the emergence time of nesting females. In addition, the extent of erosion is generally also correlated with activity of waves.
Spatial Distribution between the Zones
It is interesting to compare the spatial distribution of green turtles in 1986, 1988, 1989 at this study site with similar data collected in a different area of the Tortuguero beach (northern-most
8 kin) in 1986, 1987, 1988 (Bjorndal and Bolten, 1992). Although there was a slight difference in definition of the zonation of the beach (my vegetation/border zone is probably slightly larger than their combined vegetation zone and border zone), both studies revealed significant yearly variation of spatial distribution between the shaded area and open area over years, and that there was no common seasonal trend of fluctuation among the years. Comparison between the common years (1986 vs 1988) reveals that the yearly fluctuation of the spatial distribution showed the same trend: a higher proportion of nests in the open area in 1986 than in 1988.
Bjorndal and Bolten (1992) hypothesized that the 1986 season's higher rainfall during their study period (July to September), and consequently the higher sand moisture, may have physically facilitated nest digging in the open area relative to the drier years in 1987, 1988, and that the wetter sand may be partially responsible for the high proportion of nests in the open area in the 1986 season. When I compared the relationship between the proportion of nests deposited in the open zone versus the




59
vegetation/border zone and the mean daily rainfall from July through October with half month intervals for my 1986, 1988 and 1989 data sets, the proportion of nests in the shaded zone was lower as a function of the amount of rainfall, but the relationship was not statistically significant (P'=arcsin/P; Y=0.899-0.003X, Rsquared=0.12, n=24, p=0.10).
Although the spatial distribution between the zones fluctuates seasonally and yearly, the extent of fluctuation was consistently limited to a small range around the middle value: the proportion of nests in the shaded zone was nearly equal or a little higher (42-60% in 1986, 51-67% in 1988, 48-71% in 1989) relative to nests in the open area throughout each season. This general trend agreed with the results at the northern section of the Tortuguero beach in 1987 (47%-50%) and 1988 (49-62%) but not in 1986 (28-38%). The constant low proportion (around 30%) of nests in the shaded area throughout the 1986 season at the northern section was unique among all other data sets in both studies. However, four additional surveys over the central part of 18 miles of the 22 mile Tortuguero beach in 1986 season by the author (7 and 23 August, 5 and 17 September) showed similar spatial distributions around the middle range (respectively 55.4, 54.1, 45.7, 49.7% of nests in shaded area), and the proportions of the 18 miles were more similar to the data of the two miles of this study area than to those of the northern section. Thus, although spatial distribution of Tortuguero green turtles varies among both seasons and years, the extent of variation might be rather conservative on a population level.




60
The area available to nesting turtles in the vegetation/border zone (about 3 m wide) is much smaller relative to the nesting area in the open zone (up to about 30 m wide) at the Tortuguero beach. Therefore, the slightly higher or equal distribution of nests in the vegetation/border zone means that the vertical distribution of green turtle nests is biased toward the upper part of the vegetated beach. Similar preference of nest site selection for this species is reported in other regions. Bustard (1972) noted that in the Australian cays, green turtles show some tendency to nest close to substantial vegetation. Cornelius (1986) noted that green turtles in Pacific Costa Rica have a tendency to nest beneath the vegetation of the upper beach and rarely in the mid beach. In an island of Ogasawara Islands, Japan, 21 of 28 green turtle nests were deposited near or under the dense vegetation on the upper beach during June (Horikoshi, unpublished data). In Surinam, green turtles deposit more nests (63%) in the border and vegetation area than in the open area, whereas leatherbacks predominantly lay in the open area (87%) (Whitmore and Dutton, 1985). In Florida, green turtles nested in a higher location near the beach dune than did loggerhead turtles (Witherington, 1986). Whitmore and Dutton (1985) suspected that the interspecific difference may be due to interspecific competition: green turtles are pushed up to the upper beach because leatherback nests were deeper and the risk of destruction of nests of green turtles is higher than that of leatherbacks. In Tortuguero, leatherback turtles also use the same beach as green turtles, but their nesting season rarely overlaps with the green turtle's season. However, the nest distribution of leatherbacks is highly skewed to




61
the open area (Leslie et al., in press). Without co-ocurrence of the two species, the trend of nest site selection of each species at Tortuguero agreed with that at Surinam. Witherington (1986) suspected that the preference of green turtles to nest near dunes may be a strategy to avoid inundation although he could not detect significant differences of emergence success between the two species during the year.
Topography and vegetation of green turtle beaches show extreme variation: from low profile naked islands (e.g., French Frigate Shoals) to beaches backed by tropical jungle (e.g., Tortuguero and Sarawak). However, preference for the upper beach might be characteristic of green turtles. It would be interesting to investigate the geographic variation of nest site selection in relation to the thermal profile and egg survivorship factors on each beach.
Overall Reproduction Rate
To date, very few studies have assessed overall egg
survivorship of green turtle nests on natural beaches. Although several studies have investigated mortality factors on natural beaches, their sample seasons were rather limited or the analysis included only those nests that successfully produced some hatchlings (Mortimer, 1981; Balazs, 1980; Schulz, 1975). In Surinam, a detailed quantitative study by Whitmore and Dutton (1985) showed a relatively high hatching success of green turtle nests (mean 80.4% in 1981 and 1982). However, their samples were




62
collected only from nests above the spring high tide. Mrosovsky (1989) calculated the overall hatching success in Surinam to be 63.5% by adjusting the spatial distribution data. Recently, a longterm quantitative investigation has been conducted for green turtles on Florida beaches, although the beaches were more heavily utilized by loggerhead turtles than by green turtles. Mean emergence success of green turtle nests at the Melbourne beach varied from 54.6 to 75.2% (overall mean=65.6%) from 1985 through 1988 (Redfoot and Ehrhart, 1989). At a nearby beach, Horton (1989) did a similar study and reported 40.1% for mean emergence success for 20 green turtle nests deposited in 1988 and 1989.
Mean emergence success of sample nests of this study was 51.2% (1986), 49.1% (1988) and 66.9% (1989). By adjusting the temporal and spatial distribution of nests for each year, overall hatching success is estimated as 57.2% (1986), 45.7% (1988) and 66.6% (1989). Overall hatching success at Tortuguero under natural conditions falls within the range reported by the above quantitative studies in other parts of the world.
In the 1977 season when dog predation was serious, 42% of 350 monitored nests produced hatchlings and 83% of the eggs of those hatched nests emerged (Fowler, 1979). Thus, the 1977 season's overall hatching success is roughly estimated to be 34.8% (0.42 X 0.83). The success of green turtle nests at Tortuguero was apparently improved by the dog control program even though the natural predator, coatis, still depredate a substantial proportion of nests in some years, such as 1989.




63
Spatial Effects on Mortality Factors
For the three years of this study, no significant difference in emergence success of green turtle nests between the two zones was found. Both abiotic factors (surf, freshwater flooding, hurricane) and biotic factors (predation, turtle digging) affected egg survivorship in Tortuguero turtles, while the extent of damage from each mortality factor varies from year to year. Among the mortality factors, only mammal predation varied significantly between the zones. It is surprising that damage by surf was not very different between the zones; the open zone was 7.1% (egg loss for three year weighted mean), whereas the vegetation/border zone was 4.2%. This is probably because on several occasions heavy erosion cut the beach up to the vegetation area and all nests in the section were damaged, regardless of their zone.
Mammalian predation on green turtle nests was greater on the nests in the vegetation/border zone (14.1%) than on nests in the open zone (2.6%) throughout all three years of this study. This spatial difference between the zones most strongly influenced the overall reproduction rate in the 1989 season when the proportion of depredated nests was the highest among three years. After dogs were removed from the beach, coatis were responsible for most of the predation. Cornelius (1986) noted that coatis are the most common predators on olive ridley nests at Nancite, in the Pacific coast in Costa Rica, and that a large group of coatis living on the nesting beach feed almost exclusively on olive ridley eggs during




64
much of the year. Although raccoons, a common predator on loggerhead nests in the southeastern United States (e.g., Stancyk et al., 1980), also exist at Tortuguero, their role as a predator on green turtle nests seems to be minimal. However, a spatial bias by predatory raccoons was observed on a Florida beach. Horton (1989) found that raccoon predation on loggerhead nests on a Florida beach was negatively related to the distance of the nests from the dune's edge in one of his two study seasons. He hypothesized that predators search for prey close to the safety of cover, namely near the dune in Florida. At Tortuguero, locals occasionally observed jaguars, Felis onca, in the protected section of the beach during the green turtle nesting season. I occasionally found unidentified foot prints of large cats within the study area. Feral dogs were once abundant throughout the 22-mile Tortuguero beach (Fowler, 1979). Dogs may be a potential threat to coatis, especially young individuals. Spatially biased depredation of nests by coatis may be partially a result of avoiding the risk from their predators.
Of the abiotic mortality factors, sporadic freshwater flooding was the most consistent factor at the Tortuguero beach throughout this study, but probably it is very unusual in sea turtle rookeries elsewhere. A single excessive rain event can force the level of ground water near or over the depth of the egg mass of green turtles and suffocate eggs and emerging hatchlings. Even if the eggs are not soaked in water, the nearby water might hinder gas exchange by saturating sand space by capillary action. Similar flooding phenomena were reported on a loggerhead beach on the Georgia coast (Ragotskie, 1959; Kraemer and Bell, 1984). Ragotskie (1959)




65
suggested that because of the higher probability of excessive rain late in the incubation season (September) on the Georgia coast, successful emergence from nests deposited after the middle of the nesting season (1 July) decreases. Kraemer and Bell (1984) found that the temporal distribution of loggerhead nests did not avoid the excessive rainfall events in September. At Tortuguero, the probability of excessive rain events (>140 mm/day) also increased during the latter part of the incubation season (October and November) (Table 13). However, most nests deposited during the peak nesting period of Tortuguero (August through September) are very much affected if flooding occurs in October and November. Tortuguero green turtles have not adjusted their nesting season relative to excessive rainfall events. Kraemer and Bell (1984) proposed a hypothesis that nesting by loggerhead turtles near the beach dune at the upper beach may be an adaptive response to escape the excessive rains. This is a very tempting hypothesis to explain partially the nesting preference for the upper vegetated beach by Tortuguero green turtles. However, the difference in damage by flooding between the two zones was inconsistent among the study years, and the extent of the differences was rather small.
Seasonal Fluctuation of Emergence Success
It is not clear to what extent emergence success of green
turtle nests significantly fluctuates within a season at Tortuguero. During the three year study, a significant difference was found only in 1988. The main cause of those fluctuations was the two




66
inundation events in close succession in October from excessive rain and Hurricane Joan. However, hurricanes on the Caribbean coast of Costa Rica are relatively rare. Coen (1983) noted that Hurricane Martha, 21-25 November 1969, was the only one to hit the coast during the hundred years of records. Therefore, it would appear that the influence of hurricanes on the Tortuguero green turtle colony would be negligible, and that the 1988 season was an atypical year. The extent of damage by Hurricane Joan was not accurately assessed because of the overlapping damage by the excessive rain on 6 October. Therefore it is unknown whether the seasonal fluctuation in emergence success in the 1988 season would have occurred without Hurricane Joan. Separate flooding events in August 1986 and in October 1989 affected the sample nests, but these events did not result in drastic seasonal changes in emergence success within each season.
This lack of seasonal variation can be partially explained by the observation that all stages of embryos and emerging hatchlings are susceptible to flooding. One flooding event can affect nests deposited over long periods, possibly up to four months, if the flooding occurs in the middle of the incubation season. The same principle can apply to the damage by surf, and even for predation, because mammal predators destroyed the nests at various stages. Thus, favorable weather conditions in September (e.g., calm seas, less chance of excessive rain) can have a positive effect on emergence success of nests deposited over the long period of time from July through September, which covers the majority of the nesting season. In 1977, Fowler (1979) analyzed emergence




67
success of the nests that produced hatchlings, and she found no seasonal differences in the emergence success at Tortuguero. Fowler did not report any flooding events. Seasonal fluctuation in egg survivorship of Tortuguero green turtle nests might not be apparent, except in years of hurricanes and excessive seasonal flooding.
Spatial Effects on Sex Ratio
At the Tortuguero, beach, a distinct thermal difference at the depth of green turtle nests between the open zone and the vegetation/border zone was very consistent throughout the incubation season in the three years of this study. However, the effect of these thermal zones on sex ratio depends on the relation of sand temperatures in each zone to the pivotal temperature. It appears that this spatial effect on sex ratio can be detected only during a period of dry weather. Seasonal change in this spatial effect was clearly observed in the 1988 season. The mechanism may be the following. During prolonged dry weather, sand temperatures in the open zone clearly exceed the pivotal temperature, whereas sand temperatures in the vegetation/border only approach or slightly exceed the pivotal temperature. Consequently, most nests in the open zone show female-biased sex ratios, whereas nearly equal or moderately female-biased sex ratios occur in the vegetation/border zone. On the other hand, during prolonged wet weather, sand temperatures in both zones decrease below the pivotal temperature. As a result, the difference in sex ratios between zones becomes




68
negligible, and most nests in both zones show male-biased sex ratios. The latter case was observed throughout the 1986 season.
In 1980, Spotila et al. (1987) found a distinct spatial
difference in sex ratios of Tortuguero green turtle nests between the open zone (mean 67.4% of females, n=9) and the vegetation zone (7.6% of females, n=6). The rainfall record revealed that the weather was very dry (281 mm in August, 313 mm in September) throughout the critical periods of their samples.
In Surinam, the different topographic beach zones showed small thermal differences (0.5-1.00C), and the location had little effect on sex ratio of green turtles through the season (Mrosovsky et al., 1984). This is probably due to the fact that the vegetation is composed of sparse and low shrubs in Surinam, compared to Tortuguero's dense vegetation (Schulz, 1975; P. Dutton, pers. comm.). The existence of dense vegetation and associated shade is one requirement for a spatial effect on sex ratios. The barrier island beaches on the east coast of Florida have primary dunes at the upper part of the beach, but lack dense vegetation. Witherington (1986) noted that the thermal differences on the beach must be slight because there is no significant difference in incubation periods between beach zones for loggerhead nests. The nesting beaches of Ascension Island are virtually without vegetation (Mortimer, 1981), but differences in the mineral composition of the various beachs (Mortimer, 1990) produce different thermal regimes (Hays and Mortimer, MS in prep.).
At some island rookeries, a significant effect of the
orientation of the coast on sex ratio has been reported. At Heron




69
Island, Australia, Limpus et al. (1983) observed that the north facing beach on the island was warmer than the beach that faces south. Consequently, the northern beach produced a significantly higher proportion of green turtle females (63.1% females) than the cooler southern beach (29.5% females). In Barbados, West Indies, the warmer west coast beach produced more hawksbill females (80.6% females), whereas the cooler south coast produced more males (32.7% females) (Horrocks and Scott, 1991).
Temporal Effects on Sex Ratio
At Tortuguero, a short dry season in September (mean 320 mm) usually occurs between the wetter periods of July through August, and October through November. For the Tortuguero green turtles, the impact of this dry season seems to be very important because a large proportion of the nests are deposited during August and the most critical period for these nests occurs in September. A dry September produces enough females to avoid a highly skewed sex ratio for the entire season.
Seasonal fluctuation in sex ratios resulting from the shift
between rainy and dry periods was observed at two tropical beaches: Surinam for green turtles and leatherback turtles (Mrosovsky et al., 1984) and French Guiana for leatherback turtles (Rimblot-Baly et al., 1987). In Sarawak, another tropical beach, incubation periods of green turtle nests varied with the rainy dry weather cycle (Hendrickson, 1958). Therefore, Standora and Spotila (1985) suspected that the sex ratios of green turtles at Sarawak also show




70
seasonal fluctuation. However, the pattern of fluctuation is very different from rookery to rookery. At Tortuguero and Sarawak, the dry season occurs somewhere in the middle of the nesting season, whereas the wet season occurs near the middle of the nesting season in Surinam. In French Guiana, the wet season occurs early in the nesting season. Regardless of the pattern, these sea turtle rookeries probably produce variable sex ratios throughout the season.
Annual Variation in Primaryt Sex Ratio
In the 1986 season, the sand temperature in both zones
remained below the pivotal temperature for most of the season. In the open zone, although the sand temperature in September and later intermittently rose to the level of the pivotal temperature, these periods were too short to produce many females. Therefore, in spite of the small sample size, the 10.1% proportion of females from the samples that were sexed directly is probably a good estimate. The weather pattern is one way to assess whether the year was a typical one for Tortuguero. The rainfall in 1986 in August (846 mm) and in September (514 mm) was the highest on record, and the total amount of rainfall throughout the incubation season (3871 mm, from 16 June through 10 December) was the second highest for the last 12 years. Thus, it appears that the weather in the 1986 season was atypically wet. Therefore the highly male-biased sex ratio is probably atypical for the Tortuguero population.




71
In 1988, the total amount of rainfall throughout the incubation period was very close to the mean value of the last 12 years, but the 1988 season had much higher seasonal fluctuation in rainfall than the average of the last 12 years. Rainfall in August 1988 (256 mm) was the lowest on record whereas rainfall in October (790 mm) was the second highest since 1978. Above all, Hurricane Joan greatly influenced the survivorship and sex ratio of the nests in the latter half of the 1988 season. The weather in 1988 was also atypical for the Tortuguero beach. If Hurricane Joan had not occurred in 1988, the proportion of females estimated for the primary sex ratio of the 1988 season (40.6%) probably would have been higher.
Predictability of Sex Ratio by Environmental Factors
Mrosovsky et al. (1984) cautioned that accurate estimation of the primary sex ratio of sea turtles from a combination of data on the pivotal temperature from laboratory experiments and thermal profiles on the beach is impossible because of the many thermal influences and their possible interactions throughout the season. Some of the critical factors are metabolic heat, exact pivotal levels and critical periods, nest depth, and spatial factors. This study's direct thermal measurement in individual nests is the most accurate measure available to obtain the thermal condition of the nest, and the thermal data included the variation from metabolic heat, nest depth and spatial factors. However, even the thermal data in the nest were not sufficient to predict the sex ratio of Tortuguero green turtles.




72
The high variability of sex ratio in relation to incubation temperature was not surprising. First, even under constant temperature in the laboratory, high inter-clutch variation in sex ratios was found between two loggerhead clutches from the same Florida beach (Mrosovsky, 1988). Limpus et al. (1985) reported a similar high variation in sex ratio at the middle range of temperature among three loggerhead clutches sampled from Mon Repos, Australia, at constant temperature conditions. However, he did not find such large variation among the loggerhead clutches from Heron Island. Mrosovsky and Pieau (1991) hypothesized that the response of gonad differentiation to the incubation temperature might be different depending on the genotype of the eggs, and that genetic factors can override the temperature effect at the middle temperature range. This theory was inferred from the experimental results in a freshwater turtle, Emys orbicularis, in which the expression of H-Y antigen was strongly correlated with the expressed sex of gonads at the middle range of temperature where both sexes were produced (Zaborski et al., 1988). Mean nest temperatures during the middle third of development in 1986 and 1988 at Tortuguero fell within a narrow range (27.0-30.70C) compared to the much wider temperature range over which live hatchlings of this species can be produced (25-33C) (Miller, 1985). Both sexes were produced throughout the observed range at Tortuguero. Thus, the high variation in sex ratio over the observed range can be explained if the mechanism of genetic influence operates in green turtle eggs.




73
Second, because of seasonal fluctuation in sand temperatures and the general increase in metabolic heat throughout incubation, individual nest temperatures on the beach were rarely constant throughout the critical period. The extent and the pattern of variance differed greatly among nests. Standora et al. (1982) reported spatial variation in sex ratio within natural green turtle nests incubated at the pivotal temperature. Due to metabolic heating, eggs near the center of the clutch produced females, whereas those at the periphery produced males. In the present study, I measured nest temperatures at the center of each clutch, but determined the sex of eggs collected randomly throughout the clutch. Therefore, my nest temperature readings may not have accurately represented the incubation temperature of all sampled eggs within a nest. Had I collected the sample eggs from only the center mass of each clutch near the temperature probe, the variation in sex ratio relative to the nest temperatures might have been smaller. Furthermore, in this study the critical period was calculated based on an assumption that the time from pipping to emergence was five days for all sample nests. Hendrickson (1958) indicated that heavy rains, which pack the upper layers of beach sand, might prolong the emerging period of green turtle hatchlings in Sarawak. Because Tortuguero had high seasonal fluctuation in rainfall, the emergence period of green turtles might also vary to some extent. Thus the calculated critical period of each nest might not be accurate, and the mean nest temperature calculated for each nest may only roughly represent its thermal environment.




74
Miller (1985) emphasized that temperature, hydric
environment, and gas exchange interact cooperatively to influence embryonic metabolism. Standora and Spotila (1985) suggested that while temperature is the primary environmental factor that determines sex in sea turtles, other factors, such as osmotic stress, and 02 and C02 levels, could play a role in sex determination in the temperature range over which both sexes are produced. The impact of those other factors, however, has yet to be demonstrated. The influence of hydric conditions on sex determination in painted turtles, Chrysemys picta, was inconsistent even within one population (Gutzke and Paukstis, 1983; Packard et al., 1991). Oxygen concentration did not influence the sex ratio of red-eared slider turtles, Trachemys scripta (Etchberger et al., 1991). It is highly probable however that any environmental influence and their possible interactions in determining sexes of green turtle hatchlings vary at the inter-clutch level because of the highly heterogeneous environment over time at the Tortuguero beach.
The previous study of the relationship between sex ratio and mean incubation temperature of natural nests in the 1980 season at Tortuguero (Spotila et al., 1987) showed less variability than did this study. This is probably due to their smaller sample size (15 nests) and also their shorter sampling period. Most of their samples were collected within a half month period, whereas my sample period extended over three months in each of two years. It is possible that their mean nest temperature values were better correlated with sex ratio than were mine because most of their




75
nests were exposed to less environmental fluctuation than were those in my study.
The first multiple regression model (pseudo R2=0.24) with the significant factors selected (mean nest temperatures, incubation period, rainfall data, and zone) to predict sex ratio was only a slight improvement over the nest temperature model (pseudo R2=0.20) because rainfall data, incubation period and zonation are correlated with nest temperature data. That is, most factors only provided redundant information to various degrees. As I discussed above, the genetic factor might also be responsible for the poor fit of the data to the model.
It should be noted that another multiple regression model (pseudo R2=0.23) using rainfall data, zone and their interaction explained the sex ratios of hatchlings as well as the nest temperature model did. Rainfall fluctuation seemed to more strongly influence the sex ratio of nests in the open zone than in the vegetation/border zone. In the open zone, sex ratios ranged from 100% male to 100% female, whereas in the vegetation/border zone the sex ratios of most nests showed less variation, ranging only from 100% male to slightly female biased. My study suggests that data describing rainfall and zonation could be a rough predictor of sex ratio of green turtle hatchlings at Tortuguero. Furthermore, it might be possible to estimate very roughly the primary sex ratio using the total amount of rainfall through the entire incubation period. To construct a useful model, more data need to be gathered on the correlation between rainfall and sex ratio produced in each zone.




76
Mrosovsky et al. (1984) found as great a variability in the
relationship between sex ratios and incubation periods for Surinam green turtles as I did for Tortuguero green turtles. They suggested that if the average of a large number of samples was used, it might be possible to make a reasonable prediction of primary sex ratio from the above model. However, collecting a large sample in each thermal zone in each seasonal segment is very labor intensive. In Tortuguero, where thermal zones have a significant effect on sex ratio, the number of samples needed is doubled, and the reliability of the estimate of primary sex ratio is reduced by adding another variable.
In conclusion, due to the high level of variation, the sex ratio of green turtle nests at Tortuguero cannot yet be predicted with accuracy from the nest temperature and other environmental factors, such as sand temperature, nest depth, incubation period, rainfall, and nesting zone. In the future, long term accumulation of data correlating rainfall and sex ratios might be used to roughly estimate primary sex ratio at Tortuguero.
Currently, direct sexing of periodic subsamples is the only
accurate method to investigate primary sex ratio on a natural beach. Moreover, an advantage of direct sexing is that any variation resulting from genetic influences will be incorporated into the sex ratio derived. Unfortunately, immature sea turtles are not sexually dimorphic. Because they have homomorphic sex chromosomes, reliable sexing of sea turtle hatchlings is only possible by sacrificing turtles and performing a time-consuming histological examination of their gonads (Jackson et al., 1988). Nonsacrificial




77
methods including laparoscopy and testosterone levels (Wibbels et al., 1989) are useful in sexing larger immature sea turtles, but not hatchlings. Recently, Demas and Wachtel (1989) reported a sexspecific satellite DNA, called "Bkm", in green turtles and Kemp's ridleys. Because only a few drops of blood are needed to detect this molecule, it could be used to harmlessly determine the sex of sea turtle hatchlings. At the present time, however, this molecular method is probably too expensive and labor-intensive to process the large numbers of hatchlings needed to estimate the primary sex ratio on a beach where many variables affect sex ratio, such as on the Tortuguero beach.




CHAPTER 5
SUMMARY AND CONCLUSIONS
The major nesting activity of green turtles occurred from July through the first half of October at Tortuguero during the three-year study, although the frequency of seasonal distribution shifted among years. Nest density also varied among years. The nesting densities recorded in 1986 (7.3 nests/rn) and in 1988 (6.5 nests/rn) from July through November were two of the highest known for this species worldwide.
Tortuguero beach was divided into two zones. The
vegetation/border zone lies within 2 m of the border of dense vegetation, and the open zone lies between the vegetation/border zone and the shore line. Nest distribution between the zones varied significantly yearly and seasonally, and there was no common seasonal trend of fluctuation among the years. The vertical distribution of green turtle nests was biased toward the upper part of the vegetated beach throughout each season. A preference for nest sites on the upper beach may be rather conservative for this species.
Throughout this study, a distinct thermal difference at the
depth of green turtle nests between the two zones was consistent, with nests in the open zone being warmer than those in the vegetation/border zone. The differences in overall mean sand
78




79
temperature at 60 cm depth were 0.80C in 1986, 1.1 OC in 1988, and
1.50C in 1989. This is due to the dense vegetation and associated shade at the upper part of the beach. However, the spatial influence of distribution on sex ratio probably occurs only during a dry period. During the middle part of the 1988 season, under prolonged dry weather conditions, sand temperatures in the open zone clearly exceeded the pivotal temperature, whereas sand temperatures in the vegetation/border zone only slightly exceeded the pivotal temperature. Consequently, most nests in the open zone produced female-biased sex ratios, whereas nearly equal or moderately female-biased sex ratios occurred in the vegetation/border zone. On the other hand, during prolonged wet weather, sand temperatures in both zones decreased below the pivotal temperatures, causing malebiased sex ratio in both zones. This was observed in the latter part of the 1988 season and throughout the 1986 season.
Tortuguero, with 5400 mm of rain each year, is one of the wettest sea turtle rookeries in the world. A short dry period in September occurs between the wet months of July through August and October through November during the major incubation season of green turtle nests. This dry season seems to be very important in determining the primary sex ratio of the Tortuguero population. Because a large proportion of the nests is deposited during August and the most critical period for sex determination in these nests occurs in September, a dry September can produce enough females, particularly in the open zone, to avoid a highly male-biased sex ratio for the season.




80
Overall hatchling emergence success is estimated as 57.2% in 1986, 45.7% in 1988, and 66.6% in 1989. No significant difference in emergence success between the two zones was found in any year. A significant seasonal difference in emergence success was found only in 1988, when a rare hurricane hit the Tortuguero beach. Thus, egg survivorship probably does not have a great effect on the primary sex ratio produced on the Tortuguero beach. Both abiotic (surf, freshwater inundation, hurricane) and biotic factors (predation, turtle digging) were responsible for reducing the survivorship of green turtle nests. The relative importance of each mortality factor varied from year to year.
Freshwater inundation when ground water levels were raised by excessive rainfall, was the most consistent mortality factor of green turtle nests throughout this study. The probabilities of excessive rain events (>140 mm/day) were highest in the latter part of the incubation season (October and November). Thus, most nests deposited during the peak nesting period (August through September) are affected if the flooding occurs in October and November. Tortuguero green turtles have not adjusted their nesting season relative to excessive rainfall events.
Mammalian predation was the only significant mortality
factor that showed a distinct spatial difference between the zones. Mammalian predation, mostly by coatis, was greater on the nests in the vegetation/border zone than on nests in the open zone throughout each year. The extent of predation was most severe in the 1989 season when the nest density was about half that in the 1986 and 1988 seasons.




81
To calculate the primary sex ratio in the 1988 season,
temporal and spatial nest distribution, emergence success, and sex ratio data were combined by half month intervals. The overall proportion of females was estimated to be 40.6% in the 1988 season. During 1986, because no significant seasonal effect was detected, the overall proportion of females was calculated to be 10.1% from the mean value of sample nests.
The relationship between sex ratio and mean nest temperature during the middle third of development, mean sand temperature, incubation period, three sets of rainfall records, depth of clutch, and nesting zone were analyzed by logistic models. The pivotal temperatures were calculated as 29.40C for mean nest temperature, and 28.50C for mean sand temperature. The primary environmental factor in determining sex of green turtles--mean nest temperature-had the best fit in the single regression model. However, the variability of the model was high (pseudo R2=0.20). Mean nest temperature, incubation period, rainfall from egg deposition through the middle third of development, and nesting zone were significant parameters in the multiple regression analysis (pseudo R2=0.24). Another multiple regression model with mean daily rainfall from the time of egg deposition through the middle third of development, nesting zone, and their interaction (pseudo R2=0.23) was found to be as good as the mean nest temperature model. This rainfall/zone model might have value as a conservation tool to estimate roughly the sex ratio of green turtle nests. In conclusion, the variability of sex ratio by all of the above models was still very high for predictive




82
purposes. Genetic influence and heterogenous environmental conditions are probably partially responsible for such high variability.
To date, we have little knowledge of sex ratio in sea turtle populations and its dynamics. Because the character of sexual dimorphism, tail elongation, appears just prior to sexual maturity of individuals, only gonadal examination or testosterone level can identify the sex of immature turtles. Thus, previously reported sex ratio data based on external morphology (e.g, Ross, 1984) are probably not reliable. Levels of testosterone have been used to examine the sex ratios of immature green turtles in several feeding grounds (Wibbels et al, 1989; Meylan et al., 1992; Bolten et al., in press). However the definition of each population at the feeding grounds is not known.
The primary sex ratio data obtained in this study for
Tortuguero is the first reliable information on this subject and also is the first stage of the determination of the sex ratio of the entire population. However, we need to be cautious in assessing the primary sex ratio of the Tortuguero population. The density of nests, and thus the number of hatchlings produced each year, varies widely at Tortuguero. Because the sex ratio shows yearly variation as this study has demonstrated, the primary sex ratio of the Tortuguero green turtle population can only be calculated from the product of the number of hatchlings and the sex ratio each year for many years. To achieve this task, periodic collection of samples for sexing for many years is still essential although the procedure is labor-intensive.




83
Table 1. Ranges of daily fluctuation of sand and green turtle nest temperatures during 24 hours at Tortuguero, Costa Rica.
Mean SE Range No. Sample
(oc) (oc)
Sand temperatures on transects Open Zone
60cm depth 0.49 0.03 0.3-0.8 n=20
80cm depth 0.49 0.04 0.2-0.8 n=20
Vegetation/border Zone
60cm depth 0.48 0.03 0.4-0.6 n=8
80cm depth 0.41 0.07 0.1-0.8 n=8
Temperatures at the center of clutches Open Zone 0.46 0.04 0.2-0.8 n=14
Vegetation/border Zone 0.49 0.04 0.2-0.8 n=14 Samples were collected and pooled from four separate days on 13 August and 29 September 1988 and 12 August and 14 October 1989.




84
Table 2. Overall mean sand temperatures (0C SE) throughout the season on transects from 1 July through 10 December in 1986, 1988 and 1989 at the Tortuguero beach.
1986 1988 1989
n=76 n=73 n=76
Open Zone
60cm Depth 27.60.1 28.80.1 28.60.1
80cm 28.50.1 28.20.1
Vegetation/border Zone
60cm Depth 26.80.1 27.70.1 27.10.1
80cm 27.40.1 26.90.1




85
Table 3. Multiple comparison table of the monthly sand temperatures at a depth of 60cm in the open zone and in the vegetation/border zone in 1986, 1988 and 1989 (Fisher's procedure of least significant difference, alpha=0.05).
1986
3 2 111
Open Zone JUL AUG SEP OCT NOV
3 2 1 1 1
Vegetation! JUL AUG SEP OCT NOV
border Zone
1988
2 2 1 3 3
Open Zone JUL AUG SEP OCT NOV
2 2 1 3 3
Vegetation! JUL AUG SEP OCT NOV
border Zone
3 2 1 1,2 3
Open Zone JUL AUG SEP OCT NOV
2 1 1 1 3
Vegetation! JUL AUG SEP OCT NOV
border Zone
Those months that do not share a common number have significantly different mean temperatures. The numbers are ranked from highest to lowest monthly mean temperatures.




86
Table 4. Effects of seasonal period and nest location in the open and the vegetation/border zones on egg survivorship at the Tortuguero beach in 1986, 1988 and 1989. Analysis is by nonparametric twofactor ANOVA with unbalanced factorial design (Zar, 1984).
Source SS d f H p
1986
Period 1116.91 3 201.19 0.10 Zone 45.95 1 5.55 0.50 Period*Zone 1104.67 3 5.49 0.10 Total 9657.00 48
1988
Period 11568.96 5 18.32 **0.001 Zone 1812.10 1 2.87 0.05 Period*Zone 2800.01 5 4.43 0.25 Total 54954.50 87
1989
Period 10905.85 5 10.24 0.05 Zone 3079.23 1 2.89 0.05 Period*Zone 2781.27 5 2.61 0.75 Total 119296.96 112
H=Source SS/Total MS, Total MS=Total SS/df




87
Table 5. Descriptive fate of green turtle eggs from the sample nests
at the Tortuguero beach during the 1986, 1988 and 1989 seasons.
1986 1988 1989
Egg Fate Open Vegetation Open Vegetation Open Vegetation
border border border
Unhatched eggs
Destroyed by mammals 79 19 0 606 296 1920
Destroyed by ghost crabs 0 109 93 0 0 0
Destroyed by termits 0 3 2 0 1 20
Destroyed by ants? 30 6 21 7 67 13
Destroyed by female turtles 0 256 124 296 0 0
Destroyed by plant roots 0 0 0 0 0 3
Destroyed by unknown cause 0 0 49 97 0 0
Washed out by surf 445 244 0 0 107 0
Embryonic death by surf 3 7 111 112 0 0 0
Embryonic death by flooding 396 498 118 623 481 221
Embryonic death by Hurricane Joan 0 0 61 482 0 0
Embryonic death by flooding or 0 0 299 143 0 0
Hurricane Joan
Embryonic death with no 84 86 174 223 202 140
apparent physical disturbance
No apparent development 57 36 224 168 133 92
Rotten intact or ruptured eggs 184 125 294 359 118 133
Death at pipping 0 0 0 6 2 2
Total dead eggs 1312 1493 1571 3010 1407 2544
Unemerged Hatchlings
Hatchlings in egg chamber 1 2 7 9 18 8 1 8
Hatchlings died above egg 0 0 0 97 20 1 5
chamber by flooding
Hatchlings died above egg 0 0 0 134 0 0
chamber by Hurricane Joan
Hatchlings tangled with rope 0 0 0 0 10 0
above egg chamber
Hatchlings depredated by mammals 0 0 0 0 0 1
Total hatchlings unemerged 1 2 7 9 249 38 34
Emerged Hatchlings 1591 1301 2319 2456 4445 3638
Total eggs 2915 2801 3899 5715 5890 6216




88
Table 6. Characteristics of green turtle sample nests for sex ratio analysis. The data were collected at Tortuguero, Costa Rica, during the 1986 and 1988 seasons.
Deposition date Zonea Sample sizeb % Female
18 July 1986 V/B 16 68.8
18 July V/B 19 0
21 July 0 2c 0
28 July V/B 20 0
14 Aug. 0 20 0
21 Aug. 0 17 47.1
10 Sept. 0 19 0
10 Sept. V/B 20 0
16 Sept. 0 20 0
25 Sept. 0 20 0
28 Sept. 0 1 9 5.3
3 Oct. 0 20 0
3 July 1988 V/B 13 15.4
5 July 0 20 55.0
6 July 0 19 63.2
9 July 0 20 10.0
13 July V/B 20 30.0
14 July V/B 19 10.5
15 July 0 19 73.7
16 July V/B 20 55.0
21 July V/B 19 21.1
23 July 0 20 90.0
23 July 0 14 78.6
27 July 0 20 100.0
28 July 0 20 100.0
29 July V/B 14 7.1
30 July V/B 19 10.5
3 Aug. V/B 12 16.7
3 Aug. V/B 19 (1) 89.5
5 Aug. V/B 18 22.2
6 Aug. 0 15 100.0
7 Aug. 0 16 25.0
8 Aug 0 18 (1) 61.1




89
Table 6--continued.
Date deposited Zonea Sample sizeb % Female
8 Aug. 1988 O 20 100.0
11 Aug. O 16 100.0
15 Aug. O 12 66.7
17 Aug. O 20 60.0
17 Aug. O 20 5.0
26 Aug. V/B 20 55.0
27 Aug. O 13 (1) 15.4
28 Aug. V/B 17 47.1
30 Aug. V/B 16 (2) 56.3
30 Aug. O 20 (1) 80.0
31 Aug. V/B 19 0.0
3 Sept. V/B 18 44.4
8 Sept. O 1 8 27.8
9 Sept. V/B 18 0.0
14 Sept. O 17 0.0
15 Sept. O 20 60.0
19 Sept. O 20 (1) 25.0
20 Sept. O 18 33.3
22 Sept. V/B 17 23.5
24 Sept. O 19 0.0
27 Sept. O 1 5 0.0
27 Sept. V/B 20 0.0
a: V/B-the vegetation/border zone, O-the open zone. b: values in parentheses are the number of turtles in which gonads showed both seminiferous tubles and developed cortex. c: missed emergence; two hatchlings remained in the nest.




90
Table 7. Effects of seasonal period and nest location in the open and the vegetation/border zones on sex ratios of green turtle hatchlings at the Tortuguero beach in 1988. Analysis was by two-factor ANOVA with unbalanced design. The ratios of females were transformed (p'=1/2(arcsin/(X/(n+1l)+arcsin/(X+1)/(n+1 ))(Zar,1984). Fmax test showed homogeneity of variances.
So u rce SS df MS F p
Period 1.7385 5 0.3477 3.30 0.017
Zone 0.7695 1 0.7695 7.30 0.011
Period*Zone 0.8093 5 0.1619 1.54 0.208 ns
Total 3.2685 31 0.1054




Table 8. Estimation of overall sex ratio of green turtle hatchlings at Tortuguero, Costa Rica, in the 1988 season.
Period #Nest #Nest %Egg Female% Female% #Hatchling #Hatchling #Female #Female Survivorship (x109.1) (x109.1) (x109.1) (x109.1)
Open V/B Pooled Open V/B Open V/B Open V/B
July 1-15 294 489 48.9 50.5 18.6 144 239 73 45
July 16-31 725 766 79.7 92.1 23.4 578 610 532 143
Aug. 1-15 771 885 57.2 75.5 42.8 441 506 333 217
Aug. 16-31 746 1494 16.9 40.1 39.6 126 252 50 100
Sep. 1-15 1182 1566 45.4 29.3 22.2 536 710 157 158
Sep. 16-30 498 645 40.0 14.6 11.8 199 258 29 30
2024 2576 1174 692
Total hatchlings 4600 1866
Overall female ratio 40.6%
Open: open zone. V/B: vegetation/border zone.
109.1: mean clutch size in 1988.
Number of hatchlings = Number of nests x mean clutch size (109.1) x % Egg Survivorship. Number of females = Number of hatchlings x % Female.




92
Table 9. Logistic regression models of fluctuation in sex ratio for green turtle hatchlings with single independent variables. Data collected at Tortuguero, Costa Rica, 1986 and 1988.
Independant Variable pseudo R2 Slope Intercept
Mean nest temperature 0.202 1.1709 -34.4594 Mean sand temperature* 0.172 1.0636 -30.3740 Incubation period 0.177 -0.2630 15.4679
Rainfall (R-1) 0.079 -0.0954 0.5907
Rainfall (R-11) 0.114 -0.1290 1.0477
Rainfall (R-Ill) 0.171 -0.1912 1.8052
Bottom depth of nests 0.010 0.0250 -2.5171
All model, df=1, p<0.001
*n=728, else n=924




93
Table 10. Logistic regression model of fluctuation in sex ratio for green turtle hatchlings with all independent variables. Final model follows in Table 11. Data collected on Tortuguero, Costa Rica, 1986 and 1988.
Parameters Slope (SE) ChiSqure P
Mean nest temperature 0.5005 (0.1314) 14.50 0.0001 Incubation day -0.0801 (0.0330) 5.94 0.0149
Rainfall (R-1) -0.0136 (0.0234) 0.34 0.5607 ns
Rainfall (R-11) 0.0510 (0.0454) 1.26 0.2617 ns Rainfall (R-Ill) -0.1359 (0.0585) 5.40 0.0202 Bottom depth of nest 0.0097 (0.0090) 1.16 0.2815 ns Zone 0.4701 (0.1958) 5.76 0.0164
Intercept -9.9764 (4.8321) 4.26 0.0390
-2LogLikelihood=292.3, df=7, overall P<0.001 pseudo R2=0.241, n=924




Full Text

PAGE 1

123 VEGETATION OPENZONE BORDER ZONE 16.0% ,o 4.2% 1986 .0 4.2% 76.0% 75.0% N=25 N=24 2.7% 2.7%25% 1988 78.4% 60.8% 39% N=37 N=51 1.9% 3.8% 1.6% 3.8%11 1989 N= 53 N=60 KILLED BY SURF E DEPREDATED BY MAMMALS 0 KILLED BY FLOODING 0 DEPREDATED BY GHOST CRABS O KILLED BY FLOODING OR HURRICANE 0 EXCAVATED BY TURTLES O NEST PRODUCED SOME HATCHUNGS



PAGE 1

ACKNOWLEDGEMENTS First I thank my parents Shigeru and Kiku Horikoshi, and my wife Harumi Horikoshi, whose love and encouragement strengthened my resolve to complete this task. I would like to thank the members of my committee: Dr. Martha Crump, Dr. Karen Bjorndal, Dr. Jack Kaufmann, Dr. Clay Montague and Dr. Jeanne Mortimer who offered continuous encouragement and advice throughout this study. I feel fortunate to have been able to work under their guidance. I also would like to thank the late Dr. Archie Carr who offered invaluable encouragement and an opportunity to undertake this study at the Tortuguero beach. Dr. Takako Oshima provided advice on statistics. I am very grateful to Dr. Louis Guillette who was most generous in allowing me use of his histology laboratory. Dr. Alan Bolten and Dr. Blair Witherington offered much appreciated advice by sharing their knowledge of sea turtles. My field work in Costa Rica greatly depended on the aid of Elver and Elvin Gutierrez, Robert Carlson, Harumi Horikoshi, and Carlos Diez. Stephen Morreale kindly recorded some weather data in 1986. I would like to thank the Caribbean Conservation Corporation, Sigma Xi, and the Archie Carr Center for Sea Turtle Research, the Center for Latin American Studies, and the Department of Zoology, University of Florida for partial funding for this project. I would ii



PAGE 1

15 placed around the nest; these logs were placed in such a way as to avoid creating shade on the sand surface above the egg chamber. The temperatures of nests and the sites 1 m away from the nests were monitored once a day at an interval of two to four days in 1986 and at an interval of two days in 1988 and 1989. During the 1986 and 1988 seasons, I collected samples of eggs to be sexed; during 1989, 1 monitored temperatures of the nests, but did not collect any eggs. After 50 -60 days of incubation, I uncovered the nests and randomly selected 20 developing eggs. The distance from the bottom of the egg chamber to the sand surface was measured. The numbers of developing eggs and dead eggs were counted, and dead eggs were opened to check for any sign of development. Remaining developing eggs were reburied in the egg chamber, and eventually the day of emergence was recorded. In 1986, 24 nests were equipped with thermocouples, but subsamples of embryos to be sexed were collected from only 12 of these nests. In 1988, 92 nests were equipped with thermocouples, and sub-samples were collected from 42 of these nests. Loss of the sample nests was due to destruction by nesting females, drowning of whole clutches by flooding or hurricane, and depredation by animals. Nest temperature data were obtained from all the nests intended for future sexing of the hatchlings, except one nest in 1986. In 1989, 74 nests were set up only for temperature monitoring; nest temperature data were collected from 69 of these nests. Sand temperature data near each sample nest were unobtainable for some of the nest temperature monitored nests in all years: 3 nests in 1986, 8 nests in 1988, and 13 nests in 1989.



PAGE 1

Figure 36. Sex ratio of green turtle hatchlings as a function of amount of rainfall. Data were collected at Tortuguero, Costa Rica, in the 1986 and 1988 seasons. In Figure 36A, the X axis represents mean daily rainfall during the middle third of development. In Figure 36B, the X axis shows mean daily rainfall during the middle third of development plus the previous 10 days. In Figure 36C, the X axis represents mean daily rainfall from the deposition of the clutch through the middle third of development. The superimposed line shows logistic regression models to fit all plots. In Figure 36A, Y = exp(0.59-0.10X)/(1+ exp(0.59-0.10X)), pseudo R2 = 0.08. In Figure 36B, Y = exp(1.05-0.13X)/(1+ exp(1.05-0.13X)), pseudo R2 = 0.11. In Figure 36C, Y = exp(1.81-0.19X)/(1+ exp(1.81-0.19X)), pseudo R2 = 0.17. Closed circles represent the nests deposited in the open zone, open circles represent ones in the vegetation/border zone. The small numbers indicate the numbers of overplotted data points in the graph.



PAGE 1

63 Spatial Effects on Mortality Factors For the three years of this study, no significant difference in emergence success of green turtle nests between the two zones was found. Both abiotic factors (surf, freshwater flooding, hurricane) and biotic factors (predation, turtle digging) affected egg survivorship in Tortuguero turtles, while the extent of damage from each mortality factor varies from year to year. Among the mortality factors, only mammal predation varied significantly between the zones. It is surprising that damage by surf was not very different between the zones; the open zone was 7.1% (egg loss for three year weighted mean), whereas the vegetation/border zone was 4.2%. This is probably because on several occasions heavy erosion cut the beach up to the vegetation area and all nests in the section were damaged, regardless of their zone. Mammalian predation on green turtle nests was greater on the nests in the vegetation/border zone (14.1%) than on nests in the open zone (2.6%) throughout all three years of this study. This spatial difference between the zones most strongly influenced the overall reproduction rate in the 1989 season when the proportion of depredated nests was the highest among three years. After dogs were removed from the beach, coatis were responsible for most of the predation. Cornelius (1986) noted that coatis are the most common predators on olive ridley nests at Nancite, in the Pacific coast in Costa Rica, and that a large group of coatis living on the nesting beach feed almost exclusively on olive ridley eggs during



PAGE 1

16 Histology and Sexing Sample eggs were kept in plastic bags with holes until pipping. Hatchlings were killed, and the gonads were fixed in 10% neutral buffered formalin. The gonads were transferred to the University of Florida for histological analysis. Collection permission was issued by Servicio de Parques Nationales, Costa Rica through Mr. Fernando Cortes; the CITES permit number is 092-88 in Costa Rica, PRT725607 in United States. Transverse serial paraffin sections, 5-8 I.m thick, were prepared from the median portion of each left gonad. Harris' haematoxylin and eosin were used for staining. Criteria for sexing green turtle hatchlings were the same as those reported by Spotila et al. (1983). Sand Temperatures To investigate the seasonal thermal profile of sand temperatures at the same depths as egg chambers of green turtles on the beach, temperatures at depths of 60 cm (all three years) and 80 cm (1988, 1989) in each zone were monitored along two (1986) or four (1988 and 1989) transect lines from July through early December in all three years. These transect lines included both relatively wide and narrow portions of the beach. I placed a set of thermal probes at the border (50% shade) in the vegetation/border zone and at two to three points (0% shade) in the open sand zone: 5 m from the border point and others at 5 m intervals along the transect.



PAGE 1

59 vegetation/border zone and the mean daily rainfall from July through October with half month intervals for my 1986, 1988 and 1989 data sets, the proportion of nests in the shaded zone was lower as a function of the amount of rainfall, but the relationship was not statistically significant (P'=arcsin/P; Y=0.899-0.003X, Rsquared=0.12, n=24, p=0.10). Although the spatial distribution between the zones fluctuates seasonally and yearly, the extent of fluctuation was consistently limited to a small range around the middle value: the proportion of nests in the shaded zone was nearly equal or a little higher (42-60% in 1986, 51-67% in 1988, 48-71% in 1989) relative to nests in the open area throughout each season. This general trend agreed with the results at the northern section of the Tortuguero beach in 1987 (47%-50%) and 1988 (49-62%) but not in 1986 (28-38%). The constant low proportion (around 30%) of nests in the shaded area throughout the 1986 season at the northern section was unique among all other data sets in both studies. However, four additional surveys over the central part of 18 miles of the 22 mile Tortuguero beach in 1986 season by the author (7 and 23 August, 5 and 17 September) showed similar spatial distributions around the middle range (respectively 55.4, 54.1, 45.7, 49.7% of nests in shaded area), and the proportions of the 18 miles were more similar to the data of the two miles of this study area than to those of the northern section. Thus, although spatial distribution of Tortuguero green turtles varies among both seasons and years, the extent of variation might be rather conservative on a population level.



PAGE 1

Figure 24. Fate of green turtle eggs from the sample nests in 1986, 1988, and 1989 at Tortuguero, Costa Rica. The category proportions were calculated as weighted means among the three years.



PAGE 1

13 Furthermore, in spite of intensive marking, some marked nests (2 in 1986, 3 in 1988 and 7 in 1989) were lost. Some of these lost nests may have been completely excavated by other nesting females. In 1986, 49 marked nests were successfully sampled from the latter half of July through the early half of September. Although the actual sampling period was longer, only these periods contained adequate numbers of sample nests in both zones for analysis. In 1988, 88 marked nests were successfully sampled from July through September. In 1989, 113 marked nests from the latter half of July through the early half of October were sampled. The difference in sampling periods was due to the adjustment for the shift of temporal nest distribution between the two years. Marked nests were examined during the beach census for signs of emergence, depredation, inundation, and any disturbance. When the nests were predated, the species of predator was identified by tracks. Emergence success reported in this study was the fraction of eggs that resulted in hatchlings that emerged from the sand. After emergence of hatchlings, the marked nest was excavated. The numbers of hatched and unhatched eggs, and the numbers of dead or live hatchlings remaining underground were determined. Unhatched eggs were opened to check for development; if development had occurred, size of the embryos was recorded. Unhatched eggs lacking visible embryos or blood formation were classified as infertile eggs. Evidence of underground disturbance such as by ghost crabs and flooding, and the number of eggs affected, were determined.



PAGE 1

CHAPTER 3 RESULTS Physical Environment Surrounding Nests Air Temperatures Monthly minimum and maximum air temperatures from July through November in the 1986, 1988 and 1989 seasons are shown in Figure 3. Whereas minimum temperat tIres varied little throughout the season (between 23.3 and 24.1 OC), maximum temperatures showed slight increases in August and September for the three years. The ranges of monthly maximum temperatures for each year were 28.400 (July) -29.90C (September) in 1986, 29.200 (July and November) -30.50C (September) in 1988 and 28.700 (November) 30.800 (August) in 1989. The highest daily air temperatures occurred during September of each year: 35.000 on 1 September 1986, 34.000 on 23 September 1988 and 33.000 on 2 and 3 September 1989. The lowest daily air temperatures were recorded as 22.100 on 10 July 1986, 20.500 on 7 July 1988 and 22.500 on seven occasions throughout the 1989 season, except in September. Yearly variation in mean temperature throughout each season among three years was very low. The mean minimum and maximum temperatures during each five-month period were 23.9 and 29.300 in 18



PAGE 1

47 half of July to the middle of August encountered a higher sand temperature period during the critical stage and produced femalebiased sex ratios. However, the nests in the vegetation/border zone during the same period showed only a slight increase towards a female-biased ratio. The difference in sex ratios between zones during the high sand temperature period was significant (the open zone, mean=82.1%, SE=7.9, range=25.0-100.0, n=10; the vegetation/border zone, mean=31.7%, SE=11.3, range=7.1-89.5, n=7, one-way ANOVA, df=l, F=14.24, p=0.018, sample nests deposited from the latter half of July to the first half of August). A heavy rain event on 6 October (221 mm/day) and inundation by Hurricane Joan during 21 to 22 October decreased the sand temperatures (Figure 13). In both zones, two of eight sample nests deposited during the latter half of August and all 11 sample nests in September were affected during the temperature sensitive period by this cooling. Eleven of those 13 sample nests showed male-biased sex ratios in both zones (the open zone, mean=20.9%, SE=8.5, range=0.0-60.0, n=7; the vegetation/border zone, mean=20.7%, SE=10.2, range=0.0-56.3, n=6), and there was no significant difference between the zones (one-way ANOVA, df=l, F=0.01, p=0.923). Overall Sex Ratio in 1988 To estimate overall sex ratio, frequency of temporal nest distribution was combined with egg survivorship and sex ratio data by half month intervals. Because a significant effect was found in the seasonal factor, but not in the zone factor on egg survivorship



PAGE 1

144 3 1.0 -0 0 0 00 S 0.8 -o o


PAGE 1

6 border vegetation suffered greater predation than did nests on the lower part of the beach. However, because introduced dogs destroyed about one fourth of the nests in the 1977 season, Fowler's results do not represent natural survivorship of the green turtle nests at Tortuguero. Recently, due to success of a dog control program by Tortuguero park guards, the beach has almost returned to the natural condition so that the coati, Nasua narica, has become the main predator. No quantitative study of egg survivorship in the green turtle colony has been conducted since the dogs have been controlled. There is a large yearly fluctuation in the number of nesting green turtle females at Tortuguero; 1977 was a year with relatively few nesting females (Bjorndal et al., 1985). Nest density might affect selection of nest sites on the beach by turtles and also influence spatial and temporal predation patterns. Therefore, a prolonged study to cover several years is needed to understand fully the reproduction of Tortuguero green turtles and to document the biotic and abiotic factors contributing to their natural mortality. A combination of (1) distribution of nests among thermal zones; (2) egg survivorship; and (3) seasonal trend of sand temperature and its consequence -the seasonal trend of sex ratio in the different thermal zones, are recognized as major factors in determining primary sex ratios. The primary objective of this study was to estimate the primary sex ratios of the green turtle population at the Tortuguero beach and to obtain information concerning egg survivorship and sex ratio of the nests throughout the entire season for three years. An additional objective was to accumulate information on the physical



PAGE 1

2 warmer temperatures, males at lower temperatures and both sexes at intermediate temperatures (most turtles); Type C pattern results in females at warmer and lower temperatures and males at intermediate temperatures (three crocodile species, one lizard species and three freshwater turtle species) (see list of species in Janzen and Paukstis, 1991). All seven species of sea turtles with TSD show the Type B pattern. At present, no species with heteromorphic sex chromosomes is known to show TSD. Karyotypes of five species of sea turtles have been examined microscopically, but no discernible heteromorphic sex chromosomes have been confirmed (Caretta caretta, Bickham, 1981; Chelonia mydas, Bickham et al., 1980; Dermochelys coriacea, Medrano et al., 1987; Eretmochelys imbricata, Kamezaki, 1990; Lepidochelys olivacea, Bhunya and Mohanty-Hejmadi, 1986). However, there is evidence that the two sexes of sea turtles are genetically different at the molecular level. In adults of Caretta caretta and Chelonia mydas, males had a higher level of H-Y antigen in their blood cells than did females (Wellins, 1987). A series of experiments on a European freshwater turtle, Emys orbicularis (Zaborski et al., 1982, 1988), which shows Type B TSD, indicated that genetic and environmental factors can operate simultaneously. It appears that incubation temperatures at both extremes can overide genotypic influence, whereas at intermediate temperatures the genetic factor can influence the sex of turtles to some extent. There was a strong correlation between sexual phenotype of gonads and H-Y antigen phenotype of blood cells (male negative / female -positive) from eggs incubated at the



PAGE 1

124 100ooOPENZONE 5 6 8060 95 4020 0 0 U 100VEGETATION/BORDER ZONE Oi6 LU 80 1 U5 40 Uj 20 100 cr o LUJ 100 0 80 20 0 16-31 JUL. 1-15 AUG. 16-31 AUG. 1-15 SEP. PERIOD OF SAMPLE NESTS DEPOSITED Figure 20. Seasonal fluctuation of emergence success of green turtle sample nests in 1986 at Tortuguero, Costa Rica. Monthly mean + standard error with the sample size are shown.



PAGE 1

148 0.7 0.6 0.5 0.4 -++ 0.3+ + + + ++++ -1 0.2 -+ ++ + ++ S 0.1 -* 0 0.0+ U+ LU -0.1 --+ + o" -0.2-"' ++4++ + + -0.3+ -0.4 + +-0.5 -0.6+ -0.7 -, .. 0.0 0.2 0.4 0.6 0.8 1.0 PREDICTED PROBABILITY OF FEMALE Figure 38. Residuals of the selected logistic model that explains sex ratio fluctuation of green turtle hatchlings. Data were collected at Tortuguero, Costa Rica, in the 1986 and 1988 seasons. Y=exp(-9.81 + 0.52X1-0.08X2-0.09X3+0.55X4)/(l1+ exp(9.81 + 0.52X1-O.08X2-O.09X3+O.55X4)). pseudo R2=0.24. Xi: mean nest temperatures during the middle third of development. X2: incubation duration. X3: mean daily rainfall from the deposition of the clutch through the middle third of development. X4: zone of clutch (1 -the open zone, 0 -the vegetation/border zone).



PAGE 1

12 Sampling of Egg Survivorship Nests A representative sample of nests deposited within the study area was marked and followed throughout the incubation period. As much as possible, I attempted to sample equal numbers of nests in the two zones and for each half month period of time to analyze seasonal and spatial variances of egg survivorship. Sampling at night was conducted seven to 10 times during each half month period to locate nesting females. During 1986 and 1988, only females found during the early stage of nesting behavior (before digging the egg chamber) were sampled; for these females I counted the number of eggs during deposition. Because fewer females nested during 1989 than during 1986 and 1988, it was necessary to include nesting females that were already in the process of depositing eggs; thus during 1989 the number of eggs was not counted. Sample nests were marked with a numbered stake placed 1 m from the egg chamber. I also placed two pieces of vinyl tape in the vegetation to triangulate the location. A piece of numbered flagging tape was placed in the egg chamber to confirm the location in the event of nest destruction. Due to the difficulty of finding nesting females in a specific zone during a specific period, the numbers of marked nests in each block are not equal. I omitted from analysis several marked nests (1 in 1986, 2 in 1988 and 1 in 1989) that were located so close to other egg chambers that I was unable to determine hatching success for the individual nests because the eggshells were inseparable.



PAGE 1

27 zone. During the entire five days, the sky was overcast during the daytime. Prolonged rainy days kept sand temperatures low for the duration and caused long-term fluctuation. The continuous low temperature during July through August in 1986 was one such situation. Table 2 shows overall mean sand temperatures throughout each season at depths of 60 cm and 80 cm (only 1988, 1989) in the two zones. For all three seasons, the trends of spatial thermal profiles were identical. Sand temperatures at depths of 60 cm and 80 cm in the open zone were significantly higher than those in the vegetation/border zone. Within the same zone, sand temperatures at a depth of 60 cm were significantly higher than those at a depth of 80 cm (one-way ANOVA with repeated measure; 1986, 1988, 1989, all tests p < 0.0001; one-way ANOVA of repeated measure with Bonferroni adjustment of P for multiple comparison, overall alpha 0.05, 1988, 1989, all tests p<0.008). For overall mean sand temperatures throughout the season, the differences between the two zones ranged from 0.80C (1986) to 1.50C (1989) at a depth of 60 cm and from 1.10C (1988) to 1.30C (1989) at a depth of 80 cm. The differences between the depths of 60 cm and 80 cm ranged below 0.40C in both zones in 1988 and 1989. For overall mean sand temperatures throughout the season, the 1988 season showed the highest, the 1989 season the intermediate, and the 1986 season the lowest sand temperatures at a depth of 60 cm in both zones during the study. Figure 14 shows seasonal fluctuations of mean monthly sand temperatures at a depth of 60 cm for the three years in both zones.



PAGE 1

66 inundation events in close succession in October from excessive rain and Hurricane Joan. However, hurricanes on the Caribbean coast of Costa Rica are relatively rare. Coen (1983) noted that Hurricane Martha, 21-25 November 1969, was the only one to hit the coast during the hundred years of records. Therefore, it would appear that the influence of hurricanes on the Tortuguero green turtle colony would be negligible, and that the 1988 season was an atypical year. The extent of damage by Hurricane Joan was not accurately assessed because of the overlapping damage by the excessive rain on 6 October. Therefore it is unknown whether the seasonal fluctuation in emergence success in the 1988 season would have occurred without Hurricane Joan. Separate flooding events in August 1986 and in October 1989 affected the sample nests, but these events did not result in drastic seasonal changes in emergence success within each season. This lack of seasonal variation can be partially explained by the observation that all stages of embryos and emerging hatchlings are susceptible to flooding. One flooding event can affect nests deposited over long periods, possibly up to four months, if the flooding occurs in the middle of the incubation season. The same principle can apply to the damage by surf, and even for predation, because mammal predators destroyed the nests at various stages. Thus, favorable weather conditions in September (e.g., calm seas, less chance of excessive rain) can have a positive effect on emergence success of nests deposited over the long period of time from July through September, which covers the majority of the nesting season. In 1977, Fowler (1979) analyzed emergence



PAGE 1

19 1986, 23.4 and 29.70C in 1988 and 23.7 and 29.80C in 1989, respectively. Overall mean three-year minimum and maximum temperatures during the same period were 23.7 and 29.80C, respectively. Rainfall Daily rainfall records from 16 June through 10 December during 1986, 1988 and 1989 are shown in Figure 4. The sampling period covered the entire incubation season of green turtle eggs in Tortuguero. For all three years, heavy rain events sporadically occurred. Rainfall greater than 100 mm during 24 hours was recorded on seven occasions in 1986, five occasions in 1988 and five occasions in 1989. The heaviest 24-hour precipitation each year was 186 mm on 21 October in 1986, 221 mm on 6 October 1988 and 155 mm on 25 October 1989. Among those intense rainfall events, at least two events in 1986 (5 August, 141 mm/day; 6 December, 178 mm/day), one event in 1988 (6 October, 221 mm/day) and one consecutive three-day event in 1989 (30 October-1 November, total 367 mm for three days) caused freshwater flooding on the beach and consequently damaged some green turtle nests (see section on ground water). The 1986 season overall was the wettest among the three years. The total rainfall during the nearly six month period (16 June-10 December, 178 days) was 3871 mm in 1986, 2578 mm in 1988 and 2896 mm in 1989. Mean rainfall taken at the Tortuguero park station during the same period from 1978 through 1989 (I have



PAGE 1

42 As a supplemental observation on excavation by female turtles, I recorded the number of recently excavated nests by turtles in the study area during the beach census. In total, 238 nests in 1986 (48 surveys), 123 nests in 1988 (56 surveys), and 19 nests in 1989 (80 surveys) were observed to be excavated by female turtles. The percentage of the excavated nests of total nests in each season was calculated as 6.5% in 1986, 3.5% in 1988, and 0.9% in 1989, respectively. In the 1989 season when nest density was the lowest among three years, the intensity of excavation by turtles decreased to a very low level relative to the other two years (Figure 28), and none of the sample nests was excavated. The seasonal fluctuation of excavation by turtles was positively related to the extent of nesting activities (Figure 29). Only one sample nest was damaged by plant roots. A bundle of young coconut tree roots infiltrated three eggs of a sample nest, which was located close to the trunk of the tree. Several unsampled nests were found to be partially tangled by roots of sea oats, but no apparent damage by the roots was recognized. These roots did not penetrate any egg shells, and those nests had high emergence success. Natural damage by plant roots was probably minimal on green turtle eggs at Tortuguero. Other Categories of Mortality Over the three years, a mean of 3.6% of eggs in the open zone and 3.1% of eggs in the vegetation/border zone ceased embryonic development from no apparent cause. Actual causes of mortality of these eggs, whether genetically inherent or environmentally induced,



PAGE 1

70 seasonal fluctuation. However, the pattern of fluctuation is very different from rookery to rookery. At Tortuguero and Sarawak, the dry season occurs somewhere in the middle of the nesting season, whereas the wet season occurs near the middle of the nesting season in Surinam. In French Guiana, the wet season occurs early in the nesting season. Regardless of the pattern, these sea turtle rookeries probably produce variable sex ratios throughout the season. Annual Variation in Primaryt Sex Ratio In the 1986 season, the sand temperature in both zones remained below the pivotal temperature for most of the season. In the open zone, although the sand temperature in September and later intermittently rose to the level of the pivotal temperature, these periods were too short to produce many females. Therefore, in spite of the small sample size, the 10.1% proportion of females from the samples that were sexed directly is probably a good estimate. The weather pattern is one way to assess whether the year was a typical one for Tortuguero. The rainfall in 1986 in August (846 mm) and in September (514 mm) was the highest on record, and the total amount of rainfall throughout the incubation season (3871 mm, from 16 June through 10 December) was the second highest for the last 12 years. Thus, it appears that the weather in the 1986 season was atypically wet. Therefore the highly male-biased sex ratio is probably atypical for the Tortuguero population.



PAGE 1

CHAPTER 2 METHODS Study Area My study area was located at Tortuguero, Costa Rica. The Tortuguero beach extends 22 miles (35.4 km) from the mouth of the Rio Tortuguero south to the Rio Parismina on the Caribbean coast of Costa Rica (Figure 1). The beach is located on a long, narrow island that is separated from the mainland by a freshwater lagoon and estuaries of the rivers. Tortuguero beach is the nesting beach for the largest surviving nesting population of green turtles in the Atlantic Ocean (Groombridge and Luxmoore, 1989). Leatherback turtles, Dermochelys coriacea, and hawksbill turtles, Eretmochelys imbricata, also nest on the same beach, but are much less abundant than green turtles. The shoreline is composed of a continuous series of spits and guts. The shape of the shoreline shifts considerably with surf erosion and rebuilding, sometimes in a short period of time. High waves occur almost throughout the nesting and incubation season of green turtles (June to December). The beach is classified as a high energy beach because of constant high wave activity. The beach was divided into 22 one-mile sections from the northern end to the southern end. The central area of the beach (mile 8



PAGE 1

CHAPTER 1 INTRODUCTION Recently it has become apparent that there are two types of sex determination mechanisms in reptiles: genotypic and environmental. In the latter case, sex of the offspring is decided after fertilization by environmental factors. The common form of environmental sex determination in turtles is temperaturedependent sex determination (TSD), in which incubation temperature during a critical portion of embryonic development controls gonadal differentiation. The physiological mechanism involved in this phenomenon is not understood (reviewed in Bull, 1983: Ewert and Nelson, 1991; Janzen and Paukstis, 1991). At present, seven of eight species of sea turtles are known to show the TSD phenomenon. These species are: Caretta caretta (Yntema and Mrosovsky 1980), Chelonia mydas (Miller and Limpus 1981), Chelonia agassizi (Alvarado and Figueroa, 1989), Dermochelys coriacea (Mrosovsky et al., 1984), Eretmochelys imbricata (Dalrymple et al., 1985), Lepidochelys olivace (McCoy et al., 1983), and Lepidochelys kempi (Shaver et al., 1988). There are three patterns of TSD in reptiles (Bull, 1983). Type A pattern results in females at lower temperatures, males at higher temperatures and both sexes at intermediate temperatures (most crocodilians and lizards); Type B pattern results in females at 1



PAGE 1

Figure 10. Frequency distribution of wave height in the 1986, 1988, and 1989 incubation seasons of green turtle eggs at Tortuguero, Costa Rica.



PAGE 1

30 X2=19.9 P<0.01 in 1989). Tukey-type multiple comparison tests for the proportional data (Zar, 1984) identified the differences among each seasonal period. There was no common seasonal trend among the three years. In 1986, the proportion of clutches deposited in the vegetation/border zone was significantly higher early in the season than during the rest of the season, and the proportion decreased as the season progressed. In 1988 and 1989, the proportion of nests in the two zones oscillated throughout each season in a similar way. The proportions of clutches deposited in the vegetation/border zone shifted from high to low, and then again from high to low throughout each season. Analysis of Egg Survivorship Samples Fates of Sample Nests The fates of 49 nests in 1986, 88 nests in 1988 and 113 nests in 1989 were determined. Mean clutch size for the sample nests was 116.9 eggs (SE=3.4, range=28-165, n=49) in 1986, 109.1 eggs (SE=2.1, range=53-148, n=88) in 1988, and 107.1 eggs (SE=2.1, range 53-152, n=89) in 1989. The frequency distributions of the emergence success rate of these nests were far from a normal curve and instead showed concave bimodal shapes in each year (Figure 18). For all three years, most of the sample nests fell into one of two extremes: highly successful emergence percentage (>70%) or entire dead clutches. No hatchlings emerged from 24.0% and 25.0% of sample nests deposited in the open zone and in the vegetation/border



PAGE 1

LITERATURE CITED Aldrich, J.H. and F.D. Nelson. 1984. Linear Probability, Logit, and Probit Models. Sage Publications, Beverly Hills, CA. Alvarado, J. and A. Figueroa. 1989. The ecological recovery of sea turtles of Michoacan, Mexico. Special attention: the black turtle, Chelonia agassizi. U.S.F.W.S. Endangered Species Report. Balazs, G.H. 1980. Synopsis of biological data on the green turtle in the Hawaiian Islands. NOAA Technical Memorandum NMFS. NOAA-TM-NMFS-SWFC-7. Balazs, G.H. and E. Ross. 1974. Observations on the preemergence behavior of the green turtle. Copeia 1974:986-988. Bhunya, S.P. and P. Mohanty-Hejmadi. 1986. Somatic chromosome study of a sea turtle, Lepidochelys olivacea (Chelonia, Reptilia). Chrom. Inf. Serv. 40:12-14. Bickham, J.W. 1981. Two hundred million year old chromosomes: deceleration of the rate of karyotypic evolution in turtles. Science 212:1291-1293. Bickham, J.W., K.A. Bjorndal, M.W. Haiduk, and W.E. Rainey. 1980. The karyotype and chromosomal banding patterns of the green turtle (Chelonia mydas). Copeia 1980:540-543. Bjorndal, K.A. and A.B. Bolten. 1992. Spatial distribution of green turtle (Chelonia mydas) nests at Tortuguero, Costa Rica. Copeia 1992:45-53. 149



PAGE 1

3 intermediate temperature. Girondot and Pieau (1990) proposed a similar mechanism to explain the variation of the pivotal temperatures observed in loggerhead turtles (Limpus et al., 1985; Mrosovsky, 1988). Further investigation of the effect of genetic factors over TSD in species of sea turtles has not been conducted. Tortuguero beach on the Caribbean coast of Costa Rica is one of the last major nesting sites in the western Atlantic for green turtles, Chelonia mydas. Spotila et al. (1987) found that the temperatures of the nest during the middle third of development influenced the sex ratio of Tortuguero green turtles in natural nests. Mean temperatures less than 28.50C produced mainly males, mean temperatures greater than 30.30C produced only females, and mean temperaturess between 28.5 and 30.30C produced both sexes. They also found that the different thermal zones at the beach yielded different sex ratios; nests under vegetation produced a high percentage of male hatchlings, whereas nests in the open beach produced mainly females. No seasonal trend of sand temperature was apparent during their monitoring period (24 July -22 September 1980). Assuming no seasonal fluctuation of sex ratio, and using 1977 data for the distribution of nests between the shady zone and the open sand zone at Tortuguero (Fowler, 1979), they estimated that 67% of the hatchlings were females in the 1977 season. However, their study did not cover the full seasonal profile at Tortuguero, and the sample size was very limited (15 nests). The main incubation season of the green turtle colony at Tortuguero extends from July through November (Fowler, 1979). The rainfall at Tortuguero is extremely variable from year to year (Myers, 1981).



PAGE 1

153 Limpus, C.J., P. Reed, and J.D. Miller. 1983. Island and turtles. The influence of choice of nesting beach on sex ratio. In J.T. Baker, R.M. Carter, P.W. Sammarco, and K.P. Stark (eds.), Proceedings: Inaugural Great Barrier Reef Conference, pp.397-402. James Cook University Press, Townsville, Queensland, Australia. Limpus, C.J., P. Reed, and J.D. Miller. 1985. Temperature dependent sex determination in Queensland sea turtles: intraspecific variation in Caretta caretta. In G. Grigg, R. Shine, and H. Ehmann (eds.), Biology of Australasian Frogs and Reptiles, pp. 343-351. Royal Zoological Society, New South Wales. Maxwell, J.A., M.A. Motara, and G.H. Frank. 1988. A microenvironmental study of the effect of temperature on the sex ratios of the loggerhead turtle, Caretta caretta, from Tongaland, Natal. S. Afr. J. Zool. 23:342-350. McCoy, C.J., R.C. Vogt, and E.J. Censky. 1983. Temperaturecontrolled sex determination in the sea turtle LepidQochelys olivacea. J. Herpetol. 17:404-406. Medrano, L., M. Dorizzi, F. Rimblot, and C. Pieau. 1987. Karyotype of the sea turtle Dermochelys coriacea (Vandelli, 1761). Amphibia-Reptilia 8:171-178. Meylan, A.B., P.A. Meylan, H.C. Frick, and J.N. Burnett-Herkes. 1992. In Salmon, M. and J. Wyneken (compilers), Proceedings of the 11th Annual Workshop on Sea Turtle Conservation and Biology. pp. 73. NOAA Technical Memorandum NMFS-SEFSC-302. Miller, J.D. 1985. Embryology of marine turtles. In C. Gans, F.Billett, and P.F.A. Maderson (eds.), Biology of the Reptilia, Vol. 14. pp.269-328. Academic Press, New York. Miller, J.D. and C.J. Limpus. 1981. Incubation period and sexual differentiation in the green turtle Chelonia mydas L. In C.B. Banks, and A.A. Martin (eds.), Proceedings of the Melbourne Herpetological Symposium. pp.66-73. The Zoological Board of Victoria, Melbourne, Australia. Morreale, S.J., G.J. Ruiz, J.R. Spotila, and E.A. Standora. 1982. Temperature-dependent sex determination: current practices threaten conservation of sea turtles. Science 216:1245-1247.



PAGE 1

4 During the middle of the season, a short dry period occurs (usually in September). Carr (1979) described the June -August period, the early incubation season, as "monsoonlike, with many consecutive days of heavy overcast and rain". In general, rainfall and sand temperature are inversely related on the beach. Therefore, it is very probable that seasonal fluctuation of sand temperature, and consequently seasonal fluctuation of sex ratios, can occur regularly at Tortuguero. Therefore, in addition to evaluating whether the 1980 season is typical or atypical, it is essential to examine the entire season over several years to document seasonal and yearly variation for a better estimation of the primary sex ratio at Tortuguero. All species of sea turtles are considered to be threatened or endangered because of over-exploitation and destruction of their habitats. The green turtle, a circumtropical large species, is a valuable food source for many coastal people, and thus has been harvested for a long time. Current management practices for sea turtles focus on conservation of each population. Among the conservation practices, protection of incubating eggs in several forms of artificial hatcheries is presently the most common measure. However, little regard has been given to the incubation temperature until recently. Incubation of sea turtle eggs in styrofoam boxes subjected the eggs to a different thermal environment from that experienced on a natural beach and showed a masculinizing effect on the embryos (Mrosovsky, 1982; Dutton et al., 1985). Because of this incubation method, the early years of the Kemp's ridley head starting project unintentionally produced male



PAGE 1

31 zone respectively in 1986, 21.6% and 39.2% in 1988, 9.4% and 28.3% in 1989. Total nest loss was caused by beach erosion, inundation by excessive rain or Hurricane Joan, depredation by coatis (Nasua narica) and ghost crabs (Ocypode quadrata), and excavation by nesting female turtles (Figure 19). For each nest, emergence success was defined as the percentage of its egg clutch that produced hatchlings that successfully emerged from the sand column. Mean emergence success throughout each season was 54.8% (SE=7.3, n=25) in 1986, 57.8% (SE=6.2, n=37) in 1988, and 74.7% (SE=4.8, n=53) in 1989 in the open zone; emergence success was 47.3% (SE=8.1, n=24) in 1986, 42.7% (SE=5.6, n=51) in 1988, 60.0% (SE=5.4, n=60) in 1989 in the vegetation/border zone. Although the mean emergence success in the open zone was slightly higher than that in the vegetation/border zone during each year, the differences between the two zones were not significant (Mann-Whitney U test, p=0.778 in 1986, p=0.069 in 1988, p=0.061 in 1989). With pooling the two zones, overall mean emergence success of sample nests throughout each season was 51.2% (SE=5.4, n=49) in 1986, 49.1% (SE=4.2, n=88) in 1988 and 66.9% (SE=3.7, n=113) in 1989. The overall mean emergence success in 1989 was significantly higher than that in either 1986 and 1988 [Kruskal-Wallis Test, df=2, H=21.372, p<0.0001; nonparametric multiple comparison test with unequal sample sizes (Zar, 1984), 1989 vs 1988, Q=4.121, 1989 vs 1986, Q=3.465, Q 0.05,3=2.394]. Emergence success exhibited a different seasonal trend for each year (Figures 20, 21 and 22). For analyzing the effects of



PAGE 1

62 collected only from nests above the spring high tide. Mrosovsky (1989) calculated the overall hatching success in Surinam to be 63.5% by adjusting the spatial distribution data. Recently, a longterm quantitative investigation has been conducted for green turtles on Florida beaches, although the beaches were more heavily utilized by loggerhead turtles than by green turtles. Mean emergence success of green turtle nests at the Melbourne beach varied from 54.6 to 75.2% (overall mean=65.6%) from 1985 through 1988 (Redfoot and Ehrhart, 1989). At a nearby beach, Horton (1989) did a similar study and reported 40.1% for mean emergence success for 20 green turtle nests deposited in 1988 and 1989. Mean emergence success of sample nests of this study was 51.2% (1986), 49.1% (1988) and 66.9% (1989). By adjusting the temporal and spatial distribution of nests for each year, overall hatching success is estimated as 57.2% (1986), 45.7% (1988) and 66.6% (1989). Overall hatching success at Tortuguero under natural conditions falls within the range reported by the above quantitative studies in other parts of the world. In the 1977 season when dog predation was serious, 42% of 350 monitored nests produced hatchlings and 83% of the eggs of those hatched nests emerged (Fowler, 1979). Thus, the 1977 season's overall hatching success is roughly estimated to be 34.8% (0.42 X 0.83). The success of green turtle nests at Tortuguero was apparently improved by the dog control program even though the natural predator, coatis, still depredate a substantial proportion of nests in some years, such as 1989.



PAGE 1

147 1.0 o o00 o o 0 S0.8o 0 0 C 0.6 o O o*** 0 o c. 0 LU 0.4 0


PAGE 1

39 vegetation/border zone in the 1989 season was the highest loss factor among all the categories throughout the three years. The mean time of predation for the sample nests for which the predation date was confirmed was 41.1 days after deposition (range=Day 0 to Day 72, SE=4.0, n=26) (Figure 25). The distribution of predation days was composed of two groups: before 11 days and after 32 days. Most (79.2%) of the egg depredation occurred between 32 and 68 days of incubation. One nest was depredated at Day 72 when the hatchlings were emerging to the surface. For a supplemental observation on mammal predation, I recorded the number of recently depredated nests in the study area during the beach census. For all three years, more nests depredated by mammals were observed in the vegetation/border zone than in the open zone (X2=82.9, p<0.00O1, n=233, 1986: X2=152.0, p<0.0001, n=330, 1988: X2=174.3, p<0.0001, n=456, 1989) (Figure 26). Although nest predation by mammals was observed throughout the incubation period, the seasonal trend of predation intensity varied considerably among the three years (Figure 27). The trend in 1988 showed an almost opposite trend to that in 1989. While predation intensity in the 1988 season increased to the highest rate in the early part of the season and gradually decreased as the season progressed, nest predation in the 1989 season steadily increased as the season progressed almost to the end. The peak of nesting activities occurred between the latter half of August and September for the three years (Figure 16). The seasonal fluctuation of predation intensity did not match the seasonal fluctuation of nesting act ivi t ies.



PAGE 1

98 Border/Veg. Open Sand Zone Zone • fd ,-* .*. • .. Figure 2. Beach zones at the study area, Tortuguero, Costa Rica. The vegetation/border zone lies within 2 m of the border of dense vegetation (5-100% cover). The open zone lies below the vegetation/border zone (<5% cover).



PAGE 1

73 Second, because of seasonal fluctuation in sand temperatures and the general increase in metabolic heat throughout incubation, individual nest temperatures on the beach were rarely constant throughout the critical period. The extent and the pattern of variance differed greatly among nests. Standora et al. (1982) reported spatial variation in sex ratio within natural green turtle nests incubated at the pivotal temperature. Due to metabolic heating, eggs near the center of the clutch produced females, whereas those at the periphery produced males. In the present study, I measured nest temperatures at the center of each clutch, but determined the sex of eggs collected randomly throughout the clutch. Therefore, my nest temperature readings may not have accurately represented the incubation temperature of all sampled eggs within a nest. Had I collected the sample eggs from only the center mass of each clutch near the temperature probe, the variation in sex ratio relative to the nest temperatures might have been smaller. Furthermore, in this study the critical period was calculated based on an assumption that the time from pipping to emergence was five days for all sample nests. Hendrickson (1958) indicated that heavy rains, which pack the upper layers of beach sand, might prolong the emerging period of green turtle hatchlings in Sarawak. Because Tortuguero had high seasonal fluctuation in rainfall, the emergence period of green turtles might also vary to some extent. Thus the calculated critical period of each nest might not be accurate, and the mean nest temperature calculated for each nest may only roughly represent its thermal environment.



PAGE 1

140 100 80 -0 60 IJ 40 3/4 20 1,1 4/0 0 16 JUL. 16 AUG. 16 SEP. -15 AUG. -15 SEP. -15 OCT. PERIOD OF SAMPLE NESTS DEPOSITED Figure 32. Seasonal fluctuation of sex ratio of green turtle hatchlings in 1986 at Tortuguero, Costa Rica. Shown is the monthly mean one standard error. The number on the left side above SE bar shows the number of sample nests deposited in the open zone and the number on the right side shows ones in the vegetation/border zone.



PAGE 1

88 Table 6. Characteristics of green turtle sample nests for sex ratio analysis. The data were collected at Tortuguero, Costa Rica, during the 1986 and 1988 seasons. Deposition date Zonea Sample sizeb % Female 18 July 1986 V/B 16 68.8 18 July V/B 19 0 21 July 0 2c 0 28 July V/B 20 0 14 Aug. 0 20 0 21 Aug. 0 17 47.1 10 Sept. 0 19 0 10 Sept. V/B 20 0 16 Sept. 0 20 0 25 Sept. 0 20 0 28 Sept. 0 1 9 5.3 3 Oct. 0 20 0 3 July 1988 V/B 13 15.4 5 July 0 20 55.0 6 July 0 19 63.2 9 July 0 20 10.0 13 July V/B 20 30.0 14 July V/B 19 10.5 15 July 0 19 73.7 16 July V/B 20 55.0 21 July V/B 19 21.1 23 July 0 20 90.0 23 July 0 14 78.6 27 July 0 20 100.0 28 July 0 20 100.0 29 July V/B 14 7.1 30 July V/B 19 10.5 3 Aug. V/B 12 16.7 3 Aug. V/B 19 (1) 89.5 5 Aug. V/B 18 22.2 6 Aug. 0 15 100.0 7 Aug. 0 16 25.0 8 Aug 0 18 (1) 61.1



PAGE 1

Figure 8. Seasonal fluctuation of ground water table and rainfall in the 1986 incubation season of green turtle eggs at Tortuguero, Costa Rica. The line of mean bottom depth of green turtle clutches is 77.9 cm (SD=11.1) from the surface. The shaded area shows the maximum and minimum range of ground water table recorded at two wells in the open zone and at three wells in the vegetation/border zone. Before 4 August in the open zone and before 14 August in the vegetation/border zone, only one well was measured in each zone.



PAGE 1

157 Witherington, B.E. 1986. Human and natural causes of marine turtle clutch and hatchling mortality and their relationship to hatchling production on an important Florida nesting beach. M.S. thesis. University of Central Florida, Orlando, FL. Yntema, C.L. and N. Mrosovsky. 1980. Sexual differentiation in hatchling loggerheads (Caretta caretta) incubated at different controlled temperatures. Herpetologica 36:33-36. Yntema, C.L. and N. Mrosovsky. 1982. Critical periods and pivotal temperatures for sexual differentiation in loggerhead sea turtles. Can. J. Zool. 60:1012-11016. Zaborski, P., M. Dorizzi, and C. Pieau. 1982. H-Y antigen expression in temperature sex-reversed turtles (Emys orbicularis). Differentiation 22:73-78. Zaborski, P., M. Dorizzi, and C. Pieau. 1988. Temperature-dependent gonadal differentiation in the turtle Emys orbicularis: Concordance between sexual phenotype and serological H-Y antigen expression at threshold temperature. Differentiation 38:17-20. Zar, J.H. 1984. Biostatistical Analysis. Prentice-Hall Inc. Englewood Cliffs, NJ.



PAGE 1

154 Mortimer, J.A. 1981. Reproductive ecology of the green turtle, Chelonia mydas, at Ascension Island. Ph.D. Dissertation. University of Florida, Gainesville, FL. Mortimer, J.A. 1990. The influence of beach sand characteristics on the nesting behavior and clutch survival of green turtles (Chelonia myda$). Copeia 1990:802-817. Mortimer, J.A. and A. Carr. 1987. Reproduction and migration of the Ascension Island green turtle (Chelonia mydas). Copeia 1987:103-113. Mrosovsky, N. 1982. Sex ratio bias in hatchling sea turtles from artificially incubated eggs. Biol. Conserv. 23:309-314. Mrosovsky, N. 1988. Pivotal temperatures for loggerhead turtles (Caretta caretta) from northern and southern nesting beaches. Can. J. Zool. 66:661-669. Mrosovsky, N., P.H. Dutton, and C.P. Whitmore. 1984. Sex ratio of two species of sea turtle nesting in Suriname. Can. J. Zool. 62:2227-2239. Mrosovsky, N. and J. Provancha. 1989. Sex ratio of loggerhead sea turtles hatching on a Florida beach. Can. J. Zool. 67:25332539. Mrosovsky, N. and C.L. Yntema. 1980. Temperature dependence of sexual differentiation in sea turtles: implications for conservation practices. Biol. Conserv. 18:271-280. Myers, R.L. 1981. The ecology of low diversity palm swamps near Tortuguero, Costa Rica. Ph.D. Dissertation. University of Florida, Gainesville, FL. Packard, G.C., M.J. Packard, and G.F. Birchard. 1989. Sexual differentiation and hatching success by painted turtles incubating in different thermal and hydric environments. Herpetologica 45:385-392.



PAGE 1

also like to thank the Costa Rican National Park System for allowing me to work in Tortuguero National Park. III



PAGE 1

89 Table 6--continued. Date deposited Zonea Sample sizeb % Female 8 Aug. 1988 O 20 100.0 11 Aug. O 16 100.0 15 Aug. O 12 66.7 17 Aug. O 20 60.0 17 Aug. O 20 5.0 26 Aug. V/B 20 55.0 27 Aug. O 13 (1) 15.4 28 Aug. V/B 17 47.1 30 Aug. V/B 16 (2) 56.3 30 Aug. O 20 (1) 80.0 31 Aug. V/B 19 0.0 3 Sept. V/B 18 44.4 8 Sept. O 1 8 27.8 9 Sept. V/B 18 0.0 14 Sept. O 17 0.0 15 Sept. O 20 60.0 19 Sept. O 20 (1) 25.0 20 Sept. O 18 33.3 22 Sept. V/B 17 23.5 24 Sept. O 19 0.0 27 Sept. O 1 5 0.0 27 Sept. V/B 20 0.0 a: V/B-the vegetation/border zone, O-the open zone. b: values in parentheses are the number of turtles in which gonads showed both seminiferous tubles and developed cortex. c: missed emergence; two hatchlings remained in the nest.



PAGE 1

Table 8. Estimation of overall sex ratio of green turtle hatchlings at Tortuguero, Costa Rica, in the 1988 season. Period #Nest #Nest %Egg Female% Female% #Hatchling #Hatchling #Female #Female Survivorship (x109.1) (x109.1) (x109.1) (x109.1) Open V/B Pooled Open V/B Open V/B Open V/B July 1-15 294 489 48.9 50.5 18.6 144 239 73 45 July 16-31 725 766 79.7 92.1 23.4 578 610 532 143 Aug. 1-15 771 885 57.2 75.5 42.8 441 506 333 217 Aug. 16-31 746 1494 16.9 40.1 39.6 126 252 50 100 Sep. 1-15 1182 1566 45.4 29.3 22.2 536 710 157 158 Sep. 16-30 498 645 40.0 14.6 11.8 199 258 29 30 2024 2576 1174 692 Total hatchlings 4600 1866 Overall female ratio 40.6% Open: open zone. V/B: vegetation/border zone. 109.1: mean clutch size in 1988. Number of hatchlings = Number of nests x mean clutch size (109.1) x % Egg Survivorship. Number of females = Number of hatchlings x % Female.



PAGE 1

143 0 0 0 30.50 80 1-.18 oo0 Lu0 00 D0 0O 00 < 29.50 0 00000 L_. 28.5" 0 o z8 Z 27.5000


PAGE 1

81 To calculate the primary sex ratio in the 1988 season, temporal and spatial nest distribution, emergence success, and sex ratio data were combined by half month intervals. The overall proportion of females was estimated to be 40.6% in the 1988 season. During 1986, because no significant seasonal effect was detected, the overall proportion of females was calculated to be 10.1% from the mean value of sample nests. The relationship between sex ratio and mean nest temperature during the middle third of development, mean sand temperature, incubation period, three sets of rainfall records, depth of clutch, and nesting zone were analyzed by logistic models. The pivotal temperatures were calculated as 29.40C for mean nest temperature, and 28.50C for mean sand temperature. The primary environmental factor in determining sex of green turtles--mean nest temperature-had the best fit in the single regression model. However, the variability of the model was high (pseudo R2=0.20). Mean nest temperature, incubation period, rainfall from egg deposition through the middle third of development, and nesting zone were significant parameters in the multiple regression analysis (pseudo R2=0.24). Another multiple regression model with mean daily rainfall from the time of egg deposition through the middle third of development, nesting zone, and their interaction (pseudo R2=0.23) was found to be as good as the mean nest temperature model. This rainfall/zone model might have value as a conservation tool to estimate roughly the sex ratio of green turtle nests. In conclusion, the variability of sex ratio by all of the above models was still very high for predictive



PAGE 1

121 401 9 8 6 VEGETATION/BORDER ZONE, N=24 [] OPEN ZONE, N=25 301 20, N 10 LU 0i Lko Ji E o 0 10 20 30 40 50 60 70 80 90 100 N 401 9 8 9 VEGETATION/BORDER ZONE, N=-51 O[0 OPEN ZONE, N=37 LU 30z C/) 20 Cn w4 C/) 0 10 20 30 40 50 60 70 80 90 100 Rica. -J <. 1699 VEGETATION/BORDER ZONE, N=60 o E]1 OPEN ZONE, 5 .14 30 20. 10 0 10 20 30 40 50 60 70 80 90 100 EMERGENCE SUCCESS (%) Figure 18. Percentage distribution of green turtle sample nests by the emergence success in 1986, 1988, and 1989 at Tortuguero, Costa Rica.



PAGE 1

41 out logs on the beach, and all sample nests depredated by termites were located near such logs. An unidentified animal was responsible for minor damage on eggs (mean egg loss of 0.9% in the open zone, 0.2% in the vegetation/border zone throughout the three years). Typically, there were a few tiny holes (diameter 1-2 mm) on the egg shells, which were different from the nipped holes by ghost crabs, and the insides of the shells were often partially stuffed with dry sand. There were no visible animals in the shells at excavation. Mean number of eggs damaged by this cause was 6.2 per clutch (SE=2.1, range=1-47, n=23). I suspect that fire ants, which were very common on the Tortuguero beach, might be responsible for these damages, but I have no confirmation. Excavation of sample nests by nesting female turtles occurred twice in the 1986 season and six times in the 1988 season, but never in the 1989 season. The extent of damage by this cause (egg loss of 1.1% in the open zone and 4.8% in the vegetation/border zone throughout the three years) was minor relative to predation by mammals. Generally the damage occurred when a female dug an egg chamber overlapping a previously laid clutch. In such a case, numerous eggs were excavated to the surface. Two of the damaged sample nests did not produce any hatchlings because mammal and other predators destroyed the rest of the eggs. Another six sample nests produced some hatchlings; those nests were well covered by sand from another or the same female turtle's activities before predators invaded the nests.



PAGE 1

20 omitted 1984 data because of an incomplete data set, Instituto Meteorogico Nacional) was 2882 mm (SE=622, n=11). The rainfall (3871 mm) in the 1986 season was the second highest on record; the greatest rainfall (3929 mm) was recorded in 1982. Figure 5 shows seasonal variation of monthly rainfall taken at the Tortuguero National Park Station from 1978 through 1989 (Instituto Meteorogico Nacional, Costa Rica). The major incubation season of green turtles (July through November) includes several of the wettest months (July and November) during the year. September, however, which was the middle of the incubation period, had the least rain. Within the five-month comparison (July through November), the monthly rainfall during September was significantly less than that during July and November, but not significantly less than in August and October (one-way ANOVA, df=4, p=0.064; Fisher's procedure of least significant difference, alpha=0.05). These data indicate that a drier period normally occurs during the middle of the incubation season of green turtles at Tortuguero. The monthly rainfall for the three study years and the mean rainfall from 1978 through 1989 are shown in Figure 6. During the 1986 season, the monthly rainfall during the early to middle incubation season well exceeded the mean values. The rainfall in August (846 mm) and in September (514 mm) during 1986 were the highest monthly records. As a result, the 1986 season did not experience a dry period in the middle of the season, and the monthly rainfall exceeded 500 mm throughout the entire season. On the other hand, the 1988 and 1989 seasons were much drier than the 1986 season during the middle of the incubation season. The rainfall in



PAGE 1

52 models, n=52, p<0.05). As expected, the proportion of females decreased with an increase of mean daily rainfall (Figure 36). This trend was the most apparent for the nests in the open zone and as a whole in the data set of R-111, which showed the highest correlation with mean nest temperatures. There seemed to be an interaction between rainfall data and zones. When the mean daily rainfall was low (< 8 mm/day), the nests in the open zone showed higher female sex ratios than ones in the vegetation/border zone in general. On the other hand, as the mean daily rainfall increased, there was no apparent difference of sex ratios between the zones. R-111 pooled zone data (pseudo R2=0.17) fit a regression model almost as well as incubation period and mean sand temperature data did. If only data for the nests in the open zone were used, the pseudo R2 increased to 0.27. However, variation of the data was still high for the whole range. Bottom depth of clutch Bottom depth of sample nests varied from 59 to 105 cm (average=78.3, SE=1.4, n=52). In general, sand temperatures decreased with depth of the nest. However, no significant correlation between mean nest temperatures and bottom depth was observed (p=0.215, n=51). As expected, there was no apparent trend between the sex ratios and the bottom depth of the sample nests (Figure 37).



PAGE 1

Figure 19. Factors causing total destruction of green turtle sample nests in 1986, 1988, and 1989 at Tortuguero, Costa Rica.



PAGE 1

133 12 1986 10 8 O 1 6 l) LUJ 4 UJ --2 0 0 r¢ 12LL a_ 1988 8 OlO 6 m -4 LU 2 ULj cc U) 12I U) 1989 LU 10 z U 0 8 cr LU M 2 6 z 4 2 0 a a a a a a e JUN. JUL. JUL. AUG. AUG. SEP. SEP. OCT. OCT. NOV. NOV. 16-31 1-15 16-31 1-15 16-31 1-15 16-30 1-15 16-31 1-15 16-30 Figure 27. Temporal distribution of mammalian predation on green turtle nests at Tortuguero, Costa Rica.



PAGE 1

96 Table 13. Frequency of excessive rainfall events (>140mm/day) at Tortuguero, Costa Rica, from 1978-1989. Data from Instituto Meteorogico Nacional, Costa Rica. Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 1978 n/d n/d n/d n/d 1979 1 1980 1 1 1981 1 2 1 1982 2 1 1983 2 1 1984 2 1 1 n/d ------------------------------------------------------------1985 1 1986------------------2--------1--------1-------1--------1--1986 2 1 1 1 1 ------------------------------------------------------------1987 1 1 1 1 1988 1 1 1 1989 1 1 Frequency of excessive rainfall event per year .36 0 .09 .09 .17 0 .17 .17 .08 .50 .41 .36 n/d: no data available



PAGE 1

56 this trend, but substantial nesting activity was extended at least until the middle of October in the 1986 and 1989 seasons. The number of nests deposited in October (662) exceeded that in July (389) in the 1989 season. Carr et al. (1978) presented data on the mean seasonal profile of nesting arrivals between 1956-1976, in which the peak of nesting arrival occurred at the end of August. However, because green turtles deposit multiple nests (1-7) within a season (Carr et al., 1978), the pattern of nest number profile cannot be clearly estimated from their figure. So far, the only available information on the full seasonal profile of green turtle nests at Tortuguero is for the 1977 season taken by Fowler (1979). She noted that nesting began in June, peak activity occurred in early August, and only a few turtle nests were found by late November in 1977; she showed the actual nest profile only from 13 July through 14 September. The three years of this study also showed the beginning of nesting activity to be sometime in June, and minimum activity to be in November, but the timing of peak activity was different; in this study the peak was in the latter half of August in 1986, in the first half of September in 1988, with a plateau from the latter half of August through September in 1989. The principal nesting season in Tortuguero green turtles seems to be confined to June through October, with peak activity sometime between August and September, but the actual dates vary from year to year.



PAGE 1

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. e A. Mortimer Courtesy Assistant Professor of Zoology This dissertation was submitted to the Graduate Faculty of the Department of Zoology in the College of Liberal and Science and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1992 Dean, Graduate School



PAGE 1

37 green turtle eggs. Altogether, 5.4% of the eggs in the open zone and 20.5% of eggs in the vegetation/border zone were destroyed throughout the three years by these causes. Significantly more eggs in the vegetation/border zone were destroyed than in the open zone for all three years (3.7% vs 14.0% in 1986, 6.2% vs 15.9% in 1988, 6.2% vs 31.4% in 1989; X2=188.9, p<0.0001, 1986; X2=209.4, p<0.0001, 1988; X2=1246.0, p<0.0001, 1989). Coatis, Nasua narica, were responsible for the majority of nest loss by mammal predation. However, because it was sometimes difficult to confirm the mammal species on the sample nests only by their footprints, all nests excavated by mammals were classed as "Destroyed by mammals." Coatis are diurnal predators on green turtle nests in Tortuguero. In the protected park area, solitary or more often a band of coatis were frequently sighted walking along the upper part of the beach during the daytime. The band membership typically consisted of two to six cubs and several adults. They generally excavated a series of adjacent nests along the border of the vegetation. Although they rarely consumed all the eggs in a nest, the rest of the eggs were soon eaten by other animals. For the sample depredated nests, 72.7% of nests depredated by mammals (n=33) did not produce any hatchlings. Raccoons, Procyon lotor, were also responsible for destruction of nests, but to a lesser extent. The extent of damage was not accurately obtained because of the difficulties of species identification. For the depredated sample nests, only one nest in the vegetation/border zone in 1989 was confirmed to be damaged by



PAGE 1

46 Seasonal Variation and Zone Effect in 1988 In 1988, the sex ratios varied from 0 to 100% female for the sample nests in the open zone and from 0 to 89.5% female for the sample nests in the vegetation/border zone (Table 6). Figure 30 shows the seasonal variation of sex ratios in the 1988 season. The extent of seasonal fluctuation was more apparent in the open zone than in the vegetation/border zone. For analyzing the effects of seasonal period (half-month interval) and nest location (the open zone vs the vegetation/border zone) on sex ratio throughout the season, two-way factorial ANOVA with unequal replication was applied (Table 7). Data were transformed (p'=1 /2(arcsin1(X/(n+1 )+arcsin (X+1 )/(n+1)) (Zar, 1984). An Fmax test showed homogeneity of variances of the transformed data. The analysis revealed significant differences in seasonal fluctuation of sex ratio and in effects of Zone, but no significance in Period x Zone interaction. Fisher's procedure of least significant difference revealed a significant seasonal difference of sex ratios in the open zone (Figure 31), but not in the vegetation/border zone. In the open zone, the sample nests deposited in the latter half of July had significantly more females than during other periods within the season except for the nests deposited in the first half of August. From the end of August through September 1988, the sand temperatures, at least in the open zone, rose above the pivotal temperature (28.50C), associated with less rainfall (Figure 6, Figure 13). Most sample nests in the open zone deposited from the latter



PAGE 1

43 are not known. Possibly, some previous physical disturbance, such as a minor inundation, caused the arrested development of some of these eggs at a later stage. Over the three years, 3.3% of the eggs in the open zone and 1.9% of the eggs in the vegetation/border zone did not show any sign of visible embryos or blood formation. I did not use a white circle or patch on the egg shell as a criterion of development. Therefore, the above figures probably overestimate the percentage of infertility. Over the three years, 5.3% of the eggs in the open zone and 4.3% of the eggs in the vegetation/border zone were rotten with intact shells or were ruptured at the time of excavation. Most late stage embryos were not included in this category because the remaining shell and bones were detectable even in a ruptured shell. However, some of the earlier stage developing eggs and infertile eggs were possibly included in this category. Analysis of Sex Ratio Sexed Samples A total of 55 nests (12 in 1986 and 43 in 1988) were sampled for sexing (Table 6), and the temperatures of 52 of those nests were monitored successfully. Twenty eggs were collected as a subsample from each clutch during 50-60 days except from one sample nest in 1986. Because I missed sampling eggs in this nest before the emergence of the clutch, only the two remaining hatchlings in the



PAGE 1

Figure 13. Seasonal fluctuation of sand temperatures at a depth of 60 cm and 80 cm in the open zone and in the vegetation/border zone in 1986, 1988, and 1989 at the Tortuguero beach, Costa Rica.



PAGE 1

101 2001986 100 0 JUN. JUL. AUG. SEP. OCT. NOV. DEC. E E 200 1988 .-J -j U z 100 c JUN. JUL. AUG. SEP. OCT. NOV. DEC. 200 1989 100 JUN. JUL. AUG. SEP. OCT. NOV. DEC. Figure 4. Daily rainfall from 15 June through 10 December 1986, 1988, 1989 at Tortuguero, Costa Rica. Fresh water flooding occurred at the rain events marked *.



PAGE 1

17 The temperature readings were taken once every two to three days, with a Bairly BAT12 thermocouple meter. To determine the amount of diurnal fluctuation, several monitoring sessions lasting 24 h each, at 2 h intervals, were conducted during the study: 13 August and 29 September 1988, and 12 August and 14 October 1989. Statistical Analysis Both parametric and nonparametric statistical tests were used for data analysis. The arcsine transformation was applied to the percentage data (e.g., emergence success and sex ratios). Whenever the required assumption of homogeneity of variance was met by using the Fmax test, parametric statistical tests were utilized. If the assumption was not met, nonparametric tests were substituted. To test binomial data (e.g., nest site selection between the two zones), Chi-Square tests were utilized. The rejection level for the null hypothesis in all tests was alpha = 0.05. For those nests where complete data sets were not available, the existing data were included in analyses whenever possible. Each statistical test is mentioned in the results section.



PAGE 1

24 activity varied considerably with a similar seasonal trend throughout each season. Generally, there were more calm days during the middle of the season. For the three years combined, September was the month most likely to have a calm day. The amount of monthly rainfall was negatively related to the number of calm days (wave height 0.5 m) in the month (Y=17.147-1.656X, R=0.663, n=15; t=3.19, p=0.007; Figure 11). The greater the rainfall, the higher were the waves on the beach. Continuous days of calm waves resulted in sand accretion on the lower part of the beach, and the entire beach widened gradually. On the other hand, when the wave activity was higher (> 0.5 in), beach erosion was apparent and the shape of the beach changed rapidly. Heavy beach erosion frequently constructed beach platforms (up to 1.5 m height) along part of the beach. Green turtles either gave up trying to nest in these areas, or they crawled over the platforms and laid their clutches in a site protected from waves. Very few females nested below the platforms at the time of high waves. However, nests that had been deposited before the high platforms were formed or advanced well inland were washed away or completely inundated. Daily Fluctuations of Sand and Nest Temperatures To determine the extent of daily fluctuation of sand and nest temperatures, temperatures along two transects (two points in the vegetation/border zone, five points in the open zone) and temperatures at the center of clutches (14 nests in each zone) were



PAGE 1

34 For assessing the cause of embryonic death, when the timing of inundation reasonably matched the stage of the arrested embryos, I assumed these eggs were killed by the inundation event. In the 1988 season, because two inundations (6 October excessive rain and 21-22 October Hurricane) occurred so close to each other, I could not determine how much of egg mortality was caused by each inundation. Abiotic factors Among abiotic factors, erosion and inundation by surf, excessive rainfall resulting in inundation from ground water, and Hurricane Joan were mainly responsible for reducing egg survivorship. These abiotic extremes, as a whole, were responsible for loss of 18.5% of eggs in the open zone and 20.0% of eggs in the vegetation/border zone throughout the three years (the percentages of egg loss for the three years were calculated as the weighted mean in each zone). In addition to these abiotic extremes, artificial debris prevented a few hatchlings from emer ging. Beach erosion washed away 5.7% (open zone) and 2.9% (vegetation/border zone) of the eggs throughout the three years. The damage from erosion showed considerable yearly variation, and the 1986 season suffered the highest loss (15.3% of eggs in open zone, 8.7% in vegetation/border zone in the 1986 season). Surf inundation did relatively minor damage to the eggs throughout the three years (1.4% in open zone, 1.3% in vegetation/border zone). Hurricane Joan, whose center passed about 180 km off the coast of the Tortuguero beach in the Caribbean Sea during the night


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EXXNVCLF1_U631IK INGEST_TIME 2018-08-09T18:16:24Z PACKAGE AA00063501_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

75 nests were exposed to less environmental fluctuation than were those in my study. The first multiple regression model (pseudo R2=0.24) with the significant factors selected (mean nest temperatures, incubation period, rainfall data, and zone) to predict sex ratio was only a slight improvement over the nest temperature model (pseudo R2=0.20) because rainfall data, incubation period and zonation are correlated with nest temperature data. That is, most factors only provided redundant information to various degrees. As I discussed above, the genetic factor might also be responsible for the poor fit of the data to the model. It should be noted that another multiple regression model (pseudo R2=0.23) using rainfall data, zone and their interaction explained the sex ratios of hatchlings as well as the nest temperature model did. Rainfall fluctuation seemed to more strongly influence the sex ratio of nests in the open zone than in the vegetation/border zone. In the open zone, sex ratios ranged from 100% male to 100% female, whereas in the vegetation/border zone the sex ratios of most nests showed less variation, ranging only from 100% male to slightly female biased. My study suggests that data describing rainfall and zonation could be a rough predictor of sex ratio of green turtle hatchlings at Tortuguero. Furthermore, it might be possible to estimate very roughly the primary sex ratio using the total amount of rainfall through the entire incubation period. To construct a useful model, more data need to be gathered on the correlation between rainfall and sex ratio produced in each zone.



PAGE 1

33 In addition, predation loss of three nests by mammals and one nest by ghost crabs further decreased the emergence success during this period. Abiotic and Biotic Factors Affecting Eag Survivorship Fate of eggs from sample nests Table 5 and Figures 23-24 show the fate of green turtle eggs from sample nests in 1986, 1988 and 1989. The emergence success of eggs from the sample nests throughout each season was 54.6% (n=2915) in 1986, 59.5% (n=3899) in 1988, and 75.5% (n=5890) in 1989 in the open zone; comparable figures for the vegetation/border zone are 46.4% (n=2801) in 1986, 43.0% (n=5715) in 1988, and 58.5% (n=6216) in 1989. Criteria for assessing the cause of egg mortait The category of "Destroyed by mammals" includes all eggs lost because of initial damage by mammals. For example, when coatis excavated green turtle nests, they rarely consumed all the eggs. However, soon other animals such as black vultures, ghost crabs and ants often destroyed the remaining eggs in the half excavated nest. In this case, I categorized the entire clutch as lost under "Destroyed by mammals." If the partially depredated nest successfully produced hatchlings, the numbers of missing egg shells (clutch size minus counted egg shells at excavation), which were assumed to be removed from the nests by mammals, were classed under this category. The same principle was applied to the category of "Destroyed by female turtles."



PAGE 1

14 During 1986 and 1988, clutch size was counted during deposition, and emergence success was determined directly. In 1989, when the clutch size was not counted, clutch size was estimated by the number of remaining eggshells. In most undisturbed nests, the hatched shells remained in one piece. If the shells were fragmented, the pieces were put together to represent one shell. Fowler (1979) used this method to estimate the clutch size of green turtles in Tortuguero and found that the error was no more than 8 eggs. In the 1989 season, for nests that were partially destroyed or completely lost by any disturbance, the mean clutch size from 1989 (107.1) was used as the clutch size to calculate emergence success. Sampling of Nests to Determine Sex Ratio I originally planned to obtain five nests to determine sex ratio in each zone and for each half month period in the nesting season. Night sampling was conducted seven to 10 times during each half month period to locate nesting females. Only females during the early stage of nesting behavior (before digging the egg chamber) were sampled. During the deposition of eggs, a thermocouple probe was placed at the approximate center of the nest. For comparison, and to allow measurement of metabolic heat from the eggs, another thermocouple probe was placed at the same depth 1 m along the beach from the egg chamber. To protect the temperature-monitored nests from activities of other nesting females, several logs found on the beach were



PAGE 1

69 Island, Australia, Limpus et al. (1983) observed that the north facing beach on the island was warmer than the beach that faces south. Consequently, the northern beach produced a significantly higher proportion of green turtle females (63.1% females) than the cooler southern beach (29.5% females). In Barbados, West Indies, the warmer west coast beach produced more hawksbill females (80.6% females), whereas the cooler south coast produced more males (32.7% females) (Horrocks and Scott, 1991). Temporal Effects on Sex Ratio At Tortuguero, a short dry season in September (mean 320 mm) usually occurs between the wetter periods of July through August, and October through November. For the Tortuguero green turtles, the impact of this dry season seems to be very important because a large proportion of the nests are deposited during August and the most critical period for these nests occurs in September. A dry September produces enough females to avoid a highly skewed sex ratio for the entire season. Seasonal fluctuation in sex ratios resulting from the shift between rainy and dry periods was observed at two tropical beaches: Surinam for green turtles and leatherback turtles (Mrosovsky et al., 1984) and French Guiana for leatherback turtles (Rimblot-Baly et al., 1987). In Sarawak, another tropical beach, incubation periods of green turtle nests varied with the rainy -dry weather cycle (Hendrickson, 1958). Therefore, Standora and Spotila (1985) suspected that the sex ratios of green turtles at Sarawak also show



PAGE 1

67 success of the nests that produced hatchlings, and she found no seasonal differences in the emergence success at Tortuguero. Fowler did not report any flooding events. Seasonal fluctuation in egg survivorship of Tortuguero green turtle nests might not be apparent, except in years of hurricanes and excessive seasonal flooding. Spatial Effects on Sex Ratio At the Tortuguero, beach, a distinct thermal difference at the depth of green turtle nests between the open zone and the vegetation/border zone was very consistent throughout the incubation season in the three years of this study. However, the effect of these thermal zones on sex ratio depends on the relation of sand temperatures in each zone to the pivotal temperature. It appears that this spatial effect on sex ratio can be detected only during a period of dry weather. Seasonal change in this spatial effect was clearly observed in the 1988 season. The mechanism may be the following. During prolonged dry weather, sand temperatures in the open zone clearly exceed the pivotal temperature, whereas sand temperatures in the vegetation/border only approach or slightly exceed the pivotal temperature. Consequently, most nests in the open zone show female-biased sex ratios, whereas nearly equal or moderately female-biased sex ratios occur in the vegetation/border zone. On the other hand, during prolonged wet weather, sand temperatures in both zones decrease below the pivotal temperature. As a result, the difference in sex ratios between zones becomes



PAGE 1

20 0O Y = 17.147 -1.656X R = 0.663 18 n = 15, t =3.189, p = 0.007 0 016 z 14 0 -0 12 0 0 0 100 o a1 8oVI 8 0 OE e-D 0 56 0 WI 4 0 > OO 2 z0 1 0 -0 200 400 600 800 1000 MONTHLY RAINFALL (MM) Figure 11. Relationship between the number of calm days in a month and monthly rainfall from July through November in 1986, 1988, and 1989 at Tortuguero, Costa Rica.



PAGE 1

68 negligible, and most nests in both zones show male-biased sex ratios. The latter case was observed throughout the 1986 season. In 1980, Spotila et al. (1987) found a distinct spatial difference in sex ratios of Tortuguero green turtle nests between the open zone (mean 67.4% of females, n=9) and the vegetation zone (7.6% of females, n=6). The rainfall record revealed that the weather was very dry (281 mm in August, 313 mm in September) throughout the critical periods of their samples. In Surinam, the different topographic beach zones showed small thermal differences (0.5-1.00C), and the location had little effect on sex ratio of green turtles through the season (Mrosovsky et al., 1984). This is probably due to the fact that the vegetation is composed of sparse and low shrubs in Surinam, compared to Tortuguero's dense vegetation (Schulz, 1975; P. Dutton, pers. comm.). The existence of dense vegetation and associated shade is one requirement for a spatial effect on sex ratios. The barrier island beaches on the east coast of Florida have primary dunes at the upper part of the beach, but lack dense vegetation. Witherington (1986) noted that the thermal differences on the beach must be slight because there is no significant difference in incubation periods between beach zones for loggerhead nests. The nesting beaches of Ascension Island are virtually without vegetation (Mortimer, 1981), but differences in the mineral composition of the various beachs (Mortimer, 1990) produce different thermal regimes (Hays and Mortimer, MS in prep.). At some island rookeries, a significant effect of the orientation of the coast on sex ratio has been reported. At Heron



PAGE 1

130 OPEN ZONE 7.1% 11.4% 2.6% 1.7% 1.1% 63.2% 3.3% 3.6% VEGETATION/BORDER ZONE 4.2% 16.2% 49.3% 1.5% 4.8% 1.9% 5.2% 3.1% DAMAGED BY WAVE ACTION DAMAGED BY FLOODING/HURRICANE JOAN 0 PREDATION BY MAMMALS O PREDATION BY CRABS, OTHERS E DESTROYED BY TURTLES 1 NO APPARENT DEVELOPMENT 0 CAUSE OF EMBRIONIC DEATH UNKNOWN OTHER O HATCHLNG EMERGED



PAGE 1

Figure 3. Seasonal fluctuation of monthly minimum and maximum air temperatures from July through November 1986, 1988, 1989 at Tortuguero, Costa Rica.



PAGE 1

45 turtles. For this study, I assumed 5 days as the emergence lag in Tortuguero green turtles, and I subtracted 5 days from the deposition to emergence period to obtain an approximate developmental period for each clutch. Mean temperature during the middle third of the development period was calculated for each clutch. Figure 29 shows the relationship between sex ratios of sample nests and mean nest temperatures in the center of the clutch and mean sand temperature (1 m away from the clutch, at the same depth) during the critical period. The proportions of females were positively correlated with ambient temperatures. Both sexes were produced over the entire temperature range of the sample nests (mean nest temperature range=27.0-30.70C, mean sand temperature range=26.4 -30.20C). Because metabolic heat is produced during the middle of development, the mean nest temperatures were slightly higher than the sand temperatures at the same nest during the critical period (mean difference=0.71C, SE=0.05, range=0.18-1.48, n=41). As a result, the regression of sex ratios and mean sand temperature was shifted down approximately 0.70C on the function of temperatures, as compared to mean nest temperature data. The pivotal temperatures (expected equal sex ratio) were calculated by logistic regressions as 29.40C for the mean nest temperatures, and 28.50C for the mean sand temperatures (see Prediction of Sex Ratio from Independent Variables for a detailed explanation of the logistic regression model). For both data sets, there was a substantial amount of variation in sex ratio over the range of incubation temperatures.



PAGE 1

35 of 21 through the morning of 22 October 1988, caused substantial damage to green turtle nests. When Hurricane Joan passed, the Tortuguero beach was at the outer margin of a 70 km/h wind speed zone (Instituto Meteorologico Nacional, Costa Rica). During the first beach census after the hurricane, I confirmed that the waves apparently washed up to the vegetation area along most of the beach and that all developing sample nests were covered by waves to various extents. However, beach erosion was not severe and none of the sample nests was washed away. All damage by Hurricane Joan on the sample nests was caused by surf inundation. Hurricane Joan brought only a little rainfall to the Tortuguero beach (total 25 mm during 21 to 24 October 1988). It is not known whether the ground water raised by the high waves or the surf salt water itself suffocated the eggs. The extent of damage caused by Hurricane Joan in 1988 was not accurately assessed because the damage to several nests was inseparable from that caused by excessive rain on October 6. The minimum estimated damage by Hurricane Joan was 1.6 % egg loss in the open zone and 10.8% egg loss in the vegetation/border zone. However, if I include the undetermined dead eggs killed by either the hurricane or the excessive rain, the figures increase to 9.5% egg loss in the open zone and 13.3% egg loss in the vegetation/border zone. No other hurricane threatened the Caribbean coast of Costa Rica during the three-year study. Freshwater inundation by ground water, mainly associated with excessive rainfall, was a major and constant abiotic factor causing mortality of eggs throughout all three years. All developing



PAGE 1

114 32 1986 31 30 2928 27 -* 2625 S 24 JUL. I AUG. I SEP. I OCT. I NOV. I DEC. O III 32M. .1988 31 30 LLJ 29", 28' 27 24 JUL. AUG. I SEP. OCT. I NOV. I DEC. O 32 1989 313029 727" 25 25 2 JUL. I AUG. I SEP. I OCT. I NOV. I DEC. -OPEN ZONE, 60 CM OPEN ZONE, 80 CM ************** VEGETATION/BORDER ZONE, 60CM -..* *.-*.*-VEGETATION/BORDER ZONE. 80CM



PAGE 1

Figure 17. Yearly and seasonal fluctuation of green turtle nest distribution between the open zone and the vegetation/border zone in 1986, 1988, and 1989 at Tortuguero, Costa Rica. Within each season, Tukey-type multiple comparison test for proportional data (Zar,1984) ranked from the highest to the lowest the proportion of nests deposited in the vegetation/border zone. Those months that do not share a common number have significantly different proportions.



PAGE 1

26 temperatures at depths of 60 cm and 80 cm and nest temperatures in both zones showed a similar trend of daily profiles. While the period of highest temperatures was not distinct, the lowest temperatures were recorded from 1000 to 1200 in the morning. Times when the temperature was closest to the 24-hour mean were around 0600 and 1500. Most of my temperature readings were conducted during the two periods between 0700 to 1000 and between 1500 and 1700. Overall, the temperatures collected for this study were probably not far from the values of 24 hour means. Seasonal Fluctuation of Sand Temperatures In Tortuguero, local meteorological conditions affected the fluctuations of sand temperatures. Rainy days resulted in lower sand temperature, while sunny days resulted in higher temperatures. In general, amount of rainfall was inversely related with sand temperatures. Being associated with variable rainfall (Figure 4), the sand temperatures showed considerable seasonal fluctuations for the three years (Figure 13). Occasional heavy rain events, particularly those that caused freshwater flooding on the beach (Figure 4), resulted in rapid and substantial short-term declines of sand temperatures along the entire beach (Figure 13). In addition to such flooding rain events, overcast days with moderate rainfall sometimes resulted in a substantial decline of sand temperatures. A continuous rainy event from 21 through 25 July 1988 accumulated 165 mm of rainfall, and lowered the sand temperature up to 2.90C at 60 cm depth in the open



PAGE 1

I analyzed the emergence success of 49 clutches in 1986, 88 in 1988, and 113 in 1989. Emergence success from these nests showed a bimodal distribution (0% and >70%) and did not differ significantly between the zones. Abiotic factors including surf, freshwater inundation, and hurricane killed approximately 20% of sample eggs. Biotic factors including predation and turtle digging killed 5% of sample eggs in the open zone and 19% of sample eggs in the vegetation/border zone. The relative importance of each factor varied from year to year. Mammals destroyed four times as many nests in the vegetation/border zone as in the open zone. My estimation of hatchling emergence success was 57.2% (1986), 45.7% (1988) and 66.6% (1989). A sample of hatchlings was collected for direct sexing from each of 12 nests in 1986 and 56 in 1988. The sex ratios of hatchlings fluctuated intra-seasonally in 1988, but not in 1986. The estimated overall proportion of female hatchlings was 10.1% in 1986 and 40.6% in 1988. 1 used single and multiple logistic regression models to test environmental and clutch variables for the prediction of sex ratio. The pivotal temperatures were 29.400 for mean nest temperature, and 28.500 for mean sand temperature. Mean nest temperature during the middle third of development was the best fit model (pseudo R2=0.20) in the single regression analysis. Mean nest temperature, incubation period, rainfall, and nesting zone were significant parameters for multiple regression (pseudo R2=0.24). However, even the model with the best fit was still not reliable for predictive purposes. vii



PAGE 1

50 Logistic Regression Model For dichotomous dependent variables (i.e., female or male), a logistic model (F(Z)=exp(Z)/(l+ exp(Z)), z=B1+BO*X) is an attractive alternative to the linear probability model because the logistic model satisfies 0-1 constraint on probability unlike the linear specification, and provides a smooth symmetric S-shaped curve (Aldrich and Nelson 1984). Because this model is one of the most widely used nonlinear models, the availability and flexibility of computer programs are better. The logistic models were calculated with maximum likelihood estimation, and pseudo R2 was computed as (-loglikelihood for Model) / (-loglikelihood for C Total) (JMP version 2.02, SAS Institute Inc., 1989). Logistic Regression with a Single Variable Nest and sand temperatures Mean nest temperatures were highly correlated with mean sand temperatures during the critical period (mean nest temperatures=0.99 + 1.02 mean sand temperatures, R2=0.92, n=41, p<0.001). Mean nest temperature data showed the best fit (pseudo R2=0.20) to a logistic model among the variables for a single regression model (Table 9, Figure 33). However, the variability of sex ratios with mean nest temperatures was still very high. Particularly at the middle range of temperatures, residuals spread up to 0.4 probability (Figure 33). Mean sand temperature data



PAGE 1

57 Seasonality of Environmental Parameters Because the Tortuguero beach is located in the equatorial zone, seasonal fluctuation in air temperatures is very low relative to the high fluctuation in rookeries in the temporal zone. However, there is predictable seasonal change in several environmental factors that are important in influencing egg survivorship, sex ratio, and possibly nest site selection between different thermal zones in Tortuguero green turtles during their incubation period (June through December). The rainfall record indicated that there are two cycles of rainy/dry seasons in Tortuguero and that a high level of rainfall fluctuation occurs at the middle of the green turtle incubation period; rainfall decreases in September. Since sand temperature was inversely related to rainfall, the warmer sand temperature on the beach was expected to occur in September relative to the rest of the season. This tendency was observed during all three years of this study. Although seasonal trends of wave activity varied among the three study years, the wave activity was relatively low in September for each season. The positive correlation found between wave activity and rainfall suggests that fluctuation of wave activity also has a general seasonal trend; September has the most calm days at Tortuguero. At Tortuguero, the mean tidal range is 0.2 m (Limon/Bluefield, Tide Table 1986: East Coast of North and South America). Therefore, the high fluctuation of wave size (almost none to >2.0 m) has more influence on the width of the beach than the tide



PAGE 1

BIOGRAPHICAL SKETCH Kazuo Horikoshi was born in Tokyo, Japan, on 9 July 1956 to parents Shigeru and Kiku Horikoshi. He received a Bachelor of Fisheries in aquaculture from the Tokyo University of Fisheries, Tokyo, Japan in 1979. He continued his graduate study at the Tokyo University of Fisheries and was awarded a Master of Fisheries in aquaculture in 1982. He enrolled in the Ph.D. program at the University of Florida in 1984. 158



PAGE 1

74 Miller (1985) emphasized that temperature, hydric environment, and gas exchange interact cooperatively to influence embryonic metabolism. Standora and Spotila (1985) suggested that while temperature is the primary environmental factor that determines sex in sea turtles, other factors, such as osmotic stress, and 02 and C02 levels, could play a role in sex determination in the temperature range over which both sexes are produced. The impact of those other factors, however, has yet to be demonstrated. The influence of hydric conditions on sex determination in painted turtles, Chrysemys picta, was inconsistent even within one population (Gutzke and Paukstis, 1983; Packard et al., 1991). Oxygen concentration did not influence the sex ratio of red-eared slider turtles, Trachemys scripta (Etchberger et al., 1991). It is highly probable however that any environmental influence and their possible interactions in determining sexes of green turtle hatchlings vary at the inter-clutch level because of the highly heterogeneous environment over time at the Tortuguero beach. The previous study of the relationship between sex ratio and mean incubation temperature of natural nests in the 1980 season at Tortuguero (Spotila et al., 1987) showed less variability than did this study. This is probably due to their smaller sample size (15 nests) and also their shorter sampling period. Most of their samples were collected within a half month period, whereas my sample period extended over three months in each of two years. It is possible that their mean nest temperature values were better correlated with sex ratio than were mine because most of their



PAGE 1

Fate of eggs from sample nests ....................................... 33 Criteria for assessing the cause of egg mortality ... 33 Abiotic factors .......................................................................... 34 Biotic factors ............................................................................. 36 Other Categories of Mortality ...................................................... 42 Analysis of Sex Ratio ............................................................................. 43 Sexed Samples ..................................................................................... 43 Sex Ratio vs Temperature .............................................................. 44 Seasonal Variation and Zone Effect in 1988 .......................... 46 Overall Sex Ratio in 1988 .............................................................. 47 Seasonal Variation and Zone Effect in 1986 .......................... 48 Overall Sex Ratio in 1986 .............................................................. 49 Prediction of Sex Ratio from Independent Variables .............. 49 Logistic Regression Model ............................................................. 50 Logistic Regression with a Single Variable .......................... 50 Nest and sand temperatures ................................................ 50 Incubation period ...................................................................... 51 R a infa ll ......................................................................................... 5 1 Bottom depth of clutch .......................................................... 52 Multiple Logistic Regression Model ........................................... 53 4 DISCUSSION .................................................................................................... 55 Nesting Density and Tem poral Nest Distribution ...................... 55 Seasonality of Environm ental Parameters ................................... 57 Spatial Distribution between the Zones ........................................ 58 Overall Reproduction Rate ................................................................... 61 Spatial Effects on Mortality Factors .............................................. 63 Seasonal Fluctuation of Em ergence Success ............................... 65 Spatial Effects on Sex Ratio ............................................................... 67 Tem poral Effects on Sex Ratio .......................................................... 69 Annual Variation in Prim ary Sex Ratio .......................................... 70 Predictability of Sex Ratio by Environmental Factors ........... 71 5 SUMMARY AND CONCLUSIONS ..................................................................... 78 LITERATURE CITED ................................................................................................ 149 BIOGRAPHICAL SKETCH ...................................................................................... 158 v



PAGE 1

139 100 OPEN 4 6 ZONE 80 44 60 -4 3 40 20 0 100 VEGETATION/BORDER ZONE 80 00 3 060 2 Lu40 -43 2 20 100 100TOTAL 80 9 88 60 405 20 0 I I Io JUL. JUL. AUG. AUG. SEP. SEP. 1988 1-15 16-31 1-15 16-31 1-15 16-30 PERIOD OF NESTS DEPOSITED



PAGE 1

25 monitored at two-hour intervals for 24 hours for four days during the study. Those clutches monitored averaged 23.0 days from deposition (SE=2.2, range=3-54, n=28) at the time of temperature measurement. Two of the days (13 August and 29 September 1988) had no rainfall and two days (12 August and 14 October 1989) had minor rainfall (2 mm and 25 mm, respectively) during the 24-hour monitoring period. In general, these four days are within the range of normal weather in Tortuguero. The sand temperatures at depths of 60 cm and 80 cm were relatively stable in both zones (Table 1). Mean ranges of daily fluctuation in these four categories were from 0.410C to 0.490C. The range of individual monitoring points on the transects was from 0.10C to 0.80C. The daily fluctuations in temperatures were not significantly different between the two zones, or between depths of 60 cm and 80 cm (two-factor ANOVA; zone, p=0.3515; depth, p=0.4716). The daily fluctuation of temperatures in the center of the clutches showed low ranges similar to the fluctuation in sand temperatures along the transects (Table 1). The range of daily fluctuation averaged 0.460C in the open zone and 0.490C in the vegetation/border zone with the same individual range (0.2-0.80C). The mean daily fluctuations of nest temperatures were not significantly different between the two zones (unpaired t-test, t=0.377, p=0.710). Although daily temperature profiles varied with different weather, mean temperature profile among the four separate days showed a general trend of daily fluctuation (Figure 12). Sand



PAGE 1

77 methods including laparoscopy and testosterone levels (Wibbels et al., 1989) are useful in sexing larger immature sea turtles, but not hatchlings. Recently, Demas and Wachtel (1989) reported a sexspecific satellite DNA, called "Bkm", in green turtles and Kemp's ridleys. Because only a few drops of blood are needed to detect this molecule, it could be used to harmlessly determine the sex of sea turtle hatchlings. At the present time, however, this molecular method is probably too expensive and labor-intensive to process the large numbers of hatchlings needed to estimate the primary sex ratio on a beach where many variables affect sex ratio, such as on the Tortuguero beach.



PAGE 1

146 A 1.0 0 0 0 0 0 0.8 o 0 0 0.6" .o o 0 0.4 0 0 a o o 00 0.2 -• 1 0 0.0o e I


PAGE 1

9 3 to mile 18), has been protected as a part of Parque Nacional Tortuguero since 1975. To investigate egg survivorship and primary sex ratio of the Tortuguero population of green turtles, I selected two miles of beach (mile 6 to mile 8) for my study area. This area is one of the most highly utilized areas by nesting green turtles. Approximately 17% of nests laid on the beach are deposited in this two-mile section (Carr et al., 1978). The beach shape, vegetation, and animals in the central area remain undisturbed because the area is successfully protected as part of the national park. There is no human habitation in the park area. Railroad vine, lpomoea pes-caprae, and beach lily, Hymenocallis littoralis, predominate on the rear of the open beach. Behind the beach, cocoplum, Chrysobalanus icaco, seagrape bushes, Cocoloba uvifera, and coconut palms, Cocos nucifera, are common. Inland, there is a well developed and largely undisturbed tropical wet forest. The dense canopy of the inland forest is high (> 8 m). Weather Record and Ground Water Ambient temperature and rainfall data were collected with standard meteorological instruments. In 1986 and 1989, data were collected at the Green Turtle Research Station adjacent to the beach, located about 9 km north of the study area. During the 1988 season, data were taken at the study area from July through the middle of October, and later at the station. Rainfall and maximum



PAGE 1

93 Table 10. Logistic regression model of fluctuation in sex ratio for green turtle hatchlings with all independent variables. Final model follows in Table 11. Data collected on Tortuguero, Costa Rica, 1986 and 1988. Parameters Slope (SE) ChiSqure P Mean nest temperature 0.5005 (0.1314) 14.50 0.0001 Incubation day -0.0801 (0.0330) 5.94 0.0149 Rainfall (R-1) -0.0136 (0.0234) 0.34 0.5607 ns Rainfall (R-11) 0.0510 (0.0454) 1.26 0.2617 ns Rainfall (R-Ill) -0.1359 (0.0585) 5.40 0.0202 Bottom depth of nest 0.0097 (0.0090) 1.16 0.2815 ns Zone 0.4701 (0.1958) 5.76 0.0164 Intercept -9.9764 (4.8321) 4.26 0.0390 -2LogLikelihood=292.3, df=7, overall P<0.001 pseudo R2=0.241, n=924



PAGE 1

95 Table 12. Logistic regression model in sex ratio for green turtle hatchlings with rainfall data and nesting zone variables. Data collected at Torttuguero, Costa Rica, 1986 and 1988. Parameters Slope (SE) ChiSqure P Rainfall (R-1Il) -0.1104 (0.0225) 24.03 0.0001 Zone 2.7738 (0.4011) 47.82 0.0001 (R-1ll) X Zone -0.1510 (0.0310) 23.75 0.0001 Intercept 0.2680 (0.2870) 0.87 0.3503 -2LogLikelihood=272.2, df=4, overall P<0.0001 pseudo R2=0.225, n=924



PAGE 1

Figure 16. Seasonal distribution of green turtle nests in 1986, 1988, and 1989 in the two-mile study area at Tortuguero, Costa Rica. Data in the latter half of June 1986 were not available.



PAGE 1

49 Overall Sex Ratio in 1986 Because a significant seasonal effect was not detected (as mentioned above), overall proportion of females was calculated as a mean value of all sexed sample nests to be 10.1% (n=11) in the 1986 season. The number of sexed sample nests was biased towards nests in the open zone (8 of 12, 66.7%) as compared to the observed nearly 1:1 distribution between the two zones (Figure 17). Because the open zone showed higher sand temperatures than the vegetation/border zone (Figure 13), the 10.1% proportion of females may be an over-estimate. Prediction of Sex Ratio from Independent Variables To assess predictability of sex ratios of Tortuguero green turtle hatchlings from independent variables, simple and multiple logistic regression models were applied. Linear correlations between nest temperature, an assumed main environmental factor, and other variables were also analyzed. For the independent variables, mean nest temperatures and mean sand temperatures during both the critical period and the incubation periods, three different sets of rainfall data, bottom depth of nests, and zone of the nests were included. The data set included 52 nests consisting of 944 sexed turtles, for which nest temperatures were collected in 1986 and 1988. The data set on sand temperatures was reduced to 41 nests consisting of 728 sexed turtles from the above data set.



PAGE 1

118 000 ......... 1 9 8 6 3000 18 2000 LM LI. o .... .... 1000 -J JUN. JUL. AUG. SEP. OCT. NOV. 16-30 1-15 16-31 1-1516-31 1-15 16-30 1-15 16-31 1-15 16-30 6 3000 1988 Z 2000O l C € 1......... .. OO UJ JUN. JUL. AUG. SEP. OCT. NOV. w 16-30 1-15 16-31 1-15 16-31 1-15 16-30 1-15 16-31 1-15 16-30 H 3000. 1 989 ..J O 2000O z 1000 0 o 0. JUN. JUL. AUG. SEP. OCT. NOV. 16-30 1-15 16-31 1-1516-31 1-1516-30 1-1516-31 1-15 16-30 PERIOD (HALF MONTH)



PAGE 1

32 seasonal period (half-month interval) and nest location (open zone vs vegetation/border zone) on egg survivorship, nonparametric twoway factorial ANOVA with unequal replication (Zar, 1984) was applied for each year's data (Table 4). The ranks of the data, rather than the raw data, were used. The analysis revealed a significant difference in seasonal emergence success in 1988, but not in 1986 or 1989. There were no significant effects of Zone or Zone x Period interaction in any of the three years. These results agree with an analysis carried out with a parametric ANOVA in which case the data were transformed to their arcsine, but the assumption of homogeneity was not satisfied. For the 1988 data, a nonparametric multiple comparison test with unequal sample sizes (Zar, 1984) identified a significant seasonal difference of emergence success between the clutches deposited in the latter half of July (79.7%, SE=4.6, n=16) and ones in the latter half of August (16.9%, SE=9.2, n=14)(Q=4.10 > Qo.o5,6=2.94; Figure 21, total). The other comparisons between each period were not significant. The significant seasonal difference in the 1988 season was mainly accounted for by two heavy inundations that occurred close together in October. These inundations caused by excessive rain on 6 October (221 mm/day) and by the surf of Hurricane Joan on 21-22 October caused substantial loss of nests deposited after the latter half of the three month nesting season. The sample nests deposited in the latter half of July in 1988 hatched before these inundations, and all these clutches produced hatchlings. However, the inundations caused the complete loss of seven of 14 nests deposited during the latter half of August 1988.



PAGE 1

10 and minimum air temperatures were taken daily around 0900. Also at this time, height of shore waves was estimated by eye. Seasonal fluctuations of ground water level on the beach were monitored during the incubation season from July through 10 December in 1986 and 1989. During the 1988 season, the wells were stolen several times during the study, therefore only very scattered data were taken. Plastic PVC pipes (diameter 5 cm) were placed to a depth of 140 cm in 1986 and 150 cm in 1989. Two wells were set in places of 50% shade for the border/vegetation zone and three wells in places of 0% shade for the open zone for each year. This measurement revealed the level of ground water only when it was above the depth of the well. Beach Zones The beach was divided into two zones (Figure 2). The vegetation/border zone lies within 2 m of the border of dense vegetation; there is 5-100% cover, usually of sea grapes and coconut palms. The open sand zone lies below the vegetation/border zone, with <5% cover, usually of sparse herbaceous vines. .Nestina Census To obtain information concerning the seasonal distribution of nests within the two zones, I conducted a census within the study area throughout most of the nesting season during 1986, 1988 and 1989. 1 recorded the number and zone of all nests deposited from



PAGE 1

29 during the latter half of August in 1986 and the first half of September in 1988. In 1989, nesting activity reached a plateau from the latter half of August through September. The last nest in the study area was observed on 27 November 1986, 15 November 1988 and 20 November 1989. An insignificant amount of nesting activity was observed during November of all three years. Approximately 11,724 nests in 1986, 10,509 nests in 1988, and 4,491 nests in 1989 were deposited in the two-mile study area of the beach during the census period. The density of nests along the beach was 7.3 nests/m in 1986, 6.5 nests/m in 1988, and 2.8 nests/m in 1989. The proportion of nests deposited in the vegetation/border zone and in the open sand zone across the season was, respectively, 51.1% and 48.9% in 1986 (n=3,543), and 58.0% and 42.0% in 1988 (n=3,521) and 63.2% and 36.8% in 1989 (n=2,041) (Figure 17). The proportions among these three years were significantly different (df=2, X2=81.8, P<0.0001). Tukey-type multiple comparison tests for proportional data revealed that the proportion of clutches deposited in the vegetation/border zone in 1989 was significantly the highest, intermediate in 1988 and the lowest in 1986 among the three years [q 0.05, -, 3 = 3.314, q=24.883 (86 vs 89), 16.226 (86 vs 88), 10.976 (88 vs 89)]. [Data were transformed by the equation: p'=1/2(arcsin/(X/(n+1 )+arcsin(X+1 )/(n+1 )), Zar (1984).] Figure 17 also shows the seasonal fluctuation of nest site selection. The proportions of nests found in each zone were significantly different at each half-month period in each year (df=7, X2=31.7, P<0.0001 in 1986, df=8, X2=41.2, P<0.0001 in 1988 and df=8,



PAGE 1

92 Table 9. Logistic regression models of fluctuation in sex ratio for green turtle hatchlings with single independent variables. Data collected at Tortuguero, Costa Rica, 1986 and 1988. Independant Variable pseudo R2 Slope Intercept Mean nest temperature 0.202 1.1709 -34.4594 Mean sand temperature* 0.172 1.0636 -30.3740 Incubation period 0.177 -0.2630 15.4679 Rainfall (R-1) 0.079 -0.0954 0.5907 Rainfall (R-11) 0.114 -0.1290 1.0477 Rainfall (R-Ill) 0.171 -0.1912 1.8052 Bottom depth of nests 0.010 0.0250 -2.5171 All model, df=1, p<0.001 *n=728, else n=924



PAGE 1

86 Table 4. Effects of seasonal period and nest location in the open and the vegetation/border zones on egg survivorship at the Tortuguero beach in 1986, 1988 and 1989. Analysis is by nonparametric twofactor ANOVA with unbalanced factorial design (Zar, 1984). Source SS d f H p 1986 Period 1116.91 3 201.19 0.10


PAGE 1

EGG SURVIVORSHIP AND PRIMARY SEX RATIO OF GREEN TURTLES, CHELONIA MYDAS, AT TORTUGUERO, COSTA RICA BY KAZUO HORIKOSHI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1992



PAGE 1

108 20 -1989 LJL z 1000JUL. AUG SEP. 01x. NO. DC SUFiE1 989/BORDERNEGETATION ZONE 40 1 SD ____ ___ ____ ___ ___ ____ ___ ___ MEAN DEPTH OF S80CLUTCH BOlTOM 1 SD 120 BOTTOM OF WELJUL. AUG SEP. OCT. NO. DEC. SUR:FACE1989/OPEN ZONE 40 1 SD ____ ___ ____ ___ ____ ___ ____ ___ MEAN DEPTH OF 80CLUTCH BOTTOM I SD 120 BOTTOM OF Vw:LJUL. AUG. SEP. OCT. NOVJ. DEC.



PAGE 1

CHAPTER 4 DISCUSSION Nesting Density and Temporal Nest Distribution High yearly fluctuation of nesting density is known for Tortuguero green turtles (Bjorndal et al., 1985). The 1986 and 1988 seasons were two of the highest density years since 1956, whereas the 1989 season was in the middle of the range according to the long term monitoring project at the northernmost 8.1 km section (K. Bjorndal, pers. comm.). My study site was part of the highest nest density area along the beach (Carr et al., 1978), and this trend was observed throughout the study (Horikoshi, unpublished data). Thus, the recorded densities in 1986 (7.3 nests/m) and in 1988 (6.5 nests/m) throughout each season were probably among the highest existing at the Tortuguero beach since 1956. As far as I know, nesting densities of green turtles of this magnitude are only known on an Australian island, an Oman beach, and an Ascension beach (Groombridge and Luxmoore, 1989; Mortimer, 1981; Mortimer and Carr, 1987). Temporal distribution of green turtle nests varied greatly in the frequency pattern and slightly in the range of season among the three years. Carr et al. (1978) noted that the main breeding activity takes place in July, August and September, although a few turtles nest throughout the year. The 1988 season in general agreed with 55



PAGE 1

36 stages, including emerging hatchlings, were vulnerable to suffocation. Clutches in both zones were vulnerable to flooding. Excluding eggs that died from unknown causes between the flooding and Hurricane Joan in 1988, freshwater inundation was responsible for 8.4% of egg loss in the open zone and 11.3% of egg loss in the vegetation/border zone throughout the three years. When eggs of undetermined fate are included, the figures increase to 11.0% and 12.1%, respectively. The risk of mortality by flooding between the two zones was not consistent among years. The percentages of eggs lost by freshwater inundation are as follows: 13.6% and 17.8% in 1986, 3.0-10.7% and 12.5-15.0% in 1988, 8.5% and 3.8% in 1989 for eggs deposited in the open zone and in the vegetation/border zone, respectively. In 1986 and 1988, significantly more eggs were killed by flooding in the vegetation/border zone than in the open zone (X2=18.7, p<0.0001, 1986; X2=7.3, p=0.0068, using possible closest rate in 1988), while the opposite trend occurred in 1989 (X2=116.5, p<0.0001, 1989). Artificial debris, such as tangled fishing nets and ropes were frequently seen on the beach. On one occasion, a buried tangled rope physically prevented 10 hatchlings of a sample nest from emerging to the surface, although the remaining 119 hatchlings escaped through the rope. Fishing nets and ropes have a high potential to cause substantial damage to emerging hatchlings. Biotic factors Predation by several types of animals and excavation by female turtles were major factors responsible for mortality of



PAGE 1

102 (A) 800 600 -J SE S200 JAN. FEB.MAR. APR. MAY. JUN. JUL. AUG. SEP. OCT. NOV. DEC. (B) 40 30 I LUl 20z -00 0 -." JAN. FEB. MAR. APR. MAY. JUN. JUL. AUG. SEP. OCT. NOV. DEC. Figure 5. Seasonal relationship between rainfall and green turtle nesting. (A) Monthly mean rainfall at Tortuguero, Costa Rica, from 1978 through 1989. Months with hatched bars represent the major incubation season of green turtle eggs. The bars show one standard error for each month. (B) Mean nesting frequency among the 1986, 1988, and 1989 seasons. Details for each season are presented in Figure 16.



PAGE 1

54 variable of incubation period became insignificant. Moreover, improved fitness was negligible (pseudo R2=O.25 for both models). Finally, two easily measured param ete rs-mean daily rainfall (R-111) and zone--were entered into a multiple regression model to assess their predictive ability relative to that of nest temperature data. Because there seems to be an interaction between rainfall data (R-111) and zones in affecting the sex ratios (Figure 36), their interaction was also included. Mean nest temperature is biologically the primary factor for determining sex of green turtle hatchlings and the single regression model showed it to be the best single fit model among the applied independent factors. The fitness of the multiple regression model using rainfall and zone (pseudo R2=O.23) (Table 12) was slightly better than that of the single regression model using mean nest temperatures (pseudo R2=O.20).



PAGE 1

38 raccoons. The density of raccoons in Tortuguero was probably much lower than that of coatis, because I sighted only two adult raccoons at night and one juvenile during the daytime during the three years of study. The juvenile was seen eating several eggs of an unsampled nest under the vegetation. In addition to the above two predator species, one skunk and a few feral dogs were sighted excavating several unsampled nests in the study area. It is not known whether some of the sample nests were depredated by these species. A skunk (unknown species) was seen excavating an unsampled nest in 1986; this was the only time a skunk predator was sighted during the study. Four unsampled nests in 1988 were predated by feral dogs. Feral dogs were rarely seen in the park area throughout the study because of a successful dog control program on the beach by personnel of Parque Tortuguero. Predation by mammals was responsible for loss of 3.0% of the eggs in the open zone and 17.3% of the eggs in the vegetation/border zone throughout the three years. Only four of 115 (3.5%) sample nests in the open zone were partially or completely preyed upon during the three years as compared to 29 of 135 (21.5%) sample nests in the vegetation/border zone (X2=16.0, p<0.0001). There was considerable yearly variation in predation by mammals. In the 1989 season when the number of nests deposited was about half that of the 1986 and 1988 seasons, predation by mammals increased to the highest percentage in both zones (egg loss of 5.0% in the open zone, 30.9% in the vegetation/border zone in the 1989 season). The 30.9% egg loss by mammal predators in the



PAGE 1

CHAPTER 5 SUMMARY AND CONCLUSIONS The major nesting activity of green turtles occurred from July through the first half of October at Tortuguero during the three-year study, although the frequency of seasonal distribution shifted among years. Nest density also varied among years. The nesting densities recorded in 1986 (7.3 nests/rn) and in 1988 (6.5 nests/rn) from July through November were two of the highest known for this species worldwide. Tortuguero beach was divided into two zones. The vegetation/border zone lies within 2 m of the border of dense vegetation, and the open zone lies between the vegetation/border zone and the shore line. Nest distribution between the zones varied significantly yearly and seasonally, and there was no common seasonal trend of fluctuation among the years. The vertical distribution of green turtle nests was biased toward the upper part of the vegetated beach throughout each season. A preference for nest sites on the upper beach may be rather conservative for this species. Throughout this study, a distinct thermal difference at the depth of green turtle nests between the two zones was consistent, with nests in the open zone being warmer than those in the vegetation/border zone. The differences in overall mean sand 78



PAGE 1

Figure 9. Seasonal fluctuation of ground water table and rainfall in the 1989 incubation season of green turtle eggs at Tortuguero, Costa Rica. The line of mean bottom depth of green turtle clutches is 77.9 cm (SD=11.1) from the surface. The shaded area shows the maximum and minimum range of ground water table recorded at two wells in the open zone and at three wells in the vegetation/border zone.



PAGE 1

90 Table 7. Effects of seasonal period and nest location in the open and the vegetation/border zones on sex ratios of green turtle hatchlings at the Tortuguero beach in 1988. Analysis was by two-factor ANOVA with unbalanced design. The ratios of females were transformed (p'=1/2(arcsin/(X/(n+1l)+arcsin/(X+1)/(n+1 ))(Zar,1984). Fmax test showed homogeneity of variances. So u rce SS df MS F p Period 1.7385 5 0.3477 3.30 0.017 Zone 0.7695 1 0.7695 7.30 0.011 Period*Zone 0.8093 5 0.1619 1.54 0.208 ns Total 3.2685 31 0.1054



PAGE 1

79 temperature at 60 cm depth were 0.80C in 1986, 1.1 OC in 1988, and 1.50C in 1989. This is due to the dense vegetation and associated shade at the upper part of the beach. However, the spatial influence of distribution on sex ratio probably occurs only during a dry period. During the middle part of the 1988 season, under prolonged dry weather conditions, sand temperatures in the open zone clearly exceeded the pivotal temperature, whereas sand temperatures in the vegetation/border zone only slightly exceeded the pivotal temperature. Consequently, most nests in the open zone produced female-biased sex ratios, whereas nearly equal or moderately female-biased sex ratios occurred in the vegetation/border zone. On the other hand, during prolonged wet weather, sand temperatures in both zones decreased below the pivotal temperatures, causing malebiased sex ratio in both zones. This was observed in the latter part of the 1988 season and throughout the 1986 season. Tortuguero, with 5400 mm of rain each year, is one of the wettest sea turtle rookeries in the world. A short dry period in September occurs between the wet months of July through August and October through November during the major incubation season of green turtle nests. This dry season seems to be very important in determining the primary sex ratio of the Tortuguero population. Because a large proportion of the nests is deposited during August and the most critical period for sex determination in these nests occurs in September, a dry September can produce enough females, particularly in the open zone, to avoid a highly male-biased sex ratio for the season.



PAGE 1

84 Table 2. Overall mean sand temperatures (0C SE) throughout the season on transects from 1 July through 10 December in 1986, 1988 and 1989 at the Tortuguero beach. 1986 1988 1989 n=76 n=73 n=76 Open Zone 60cm Depth 27.60.1 28.80.1 28.60.1 80cm 28.50.1 28.20.1 Vegetation/border Zone 60cm Depth 26.80.1 27.70.1 27.10.1 80cm 27.40.1 26.90.1



PAGE 1

94 Table 11. Logistic regression model of fluctuation in sex ratio for green turtle hatchlings with significant variables. Data collected at Tortuguero, Costa Rica, 1986 and 1988. Parameters Slope (SE) ChiSqure P Mean nest temperature 0.5230 (0.1250) 17.50 <0.0001 Incubation day -0.0836 (0.0299) 7.83 0.0051 Rainfall (R-Ill) -0.0906 (0.0224) 16.40 0.0001 Zone 0.5532 (0.1822) 9.21 0.0024 Intercept -9.8173 (4.6810) 4.40 0.0360 -2LogLikelihood=290.1, df=4, overall P<0.0001 pseudo R2=0.239, n=924



PAGE 1

97 MILE 0 2 4 TORTUGUERO 6 STUDY AREA 83 20'W 10 10 20'N 14 COSTA 16 RICA 18 20 22 PAPISMINA Figure 1. The two-mile study area at Tortuguero, Costa Rica.



PAGE 1

72 The high variability of sex ratio in relation to incubation temperature was not surprising. First, even under constant temperature in the laboratory, high inter-clutch variation in sex ratios was found between two loggerhead clutches from the same Florida beach (Mrosovsky, 1988). Limpus et al. (1985) reported a similar high variation in sex ratio at the middle range of temperature among three loggerhead clutches sampled from Mon Repos, Australia, at constant temperature conditions. However, he did not find such large variation among the loggerhead clutches from Heron Island. Mrosovsky and Pieau (1991) hypothesized that the response of gonad differentiation to the incubation temperature might be different depending on the genotype of the eggs, and that genetic factors can override the temperature effect at the middle temperature range. This theory was inferred from the experimental results in a freshwater turtle, Emys orbicularis, in which the expression of H-Y antigen was strongly correlated with the expressed sex of gonads at the middle range of temperature where both sexes were produced (Zaborski et al., 1988). Mean nest temperatures during the middle third of development in 1986 and 1988 at Tortuguero fell within a narrow range (27.0-30.70C) compared to the much wider temperature range over which live hatchlings of this species can be produced (25-33C) (Miller, 1985). Both sexes were produced throughout the observed range at Tortuguero. Thus, the high variation in sex ratio over the observed range can be explained if the mechanism of genetic influence operates in green turtle eggs.



PAGE 1

152 Horrocks, J.A. and N. McA. Scott. 1991. Nest site location and nest success in the hawksbill turtle Eretmochelys imbricata in Barbados, West Indies. Mar. Ecol. Prog. Ser. 69:1-8. Horton, M. 1989. Reproductive success of sea turtles nesting on Wabasso beach, east-central Florida. M. S. thesis. The Virginia Polytechnic Institute and State University, Blacksburg, Virginia. Hosmer, D.W., Jr. and S. Lemeshow. 1989. Applied Logistic Regression. A Wiley-lnterscience Publication, New York. Jackson, M.E., L.U. Williamson, and J.R. Spotila. 1988. Gross morphology vs. histology: sex determination of hatchling sea turtles. Marine Turtle Newsletter. 40:10-11. Janzen, F.J. and G.L. Paukstis. 1991. Environmental sex determination in reptiles: ecology, evolution, and experimental design. Quart. Rev. Biol. 66:149-179. Kamezaki, N. 1990. Karyotype of the hawksbill turtle, Eretmochelys imbricata, from Japan, with notes on a method for preparation of chromosomes from liver cells. Jap. J. Herpetol. 13:111113. Kraemer, J.E. and R. Bell. 1984. Rain-induced mortality of eggs and hatchlings of loggerhead sea turtles (Caretta caretta) on the Georgia coast. Herpetologica 36:72-77. Leh, C.M.U., S.K. Poon, and Y.C. Siew. 1985. Temperature-related phenomena affecting the sex of green turtle (Chelonia mydas) hatchlings in the Sarawak Turtle Islands. Sarawak Mus. J. 34:183-193. Leslie, A.J., D.N. Penick, and J.R. Spotila. In Press. Abiotic and biotic factors affecting hatching of the leatherback turtle at Tortuguero, Costa Rica. In Proceedings of the 12th Annual Workshop on Sea Turtle Conservation and Biology. NOAA Technical Memorandum.



PAGE 1

150 Bjorndal, K.A., A. Carr., A.B. Meylan, and J.A. Mortimer. 1985. Reproductive biology of the hawksbill (Eretmochelys imbricata) at Tortuguero, Costa Rica, with notes on the ecology of the species in the Caribbean. Biol. Conserv. 34:353-368. Bolten, A.B., K.A. Bjorndal, J.S. Grumbles, and D.W. Owens. In Press. Sex ratio and sex-specific growth rates in immature green turtles, Chelonia mydas, in the southern Bahamas. Copeia. Bull, J.J. 1983. Evolution of Sex Determining Mechanisms. Benjamin/Cummings, Menlo Park, CA. Bull, J.J. and R.C. Vogt. 1979. Temperature-dependent sex determination in turtles. Science 206:1186-1188. Bustard, H.R. 1972. Sea Turtles: Natural History and Conservation. Taplinger, New York. Carr, A.F., M.H. Carr, and A.B. Meylan. 1978. The ecology and migrations of sea turtles, 7. The west Caribbean green turtle colony. Bull. Amer. Mus. Nat. Hist. 162:1-46. Carr, A.F., III. 1979. The ecology of the prawn, Macrobrachium acanthurus (Weigmann) and its implications for tropical esturine management. Ph.D. Dissertation. The University of Michigan, Ann Arbor, MI. Christens, E. 1990. Nest emergence lag in loggerhead sea turtles. J. Herpetol. 24:400-402. Coen, E. 1983. Climate. In Janzen, D.H. (ed), Costa Rican Natural History. pp.35-46. The University of Chicago Press, Chicago. Cornelius, S.E. 1986. The Sea Turtles of Santa Rosa National Park. Fundacion de Parques Nacionales, Costa Rica. Dalrymple, G.H., J.C. Hampp, and D.J. Wellins. 1985. Male-biased sex ratio in a cold nest of a hawksbill turtle (Eretmochelys imbricata). J. Herpetol. 19:158-159.



PAGE 1

EGG SURVIVORSHIP AND PRIMARY SEX RATIO OF GREEN TURTLES, CHELONIA MYDAS, ATTORTUGUERO, COSTA RICA BY KAZUO HORIKOSHI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1992

PAGE 2

ACKNOWLEDGEMENTS First I thank my parents Shigeru and Kiku Horikoshi, and my wife Harumi Horikoshi, whose love and encouragement strengthened my resolve to complete this task. I would like to thank the members of my committee: Dr. Martha Crump, Dr. Karen Bjorndal, Dr. Jack Kaufmann, Dr. Clay Montague and Dr. Jeanne Mortimer who offered continuous encouragement and advice throughout this study. I feel fortunate to have been able to work under their guidance. I also would like to thank the late Dr. Archie Carr who offered invaluable encouragement and an opportunity to undertake this study at the Tortuguero beach. Dr. Takako Oshima provided advice on statistics. I am very grateful to Dr. Louis Guillette who was most generous in allowing me use of his histology laboratory. Dr. Alan Bolten and Dr. Blair Witherington offered much appreciated advice by sharing their knowledge of sea turtles. My field work in Costa Rica greatly depended on the aid of Elver and Elvin Gutierrez, Robert Carlson, Harumi Horikoshi, and Carlos Diez. Stephen Morreale kindly recorded some weather data in 1986. I would like to thank the Caribbean Conservation Corporation, Sigma Xi, and the Archie Carr Center for Sea Turtle Research, the Center for Latin American Studies, and the Department of Zoology, University of Florida for partial funding for this project. I would i i

PAGE 3

also like to thank the Costa Rican National Park System for allowing me to work in Tortuguero National Park iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGEMENTS ............................................................................................... i i ABSTRACT .................................................................................................................... V i CHAPTERS 1 INTRODUCTION ... .. .. ...... .. .. .... .. .. .. .. .... .. .. .. ...... .. ...... .. .... .. .. .. .. ... ..... ... .. .. .... .. .. 1 2 METHODS ............................................................................................................. 8 Study Area .. . . . . .. . . . . .. . .. . . .. . . . . .. . . . . .. 8 Weather Record .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 9 Beach Zones .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 1 0 Nesting Census .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 1 0 Sampling of Egg Survivorship Nests ............................................... 1 2 Sampling of Nests to Determine Sex Ratio .. ... .. .. .. .. .. .. .... .. .. .. .. .. 1 4 Histology and Sexing .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 1 6 Sand Temperatures ................................................................................. 1 6 Statistical Analysis ............................................................................... 1 7 3 RES UL TS ............................................................................................................ 1 8 Physical Environment Surrounding Nests ..................................... 1 8 Air Temperatures .............................................................................. 1 8 Rainfall .................................................................................................. 1 9 Vertical Position of Clutches ...................................................... 2 1 Ground Water ....................................................................................... 2 2 Waves ...................................................................................................... 2 3 Daily Fluctuation of Sand and Nest Temperatures ............... 2 4 Seasonal Fluctuation of Sand Temperatures ......................... 2 6 Seasonal Nest Distribution and Nest-Site Selection .............. 2 8 Analysis of Egg Survivorship Samples .......................................... 3 O Fates of Sample Nests .................................................................... 3 O Abiotic and Biotic Factors Affecting Egg Survivorship .... 3 3 iv

PAGE 5

Fate of eggs from sample nests ....................................... 3 3 Criteria for assessing the cause of egg mortality ... 3 3 Abiotic factors .......................................................................... 3 4 Biotic factors ............................................................................. 3 6 Other Categories of Mortality ...................................................... 4 2 Analysis of Sex Ratio ............................................................................. 4 3 Sexed Samples ..................................................................................... 4 3 Sex Ratio vs Temperature .............................................................. 4 4 Seasonal Variation and Zone Effect in 1988 .......................... 4 6 Overall Sex Ratio in 1988 .............................................................. 4 7 Seasonal Variation and Zone Effect in 1986 ......................... .4 8 Overall Sex Ratio in 1986 .............................................................. 4 9 Prediction of Sex Ratio from Independent Variables .............. 4 9 Logistic Regression Model ............................................................. 5 O Logistic Regression with a Single Variable .......................... 5 0 Nest and sand temperatures ................................................ 5 0 Incubation period ...................................................................... 51 Rainfall ......................................................................................... 51 Bottom depth of clutch . . . . . . . . . . . . . . 5 2 Multiple Logistic Regression Model ........................................... 5 3 4 DISCUSSION .................................................................................................... 5 5 Nesting Density and Temporal Nest Distribution ...................... 5 5 Seasonality of Environmental Parameters ................................... 5 7 Spatial Distribution between the Zones ........................................ 5 8 Overall Reproduction Rate ................................................................... 6 1 Spatial Effects on Mortality Factors .............................................. 6 3 Seasonal Fluctuation of Emergence Success ............................... 6 5 Spatial Effects on Sex Ratio .............................................................. 6 7 Temporal Effects on Sex Ratio .......................................................... 6 9 Annual Variation in Primary Sex Ratio .......................................... 7 0 Predictability of Sex Ratio by Environmental Factors ........... 71 5 SUMMARY AND CONCLUSIONS ..................................................................... 78 LITERATURE CITED ................................................................................................ 149 BIOGRAPHICAL SKETCH ...................................................................................... 1 58 V

PAGE 6

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 EGG SURVIVORSHIP AND PRIMARY SEX RATIO OF GREEN TURTLES, CHELONIA MYDAS, ATTORTUGUERO, COSTA RICA By KAZUO HORIKOSHI AUGUST, 1992 Chair: Martha L. Crump Cochair: Karen A. Bjorndal Major Department: Zoology Tortuguero beach in Costa Rica is one of the last major nesting sites in the western Atlantic for green turtles. During 1986, 1988 and 1989, I studied temporal and spatial nest distribution, egg survivorship, and sex ratio of hatchlings in two thermal zones: the vegetation/border zone (within 2 m of the border of dense vegetation) and the open zone (between the vegetation/border zone and the sea). Although the distribution of nests between the two zones differed significantly from year to year, it was biased toward the vegetated area of the beach throughout each season vi

PAGE 7

I analyzed the emergence success of 49 clutches in 1986, 88 in 1988, and 113 in 1989 Emergence success from these nests showed a bimodal distribution (0% and > 70%) and did not differ significantly between the zones. Abiotic factors including surf, freshwater inundation, and hurricane killed approximately 20% of sample eggs. Biotic factors including predation and turtle digging killed 5% of sample eggs in the open zone and 19% of sample eggs in the vegetation/border zone. The relative importance of each factor varied from year to year. Mammals destroyed four times as many nests in the vegetation/border zone as in the open zone. My estimation of hatchling emergence success was 57.2% (1986), 45.7% (1988) and 66.6% (1989). A sample of hatchlings was collected for direct sexing from each of 12 nests in 1986 and 56 in 1988 The sex ratios of hatchlings fluctuated intra-seasonally in 1988, but not in 1986. The estimated overall proportion of female hatchlings was 10 1 % in 1986 and 40.6% in 1988. used single and multiple logistic regression models to test environmental and clutch variables for the prediction of sex ratio. The pivotal temperatures were 29.4C for mean nest temperature, and 28 5C for mean sand temperature. Mean nest temperature during the middle third of development was the best fit model (pseudo R 2 =0.20) in the single regression analysis Mean nest temperature, incubation period, rainfall, and nesting zone were significant parameters for multiple regression (pseudo R 2 =0.24). However, even the model with the best fit was still not reliable for predictive purposes vii

PAGE 8

CHAPTER 1 INTRODUCTION Recently it has become apparent that there are two types of sex determination mechanisms in reptiles: genotypic and environmental. In the latter case, sex of the offspring is decided after fertilization by environmental factors. The common form of environmental sex determination in turtles is temperature dependent sex determination (TSO), in which incubation temperature during a critical portion of embryonic development controls gonadal differentiation. The physiological mechanism involved in this phenomenon is not understood (reviewed in Bull, 1983: Ewert and Nelson, 1991 ; Janzen and Paukstis, 1991 ). At present, seven of eight species of sea turtles are known to show the TSO phenomenon. These species are: Caretta caretta (Yntema and Mrosovsky 1980), Chelonia mydas (Miller and Limpus 1981 ) Chelonia agassizi (Alvarado and Figueroa, 1989), Dermochelys coriacea (Mrosovsky et al., 1984) Eretmochelys imbricata (Dalrymple et al., 1985), Lepidochelys olivacea (McCoy et al., 1983), and Lepidochelys kempi (Shaver et al., 1988). There are three patterns of TSO in reptiles (Bull, 1983). Type A pattern results in females at lower temperatures, males at higher temperatures and both sexes at intermediate temperatures (most crocodilians and lizards); Type B pattern results in females at 1

PAGE 9

2 warmer temperatures, males at lower temperatures and both sexes at intermediate temperatures (most turtles); Type C pattern results in females at warmer and lower temperatures and males at intermediate temperatures (three crocodile species, one lizard species and three freshwater turtle species) (see list of species in Janzen and Paukstis, 1991 ). All seven species of sea turtles with TSO show the Type B pattern. At present, no species with heteromorphic sex chromosomes is known to show TSO Karyotypes of five species of sea turtles have been examined microscopically, but no discernible heteromorphic sex chromosomes have been confirmed (Caretta caretta, Bickham, 1981; Chelonia mydas, Bickham et al., 1980; Oermochelys coriacea, Medrano et al., 1987; E retmochelys imbricata, Kamezaki, 1990; Lepidochelys olivacea, Bhunya and Mohanty-Hejmadi, 1986). However, there is evidence that the two sexes of sea turtles are genetically different at the molecular level. In adults of Caretta caretta and Chelonia mydas, males had a higher level of H-Y antigen in their blood cells than did females (Wellins, 1987). A series of experiments on a European freshwater turtle, Emys orbicularis (Zaborski et al., 1982, 1988), which shows Type B TSO, indicated that genetic and environmental factors can operate simultaneously. It appears that incubation temperatures at both extremes can overide genotypic influence, whereas at intermediate temperatures the genetic factor can influence the sex of turtles to some extent. There was a strong correlation between sexual phenotype of gonads and H-Y antigen phenotype of blood cells (male negative / female positive) from eggs incubated at the

PAGE 10

3 intermediate temperature. Girondot and Pieau ( 1990) proposed a similar mechanism to explain the variation of the pivotal temperatures observed in loggerhead turtles (Limpus et al., 1985; Mrosovsky, 1988). Further investigation of the effect of genetic factors over TSD in species of sea turtles has not been conducted. Tortuguero beach on the Caribbean coast of Costa Rica is one of the last major nesting sites in the western Atlantic for green turtles, Chelonia mydas. Spotila et al. (1987) found that the temperatures of the nest during the middle third of development influenced the sex ratio of Tortuguero green turtles in natural nests. Mean temperatures less than 28.5C produced mainly males, mean temperatures greater than 30.3C produced only females, and mean temperaturess between 28.5 and 30.3C produced both sexes. They also found that the different thermal zones at the beach yielded different sex ratios; nests under vegetation produced a high percentage of male hatchlings, whereas nests in the open beach produced mainly females. No seasonal trend of sand temperature was apparent during their monitoring period (24 July 22 September 1980). Assuming no seasonal fluctuation of sex ratio, and using 1977 data for the distribution of nests between the shady zone and the open sand zone at Tortuguero (Fowler, 1979), they estimated that 67% of the hatchlings were females in the 1977 season. However, their study did not cover the full seasonal profile at Tortuguero, and the sample size was very limited (15 nests). The main incubation season of the green turtle colony at Tortuguero extends from July through November (Fowler, 1979). The rainfall at Tortuguero is extremely variable from year to year (Myers, 1981 ).

PAGE 11

4 During the middle of the season, a short dry period occurs (usually in September). Carr (1979) described the June August period, the early incubation season, as "monsoonlike, with many consecutive days of heavy overcast and rain". In general, rainfall and sand temperature are inversely related on the beach. Therefore, it is very probable that seasonal fluctuation of sand temperature, and consequently seasonal fluctuation of sex ratios, can occur regularly at Tortuguero. Therefore, in addition to evaluating whether the 1980 season is typical or atypical, it is essential to examine the entire season over several years to document seasonal and yearly variation for a better estimation of the primary sex ratio at Tortuguero. All species of sea turtles are considered to be threatened or endangered because of over-exploitation and destruction of their habitats. The green turtle, a circumtropical large species, is a valuable food source for many coastal people, and thus has been harvested for a long time. Current management practices for sea turtles focus on conservation of each population. Among the conservation practices, protection of incubating eggs in several forms of artificial hatcheries is presently the most common measure. However, little regard has been given to the incubation temperature until recently. Incubation of sea turtle eggs in styrofoam boxes subjected the eggs to a different thermal environment from that experienced on a natural beach and showed a masculinizing effect on the embryos (Mrosovsky, 1982; Dutton et al., 1985). Because of this incubation method, the early years of the Kemp's ridley head starting project unintentionally produced male

PAGE 12

5 dominated turtles much of the time (Shaver et al., 1988). On the other hand, the hatchery on a Sarawak beach produced highly female biased green turtle hatchlings for many years (Leh et al., 1985). Location and type of artificial hatchery can easily alter the sex ratio of hatchlings produced. Currently there is no consensus on the best sex ratio to produce in hatcheries because the sex ratios of sea turtle populations and their dynamics are uncertain. Therefore, to understand better the ecological and conservation implications of TSO among sea turtles, it is critical to know the process of sex determination and the primary sex ratio of hatchlings on a natural beach. However, natural sex ratios of hatchling sea turtles have been studied in only a few populations, and estimation of primary sex ratio of a population has not been feasible, partially because of small sample sizes and limited seasonal coverage (Lim pus et al., 1983; Maxwell et al., 1988; Rimblot-Baly et al., 1987; Spotila et al. 1987) Quantitative investigations that cover the full seasonal profile exist only for the Suriname green turtle and leatherback turtle colonies (Mrosovsky et al., 1984) and a Florida loggerhead turtle colony (Mrosovsky and Provancha 1989; Provancha and Mrosovsky, 1992) However, these studies do not include information on egg survivorship on the beach. To estimate reliably the primary sex ratio, egg survivorship is an essential factor because survivorship might differ in different thermal zones on the beach or within seasons. Fowler (1979) found that the nest position affected survival rate of green turtle nests at Tortuguero because of differential mammal predation and inundation rates. Nests near or in the beach

PAGE 13

6 border vegetation suffered greater predation than did nests on the lower part of the beach. However, because introduced dogs destroyed about one fourth of the nests in the 1977 season, Fowler's results do not represent natural survivorship of the green turtle nests at Tortuguero. Recently, due to success of a dog control program by Tortuguero park guards, the beach has almost returned to the natural condition so that the coati, Nasua narica, has become the main predator. No quantitative study of egg survivorship in the green turtle colony has been conducted since the dogs have been controlled. There is a large yearly fluctuation in the number of nesting green turtle females at Tortuguero; 1977 was a year with relatively few nesting females (Bjorndal et al., 1985). Nest density might affect selection of nest sites on the beach by turtles and also influence spatial and temporal predation patterns. Therefore, a prolonged study to cover several years is needed to understand fully the reproduction of Tortuguero green turtles and to document the biotic and abiotic factors contributing to their natural mortality. A combination of (1) distribution of nests among thermal zones; (2) egg survivorship; and (3) seasonal trend of sand temperature and its consequence the seasonal trend of sex ratio in the different thermal zones, are recognized as major factors in determining primary sex ratios. The primary objective of this study was to estimate the primary sex ratios of the green turtle population at the Tortuguero beach and to obtain information concerning egg survivorship and sex ratio of the nests throughout the entire season for three years. An additional objective was to accumulate information on the physical

PAGE 14

7 characteristics of natural nests on the beach on which the TSO phenomenon operates and to assess the predictability of sex ratio from environmental factors. This parameter, primary sex ratio, is not only necessary for improving conservation practices, but it also will increase our understanding of the dynamics of the most important population of green turtles in the western Atlantic Ocean Furthermore, I believe that information on primary sex ratio will also give us a key to assess adult sex ratio of all sea turtles.

PAGE 15

CHAPTER2 METHODS Study Area My study area was located at Tortuguero, Costa Rica. The Tortuguero beach extends 22 miles (35.4 km) from the mouth of the Rio Tortuguero south to the Rio Parismina on the Caribbean coast of Costa Rica (Figure 1 ). The beach is located on a long, narrow island that is separated from the mainland by a freshwater lagoon and estuaries of the rivers Tortuguero beach is the nesting beach for the largest surviving nesting population of green turtles in the Atlantic Ocean (Groombridge and Luxmoore, 1989). Leatherback turtles, Dermochelys coriacea, and hawksbill turtles, Eretmochelys imbricata, also nest on the same beach, but are much less abundant than green turtles. The shoreline is composed of a continuous series of spits and guts The shape of the shoreline shifts considerably with surf erosion and rebuilding, sometimes in a short period of time. High waves occur almost throughout the nesting and incubation season of green turtles (June to December). The beach is classified as a high energy beach because of constant high wave activity. The beach was divided into 22 one-mile sections from the northern end to the southern end. The central area of the beach (mile 8

PAGE 16

9 3 to mile 18), has been protected as a part of Parque Nacional Tortuguero since 1975. To investigate egg survivorship and primary sex ratio of the Tortuguero population of green turtles, I selected two miles of beach (mile 6 to mile 8) for my study area This area is one of the most highly utilized areas by nesting green turtles. Approximately 17% of nests laid on the beach are deposited in this two-mile section (Carr et al., 1978). The beach shape, vegetation, and animals in the central area remain undisturbed because the area is successfully protected as part of the national park. There is no human habitation in the park area. Railroad vine, lpomoea pes-caprae, and beach lily, Hymenocallis littoralis, predominate on the rear of the open beach. Behind the beach, cocoplum, Chrysobalanus icaco, seagrape bushes, Cocoloba uvifera, and coconut palms, Cocos nucifera, are common Inland, there is a well developed and largely undisturbed tropical wet forest. The dense canopy of the inland forest is high (> 8 m). Weather Record and Ground Water Ambient temperature and rainfall data were collected with standard meteorological instruments. In 1986 and 1989, data were collected at the Green Turtle Research Station adjacent to the beach, located about 9 km north of the study area. During the 1988 season, data were taken at the study area from July through the middle of October, and later at the station. Rainfall and maximum

PAGE 17

10 and minimum air temperatures were taken daily around 0900. Also at this time, height of shore waves was estimated by eye. Seasonal fluctuations of ground water level on the beach were monitored during the incubation season from July through 10 December in 1986 and 1989. During the 1988 season, the wells were stolen several times during the study, therefore only very scattered data were taken. Plastic PVC pipes (diameter 5 cm) were placed to a depth of 140 cm in 1986 and 150 cm in 1989. Two wells were set in places of 50% shade for the border/vegetation zone and three wells in places of 0% shade for the open zone for each year. This measurement revealed the level of ground water only when it was above the depth of the well. Beach Zones The beach was divided into two zones (Figure 2). The vegetation/border zone lies within 2 m of the border of dense vegetation; there is 5-100% cover, usually of sea grapes and coconut palms. The open sand zone lies below the vegetation/border zone, with <5% cover, usually of sparse herbaceous vines. Nesting Census To obtain information concerning the seasonal distribution of nests within the two zones, I conducted a census within the study area throughout most of the nesting season during 1986, 1988 and 1989. I recorded the number and zone of all nests deposited from

PAGE 18

1 1 the previous night. Crawl tracks and nest marks were used to locate nests and to identify species (Pritchard et al., 1983). In 1986 the census was carried out from 2 July through 29 November at an interval of once every two to four days. Two surveys on 1 O and 14 July were not completed because high waves and strong wind erased most of the tracks before the surveys began. Therefore, these surveys were eliminated from analysis. In 1988, the census was carried out from 16 June through 30 November. The interval between surveys was two to five days; most were two to three days No survey was conducted during the evacuation for Hurricane Joan between 19 and 23 October. In 1989, the census was conducted from 17 June through 30 November. The interval between surveys was two to three days; most were two days. The census data were pooled into half month periods (1st to 15th, and 16th to 30th or 31st). The daily mean number of nests deposited per period was calculated by dividing the total number of nests counted by the number of censuses during each period. The total number of nests deposited during each period was calculated by multiplying the daily mean number of nests by the number of days during the period. Total number of nests over the season was the sum of the numbers of nests during each period The proportion of nests deposited in the vegetation/border zone and in the open sand zone was calculated for each half month period. The proportion of nests in the two zones for the entire season was calculated from pooled numbers of nests counted during the censuses

PAGE 19

12 Sampling of Egg Survivorship Nests A representative sample of nests deposited within the study area was marked and followed throughout the incubation period. As much as possible, I attempted to sample equal numbers of nests in the two zones and for each half month period of time to analyze seasonal and spatial variances of egg survivorship. Sampling at night was conducted seven to 1 O times during each half month period to locate nesting females. During 1986 and 1988, only females found during the early stage of nesting behavior (before digging the egg chamber) were sampled; for these females I counted the number of eggs during deposition. Because fewer females nested during 1989 than during 1986 and 1988, it was necessary to include nesting females that were already in the process of depositing eggs; thus during 1989 the number of eggs was not counted. Sample nests were marked with a numbered stake placed 1 m from the egg chamber. I also placed two pieces of vinyl tape in the vegetation to triangulate the location. A piece of numbered flagging tape was placed in the egg chamber to confirm the location in the event of nest destruction. Due to the difficulty of finding nesting females in a specific zone during a specific period, the numbers of marked nests in each block are not equal. I omitted from analysis several marked nests (1 in 1986, 2 in 1988 and 1 in 1989) that were located so close to other egg chambers that I was unable to determine hatching success for the individual nests because the eggshells were inseparable

PAGE 20

1 3 Furthermore, in spite of intensive marking, some marked nests (2 in 1986, 3 in 1988 and 7 in 1989) were lost. Some of these lost nests may have been completely excavated by other nesting females. In 1986, 49 marked nests were successfully sampled from the latter half of July through the early half of September. Although the actual sampling period was longer only these periods contained adequate numbers of sample nests in both zones for analysis. In 1988, 88 marked nests were successfully sampled from July through September. In 1989, 113 marked nests from the latter half of July through the early half of October were sampled. The difference in sampling periods was due to the adjustment for the shift of temporal nest distribution between the two years Marked nests were examined during the beach census for signs of emergence, depredation, inundation, and any disturbance. When the nests were predated, the species of predator was identified by tracks. Emergence success reported in this study was the fraction of eggs that resulted in hatchlings that emerged from the sand. After emergence of hatchlings, the marked nest was excavated. The numbers of hatched and unhatched eggs, and the numbers of dead or live hatchlings remaining underground were determined. Unhatched eggs were opened to check for development; if development had occurred, size of the embryos was recorded. Unhatched eggs lacking visible embryos or blood formation were classified as infertile eggs. Evidence of underground disturbance such as by ghost crabs and flooding, and the number of eggs affected, were determined.

PAGE 21

14 During 1986 and 1988, clutch size was counted during deposition, and emergence success was determined directly. In 1989, when the clutch size was not counted, clutch size was estimated by the number of remaining eggshells. In most undisturbed nests, the hatched shells remained in one piece. If the shells were fragmented, the pieces were put together to represent one shell. Fowler (1979) used this method to estimate the clutch size of green turtles in Tortuguero and found that the error was no more than 8 eggs In the 1989 season, for nests that were partially destroyed or completely lost by any disturbance, the mean clutch size from 1989 (107.1) was used as the clutch size to calculate emergence success. Sampling of Nests to Determine Sex Ratio originally planned to obtain five nests to determine sex ratio in each zone and for each half month period in the nesting season. Night sampling was conducted seven to 10 times during each half month period to locate nesting females. Only females during the early stage of nesting behavior (before digging the egg chamber) were sampled. During the deposition of eggs, a thermocouple probe was placed at the approximate center of the nest. For comparison, and to allow measurement of metabolic heat from the eggs, another thermocouple probe was placed at the same depth 1 m along the beach from the egg chamber. To protect the temperature-monitored nests from activities of other nesting females, several logs found on the beach were

PAGE 22

1 5 placed around the nest; these logs were placed in such a way as to avoid creating shade on the sand surface above the egg chamber. The temperatures of nests and the sites 1 m away from the nests were monitored once a day at an interval of two to four days in 1986 and at an interval of two days in 1988 and 1989. During the 1986 and 1988 seasons, I collected samples of eggs to be sexed; during 1989, I monitored temperatures of the nests, but did not collect any eggs. After 50 60 days of incubation, I uncovered the nests and randomly selected 20 developing eggs. The distance from the bottom of the egg chamber to the sand surface was measured. The numbers of developing eggs and dead eggs were counted, and dead eggs were opened to check for any sign of development. Remaining developing eggs were reburied in the egg chamber, and eventually the day of emergence was recorded. In 1986, 24 nests were equipped with thermocouples, but subsamples of embryos to be sexed were collected from only 12 of these nests. In 1988, 92 nests were equipped with thermocouples, and sub-samples were collected from 42 of these nests. Loss of the sample nests was due to destruction by nesting females, drowning of whole clutches by flooding or hurricane, and depredation by animals. Nest temperature data were obtained from all the nests intended for future sexing of the hatchlings, except one nest in 1986. In 1989, 74 nests were set up only for temperature monitoring; nest temperature data were collected from 69 of these nests. Sand temperature data near each sample nest were unobtainable for some of the nest temperature monitored nests in all years: 3 nests in 1986, 8 nests in 1988 and 13 nests in 1989

PAGE 23

16 Histology and Sexing Sample eggs were kept in plastic bags with holes until pipping. Hatchlings were killed, and the gonads were fixed in 10% neutral buffered formalin. The gonads were transferred to the University of Florida for histological analysis. Collection permission was issued by Servicio de Parques Nationales Costa Rica through Mr. Fernando Cortes; the CITES permit number is 092-88 in Costa Rica, PRT725607 in United States. Transverse serial paraffin sections, 5-8 m thick were prepared from the median portion of each left gonad. Harris' haematoxylin and eosin were used for staining. Criteria for sexing green turtle hatchlings were the same as those reported by Spotila et al. (1983). Sand Temperatures To investigate the seasonal thermal profile of sand temperatures at the same depths as egg chambers of green turtles on the beach, temperatures at depths of 60 cm (all three years) and 80 cm (1988, 1989) in each zone were monitored along two (1986) or four (1988 and 1989) transect lines from July through early December in all three years. These transect lines included both relatively wide and narrow portions of the beach I placed a set of thermal probes at the border (50% shade) in the vegetation/border zone and at two to three points (0% shade) in the open sand zone: 5 m from the border point and others at 5 m intervals along the transect.

PAGE 24

1 7 The temperature readings were taken once every two to three days, with a Bairly BAT12 thermocouple meter. To determine the amount of diurnal fluctuation, several monitoring sessions lasting 24 h each, at 2 h intervals, were conducted during the study: 13 August and 29 September 1988, and 12 August and 14 October 1989. Statistical Analysis Both parametric and nonparametric statistical tests were used for data analysis. The arcsine transformation was applied to the percentage data (e.g ., emergence success and sex ratios). Whenever the required assumption of homogeneity of variance was met by using the Fmax test, parametric statistical tests were utilized. If the assumption was not met, nonparametric tests were substituted. To test binomial data (e.g nest site selection between the two zones), Chi-Square tests were utilized. The rejection level for the null hypothesis in all tests was alpha = 0.05. For those nests where complete data sets were not available, the existing data were included in analyses whenever possible. Each statistical test is mentioned in the results section.

PAGE 25

CHAPTER 3 RESULTS Physical Environment Surrounding Nests Air Temperatures Monthly minimum and maximum air temperatures from July through November in the 1986, 1988 and 1989 seasons are shown in Figure 3. Whereas minimum temperat r es varied little throughout O the season (between 23.3 and 24 1 c c), maximum temperatures showed slight increases in August and September for the three years. The ranges of monthly maximum temperatures for each year were 28.4 c c (July) 29.9 c c (September) in 1986, 29 2 c c (July and November) 30.s c c (September) in 1988 and 28.7 c c (November) 30.8 c c (August) in 1989. The highest daily air temperatures occurred during September of each year: 35.o c c on 1 September 1986 34 o c c on 23 September 1988 and 33.o c c on 2 and 3 September 1989 The lowest daily air temperatures were recorded as 22.1 c c on 10 July 1986 20.s c c on 7 July 1988 and 22.s c c on seven occasions throughout the 1989 season, except in September. Yearly variation in mean temperature throughout each season among three years was very low The mean minimum and maximum temperatures during each five-month period were 23 9 and 29.3 c c in 1 8

PAGE 26

19 1986, 23.4 and 29.7C in 1988 and 23.7 and 29.8 C in 1989, respectively. Overall mean three-year minimum and maximum temperatures during the same period were 23.7 and 29.8C, respectively. Rainfall Daily rainfall records from 16 June through 10 December during 1986, 1988 and 1989 are shown in Figure 4. The sampling period covered the entire incubation season of green turtle eggs in Tortuguero. For all three years, heavy rain events sporadically occurred. Rainfall greater than 100 mm during 24 hours was recorded on seven occasions in 1986 five occasions in 1988 and five occasions in 1989. The heaviest 24-hour precipitation each year was 186 mm on 21 October in 1986, 221 mm on 6 October 1988 and 155 mm on 25 October 1989. Among those intense rainfall events at least two events in 1986 (5 August, 141 mm/day; 6 December, 178 mm/day) one event in 1988 (6 October 221 mm/day) and one consecutive three-day event in 1989 (30 October-1 November, total 367 mm for three days) caused freshwater flooding on the beach and consequently damaged some green turtle nests (see section on ground water). The 1986 season overall was the wettest among the three years. The total rainfall during the nearly six month period (16 June-10 December, 178 days) was 3871 mm in 1986, 2578 mm in 1988 and 2896 mm in 1989. Mean rainfall taken at the Tortuguero park station during the same period from 1978 through 1989 (I have

PAGE 27

20 om itted 1984 data because of an incomplete data set, lnstituto Meteorogico Nacional) was 2882 mm (SE=622, n=11 ). The rainfall (3871 mm) in the 1986 season was the second highest on record; the greatest rainfall (3929 mm) was recorded in 1982. Figure 5 shows seasonal variation of monthly rainfall taken at the Tortuguero National Park Station from 1978 through 1989 (lnstituto Meteorogico Nacional, Costa Rica). The major incubation season of green turtles (July through November) includes several of the wettest months (July and November) during the year. September, however, which was the middle of the incubation period, had the least rain. Within the five-month comparison (July through November), the monthly rainfall during September was significantly less than that during July and November, but not significantly less than in August and October ( one-way ANO VA, df=4, p=0.064; Fisher's procedure of least significant difference, alpha=0.05). These data indicate that a drier period normally occurs during the middle of the incubation season of green turtles at Tortuguero. The monthly rainfall for the three study years and the mean rainfall from 1978 through 1989 are shown in Figure 6. During the 1986 season, the monthly rainfall during the early to middle incubation season well exceeded the mean values. The rainfall in August (846 mm) and in September (514 mm) during 1986 were the highest monthly records. As a result, the 1986 season did not experience a dry period in the middle of the season, and the monthly rainfall exceeded 500 mm throughout the entire season. On the other hand, the 1988 and 1989 seasons were much drier than the 1986 season during the middle of the incubation season. The rainfall in

PAGE 28

21 August 1988 (256 mm) and in September 1989 (157 mm) were the lowest ones since 1978. The rainfall in October of both years exceeded the average values. Consequently the 1988 and 1989 seasons showed distinctive seasonal fluctuation in rainfall. Vertical Position of Clutches A total of 344 sample nests (including both egg survivorship samples and sexing/temperature monitoring samples) during the three seasons was measured for the distance between the bottom of the egg chamber and the level of the sand surface on the beach at the time of excavation. Nest mounds above the clutches made by females were generally flattened to the level of the beach surface by the time of excavation. In addition, 43 sexing/temperature monitoring sample nests in the 1988 season were measured for the height of the clutch mass (a distance between the bottom and the top of egg mass) when 20 sub-sample eggs were taken for sexing at around 50-60 days of incubation. Bottom depth of green turtle clutches during the three years averaged 76.8 cm (SE=0.8, range=52105, n=176) in the open zone and 79.0 cm (SE=0.8, range=45-120, n=168) in the vegetation/border zone. The difference between the two zones was almost significant (t=1.93, p=0.0549). Overall bottom depths of green turtle clutches (two zones combined) averaged 77.9 cm (SE=0.6, range=45-120, n=344; Figure 7). Mean height of the clutch mass was 19.8 cm (SE=0.9, range=8-33, n=43). If the center of the clutch is assumed to be located at half of the height of the clutch (about 8.9 cm from the bottom), mean vertical

PAGE 29

22 position of the center of the clutch is 69.0 cm below the sand surface. The top of the clutch is 58.1 cm below the sand surface. Ground Water Figures 8 and 9 show seasonal fluctuations of the ground water table and rainfall during the 1986 and 1989 seasons. During the 1988 season, the wells were stolen several times during the study, therefore only very scattered data were available. The level of the water table on the beach was affected by rainfall. Continuous rainfall and occasional heavy rainfall were associated with a rise in the ground water table. During the 1986 season, which was very wet throughout, the level of the water table frequently rose close to the general depth of the clutches in both zones ( overall mean bottom depth of clutches was 77.9 cm; Figure 7). On the other hand, the level of ground water during the 1989 season rose close to the depth of clutches only early and late during the incubation season when rainfall was heavy. On several occasions, excessive rainfall caused freshwater flooding on the beach. Flooding events were confirmed by water marks that remained at the margin of a depression area on the beach that recorded the highest level of water pools during the excessive rain events. However, only one event (5 August 1986; 141 mm/day) was actually recorded on the well as a sharp increase of the water table much above the general depth of clutches on the beach. Because the measurements of the water table at the time of other excessive rain events were taken 6 to 9 hours after the rain had

PAGE 30

23 stopped, the levels of the water table recorded were probably much lower than the actual maximum level. The water marks remaining in the depression area suggested that the water tables on 6 December 1986 (178 mm/day), and on 1 November 1989 (total 367 mm for three days) also rose above the general depth of the clutches. During the 1988 season, freshwater flooding was observed on one occasion, associated with a heavy rain event on 6 October (221 mm/day) In addition to rainfall, high waves probably raised the water table. While the amount of rainfall was relatively small (28.5 to 37.0 mm) on 19 and 20 November 1989 the water table showed a substantial high level on 20 November. At the same time, very high waves (>2 m) were recorded. This height of waves rarely occurred in Tortuguero during the study. Except on this one occasion, waves of this height only occurred during the time that Hurricane Joan passed near Tortuguero on 21-22 October 1988. Data on the water table during the hurricane are not available. Waves Figure 1 O shows frequency distributions of relative heights of waves throughout each incubation season in 1986, 1988 and 1989. On the Tortuguero beach, heights of waves were primarily associated with the size of swells from offshore rather than with the wind speed over the beach. Wave actions were generally high in Tortuguero during most of each season. Less than one third of the days were identified as calm throughout each season (wave height :::; 0.5 m; 17. 7% in 1986, 31.4% in 1988 and 31.5% in 1989) Wave

PAGE 31

24 activity varied considerably with a similar seasonal trend throughout each season. Generally, there were more calm days during the middle of the season. For the three years combined, September was the month most likely to have a calm day. The amount of monthly rainfall was negatively related to the number of calm days (wave height :s:; 0.5 m) in the month (Y=17.147-1.656X, R=0.663, n=15; t=3.19, p=0 007; Figure 11 ). The greater the rainfall, the higher were the waves on the beach. Continuous days of calm waves resulted in sand accretion on the lower part of the beach, and the entire beach widened gradually. On the other hand, when the wave activity was higher (> 0.5 m), beach erosion was apparent and the shape of the beach changed rapidly. Heavy beach erosion frequently constructed beach platforms (up to 1.5 m height) along part of the beach. Green turtles either gave up trying to nest in these areas, or they crawled over the platforms and laid their clutches in a site protected from waves. Very few females nested below the platforms at the time of high waves. However, nests that had been deposited before the high platforms were formed or advanced well inland were washed away or completely inundated. Daily Fluctuations of Sand and Nest Temperatures To determine the extent of daily fluctuation of sand and nest temperatures, temperatures along two transects (two points in the vegetation/border zone, five points in the open zone) and temperatures at the center of clutches (14 nests in each zone) were

PAGE 32

25 monitored at two-hour intervals for 24 hours for four days during the study. Those clutches monitored averaged 23.0 days from deposition (SE=2.2, range=3-54, n=28) at the time of temperature measurement. Two of the days (13 August and 29 September 1988) had no rainfall and two days ( 12 August and 14 October 1989) had minor rainfall (2 mm and 25 mm, respectively) during the 24-hour monitoring period. In general, these four days are within the range of normal weather in Tortuguero. The sand temperatures at depths of 60 cm and 80 cm were relatively stable in both zones {Table 1 ). Mean ranges of daily fluctuation in these four categories were from 0.41 c to 0.49 C. The range of individual monitoring points on the transects was from 0.1 c to 0.8 C. The daily fluctuations in temperatures were not significantly different between the two zones, or between depths of 60 cm and 80 cm (two-factor ANOVA; zone, p=0.3515; depth, p=0.4716). The daily fluctuation of temperatures in the center of the clutches showed low ranges similar to the fluctuation in sand temperatures along the transects (Table 1 ). The range of daily fluctuation averaged 0.46 C in the open zone and 0.49C in the vegetation/border zone with the same individual range (0.2-0.8 C). The mean daily fluctuations of nest temperatures were not significantly different between the two zones (unpaired t-test, t=0 377, p=0. 710). Although daily temperature profiles varied with different weather, mean temperature profile among the four separate days showed a general trend of daily fluctuation (Figure 12). Sand

PAGE 33

26 temperatures at depths of 60 cm and 80 cm and nest temperatures in both zones showed a similar trend of daily profiles. While the period of highest temperatures was not distinct, the lowest temperatures were recorded from 1000 to 1200 in the morning. Times when the temperature was closest to the 24-hour mean were around 0600 and 1500. Most of my temperature readings were conducted during the two periods between 0700 to 1000 and between 1500 and 1700. Overall, the temperatures collected for this study were probably not far from the values of 24 hour means. Seasonal Fluctuation of Sand Temperatures In Tortuguero, local meteorological conditions affected the fluctuations of sand temperatures. Rainy days resulted in lower sand temperature, while sunny days resulted in higher temperatures. In general, amount of rainfall was inversely related with sand temperatures. Being associated with variable rainfall (Figure 4), the sand temperatures showed considerable seasonal fluctuations for the three years (Figure 13) Occasional heavy rain events, particularly those that caused freshwater flooding on the beach (Figure 4), resulted in rapid and substantial short-term declines of sand temperatures along the entire beach (Figure 13). In addition to such flooding rain events, overcast days with moderate rainfall sometimes resulted in a substantial decline of sand temperatures. A continuous rainy event from 21 through 25 July 1988 accumulated 165 mm of rainfall, and lowered the sand temperature up to 2.9 C at 60 cm depth in the open

PAGE 34

27 zone. During the entire five days, the sky was overcast during the daytime. Prolonged rainy days kept sand temperatures low for the duration and caused long-term fluctuation. The continuous low temperature during July through August in 1986 was one such situation. Table 2 shows overall mean sand temperatures throughout each season at depths of 60 cm and 80 cm ( only 1988, 1989) in the two zones. For all three seasons, the trends of spatial thermal profiles were identical. Sand temperatures at depths of 60 cm and 80 cm in the open zone were significantly higher than those in the vegetation/border zone. Within the same zone, sand temperatures at a depth of 60 cm were significantly higher than those at a depth of 80 cm ( one-way ANO VA with repeated measure; 1986, 1988, 1989, all tests p < 0.0001; one-way ANOVA of repeated measure with Bonferroni adjustment of P for multiple comparison, overall alpha = 0.05, 1988, 1989, all tests p<0.008). For overall mean sand temperatures throughout the season, the differences between the two zones ranged from 0.8C (1986) to 1.5C (1989) at a depth of 60 cm and from 1.1C (1988) to 1.3C (1989) at a depth of 80 cm. The differences between the depths of 60 cm and 80 cm ranged below 0.4C in both zones in 1988 and 1989. For overall mean sand temperatures throughout the season, the 1988 season showed the highest, the 1989 season the intermediate, and the 1986 season the lowest sand temperatures at a depth of 60 cm in both zones during the study. Figure 14 shows seasonal fluctuations of mean monthly sand temperatures at a depth of 60 cm for the three years in both zones.

PAGE 35

28 Mean monthly sand temperatures from July through November differed among years in both zones ( one-way ANOV A for each depth and zone; same results for all six tests, df=4, p<0.0001 ). Fisher's procedure of least significant difference identified those months where mean temperatures are significantly different from each other (Table 3). Although the seasonal trends of monthly sand temperatures varied from year to year, September's sand temperatures in both zones resulted in the highest rank within each season for all three years. Figure 15 shows the negative relationship between the amount of monthly rainfall and mean monthly sand temperatures at a depth of 60 cm on transects along the beach during the study (the open zone, Y=30.020-0.003X, R=0.718, n=15, 1=3.718, p=0.003; the vegetation/border zone, Y=28.524-0.002X, R=0.665, n=15, t=3.210 p=0.007). Seasonal Nest Distribution and Nest-Site Selection By the first censuses, minor nesting activity already had occurred in all three years. At least 16 nests in 1988 and five nests in 1989 were counted at the study area before the censuses started in the middle of June. Because the census in 1986 started in July, at least a half month of nesting activity at the beginning of the season was missed. The number of unchecked nests in 1986 was not known but was probably very minor relative to the whole season. Most nesting activity occurred from July through the first half of October in all three years (Figure 16). However, the seasonal distribution shifted among the years. Peak nesting activity occurred

PAGE 36

29 during the latter half of August in 1986 and the first half of September in 1988. In 1989, nesting activity reached a plateau from the latter half of August through September. The last nest in the study area was observed on 27 November 1986, 15 November 1988 and 20 November 1989. An insignificant amount of nesting activity was observed during November of all three years. Approximately 11,724 nests in 1986, 10,509 nests in 1988, and 4,491 nests in 1989 were deposited in the two-mile study area of the beach during the census period. The density of nests along the beach was 7.3 nests/m in 1986, 6 5 nests/m in 1988, and 2.8 nests/m in 1989. The proportion of nests deposited in the vegetation/border zone and in the open sand zone across the season was, respectively, 51.1 % and 48.9% in 1986 (n=3,543), and 58.0% and 42.0% in 1988 (n=3,521) and 63.2% and 36.8% in 1989 (n=2,041) (Figure 17). The proportions among these three years were significantly different (df=2, X 2 =81.8, P<0.0001 ). Tukey-type multiple comparison tests for proportional data revealed that the proportion of clutches deposited in the vegetation/border zone in 1989 was significantly the highest, intermediate in 1988 and the lowest in 1986 among the three years [q 0 05, 00 3 = 3.314, q=24.883 (86 vs 89), 16.226 (86 vs 88), 10.976 (88 vs 89)]. [Data were transformed by the equation: p'= 1 1 1 )/(n+ 1 )), Zar ( 1984 ).] Figure 17 also shows the seasonal fluctuation of nest site selection. The proportions of nests found in each zone were significantly different at each half-month period in each year (df=7, X 2 =31.7, P<0.0001 in 1986, df=8, X 2 =41.2, P<0.0001 in 1988 and df=8,

PAGE 37

30 X2=19.9 P<0.01 in 1989). Tukey-type multiple comparison tests for the proportional data (Zar, 1984) identified the differences among each seasonal period There was no common seasonal trend among the three years. In 1986, the proportion of clutches deposited in the vegetation/border zone was significantly higher early in the season than during the rest of the season and the proportion decreased as the season progressed. In 1988 and 1989, the proportion of nests in the two zones oscillated throughout each season in a similar way. The proportions of clutches deposited in the vegetation/border zone shifted from high to low, and then again from high to low throughout each season. Analysis of Egg Survivorship Samples Fates of Sample Nests The fates of 49 nests in 1986, 88 nests in 1988 and 113 nests in 1989 were determined. Mean clutch size for the sample nests was 116 9 eggs (SE=3.4 range=28-165 n=49) in 1986, 109.1 eggs (SE=2 1 range=53-148 n=88) in 1988 and 107.1 eggs (SE=2.1, range 53-152 n=89) in 1989 The frequency distributions of the emergence success rate of these nests were far from a normal curve and instead showed concave bimodal shapes in each year (Figure 18). For all three years, most of the sample nests fell into one of two extremes: highly successful emergence percentage (>70%) or entire dead clutches. No hatchlings emerged from 24.0% and 25.0% of sample nests deposited in the open zone and in the vegetation/border

PAGE 38

31 zone respectively in 1986, 21.6% and 39.2% in 1988, 9.4% and 28.3% in 1989. Total nest loss was caused by beach erosion inundation by excessive rain or Hurricane Joan, depredation by coatis (Nasua narica) and ghost crabs (Ocypode guadrata) and excavation by nesting female turtles (Figure 19). For each nest, emergence success was defined as the percentage of its egg clutch that produced hatchlings that successfully emerged from the sand column Mean emergence success throughout each season was 54 8% (SE=7 3, n=25) in 1986, 57.8% (SE=6.2, n=37) in 1988, and 74.7% (SE=4 8 n=53) in 1989 in the open zone ; emergence success was 47.3% (SE=8.1, n=24) in 1986 42.7% (SE=S 6, n=51) in 1988 60.0% (SE=S.4, n=60) in 1989 in the vegetation/border zone. Although the mean emergence success in the open zone was slightly higher than that in the vegetation/border zone during each year, the differences between the two zones were not significant (Mann-Whitney U test, p=0.778 in 1986 p=0.069 in 1988, p=0.061 in 1989). With pooling the two zones, overall mean emergence success of sample nests throughout each season was 51.2% (SE=S.4, n=49) in 1986, 49.1% (SE=4.2, n=88) in 1988 and 66 9% (SE=3.7, n=113) in 1989. The overall mean emergence success in 1989 was significantly higher than that in either 1986 and 1988 [Kruskal-Wallis Test, df=2, H=21.372, p<0.0001; nonparametric multiple comparison test with unequal sample sizes (Zar, 1984), 1989 vs 1988, 0=4.121 1989 vs 1986, 0=3.465, 0 0 05 3=2.394]. Emergence success exhibited a different seasonal trend for each year (Figures 20, 21 and 22). For analyzing the effects of

PAGE 39

32 seasonal period (half-month interval) and nest location (open zone vs vegetation/border zone) on egg survivorship, nonparametric two way factorial ANOVA with unequal replication (Zar, 1984) was applied for each year's data (Table 4). The ranks of the data, rather than the raw data, were used. The analysis revealed a significant difference in seasonal emergence success in 1988, but not in 1986 or 1989. There were no significant effects of Zone or Zone x Period interaction in any of the three years These results agree with an analysis carried out with a parametric ANOV A in which case the data were transformed to their arcsine, but the assumption of homogeneity was not satisfied. For the 1988 data, a nonparametric multiple comparison test with unequal sample sizes (Zar, 1984) identified a significant seasonal difference of emergence success between the clutches deposited in the latter half of July (79.7%, SE=4.6, n=16) and ones in the latter half of August (16.9%, SE=9.2, n=14)(O=4.10 > Qo.os,6=2.94; Figure 21, total). The other comparisons between each period were not significant. The significant seasonal difference in the 1988 season was mainly accounted for by two heavy inundations that occurred close together in October These inundations caused by excessive rain on 6 October (221 mm/day) and by the surf of Hurricane Joan on 21-22 October caused substantial loss of nests deposited after the latter half of the three month nesting season. The sample nests deposited in the latter half of July in 1988 hatched before these inundations, and all these clutches produced hatchlings. However, the inundations caused the complete loss of seven of 14 nests deposited during the latter half of August 1988.

PAGE 40

33 In addition, predation loss of three nests by mammals and one nest by ghost crabs further decreased the emergence success during this period. Abiotic and Biotic Factors Affecting Egg Survivorship Fate of eggs from sample nests Table 5 and Figures 23-24 show the fate of green turtle eggs from sample nests in 1986, 1988 and 1989. The emergence success of eggs from the sample nests throughout each season was 54.6% (n=2915) in 1986, 59.5% (n=3899) in 1988, and 75.5% (n=5890) in 1989 in the open zone; comparable figures for the vegetation/border zone are 46.4% (n=2801) in 1986, 43.0% (n=5715) in 1988, and 58.5% (n=6216) in 1989. Criteria for assessing the cause of egg mortality The category of "Destroyed by mammals" includes all eggs lost because of initial damage by mammals. For example, when coatis excavated green turtle nests, they rarely consumed all the eggs. However, soon other animals such as black vultures, ghost crabs and ants often destroyed the remaining eggs in the half excavated nest. In this case, I categorized the entire clutch as lost under "Destroyed by mammals." If the partially depreciated nest successfully produced hatchlings, the numbers of missing egg shells (clutch size minus counted egg shells at excavation), which were assumed to be removed from the nests by mammals, were classed under this category. The same principle was applied to the category of "Destroyed by female turtles."

PAGE 41

34 For assessing the cause of embryonic death, when the timing of inundation reasonably matched the stage of the arrested embryos, I assumed these eggs were killed by the inundation event. In the 1988 season, because two inundations (6 October excessive rain and 21-22 October Hurricane) occurred so close to each other, I could not determine how much of egg mortality was caused by each inundation. Abiotic factors Among abiotic factors, erosion and inundation by surf, excessive rainfall resulting in inundation from ground water, and Hurricane Joan were mainly responsible for reducing egg survivorship. These abiotic extremes, as a whole, were responsible for loss of 18.5% of eggs in the open zone and 20.0% of eggs in the vegetation/border zone throughout the three years (the percentages of egg loss for the three years were calculated as the weighted mean in each zone). In addition to these abiotic extremes, artificial debris prevented a few hatchlings from emerging. Beach erosion washed away 5.7% (open zone) and 2.9% (vegetation/border zone) of the eggs throughout the three years. The damage from erosion showed considerable yearly variation, and the 1986 season suffered the highest loss (15.3% of eggs in open zone, 8. 7% in vegetation/border zone in the 1986 season). Surf inundation did relatively minor damage to the eggs throughout the three years (1.4% in open zone, 1.3% in vegetation/border zone). Hurricane Joan, whose center passed about 180 km off the coast of the Tortuguero beach in the Caribbean Sea during the night

PAGE 42

35 of 21 through the morning of 22 October 1988 caused substantial damage to green turtle nests. When Hurricane Joan passed, the Tortuguero beach was at the outer margin of a 70 km/h wind speed zone (lnstituto Meteorologico Nacional, Costa Rica) During the first beach census after the hurricane, I confirmed that the waves apparently washed up to the vegetation area along most of the beach and that all developing sample nests were covered by waves to various extents. However, beach erosion was not severe and none of the sample nests was washed away All damage by Hurricane Joan on the sample nests was caused by surf inundation. Hurricane Joan brought only a little rainfall to the Tortuguero beach (total 25 mm during 21 to 24 October 1988). It is not known whether the ground water raised by the high waves or the surf salt water itself suffocated the eggs. The extent of damage caused by Hurricane Joan in 1988 was not accurately assessed because the damage to several nests was inseparable from that caused by excessive rain on October 6. The minimum estimated damage by Hurricane Joan was 1.6 % egg loss in the open zone and 10.8% egg loss in the vegetation/border zone. However if I include the undetermined dead eggs killed by either the hurricane or the excessive rain, the figures increase to 9.5% egg loss in the open zone and 13.3% egg loss in the vegetation/border zone. No other hurricane threatened the Caribbean coast of Costa Rica during the three-year study. Freshwater inundation by ground water, mainly associated with excessive rainfall, was a major and constant abiotic factor causing mortality of eggs throughout all three years. All developing

PAGE 43

36 stages, including emerging hatchlings, were vulnerable to suffocation. Clutches in both zones were vulnerable to flooding. Excluding eggs that died from unknown causes between the flooding and Hurricane Joan in 1988, freshwater inundation was responsible for 8.4% of egg loss in the open zone and 11.3% of egg loss in the vegetation/border zone throughout the three years. When eggs of undetermined fate are included, the figures increase to 11.0% and 12.1 %, respectively. The risk of mortality by flooding between the two zones was not consistent among years. The percentages of eggs lost by freshwater inundation are as follows: 13.6% and 17 .8% in 1986, 3.0-10.7% and 12.5-15.0% in 1988, 8.5% and 3.8% in 1989 for eggs deposited in the open zone and in the vegetation/border zone, respectively. In 1986 and 1988, significantly more eggs were killed by flooding in the vegetation/border zone than in the open zone (X 2 =18.7, p<0.0001, 1986; X2=7.3, p=0.0068, using possible closest rate in 1988), while the opposite trend occurred in 1989 (X2= 116.5, p<0.0001, 1989). Artificial debris, such as tangled fishing nets and ropes were frequently seen on the beach. On one occasion, a buried tangled rope physically prevented 1 O hatchlings of a sample nest from emerging to the surface, although the remaining 119 hatchlings escaped through the rope. Fishing nets and ropes have a high potential to cause substantial damage to emerging hatchlings. Biotic factors Predation by several types of animals and excavation by female turtles were major factors responsible for mortality of

PAGE 44

37 green turtle eggs. Altogether, 5.4% of the eggs in the open zone and 20.5% of eggs in the vegetation/border zone were destroyed throughout the three years by these causes. Significantly more eggs in the vegetation/border zone were destroyed than in the open zone for all three years (3.7% vs 14.0% in 1986, 6.2% vs 15.9% in 1988, 6.2% vs 31.4% in 1989; X 2 =188.9, p<0.0001, 1986; X 2 =209.4, p<0.0001, 1988; X2= 1246.0, p<0.0001, 1989). Coatis, Nasua narica, were responsible for the majority of nest loss by mammal predation. However, because it was sometimes difficult to confirm the mammal species on the sample nests only by their footprints, all nests excavated by mammals were classed as "Destroyed by mammals." Coatis are diurnal predators on green turtle nests in Tortuguero. In the protected park area, solitary or more often a band of coatis were frequently sighted walking along the upper part of the beach during the daytime. The band membership typically consisted of two to six cubs and several adults. They generally excavated a series of adjacent nests along the border of the vegetation. Although they rarely consumed all the eggs in a nest, the rest of the eggs were soon eaten by other animals. For the sample depredated nests, 72.7% of nests depredated by mammals (n=33) did not produce any hatchlings. Raccoons, Procyon lotor, were also responsible for destruction of nests, but to a lesser extent. The extent of damage was not accurately obtained because of the difficulties of species identification. For the depredated sample nests, only one nest in the vegetation/border zone in 1989 was confirmed to be damaged by

PAGE 45

38 raccoons. The density of raccoons in Tortuguero was probably much lower than that of coatis, because I sighted only two adult raccoons at night and one juvenile during the daytime during the three years of study. The juvenile was seen eating several eggs of an unsampled nest under the vegetation. In addition to the above two predator species, one skunk and a few feral dogs were sighted excavating several unsampled nests in the study area. It is not known whether some of the sample nests were depredated by these species. A skunk (unknown species) was seen excavating an unsampled nest in 1986; this was the only time a skunk predator was sighted during the study. Four unsampled nests in 1988 were predated by feral dogs. Feral dogs were rarely seen in the park area throughout the study because of a successful dog control program on the beach by personnel of Parque Tortuguero. Predation by mammals was responsible for loss of 3.0% of the eggs in the open zone and 17.3% of the eggs in the vegetation/border zone throughout the three years. Only four of 115 (3.5%) sample nests in the open zone were partially or completely preyed upon during the three years as compared to 29 of 135 (21.5%) sample nests in the vegetation/border zone (X 2 =16.0, p<0.0001 ). There was considerable yearly variation in predation by mammals. In the 1989 season when the number of nests deposited was about half that of the 1986 and 1988 seasons, predation by mammals increased to the highest percentage in both zones ( egg loss of 5.0% in the open zone, 30.9% in the vegetation/border zone in the 1989 season). The 30 9% egg loss by mammal predators in the

PAGE 46

39 vegetation/border zone in the 1989 season was the highest loss factor among all the categories throughout the three years. The mean time of predation for the sample nests for which the predation date was confirmed was 41.1 days after deposition (range=Day O to Day 72, SE=4.0, n=26) (Figure 25). The distribution of predation days was composed of two groups: before 11 days and after 32 days. Most (79.2%) of the egg depredation occurred between 32 and 68 days of incubation. One nest was de predated at Day 72 when the hatchlings were emerging to the surface. For a supplemental observation on mammal predation, recorded the number of recently depredated nests in the study area during the beach census. For all three years, more nests depredated by mammals were observed in the vegetation/border zone than in the open zone (X 2 =82.9, p<0.0001, n=233, 1986: X 2 =152.0, p<0.0001, n=330, 1988: X 2 = 174.3, p<0.0001, n=456, 1989) (Figure 26). Although nest predation by mammals was observed throughout the incubation period, the seasonal trend of predation intensity varied considerably among the three years (Figure 27). The trend in 1988 showed an almost opposite trend to that in 1989. While predation intensity in the 1988 season increased to the highest rate in the early part of the season and gradually decreased as the season progressed, nest predation in the 1989 season steadily increased as the season progressed almost to the end. The peak of nesting activities occurred between the latter half of August and September for the three years (Figure 16). The seasonal fluctuation of predation intensity did not match the seasonal fluctuation of nesting activities.

PAGE 47

40 Although ghost crabs, Ocypode guadrata, were very common on the Tortuguero beach, predation by ghost crabs was not a serious factor affecting egg survivorship (mean egg loss of 0.8% in the open zone and 1.3% in the vegetation/border zone throughout the three years) Only three sample nests were confirmed to have been damaged by ghost crabs during the three-year study. Two of them were completely destroyed, and the other was partially destroyed by ghost crabs. For those nests that were completely destroyed, numerous crab burrows were observed around the nests during the incubation period; at the time I excavated the nests to check on the eggs, only the flagging tape marker and many scattered torn shells remained at the bottom of the nest. For the partially destroyed nest, I used a characteristic nipped hole on the egg shells for identifying the eggs damaged by ghost crabs. However, my s e. s_ ~ment rate of .----" ghost crab predation for this study might b f conserv ~t ~~ ~ ) In cases where only a few eggs from a clutch were d~'dated by crabs and the typical nipped holes were not clear enough to identify, torn shells were classed under the "Rotten intact or ruptured eggs" category. Predation by termites, unknown species, was a very minor factor causing mortality of eggs. Eight sample nests were found to be partially included in termite nests, but generally only small numbers of eggs from each clutch (mean 3.7 eggs, SE=2.1, range=1 to 16, n=7) were killed by surrounding clay materials. The insides of most eggs were stuffed by the clay material through a few tiny holes. The termite nests were typically constructed under washed

PAGE 48

41 out logs on the beach, and all sample nests depredated by termites were located near such logs. An unidentified animal was responsible for minor damage on eggs (mean egg loss of 0.9% in the open zone, 0.2% in the vegetation/border zone throughout the three years). Typically, there were a few tiny holes (diameter 1-2 mm) on the egg shells, which were different from the nipped holes by ghost crabs, and the insides of the shells were often partially stuffed with dry sand. There were no visible animals in the shells at excavation. Mean number of eggs damaged by this cause was 6.2 per clutch (SE=2 1, range=1-47, n=23). I suspect that fire ants, which were very common on the Tortuguero beach, might be responsible for these damages, but I have no confirmation. Excavation of sample nests by nesting female turtles occurred twice in the 1986 season and six times in the 1988 season, but never in the 1989 season. The extent of damage by this cause (egg loss of 1 .1 % in the open zone and 4.8% in the vegetation/border zone throughout the three years) was minor relative to predation by mammals. Generally the damage occurred when a female dug an egg chamber overlapping a previously laid clutch. In such a case, numerous eggs were excavated to the surface. Two of the damaged sample nests did not produce any hatchlings because mammal and other predators destroyed the rest of the eggs. Another six sample nests produced some hatchlings; those nests were well covered by sand from another or the same female turtle's activities before predators invaded the nests.

PAGE 49

42 As a supplemental observation on excavation by female turtles, I recorded the number of recently excavated nests by turtles in the study area during the beach census. In total, 238 nests in 1986 (48 surveys), 123 nests in 1988 (56 surveys), and 19 nests in 1989 (80 surveys) were observed to be excavated by female turtles. The percentage of the excavated nests of total nests in each season was calculated as 6.5% in 1986, 3.5% in 1988, and 0.9% in 1989, respectively. In the 1989 season when nest density was the lowest among three years, the intensity of excavation by turtles decreased to a very low level relative to the other two years (Figure 28), and none of the sample nests was excavated. The seasonal fluctuation of excavation by turtles was positively related to the extent of nesting activities (Figure 29). Only one sample nest was damaged by plant roots. A bundle of young coconut tree roots infiltrated three eggs of a sample nest, which was located close to the trunk of the tree. Several unsampled nests were found to be partially tangled by roots of sea oats, but no apparent damage by the roots was recognized. These roots did not penetrate any egg shells, and those nests had high emergence success. Natural damage by plant roots was probably minimal on green turtle eggs at Tortuguero. Other Categories of Mortality Over the three years, a mean of 3.6% of eggs in the open zone and 3.1 % of eggs in the vegetation/border zone ceased embryonic development from no apparent cause. Actual causes of mortality of these eggs, whether genetically inherent or environmentally induced,

PAGE 50

43 are not known. Possibly, some previous physical disturbance, such as a minor inundation, caused the arrested development of some of these eggs at a later stage. Over the three years, 3.3% of the eggs in the open zone and 1.9% of the eggs in the vegetation/border zone did not show any sign of visible embryos or blood formation. I did not use a white circle or patch on the egg shell as a criterion of development. Therefore, the above figures probably overestimate the percentage of infertility. Over the three years, 5.3% of the eggs in the open zone and 4.3% of the eggs in the vegetation/border zone were rotten with intact shells or were ruptured at the time of excavation. Most late stage embryos were not included in this category because the remaining shell and bones were detectable even in a ruptured shell. However, some of the earlier stage developing eggs and infertile eggs were possibly included in this category. Analysis of Sex Ratio Sexed Samples A total of 55 nests ( 12 in 1986 and 43 in 1988) were sampled for sexing (Table 6), and the temperatures of 52 of those nests were monitored successfully. Twenty eggs were collected as a subsample from each clutch during 50-60 days except from one sample nest in 1986. Because I missed sampling eggs in this nest before the emergence of the clutch, only the two remaining hatchlings in the

PAGE 51

44 nest were collected for sexing. Some embryos were not fully developed and some tissues deteriorated before being fixed, thus the number of sexed gonads was reduced to an average of 17 .8 gonads per clutch (SE=0.4, range=2 to 20, n=55 nests, total turtles=979). Undetermined gonads, which possessed both germinal epithelium (female component) and seminiferous tubules (male component), were found in seven of the 979 individuals (0 7%). Sex Ratio vs Temperature To date no laboratory experiment to determine accurately the critical period of green turtle eggs in sex determination has been conducted. Spotila et al. (1987) found that mean incubation temperatures during the middle third of development explained the observed sex ratio in green turtles on the natural beach in Tortuguero. Results of temperature shift experiments in the laboratory with loggerhead turtles, Caretta caretta. (Yntema and Mrosovsky, 1982) also suggested that the thermosensitive period for sex determination occurs during the middle third of development. The middle third of development is also the critical period in freshwater turtle species (e.g., Bull and Vogt, 1979). Thus, I used the middle third of the development period as a critical period in green turtles for this study. Emergence lag (interval between pipping and emergence) in green turtles is believed to take from 3 to 7 days (Balazs and Ross, 1974), but no quantitative data are available. Recently, Christens (1990) found the mean lag period to be 5.4 days for loggerhead

PAGE 52

45 turtles. For this study, I assumed 5 days as the emergence lag in Tortuguero green turtles, and I subtracted 5 days from the deposition to emergence period to obtain an approximate developmental period for each clutch. Mean temperature during the middle third of the development period was calculated for each clutch. Figure 29 shows the relationship between sex ratios of sample nests and mean nest temperatures in the center of the clutch and mean sand temperature (1 m away from the clutch, at the same depth) during the critical period. The proportions of females were positively correlated with ambient temperatures. Both sexes were produced over the entire temperature range of the sample nests (mean nest temperature range=27.0-30 7C, mean sand temperature range=26.4 30.2C}. Because metabolic heat is produced during the middle of development, the mean nest temperatures were slightly higher than the sand temperatures at the same nest during the critical period (mean difference=0.71 c, SE=0.05, range=0.18-1.48, n=41 ). As a result, the regression of sex ratios and mean sand temperature was shifted down approximately 0.7C on the function of temperatures, as compared to mean nest temperature data. The pivotal temperatures (expected equal sex ratio) were calculated by logistic regressions as 29.4C for the mean nest temperatures, and 28.5C for the mean sand temperatures (see Prediction of Sex Ratio from Independent Variables for a detailed explanation of the logistic regression model). For both data sets, there was a substantial amount of variation in sex ratio over the range of incubation temperatures.

PAGE 53

46 Seasonal Variation and Zone Effect in 1988 In 1988, the sex ratios varied from O to 100% female for the sample nests in the open zone and from O to 89.5% female for the sample nests in the vegetation/border zone (Table 6). Figure 30 shows the seasonal variation of sex ratios in the 1988 season. The extent of seasonal fluctuation was more apparent in the open zone than in the vegetation/border zone. For analyzing the effects of seasonal period (half-month interval) and nest location (the open zone vs the vegetation/border zone) on sex ratio throughout the season, two-way factorial ANOVA with unequal replication was applied (Table 7) Data were transformed 1 1)/(n+ 1 )) (Zar, 1984 ). An Fmax test showed homogeneity of variances of the transformed data. The analysis revealed significant differences in seasonal fluctuation of sex ratio and in effects of Zone, but no significance in Period x Zone interaction. Fisher's procedure of least significant difference revealed a significant seasonal difference of sex ratios in the open zone (Figure 31), but not in the vegetation/border zone. In the open zone, the sample nests deposited in the latter half of July had significantly more females than during other periods within the season except for the nests deposited in the first half of August. From the end of August through September 1988, the sand temperatures, at least in the open zone, rose above the pivotal temperature (28.5C), associated with less rainfall (Figure 6, Figure 13). Most sample nests in the open zone deposited from the latter

PAGE 54

47 half of July to the middle of August encountered a higher sand temperature period during the critical stage and produced female biased sex ratios. However, the nests in the vegetation/border zone during the same period showed only a slight increase towards a female-biased ratio. The difference in sex ratios between zones during the high sand temperature period was significant (the open zone, mean=82.1%, SE=7.9, range=25.0-100.0, n=10; the vegetation/border zone, mean=31.7%, SE=11.3, range=7.1-89.5, n=7, one-way ANOVA, df=1, F=14.24, p=0.018, sample nests deposited from the latter half of July to the first half of August). A heavy rain event on 6 October (221 mm/day) and inundation by Hurricane Joan during 21 to 22 October decreased the sand temperatures (Figure 13). In both zones, two of eight sample nests deposited during the latter half of August and all 11 sample nests in September were affected during the temperature sensitive period by this cooling. Eleven of those 13 sample nests showed male-biased sex ratios in both zones (the open zone, mean=20.9%, SE=8.5, range=0.0-60.0, n=7; the vegetation/border zone, mean=20.7%, SE=10.2, range=0.0-56.3, n=6), and there was no significant difference between the zones (one-way ANOVA, df=1, F=0.01, p=0.923). Overall Sex Ratio in 1988 To estimate overall sex ratio, frequency of temporal nest distribution was combined with egg survivorship and sex ratio data by half month intervals. Because a significant effect was found in the seasonal factor, but not in the zone factor on egg survivorship

PAGE 55

48 (Table 4), the egg survivorship data in both zones were pooled. Since both the seasonal and zone factors significantly affected sex ratio (Table 8), each factor was treated separately. Because multiple comparison tests on both egg survivorship and sex ratio data for seasonal analysis failed to assign each half-month period to significantly different subsets, mean values on each bimonthly base were applied. Table 8 shows the value of each parameter applied and the process of estimation. Overall sex ratio was calculated to be 40.6% females in the 1988 season. Seasonal Variation and Zone Effect in 1986 In 1986, numerous rainy days and sporadic heavy rains kept sand temperatures lower than the pivotal temperature for most of the season (Figure 13). Eight of 12 sample nests produced only males (Table 6). The proportion of females in sample nests varied from O to 4 7 .1 % in the open zone and from O to 68.8% in the vegetation/border zone. Because of the small sample size and sampling bias, assessing the difference between sex ratios between the zones was not possible. To analyze the seasonal factor, data in both zones were pooled by monthly intervals. Figure 32 shows the seasonal variation of sex ratios in the 1986 season. Throughout the season, sex ratios were strongly biased toward males. There was no significant difference in the sex ratios observed throughout the season (Kruskal-Wallis test, df=2, Hc=0.983, p=0.612; because of heterogeneity of variance, nonparametric analysis was applied).

PAGE 56

49 Overall Sex Ratio in 1986 Because a significant seasonal effect was not detected (as mentioned above), overall proportion of females was calculated as a mean value of all sexed sample nests to be 10.1% (n=11) in the 1986 season. The number of sexed sample nests was biased towards nests in the open zone (8 of 12, 66.7%) as compared to the observed nearly 1 :1 distribution between the two zones (Figure 17). Because the open zone showed higher sand temperatures than the vegetation/border zone (Figure 13), the 10.1 % proportion of females may be an over-estimate. Prediction of Sex Ratio from Independent Variables To assess predictability of sex ratios of Tortuguero green turtle hatchlings from independent variables, simple and multiple logistic regression models were applied. Linear correlations between nest temperature, an assumed main environmental factor, and other variables were also analyzed. For the independent variables, mean nest temperatures and mean sand temperatures during both the critical period and the incubation periods, three different sets of rainfall data, bottom depth of nests, and zone of the nests were included. The data set included 52 nests consisting of 944 sexed turtles, for which nest temperatures were collected in 1986 and 1988. The data set on sand temperatures was reduced to 41 nests consisting of 728 sexed turtles from the above data set.

PAGE 57

50 Logistic Regression Model For dichotomous dependent variables (i.e., female or male), a logistic model (F(Z)=exp(Z)/(1 + exp(Z)), z=B1 +B0*X) is an attractive alternative to the linear probability model because the logistic model satisfies 0-1 constraint on probability unlike the linear specification, and provides a smooth symmetric S-shaped curve (Aldrich and Nelson 1984). Because this model is one of the most widely used nonlinear models, the availability and flexibility of computer programs are better. The logistic models were calculated with maximum likelihood estimation, and pseudo R 2 was computed as (-loglikelihood for Model) / (-loglikelihood for C Total) (JMP version 2.02, SAS Institute Inc., 1989). Logistic Regression with a Single Variable Nest and sand temperatures Mean nest temperatures were highly correlated with mean sand temperatures during the critical period (mean nest temperatures=0 99 + 1.02 mean sand temperatures, R 2 =0.92, n=41, p<0.001 ). Mean nest temperature data showed the best fit (pseudo R 2 =0.20) to a logistic model among the variables for a single regression model (Table 9, Figure 33). However, the variability of sex ratios with mean nest temperatures was still very high. Particularly at the middle range of temperatures, residuals spread up to 0.4 probability (Figure 33). Mean sand temperature data

PAGE 58

51 showed a slightly lower fit to the logistic model (pseudo R 2 =0 .17, Table 9) than did mean nest temperature data. Incubation period Incubation periods of the sexed sample nests varied from 51 to 74 days (mean=61.7, SE=0.7, n=52). Incubation periods were negatively correlated with mean nest temperatures (mean nest temperatures=38.3 0.15 incubation days, R 2 =0.49, n=52, p<0.001, Figure 34). Therefore, female sex ratio was also negatively correlated with incubation period (Figure 35). Only males were observed in nests that had over 68 days of incubation, whereas both sexes were found in nests that had incubation periods of between 56 and 66 days. Incubation period data were the second best fitted regression model (pseudo R 2 =0.18, Table 9). However, the variability of sex ratio for the middle range of incubation days was so high that the observed ratios almost covered the entire range of sex ratios. Therefore, predictability of sex ratio by incubation period was poor. Rainfall The amount of rainfall recorded at each sample nest was calculated three ways for statistical testing: R-l=mean daily rainfall throughout the critical period, R-ll=m~an daily rainfall throughout the critical period plus the previous 1 O days, R-Ill=mean daily rainfall from the deposition of eggs throughout the critical period. Rainfall decreased the sand temperature at the depth of nests. All three data sets of rainfall were negatively correlated with the mean sand temperatures (R2=0.16 (R-I), 0.33 (R-11), 0.47 (R-11I), all

PAGE 59

52 models, n=52, p<0.05). As expected, the proportion of females decreased with an increase of mean daily rainfall (Figure 36). This trend was the most apparent for the nests in the open zone and as a whole in the data set of R-111, which showed the highest correlation with mean nest temperatures. There seemed to be an interaction between rainfall data and zones. When the mean daily rainfall was low ( < 8 mm/day), the nests in the open zone showed higher female sex ratios than ones in the vegetation/border zone in general. On the other hand, as the mean daily rainfall increased, there was no apparent difference of sex ratios between the zones. R-111 pooled zone data (pseudo R 2 =0.17) fit a regression model almost as well as incubation period and mean sand temperature data did. If only data for the nests in the open zone were used, the pseudo R 2 increased to 0.27. However, variation of the data was still high for the whole range. Bottom depth of clutch Bottom depth of sample nests varied from 59 to 105 cm (average=78.3, SE=1.4, n=52). In general, sand temperatures decreased with depth of the nest. However, no significant correlation between mean nest temperatures and bottom depth was observed (p=0.215, n=51 ). As expected, there was no apparent trend between the sex ratios and the bottom depth of the sample nests (Figure 37).

PAGE 60

53 Multiple Logistic Regression Model Because of a strong correlation between mean nest temperatures and mean sand temperatures (R 2 =0.92), only mean nest temperature was used for the multiple regression model to avoid a problem of multi-colinearity. Mean nest temperature, incubation period, three sets of rainfall data, bottom depth of nests, and zone of the nests (the open zone=1, the vegetation/border zone=0) were entered into logistic multiple regression models. Backward elimination (Table 10) and stepwise regression procedures were used to eliminate nonsignificant variables (alpha=0.05). For both procedures, the same model was selected, in which mean nest temperatures, incubation period, mean daily rainfall from deposition through the critical period (R-I11), and zone significantly explained some variation of sex ratios (pseudo R 2 =0.24, Table 11 ). Figure 38 shows residuals on this selected model. Although the selected model was slightly improved from the best single variable model of mean nest temperatures (Figure 33), fitness of the model was still low. All first order interactions of the selected variables were tested individually by a likelihood ratio test (Hosmer and Lemesho, 1989) to determine their significant contribution to the selected model. Two interactions, between incubation period and zone (G=10.42, df=2, p<0.025) and between mean daily rainfall (R-11I) and zone (G=7.44, df=2, p<0.01 ), were significant. However, in both cases when each interaction was included in the selected model, the

PAGE 61

54 variable of incubation period became insignificant. Moreover, improved fitness was negligible (pseudo R2=0.25 for both models). Finally, two easily measured parameters--mean daily rainfall (R-11I) and zone--were entered into a multiple regression model to assess their predictive ability relative to that of nest temperature data. Because there seems to be an interaction between rainfall data (R-III) and zones in affecting the sex ratios (Figure 36), their interaction was also included. Mean nest temperature is biologically the primary factor for determining sex of green turtle hatchlings and the single regression model showed it to be the best single fit model among the applied independent factors. The fitness of the multiple regression model using rainfall and zone (pseudo R2=0.23) (Table 12) was slightly better than that of the single regression model using mean nest temperatures (pseudo R2=0 20).

PAGE 62

CHAPTER4 DISCUSSION Nesting Density and Temporal Nest Distribution High yearly fluctuation of nesting density is known for Tortuguero green turtles (Bjorndal et al., 1985). The 1986 and 1988 seasons were two of the highest density years since 1956, whereas the 1989 season was in the middle of the range according to the long term monitoring project at the northernmost 8.1 km section (K. Bjorndal, pers. comm.). My study site was part of the highest nest density area along the beach (Carr et al., 1978), and this trend was observed throughout the study (Horikoshi, unpublished data). Thus, the recorded densities in 1986 (7.3 nests/m) and in 1988 (6.5 nests/m) throughout each season were probably among the highest existing at the Tortuguero beach since 1956. As far as I know, nesting densities of green turtles of this magnitude are only known on an Australian island, an Oman beach, and an Ascension beach (Groombridge and Luxmoore, 1989; Mortimer, 1981; Mortimer and Carr, 1987). Temporal distribution of green turtle nests varied greatly in the frequency pattern and slightly in the range of season among the three years. Carr et al. ( 1978) noted that the main breeding activity takes place in July, August and September, although a few turtles nest throughout the year. The 1988 season in general agreed with 55

PAGE 63

56 this trend, but substantial nesting activity was extended at least until the middle of October in the 1986 and 1989 seasons. The number of nests deposited in October (662) exceeded that in July (389) in the 1989 season. Carr et al. (1978) presented data on the mean seasonal profile of nesting arrivals between 1956-1976, in which the peak of nesting arrival occurred at the end of August. However, because green turtles deposit multiple nests ( 1-7) within a season (Carr et al., 1978), the pattern of nest number profile cannot be clearly estimated from their figure. So far, the only available information on the full seasonal profile of green turtle nests at Tortuguero is for the 1977 season taken by Fowler (1979). She noted that nesting began in June, peak activity occurred in early August, and only a few turtle nests were found by late November in 1977; she showed the actual nest profile only from 13 July through 14 September. The three years of this study also showed the beginning of nesting activity to be sometime in June, and minimum activity to be in November, but the timing of peak activity was different; in this study the peak was in the latter half of August in 1986, in the first half of September in 1988, with a plateau from the latter half of August through September in 1989. The principal nesting season in Tortuguero green turtles seems to be confined to June through October, with peak activity sometime between August and September, but the actual dates vary from year t o year.

PAGE 64

57 Seasonality of Environmental Parameters Because the Tortuguero beach is located in the equatorial zone, seasonal fluctuation in air temperatures is very low relative to the high fluctuation in rookeries in the temporal zone. However, there is predictable seasonal change in several environmental factors that are important in influencing egg survivorship, sex ratio, and possibly nest site selection between different thermal zones in Tortuguero green turtles during their incubation period (June through December). The rainfall record indicated that there are two cycles of rainy/dry seasons in Tortuguero and that a high level of rainfall fluctuation occurs at the middle of the green turtle incubation period; rainfall decreases in September. Since sand temperature was inversely related to rainfall, the warmer sand temperature on the beach was expected to occur in September relative to the rest of the season. This tendency was observed during all three years of this study. Although seasonal trends of wave activity varied among the three study years, the wave activity was relatively low in September for each season. The positive correlation found between wave activity and rainfall suggests that fluctuation of wave activity also has a general seasonal trend; September has the most calm days at Tortuguero. At Tortuguero, the mean tidal range is 0.2 m (Limon/Bluefield, Tide Table 1986: East Coast of North and South America). Therefore, the high fluctuation of wave size (almost none to >2.0 m) has more influence on the width of the beach than the tide

PAGE 65

58 does on the emergence time of nesting females. In addition, the extent of erosion is generally also correlated with activity of waves. Spatial Distribution between the Zones It is interesting to compare the spatial distribution of green turtles in 1986, 1988, 1989 at this study site with similar data collected in a different area of the Tortuguero beach (northern-most 8 km) in 1986, 1987, 1988 (Bjorndal and Bolten, 1992). Although there was a slight difference in definition of the zonation of the beach (my vegetation/border zone is probably slightly larger than their combined vegetation zone and border zone), both studies revealed significant yearly variation of spatial distribution between the shaded area and open area over years, and that there was no common seasonal trend of fluctuation among the years. Comparison between the common years ( 1986 vs 1988) reveals that the yearly fluctuation of the spatial distribution showed the same trend: a higher proportion of nests in the open area in 1986 than in 1988. Bjorndal and Bolten (1992) hypothesized that the 1986 season's higher rainfall during their study period (July to September), and consequently the higher sand moisture, may have physically facilitated nest digging in the open area relative to the drier years in 1987, 1988, and that the wetter sand may be partially responsible for the high proportion of nests in the open area in the 1986 season. When I compared the relationship between the proportion of nests deposited in the open zone versus the

PAGE 66

59 vegetation/border zone and the mean daily rainfall from July through October with half month intervals for my 1986, 1988 and 1989 data sets, the proportion of nests in the shaded zone was lower as a function of the amount of rainfall, but the relationship was not statistically significant Y=0.899-0.003X, squared=0.12, n=24, p=0.10). Although the spatial distribution between the zones fluctuates seasonally and yearly, the extent of fluctuation was consistently limited to a small range around the middle value: the proportion of nests in the shaded zone was nearly equal or a little higher ( 42-60% in 1986, 51-67% in 1988, 48-71% in 1989) relative to nests in the open area throughout each season. This general trend agreed with the results at the northern section of the Tortuguero beach in 1987 (47%-50%) and 1988 (49-62%) but not in 1986 (28-38%). The constant low proportion (around 30%) of nests in the shaded area throughout the 1986 season at the northern section was unique among all other data sets in both studies. However, four additional surveys over the central part of 18 miles of the 22 mile Tortuguero beach in 1986 season by the author (7 and 23 August, 5 and 17 September) showed similar spatial distributions around the middle range (respectively 55.4, 54.1, 45.7, 49.7% of nests in shaded area), and the proportions of the 18 miles were more similar to the data of the two miles of this study area than to those of the northern section. Thus, although spatial distribution of Tortuguero green turtles varies among both seasons and years, the extent of variation might be rather conservative on a population level.

PAGE 67

60 The area available to nesting turtles in the vegetation/border zone (about 3 m wide) is much smaller relative to the nesting area in the open zone (up to about 30 m wide) at the Tortuguero beach Therefore, the slightly higher or equal distribution of nests in the vegetation/border zone means that the vertical distribution of green turtle nests is biased toward the upper part of the vegetated beach. Similar preference of nest site selection for this species is reported in other regions. Bustard ( 1972) noted that in the Australian cays, green turtles show some tendency to nest close to substantial vegetation. Cornelius (1986) noted that green turtles in Pacific Costa Rica have a tendency to nest beneath the vegetation of the upper beach and rarely in the mid beach. In an island of Ogasawara Islands, Japan, 21 of 28 green turtle nests were deposited near or under the dense vegetation on the upper beach during June (Horikoshi, unpublished data). In Surinam, green turtles deposit more nests (63%) in the border and vegetation area than in the open area, whereas leatherbacks predominantly lay in the open area (87%) (Whitmore and Dutton, 1985) In Florida, green turtles nested in a higher location near the beach dune than did loggerhead turtles (Witherington, 1986). Whitmore and Dutton (1985) suspected that the interspecific difference may be due to interspecific competition: green turtles are pushed up to the upper beach because leatherback nests were deeper and the risk of destruction of nests of green turtles is higher than that of leatherbacks. In Tortuguero, leatherback turtles also use the same beach as green turtles, but their nesting season rarely overlaps with the green turtle's season. However, the nest distribution of leatherbacks is highly skewed to

PAGE 68

61 the open area (Leslie et al., in press). Without co-ocurrence of the two species, the trend of nest site selection of each species at Tortuguero agreed with that at Surinam. Witherington (1986) suspected that the preference of green turtles to nest near dunes may be a strategy to avoid inundation although he could not detect significant differences of emergence success between the two species during the year. Topography and vegetation of green turtle beaches show extreme variation: from low profile naked islands (e.g., French Frigate Shoals) to beaches backed by tropical jungle (e.g., Tortuguero and Sarawak). However, preference for the upper beach might be characteristic of green turtles. It would be interesting to investigate the geographic variation of nest site selection in relation to the thermal profile and egg survivorship factors on each beach. Overall Reproduction Rate To date, very few studies have assessed overall egg survivorship of green turtle nests on natural beaches. Although several studies have investigated mortality factors on natural beaches, their sample seasons were rather limited or the analysis included only those nests that successfully produced some hatchlings (Mortimer, 1981; Balazs, 1980; Schulz, 1975). In Surinam, a detailed quantitative study by Whitmore and Dutton (1985) showed a relatively high hatching success of green turtle nests (mean 80.4% in 1981 and 1982). However, their samples were

PAGE 69

62 collected only from nests above the spring high tide. Mrosovsky (1989) calculated the overall hatching success in Surinam to be 63.5% by adjusting the spatial distribution data. Recently, a long term quantitative investigation has been conducted for green turtles on Florida beaches, although the beaches were more heavily utilized by loggerhead turtles than by green turtles. Mean emergence success of green turtle nests at the Melbourne beach varied from 54.6 to I 75.2% (overall mean=65.6%) from 1985 through 1988 (Redfoot and Ehrhart, 1989). At a nearby beach, Horton (1989) did a similar study and reported 40.1 % for mean emergence success for 20 green turtle nests deposited in 1988 and 1989. Mean emergence success of sample nests of this study was 51.2% (1986), 49.1% (1988) and 66.9% (1989). By adjusting the temporal and spatial distribution of nests for each year, overall hatching success is estimated as 57 .2% (1986), 45.7% (1988) and 66.6% (1989). Overall hatching success at Tortuguero under natural conditions falls within the range reported by the above quantitative studies in other parts of the world. In the 1977 season when dog predation was serious, 42% of 350 monitored nests produced hatchlings and 83% of the eggs of those hatched nests emerged (Fowler, 1979). Thus, the 1977 season's overall hatching success is roughly estimated to be 34.8% (0.42 X 0.83). The success of green turtle nests at Tortuguero was apparently improved by the dog control program even though the natural predator, coatis, still depredate a substantial proportion of nests in some years, such as 1989.

PAGE 70

63 Spatial Effects on Mortality Factors For the three years of this study, no significant difference in emergence success of green turtle nests between the two zones was found. Both abiotic factors (surf, freshwater flooding, hurricane) and biotic factors (predation, turtle digging) affected egg survivorship in Tortuguero turtles, while the extent of damage from each mortality factor varies from year to year. Among the mortality factors, only mammal predation varied significantly between the zones. It is surprising that damage by surf was not very different between the zones; the open zone was 7.1% (egg loss for three year weighted mean), whereas the vegetation/border zone was 4.2%. This is probably because on several occasions heavy erosion cut the beach up to the vegetation area and all nests in the section were damaged, regardless of their zone. Mammalian predation on green turtle nests was greater on the nests in the vegetation/border zone (14.1 %) than on nests in the open zone (2.6%) throughout all three years of this study. This spatial difference between the zones most strongly influenced the overall reproduction rate in the 1989 season when the proportion of depredated nests was the highest among three years. After dogs were removed from the beach, coatis were responsible for most of the predation. Cornelius (1986) noted that coatis are the most common predators on olive ridley nests at Nancite, in the Pacific coast in Costa Rica, and that a large group of coatis living on the nesting beach feed almost exclusively on olive ridley eggs during

PAGE 71

64 much of the year. Although raccoons, a common predator on loggerhead nests in the southeastern United States (e.g., Stancyk et al., 1980), also exist at Tortuguero, their role as a predator on green turtle nests seems to be minimal. However, a spatial bias by predatory raccoons was observed on a Florida beach. Horton (1989) found that raccoon predation on loggerhead nests on a Florida beach was negatively related to the distance of the nests from the dune's edge in one of his two study seasons. He hypothesized that predators search for prey close to the safety of cover, namely near the dune in Florida. At Tortuguero, locals occasionally observed jaguars, Felis onca, in the protected section of the beach during the green turtle nesting season. I occasionally found unidentified foot prints of large cats within the study area. Feral dogs were once abundant throughout the 22-mile Tortuguero beach (Fowler, 1979). Dogs may be a potential threat to coatis, especially young individuals. Spatially biased depredation of nests by coatis may be partially a result of avoiding the risk from their predators. Of the abiotic mortality factors, sporadic freshwater flooding was the most consistent factor at the Tortuguero beach throughout this study, but probably it is very unusual in sea turtle rookeries elsewhere. A single excessive rain event can force the level of ground water near or over the depth of the egg mass of green turtles and suffocate eggs and emerging hatchlings. Even if the eggs are not soaked in water, the nearby water might hinder gas exchange by saturating sand space by capillary action. Similar flooding phenomena were reported on a loggerhead beach on the Georgia coast (Ragotskie, 1959; Kraemer and Bell, 1984). Ragotskie (1959)

PAGE 72

65 suggested that because of the higher probability of excessive rain late in the incubation season (September) on the Georgia coast, successful emergence from nests deposited after the middle of the nesting season (1 July) decreases. Kraemer and Bell (1984) found that the temporal distribution of loggerhead nests did not avoid the excessive rainfall events in September. At Tortuguero the probability of excessive rain events (> 140 mm/day) also increased during the latter part of the incubation season (October and November) (Table 13). However, most nests deposited during the peak nesting period of Tortuguero (August through September) are very much affected if flooding occurs in October and November. Tortuguero green turtles have not adjusted their nesting season relative to excessive rainfall events. Kraemer and Bell (1984) proposed a hypothesis that nesting by loggerhead turtles near the beach dune at the upper beach may be an adaptive response to escape the excessive rains. This is a very tempting hypothesis to explain partially the nesting preference for the upper vegetated beach by Tortuguero green turtles. However, the difference in damage by flooding between the two zones was inconsistent among the study years, and the extent of the differences was rather small. Seasonal Fluctuation of Emergence Success It is not clear to what extent emergence success of green turtle nests significantly fluctuates within a season at Tortuguero. During the three year study, a sign i ficant difference was found only in 1988. The main cause of those fluctuations was the two

PAGE 73

66 inundation events in close succession in October from excessive rain and Hurricane Joan. However, hurricanes on the Caribbean coast of Costa Rica are relatively rare. Coen (1983) noted that Hurricane Martha, 21-25 November 1969, was the only one to hit the coast during the hundred years of records. Therefore, it would appear that the influence of hurricanes on the Tortuguero green turtle colony would be negligible, and that the 1988 season was an atypical year. The extent of damage by Hurricane Joan was not accurately assessed because of the overlapping damage by the excessive rain on 6 October Therefore it is unknown whether the seasonal fluctuation in emergence success in the 1988 season would have occurred without Hurricane Joan. Separate flooding events in August 1986 and in October 1989 affected the sample nests, but these events did not result in drastic seasonal changes in emergence success within each season This lack of seasonal variation can be partially explained by the observation that all stages of embryos and emerging hatchlings are susceptible to flooding. One flooding event can affect nests deposited over long periods, possibly up to four months, if the flooding occurs in the middle of the incubation season. The same principle can apply to the damage by surf, and even for predation, because mammal predators destroyed the nests at various stages. Thus, favorable weather conditions in September (e g., calm seas, less chance of excessive rain) can have a posit i ve effect on emergence success of nests deposited over the long period of time from July through September, which covers the majority of the nesting season In 1977, Fowler (1979) analyzed emergence

PAGE 74

67 success of the nests that produced hatchlings, and she found no seasonal differences in the emergence success at Tortuguero. Fowler did not report any flooding events. Seasonal fluctuation in egg survivorship of Tortuguero green turtle nests might not be apparent, except in years of hurricanes and excessive seasonal flooding. Spatial Effects on Sex Ratio At the Tortuguero beach, a distinct thermal difference at the depth of green turtle nests between the open zone and the vegetation/border zone was very consistent throughout the incubation season in the three years of this study. However, the effect of these thermal zones on sex ratio depends on the relation of sand temperatures in each zone to the pivotal temperature. It appears that this spatial effect on sex ratio can be detected only during a period of dry weather. Seasonal change in this spatial effect was clearly observed in the 1988 season. The mechanism may be the following. During prolonged dry weather, sand temperatures in the open zone clearly exceed the pivotal temperature, whereas sand temperatures in the vegetation/border only approach or slightly exceed the pivotal temperature. Consequently, most nests in the open zone show female-biased sex ratios, whereas nearly equal or moderately female-biased sex ratios occur in the vegetation/border zone. On the other hand, during prolonged wet weather, sand temperatures in both zones decrease below the pivotal temperature. As a result, the difference in sex ratios between zones becomes

PAGE 75

68 negligible, and most nests in both zones show male-biased sex ratios. The latter case was observed throughout the 1986 season. In 1980, Spotila et al. ( 1987) found a distinct spatial difference in sex ratios of Tortuguero green turtle nests between the open zone (mean 67.4% of females, n=9) and the vegetation zone (7.6% of females, n=6) The rainfall record revealed that the weather was very dry (281 mm in August, 313 mm in September) throughout the critical periods of their samples. In Surinam, the different topographic beach zones showed small thermal differences (0.5-1.0 C), and the location had little effect on sex ratio of green turtles through the season (Mrosovsky et al., 1984). This is probably due to the fact that the vegetation is composed of sparse and low shrubs in Surinam, compared to Tortuguero's dense vegetation (Schulz, 1975; P. Dutton, pers. comm.). The existence of dense vegetation and associated shade is one requirement for a spatial effect on sex ratios. The barrier island beaches on the east coast of Florida have primary dunes at the upper part of the beach, but lack dense vegetation. Witherington (1986) noted that the thermal differences on the beach must be slight because there is no significant difference in incubation periods between beach zones for loggerhead nests The nesting beaches of Ascension Island are virtually without vegetation (Mortimer, 1981), but differences in the mineral composition of the various beachs (Mortimer, 1990) produce different thermal regimes (Hays and Mortimer, MS in prep.) At some island rookeries, a significant effect of the orientation of the coast on sex ratio has been reported. At Heron

PAGE 76

69 Island, Australia, Limpus et al. (1983) observed that the north facing beach on the island was warmer than the beach that faces south. Consequently, the northern beach produced a significantly higher proportion of green turtle females (63.1 % females) than the cooler southern beach (29.5% females). In Barbados, West Indies, the warmer west coast beach produced more hawksbill females (80.6% females), whereas the cooler south coast produced more males (32.7% females) (Horrocks and Scott, 1991 ). Temporal Effects on Sex Ratio At Tortuguero, a short dry season in September (mean 320 mm) usually occurs between the wetter periods of July through August, and October through November. For the Tortuguero green turtles, the impact of this dry season seems to be very important because a large proportion of the nests are deposited during August and the most critical period for these nests occurs in September. A dry September produces enough females to avoid a highly skewed sex ratio for the entire season. Seasonal fluctuation in sex ratios resulting from the shift between rainy and dry periods was observed at two tropical beaches: Surinam for green turtles and leatherback turtles (Mrosovsky et al., 1984) and French Guiana for leatherback turtles (Rimblot-Baly et al., 1987). In Sarawak, another tropical beach, incubation periods of green turtle nests varied with the rainy dry weather cycle (Hendrickson, 1958). Therefore, Standora and Spotila ( 1985) suspected that the sex ratios of green turtles at Sarawak also show

PAGE 77

70 seasonal fluctuation. However, the pattern of fluctuation is very different from rookery to rookery. At Tortuguero and Sarawak, the dry season occurs somewhere in the middle of the nesting season, whereas the wet season occurs near the middle of the nesting season in Surinam. In French Guiana, the wet season occurs early in the nesting season. Regardless of the pattern, these sea turtle rookeries probably produce variable sex ratios throughout the season. Annual Variation in Primary Sex Ratio In the 1986 season, the sand temperature in both zones remained below the pivotal temperature for most of the season. In the open zone, although the sand temperature in September and later intermittently rose to the level of the pivotal temperature, these periods were too short to produce many females. Therefore, in spite of the small sample size, the 10.1 % proportion of females from the samples that were sexed directly is probably a good estimate. The weather pattern is one way to assess whether the year was a typical one for Tortuguero. The rainfall in 1986 in August (846 mm) and in September (514 mm) was the highest on record, and the total amount of rainfall throughout the incubation season (3871 mm, from 16 June through 1 0 December) was the second highest for the last 12 years. Thus, it appears that the weather in the 1986 season was atypically wet. Therefore the highly male-biased sex ratio is probably atypical for the Tortuguero population.

PAGE 78

71 In 1988, the total amount of rainfall throughout the incubation period was very close to the mean value of the last 12 years, but the 1988 season had much higher seasonal fluctuation in rainfall than the average of the last 12 years. Rainfall in August 1988 (256 mm) was the lowest on record whereas rainfall in October (790 mm) was the second highest since 1978. Above all Hurricane Joan greatly influenced the survivorship and sex ratio of the nests in the latter half of the 1988 season. The weather in 1988 was also atypical for the Tortuguero beach. If Hurricane Joan had not occurred in 1988, the proportion of females estimated for the primary sex ratio of the 1988 season (40.6%) probably would have been higher. Predictability of Sex Ratio by Environmental Factors Mrosovsky et al. (1984) cautioned that accurate estimation of the primary sex ratio of sea turtles from a combination of data on the pivotal temperature from laboratory experiments and thermal profiles on the beach is impossible because of the many thermal influences and their possible interactions throughout the season. Some of the critical factors are metabolic heat, exact pivotal levels and critical periods, nest depth, and spatial factors. This study's direct thermal measurement in individual nests is the most accurate measure available to obtain the thermal condition of the nest, and the thermal data included the variation from metabolic heat, nest depth and spatial factors. However, even the thermal data in the nest were not sufficient to predict the sex ratio of Tortuguero green turtles.

PAGE 79

72 The high variability of sex ratio in relation to incubation temperature was not surprising. First, even under constant temperature in the laboratory, high inter-clutch variation in sex ratios was found between two loggerhead clutches from the same Florida beach (Mrosovsky, 1988). Limpus et al. (1985) reported a similar high variation in sex ratio at the middle range of temperature among three loggerhead clutches sampled from Mon Repos, Australia, at constant temperature conditions. However, he did not find such large variation among the loggerhead clutches from Heron Island. Mrosovsky and Pieau (1991) hypothesized that the response of gonad differentiation to the incubation temperature might be different depending on the genotype of the eggs, and that genetic factors can override the temperature effect at the middle temperature range This theory was inferred from the experimental results in a freshwater turtle, Emys orbicularis, in which the expression of H-Y antigen was strongly correlated with the expressed sex of gonads at the middle range of temperature where both sexes were produced (Zaborski et al. 1988). Mean nest temperatures during the middle third of development in 1986 and 1988 at Tortuguero fell within a narrow range (27 .0-30. 7 C) compared to the much wider temperature range over which live hatchlings of this species can be produced (25-33 C) (Miller 1985). Both sexes were produced throughout the observed range at Tortuguero. Thus, the high variation in sex ratio over the observed range can be explained if the mechanism of genetic influence operates in green turtle eggs.

PAGE 80

73 Second, because of seasonal fluctuation in sand temperatures and the general increase in metabolic heat throughout incubation, individual nest temperatures on the beach were rarely constant throughout the critical period. The extent and the pattern of variance differed greatly among nests. Standora et al. (1982) reported spatial variation in sex ratio within natural green turtle nests incubated at the pivotal temperature. Due to metabolic heating, eggs near the center of the clutch produced females, whereas those at the periphery produced males. In the present study, I measured nest temperatures at the center of each clutch but determined the sex of eggs collected randomly throughout the clutch. Therefore, my nest temperature readings may not have accurately represented the incubation temperature of all sampled eggs within a nest. Had I collected the sample eggs from only the center mass of each clutch near the temperature probe, the variation in sex ratio relative to the nest temperatures might have been smaller. Furthermore, in this study the critical period was calculated based on an assumption that the time from pipping to emergence was five days for all sample nests. Hendrickson (1958) indicated that heavy rains, which pack the upper layers of beach sand, might prolong the emerging period of green turtle hatchlings in Sarawak Because Tortuguero had high seasonal fluctuation in rainfall, the emergence period of green turtles might also vary to some extent. Thus the calculated critical period of each nest might not be accurate, and the mean nest temperature calculated for each nest may only roughly represent its thermal environment.

PAGE 81

74 Miller (1985) emphasized that temperature, hydric environment, and gas exchange interact cooperatively to influence embryonic metabolism Standora and Spotila (1985) suggested that while temperature is the primary environmental factor that determines sex in sea turtles, other factors such as osmotic stress, and 02 and CO2 levels, could play a role in sex determination in the temperature range over which both sexes are produced. The impact of those other factors, however has yet to be demonstrated. The influence of hydric conditions on sex determination in painted turtles, Chrysemys picta, was inconsistent even within one population (Gutzke and Paukstis, 1983; Packard et al., 1991 ). Oxygen concentration did not influence the sex ratio of red-eared slider turtles, Trachemys scripta (Etchberger et al. 1991 ). It is highly probable however that any environmental influence and their possible interactions in determining sexes of green turtle hatchlings vary at the inter-clutch level because of the highly heterogeneous environment over time at the Tortuguero beach. The previous study of the relationship between sex ratio and mean incubation temperature of natural nests in the 1980 season at Tortuguero (Spotila et al., 1987) showed less variability than did this study This is probably due to their smaller sample size ( 15 nests) and also their shorter sampling period. Most of their samples were collected within a half month period, whereas my sample period extended over three months in each of two years. It is possible that their mean nest temperature values were better correlated with sex ratio than were mine because most of their

PAGE 82

75 nests were exposed to less environmental fluctuation than were those in my study. The first multiple regression model (pseudo R 2 =0.24) with the significant factors selected (mean nest temperatures, incubation period, rainfall data, and zone) to predict sex ratio was only a slight improvement over the nest temperature model (pseudo R 2 =0.20) because rainfall data, incubation period and zonation are correlated with nest temperature data. That is, most factors only provided redundant information to various degrees. As I discussed above, the genetic factor might also be responsible for the poor fit of the data to the model. It should be noted that another multiple regression model (pseudo R 2 =0.23) using rainfall data, zone and their interaction explained the sex ratios of hatchlings as well as the nest temperature model did. Rainfall fluctuation seemed to more strongly influence the sex ratio of nests in the open zone than in the vegetation/border zone. In the open zone, sex ratios ranged from 100% male to 100% female, whereas in the vegetation/border zone the sex ratios of most nests showed less variation, ranging only from 100% male to slightly female biased. My study suggests that data describing rainfall and zonation could be a rough predictor of sex ratio of green turtle hatchlings at Tortuguero. Furthermore, it might be possible to estimate very roughly the primary sex ratio using the total amount of rainfall through the entire incubation period. To construct a useful model, more data need to be gathered on the correlation between rainfall and sex ratio produced in each zone.

PAGE 83

76 Mrosovsky et al. ( 1984) found as great a variability in the relationship between sex ratios and incubation periods for Surinam green turtles as I did for Tortuguero green turtles. They suggested that if the average of a large number of samples was used, it might be possible to make a reasonable prediction of primary sex ratio from the above model. However, collecting a large sample in each thermal zone in each seasonal segment is very labor intensive. In Tortuguero, where thermal zones have a significant effect on sex ratio, the number of samples needed is doubled, and the reliability of the estimate of primary sex ratio is reduced by adding another variable. In conclusion, due to the high level of variation, the sex ratio of green turtle nests at Tortuguero cannot yet be predicted with accuracy from the nest temperature and other environmental factors, such as sand temperature, nest depth, incubation period, rainfall, and nesting zone. In the future, long term accumulation of data correlating rainfall and sex ratios might be used to roughly estimate primary sex ratio at Tortuguero. Currently, direct sexing of periodic subsamples is the only accurate method to investigate primary sex ratio on a natural beach. Moreover, an advantage of direct sexing is that any variation resulting from genetic influences will be incorporated into the sex ratio derived. Unfortunately, immature sea turtles are not sexually dimorphic. Because they have homomorphic sex chromosomes, reliable sexing of sea turtle hatchlings is only possible by sacrificing turtles and performing a time-consuming histological examination of their gonads (Jackson et al., 1988). Nonsacrificial

PAGE 84

77 methods including laparoscopy and testosterone levels (Wibbels et al., 1989) are useful in sexing larger immature sea turtles, but not hatchlings. Recently, Demas and Wachtel (1989) reported a sex specific satellite DNA, called "Bkm", in green turtles and Kemp's ridleys. Because only a few drops of blood are needed to detect this molecule, it could be used to harmlessly determine the sex of sea turtle hatchlings At the present time, however, this molecular method is probably too expensive and labor-intensive to process the large numbers of hatchlings needed to estimate the primary sex ratio on a beach where many variables affect sex ratio, such as on the Tortuguero beach.

PAGE 85

CHAPTER 5 SUMMARY AND CONCLUSIONS The major nesting activity of green turtles occurred from July through the first half of October at Tortuguero during the three-year study, although the frequency of seasonal distribution shifted among years. Nest density also varied among years. The nesting densities recorded in 1986 (7 .3 nests/m) and in 1988 (6.5 nests/m) from July through November were two of the highest known for this species worldwide. Tortuguero beach was divided into two zones The vegetation/border zone lies within 2 m of the border of dense vegetation, and the open zone lies between the vegetation/border zone and the shore line. Nest distribution between the zones varied significantly yearly and seasonally, and there was no common seasonal trend of fluctuation among the years. The vertical distribution of green turtle nests was biased toward the upper part of the vegetated beach throughout each season. A preference for nest sites on the upper beach may be rather conservative for this species Throughout this study, a distinct thermal difference at the depth of green turtle nests between the two zones was consistent, with nests in the open zone being warmer than those in the vegetation/border zone. The differences in overall mean sand 78

PAGE 86

79 temperature at 60 cm depth were 0.8C in 1986, 1.1 C in 1988, and 1.5C in 1989. This is due to the dense vegetation and associated shade at the upper part of the beach. However, the spatial influence of distribution on sex ratio probably occurs only during a dry period. During the middle part of the 1988 season, under prolonged dry weather conditions, sand temperatures in the open zone clearly exceeded the pivotal temperature, whereas sand temperatures in the vegetation/border zone only slightly exceeded the pivotal temperature. Consequently, most nests in the open zone produced female-biased sex ratios, whereas nearly equal or moderately female-biased sex ratios occurred in the vegetation/border zone. On the other hand, during prolonged wet weather, sand temperatures in both zones decreased below the pivotal temperatures, causing male biased sex ratio in both zones. This was observed in the latter part of the 1988 season and throughout the 1986 season. Tortuguero, with 5400 mm of rain each year, is one of the wettest sea turtle rookeries in the world. A short dry period in September occurs between the wet months of July through August and October through November during the major incubation season of green turtle nests. This dry season seems to be very important in determining the primary sex ratio of the Tortuguero population Because a large proportion of the nests is deposited during August and the most critical period for sex determination in these nests occurs in September, a dry September can produce enough females, particularly in the open zone, to avoid a highly male-biased sex ratio for the season.

PAGE 87

80 Overall hatchling emergence success is estimated as 57.2% in 1986, 45. 7% in 1988, and 66.6% in 1989. No significant difference in emergence success between the two zones was found in any year. A significant seasonal difference in emergence success was found only in 1988, when a rare hurricane hit the Tortuguero beach. Thus, egg survivorship probably does not have a great effect on the primary sex ratio produced on the Tortuguero beach. Both abiotic (surf, freshwater inundation, hurricane) and biotic factors (predation, turtle digging) were responsible for reducing the survivorship of green turtle nests. The relative importance of each mortality factor varied from year to year. Freshwater inundation when ground water levels were raised by excessive rainfall, was the most consistent mortality factor of green turtle nests throughout this study. The probabilities of excessive rain events (> 140 mm/day) were highest in the latter part of the incubation season (October and November). Thus, most nests deposited during the peak nesting period (August through September) are affected if the flooding occurs in October and November. Tortuguero green turtles have not adjusted their nesting season relative to excessive rainfall events. Mammalian predation was the only significant mortality factor that showed a distinct spatial difference between the zones. Mammalian predation, mostly by coatis, was greater on the nests in the vegetation/border zone than on nests in the open zone throughout each year. The extent of predation was most severe in the 1989 season when the nest density was about half that in the 1986 and 1988 seasons.

PAGE 88

81 To calculate the primary sex ratio in the 1988 season, temporal and spatial nest distribution, emergence success, and sex ratio data were combined by half month intervals. The overall proportion of females was estimated to be 40.6% in the 1988 season. During 1986, because no significant seasonal effect was detected, the overall proportion of females was calculated to be 10.1 % from the mean value of sample nests. The relationship between sex ratio and mean nest temperature during the middle third of development, mean sand temperature, incubation period, three sets of rainfall records, depth of clutch, and nesting zone were analyzed by logistic models. The pivotal temperatures were calculated as 29.4 C for mean nest temperature, and 28.5 C for mean sand temperature. The primary environmental factor in determining sex of green turtles--mean nest temperaturehad the best fit in the single regression model. However, the variability of the model was high (pseudo R 2 =0 20). Mean nest temperature, incubation period, rainfall from egg deposition through the middle third of development, and nesting zone were significant parameters in the multiple regression analysis (pseudo R2=0.24 ). Another multiple regression model with mean daily rainfall from the time of egg deposition through the middle third of development nesting zone, and their interaction (pseudo R 2 =0.23) was found to be as good as the mean nest temperature model. This rainfall/zone model might have value as a conservation tool to estimate roughly the sex ratio of green turtle nests. In conclusion, the variability of sex ratio by all of the above models was still very high for predictive

PAGE 89

82 purposes. Genetic influence and heterogenous environmental conditions are probably partially responsible for such high variability. To date, we have little knowledge of sex ratio in sea turtle populations and its dynamics. Because the character of sexual dimorphism, tail elongation, appears just prior to sexual maturity of individuals, only gonadal examination or testosterone level can identify the sex of immature turtles. Thus, previously reported sex ratio data based on external morphology (e.g, Ross, 1984) are probably not reliable. Levels of testosterone have been used to examine the sex ratios of immature green turtles in several feeding grounds (Wibbels et al, 1989; Meylan et al., 1992; Bolten et al., in press) However the definition of each population at the feeding grounds is not known. The primary sex ratio data obtained in this study for Tortuguero is the first reliable information on this subject and also is the first stage of the determination of the sex ratio of the entire population. However, we need to be cautious in assessing the primary sex ratio of the Tortuguero population. The density of nests, and thus the number of hatchlings produced each year, varies widely at Tortuguero. Because the sex ratio shows yearly variation as this study has demonstrated, the primary sex ratio of the Tortuguero green turtle population can only be calculated from the product of the number of hatchlings and the sex ratio each year for many years. To achieve this task, periodic collection of samples for sexing for many years is still essential although the procedure is labor-intensive.

PAGE 90

83 Table 1. Ranges of daily fluctuation of sand and green turtle nest temperatures during 24 hours at Tortuguero, Costa Rica. Mean SE (OC) Sand temperatures on transects Open Zone 60cm depth 0.49 0.03 80cm depth 0.49 0.04 Vegetation/border Zone 60cm depth 0.48 0.03 80cm depth 0.41 0.07 Temperatures at the center of clutches Open Zone 0.46 0.04 Vegetation/border Zone 0.49 0.04 Range (OC) 0.3-0.8 0 2-0.8 0.4-0.6 0.1-0.8 0.2-0 8 0.2-0.8 No. Sample n=20 n=20 n=8 n=8 n=14 n=14 Samples were collected and pooled from four separate days on 13 August and 29 September 1988 and 12 August and 14 October 1989

PAGE 91

84 Table 2. Overall mean sand temperatures (C SE) throughout the season on transects from 1 July through 10 December in 1986, 1988 and 1989 at the Tortuguero beach. Open Zone 60cm Depth 80cm Vegetation/border Zone 60cm Depth 80cm 1986 n=76 27.6.1 26.8.1 1988 n=73 28.8.1 28.5.1 27.7.1 27.4.1 1989 n=76 28.6.1 28.2.1 27 .1 .1 26.9.1

PAGE 92

85 Table 3. Multiple comparison table of the monthly sand temperatures at a depth of 60cm in the open zone and in the vegetation/border zone in 1986, 1988 and 1989 (Fisher's procedure of least significant difference, alpha=0.05). 1986 Open Zone 3 JUL 3 Vegetation/ JUL border Zone 1988 Open Zone 2 JUL 2 Vegetation/ JUL border Zone 1989 Open Zone 3 JUL 2 Vegetation/ JUL border Zone 2 AUG 2 AUG 2 AUG 2 AUG 2 AUG 1 AUG 1 SEP 1 SEP 1 SEP 1 SEP 1 SEP 1 SEP 1 OCT 1 OCT 3 OCT 3 OCT 1 2 OCT 1 OCT 1 NOV 1 NOV 3 NOV 3 NOV 3 NOV 3 NOV Those months that do not share a common number have significantly different mean temperatures The numbers are ranked from highest to lowest monthly mean temperatures

PAGE 93

86 Table 4 Effects of seasonal period and nest location in the open and the vegetation/border zones on egg survivorship at the Tortuguero beach in 1986 1988 and 1989. Analysis is by nonparametric two factor ANOVA with unbalanced factorial design (Zar, 1984) Source ss df H p 1986 Period 1116 91 3 201.19 0.1 0
PAGE 94

87 Table 5. Descriptive fate of green turtle eggs from the sample nests at the Tortuguero beach during the 1986, 1988 and 1989 seasons. 1986 1988 1989 Egg Fate Open Vegetation Open Vegetation Open Vegetation border border border Unhatched eggs Destroyed by mammals 79 19 0 606 296 1920 Destroyed by ghost crabs 0 109 93 0 0 0 Destroyed by termits 0 3 2 0 1 20 Destroyed by ants? 30 6 21 7 67 1 3 Destroyed by female turtles 0 256 124 296 0 0 Destroyed by plant roots 0 0 0 0 0 3 Destroyed by unknown cause 0 0 49 97 0 0 Washed out by surf 445 244 0 0 107 0 Embryonic death by surf 37 111 11 2 0 0 0 Embryonic death by flooding 396 498 118 623 481 221 Embryonic death by Hurricane Joan 0 0 61 482 0 0 Embryonic death by flooding or 0 0 299 143 0 0 Hurricane Joan Embryonic death with no 84 86 174 223 202 140 apparent physical disturbance No apparent development 57 36 224 168 133 92 Rotten intact or ruptured eggs 184 125 294 359 118 133 Death at p i pping 0 0 0 6 2 2 Total dead eggs 1312 1493 1571 3010 1407 2544 Unemerged Hatchlings Hatchlings in egg chamber 12 7 9 18 8 18 Hatchlings died above egg 0 0 0 97 20 15 chamber by flooding Hatchlings died above egg 0 0 0 134 0 0 chamber by Hurricane Joan Hatchlings tangled with rope 0 0 0 0 10 0 above egg chamber Hatchlings depreciated by mammals 0 0 0 0 0 Total hatchlings unemerged 1 2 7 9 249 38 34 Emerged Hatchlings 1591 1301 2319 2456 4445 3638 Total eggs 2915 2801 3899 5715 5890 6216

PAGE 95

88 Table 6 Characteristics of green turtle sample nests for sex ratio analysis The data were collected at Tortuguero Costa Rica, during the 1986 and 1988 seasons. Deposition date Zonea 18 July 1986 V/B 18 July V/B 21 July 0 28 July V /B 14 Aug. 0 21 Aug 0 10 Sept. 0 10 Sept. V/B 16 Sept. 0 25 Sept. 0 28 Sept. 0 3 Oct. 0 3 July 1988 5 July 6 July 9 July 13 July 14 July 15 July 16 July 21 July 23 July 23 July 27 July 28 July 29 July 30 July 3 Aug 3 Aug. 5 Aug. 6 Aug. 7 Aug. 8 Aug V/B 0 0 0 V/B V/B 0 V/B V/B 0 0 0 0 V/B V/B V/B V/B V/B 0 0 0 Sample sizeb 1 6 1 9 2c 20 20 1 7 1 9 20 20 20 1 9 20 1 3 20 1 9 20 20 1 9 1 9 20 1 9 20 14 20 20 1 4 1 9 1 2 19 (1) 1 8 1 5 1 6 18 (1) % Female 68 8 0 0 0 0 47.1 0 0 0 0 5 3 0 15.4 55 0 63.2 10.0 30.0 10 5 73 7 55.0 21 .1 90.0 78.6 100.0 100.0 7.1 10 5 16. 7 89.5 22 2 100 0 25.0 61 1

PAGE 96

Table 6--continued. Date deposited 8 Aug. 1988 11 Aug. 15 Aug. 17 Aug. 17 Aug 26 Aug. 27 Aug. 28 Aug. 30 Aug 30 Aug. 31 Aug 3 Sept. 8 Sept. 9 Sept. 14 Sept. 15 Sept. 19 Sept. 20 Sept. 22 Sept. 24 Sept. 27 Sept. 27 Sept. Zonea 0 0 0 0 0 V/B 0 V/B V/B 0 V/B V/B 0 V/B 0 0 0 0 V/B 0 0 V/B 89 Sample sizeb 20 1 6 1 2 20 20 20 13 (1) 1 7 16 (2) 20 (1) 1 9 1 8 1 8 1 8 1 7 20 20 (1) 1 8 1 7 1 9 1 5 20 % Female 100.0 100.0 66.7 60 0 5.0 55.0 15 4 47.1 56.3 80.0 0 0 44.4 27.8 0 0 0 0 60 0 25 0 33.3 23.5 0.0 0 0 0.0 a: V/B-the vegetation/border zone, O-the open zone b : values in parentheses are the number of turtles in which gonads showed both seminiferous tubles and developed cortex. c : missed emergence; two hatchlings remained in the nest.

PAGE 97

90 Table 7. Effects of seasonal period and nest location in the open and the vegetation/border zones on sex ratios of green turtle hatchlings at the Tortuguero beach in 1988. Analysis was by two-factor ANOVA with unbalanced design. The ratios of females were transformed (p'= 1 (X/(n+ 1 1 )/(n+ 1)) (Zar, 1984). Fmax test showed homogeneity of variances. Source Period Zone Period*Zone Total ss 1. 7385 0.7695 0.8093 3.2685 df 5 1 5 31 MS 0.3477 0.7695 0.1619 0.1054 F 3.30 7.30 1.54 p 0.017 0.011 0.208 ns

PAGE 98

Table 8 Estimation of overall sex ratio of green turtle hatchlings at Tortuguero, Costa Rica, in the 1988 season Period #Nest #Nest %Egg Female% Female% #Hatch ling #Hatch ling #Female #Female Survivorship (x109.1) (x109.1) (x109 1) (x109 1) Open V/B Pooled Open V/B Open V/8 Open V/B July 1-15 294 489 48.9 50 5 18.6 144 239 73 45 July 16-31 725 766 79.7 92.1 23.4 578 610 532 143 Aug 1-15 771 885 57.2 75.5 42.8 441 506 333 217 Aug 16-31 746 1494 16.9 40 1 39.6 126 252 50 100 Sep. 1-15 1182 1566 45.4 29 3 22 2 536 710 157 158 Sep. 16-30 498 645 40.0 14.6 11 .8 199 258 29 30 -------------------------------------------------------------------------------Total hatch lings Overall female ratio Open : open zone. V/B: vegetation/border zone 109.1: mean clutch size in 1988. 2024 2576 117 4 4600 Number of hatch lings = Number of nests x mean clutch size (109 1) x % Egg Survivorship Number of females = Number of hatchlings x % Female 692 1866 40.6% (0 ......

PAGE 99

92 Table 9. Logistic regression models of fluctuation in sex ratio for green turtle hatchlings with single independent variables. Data collected at Tortuguero, Costa Rica, 1986 and 1988. lndependant Variable Mean nest temperature Mean sand temperature* Incubation period Rainfall (R-1) Rainfall (R-11) Rainfall (R-111) Bottom depth of nests All model, df=1, p<0.001 *n=728, else n=924 pseudo R2 0.202 0.172 0.177 0.079 0 114 0.171 0.010 Slope 1 .1709 1 .0636 -0.2630 -0.0954 -0 1290 -0.1912 0 0250 Intercept -34 .4594 -30.3740 15.4679 0.5907 1.0477 1 .8052 -2.5171

PAGE 100

93 Table 10. Logistic regression model of fluctuation in sex ratio for green turtle hatchlings with all independent variables. Final model follows in Table 11 Data collected on Tortuguero, Costa Rica, 1986 and 1988. Parameters Slope (SE) Mean nest temperature 0.5005 (0.1314) Incubation day -0.0801 (0.0330) Rainfall (R-1) -0 0136 (0.0234) Rainfall (R-11) 0.0510 (0.0454) Rainfall (R-111) -0.1359 (0.0585) Bottom depth of nest 0.0097 (0.0090) Zone 0.4701 (0.1958) Intercept -9.9764 (4.8321) -2Loglikelihood=292 3, df=7, overall P<0 001 pseudo R2=0 241 n=924 ChiSqure p 14.50 0 0001 5.94 0.0149 0.34 0 5607 ns 1 26 0.2617 ns 5.40 0.0202 1 .16 0.2815 ns 5.76 0 0164 4.26 0.0390

PAGE 101

94 Table 11 Logistic regression model of fluctuation in sex ratio for green turtle hatchlings with significant variables Data collected at Tortuguero, Costa Rica, 1986 and 1988. Parameters Slope (SE) ChiSqure p Mean nest temperature 0.5230 (0 1250) 17.50 <0.0001 Incubation day -0.0836 (0.0299) 7 83 0.0051 Rainfall (R-111) -0.0906 (0.0224) 16.40 0.0001 Zone 0.5532 (0.1822) 9 21 0.0024 Intercept -9.8173 (4.6810) 4.40 0 0360 -2Loglikelihood=290 1, df=4, overall P<0.0001 pseudo R2=0.239, n=924

PAGE 102

95 Table 12. Logistic regression model in sex ratio for green turtle hatchlings with rainfall data and nesting zone variables. Data collected at Torttuguero, Costa Rica, 1986 and 1988. Parameters Slope (SE) ChiSqure Rainfall (R-111) -0.1104 (0.0225) Zone 2.7738 (0.4011) (R-111) X Zone -0.1510 (0.0310) Intercept 0.2680 (0.2870) -2Loglikelihood=272.2, df=4 overall P<0.0001 pseudo R2=0.225, n=924 24.03 47.82 23 75 0.87 p 0.0001 0 0001 0.0001 0 3503

PAGE 103

96 Table 13. Frequency of excessive rainfall events (> 140mm/day) at Tortuguero, Costa Rica, from 1978-1989 Data from Institute Meteorogico Nacional, Costa Rica. Year Jan Feb Mar Apr May Jun Ju I Aug Sep Oct Nov Dec 1978 n/d n/d n/d n/d 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1 2 2 1 1 1 1 1 1 2 1 1 Frequency of excessive rainfall event per year 1 2 1 1 1 1 1 2 1 1 1 1 n/d 1 1 1 1 .36 0 .09 .09 .17 0 .17 17 .08 .50 .41 .36 n/d: no data available

PAGE 104

97 MILE 0 Figure 1. The two-mile study area at Tortuguero, Costa Rica.

PAGE 105

Border/Veg. Zone '2m 98 Open Sand Zone Figure 2. Beach zones at the study area, Tortuguero Costa Rica. The vegetation/border zone lies within 2 m of the border of dense vegetation (5-100% cover) The open zone lies below the vegetation/border zone ( <5% cover).

PAGE 106

Figure 3. Seasonal fluctuation of monthly minimum and maximum air temperatures from July through November 1986 1988, 1989 at Tortuguero, Costa Rica.

PAGE 107

100 32 1986 30 21 2t 2' MIN 22 20 32 0 1988 0 30 w a: 21 ::> 2t a: w 2, MIN a.. 22 w Ia: 20 21 _J I 2t MIN 0 2, 22 20 32 86, 88, 89 MEAN 30 21 26 2, MIN 22 20 JUL. AUG. SEP. otr. NOV.

PAGE 108

E E .....J .....J <: u. z <: a: .....J <: 0 200 1 9 8 6 100 200 1988 100 200 1 9 8 9 100 101 * DEC. Figure 4. Daily rainfall from 15 June through 10 December 1986 1988, 1989 at Tortuguero, Costa Rica Fresh water flooding occurred at the rain events marked

PAGE 109

(A) (B) 800 600 ...J ...J z <( 400 a: 200 z 0 102 JAN FEB MAR APR MAY. JUN JUL. AUG SEP OCT. NOV. DEC JAN FEB MAR. APR. MAY. JUN JUL. AUG SEP OCT. NOV. DEC Figure 5. Seasonal relationship between rainfall and green turtle nesting. (A) Monthly mean rainfall at Tortuguero, Costa Rica from 1978 through 1989. Months with hatched bars represent the major incubation season of green turtle eggs. The bars show one standard error for each month. (B) Mean nesting frequency among the 1986, 1988, and 1989 seasons. Details for each season are presented in Figure 16.

PAGE 110

....J ....J < LL z < a: >....J I ..... z 0 103 1000 800 600 400 200 C .'d, I/ ,-. ._ ,' j ', ......... I : \ I \ I : \ I : \ I / \ I : I I I J 0 --..-----,------,...----.-------,,..... JUL AUG. SEP. OCT. NOV. MEAN -----.1986 ----c--1988 ---<>----1989 Figure 6. Monthly rainfall in the 1986 1988 and 1989 incubation seasons of green turtle eggs at Tortuguero, Costa Rica. Mean values from 1978 through 1989 are shown for comparison.

PAGE 111

w z LL 0 a: w co ::> z 104 40 50 60 70 80 90 100 110 120 130 BOTTOM OF EGG CHAMBER (CM) Figure 7. Frequency distribution of the bottom depth of green turtle clutches at Tortuguero, Costa Rica.

PAGE 112

Figure 8. Seasonal fluctuation of ground water table and rainfall in the 1986 incubation season of green turtle eggs at Tortuguero, Costa Rica. The line of mean bottom depth of green turtle clutches is 77.9 cm (S0=11.1) from the surface. The shaded area shows the maximum and minimum range of ground water table recorded at two wells in the open zone and at three wells in the vegetation/border zone. Before 4 August in the open zone and before 14 August in the vegetation/border zone, only one well was measured in each zone

PAGE 113

106 200-.-----------------------, SUFFACE ....... ---------------------1986/BORDERIVEGETATION ZONE 40 1 SD MEAN CEPTH OF 80 t-------------------------.--1 CLUTCH BOTTOM 1 so SURCACE ....... ---------------------1986/OPEN ZONE 40 1 SD t--------1-------------------1 MEANCEPTHOF 8 0 CLUTCH BOTTOM 1 SD

PAGE 114

Figure 9. Seasonal fluctuation of ground water table and rainfall in the 1989 incubation season of green turtle eggs at Tortuguero, Costa Rica. The line of mean bottom depth of green turtle clutches is 77 .9 cm (S0=11.1) from the surface. The shaded area shows the maximum and minimum range of ground water table recorded at two wells in the open zone and at three wells in the vegetation/border zone.

PAGE 115

108 200...------------------------, ....J ....J <( LL Z 100 < er ....J < 0 1989 SUFFJICE ---------------------1989/BORDER/VEGETATION ZONE 40 1 SD ----------------------MEANDEPTHCF 8 0 CLUTCH BOTTOM 1 SD 120 BOTTOM CF ~----lll.-.l...Jo.,1,..1...K..;Jo _________ ..,ui~i....u::..----l~.k:I Vv9.1. JUL. ALG. SEP ccr ~~ SlffACE ---------------------1989/OPEN ZONE 40 1 SD 1----------------------~ MEAN DEPTH OF 8 0 CLUTCH BOTTOM 1 SD 120

PAGE 116

Figure 10. Frequency distribution of wave height in the 1986, 1988, and 1989 incubation seasons of green turtle eggs at Tortuguero Costa Rica.

PAGE 117

110 WAVE HEIGHl"> 1 Sm 1986 CJ 1 Sm> WAVE HEIGHT > 0 5m D WAVE HEI~ 0 Sm % 100 80 80 40 20 0 1988 JUL AUG SEP OCT '' % 100 80 80 40 20 0 1989 JU L AUG SEP OCT '' % 100 80 80 40 20 0 JUL AUG SEP OCT '' MEAN (86, 88 AND 89) % 100 80 80 40 20 0 JUI AUG SEP OCT ''

PAGE 118

111 20 0 Y = 17.147 1.656X R = 0.663 I 18 n = 15, t = 3 189 p = 0.007 0 16 :? < z 14 Cl) >:? 12 < I!') Cl 0 10 :? VI .....J < ..... 8 0 I LL (!) 0 LU 6 a: I LU LU 4 co > :? < 2 ::, 3: Z0 0 200 400 600 800 1000 MONTHLY RAINFALL (MM) Figure 11. Relationship between the number of calm days in a month and monthly rainfall from July through November in 1986, 1988, and 1989 at Tortuguero, Costa Rica.

PAGE 119

w a: ::, ..... < a: w a.. w ..... 0 z z u) 6 w a: ::, ..... < a: w a.. w ..... ..... CJ) w z z u) 112 29 0 28 5 t A, ... 28.0 27 5 -i)-OPEN, 60cm OPEN, 8 0 cm -VEGETATION.13ORDER 60cm VEGETATION.13ORDER 8 0 cm 27 0 ..._----~-~----------2 4 6 8 10 12 14 16 18 20 22 24 TIME(HOURS) 29.5 ts a'EN VEGETATIONIBORDER 29.0 28 5 28.0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 TIME (HOURS) Figure 12 Daily fluctuation of green turtle nest temperatures and sand temperatures at the Tortuguero beach, Costa Rica. Data were pooled from four separate days; 13 August 1988, 29 September 1988, 12 August 1989 and 14 October 1989. Each arrow shows the time representing the 24-hour mean value.

PAGE 120

Figure 13 Seasonal fluctuation of sand temperatures at a depth of 60 cm and 80 cm in the open zone and in the vegetation/border zone in 1986, 1988, and 1989 at the Tortuguero beach Costa Rica

PAGE 121

32 1986 31 30 29 28 27 26 25 24 0 JUL. 0 w 32 a: 1988 :J 31 I<( 30 a: w 29 a.. 28 w I27 0 z 26 <( Cl) 25 >_J 24 <( JUL. 0 32 1989 31 30 29 28 27 26 25 24 JUL. 114 AUG. SEP. AUG. SEP. AUG. SEP ;l OCT. NOV. DEC. OCT. NOV. DEC OCT. NOV. DEC OPEN ZONE 60 CM OPEN ZONE 80 CM VEGETATION/BORDER ZONE 60CM VEGETATION/BORDER ZO NE BO CM

PAGE 122

115 30 OPEN ZONE, 60CM DEPTH --0-1986 ---r1988 6 2..29 1989 LU a: :::> LU 28 CL ::? 0 z <( 27 Cl) 26 ..._---~----------......-JUL AUG. SEP. OCT OOV 30 VEGETATION/BORDER ZONE 60CMDEPTH ---.-1986 6 -.1988 2..29 LU ----1989 a: a: 28 LU CL ::? LU .... 0 z 27 <( en JUL AUG. SEP OCT. OOV. Figure 14. Seasonal fluctuation of monthly sand temperatures at a depth of 60 cm in 1986, 1988, and 1989 at the Tortuguero beach, Costa Rica.

PAGE 123

116 0 OPEN 2.0'JE e VEGETATION/BORDER n 30 0 w a: => i w 29 0 a.. ~"o' w () t-0~ z () ~o Cl) (0 II g~ ~w 28 0 27 0 26 0 25 0-+---------------------------0 200 400 600 800 1000 MONTHLY RAINFALL (MM) Figure 15. Relationship between mean monthly sand temperature at a depth of 60 cm and monthly rainfall from July through November in 1986, 1988 and 1989 at Tortuguero, Costa Rica. Regression analysis (open zone, Y=30 0199-0.0031X, R=0 718, n=15, t=3.718, p=0.003; vegetation/border zone, Y=28.5249-0.0024X, R=0.665, n=15, t=3.210 p=0.007).

PAGE 124

Figure 16. Seasonal distribution of green turtle nests in 1986, 1988 and 1989 in the two-mile study area at Tortuguero Costa Rica Data in the latter half of June 1986 were not available.

PAGE 125

w a: >0 :::> fCl) w _J I 0 w I fz 0 w fCl) 0 a.. w 0 Cl) w I 0 f::> _J 0 LL. 0 0 z 118 3000 1986 2000 1000 ? 0 JUN. JUL. AUG sEP. OCT. NOV. 16 30 1 15 16 31 1 15 16-31 1 15 16 30 1 15 16 31 1 15 16-30 3000 1988 2000 1000 0 JUN. JUL. AUG. SEP. OCT NOV. 16 30 1 15 16 31 1 15 16-31 1 15 16 30 1 15 16-31 1-15 16-30 3000 1989 2000 1000 0 -1==""""'=~~ JUN. JUL. AUG. SEP OCT NOV 16 30 1 15 16 31 1-15 16 31 1-15 1 6-30 1-15 16 3 1 1 -1 5 16 30 PERIOD (HALF MONTH)

PAGE 126

Figure 17. Yearly and seasonal fluctuation of green turtle nest distribution between the open zone and the vegetation/border zone in 1986, 1988, and 1989 at Tortuguero Costa Rica W i thin each season, Tukey-type multiple comparison test for proportional data (Zar, 1984) ranked from the highest to the lowest the proportion of nests deposited in the vegetation/border zone. Those months that do not share a common number have significantly different proportions.

PAGE 127

120 %100 2,3 1,2 3 4 3 4 3 4 5 4 5 1986 50 .R.fr,j t&30 Jll. HS Jll. 16 31 AUG t-15 AUO 1&31 SEP 1-15 SEP 16'30 OCT MS OCT 16-31 TOTAL %100 1,2 2 3 5 4,5 2,3,4 3,4,5 2 ,3,4,5 N S 1988 50 0 Jl.1,,118-30 JU. 1-15 JU.. 18 31 AUG 1 15 AUG 18-3 1 SEP 1 15 SEP 18-30 OCT 1 15 OCT 18-31 TOTAL %100 1,2 1 2 3 2 1 2 1, 2 2 2,3 1989 50 N 0 JUii 1630 JU.. HS Jll. 16 31 Al.JO 1 15 AUG 16-31 SEP 1 1 5 SEP 16-30 OCT 1 1 5 OCT 1&31 TOTAL PERIOD (HALF MONTH)

PAGE 128

121 40 1986 a VEGETATION/BORDER ZONE N=24 El OPEN Z ONE N=25 30 20 10 w z 0 0 0 10 20 30 4 0 50 60 7 0 80 90 1 00 N I 40 1988 l!I VEGETATION/BORDER Z O NE. N=51 () l3 OPEN ZONE N:37 <( w 30 z Cf) I20 Cf) w z w 10 .....J 0. <( 0 Cf) 0 10 2 0 30 4 0 50 60 7 0 80 90 100 .....J <( 60 I1989 w VEGETATION/B O RDER ZONE, N=60 f2 !::'.] OPEN ZONE, N=53 50 0 40 30 20 10 0 0 10 20 30 40 50 60 7 0 80 90 1 00 EMERGENCESUCCESS(o/~ Figure 18. Percentage distribution of green turtle sample nests by the emergence success in 1986, 1988, and 1989 at Tortuguero, Costa Rica.

PAGE 129

Figure 19. Factors causing total destruction of green turtle sample nests in 1986, 1988, and 1989 at Tortuguero, Costa Rica.

PAGE 130

123 OPENZO'JE 1986 76 0% 2 7 % 2 7 % 1988 78 4% 1989 90 5% II KILLED BY SURF 0 KILLED BY FLOODING El KILLED BY FLOODING OR HURRICANE VEGErATlGJ BORDERZGJE 75 0% 1. 6" ,4 (:) DEPREDA TED BY MAMMALS 0 DEPREDA TED BY GHOST CRABS 0 EXCAVATED BY TURTLES 0 NEST PRODUCED SOME HA TCHUNGS

PAGE 131

0 C/) C/) w 0 0 :::> C/) w 0 z w (!) a: w w z <( w 0 [Il 124 1 oo OPEN ZONE 5 6 80 60 9 40 20 0 ..,__-......------.----......------,-100 VEGETATION/BORDER ZONE 5 8 0 60 9 4 0 2 0 0 ....____,,,_-........ ---......------.-100 TOTAL 11 8 0 60 18 40 20 o ..,__ ____________ 16-31 JUL 1-15 AUG. 16-31 AUG 1-15 SEP PERIOD OF SAMPLE NESTS DEPOSITED Figure 20. Seasonal fluctuation of emergence success of green turtle sample nests in 1986 at Tortuguero, Costa Rica. Monthly mean standard error with the sample size are shown.

PAGE 132

g__ Cl) Cl) w (.) (.) :::::> Cl) w (.) z w C) a: w w z 0 CJ 125 100 OPENZQ\JE 8 4 80 60 40 20 o..,__---.---....----~----r------,----,--100 VEGETATION/BORDER ZONE 8 80 60 40 20 7 9 o..,__---...------.-------.----...-------r------100 TOTAL 16 80 60 40 20 0 ......_---,---....--------,-----.------.-----,,--1-15 JUL. 16 31 JUL 1-15 AUG 16 31 AUG 1-15 SEP 16 30 SEP PERIOD OF SArv1PLE NESTS DEPOSITED Figure 21. Seasonal fluctuation of emergence success of green turtle sample nests in 1988 at Tortuguero, Costa Rica. Monthly mean standard error with the sample size are shown.

PAGE 133

126 OPENZONE 100 8 9 9 80 60 40 20 -;!!_ 0 0 en en w VEGETATION/BORDER ZONE () () 100 8 ::::> 1 0 en w 80 () z w 60 (!) a: w 40 w z 20 < w 0 >_J I 100 TOTAL !z 0 80 co 60 40 20 0 16 -31 JUL 1-15 AUG 16 -31 AUG 1-15 SEP 16 30 SEP 1-15 OCT PERIOD OF SAMPLE NESTS DEPOSITED Figure 22. Seasonal fluctuation of emergence success of green turtle sample nests in 1989 at Tortuguero, Costa Rica Monthly mean standard error with the sample size are shown.

PAGE 134

Figure 23 Fate of green tu r tle eggs from the sample nests in 1986, 1988, and 1989 at Tortuguero Costa Rica.

PAGE 135

128 OPEN ZONE 1986 1988 1989 KILLED BY SURF m KILLED B Y FNI FLOODING ra KILLED B Y FNI FLOODING & HURR I CANE PRED A TION B Y MA M MALS PREDATION BY CRABS. OTHERS Ii; 0 VEGETATION BORDER ZONE DES"TROYED B Y TURTlES NO APPARENT DEVELOPMENT CAUSE OF E"1BRYONIC DEATH UNKNOWN 011-ER HATCHUNG EMERGED

PAGE 136

Figure 24. Fate of green turtle eggs from the sample nests in 1986, 1988, and 1989 at Tortuguero, Costa Rica The category proportions were calculated as weighted means among the three years.

PAGE 137

130 OPEN ZONE 2.6 % 1 7 % 1 1 % 3.3 % 3 6 % VEGETATION/BORDER ZONE 4 2 % DAMAG E D BY WAVE ACTION Ill DAMAGED BY FLOODING/HURRICANE JOAN E] PREDATION BY MAMMALS 0 PREDATION BY CRABS OTHERS [] DESTROYED BY TURTlES 8 NO APPARENT DEVELOPMENT CAUSE OF E MBRIONIC DEATH UNKNOWN OTHER 0 HATCHLl'JG EMERGED

PAGE 138

0 0 w a: a.. w 0 Cf) 8 6 w Zcn u. _j 4 0 <( a: w~ CD <( 2 ::> z CD 131 0-9 10-19 20-29 30-39 40-49 50-59 60-69 72 DAYS AFTER DEPOSITION OF EGGS Figure 25. Timing of mammalian predation on sample green turtle nests at Tortuguero, Costa Rica.

PAGE 139

w z LL 0 'if!. 80 60 40 20 N=233 1986 132 N=330 N=456 1988 1989 Figure 26. Relative proportion of green turtle nests depredated by mammals that occurred in the open zone and in the vegetation/border zone. Data were gathered during the two mile beach censuses at Tortuguero, Costa Rica; 48 censuses were made in 1986, 56 in 1988, and 80 in 1989.

PAGE 140

133 12 1986 10 8 Cl) 6 :::, Cl) z w 4 () w ....J 2 0 0 a: 12 w 1988 Q.. 10 co 0 4 w 0 w 2 a: Q.. 0 w 0 Cl) 12 I1989 Cl) w 10 z LL 0 8 a: w co 6 :::, z 4 2 0 JUN JUL JUL AUG AUG. SEP SEP OCT. OCT NOV. NOV. 16-31 1-15 16 31 1-15 16 31 1-15 16 30 1 15 16-31 1-15 16 30 Figure 27. Temporal distribution of mammalian predation on green turtle nests at Tortuguero, Costa Rica.

PAGE 141

134 16 1986 14 12 Cf) 10 => 8 Cf) z w 6 () w 4 ...J 2 0 0 a: w 16 a.. 1988 Cf) w 14 ....J I12 a: => I10 >co 8 0 w 6 4 <{ () 2 Cf) 0 ICf) w 16 z 1989 u.. 0 14 a: w 12 co 10 => z 8 6 4 2 0 JUN JUL JUL. AUG AUG SEP. SEP. OCT. OCT. NOV. NOV 16-31 1 -15 16-31 1-15 16-31 1 -1 5 16-30 1-15 16-31 1 -1 5 16-30 Figure 28. Temporal distribution of excavation of green turtle nests by female green turtles at Tortuguero, Costa Rica.

PAGE 142

fD 0 w Ig Cl) Cl) w ::> z Cl) u.. lri (.) w a: CD W a.. ::::, Cl) zw ~g 135 20 .......-------------------, 15 10 5 Y = 1 0564 + 4 214 X, R"2 = 0.757 n = 31, t = 9.50 p < 0 0001 0 0 0 0 0 00 0 0 il[[~---Ql;--c>--T--r---r---,----,~-T----1 0 1000 2000 3000 NUMBER OF NESTS DEPOSITED IN TWO-MILE SECTION OF BEACH IN A HALF-MONTH PERIOD 4000 Figure 29. Relationship between the number of green turtle nests excavated by female green turtles per census and the number of green turtle nests deposited within the two-mile study area of the Tortuguero beach during a half-month period. Data were collected in the 1986, 1988, and 1989 incubation seasons of green turtle eggs.

PAGE 143

Figure 30. Sex ratio of green turtle hatchlings as a function of mean incubation temperatures during the middle third of the development period. Data were collected at Tortuguero, Costa Rica during the 1986 and 1988 seasons. Nest temperatures were taken at the approximate center of each clutch, and sand temperatures were taken 1 m apart from each clutch at the same depth. The superimposed line shows the fit to a logistic regression model. In Figure 29A, Y = exp(-34.45 + 1.17X)/(1+ exp(-34.45 + 1.17X)), pseudo R 2 = 0.20. In Figure 29B, Y = exp(-30.37 + 1 06X)/(1 + exp(-30.37 + 1.06X)), pseudo R 2 = 0 17. The small number indicates the number of overplotted data points in the graph.

PAGE 144

_J al <( al 0 er: Cl. w _J <( w LL ._ _J al <( al 0 er: Cl. w _J <( w LL 1.0 0 8 0 6 0.4 0.2 0.0 1.0 0 8 0 6 0 4 0.2 0 0 A 137 0 0 0 0 0 0 0 0 0 0 0 00 0 0 ---------------------------------------------------------26 B 0 0 0 0 0 0 0 0 0 0 O 0 00 0 0 0 0 0 0 00CX> 0 C> 0 C> 0 27 28 29 30 MEAN NEST TEMPERATURE (C) 00 0 0 0 0 0 3 00 0 0 31 ----------------------------------------------------------0 0 0 0 0 0 0 0 0 0 0 0 CDQ)O 0 26 27 28 29 30 31 MEAN SAND TEMPERATURE ( C)

PAGE 145

Figure 31. Seasonal fluctuation of sex rat i o of green turtle hatchlings in 1988 at Tortuguero, Costa Rica Shown is the bimonthly mean one standard error ; the number above each bar indicates sample size For nests in the open zone the Roman numerals represent a rank from highest to lowest female ratio analyzed by Fisher's procedure of least significant difference (alpha=0.05). Those bimonthly periods that do not share a common figure have significantly different mean sex ratios For nests in the vegetation / border zone, no significant difference was detected within the season.

PAGE 146

0 w w LL 100 80 60 40 20 OPEN ZONE 4 (11 111) (I) 139 6 4 3 ( Ill) (1 11) (Ill) (Ill) 0 ...__----..----~------.----..----.....-100 VEGETATION/BORDER ZONE 80 3 60 4 2 40 20 0 ------.----.-----.--------+-----+-100 TOTAL 80 9 8 60 40 20 o.....__-------.....---~------..--JUL 1 15 JUL 16 31 AUG 1 -1 5 AUG 16 31 SEP 1 1 5 PERIOD OF NESTS DEPOSITED SEP 1988 16 30

PAGE 147

0 <( w LL 140 100 ...-----------------80 60 40 3/3 16 JUL. -15 AUG. 16AUG -15 SEP. 16 SEP. -15 OCT. PERIOD OF SAMPLE NESTS DEPOSITED Figure 32. Seasonal fluctuation of sex ratio of green turtle hatchlings in 1986 at Tortuguero, Costa Rica. Shown is the monthly mean one standard error. The number on the left side above SE bar shows the number of sample nests deposited in the open zone and the number on the right side shows ones in the vegetation/border zone.

PAGE 148

Figure 33. Logistic regression model and its residuals to explain sex ratio fluctuation of green turtle hatchlings as a function of mean nest temperature during the middle third of the development period Data were collected at Tortuguero, Costa Rica, in the 1986 and 1988 seasons. The superimposed line shows a logistic regression line to fit: Y = exp(-34.45 + 1.17X)/(1 + exp(-34.45 + 1 .17X)) pseudo R 2 = 0.20. Closed circles represent the nests deposited in the open zone, open circles represent ones in the vegetation/border zone.

PAGE 149

142 1 0 0 00 00 0 :::i 0 8 0 aJ 0 <( 0 0 aJ 0.6 0 0 0 a: 0 a. 0.4 w 0 ....J 0 0 <( 0 0.2 o w LL 0 0 0 0 oeo 26.5 27.5 28.5 29 5 30 5 MEAN NEST TEMPERATURE (C 0 ) 0 6 0 4 + + + + + + + + + + + 0 2 + + + ++ ....J + ++ <( + + + + + :::, 0 0 0 U) +++ + + w ..... + + + + a: -0 2 + + + + + .... + + + + -0.4 + -0.6 0 0 0 2 0 4 0 6 0 8 LO PPREDICTED PROBABILITY OF FEMALE

PAGE 150

143 0 0 0 30.5 0 w a: :::, 29 5 a: w Cl.. w 28 5 Cf) w z z 27 5 <{ w 26 5 5 0 55 60 65 7 0 7 5 INCUBATION PERIOD (DAYS) Figure 34. Relationship between mean nest temperatures during the middle third of development and incubation duration of green turtle nests The superimposed line shows a linear regression l i ne : Y = 38 3 0 15X R 2 = 0.49 n= 52 p<0.001

PAGE 151

144 3 1.0 0 0 0 0 0 ....J 0 8 0 0 co 0 0 co 0 6 0 a: 0. 0. 0.4 w ....J 0.2 0 w LL 0 0 0 0 0. 0000 2 2 2 2 50 55 60 65 70 75 INCUBATION PERIOD (DAYS) Figure 35. Sex ratio of green turtle hatchlings as a function of incubation duration. Data were collected at Tortuguero, Costa Rica in the 1986 and 1988 seasons. The superimposed line shows the logistic regression model to fit: Y=exp(15.47 0.26X)/(1 + exp(15.47 -0.26X)), pseudo R2=0.18. Closed circles represent the nests deposited in the open zone, open circles represent ones in the vegetation/border zone. The small numbers indicate the numbers of overplotted data points in the graph.

PAGE 152

Figure 36 Sex ratio of green turtle hatch lings as a function of amount of rainfall. Data were collected at Tortuguero Costa Rica, in the 1986 and 1988 seasons In Figure 36A the X axis represents mean daily rainfall during the middle third of development. In Figure 368, the X axis shows mean daily rainfall during the m i ddle third of development plus the previous 1 O days In Figure 36C the X ax i s represents m ea n daily rainfall from the deposition of the clutch through the middle third of development. The super i mposed line shows logistic regression models to fit all plots In Figure 36A Y = exp(0 59-0 1 OX)/(1 + exp(0.59-0 1 OX)) pseudo R 2 = 0 08 In Figure 368, Y = exp(1 .05 0 13X)/(1 + exp(1 .05-0 13X)) pseudo R 2 = 0 11 In Figure 36C Y = exp(1 81-0.19X)/(1 + exp(1 81-0 19X)) pseudo R 2 = 0 17. Closed circ l es represent the nests deposited i n the open zone open circles represent ones in the vegetation / border zone. The small numbers indicate the numbers of overplotted data points i n the graph.

PAGE 153

146 A 1.0 0 0 CD 0 0 0 .8 0 0 0 0 0.6 0 0 0 0 0 0 4 0 0.2 0 0 0 0.0 0 CD 0 o 0 0 1 0 2 0 30 8 1.0 0 .....J 0 co 0.8 0 0
PAGE 154

147 1.0 0 00 0 0 0.8 :J 0 0 0 co c:x: 0.6 co 0 0 Cbo 0 o a: a. w 0.40 _J c:x: 0.2 w LL 0 0 :. 0 0 .. 0 0 0.0o ec.o ., 0 0 0 2 2 I I I I I 50 60 70 80 90 100 11 0 BOTTOM DEPTH OF CLUTCH (CM) Figure 37. Sex ratio of green turtle hatchlings as a function of the bottom depth of clutches. Data were collected at Tortuguero, Costa Rica, in the 1986 and 1988 seasons. Closed circles represent the nests deposited in the open zone, open circles represent ones in the vegetation/border zone. The small numbers indicate the numbers of overplotted data points in the graph.

PAGE 155

148 0 7 0.6 0.5 0.4 + 0.3 ++ + + + + + ++ + .....J 0 2 + ++ ++ <( + + + + + ::> 0 .1 0 + 0 0 (/) -0.1 +..., w + + a: '-+ ..... + -0 2 + + + + -0.3 + -0.4 + + + -0.5 -0 6 + -0.7 0.0 0 2 0.4 0 6 0 8 1. 0 PREDICTED PROBABILITY OF FEMALE Figure 38. Residuals of the selected logistic model that explains sex ratio fluctuation of green turtle hatchlings Data were collected at Tortuguero, Costa Rica, in the 1986 and 1988 seasons. Y=exp(-9.81 + 0.52X1-0.08X2-0.09X3+0.55X4)/(1+ exp(9.81 + 0.52X1-0.08X2-0.09X3+0.55X4)). pseudo R 2 =0.24. X1 : mean nest temperatures during the middle third of development. X2 : incubation duration. X3 : mean daily rainfall from the deposition of the clutch through the middle third of development. X4 : zone of clutch (1 the open zone, O the vegetation/border zone).

PAGE 156

LITERATURE CITED Aldrich J.H and F.D Nelson. 1984. Linear Probability Logit and Probit Models. Sage Publications Beverly Hills CA Alvarado, J and A. Figueroa. 1989. The ecological recovery of sea turtles of Michoacan, Mexico. Special attention: the black turtle Chelonia agassizi. U.S.F.W.S. Endangered Species Report. Balazs G.H. 1980. Synopsis of biological data on the green turtle in the Hawaiian Islands NOAA Technical Memorandum NMFS. NOAA-TM NMFS-SWFC-7 Balazs G.H. and E. Ross. 1974. Observations on the preemergence behavior of the green turtle. Copeia 1974:986-988 Bhunya, S P and P Mohanty-Hejmadi. 1986. Somatic chromosome study of a sea turtle, Lepidochelys olivacea (Chelonia, Reptilia). Chrom. Inf Serv. 40:12-14 Bickham J W. 1981 Two hundred million year old chromosomes: deceleration of the rate of karyotypic evolution in turtles Science 212:1291-1293. Bickham, J.W. K.A. Bjorndal M.W. Haiduk, and W.E. Rainey. 1980 The karyotype and chromosomal banding patterns of the green turtle (Chelonia mydas ). Copeia 1980:540 543 Bjorndal K.A and A.B. Bolten. 1992 Spatial distribution of green turtle (Chelonia mydas) nests at Tortuguero, Costa Rica. Copeia 1992:45-53. 149

PAGE 157

150 Bjorndal, K.A., A. Carr., A.B. Meylan, and J A. Mortimer. 1985. Reproductive biology of the hawksbill (E retmochelys imbricata) at Tortuguero, Costa Rica, with notes on the ecology of the species in the Caribbean. Biol. Conserv. 34 : 353-368. Bolten, A.B., K.A. Bjorndal J.S. Grumbles, and D.W. Owens. In Press. Sex ratio and sex-specific growth rates in immature green turtles, Chelonia mydas, in the southern Bahamas. Copeia. Bull, J.J. 1983. Evolution of Sex Determining Mechanisms. Benjamin/Cummings, Menlo Park, CA. Bull, J.J. and R.C. Vogt. 1979. Temperature-dependent sex determination in turtles. Science 206: 1186-1188. Bustard H. R. 1972. Sea Turtles: Natural History and Conservation Taplinger, New York. Carr, A.F., M.H. Carr, and A.B. Meylan. 1978. The ecology and migrations of sea turtles, 7. The west Caribbean green turtle colony. Bull. Amer Mus. Nat. Hist. 162: 1-46. Carr, A.F., Ill. 1979. The ecology of the prawn, Macrobrachium acanthurus (Weigmann) and its implications for tropical esturine management. Ph.D. Dissertation. The University of Michigan, Ann Arbor, Ml. Christens, E. 1990. Nest emergence lag in loggerhead sea turtles. J. Herpetol. 24:400-402. Coen, E. 1983. Climate. In Janzen, D.H. (ed), Costa Rican Natural History. pp.35-46. The University of Chicago Press Chicago. Cornelius, S.E. 1986. The Sea Turtles of Santa Rosa National Park Fundacion de Parques Nacionales Costa Rica. Dalrymple, G.H., J.C. Hampp, and D.J. Wellins. 1985. Male-biased sex ratio in a cold nest of a hawksbill turtle (Eretmochelys imbricata). J. Herpetol. 19:158-159.

PAGE 158

151 Demas, S and S Wachtel. 1989. Sexing the sea turtle In S.A. Eckert K.L. Eckert, and T.H. Richardson (compilers), Proceedings of the 9th Annual Workshop on Sea Turtle Conservation and Biology pp.37-39. NOAA Technical Memorandum NMFS-SEFC-232. Dutton, P.H., C.P. Whitmore, and N. Mrosovsky. 1985. Masculinisation of leatherback turtle Dermochelys coriacea hatchlings from eggs incubated in styrofoam boxes. Biol. Conserv. 31 :249-264. Etchberger, C.R., J.B. Phillips M.A Ewert, C.E. Nelson, and H.D. Prange. 1991. Effects of oxygen concentration and clutch on sex determination and physiology in red-eared slider turtles (Trachemys scripta) J. Exper. Zool. 258 : 394-403. Ewert, M. and C.E. Nelson 1991. Sex determination in turtles: diverse patterns and some possible adaptive values. Copeia 1991 :50-69. Fowler, L.E 1979 Hatching success and nest predation in the green turtles, Chelonia mydas. at Tortuguero, Costa Rica. Ecology 60:946-955. Girondot M and C. Pieau 1990. Sex determination in the critical range of temperature for marine turtles In T.H. Richardson, J.I. Richardson, and M. Donnelly (compilers), Proceedings of the 10th Annual Workshop on Sea Turtle Conservation and Biology pp.77-80. NOAA Technical Memorandum NMFS-SEFC-278. Groombridge B. and R. Luxmoore. 1989. The Green Turtle and Hawksbill (Reptilia: Cheloniidae): World Status Exploitation and Trade IUCN Conservation Monitoring Center. Cambridge, UK. Gutzke, W H N and G.L. Paukstis 1983. Influence of the hydric environment on sexual differentiation of turtles J. Exper. Zool. 226:467-469 Hendrickson, J.R. 1958. The green sea turtle, Chelonia mydas (Linn.) in Malaya and Sarawak. Proc. Zool. Soc. Lond 130:455-535.

PAGE 159

152 Horrocks, J.A. and N. McA. Scott. 1991 Nest site location and nest success in the hawksbill turtle E retmochelys imbricata in Barbados, West Indies. Mar. Ecol. Prog. Ser. 69:1-8. Horton, M. 1989. Reproductive success of sea turtles nesting on Wabasso beach, east-central Florida M. S. thesis The Virginia Polytechnic Institute and State University, Blacksburg, Virginia. Hosmer, D.W., Jr. and S. Lemeshow. 1989. Applied Logistic Regression. A Wiley-lnterscience Publication, New York. Jackson, M.E., L.U. Williamson, and J.R. Spotila. 1988. Gross morphology vs. histology: sex determination of hatchling sea turtles Marine Turtle Newsletter. 40: 10-11. Janzen F.J. and G.L. Paukstis. 1991. Environmental sex determination in reptiles: ecology, evolution, and experimental design. Quart Rev. Biol. 66:149-179. Kamezaki, N. 1990. Karyotype of the hawksbill turtle Eretmochelys imbricata, from Japan, with notes on a method for preparation of chromosomes from liver cells. Jap. J. Herpetol. 13: 111113. Kraemer J.E and R Bell. 1984. Rain-induced mortality of eggs and hatchlings of loggerhead sea turtles (Caretta caretta) on the Georgia coast. Herpetologica 36:72-77. Leh C.M.U., S.K Poon, and Y.C. Siew. 1985. Temperature-related phenomena affecting the sex of green turtle (Chelonia mydas) hatchlings in the Sarawak Turtle Islands. Sarawak Mus. J. 34 :1 83-193. Leslie A J., D.N. Penick, and J.R. Spotila. In Press. Abiotic and biotic factors affecting hatching of the leatherback turtle at Tortuguero, Costa Rica In Proceedings of the 12th Annual Workshop on Sea Turtle Conservation and Biology. NOAA Technical Memorandum.

PAGE 160

153 Limpus, C.J. P. Reed, and J D. Miller. 1983. Island and turtles. The influence of choice of nesting beach on sex ratio In J.T. Baker R M. Carter, P.W. Sammarco, and K.P. Stark (eds.) Proceedings: Inaugural Great Barrier Reef Conference, pp.397-402. James Cook University Press, Townsville, Queensland, Australia. Limpus, C.J., P. Reed, and J.D. Miller. 1985. Temperature dependent sex determination in Queensland sea turtles: intraspecific variation in Caretta caretta. In G. Grigg, R. Shine, and H Ehmann (eds.), Biology of Australasian Frogs and Reptiles, pp 343-351 Royal Zoological Society, New South Wales. Maxwell, J.A., M.A. Motara, and G.H. Frank. 1988. A micro environmental study of the effect of temperature on the sex ratios of the loggerhead turtle Caretta caretta from Tongaland Natal. S Afr J. Zool. 23:342-350. McCoy C J., R.C. Vogt, and E J. Censky. 1983. Temperature controlled sex determination in the sea turtle Lepidochelys olivacea. J. Herpetol. 17:404-406. Medrano, L., M Dorizzi, F. Rimblot, and C. Pieau. 1987. Karyotype of the sea turtle Dermochelys coriacea (Vandelli, 1761 ) Amphibia-Reptilia 8: 171-178. Meylan, A.B., P A. Meylan H.C. Frick, and J.N. Burnett-Herkes 1992. In Salmon, M. and J Wyneken (compilers) Proceedings of the 11th Annual Workshop on Sea Turtle Conservation and Biology. pp 73. NOAA Technical Memorandum NMFS-SEFSC-302. Miller, J D 1985. Embryology of marine turtles. In C Gans, F.Billett, and P.F A Maderson (eds.) Biology of the Reptilia, Vol. 14 pp.269-328. Academic Press, New York. Miller J.D. and C.J. Limpus. 1981. Incubation period and sexual differentiation in the green turtle Chelonia mydas L. In C B. Banks, and A.A. Martin (eds ), Proceed i ngs of the Melbourne Herpetological Symposium. pp.66-73. The Zoological Board of Victoria Melbourne Australia. Morreale, S J G.J Ruiz, J R. Spotila, and E.A. Standora. 1982 Temperature-dependent sex determination: current practices threaten conservation of sea turtles. Science 216:1245-1247.

PAGE 161

154 Mortimer, J.A. 1981. Reproductive ecology of the green turtle, Chelonia mydas, at Ascension Island. Ph.D. Dissertation. University of Florida, Gainesville, FL. Mortimer, J.A. 1990. The influence of beach sand characteristics on the nesting behavior and clutch survival of green turtles (Chelonia mydas). Copeia 1990:802-817. Mortimer, J.A. and A. Carr. 1987. Reproduction and migration of the Ascension Island green turtle (Chelonia mydas). Copeia 1987:103-113 Mrosovsky, N. 1982. Sex ratio bias in hatchling sea turtles from artificially incubated eggs. Biol. Conserv. 23:309-314. Mrosovsky, N. 1988 Pivotal temperatures for loggerhead turtles (Caretta caretta) from northern and southern nesting beaches. Can. J. Zool. 66:661-669. Mrosovsky, N., P.H. Dutton, and C.P. Whitmore. 1984. Sex ratio of two species of sea turtle nesting in Suriname Can. J. Zool. 62 :2227-2239. Mrosovsky, N. and J. Provancha. 1989. Sex ratio of loggerhead sea turtles hatching on a Florida beach. Can. J. Zool. 67:25332539. Mrosovsky, N. and C.L. Yntema. 1980. Temperature dependence of sexual differentiation in sea turtles: implications for conservation practices. Biol. Conserv. 18:271-280. Myers, R.L. 1981. The ecology of low diversity palm swamps near Tortuguero, Costa Rica. Ph.D. Dissertation. University of Florida, Gainesville, FL. Packard, G.C., M.J. Packard, and G.F. Birchard. 1989. Sexual differentiation and hatching success by painted turtles incubating in different thermal and hydric environments. Herpetologica 45:385-392.

PAGE 162

155 Provancha, J. and N. Mrosovsky. 1992. Estimates of hatchling sex ratios for Caretta caretta: a five year study from Cape Canaveral, Florida, 1986-1990. In Salmon, M. and J. Wyneken (compilers), Proceedings of the 11th Annual Workshop on Sea Turtle Conservation and Biology. pp.99. NOAA Technical Memorandum NMFS-SEFSC-302. Ragotzski, R.A. 1959. Mortality of loggerhead turtle eggs from excessive rainfall. Ecology 40:303-305. Redfoot, W.E. and L.M. Ehrhart. 1989. Marine turtle nesting and reproductive success in south Brevard County, Florida, 19821988. In S.A. Eckert, K.L. Eckert, and T.H. Richardson (compilers), Proceedings of the 9th Annual Workshop on Sea Turtle Conservation and Biology. pp.249-251. NOAA Technical Memorandum NMFS-SEFC-232. Rimblot-Baly, F., J. Fretey, N. Mrosovsky, J. Lescure, and C. Pieau. 1985. Sexual differentiation as a function of the incubation temperature of eggs in the sea turtle Dermochelys coriacea (Vandelli, 1761 ). Amphibia-Reptilia 6:83-92. Rimblot-Baly, F., J. Lescure, J. Fretey, and C. Pieau. 1987. Sensibilite a la temperature de la differenciation sexuelle chez la Tortue Luth, Dermochelys coriacea (Vandelli, 1761 ); application des donnees de !'incubation artificielle a l'stude de la sex-ratio dans la nature. Ann. Sci. Nat. Zool. Paris 8:277290. Ross, J.R. 1984. Adult sex ratio in the green sea turtle. Copeia 1984:774-776. SAS Institute Inc. 1989. JMP version 2.02. Cary, NC. Schulz, J.P. 1975. Sea turtles nesting in Surinam. Stichting Natuurbehoud Suriname (STINASU), Verhandeling, Netherlands.

PAGE 163

156 Shaver, D.J., D.W. Owens, A.H. Chaney, C.W. Caillouet, Jr., P. Burchfield, and R. Marquez. 1988. Styrofoam box and beach temperatures in relation to incubation and sex ratios of Kemp's ridley sea turtles. In B.A. Schroeder (compiler), Proceedings of the 8th Annual Workshop on Sea Turtle Conservation and Biology. pp 103-108. NOAA Technical Memorandum NMFS SEFC-214. Spotila J.R., E.A. Standora, S.J. Morreale, G.J. Ruiz, and C. Puccia. 1983. Methodology for the study of temperature related phenomena affecting sea turtle eggs. U.S.F.W.S. Endangered Species Report 11. Spotila, J. R., E.A. Standora, S.J. Morreale, and G.J. Ruiz. 1987. Temperature dependent sex determination in the green turtle (Chelonia mydas): effects on the sex ratio on a natural nesting beach. Herpetologica 43:74-81. Stancyk, S.E., O.R. Talbert, Jr., and J.M. Dean. 1980. Nesting activity of the loggerhead turtle Caretta caretta in South Carolina. 11. Protection of nests from raccoon predation by transplantation. Biol. Conserv 18:289-298. Standora, E.A., S.J. Morreale, G.J. Ruiz and J.R. Spotila. 1982. Sex determination in green turtle (Chelonia mydas) hatchlings can be influenced by egg position within the nest. Abstr. Bull. Ecol. Soc. Amer. 63:83. Standora, E.A. and J.R. Spotila. 1985. Temperature dependent sex determination in sea turtles. Copeia 1985:711-722 Wellins D. 1987. Use of an H-Y antigen assay for sex determination in sea turtles. Copeia 1987:46-52. Whitmore, C.P. and P.H. Dutton. 1985. Infertility, embryonic mortality and nest-site selection in leatherback and green turtles in Suriname. Biol. Conserv. 34:251-272. Wibbels, T., D.W. Owens, C.J. Limpus, and M.S. Amoss. 1989. Field testing of a sexing technique for immature sea turtles. In Proceeding of the Second Western Atlantic Turtle Symposium. pp.349-350. NOAA Technical Memorandum NMFS-SEFC-226.

PAGE 164

157 Witherington, B.E. 1986. Human and natural causes of marine turtle clutch and hatchling mortality and their relationship to hatchling production on an important Florida nesting beach. M.S. thesis University of Central Florida, Orlando FL. Yntema, C L. and N. Mrosovsky. 1980. Sexual differentiation in hatchling loggerheads (Caretta caretta) incubated at different controlled temperatures. Herpetologica 36:33-36. Yntema, C.L. and N Mrosovsky. 1982. Critical periods and pivotal temperatures for sexual differentiation in loggerhead sea turtles Can. J. Zool. 60:1012-11016. Zaborski P., M. Dorizzi, and C. Pieau. 1982 H-Y antigen expression in temperature sex-reversed turtles (E mys orbicularis) Differentiation 22 :7378. Zaborski, P M Dorizzi, and C. Pieau. 1988. Temperature-dependent gonadal differentiation in the turtle Emys orbicularis: Concordance between sexual phenotype and serological H-Y antigen expression at threshold temperature. Differentiation 38:17-20. Zar, J.H. 1984 Biostatistical Analysis. Prentice-Hall Inc. Englewood Cliffs NJ.

PAGE 165

BIOGRAPHICAL SKETCH Kazuo Horikoshi was born in Tokyo, Japan, on 9 July 1956 to parents Shigeru and Kiku Horikoshi. He received a Bachelor of Fisheries in aquaculture from the Tokyo University of Fisheries, Tokyo, Japan in 1979. He continued his graduate study at the Tokyo University of Fisheries and was awarded a Master of Fisheries in aquaculture in 1982. He enrolled in the Ph.D. program at the University of Florida in 1984. 158

PAGE 166

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Martha L. Crump, Chair 6 Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Karen A. Bjornc1!i,cochair Assistant Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. n H. Kaufman Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Clay Montague As ociate Professor of Environmental Engineering Science

PAGE 167

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentat i on and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy i6eanne A. Mort i mer Courtesy Assistant Professor of Zoology This dissertation was submitted to the Graduate Faculty of the Department of Zoology in the College of Liberal and Science and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy August, 1992 Dean Graduate School



PAGE 1

64 much of the year. Although raccoons, a common predator on loggerhead nests in the southeastern United States (e.g., Stancyk et al., 1980), also exist at Tortuguero, their role as a predator on green turtle nests seems to be minimal. However, a spatial bias by predatory raccoons was observed on a Florida beach. Horton (1989) found that raccoon predation on loggerhead nests on a Florida beach was negatively related to the distance of the nests from the dune's edge in one of his two study seasons. He hypothesized that predators search for prey close to the safety of cover, namely near the dune in Florida. At Tortuguero, locals occasionally observed jaguars, Felis onca, in the protected section of the beach during the green turtle nesting season. I occasionally found unidentified foot prints of large cats within the study area. Feral dogs were once abundant throughout the 22-mile Tortuguero beach (Fowler, 1979). Dogs may be a potential threat to coatis, especially young individuals. Spatially biased depredation of nests by coatis may be partially a result of avoiding the risk from their predators. Of the abiotic mortality factors, sporadic freshwater flooding was the most consistent factor at the Tortuguero beach throughout this study, but probably it is very unusual in sea turtle rookeries elsewhere. A single excessive rain event can force the level of ground water near or over the depth of the egg mass of green turtles and suffocate eggs and emerging hatchlings. Even if the eggs are not soaked in water, the nearby water might hinder gas exchange by saturating sand space by capillary action. Similar flooding phenomena were reported on a loggerhead beach on the Georgia coast (Ragotskie, 1959; Kraemer and Bell, 1984). Ragotskie (1959)



PAGE 1

Figure 31. Seasonal fluctuation of sex ratio of green turtle hatchlings in 1988 at Tortuguero, Costa Rica. Shown is the bimonthly mean one standard error; the number above each bar indicates sample size. For nests in the open zone, the Roman numerals represent a rank from highest to lowest female ratio analyzed by Fisher's procedure of least significant difference (alpha=O.05). Those bimonthly periods that do not share a common figure have significantly different mean sex ratios. For nests in the vegetation/border zone, no significant difference was detected within the season.



PAGE 1

115 3oOPEN ZONE, 60CM DEPTH 0 1986 -*1988 S29 ----1989 w CC W 280 z < 27 CO 26 JUL .AUG. SEP. OCT. NOV. 30. VEGETATION/BORDER ZONE 60CM DEPTH C5~ -*-1988 de o 6, a 1989 w 28 w c 0 S 27 26 JUL. AUG. SEP. OCT. NOV. Figure 14. Seasonal fluctuation of monthly sand temperatures at a depth of 60 cm in 1986, 1988, and 1989 at the Tortuguero beach, Costa Rica.



PAGE 1

132 N=233 N=330 N=456 100 OPEN ZONE 80. 60 Z VEGETATION/ LL BORDER ZONE 0 40 0 20 0 1986 1988 1989 Figure 26. Relative proportion of green turtle nests depredated by mammals that occurred in the open zone and in the vegetation/border zone. Data were gathered during the twomile beach censuses at Tortuguero, Costa Rica; 48 censuses were made in 1986, 56 in 1988, and 80 in 1989.



PAGE 1

40 Although ghost crabs, Ocypode guadrata, were very common on the Tortuguero beach, predation by ghost crabs was not a serious factor affecting egg survivorship (mean egg loss of 0.8% in the open zone and 1.3% in the vegetation/border zone throughout the three years). Only three sample nests were confirmed to have been damaged by ghost crabs during the three-year study. Two of them were completely destroyed, and the other was partially destroyed by ghost crabs. For those nests that were completely destroyed, numerous crab burrows were observed around the nests during the incubation period; at the time I excavated the nests to check on the eggs, only the flagging tape marker and many scattered torn shells remained at the bottom of the nest. For the partially destroyed nest, I used a characteristic nipped hole on the egg shells for identifying the eggs damaged by ghost crabs. However, my assessment rate of ghost crab predation for this study might b conservative. In cases where only a few eggs from a clutch were dep0redated by crabs and the typical nipped holes were not clear enough to identify, torn shells were classed under the "Rotten intact or ruptured eggs" category. Predation by termites, unknown species, was a very minor factor causing mortality of eggs. Eight sample nests were found to be partially included in termite nests, but generally only small numbers of eggs from each clutch (mean 3.7 eggs, SE=2.1, range=1 to 16, n=7) were killed by surrounding clay materials. The insides of most eggs were stuffed by the clay material through a few tiny holes. The termite nests were typically constructed under washed-



PAGE 1

100 32 1986 3MAX 2220 22 1988 0~ 30. 2I LU 24WN w22 20 <32zMAX 1989


PAGE 1

61 the open area (Leslie et al., in press). Without co-ocurrence of the two species, the trend of nest site selection of each species at Tortuguero agreed with that at Surinam. Witherington (1986) suspected that the preference of green turtles to nest near dunes may be a strategy to avoid inundation although he could not detect significant differences of emergence success between the two species during the year. Topography and vegetation of green turtle beaches show extreme variation: from low profile naked islands (e.g., French Frigate Shoals) to beaches backed by tropical jungle (e.g., Tortuguero and Sarawak). However, preference for the upper beach might be characteristic of green turtles. It would be interesting to investigate the geographic variation of nest site selection in relation to the thermal profile and egg survivorship factors on each beach. Overall Reproduction Rate To date, very few studies have assessed overall egg survivorship of green turtle nests on natural beaches. Although several studies have investigated mortality factors on natural beaches, their sample seasons were rather limited or the analysis included only those nests that successfully produced some hatchlings (Mortimer, 1981; Balazs, 1980; Schulz, 1975). In Surinam, a detailed quantitative study by Whitmore and Dutton (1985) showed a relatively high hatching success of green turtle nests (mean 80.4% in 1981 and 1982). However, their samples were



PAGE 1

103 -MEAN 1000 ..... 1986 S----a--..1988 800 ./.O.. 1989 600 *** I ---,,---o-I.. ....... 'N '0 .. 200. 0 I I I JUL. AUG. SEP. OCT. NOV. Figure 6. Monthly rainfall in the 1986, 1988 and 1989 incubation seasons of green turtle eggs at Tortuguero, Costa Rica. Mean values from 1978 through 1989 are shown for comparison.



PAGE 1

80 Overall hatchling emergence success is estimated as 57.2% in 1986, 45.7% in 1988, and 66.6% in 1989. No significant difference in emergence success between the two zones was found in any year. A significant seasonal difference in emergence success was found only in 1988, when a rare hurricane hit the Tortuguero beach. Thus, egg survivorship probably does not have a great effect on the primary sex ratio produced on the Tortuguero beach. Both abiotic (surf, freshwater inundation, hurricane) and biotic factors (predation, turtle digging) were responsible for reducing the survivorship of green turtle nests. The relative importance of each mortality factor varied from year to year. Freshwater inundation when ground water levels were raised by excessive rainfall, was the most consistent mortality factor of green turtle nests throughout this study. The probabilities of excessive rain events (>140 mm/day) were highest in the latter part of the incubation season (October and November). Thus, most nests deposited during the peak nesting period (August through September) are affected if the flooding occurs in October and November. Tortuguero green turtles have not adjusted their nesting season relative to excessive rainfall events. Mammalian predation was the only significant mortality factor that showed a distinct spatial difference between the zones. Mammalian predation, mostly by coatis, was greater on the nests in the vegetation/border zone than on nests in the open zone throughout each year. The extent of predation was most severe in the 1989 season when the nest density was about half that in the 1986 and 1988 seasons.



PAGE 1

156 Shaver, D.J., D.W. Owens, A.H. Chaney, C.W. Caillouet, Jr., P. Burchfield, and R. Marquez. 1988. Styrofoam box and beach temperatures in relation to incubation and sex ratios of Kemp's ridley sea turtles. In B.A. Schroeder (compiler), Proceedings of the 8th Annual Workshop on Sea Turtle Conservation and Biology. pp.103-108. NOAA Technical Memorandum NMFSSEFC-214. Spotila J.R., E.A. Standora, S.J. Morreale, G.J. Ruiz, and C. Puccia. 1983. Methodology for the study of temperature related phenomena affecting sea turtle eggs. U.S.F.W.S. Endangered Species Report 11. Spotila, J. R., E.A. Standora, S.J. Morreale, and G.J. Ruiz. 1987. Temperature dependent sex determination in the green turtle (Chelonia mydas): effects on the sex ratio on a natural nesting beach. Herpetologica 43:74-81. Stancyk, S.E., O.R. Talbert, Jr., and J.M. Dean. 1980. Nesting activity of the loggerhead turtle Caretta caretta in South Carolina. II. Protection of nests from raccoon predation by transplantation. Biol. Conserv. 18:289-298. Standora, E.A., S.J. Morreale, G.J. Ruiz, and J.R. Spotila. 1982. Sex determination in green turtle (Chelonia mydas) hatchlings can be influenced by egg position within the nest. Abstr. Bull. Ecol. Soc. Amer. 63:83. Standora, E.A. and J.R. Spotila. 1985. Temperature dependent sex determination in sea turtles. Copeia 1985:711-722. Wellins, D. 1987. Use of an H-Y antigen assay for sex determination in sea turtles. Copeia 1987:46-52. Whitmore, C.P. and P.H. Dutton. 1985. Infertility, embryonic mortality and nest-site selection in leatherback and green turtles in Suriname. Biol. Conserv. 34:251-272. Wibbels, T., D.W. Owens, C.J. Limpus, and M.S. Amoss. 1989. Field testing of a sexing technique for immature sea turtles. In Proceeding of the Second Western Atlantic Turtle Symposium. pp.349-350. NOAA Technical Memorandum NMFS-SEFC-226.



PAGE 1

112 29.0 29.0 OPEN, 60cm .-.. -OPEN, 80cm w ..................... -VEGETATIONSBORDER S 28.5 ......* *....... .60cm ..... .*...F ***... .... ......... VEGETATIONBORDER W 80cm 0 z IH 28.0 z 27.5 o ..... .... .......--. 1i 1. 1. 1 1 1 1 27.0 , 2 4 6 8 10 12 14 16 18 20 22 24 TIME(HOURS) 29.5 -a-OPEN --VEGETATION/BORDER 29.0 0~ CLL Ll 28.5 H C') z S 28.0 27.5 ,II, Il 2 4 6 8 10 12 14 16 18 20 22 24 TIME(HOURS) Figure 12. Daily fluctuation of green turtle nest temperatures and sand temperatures at the Tortuguero beach, Costa Rica. Data were pooled from four separate days; 13 August 1988, 29 September 1988, 12 August 1989 and 14 October 1989. Each arrow shows the time representing the 24-hour mean value.



PAGE 1

58 does on the emergence time of nesting females. In addition, the extent of erosion is generally also correlated with activity of waves. Spatial Distribution between the Zones It is interesting to compare the spatial distribution of green turtles in 1986, 1988, 1989 at this study site with similar data collected in a different area of the Tortuguero beach (northern-most 8 kin) in 1986, 1987, 1988 (Bjorndal and Bolten, 1992). Although there was a slight difference in definition of the zonation of the beach (my vegetation/border zone is probably slightly larger than their combined vegetation zone and border zone), both studies revealed significant yearly variation of spatial distribution between the shaded area and open area over years, and that there was no common seasonal trend of fluctuation among the years. Comparison between the common years (1986 vs 1988) reveals that the yearly fluctuation of the spatial distribution showed the same trend: a higher proportion of nests in the open area in 1986 than in 1988. Bjorndal and Bolten (1992) hypothesized that the 1986 season's higher rainfall during their study period (July to September), and consequently the higher sand moisture, may have physically facilitated nest digging in the open area relative to the drier years in 1987, 1988, and that the wetter sand may be partially responsible for the high proportion of nests in the open area in the 1986 season. When I compared the relationship between the proportion of nests deposited in the open zone versus the



PAGE 1

Figure 23. Fate of green turtle eggs from the sample nests in 1986, 1988, and 1989 at Tortuguero, Costa Rica.



PAGE 1

125 100oo -OPEN ZONE 8 6 4 80 6 60 40 EO 6 7 0) 20 Lu 0 0 D7 C/) ( 100VEGETATION/BORDER ZONE LU 60 M7 Z LU 80 Z 40 8 20 '0 , z 0 1oo TOTAL 16 80 16 60 -16 13 13 40 -14 143 40 20 0 5 1-15 JUL. 16-31 JUL. 1-15 AUG. 16-31 AUG. 1-15 SEP. 16-30 SEP. PERIOD OF SAMPLE NESTS DEPOSITED Figure 21. Seasonal fluctuation of emergence success of green turtle sample nests in 1988 at Tortuguero, Costa Rica. Monthly mean standard error with the sample size are shown.



PAGE 1

137 A 1.0 o00oo oo00 0 o 0.8 0 0.6 m 0 o 0.6 0 00 00 0 0 8 CL0 0 0 u 0.4 2 0.20. 00 00o LL 0 00 0 0 0.00 0 00oo 0Oo I I | I 26 27 28 29 30 31 MEAN NEST TEMPERATURE (oC) B 3 1.0 0 o0 o o F -0.84 m0 o0 O 0.6 0 o 0o o U 0.40 0.2o o o LL. 00 0.00 o co 0 0 26 27 28 29 30 31 MEAN SAND TEMPERATURE (oC)



PAGE 1

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. L;-~~> Martha L. Grump, Chair Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Karen A. Bjorndcl, CochairAssistant Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Jkn H. Kaufman Professor of Zoology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Clay Montague As ociate Professor of Environmental Engineering Science



PAGE 1

65 suggested that because of the higher probability of excessive rain late in the incubation season (September) on the Georgia coast, successful emergence from nests deposited after the middle of the nesting season (1 July) decreases. Kraemer and Bell (1984) found that the temporal distribution of loggerhead nests did not avoid the excessive rainfall events in September. At Tortuguero, the probability of excessive rain events (>140 mm/day) also increased during the latter part of the incubation season (October and November) (Table 13). However, most nests deposited during the peak nesting period of Tortuguero (August through September) are very much affected if flooding occurs in October and November. Tortuguero green turtles have not adjusted their nesting season relative to excessive rainfall events. Kraemer and Bell (1984) proposed a hypothesis that nesting by loggerhead turtles near the beach dune at the upper beach may be an adaptive response to escape the excessive rains. This is a very tempting hypothesis to explain partially the nesting preference for the upper vegetated beach by Tortuguero green turtles. However, the difference in damage by flooding between the two zones was inconsistent among the study years, and the extent of the differences was rather small. Seasonal Fluctuation of Emergence Success It is not clear to what extent emergence success of green turtle nests significantly fluctuates within a season at Tortuguero. During the three year study, a significant difference was found only in 1988. The main cause of those fluctuations was the two



PAGE 1

142 1.0 o0oo oo0000 i 0.8oo _j 0.80 o0 Fn 0 0 0 0 O 0.o0 ao 0.4 wo w 0 0 0 0.2 O o0* o0 0.0 -o *0 ao e I0.0I I I 26.5 27.5 28.5 29.5 30.5 MEAN NEST TEMPERATURE (Co) 0.6 0.4 + + + + + + +* 0.2+ + ++ ++ + ++ + + + D 0.0 + 4. + +4+o++ + ++ + Wl + + + + + CC -0.2+ + + -0.4 + -0.6 II I II 0.0 0.2 0.4 0.6 0.8 1.0 PPREDICTED PROBABILITY OF FEMALE



PAGE 1

23 stopped, the levels of the water table recorded were probably much lower than the actual maximum level. The water marks remaining in the depression area suggested that the water tables on 6 December 1986 (178 mm/day), and on 1 November 1989 (total 367 mm for three days) also rose above the general depth of the clutches. During the 1988 season, freshwater flooding was observed on one occasion, associated with a heavy rain event on 6 October (221 mm/day). In addition to rainfall, high waves probably raised the water table. While the amount of rainfall was relatively small (28.5 to 37.0 mm) on 19 and 20 November 1989, the water table showed a substantial high level on 20 November. At the same time, very high waves (>2 m) were recorded. This height of waves rarely occurred in Tortuguero during the study. Except on this one occasion, waves of this height only occurred during the time that Hurricane Joan passed near Tortuguero on 21-22 October 1988. Data on the water table during the hurricane are not available. Waves Figure 10 shows frequency distributions of relative heights of waves throughout each incubation season in 1986, 1988 and 1989. On the Tortuguero beach, heights of waves were primarily associated with the size of swells from offshore rather than with the wind speed over the beach. Wave actions were generally high in Tortuguero during most of each season. Less than one third of the days were identified as calm throughout each season (wave height 0.5 m; 17.7% in 1986, 31.4% in 1988 and 31.5% in 1989). Wave



PAGE 1

5 dominated turtles much of the time (Shaver et al., 1988). On the other hand, the hatchery on a Sarawak beach produced highly female biased green turtle hatchlings for many years (Leh et al., 1985). Location and type of artificial hatchery can easily alter the sex ratio of hatchlings produced. Currently there is no consensus on the best sex ratio to produce in hatcheries because the sex ratios of sea turtle populations and their dynamics are uncertain. Therefore, to understand better the ecological and conservation implications of TSD among sea turtles, it is critical to know the process of sex determination and the primary sex ratio of hatchlings on a natural beach. However, natural sex ratios of hatchling sea turtles have been studied in only a few populations, and estimation of primary sex ratio of a population has not been feasible, partially because of small sample sizes and limited seasonal coverage (Limpus et al., 1983; Maxwell et al., 1988; Rimblot-Baly et al., 1987; Spotila et al., 1987). Quantitative investigations that cover the full seasonal profile exist only for the Suriname green turtle and leatherback turtle colonies (Mrosovsky et al., 1984) and a Florida loggerhead turtle colony (Mrosovsky and Provancha, 1989; Provancha and Mrosovsky, 1992). However, these studies do not include information on egg survivorship on the beach. To estimate reliably the primary sex ratio, egg survivorship is an essential factor because survivorship might differ in different thermal zones on the beach or within seasons. Fowler (1979) found that the nest position affected survival rate of green turtle nests at Tortuguero because of differential mammal predation and inundation rates. Nests near or in the beach



PAGE 1

83 Table 1. Ranges of daily fluctuation of sand and green turtle nest temperatures during 24 hours at Tortuguero, Costa Rica. Mean SE Range No. Sample (oc) (oc) Sand temperatures on transects Open Zone 60cm depth 0.49 0.03 0.3-0.8 n=20 80cm depth 0.49 0.04 0.2-0.8 n=20 Vegetation/border Zone 60cm depth 0.48 0.03 0.4-0.6 n=8 80cm depth 0.41 0.07 0.1-0.8 n=8 Temperatures at the center of clutches Open Zone 0.46 0.04 0.2-0.8 n=14 Vegetation/border Zone 0.49 0.04 0.2-0.8 n=14 Samples were collected and pooled from four separate days on 13 August and 29 September 1988 and 12 August and 14 October 1989.



PAGE 1

11 the previous night. Crawl tracks and nest marks were used to locate nests and to identify species (Pritchard et al., 1983). In 1986 the census was carried out from 2 July through 29 November at an interval of once every two to four days. Two surveys on 10 and 14 July were not completed because high waves and strong wind erased most of the tracks before the surveys began. Therefore, these surveys were eliminated from analysis. In 1988, the census was carried out from 16 June through 30 November. The interval between surveys was two to five days; most were two to three days. No survey was conducted during the evacuation for Hurricane Joan between 19 and 23 October. In 1989, the census was conducted from 17 June through 30 November. The interval between surveys was two to three days; most were two days. The census data were pooled into half month periods (1st to 15th, and 16th to 30th or 31st). The daily mean number of nests deposited per period was calculated by dividing the total number of nests counted by the number of censuses during each period. The total number of nests deposited during each period was calculated by multiplying the daily mean number of nests by the number of days during the period. Total number of nests over the season was the sum of the numbers of nests during each period. The proportion of nests deposited in the vegetation/border zone and in the open sand zone was calculated for each half month period. The proportion of nests in the two zones for the entire season was calculated from pooled numbers of nests counted during the censuses.



PAGE 1

85 Table 3. Multiple comparison table of the monthly sand temperatures at a depth of 60cm in the open zone and in the vegetation/border zone in 1986, 1988 and 1989 (Fisher's procedure of least significant difference, alpha=0.05). 1986 3 2 111 Open Zone JUL AUG SEP OCT NOV 3 2 1 1 1 Vegetation! JUL AUG SEP OCT NOV border Zone 1988 2 2 1 3 3 Open Zone JUL AUG SEP OCT NOV 2 2 1 3 3 Vegetation! JUL AUG SEP OCT NOV border Zone 3 2 1 1,2 3 Open Zone JUL AUG SEP OCT NOV 2 1 1 1 3 Vegetation! JUL AUG SEP OCT NOV border Zone Those months that do not share a common number have significantly different mean temperatures. The numbers are ranked from highest to lowest monthly mean temperatures.



PAGE 1

60 The area available to nesting turtles in the vegetation/border zone (about 3 m wide) is much smaller relative to the nesting area in the open zone (up to about 30 m wide) at the Tortuguero beach. Therefore, the slightly higher or equal distribution of nests in the vegetation/border zone means that the vertical distribution of green turtle nests is biased toward the upper part of the vegetated beach. Similar preference of nest site selection for this species is reported in other regions. Bustard (1972) noted that in the Australian cays, green turtles show some tendency to nest close to substantial vegetation. Cornelius (1986) noted that green turtles in Pacific Costa Rica have a tendency to nest beneath the vegetation of the upper beach and rarely in the mid beach. In an island of Ogasawara Islands, Japan, 21 of 28 green turtle nests were deposited near or under the dense vegetation on the upper beach during June (Horikoshi, unpublished data). In Surinam, green turtles deposit more nests (63%) in the border and vegetation area than in the open area, whereas leatherbacks predominantly lay in the open area (87%) (Whitmore and Dutton, 1985). In Florida, green turtles nested in a higher location near the beach dune than did loggerhead turtles (Witherington, 1986). Whitmore and Dutton (1985) suspected that the interspecific difference may be due to interspecific competition: green turtles are pushed up to the upper beach because leatherback nests were deeper and the risk of destruction of nests of green turtles is higher than that of leatherbacks. In Tortuguero, leatherback turtles also use the same beach as green turtles, but their nesting season rarely overlaps with the green turtle's season. However, the nest distribution of leatherbacks is highly skewed to



PAGE 1

131 10 8 a a w 0 6 ZC/) LL J 4c 0 0-9 10-19 20-29 30-39 40-49 50-59 60-69 72 DAYS AFTER DEPOSITION OF EGGS Figure 25. Timing of mammalian predation on sample green turtle nests at Tortuguero, Costa Rica.



PAGE 1

128 VEGETATION OPEN ZONE BORDER ZONE 1986 19 881 N= 2915 ~N=2801 1988 N=3899 N=571 1989 N=5890 N=621 KILLED BY SURF E DESTROYED BY TURTLES SKILLED BY F/W FLOODING NO APPARENT DEVELOPMENT KILLED BY F/W FLOODING & HURRICANE 0 CAUSE OF EMBRYONIC DEATH UNKNOWN 0 PREDATION BY MAMMALS U OhER [ PREDATION BY CRABS, OTHERS HATCHUNG EMERGED



PAGE 1

71 In 1988, the total amount of rainfall throughout the incubation period was very close to the mean value of the last 12 years, but the 1988 season had much higher seasonal fluctuation in rainfall than the average of the last 12 years. Rainfall in August 1988 (256 mm) was the lowest on record whereas rainfall in October (790 mm) was the second highest since 1978. Above all, Hurricane Joan greatly influenced the survivorship and sex ratio of the nests in the latter half of the 1988 season. The weather in 1988 was also atypical for the Tortuguero beach. If Hurricane Joan had not occurred in 1988, the proportion of females estimated for the primary sex ratio of the 1988 season (40.6%) probably would have been higher. Predictability of Sex Ratio by Environmental Factors Mrosovsky et al. (1984) cautioned that accurate estimation of the primary sex ratio of sea turtles from a combination of data on the pivotal temperature from laboratory experiments and thermal profiles on the beach is impossible because of the many thermal influences and their possible interactions throughout the season. Some of the critical factors are metabolic heat, exact pivotal levels and critical periods, nest depth, and spatial factors. This study's direct thermal measurement in individual nests is the most accurate measure available to obtain the thermal condition of the nest, and the thermal data included the variation from metabolic heat, nest depth and spatial factors. However, even the thermal data in the nest were not sufficient to predict the sex ratio of Tortuguero green turtles.



PAGE 1

22 position of the center of the clutch is 69.0 cm below the sand surface. The top of the clutch is 58.1 cm below the sand surface. Ground Wate Figures 8 and 9 show seasonal fluctuations of the ground water table and rainfall during the 1986 and 1989 seasons. During the 1988 season, the wells were stolen several times during the study, therefore only very scattered data were available. The level of the water table on the beach was affected by rainfall. Continuous rainfall and occasional heavy rainfall were associated with a rise in the ground water table. During the 1986 season, which was very wet throughout, the level of the water table frequently rose close to the general depth of the clutches in both zones (overall mean bottom depth of clutches was 77.9 cm; Figure 7). On the other hand, the level of ground water during the 1989 season rose close to the depth of clutches only early and late during the incubation season when rainfall was heavy. On several occasions, excessive rainfall caused freshwater flooding on the beach. Flooding events were confirmed by water marks that remained at the margin of a depression area on the beach that recorded the highest level of water pools during the excessive rain events. However, only one event (5 August 1986; 141 mm/day) was actually recorded on the well as a sharp increase of the water table much above the general depth of clutches on the beach. Because the measurements of the water table at the time of other excessive rain events were taken 6 to 9 hours after the rain had



PAGE 1

116 O OPEN ZONE VEGETATION/BORDER 2NE 30.0 wO S 29.0 O O R"..O O 0 2 28.0


PAGE 1

155 Provancha, J. and N. Mrosovsky. 1992. Estimates of hatchling sex ratios for Caretta caretta: a five year study from Cape Canaveral, Florida, 1986-1990. In Salmon, M. and J. Wyneken (compilers), Proceedings of the 11th Annual Workshop on Sea Turtle Conservation and Biology. pp.99. NOAA Technical Memorandum NMFS-SEFSC-302. Ragotzski, R.A. 1959. Mortality of loggerhead turtle eggs from excessive rainfall. Ecology 40:303-305. Redfoot, W.E. and L.M. Ehrhart. 1989. Marine turtle nesting and reproductive success in south Brevard County, Florida, 19821988. In S.A. Eckert, K.L. Eckert, and T.H. Richardson (compilers), Proceedings of the 9th Annual Workshop on Sea Turtle Conservation and Biology. pp.249-251. NOAA Technical Memorandum NMFS-SEFC-232. Rimblot-Baly, F., J. Fretey, N. Mrosovsky, J. Lescure, and C. Pieau. 1985. Sexual differentiation as a function of the incubation temperature of eggs in the sea turtle Dermochelys coriacea (Vandelli, 1761). Amphibia-Reptilia 6:83-92. Rimblot-Baly, F., J. Lescure, J. Fretey, and C. Pieau. 1987. Sensibilite a la temperature de la differenciation sexuelle chez Ia Tortue Luth, Dermochelys coriacea (Vandelli, 1761); application des donnees de I'incubation artificielle a I'stude de la sex-ratio dans la nature. Ann. Sci. Nat. Zool. Paris 8:277290. Ross, J.R. 1984. Adult sex ratio in the green sea turtle. Copeia 1984:774-776. SAS Institute Inc. 1989. JMP version 2.02. Cary, NC. Schulz, J.P. 1975. Sea turtles nesting in Surinam. Stichting Natuurbehoud Suriname (STINASU), Verhandeling, Netherlands.



PAGE 1

76 Mrosovsky et al. (1984) found as great a variability in the relationship between sex ratios and incubation periods for Surinam green turtles as I did for Tortuguero green turtles. They suggested that if the average of a large number of samples was used, it might be possible to make a reasonable prediction of primary sex ratio from the above model. However, collecting a large sample in each thermal zone in each seasonal segment is very labor intensive. In Tortuguero, where thermal zones have a significant effect on sex ratio, the number of samples needed is doubled, and the reliability of the estimate of primary sex ratio is reduced by adding another variable. In conclusion, due to the high level of variation, the sex ratio of green turtle nests at Tortuguero cannot yet be predicted with accuracy from the nest temperature and other environmental factors, such as sand temperature, nest depth, incubation period, rainfall, and nesting zone. In the future, long term accumulation of data correlating rainfall and sex ratios might be used to roughly estimate primary sex ratio at Tortuguero. Currently, direct sexing of periodic subsamples is the only accurate method to investigate primary sex ratio on a natural beach. Moreover, an advantage of direct sexing is that any variation resulting from genetic influences will be incorporated into the sex ratio derived. Unfortunately, immature sea turtles are not sexually dimorphic. Because they have homomorphic sex chromosomes, reliable sexing of sea turtle hatchlings is only possible by sacrificing turtles and performing a time-consuming histological examination of their gonads (Jackson et al., 1988). Nonsacrificial



PAGE 1

106 200 1986 100 0 JUL. AM SEP. OCT. NV DEC. SUFWACE 1986/BORDER/VEGETATION ZONE 40 1 iSD I _______MEAN DEPTH OF o 80 CLUTCH BOTTOM 1 SD 120 BOTTOM OF WELL JUL. AUa SEP. OCT. -NOV. -DEC. SU FKE 1986/OPEN ZONE 40 ov 1 SD I S ____MEAN DEPTH OF 80 CLUTCH BOTTOM 1 SD 120 BOTTOM OF WELL JUL. AUG. SEP. OCT. NOV. DEC.



PAGE 1

134 16 1986 14 12 S 10 U) 8z LU 6 O 0 4 2 o LU 16 1988 LI 14 ..J 12 I10 8 D LU 6 I4 2 Cl) 0 l LU Z 16 L, 1989 O 14 L 12 10 D 8 6 4 2 0 JUN. JUL. JUL. AUG. AUG. SEP. SEP. OCT. OCT. NOV. NOV. 16-31 1-15 16-31 1-15 16-31 1-15 16-30 1-15 16-31 1-15 16-30 Figure 28. Temporal distribution of excavation of green turtle nests by female green turtles at Tortuguero, Costa Rica.



PAGE 1

104 80 O 60/ z L 0 j 40* 00 20 0 LL 0 -FAN.,,' 40 50 60 70 80 90 100 110 120 130 BOTTOM OF EGG CHAMBER (CM) Figure 7. Frequency distribution of the bottom depth of green turtle clutches at Tortuguero, Costa Rica.



PAGE 1

135 c20 O Y =1.0564 + 4.214 X, RA2 = 0.757 LU n =31, t = 9.50, p < 0.0001 15 0 0 10O LL o I-5 100 n0 CD. 0 D O0 00 1000 2000 3000 4000 NUMBER OF NESTS DEPOSITED IN TWO-MILE SECTION OF BEACH IN A HALF-MONTH PERIOD Figure 29. Relationship between the number of green turtle nests excavated by female green turtles per census and the number of green turtle nests deposited within the two-mile study area of the Tortuguero beach during a half-month period. Data were collected in the 1986, 1988, and 1989 incubation seasons of green turtle eggs.



PAGE 1

126 OPEN ZONE 1008 9 9 80 604020 01 80 (0 u2i o VEGETATION/BORDER ZONE Sloo -OA L 80 1610 O Z0 20 60 2 40 20 Z 0 16-31 JUL. 1-15 AUG. 16-31 AUG. 1-15 SEP. 16-30 SEP. 1-15 OCT. PERIOD OF SAMPLE NESTS DEPOSITED Figure 22. Seasonal fluctuation of emergence success of green turtle sample nests in 1989 at Tortuguero, Costa Rica. Monthly mean + standard error with the sample size are shown.



PAGE 1

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 EGG SURVIVORSHIP AND PRIMARY SEX RATIO OF GREEN TURTLES, CHELONIA MYDAS, AT TORTUGUERO, COSTA RICA By KAZUO HORIKOSHI AUGUST, 1992 Chair: Martha L. Crump Cochair: Karen A. Bjorndal Major Department: Zoology Tortuguero beach in Costa Rica is one of the last major nesting sites in the western Atlantic for green turtles. During 1986, 1988 and 1989, I studied temporal and spatial nest distribution, egg survivorship, and sex ratio of hatchlings in two thermal zones: the vegetation/border zone (within 2 m of the border of dense vegetation) and the open zone (between the vegetation/border zone and the sea). Although the distribution of nests between the two zones differed significantly from year to year, it was biased toward the vegetated area of the beach throughout each season. vi



PAGE 1

87 Table 5. Descriptive fate of green turtle eggs from the sample nests at the Tortuguero beach during the 1986, 1988 and 1989 seasons. 1986 1988 1989 Egg Fate Open Vegetation Open Vegetation Open Vegetation border border border Unhatched eggs Destroyed by mammals 79 19 0 606 296 1920 Destroyed by ghost crabs 0 109 93 0 0 0 Destroyed by termits 0 3 2 0 1 20 Destroyed by ants? 30 6 21 7 67 13 Destroyed by female turtles 0 256 124 296 0 0 Destroyed by plant roots 0 0 0 0 0 3 Destroyed by unknown cause 0 0 49 97 0 0 Washed out by surf 445 244 0 0 107 0 Embryonic death by surf 3 7 111 112 0 0 0 Embryonic death by flooding 396 498 118 623 481 221 Embryonic death by Hurricane Joan 0 0 61 482 0 0 Embryonic death by flooding or 0 0 299 143 0 0 Hurricane Joan Embryonic death with no 84 86 174 223 202 140 apparent physical disturbance No apparent development 57 36 224 168 133 92 Rotten intact or ruptured eggs 184 125 294 359 118 133 Death at pipping 0 0 0 6 2 2 Total dead eggs 1312 1493 1571 3010 1407 2544 Unemerged Hatchlings Hatchlings in egg chamber 1 2 7 9 18 8 1 8 Hatchlings died above egg 0 0 0 97 20 1 5 chamber by flooding Hatchlings died above egg 0 0 0 134 0 0 chamber by Hurricane Joan Hatchlings tangled with rope 0 0 0 0 10 0 above egg chamber Hatchlings depredated by mammals 0 0 0 0 0 1 Total hatchlings unemerged 1 2 7 9 249 38 34 Emerged Hatchlings 1591 1301 2319 2456 4445 3638 Total eggs 2915 2801 3899 5715 5890 6216



PAGE 1

7 characteristics of natural nests on the beach, on which the TSD phenomenon operates, and to assess the predictability of sex ratio from environmental factors. This parameter, primary sex ratio, is not only necessary for improving conservation practices, but it also will increase our understanding of the dynamics of the most important population of green turtles in the western Atlantic Ocean. Furthermore, I believe that information on primary sex ratio will also give us a key to assess adult sex ratio of all sea turtles.



PAGE 1

53 Multiple Logistic Regression Model Because of a strong correlation between mean nest temperatures and mean sand temperatures (R2=0.92), only mean nest temperature was used for the multiple regression model to avoid a problem of multi-colinearity. Mean nest temperature, incubation period, three sets of rainfall data, bottom depth of nests, and zone of the nests (the open zone=1, the vegetation/border zone=0) were entered into logistic multiple regression models. Backward elimination (Table 10) and stepwise regression procedures were used to eliminate nonsignificant variables (alpha=0.05). For both procedures, the same model was selected, in which mean nest temperatures, incubation period, mean daily rainfall from deposition through the critical period (R-Ill), and zone significantly explained some variation of sex ratios (pseudo R2=0.24, Table 11). Figure 38 shows residuals on this selected model. Although the selected model was slightly improved from the best single variable model of mean nest temperatures (Figure 33), fitness of the model was still low. All first order interactions of the selected variables were tested individually by a likelihood ratio test (Hosmer and Lemesho, 1989) to determine their significant contribution to the selected model. Two interactions, between incubation period and zone (G=10.42, df=2, p<0.025) and between mean daily rainfall (R-Ill) and zone (G=7.44, df=2, p<0.01), were significant. However, in both cases when each interaction was included in the selected model, the



PAGE 1

44 nest were collected for sexing. Some embryos were not fully developed and some tissues deteriorated before being fixed, thus the number of sexed gonads was reduced to an average of 17.8 gonads per clutch (SE=0.4, range=2 to 20, n=55 nests, total turtles=979). Undetermined gonads, which possessed both germinal epithelium (female component) and seminiferous tubules (male component), were found in seven of the 979 individuals (0.7%). Sex Ratio vs Temperature To date no laboratory experiment to determine accurately the critical period of green turtle eggs in sex determination has been conducted. Spotila et al. (1987) found that mean incubation temperatures during the middle third of development explained the observed sex ratio in green turtles on the natural beach in Tortuguero. Results of temperature shift experiments in the laboratory with loggerhead turtles, Caretta ;aretta, (Yntema and Mrosovsky, 1982) also suggested that the thermosensitive period for sex determination occurs during the middle third of development. The middle third of development is also the critical period in freshwater turtle species (e.g., Bull and Vogt, 1979). Thus, I used the middle third of the development period as a critical period in green turtles for this study. Emergence lag (interval between pipping and emergence) in green turtles is believed to take from 3 to 7 days (Balazs and Ross, 1974), but no quantitative data are available. Recently, Christens (1990) found the mean lag period to be 5.4 days for loggerhead



PAGE 1

TABLE OF CONTENTS ACKNOW LEDGEMENTS ............................................................................................... i i ABSTRACT .................................................................................................................... v i CHAPTERS 1 INTRODUCTION .................................................................................................. 1 2 METHODS ............................................................................................................. 8 Study Area ................................................................................................... 8 W eather Record ........................................................................................ 9 Beach Zones ................................................................................................ 10 Nesting Census .......................................................................................... 10 Sam pling of Egg Survivorship Nests ............................................... 12 Sampling of Nests to Determine Sex Ratio ................................. 14 Histology and Sexing .............................................................................. 16 Sand Temperatures ................................................................................. 16 Statistical Analysis ............................................................................... 17 3 RESULTS ............................................................................................................ 18 Physical Environment Surrounding Nests ..................................... 18 Air Temperatures .............................................................................. 18 R a in fa ll .................................................................................................. 1 9 Vertical Position of Clutches ...................................................... 2 1 Ground W ater ....................................................................................... 22 W aves ...................................................................................................... 23 Daily Fluctuation of Sand and Nest Temperatures ............... 24 Seasonal Fluctuation of Sand Tem peratures ......................... 26 Seasonal Nest Distribution and Nest-Site Selection .............. 28 Analysis of Egg Survivorship Samples .......................................... 30 Fates of Sample Nests .................................................................... 30 Abiotic and Biotic Factors Affecting Egg Survivorship .... 33 i v



PAGE 1

Figure 33. Logistic regression model and its residuals to explain sex ratio fluctuation of green turtle hatchlings as a function of mean nest temperature during the middle third of the development period. Data were collected at Tortuguero, Costa Rica, in the 1986 and 1988 seasons. The superimposed line shows a logistic regression line to fit: Y = exp(-34.45 + 1.17X)/(1+ exp(-34.45 + 1.17X)), pseudo R2 0.20. Closed circles represent the nests deposited in the open zone, open circles represent ones in the vegetation/border zone.



PAGE 1

110 g WAVE HEIGHT1.5m 1 9 8 6 Q 1.5m > WAVE HEIGHT > .Sm [ WAVE HEIGHfI 0.5im 100 60. 00 40 20 1988 JUL. AUG NOV. % 100 so. 60 4020 1989 JUL AUG SEP OCT. NOV % 100 80 80 40 20 JUL AUG SEP. OCT. NE. MEAN (86, 88 AND 89) % 100 V/7/ i //X/ V.// 60 40 20 0 JUl. AUG. SEP. OCT. NO.



PAGE 1

Figure 30. Sex ratio of green turtle hatchlings as a function of mean incubation temperatures during the middle third of the development period. Data were collected at Tortuguero, Costa Rica during the 1986 and 1988 seasons. Nest temperatures were taken at the approximate center of each clutch, and sand temperatures were taken 1m apart from each clutch at the same depth. The superimposed line shows the fit to a logistic regression model. In Figure 29A, Y = exp(-34.45 + 1.17X)/(1+ exp(-34.45 + 1.17X)), pseudo R2 = 0.20. In Figure 29B, Y = exp(-30.37 + 1.06X)/(1+ exp(-30.37 + 1.06X)), pseudo R2 = 0.17. The small number indicates the number of overplotted data points in the graph.



PAGE 1

21 August 1988 (256 mm) and in September 1989 (157 mm) were the lowest ones since 1978. The rainfall in October of both years exceeded the average values. Consequently the 1988 and 1989 seasons showed distinctive seasonal fluctuation in rainfall. Vertical Position of Clutches A total of 344 sample nests (including both egg survivorship samples and sexing/temperature monitoring samples) during the three seasons was measured for the distance between the bottom of the egg chamber and the level of the sand surface on the beach at the time of excavation. Nest mounds above the clutches made by females were generally flattened to the level of the beach surface by the time of excavation. In addition, 43 sexing/temperature monitoring sample nests in the 1988 season were measured for the height of the clutch mass (a distance between the bottom and the top of egg mass) when 20 sub-sample eggs were taken for sexing at around 50-60 days of incubation. Bottom depth of green turtle clutches during the three years averaged 76.8 cm (SE=0.8, range=52105, n=176) in the open zone and 79.0 cm (SE=0.8, range=45-120, n=168) in the vegetation/border zone. The difference between the two zones was almost significant (t=1.93, p=0.0549). Overall bottom depths of green turtle clutches (two zones combined) averaged 77.9 cm (SE=0.6, range=45-120, n=344; Figure 7). Mean height of the clutch mass was 19.8 cm (SE=0.9, range=8-33, n=43). If the center of the clutch is assumed to be located at half of the height of the clutch (about 8.9 cm from the bottom), mean vertical



PAGE 1

28 Mean monthly sand temperatures from July through November differed among years in both zones (one-way ANOVA for each depth and zone; same results for all six tests, df=4, p<0.0001). Fisher's procedure of least significant difference identified those months where mean temperatures are significantly different from each other (Table 3). Although the seasonal trends of monthly sand temperatures varied from year to year, September's sand temperatures in both zones resulted in the highest rank within each season for all three years. Figure 15 shows the negative relationship between the amount of monthly rainfall and mean monthly sand temperatures at a depth of 60 cm on transects along the beach during the study (the open zone, Y=30.020-0.003X, R=0.718, n=15, t=3.718, p=0.003; the vegetation/border zone, Y=28.524-0.002X, R=0.665, n=1 5, t=3.21 0 p=0.007). Seasonal Nest Distribution and Nest-Site Selection By the first censuses, minor nesting activity already had occurred in all three years. At least 16 nests in 1988 and five nests in 1989 were counted at the study area before the censuses started in the middle of June. Because the census in 1986 started in July, at least a half month of nesting activity at the beginning of the season was missed. The number of unchecked nests in 1986 was not known but was probably very minor relative to the whole season. Most nesting activity occurred from July through the first half of October in all three years (Figure 16). However, the seasonal distribution shifted among the years. Peak nesting activity occurred



PAGE 1

48 (Table 4), the egg survivorship data in both zones were pooled. Since both the seasonal and zone factors significantly affected sex ratio (Table 8), each factor was treated separately. Because multiple comparison tests on both egg survivorship and sex ratio data for seasonal analysis failed to assign each half-month period to significantly different subsets, mean values on each bimonthly base were applied. Table 8 shows the value of each parameter applied and the process of estimation. Overall sex ratio was calculated to be 40.6% females in the 1988 season. Seasonal Variation and Zone Effect in 1986 In 1986, numerous rainy days and sporadic heavy rains kept sand temperatures lower than the pivotal temperature for most of the season (Figure 13). Eight of 12 sample nests produced only males (Table 6). The proportion of females in sample nests varied from 0 to 47.1% in the open zone and from 0 to 68.8% in the vegetation/border zone. Because of the small sample size and sampling bias, assessing the difference between sex ratios between the zones was not possible. To analyze the seasonal factor, data in both zones were pooled by monthly intervals. Figure 32 shows the seasonal variation of sex ratios in the 1986 season. Throughout the season, sex ratios were strongly biased toward males. There was no significant difference in the sex ratios observed throughout the season (Kruskal-Wallis test, df=2, Hc=0.983, p=0.612; because of heterogeneity of variance, nonparametric analysis was applied).



PAGE 1

120 100 2.3 1 1,2 3,4 3,4 3,4 5 4,5 1986 50 1 1 28 71 82 64<1-5-8 34 N JA•N 16-30 J 1-S15 JUL 16-31 AUG I-15 AUG 1&31 SEP I-15 SEP 1630 OCT 1-15 OCT 1631 TOTAL 1,2 2,3 5 4,5 1 2,3,4 3,4,5 2,3,4,5 N.S. o%1001988 50 N 32 60 114 285 461 394 369 246 59 2041 N 50 JMN 16-30 JL1-15 JUL 16-31 AUG 1-15 AUG 1631 SEP 1-15 SEP 1-30 OCT 1-15S OCT 1631 TOTAL PERIOD (HALF MONTH)



PAGE 1

51 showed a slightly lower fit to the logistic model (pseudo R2=0.17, Table 9) than did mean nest temperature data. Incubation period Incubation periods of the sexed sample nests varied from 51 to 74 days (mean=61.7, SE=0.7, n=52). Incubation periods were negatively correlated with mean nest temperatures (mean nest temperatures=38.3 -0.15 incubation days, R2=0.49, n=52, p<0.001, Figure 34). Therefore, female sex ratio was also negatively correlated with incubation period (Figure 35). Only males were observed in nests that had over 68 days of incubation, whereas both sexes were found in nests that had incubation periods of between 56 and 66 days. Incubation period data were the second best fitted regression model (pseudo R2=0.18, Table 9). However, the variability of sex ratio for the middle range of incubation days was so high that the observed ratios almost covered the entire range of sex ratios. Therefore, predictability of sex ratio by incubation period was poor. Rainfall The amount of rainfall recorded at each sample nest was calculated three ways for statistical testing: R-I=mean daily rainfall throughout the critical period, R-II=mean daily rainfall throughout the critical period plus the previous 10 days, R-Ill=mean daily rainfall from the deposition of eggs throughout the critical period. Rainfall decreased the sand temperature at the depth of nests. All three data sets of rainfall were negatively correlated with the mean sand temperatures (R2=0.16 (R-1), 0.33 (R-11), 0.47 (R-Ill), all



PAGE 1

82 purposes. Genetic influence and heterogenous environmental conditions are probably partially responsible for such high variability. To date, we have little knowledge of sex ratio in sea turtle populations and its dynamics. Because the character of sexual dimorphism, tail elongation, appears just prior to sexual maturity of individuals, only gonadal examination or testosterone level can identify the sex of immature turtles. Thus, previously reported sex ratio data based on external morphology (e.g, Ross, 1984) are probably not reliable. Levels of testosterone have been used to examine the sex ratios of immature green turtles in several feeding grounds (Wibbels et al, 1989; Meylan et al., 1992; Bolten et al., in press). However the definition of each population at the feeding grounds is not known. The primary sex ratio data obtained in this study for Tortuguero is the first reliable information on this subject and also is the first stage of the determination of the sex ratio of the entire population. However, we need to be cautious in assessing the primary sex ratio of the Tortuguero population. The density of nests, and thus the number of hatchlings produced each year, varies widely at Tortuguero. Because the sex ratio shows yearly variation as this study has demonstrated, the primary sex ratio of the Tortuguero green turtle population can only be calculated from the product of the number of hatchlings and the sex ratio each year for many years. To achieve this task, periodic collection of samples for sexing for many years is still essential although the procedure is labor-intensive.



PAGE 1

151 Demas, S. and S. Wachtel. 1989. Sexing the sea turtle. In S.A. Eckert, K.L. Eckert, and T.H. Richardson (compilers), Proceedings of the 9th Annual Workshop on Sea Turtle Conservation and Biology. pp.37-39. NOAA Technical Memorandum NMFS-SEFC-232. Dutton, P.H., C.P. Whitmore, and N. Mrosovsky. 1985. Masculinisation of leatherback turtle Dermochelys coriacea hatchlings from eggs incubated in styrofoam boxes. Biol. Conserv. 31:249-264. Etchberger, C.R., J.B. Phillips, M.A. Ewert, C.E. Nelson, and H.D. Prange. 1991. Effects of oxygen concentration and clutch on sex determination and physiology in red-eared slider turtles (Trachemys scripta). J. Exper. Zool. 258:394-403. Ewert, M. and C.E. Nelson. 1991. Sex determination in turtles: diverse patterns and some possible adaptive values. Copeia 1991:50-69. Fowler, L.E. 1979. Hatching success and nest predation in the green turtles, Chelonia mydas, at Tortuguero, Costa Rica. Ecology 60:946-955. Girondot, M and C. Pieau. 1990. Sex determination in the critical range of temperature for marine turtles. In T.H. Richardson, J.l. Richardson, and M. Donnelly (compilers), Proceedings of the 10th Annual Workshop on Sea Turtle Conservation and Biology. pp.77-80. NOAA Technical Memorandum NMFS-SEFC-278. Groombridge, B. and R. Luxmoore. 1989. The Green Turtle and Hawksbill (Reptilia: Cheloniidae): World Status, Exploitation and Trade. IUCN Conservation Monitoring Center. Cambridge, UK. Gutzke, W.H.N. and G.L. Paukstis. 1983. Influence of the hydric environment on sexual differentiation of turtles. J. Exper. Zool. 226:467-469. Hendrickson, J.R. 1958. The green sea turtle, Chelonia mydas (Linn.) in Malaya and Sarawak. Proc. Zool. Soc. Lond. 130:455-535.