Predicting tadpole metamorphosis


Material Information

Predicting tadpole metamorphosis ecological aspects of energy allocation and development
Physical Description:
x, 126 leaves : ill. ; 29 cm.
Hensley, Frank R., 1965-
Publication Date:


bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1994.
Includes bibliographical references (leaves 121-125).
Statement of Responsibility:
Frank R. Hensley.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002044416
notis - AKN2324
oclc - 33355057
System ID:

This item is only available as the following downloads:

Full Text







Copyright 1994


Frank R. Hensley


Sincere thanks are due to the members of my supervisory

committee, Drs. Marty Crump, Gary Meffe, Karen Bjorndal,

Larry McEdward, and Franklin Percival. Their input to this

project has been invaluable in the formative stages, in the

mid-course corrections, and in the final presentation. Each

of them has shown me their unique perspectives on ecology,

and I thank them for sharing their insights. In particular,

my coadvisors Gary Meffe and Marty Crump have provided day-

to-day instruction, assistance, and discussions that have

shaped my scientific endeavors.

Funding for this research was provided by the University

of Georgia's Savannah River Ecology Laboratory (contract DE-

AC09-76SROO-819 between the United States Department of

Energy and the University of Georgia). At SREL Gary Meffe

provided the majority of the research funds, equipment and

logistical support. I was also supported by a teaching

assistantship from the Department of Zoology. Additional

research funds were provided by a Gaige Award from the

American Society of Ichthyologists and Herpetologists, and

also from a Sigma Xi Grant-in-Aid of Research.

For assistance with field work, experiments, and

laboratory procedures I thank George Anderson, Marty Crump,

Adele Pfrimmer Hensley, Kevin Holloman, Karen Kandl, Mark

"Komo" Komoroski, John Lee, Tammy Miettunen, Roy Nagle, Susan

Rooks, David Schultz, Steve Weeks, and David Wilkins. I'd

especially like to thank David Schultz for his donation of

the DataQ0 software that saved me many hours of data entry.

The Life History Discussion Group at SREL was my

classroom for two years. Thanks are due to Karen Kandl,

Steve Weeks, Joe Pechmann, Karen Garrett, Vicki Medland,

Evelyn Gaiser, David Schultz, and the various scientists who

dropped in to chat. In addition, conversations about life

history theory, experimental design, data analysis, and

science in general with Kevin Baldwin, Joe Bernardo, Phil

Dixon, Karen Garrett, Ray Semlitsch, and Harry Tiebout helped

me over several hurdles.

Early drafts of dissertation chapters were improved by

the comments of my committee, and by Kim Babbitt, Doug Emlen,

and Adele Hensley.

Special thanks are due to Cathy Cox for handling much of

my business at the University of Florida while I was in South

Carolina. Thanks also to Kevin Baldwin and Dennis Haney for

logistical support in Gainesville.

Life in South Carolina was happy and humorous thanks to

Kevin Holloman and John Lee. Memorable field work with Joe

Pechmann, Joel Snodgrass, Mark Mills, Chris Hudson, Howard

Berna, and the whole Herp Lab made me feel like a real

biologist. The Ugly Customers provided the music. Special

recognition goes to Kevin Holloman for serving as film critic

and dive master, and to Kurt Pyle for keeping me sharp.

My parents, Roy and Nancy Hensley have never wavered in

their encouragement, love, and support. The Pfrimmer and

Puls clans rounded out my long-distance cheering sections.

Above all, my wife, Adele, endured more than I dared to ask;

her continual support was all that kept me going many times

over the years. For her love, friendship, and camaraderie, I

am forever grateful.



ACKNOWLEDGEMENTS .......................................... iii

ABSTRACT ................................................. viii


1 INTRODUCTION ........... ........................... 1


Introduction .................................... 8
Methods ........................................ 11
Results .................... ... ..................... 16
Discussion ............. ....................... 18


Introduction ........... ........................ 33
Methods ............................................ 37
Results ............ ..... ... .................... 42
Discussion ......... ........................... .. 46


Introduction ................ ....................... 65
Methods .......... ........................ ... .... 66
Results ................................ .... ... 68
Discussion ......................................... 69

BUFO TERRESTRIS TADPOLES ............................ 87

Introduction. ............................... ... 87
Methods. .......... ............. ...... ........... .. 90
Results ...... ... ................ .............. 96
Discussion. .............................................. .....***** 99


6 SUMMARY AND PROSPECTUS................................ 115

Summary ............................ ............ 115
Prospectus ........................................ 117

LITERATURE CITED ....................................... 121

BIOGRAPHICAL SKETCH ..................................... 126

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



Frank R. Hensley

December, 1994

Chairman: Martha L. Crump
Cochairman: Gary K. Meffe
Major Department: Zoology

Amphibian larvae exhibit plasticity in age and size at

metamorphosis. Ecological models of amphibian metamorphosis

might be improved by considering how tadpoles allocate energy

to storage versus growth or development.

I experimentally manipulated food availability to test

whether fat storage in tadpoles is independent of body size.

Fat reserves at metamorphosis were correlated with body mass,

but size-adjusted reserves varied significantly among feeding

treatments. Allocation to lipid storage is therefore

plastic, and warrants consideration in metamorphosis models.

In a second experiment I examined the relationship

between growth and development in Bufo terrestris tadpoles in

light of a model of metamorphosis based on changing


priorities of allocation to growth versus development.

Developmental trajectories indicated that plasticity in

development rate persists through later stages than indicated

by previous studies. Tadpoles can delay metamorphosis and

take advantage of opportunities for rapid growth. In

contrast to previous studies, loss of plasticity in age at

metamorphosis appeared to be the product of time constraints

on development rather than an actual loss of flexible

development rate. The data support the model of dynamic


The same experiment was also used to test whether lipid

storage is significantly related to developmental timing, and

how lipid storage relates to a model of proportional energy

allocation in tadpoles. Timing of metamorphosis was

positively correlated with higher size-adjusted lipid

storage. I propose an expanded model of proportional

allocation that relates lipid storage to rapid development.

Allometric effects generally dominate allocation to storage,

but stage-specific effects dominate under some conditions.

Tadpoles facultatively adjust development rate in

unpredictable habitats. I predicted that lipid storage would

be reduced when environmental conditions induced accelerated

development. This prediction was not supported in a

comparison of tadpoles from artificial ponds that dried early

versus constant-depth ponds. Mean lipid storage did not

differ between drying versus constant ponds, but tadpoles in

drying ponds stored lipid faster per unit body size. This

result is consistent with the proposed model of stage-

specific allocation to storage. Further research on the

ecology and genetics of phenotypic plasticity in complex life

cycles will help characterize the adaptive significance of

stage-specific allocation patterns.


Organisms live in a wide variety of environments that

range from very stable to those that fluctuate dramatically

and unpredictably. For example, communities around

hydrothermal vents in the ocean floor are among the most

stable, with constant temperature, no change of seasons, and

constant energy input. In contrast, many environments such

as tree-fall gaps in rainforest canopies or ephemeral rain

puddles in deserts, may be highly unpredictable in space or

time. Nonetheless, many organisms rely on such environments.

How organisms deal with environmental unpredictability is a

central theme in ecological research.

One of the defining characteristics of living organisms

is the ability to assimilate energy from the environment

(Purves and Orians 1983, Wessells and Hopson 1988). Energy

availability is one factor that often varies unpredictably in

both terrestrial and aquatic environments. Assimilated

energy must be allocated among the competing functions of

maintenance, growth, reproduction, or activity, including the

acquisition of more energy (Sibly and Calow 1986, Boggs and

Ross 1993). Assimilated energy may be allocated immediately,

or stored for future allocation. Because patterns of energy

allocation will affect an organism's survival and fecundity,

2 -

energy allocation is a fundamental component of life

histories (Stearns 1992, Perrin and Sibly 1993). Life

history research is often focused on characterizing how a

species' life history is suited to its environment, and how

life histories change in response to environmental variation.

Perrin and Sibly (1993) provided a thorough review of

current models of energy allocation, including models for

both constant and unpredictable environments. These models

begin with the assumption that selection will tend to

optimize energy allocation, thereby maximizing fitness. In

these models energy storage is considered adaptive under

certain conditions, depending on the relative benefits of

immediate versus delayed allocation. Unpredictability in the

returns from somatic and reproductive allocation may select

for energy storage (Perrin and Sibly 1993). Thus in

unpredictable environments, energy storage is likely to be


This question of how organisms deal with unpredictable

environments becomes more complex when we consider species

that exploit two or more distinct environments over the

course of a complex life cycle. Wilbur (1980) defines a

complex life cycle as one in which there is a larval stage

that undergoes a radical morphological metamorphosis

accompanied by a change in habitat. Sexual maturity is an

important life history transition for many organisms, and its

timing often varies in response to environmental variables.

Similarly, metamorphosis of species with complex life cycles

is a phenotypically plastic trait that is influenced by many

environmental factors (Wilbur and Collins 1973, Wilbur 1980).

While metamorphosis and sexual maturity often occur together,

this is not always the case in complex life cycles. For

example, in amphibians metamorphosis may occur long before

or, in some cases, after sexual maturity, and may impose

independent selective pressures on energy allocation.

Amphibians represent an excellent system for studying

how animals with complex life cycles allocate energy in

response to unpredictable environmental variation. Typically

amphibians have aquatic larvae that metamorphose into

terrestrial juveniles. Salamander larvae are carnivorous, as

are the adults, and sexual maturity in salamanders may occur

well before metamorphosis or years afterward. In contrast,

most anuran larvae are suspension-feeding herbivores or

omnivores that metamorphose into carnivorous adults. For

frogs metamorphosis is a prerequisite for sexual maturity.

Thus metamorphosis in frogs marks a radical life history

transition, a change in morphology, diet, and habitat, that

is essential for individual reproductive success.

In anurans variation in age and size at metamorphosis

has been shown to have fitness consequences. Age and size at

metamorphosis are correlated with survival to maturity

(Berven 1990) and age at first reproduction (Collins 1979,

Smith 1987, Berven 1990). Larger size at metamorphosis may

reduce the risks of mortality from predation, desiccation,

and starvation in the terrestrial habitat (Smith 1987). In

addition, size differences at metamorphosis can persist to

maturity (Smith 1987, Berven 1990) and thus affect male

mating success and female fecundity (Howard 1978, Berven


Anurans typically inhabit unpredictable environments as

larvae, and metamorphosis is an obligate life-history

transition with significant fitness consequences. So how do

tadpoles accommodate environmental variation? One factor

that must be considered is the importance of growth.

Wassersug (1975) has argued that the tadpole stage of the

anuran life cycle is adapted to exploit bursts of primary

productivity that occur when temporary ponds fill, and that

growth is the primary function in the tadpole phase. The

importance of tadpole growth in determining age and size at

metamorphosis is seen in ecological models for predicting


Wilbur and Collins (1973) proposed a model for

predicting age and size at metamorphosis which has become

central to our understanding of phenotypic plasticity in

amphibians. According to this model, the various

environmental factors that influence larval development exert

their influence by affecting tadpole growth rates. There is

a minimum body size (b) that must be attained before tadpoles

are physiologically competent to metamorphose. A tadpole's

growth rate determines how rapidly it reaches b, and thus how

soon metamorphosis is possible. Once tadpoles attain this

minimum size, however, they may metamorphose immediately or

delay metamorphosis and take advantage of opportunities for

further growth. The "decision" to metamorphose or not is

also dependent on growth rate. According to the Wilbur-

Collins model, as long as a tadpole's growth rate remains

relatively high, metamorphosis is delayed. The model

proposes a minimum threshold of mass-specific growth rate (g)

above which tadpoles delay metamorphosis; if mass-specific

growth drops below g, metamorphosis is initiated.

The Wilbur-Collins model allows predictions of how

environmental variation will influence age and size at

metamorphosis. Consider two groups of tadpoles in different

environments, where differences in habitat impose different

limits on growth rate. The slowly-growing group will reach b

later than tadpoles in a rapid-growth environment, and once

they reach b, their growth rates will fall below g more

rapidly. Thus they will metamorphose later, and smaller than

tadpoles in the high-growth environment.

In addition to comparing tadpoles with different mean

growth rates, the model can also be used to predict the

effects of changes in growth rate at different times during

the larval period. For example, a decline in growth rate

prior to attainment of b will delay metamorphosis relative to

tadpoles that do not experience the decline. In contrast, a

decline in growth rate after attainment of b will accelerate

metamorphosis. An early increase in growth rate will allow

tadpoles to reach b sooner and will result in earlier

metamorphosis. An increase in growth rate after b is

reached may result in either accelerated or delay

metamorphosis, depending on the exact values of b, g, and the

magnitude of the change in growth rate. These predictions

are presented graphically in Alford and Harris (1988) and

Hensley (1993).

This model of amphibian metamorphosis has proven to be a

robust paradigm for understanding plasticity in age and size

at metamorphosis. It emphasizes the importance of growth

rate in determining age and size at metamorphosis, which are

factors that can significantly affect individual fitness.

Growth, however, is but one function that competes for energy

assimilated from the environment. In unpredictable

environments energy storage may be an important, adaptive

allocation priority. Given that there is clearly a

relationship among growth, development, and energy storage,

how should a tadpole allocate energy in an unpredictable


I conducted three experiments to address the question of

how tadpoles allocate energy in unpredictable environments.

First, I examined the assumption that energy allocation in

tadpoles is phenotypically plastic. Most of the research on

plasticity in tadpole growth and development has emphasized

the importance of body size and growth rate. I tested

whether tadpole fat storage varies with environmental

conditions (changes in food supply) independently of body

size. Second, I tested two recent models of amphibian

plasticity (Leips and Travis 1994, Harris in press) that

emphasize energy allocation rather than growth. These models

make specific predictions about changes in age and size at

metamorphosis in response to environmental variation. Again,

I used tadpole responses to changes in food availability to

test predictions of the models. Finally, I performed an

experiment to test how fat storage is affected by

unpredictable pond duration. Previous work has demonstrated

that tadpoles can facultatively adjust development rates in

response to early pond drying. I tested whether fat storage

is affected by early pond drying as predicted by models of

energy allocation in tadpoles.



Amphibian larvae exhibit phenotypic plasticity in growth

and development rates, and thus a great deal of variation in

age and size at metamorphosis. This plasticity has received

much attention in experimental ecology (reviews in Alford in

press, Harris in press). This research has been motivated by

a desire to understand how these phenotypes respond to

environmental variation, and whether such responses are

adaptive (review in Newman 1992). Several models have been

proposed to predict age and size at metamorphosis (reviews in

Alford 1988, Hensley 1993, Harris in press). In these models

individual growth rate is a critical determinant of age and

size at metamorphosis.

The most widely applied model (Wilbur and Collins 1973)

proposes that there are minimum and maximum body size limits

on metamorphosis. Growth rate, which is determined by

various factors in the larval environment, influences how

rapidly tadpoles reach the minimum size for metamorphosis and

their tendency to accelerate or delay metamorphosis once the

threshold size has been reached. Growth rates also play an

important role in alternative models proposed by Werner

(1986) and Rowe and Ludwig (1991) that focus on population

optima of age and size at metamorphosis rather than simply

individual responses.

Although experimental work has been conducted to test

various aspects of the models (reviews in Alford 1988,

Hensley 1993, Harris in press), one factor that has not been

studied in detail is energy allocation. All organisms must

allocate assimilated energy among the competing functions of

growth, maintenance, reproduction, and storage (Sibly and

Calow 1986). Wassersug (1975) has argued that the tadpole

stage is an adaptation for exploiting transient opportunities

for rapid growth. Given that growth is a high priority for

tadpoles and is predictive of age and size at metamorphosis,

tadpoles might be expected to allocate energy accordingly.

Crump (1981) studied energy accumulation in tadpoles of

the spring peeper Pseudacris (Hvla) crucifer. In her study

tadpoles raised at low densities accumulated more total

energy and more energy per unit mass than did tadpoles raised

at high densities. Based on these results, Crump proposed

that energy accumulation may be an important determinant of

age and size at metamorphosis, and that metamorphosis may not

be possible without a minimum amount of stored fat.

Crump's study raises the basic question of how tadpoles

allocate assimilated energy between, for example, increasing

body size, and increasing energy density through fat storage.

One possibility is that energy allocation is strictly

allometric; animals follow a fixed pattern of increasing

allocation to storage as body size increases. Such a fixed

allocation rule would explain Crump's observation of higher

energy density in larger tadpoles, but phenotypic plasticity

in body size would adequately account for the differences in

energy content.

Alternatively, energy allocation may be a phenotypically

plastic trait that is part of the complex metamorphic

phenotype. If energy allocation is strictly allometric, then

it would be unnecessary to consider energy accumulation as a

possible size-independent determinant of metamorphosis. If,

on the other hand, energy allocation is a phenotypically

plastic trait independent of body size, this plasticity would

support Crump's proposal that energy accumulation is a factor

that should be considered in predicting metamorphosis.

Further, plasticity in allocation would suggest that energy

reserves at metamorphosis might influence fitness

independently of body size.

I tested whether energy allocation is phenotypically

plastic in Pseudacris crucifer tadpoles, and examined how

tadpoles allocate assimilated energy between growth and

storage in response to changing food availability. Because

genetic variation for age and size at metamorphosis has been

demonstrated in several previous studies (Travis 1980, Newman

1988a, Semlitsch et al. 1990), I measured the responses of

four separate full-sibling families in this study. Variation

among sibships in growth and development may be genetic, or

may be due to non-genetic maternal effects such as variation

in egg size or composition. Travis (1980) pointed out that

if growth is size-specific, initial hatchling sizes may have

strong effects on growth rates, and ultimately influence

metamorphic phenotypes. Kaplan (1992) found that egg size

interacted with temperature, significantly influencing embryo

development rate and hatchling survival. I therefore

measured egg size (diameter) and hatchling size (body length)

for all tadpoles used in this study.


I raised tadpoles individually from hatching to

metamorphosis on controlled food rations. For each tadpole,

age and size at metamorphosis were recorded, and lipid

storage was measured using petroleum ether extraction.

Feeding treatments were either constant high food (H),

constant low food (L), or a switch (low increased to high (I)

or high decreased to low (D)). The H and L groups are

considered control groups. Each treatment-sibship

combination was replicated twice on each of 8 laboratory

shelves that were treated as spatial blocks. Food switches

(I,D) were made after 14 days because previous work (Hensley

1993) indicated this would generate significant effects on

age and size at metamorphosis compared to unmanipulated


Pseudacris crucifer (Hylidae) is a winter-breeding

treefrog that lays its eggs in temporary ponds. I collected

11 amplectant pairs from a Carolina bay, a natural temporary

wetland, at the intersection of roads 2 and F on the Savannah

River Site in Aiken County, South Carolina, on 24 January

1993. Pairs were placed individually in plastic containers

with water and vegetation from the bay and allowed to lay

eggs. The next morning four clutches of eggs were randomly

selected for the experiment. I haphazardly selected 64 eggs

from each of the four clutches, separated them using fine

forceps, and placed each egg in a polyethylene cup (9.5 cm

tall x 9.2 cm diameter) containing 38 ml of well water. Eggs

were randomly assigned to one of the four feeding treatments.

Egg diameters were measured under a dissecting microscope

using a video camera and MorphoSys image analysis software.

For each egg I measured two approximately perpendicular

diameters and calculated a mean. During the four hours

required to measure the eggs no trend of change in egg size

within clutch or shape (elliptical eccentricity) was

detected. Additional eggs from each of the four clutches

were also analyzed for total lipid content and various lipid

classes (Komoroski, unpublished data).

I measured snout-vent length of the four females whose

eggs were chosen for the experiment, and counted the total

number of eggs laid by each female. Females were killed by

anaesthetic overdose using MS-222, preserved in formalin, and

dissected to allow counts of unlaid eggs.

Eggs hatched in 3 or 4 days. Six days after

oviposition, I measured total length (snout to tail tip) of

all hatchlings, using the video image analysis system. After

measuring lengths, I increased the volume of water in the

cups to 370 ml (7.7 cm deep), and replaced all dead tadpoles

with similarly-treated siblings (N = 18). Tadpoles were

first fed seven days after oviposition, when all tadpoles

were at stage 25 (Gosner 1960).

Food was delivered using a glass jar with a perforated

metal lid. Each shake of the jar delivered 13.3 mg (c.v. =

14.58%) of food. The diet consisted of a finely ground

mixture (1:1 by mass) of Purina rabbit chow and NutraFin

fish flakes. Tadpoles were either fed a high food level (2

shakes) or a low food level (1 shake). Tadpoles were fed on

days 7,10,14,18, and every third day thereafter. I changed

water prior to each feeding. Based on previous experience, I

anticipated that temperature differences among shelves would

exist and would contribute to differences among blocks. To

check for such a trend, I measured water temperature in a cup

near the center of each shelf on 11 haphazardly chosen days

during the experiment.

For each tadpole I recorded the day of emergence of at

least one forelimb (stage 42, Gosner 1960), and the day when

tail resorption was complete (Gosner stage 46). Size at

metamorphosis was measured as total dry mass at stage 46.

Lipid reserves of each individual were estimated using

methods modified from Reznick and Braun (1987). Whole frozen

metamorphs were dried in thrice tared, 1/2 dram glass shell

vials at 550C. Dried tadpoles were stored over CaS04

desiccant and weighed three times to the nearest 0.1 mg.

Lipids were extracted by soaking each tadpole in room-

temperature petroleum ether, which preferentially dissolves

non-polar storage lipids (triglycerides and fatty acids) (D.L

Schultz, personal communication; Hensley, unpublished). At

hourly intervals ether was pipetted off and replaced.

Previous work indicated that seven one-hour soaks was in

excess of that needed to extract the tadpoles to a constant

mass (Hensley, unpublished data). After seven extractions,

tadpoles were again oven-dried, stored over desiccant, and

weighed three times. Mean values of tare mass and mass

before and after extraction were used to calculate tadpole

total dry mass and total grams of lipid extracted. I used

this technique to extract lipids from a pooled subsample of

eggs (N = 50 80) from each female.

Statistical Analysis

Hypothesis tests for the effects of treatment and

sibship on age, size, and lipid storage at metamorphosis were

performed using multivariate analyses of variance (MANOVA)

with SuperANOVA 1.1 software (Abacus Concepts Inc. 1989).

Dependent variables were transformed to meet the assumptions

of analysis of variance; total dry mass (g) and days to

metamorphosis were loglo(x+l) transformed, and percent lipid

was arcsine(square-root(x)) transformed. Egg diameter was

significantly correlated with hatchling body length (R =

0.62) and body length had a higher variance attributable to

measurement error, so I used egg diameter rather than

hatchling body length as a covariate in the analysis.

Because egg diameters were significantly different among

sibships (F3,248 = 600.2, P < 0.001, Table 2-1), egg size

effects could not be treated as independent of other family

effects (genetic or non-genetic maternal effects such as egg

lipid content). Thus egg diameter was treated as a covariate

nested within families rather than across families. I

explored the underlying structure of the MANOVA using

univariate ANOVAs and made post hoc comparisons among

families and treatments using Scheffe's test.

To test whether lipid storage was allometric or plastic,

I used a univariate general linear model with Type I sums of

squares. This approach tests hypotheses sequentially and

thus allows for an ordered, biological interpretation of the

data. In this analysis I tested for treatment effects on

allocation to lipid storage after first removing genetic

background (sibship effects) and allometric effects (dry mass

and dry mass2). I included the quadratic term because fixed

allometric allocation might not necessarily be a linear

function. This analysis allowed me to test for plasticity in

allocation independent of plasticity in body size. In this

model size at metamorphosis was entered untransformed as a

covariate; the dependent variable (% lipid) was transformed

as before.

To quantify plasticity in responses, I calculated the

grand mean and its standard deviation for each response

within each family. Plasticity is measured as the number of

standard deviations that separate the most extreme treatment

means (Leips and Travis 1994).


A total of 193 tadpoles metamorphosed, with 10 to 14

tadpoles in each treatment-sibship category. Treatment means

of length of the larval period, dry mass at metamorphosis,

and lipid percentages are shown in Figure 2-1. In the

initial MANOVA, variation in egg diameter within sibships did

not explain a significant amount of variation in the

dependent variables. Because egg size was confounded with

sibship effects and did not explain significant variance

within sibships, it was pooled with sibship for the final


Feeding treatments, sibships, and laboratory shelves had

significant effects on the multivariate response vector

(Table 2-2). Feeding treatments significantly affected size

at metamorphosis and percent lipid, but not development rate

(Figure 2-1). The food increase treatment (I) converged with

the high food group (H) in size and lipid content,

metamorphosing larger and with more fat than tadpoles that

remained on low food (L). Likewise, the food decrease

treatment (D) converged with group (L) in size and lipid

content, but differed significantly from group H.

The responses of each sibship are plotted separately in

Figure 2-2. Sibship significantly affected length of the

larval period and percent lipid. Scheffe's tests among

sibships indicated that sibship 1 metamorphosed with

significantly less stored fat and significantly later than

the other families. There were no significant sibship-by-

treatment interactions in the experiment, indicating that all

four families responded similarly to the treatments.

Qualitatively, however, families differed in whether food

manipulations (I and D) resulted in trends toward

accelerating or delaying metamorphosis as compared to control


The significant block effect on development rate was

attributable to a consistent thermal gradient from the top

shelf to the bottom. Temperature variance among the shelves

changed over the course of the experiment, but any given

shelf was almost always warmer than lower shelves and cooler

than higher shelves throughout the experiment (Figure 2-3).

Orthogonal polynomial contrasts indicated significant linear

and quadratic trends for lower (cooler) shelves to result in

slower development.

A significant shelf-by-treatment interaction for size at

metamorphosis was detected, but the interaction of shelf-by-

family was not significant. Overall, the MANOVA model

explained 71% of the variance in development rate, 58% of the

variance in dry mass, and 53% of the variance in lipid


The ANCOVA using Type I sums of squares (Table 2-3)

indicated significant effects of genetic background (sibship)

and body size (dry mass) on allocation to lipid. After

accounting for these variables, treatment effects explained a

significant portion of the remaining variation.

Measures of plasticity for each family are shown in

Table 2-4.


The results of this study indicate that energy

allocation between growth and fat storage is a phenotypically

plastic trait that responds to conditions in the larval

environment independently of changes in body size (Table 2-

3). In a laboratory experiment Pfennig (1992) found that

higher post-metamorphic survival of Scaphiopus multiplicatus

was associated with larger fat bodies at metamorphosis.

Thus, like age and size at metamorphosis, lipid reserve at

metamorphosis is another phenotypically plastic trait that

may significantly influence fitness. Such plastic allocation

may be important for predicting metamorphosis, and may be the

result of optimized allocation that maximizes cost-benefit

ratios of competing functions.

Predicting Metamorphosis

Crump (1981) proposed that a prerequisite for successful

metamorphosis might be not only a minimum body size, but also

a minimum fat reserve, and that models for predicting

metamorphosis timing might be improved by including energy

accumulation as a predictive variable. Experiments conducted

to test models of amphibian metamorphosis have used mass or

linear dimensions as measures of growth. The data presented

here support the idea that energy accumulation is, to a

degree, independent of body mass and influenced by

environmental variables. If energy density or fat reserves

are independent of size (mass or length), perhaps a better

measure of growth would be total energy accumulated (Crump

1981, Ludwig and Rowe 1990).

Using such a measure might improve how well changes in

growth rate predict changes in the timing of development. In

this study, however, feeding treatments resulted in no

significant changes in timing of metamorphosis, unlike

previous studies (Travis 1980, Alford and Harris 1988,

Hensley 1993); mass at metamorphosis and lipid storage are

positively correlated, but not predictive of developmental

timing. Under conditions where size at metamorphosis and

developmental timing are correlated (either positively or

negatively), consideration of energy content or fat reserves

might improve how well timing of metamorphosis can be


Optimal Energy Allocation

Optimality models are widely used in the study of

plasticity of life histories (Roff 1992). Based on

expectations of optimality, and if rapid development is

advantageous, tadpoles would be predicted to optimize energy

allocation such that they would arrive at the minimum size

for metamorphosis with the requisite fat reserve (allometric

allocation), and only in subsequent growth would allocation

be plastic. Once the minimum threshold for metamorphosis is

reached, tadpoles would be expected to allocate energy

between growth and storage to maximize post-metamorphic

survival and thus maximize fitness.

An alternative to such an optimality approach to

allocation is a consideration of constraints. If maximizing

growth rates is central to the adaptive significance of the

tadpole stage (Wassersug 1975), then why do tadpoles ever

allocate energy to storage rather than to growth? Perhaps

there are other factors such as nutritional availability of

minerals that limit growth, and energy can be assimilated

faster than required for maximal growth rate. For example,

cannibalistic diets can confer growth advantages in tadpoles

when compared to very similar diets that provide as much

energy but no conspecific tissue (Crump 1990). This result

suggests that rather than energy, some nutrient limits

growth, and may be most readily available when conspecifics

are eaten. Under conditions where growth is nutrient-

limited, one might predict that over time a tadpole's energy

density would increase as the surplus of assimilated energy

outpaced growth. Studies of the nutritional ecology of

tadpoles are necessary to determine if growth is limited by

some nutrient rather than by energy accumulation, and thus

results in higher lipid storage. Such a system could produce

plastic allocation phenotypes across environments that vary

in food availability or quality. Thus, detecting apparent

plasticity in lipid storage does not necessarily indicate

that tadpoles optimize allocation on the basis of fitness

costs and benefits. If development is nutrient-limited,

energy allocation may have no influence on development rate.

Costs of Plasticity

Cost of phenotypic plasticity is an important

consideration in understanding the evolution of reaction

norms. Newman (1992) discussed the difficulty of

distinguishing between costs of plasticity across

environments and costs of individual phenotypes. Costs of

plasticity are incurred due to trade-offs between plasticity

in one trait and either the mean contribution to fitness of

another trait, or plasticity in another trait. Any single

phenotype may have associated costs that are always incurred

with the production of that phenotype, but these costs are

not true costs of plasticity. For example, if a tadpole

genotype had high plasticity in size at metamorphosis

(regardless of the range of sizes attainable) at the expense

of either plasticity in fat storage or the attain

fat levels that maximize fitness, then the plasticity could

be said to have a cost over and above the pure costs of small

body size. Newman (1988a) found a cost of plasticity in

spadefoot (Scaphiopus couchii) tadpoles; among families,

higher plasticity in growth (not simply larger size at

metamorphosis) was associated with slower development.

I measured plasticity of P. crucifer families across the

four feeding treatments in units of standard deviations of

the family grand mean (Table 2-4) (Leips and Travis 1994).

Families with higher plasticity in growth show slower, less

plastic rates of development, just as they did in Newman's

(1988a) study. In addition I found that greater plasticity

in lipid storage was associated with slower, less plastic

development and larger, more plastic size at metamorphosis.

Newman's study was limited to just 5 sibships (Newman 1988a),

and my data represent just four, making all conclusions

tentative. If such a trade-off between plasticity in growth

and plasticity in development rate is real, and not an

artifact of small sample sizes, Newman (1992) suggested that

this trade-off is probably the result of the functional

relationship between growth and development. Previous

workers have suggested a cause-effect relationship between

growth and development in tadpoles, but have differed on

which of the two is causal of the other (Wilbur and Collins

1973, Smith-Gill and Berven 1979, Stearns and Koella 1986).

The details of the relationship between changes in growth

rates and changes in development rates have only recently

been explored (Hensley 1993, Leips and Travis 1994). Further

research is necessary to determine if there is a

physiological basis for a trade-off between plasticity in

growth and plasticity in development.


Leips and Travis (1994) proposed a model of dynamic

allocation of energy in tadpoles. According to this model,

during early developmental stages energy is primarily

allocated toward development, and growth is a lesser

priority. Changes in energy income in this phase of larval

ontogeny primarily affect timing of metamorphosis, but also

affect growth. Beyond some boundary, however, allocation

patterns change, and fluctuations in energy cease to affect

development rates but strongly affect size at metamorphosis.

In the present study food level manipulations were

expected to affect both size at metamorphosis and timing of

metamorphosis, but no effect on timing was realized. Feeding

groups L and D did not suffer delayed metamorphosis, but

metamorphosed with lower levels of stored fat than did

animals that had high food availability (I, H) in later

development. This result may suggest that tadpoles raised on

low food allocated energy preferentially to maintain rapid

development and never fell behind. An alternative is that

family effects and block (temperature) effects simply swamped

any treatment effects on development rate, and thus obscured

any underlying relationship between energy allocation and

rapid development. Further studies of the details of growth,

development, and energy allocation are necessary to determine

whether the lack of a developmental response was actually due

to dynamic allocation or was simply the result of high

within-family variance in development rates.

Growth of the food increase group (I), however, suggests

that tadpoles are able to compensate for low food in early

development, without significant delays in metamorphosis or

reduction in lipid storage. One possible explanation is that

low food levels in early larval stages can condition tadpoles

to be more efficient at energy accumulation, and thus makes

them better able to take advantage of the increase in food

resources and "catch up" to the high food group.

Alternatively, these tadpoles may be paying some cost of

reduced early growth that is simply undetected.

Unfortunately, both of these arguments assume that the first

two weeks of growth resulted in differences between the high

and low food groups, but in this experiment I did not measure

tadpoles prior to the food switch. Possibly food was not a

limiting factor during the first two weeks of development,

even on the low food ration. Data from a previous study on

this species (Hensley 1993), however, suggest that

significant differences in body size between the low and high

food rations should have been attained by the fourteenth day

of the experiment. The ability of tadpoles to change energy

allocation to compensate for reduced growth and development

when resource levels increase warrants further study.

The results of this study indicate that energy

allocation is a phenotypically plastic trait in tadpoles.

Full sibling families differed in the degree of plasticity in

growth, development, and lipid storage. My data support

Newman's suggestion that plasticity in development may be

traded against plasticity in growth (and energy storage). As

proposed by Crump (1981), energy accumulation is an important

component of the larval period for tadpoles. Models for

predicting metamorphosis may be improved by considering

energy accumulation and allocation, but more detailed

research on the relationship between growth and development

is necessary for a clear picture of the role of energy

accumulation and allocation. The ability of tadpoles to

compensate for early limitations on growth and development

and the potential costs of compensatory growth and

development also warrant further exploration. Genetic trade-

offs between rapid growth and developmental plasticity are

also suggested as elements that must be considered for a full

understanding of the adaptive significance of plasticity in


Table 2-1. Characteristics of female P. crucifer and their
egg clutches.

SSL egg dia (mm) egg eggs total % lipid
sibship (mm) 3i s.d. dry mass laid clutch in eggs
1 32.5 1.20 0.023 0.51 350 926 84.6
2 28.5 1.04 0.020 0.34 446 800 88.0
3 32.0 1.14 0.022 0.45 277 1056 93.3
4 33.8 1.14 0.021 0.43 253 1270 81.6

Table 2-2. Results of MANOVA and univariate ANOVAs for age
at metamorphosis, dry mass at metamorphosis, and lipid
reserves for Pseudacris crucifer tadpoles. Wilks' K is the
multivariate test statistic. Probability values < 0.05 are
indicated with an asterisk (*). Coefficients of
determination (r2) indicate the proportion of total variance
explained by each factor.

Source Wilks' K F df P
treatment .5379 9.891 9, 306.80 0.0001*
sibship .6960 5.475 9, 306.80 0.0001*
shelf .3465 7.705 21, 362.35 10.0001*
treatment x sibship .8318 0.888 27, 368.63 0.6292
sibship x shelf .6799 0.832 63, 376.95 10.8122
treatment x shelf .5609 1.279 63, 376.95 10.0869

period dry mass % lipid
Source P P P
treatment .5584 .0001* .0001*
sibship .0001* .1958 .0086*
shelf .0001* .3911 .9778
treatment x sibship .5838 .9343 .2017
sibship x shelf .2015 .9556 .7102
treatment x shelf .4152 .0096* .1000
coefficient of
determination (r2) .7144 .5826 .5347

Table 2-3. Analysis of covariance for percent lipid in
Pseudacris crucifer tadpoles. Type I sums of squares test
sequential hypotheses explained in text.

Source df Type I SS MS F P
sibship 3 .0121 .0040 8.12 .0001*
dry mass 1 .0696 .0696 139.67 .0001*
(dry mass)2 1 .0010 .0010 1.94 .1653
treatment 3 .0066 .0022 4.43 .0049*
residual i188 .0936 .0005 _

Table 2-4. Mean responses and phenotypic plasticity in
larval period, size at metamorphosis, and lipid storage in P.
crucifer tadpoles. Plasticity is the maximum difference in
treatment means (in units of standard deviation of the family

larval period
(days) I

dry mass

(% dry mass)

sibship last. plast. plast.
1 41.60 0.3200i .0324 1.4928 9.67 1.5559
2 39.27 0.6857 .0305 1.3654 10.55 1.2270
3 39.00 0.8716i .0314 1.3182 10.74 0.9842
4 39.75 0.5864 .0328 1.3538 10.49 1.0340

O H 11.38%
0.050- D 9.42%
0.045 A I 11.21%
0 L 9.61%
--, 0.040-

S 0.035-
2 0.030-




0 .0 1 0 .
25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55


Figure 2-1. Treatment means for age and dry mass at
metamorphosis for P. crucifer tadpoles. Mean lipid
percentages are shown for each treatment. Control groups
were raised on constant high (H) or low (L) food levels.
Food increases (I) and decreases (D) were made after 14 days
on initial food levels. For clarity only one side of the 95%
confidence intervals for size at metamorphosis are shown.

Family 1
OH 11.26
ED 8.12
Al 10.84
OL 8.28

35 37 39
35 37 39

I1 43 37 39 41
41 43 37 39 41

37 39

Family 4
OH 11.24
ED 10.01
Al 11.27
OL 9.52


Figure 2-2. Mean responses to the four feeding treatments
for each sibship with mean lipid percentage given in each
legend. All plots are made to the same scale.













top shelf

bottom shelf

0 10 20 30 40

Day of Experiment

Figure 2-3. Water temperature measurements for each shelf,
measured 11 times during the experiment.



Phenotypic plasticity of growth and development in

amphibian larvae has received much attention in experimental

ecology over the last two decades. This research has focused

on two central questions. First, given the wide ranges of

sizes and ages at which larvae metamorphose within

populations, can we predict when metamorphosis will occur for

individual larvae (reviews in Alford and Harris 1988, Hensley

1993)? Second, is phenotypic plasticity in amphibian larvae

adaptive; that is, does plasticity in the timing of

metamorphosis result in greater fitness across a range of

environments than could be attained by fixed timing (reviewed

by Newman 1992)? Central to answering both of these

questions is an understanding of the functional relationship

between growth and development. Previous work has treated

growth as a prerequisite of developmental change (Wilbur and

Collins 1973, Stearns and Koella 1986), but the converse has

also been argued (Smith-Gill and Berven 1979).

Wilbur and Collins (1973) developed a model to predict

how growth and development rates of amphibian larvae are

related, and how changes in growth rate influence age and

size at metamorphosis. This model has become central to our

understanding of phenotypic plasticity in complex life cycles

(life cycles that include a larval stage followed by a

radical metamorphosis and usually a habitat change (Wilbur


The Wilbur-Collins model proposes that amphibian larvae

must reach a minimum size threshold (b) before successful

metamorphosis is possible. Above that size, metamorphosis is

initiated when the mass-specific growth rate falls below a

threshold level (g). Based on these two principles, we can

predict how development rates will be affected when changing

conditions in the larval environment induce changes in growth

rate. In the early stages of development when tadpoles have

not yet reached the size threshold (b), increased growth rate

will allow tadpoles to reach b earlier and thus result in

earlier metamorphosis, but decreased growth rate will delay

metamorphosis. Once tadpoles have crossed the size

threshold, however, the opposite developmental response is

predicted: increased growth rate will delay the time when

mass-specific growth falls below g, and thus will delay

metamorphosis. Above size b, a decrease in growth rate

should accelerate metamorphosis.

Experimental tests have generally supported the Wilbur-

Collins model (Alford and Harris 1988, Hensley 1993, Leips

and Travis 1994, Tejedo and Reques 1994), but raise questions

about the details of the correlations between growth and

development. For example, Hensley (1993) found that

development rates become fixed in the latter third of the

larval period and do not respond to subsequent changes in

growth rate. Leips and Travis's (1994) study confirms this

observation, and they proposed that during larval development

tadpoles allocate energy to either growth or development, but

the pattern of allocation changes over time.

According to their model (Leips and Travis 1994), growth

and development are competing functions, and energy allocated

to one function represents a trade-off with allocation to the

other. During development, however, a tadpole's allocation

priorities change. In the early stages of larval life

allocation of energy to development takes precedence over

allocation to growth, but the importance of allocation to

development declines over time. Eventually, at the point

when developmental timing becomes fixed (Hensley 1993), all

excess energy is allocated to growth. Leips and Travis's

(1994) model predicts asymmetric responses to changes in food

supply. Early in development a decrease in food supply is

predicted to affect growth first, because development rate is

maintained as a high allocation priority. For tadpoles on

low food, an early increase in food will first accelerate

development rate, and only after development rate is

maximized will growth be accelerated.

At later larval stages, however, allocation to growth

becomes the higher priority, developmental responses to

changing food supply are reduced, and eventually development

rate ceases to respond to changes in food supply (Hensley

1993). According to Leips and Travis's (1994) model, later

increases in food result in larger size at metamorphosis

because development places less demands on energy later in

the larval period. This prediction that later food increases

result in greater size increases is contrary to the Wilbur-

Collins model.

In this model of dynamic allocation (Leips and Travis

1994), growth and development are competing (i.e., negatively

correlated) functions; the strong positive correlation

generally seen between growth and development may be due to

both functions being highly correlated with age (Bernardo

1993). Comparisons among this model, the Wilbur-Collins

model, and experimental studies (Alford and Harris 1988,

Hensley 1993, Tejedo and Reques 1994), raise questions about

the details of this ontogenetic change in priorities.

At what stage does this change in priorities occur?

Hensley (1993) proposed that by stage 35-37 (Gosner 1960)

tadpole development rates do not respond to changes in growth

rate. Whether this point in development marks an abrupt

shift in energy allocation priorities or the end of a gradual

transition is unclear. Further, how much plasticity in

development is seen in the early and middle stages of larval

ontogeny? Are development rates particularly sensitive at

certain developmental stages, and less sensitive at other


Answers to these questions require data on the ontogeny

of individual larvae raised under a variety of constant and

changing conditions. Yet, to date, only a single study has

monitored the development of individual tadpoles in an

experimental setting (Smith-Gill and Berven 1979). That

study, however, did not examine the effects of changing food

supply on the timing of metamorphosis. Other experimental

studies have focused on growth trajectories and timing of

metamorphosis, ignoring the details of developmental


I performed an experiment to examine the details of

developmental response in individual Bufo terrestris tadpoles

reared in variable growth environments. The goal of this

experiment was to characterize the relationship between

growth and development, and to ascertain whether the dynamics

of this relationship effectively predict individual

differences in the timing of metamorphosis.


This experiment was designed to generate variation in

growth trajectories of tadpoles and to make comparisons among

tadpoles in the developmental responses to changes in growth

rate. Tadpoles were raised individually on controlled food

rations from hatching to metamorphosis. Some tadpoles were

raised on constant food levels (control treatments), while

others experienced either increases or decreases in food

availability during the experiment. For each tadpole, I

recorded wet mass and developmental stage seven times during

the larval period. Because genetic variation in larval

growth and development has been demonstrated for many

amphibians (Travis 1980, Newman 1988b, Semlitsch et al. 1990)

I used two separate full-sibling families in this experiment.

This experiment tested the responses of tadpoles to

eight different feeding regimes. I established two feeding

levels (high and low) used throughout this experiment.

Feeding treatments included two constant food level

treatments (controls, H and L) raised on the high and low

feeding rates, respectively. In addition six variable food

treatments were switched between high and low feeding rates.

Treatments II, 12, and 13 experienced increased food levels,

switching from low to high food on one of three different

days (12, 18, 24) during the experiment. The reciprocal food

decreases were made on the same schedule, designated as

treatments Dl, D2, and D3.

Bufo terrestris (Bufonidae) breeds in both temporary and

permanent ponds from spring to fall, and thus experiences a

variety of unpredictable environments in nature. On 21 April

1992 I collected two amplectant pairs of toads from a

roadside ditch on the Savannah River Site, Aiken County,

South Carolina. Each pair was housed in a plastic container

with shallow water and loose vegetation from the collection

site. Both pairs laid eggs, but hatching success was much

higher in sibship 1 than in sibship 2. On 28 April I

haphazardly selected tadpoles from each clutch and randomly

assigned them to treatments.

Tadpoles (N = 256) were raised individually in plastic

cups on 8 laboratory shelves, which were treated as spatial

blocks. Each treatment-sibship combination was replicated

twice per shelf, and cups were randomly assigned positions on

shelves. Cups were 9.5 cm tall x 9.2 cm diameter and were

filled with 370 ml of well water. Water was changed prior to

each feeding. Tadpoles were fed every third day. Food was

delivered from a glass jar with a perforated metal lid. Each

shake of the jar delivered 13.3 mg (c.v. = 14.6%) of food.

The diet consisted of a finely ground mixture (1:1 by mass)

of Purina rabbit chow and NutraFin fish flakes. Tadpoles

were either fed a high food level (2 shakes per feeding) or a

low food level (1 shake per feeding).

Throughout this experiment the growth and development of

individual tadpoles were monitored. Logistics of weighing

and determining developmental stages necessitated dividing

the experiment into two temporally staggered halves.

Tadpoles on odd numbered shelves were always weighed, staged,

and fed one day earlier than tadpoles on even numbered

shelves. Tadpoles were first fed on 28 April (designated day

0). Tadpoles on odd numbered shelves were next fed on day 3;

tadpoles on even shelves were fed on day 4, and this 1-day

difference was maintained throughout the experiment.

Henceforth all references to treatment and data collection

schedules are for the odd shelf regime, but even shelves were

always treated identically, one day later. Food increases

(II, 12, and 13) were made on days 12, 18, and 24,

respectively. Food decreases (D1, D2, D3) were made on this

same schedule.

Tadpoles were weighed and developmental stages

determined on days 9,12,18,24,30. All growth and

developmental data were pooled for each pair of days (e.g.

wet masses and stages measured on days 9 (odd shelves) and 10

(even shelves) were treated as simultaneous measurements).

To weigh tadpoles I blotted them with a moist paper towel to

remove excess surface water and weighed them to the nearest

0.1 mg in a tared beaker of water. Gosner (1960)

developmental stages were determined by examining each

tadpole in a transparent vial of water under a dissecting

microscope. The vial was held horizontally and rotated about

its long axis until the tadpole's hind limbs became visible.

Vial size was chosen to limit tadpole movement, and thus

depended on tadpole size.

For each tadpole, date and mass at forelimb emergence

(stage 42) and at tail resorption (stage 46) were recorded.

Statistical Analyses

All statistical analyses for this experiment were

performed with SuperANOVA 1.1 software (Abacus Concepts Inc.

1989). I used two repeated measures analyses of variance

(ANOVAs) to test for treatment and sibship effects on growth

and developmental trajectories. Mass measurements were

loglo(x+1) transformed to meet the assumption of homogeneous

variances. Growth trajectories included measurements of wet

mass on days 9, 12, 18, 24, 30 and wet mass at Gosner stages

42 (fore limb emergence) and 46 (complete tail resorption).

Using mass at stages 42 and 46 as the final two growth

observations standardizes the trajectories for differences in

development rate.

Developmental trajectories included Gosner stage

observations on days 9, 12, 18, 24, 30, plus the number of

days to reach Gosner stages 42 and 46. Using time to Gosner

stages 42 and 46 as the endpoints of these trajectories,

rather than a uniform date of observation, means that the

entire larval period of each tadpole was included in the

analysis, regardless of length of the larval period.

These ANOVAs were followed by a priori planned contrasts

(df = 1) between each control (H, L) and each manipulated

treatment (D1-3, 11-3) to test for differences in mass

averaged over time and developmental stages averaged over

time. The interaction of each of these contrasts (control

vs. experimental) with time was also calculated to test

whether the differences between control and treatments were

consistent over time. These interaction tests provide a

quantitative assessment of variation in trajectory shape.

Repeated measures ANOVAs provide powerful, formal tests

for overall effects of feeding treatment and sibship effects

on growth and development, but do not test for differences in

mass or developmental stage on each day of observation. To

test for such differences, I analyzed the same growth and

developmental trajectories using two multivariate analyses of

variance (MANOVAs) followed by Dunnett's tests to compare

both controls to manipulated treatments on each day of

observation. For each day's comparison the experimentwise

error rate is a = 0.05, but due to the number of tests being

made (7 for growth, 7 for development) interpretation is

limited to description of trends over the course of the



In this experiment 200 of the original 256 tadpoles

metamorphosed. Survivorship in each treatment ranged from 19

to 29 of the original 32 tadpoles, with the lowest

survivorship in both families occurring in the D1 treatment.

Sibship 1 had 105 (82%) survivors and sibship 2 had 95 (74%).

Growth Trajectories

Figures 3-1 and 3-2 show growth trajectories and mean

metamorphic responses for the eight feeding treatments. The

repeated measures ANOVA (Table 3-1) revealed highly

significant effects of feeding treatments, sibships, and

blocks (shelves) on growth trajectories. Sibships differed

significantly in overall growth, but not in growth trajectory

shape, as indicated by a lack of significant time x sibship

interactions. Block x time and treatment x time interactions

indicate that growth trajectory shapes differed across

treatments, as intended, and across blocks.

Pairwise contrasts of growth between all control-

treatment pairs indicate that food switches significantly

affected growth (Table 3-2). Contrast interactions with time

indicate that except for L vs. D1 and H vs. II, all

experimental growth trajectories had significantly different

shapes from both controls (Table 3-2).

Dunnett's tests for significant differences on each day

of observation suggest that prior to the food increases,

treatments 11-13 were not different in mass from the control

L (Table 3-3). In each case, however, by the next weighing

(six days after the food increase), the manipulated

treatments were significantly larger than the low food

control. In early growth these tadpoles were significantly

smaller than the high food control, but by Gosner stage 42

this difference in size had been made up by the first and

second food increase treatments (11-2). Throughout

development the third increase treatment (13) remained

significantly smaller than the high food control.

Prior to the food manipulations, none of the decreased

treatments (D1-D3) differed from the high food control (H),

according to Dunnett's tests. By six days after the food

decrease, each treatment was significantly smaller than H,

and remained so through metamorphosis. By day 12 of the

experiment, treatments D1-3 were significantly larger than

the low food control, L. After the food decreases, however,

all three treatments grew slowly and the low food control

caught up, resulting in no significant differences in size at

metamorphosis (stage 42).

Developmental Trajectories

Figures 3-3 and 3-4 show developmental trajectories for

the eight feeding treatments. Repeated measures ANOVA

revealed significant effects of treatments, sibships, and

blocks on developmental trajectories (Table 3-4).

Interactions with time revealed that trajectory shapes were

significantly influenced by both sibship and feeding

treatment, but not by the block (shelf) effect.

Contrasts revealed that the H and L controls did not

develop at significantly different rates, nor did their

developmental trajectories differ in shape (Table 3-5). In

general, development rates of manipulated treatments were

very similar to controls (Table 3-5), with only the earliest

decrease (Dl) significantly affecting developmental timing

compared to the controls. Time interaction contrasts (Table

3-5) showed that the developmental trajectory of D1 took a

different shape from that of the high food control. A

difference in trajectory shapes was seen in the comparisons

of D3, II, and 12 to L, but these treatments did not differ

in overall developmental timing.

Results of the Dunnett's tests for day by day

differences in development suggest that prior to food

manipulations the food increase treatments did not differ

from the low food control (Table 3-6). The first food

increase treatment, II, accelerated development in response

to the change in food supply, but did not maintain this

advantage, and the L treatment caught up. This result is

consistent with the significant contrast for trajectory shape

between these L and II (Table 3-5). Dunnett's test indicated

that treatments 12 and 13 initially fell behind the H control

(Table 3-6), then caught up by metamorphosis. Overall these

differences were not significant according to pairwise

contrasts (Table 3-5).

For the food decrease treatments, Dunnett's test is not

as helpful in interpretation. Although L and D1 differed

significantly in overall development (Table 3-5), Dunnett's

test does not indicate any significant day-to-day differences

(Table 3-6). Significant differences between D2 and L

indicated by Dunnett's tests are not reflected in the

contrasts (Table 3-5). They suggest, however, that D2 was

developing more rapidly than L prior to the decrease in food,

but lost this advantage within 12 days. The developmental

advantage was present for at least 12 days, but was slight

with respect to overall developmental timing and trajectory

shape. Significant differences between L and D3 on days 18-

30 (Table 3-6) contributed to differences in trajectory shape

(Table 3-5). Dunnett's test indicated only a single

significant difference between the H control and a food

decrease; D1 was different on day 18. The contrasts,

however, indicated that these two treatments developed at

marginally different rates (P = .0515) and had extremely

different trajectory shapes.


This experiment explores the details of the relationship

between developmental trajectories and growth trajectories

and tests the model of dynamic allocation (Leips and Travis

1994). According to the dynamic allocation model, early

manipulations of food supply should affect development rates

more strongly than later manipulations. In Bufo terrestris

experimental manipulations of food supply revealed that

earlier changes in food supply had larger effects on

developmental timing. Later food increases were predicted to

result in greater size increases (Leips and Travis 1994), but

this prediction was not supported. Results with B.

terrestris thus provide some limited support for the

hypothesis of changing allocation priorities.

A comparison between the H and L controls illuminates

the changes in allocation priorities. Although body sizes of

H and L diverged by day 9 of the experiment and remained

different through metamorphosis (Table 3-3), developmental

trajectories were not significantly different overall (Table

3-5). During days 18-30, however, there is evidence that the

H control tadpoles were at significantly more advanced stages

of development than L tadpoles (Table 3-6), but this

difference did not persist through metamorphosis. These

subtle differences in trajectories can be interpreted in

terms of changing developmental priorities.

Prior to day 18, food did not limit development, and the

two controls (L and H) followed similar developmental

trajectories, even though there was enough excess food to

allow H to grow much faster than L. This similarity in

developmental stages suggests that both groups were

maximizing development rate. From day 18-30 H tadpoles used

their additional resources to develop significantly faster

than L, indicating that development was a high priority

during this interval, and that food became a limiting factor

for development of L tadpoles. During this time period, L

tadpoles grew, indicating that the reduced development

compared to H was balanced against allocation to growth.

Between day 30 and metamorphosis the developmental difference

between the controls disappeared. Either development became

less of a priority for H, or the L control underwent a

compensatory acceleration of development, or both.

A compensatory acceleration of development by the L

control is not consistent with the hypothesis that

development becomes a progressively lower priority in later

stages. Apparently H tadpoles did not maximize development,

but delayed metamorphosis (compared to the possible

trajectories of such groups as II, 12 and D3) and took

advantage of the opportunity for greater growth. This

sacrifice of rapid development in favor of greater growth

supports the model of dynamic allocation and its hypothesis

that allocation to development declines in later stages in

favor of allocation to growth.

A second line of evidence supporting dynamic allocation

is seen in the developmental responses of the treatments that

experienced decreased food supplies. The Wilbur-Collins

(1973) model predicts that declining food availability early

in development, (i.e., before tadpoles have reached the

minimum body size (b)), will delay metamorphosis. Food

decreases that occur after tadpoles are larger than b will

stimulate accelerated development and earlier metamorphosis.

Such a switch is consistent with the hypothesis that

development is always the first allocation priority, but is

constrained by a threshold body size (Wilbur and Collins

1973) or a minimum energy reserve (Crump 1981). In contrast,

the model of dynamic allocation suggests that beyond a

certain point in development excess energy is allocated

preferentially to growth, and declines in food supply will

not affect developmental timing.

In this experiment, D1 metamorphosed significantly later

than the H control, as predicted by both models. Treatments

D2 and D3 did not differ significantly from the H control, as

predicted by dynamic allocation. The deviations of D2 and D3

from H are in opposite directions, as predicted by the

Wilbur-Collins model. This pattern suggests that tadpoles

reached the minimum size/energy threshold somewhere between

stage 34 (approximately when D2 was switched, Figure 3-3) and

stage 37 (approximately when D3 was switched). The lack of a

significant treatment effect in the latter stages coupled

with the predicted trend in treatment responses supports the

idea that the dynamic allocation model is complementary to

the Wilbur-Collins model (Leips and Travis 1994).

The importance of early growth history to later

developmental responses is evident in comparisons of the

control treatments and in the responses to food increases.

The loss of significant differences in developmental stage

between L and H (Dunnett's tests, Table 3-6) may indicate one

of three possibilities. First, the L control may have

increased allocation to development after day 30 (= stage 34)

and caught up with H. Second, L and H may both have had

fixed development rates (Hensley 1993) beyond day 30, but the

low food level may have set the development rate of L higher,

resulting in convergence. Third, H may have delayed

metamorphosis in favor of greater growth, and thus allowed L

to catch up. Of these three alternatives, the first is least

likely based on the dynamic model of allocation, since

allocation to development is predicted to decline over

developmental time. All three explanations, however, may be

accommodated by the model, and all suggest that early growth

rates may influence how allocation priorities change over


The importance of early growth to later responses is

also evident in the effects of food increases. Leips and

Travis (1994) stated that the progressive subordination of

development to growth as an allocation priority would result

in later food increases causing larger increases in size at

metamorphosis. Previous studies with temporary-pond breeding

species have found the opposite trend; later food increases

resulted in smaller size increases (Alford and Harris 1988,

Hensley 1990, Hensley 1993, this study). These two patterns

may reflect adaptation to temporary versus permanent ponds.

For temporary-pond breeders it is possible that the

cumulative effects of low food supply result in greater

allocation to development in the later stages of the larval

period. Late food increases may serve to promote development

to a greater degree in species where the risk of desiccation

is high, and thus result in smaller size at metamorphosis

than early increases. Leips and Travis (1994) interpreted

the ability to take advantage of growth opportunities that

arise late in the larval period as adaptive for permanent

ponds, but maladaptive for temporary ponds.

Leips and Travis (1994) studied Hyla cinerea, a

permanent-pond breeder, and H. qratiosa, a closely-related

temporary pond breeder. They detected the trend for

progressively later food increases to generate progressively

larger size at metamorphosis in both species, but the trend

was present only at 250C, not at 310C. They attributed this

trend to adaptation for permanent ponds, speculating that H.

qratiosa and H. cinerea had a common ancestor that bred in

permanent ponds. An alternative explanation is that this

trend is general for tadpoles growing at lower temperatures,

and it was not seen in other studies simply because higher

temperatures affected how allocation patterns changed during

development. Leips and Travis's (1994) study is the first to

examine the interaction of temperature and changes in food

supply. They found significant interactions between
temperature and feeding treatments on development rates of H.

cinerea, but not in H. qratiosa. They found no significant

interactions of temperature and feeding treatment on size at

metamorphosis. Further factorial experiments that

incorporate changes in food supply at various temperatures

will be necessary to determine the extent to which this

aspect of breeding habitat influences how allocation dynamics

change over time.

Hensley (1993) concluded that development rates of

Pseudacris crucifer tadpoles became fixed at approximately

Gosner stage 35-37, based on a lack of significant

differences in timing of metamorphosis between late

experimental food switches and controls. The same approach

would suggest that for Bufo terrestris the fixation of

development occurred by stage 34, since later manipulations

did not result in significantly different timing of

metamorphosis (Figures 3-3 and 3-4). Analysis of maturation

phenotypes without regard to trajectories that lead to them

may obscure dynamic changes during ontogeny (Bernardo 1993).

Changes in growth trajectory shape after stage 34 (L vs. H)

and direction of response (D2 vs. D3) clearly show that

plasticity persists later than would be indicated by analysis

of trajectory endpoints alone.

According to the dynamic allocation model (Leips and

Travis 1994), early food manipulations should result in

greater changes in development rate than later manipulations.

An examination of Figure 3-4, however, suggests that the

changes in development rates (trajectory slopes) were very

similar in magnitude across the entire experiment. I suggest

that changes in food availability may have similar effects on

development rate at any time during larval development, but

that earlier changes simply have more time for cumulative

effects to be manifested. If this is the case, then the

patterns of age and size at metamorphosis (Hensley 1993,

Leips and Travis 1994, this study) may be largely due to

fixed patterns of developmental timing rather than to changes

in energy allocation priorities.

An additional concern in application of the dynamic

allocation model is its assumption that growth and

development are functions that compete for allocation of

assimilated energy. While metabolic rates are definitely

influenced by body size and temperature, there is no direct

evidence that more rapid development is more metabolically

demanding than slower development. Feder (1982) measured

oxygen consumption of tadpoles and found that developmental

stage accounted for only 2% of the variation in metabolic

rate in Bufo woodhousei, and never more than 7% in other

species. Because growth and development rates of tadpoles

tend to be correlated (Wilbur and Collins 1973, Alford and

Harris 1988, Hensley 1993), rapid development may be

associated with larger body size, and thus greater metabolic

efficiency. Energy allocated to growth, therefore, may

actually result in reduced developmental costs. The actual

energy costs of rapid versus slower development, adjusted for

body size, must be measured in order to test this assumption

of the dynamic allocation model.

The results of the present study confirm that even when

developmental timing is not significantly influenced by

environmental conditions, there is a dynamic relationship

between growth and development. Development is significantly

influenced by changes in food supply early in the larval

period (prior to stage 34 in this study). Later changes in

growth rate may not significantly affect timing of

metamorphosis, but the interaction of growth and development

continues to change until at least stage 37. These results

suggest that the model of dynamic allocation and the Wilbur-

Collins model are complementary and predict how development

rates change in response to changing growth rates. Further

research is required to ascertain whether specific

relationships between growth and development are adaptive in


permanent, predictable environments versus temporary,

unpredictable environments.

Table 3-1. Repeated Measures ANOVA for growth trajectories.
Growth measurements are explained in the text.

df MS

P term

1 feeding treatment 7 .0124 29.44 .0001 4
2 sibship 1 .0253 59.97 .0001 4
3 block 7 .0019 4.59 .0001 4
4 tadpole(treatment) 183 .0004_
5 time 6 .0842 1147.40 .0001 9
6 time x treatment i 42 .0024 32.50 .0001 9
7 time x sibship 6 .0001 0.56 .7671 **** it 9
8 time x block I 42 .0003 4.34 .0001 9 1
...9 time x tadole(troea.ment) 1 8 ............... .................001 .......... ................
9 time x tadpole(treatment) 1 1098 .0001


Table 3-2. Probabilities for contrasts in total growth and
growth trajectory shape. The left side of the table shows
probabilities for the hypothesis of no difference in total
growth between treatments. The right side shows
probabilities for the hypothesis of no interaction of growth
with time for each contrast, a test of differences in
trajectory shape. The top row shows the contrast between the
two controls; lower rows show contrasts between experimental
treatments and the controls. Probabilities deemed
significant (< 0.05) are indicated by *.

total growth I trajectory shape

L vs. H



vs. L vs. H vs. L vs. H
11 .0001* .0190* .0001* .4290
12 .0001* .0001* .0001* .0001*
13 .0001* .0001* .0001* .0001*
D1 .0070* .0001* .6923 .0001*
D2 .0001* .0001* .0001* .0001*
D3 .0001* .0001* .0001* .0001*

Table 3-3. Results of Dunnett's tests for body size
differences between the two controls, and between
experimental treatments and controls. Observations to the
right of the bold line were made after the food level
changes. Comparisons marked with are statistically
significant (a = 0.05). Observation days shown are for odd
numbered shelves, but data are pooled for pairs of
observation days (see text).



DAY 9 DAY 12 DAY 18 DAY 24 DAY 30 42 46
H vs L *

L vs I *
L..... V S .....i ......................... ......... .*............. .......... ......... ...............*........... .......... *...........

H VS 12 I i *
L vs 13 i

H vs D i *

L vs D2 *
. ... 3.. ... .. .* *
L s D I i *

H vs D2 ... *
H vs D3 *

Table 3-4. Repeated Measures ANOVA for developmental
trajectories. Observations of development are explained in
the text.


P term

1 feeding treatment 7 179.77 3.74 .0008 4
2 sibship 1 416.39 8.71 .0036 4
3 block 7 108.45 2.26 .0308 4
4 tadpole(treatment) 183 47.79
5 time 6 10644.88 348.04 .0001 9
6 time x treatment 42 100.94 3.30 .0001 9
7 time x sibship 6 253.83 1 8.30 .0001 9
8 time x block 42 i 29.98 0.98 .5083 9
.9 time x tadpole(treatm.ent) 1098 ............8 ............................................

Table 3-5. Probabilities for contrasts of developmental
timing and developmental trajectory shape. The left side of
the table shows probabilities for the hypothesis of no
difference in developmental timing between treatments. The
right side shows probabilities for the hypothesis of no
interaction of development with time for each contrast, a
test of differences in trajectory shape. The top row shows
the contrast between the two controls; lower rows show
contrasts between experimental treatments and the controls.
Probabilities < 0.05 are indicated by *.

L vs. H

developmental time


trajectory shape


vs. L vs. H vs. L vs. H
II .1517 .1106 .0001* .1989
12 .0760 .0520 .0036* .7087
13 .4336 .3455 .6736 .9956
D1 .0397* .0515 .0672 .0001*
D2 .1205 .1568 .9852 .0675
D3 .4103 i .3238 .0001* .2153

Table 3-6. Results of Dunnett's tests for developmental
stage differences between the two controls, and between
experimental treatments and controls. Observations to the
right of the bold line were made after the food level
changes. Comparisons marked with are statistically
significant (a = 0.05). Observation days shown are for odd
numbered shelves, but data are pooled for pairs of
observation days (see text).

DAY 9 DAY 12 DAY 18 DAY 24 DAY 30 42 46
H vs L *

L vsI1 *
L vs 12 ...... ... ..... .......... .... ..... .
L vs I2

.... .... ... ......................... ................ ........................................................................ .........................
H vs *
L vs 132i
H vs 11
H vs I2 *
H vs 13 I. .* *

L vs D1 .
L vs D2 *
L vs D3 *

H vsD1 *
i L V S...... .................. i ..... ..

H vs D2
H vs D3

0.300 0 H

A 11



0.000- i i i i *
0 10 20 30 40 50 60

Figure 3-1. Mean growth trajectories for food increase
treatments and both controls. Growth trajectories end at
mean age and size at metamorphosis (stage 46), shown with 95%
confidence intervals. Food level increases are shown by
arrows. Weight loss during metamorphic climax is typical for
tadpoles and is due mostly to water loss.

A D1
+ D2
D3 D3
0.200- L

t: D1

0 10 20 30 40 50 60

Figure 3-2. Mean growth trajectories for food decrease
treatments and both controls. Trajectories end at mean age
and size at metamorphosis (stage 46), shown with 95%
confidence intervals. Food decreases are indicated by


----- 12
--- 13
-- L

20 30 40 50

DAY of


Figure 3-3. Mean developmental trajectories for food
increase treatments and both controls. The bold line
indicates the high food control (H). The line with filled
circles indicates the low food control (L). Food increases
are indicated by arrows.

----- D

---- D3
-- L

20 30 40 50


Figure 3-4. Mean developmental trajectories for food
decrease treatments and both controls. Control groups are
indicated as in Figure 3-3. Food decreases are indicated by



Age and size at metamorphosis are plastic phenotypic

characters in amphibian larvae that have been shown to

influence adult fitness (Collins 1979, Smith 1987, Semlitsch

et al. 1988, Berven 1990). Wilbur and Collins (1973)

proposed a model for predicting age and size at metamorphosis

that has become the dominant framework for understanding this

plasticity in an ecological context (reviews in Alford 1988,

Hensley 1993, Harris in press). The Wilbur-Collins model

proposes that metamorphosis is possible only for tadpoles

above a minimum size, and that metamorphosis is triggered by

a reduction of mass-specific growth rate. Thus,

environmental conditions that influence growth rate will

affect age and size at metamorphosis. Crump (1981) studied

energy accumulation in tadpoles of the spring peeper,

Pseudacris crucifer, a winter-breeding treefrog. She found

that tadpoles raised at low densities accumulated more energy

per unit size and developed faster than did tadpoles raised

under crowded conditions, even though both groups were fed ad

libitum. Crump proposed that energy accumulation rate may

influence timing of metamorphosis, and that accumulation of a

minimum amount of energy may be necessary for successful


In a previous study I demonstrated that energy

allocation in tadpoles is phenotypically plastic, varying

independently of body size (Chapter 2). This plasticity

suggests that energy allocation may be a size-independent

predictor of metamorphosis. Additionally, growth and

developmental trajectories of tadpoles (Chapter 3) are

generally consistent with a model of dynamic energy

allocation (Leips and Travis 1994) that predicts relative

changes in age and size at metamorphosis in response to

changes in growth rate. In the present study I test Crump's

(1981) prediction that total energy accumulation is an

important predictor of the timing of metamorphosis. In

addition, I examine the relationships among growth,

development rate, and lipid storage in light of a model of

proportional allocation of energy, and extend this model to

include allocation to storage.


I performed an experiment to examine the relationship

between development rate and lipid storage. The details of

the methods are presented in Chapter 3, along with an

analysis of resulting growth and developmental trajectories.

In this experiment Bufo terrestris tadpoles were raised

individually in plastic cups on controlled food rations from

hatching to metamorphosis.

This experiment tested the responses of tadpoles from

two sibships to eight different feeding regimes. Feeding

treatments consisted of two constant food level treatments

(low food (L) and high food (H), considered as controls) and

six variable food treatments. Treatments II, 12, and 13

experienced increased food levels, switching from low to high

food on one of three days during the experiment (days 12, 18,

and 24, respectively). The reciprocal food decreases were

made on the same schedule, designated as treatments D1, D2,

and D3.

For each tadpole, age and size at metamorphosis were

recorded. Based on the order in which they metamorphosed,

every third individual was then assigned to another study

(Hensley, unpublished). The remaining two thirds of the

metamorphs were retained for the present analysis of energy

allocation. Lipid storage at metamorphosis was measured

using petroleum ether extraction (methods modified from

Reznick and Braun (1987), presented in Chapter 2).

Statistical Analysis

Repeated measures analyses of variance on tadpoles in

this experiment revealed significant effects of feeding

treatments on the shapes of growth and developmental

trajectories (Chapter 3). Overall, however, timing of

metamorphosis was very uniform, with only a single treatment

(Dl) deviating significantly from its control group (H).

Body size (mass at metamorphosis), however, was strongly

influenced by treatment effects. Based on a previous

analysis of lipid storage in tadpoles (Chapter 2), I

anticipated that tadpole size would account for most of the

variation in lipid storage. To test whether feeding

treatments significantly influenced lipid storage, it is

necessary to account first for size effects, and also for

differences among sibships. I therefore calculated a general

linear model using Type I sums of squares that accounted

first for body size and sibship effects before testing if

treatments significantly affected lipid storage and whether

lipid storage was related to development rate. All

statistical calculations were performed using SuperANOVA 1.1

software (Abacus Concepts Inc. 1989).


In this experiment 200 tadpoles metamorphosed, and 142

were selected for lipid extraction. Growth trajectories and

mean metamorphic responses for each family are shown in

Chapter 3 (Figures 3-1 through 3-4). The relationship

between mass and lipid reserves at metamorphosis for each

treatment is plotted in Figure 4-1. The general linear model

using Type I sums of squares (Table 4-1) showed that size at

metamorphosis (dry mass) and sibship explained significant

variation in lipid storage. After accounting for these

effects, feeding treatments were marginally significant (P =

0.085). The final parameter in the model, day of

metamorphosis, was negatively related to adjusted lipid

storage (Table 4-1). This relationship is unambiguously

negative (slope = -5.39E-6, P = 0.0002) but is not strong (R2

= 0.087, Figure 4-2).


Crump (1981) proposed that incorporating energy

allocation patterns into models of amphibian metamorphosis

might increase their power for predicting timing of

metamorphosis. In Crump's study, high energy density (J/g

body mass) was associated with both large size at

metamorphosis and rapid development. This positive

correlation between growth and development rates is commonly

observed (Wilbur and Collins 1973, Collins 1979, Alford and

Harris 1988, Hensley 1993). In the present study, however,

early metamorphosis was not associated with large body size

(Chapter 3). There is evidence in the present study that

early metamorphosis was associated with greater lipid

storage, independent of body size (Figure 4-2). Although the

relationship was weak, it provides some support for the

hypothesis that energy accumulation rate influences

development independently of the effects of growth rate.

Leips and Travis (1994) proposed a model of dynamic

energy allocation to explain variation in age and size at

metamorphosis. According to their model, growth and

development are competing functions, and energy allocation

between these two functions determines age and size at

metamorphosis. In their model, early in the larval period

development is the highest priority for allocation.

Increases in food supply in this phase of the larval period

are primarily allocated to more rapid development, and

decreases in food supply result primarily in reduced

allocation to growth. Over developmental time, however,

growth becomes an increasing allocation priority and the

importance of development declines. Late in the larval

period changes in food supply affect size at metamorphosis,

but developmental timing is fixed and does not respond.

This model of dynamic allocation does not distinguish

between absolute energy allocation (Joules) and proportional

energy allocation (percent of available energy), nor does it

include an energy storage component. Harris (in press) has

proposed an alternative to Leips and Travis's (1994) model

that distinguishes between the absolute amount of energy

allocated and the proportional expenditure (Figure 4-3).

According to this model both the absolute amount of energy

and the proportion of total energy allocated to development

increase over developmental time for all tadpoles. Early in

development a greater proportion is allocated to growth, but

over time this proportion decreases as developmental demands

increase. According to this model, tadpoles in high food

environments allocate more energy to development than do

tadpoles with less food, but this represents a smaller

proportion of their total energy. Thus, tadpoles on high

food are able to grow and develop more rapidly than tadpoles

on low food.

In Harris's model, changes in food availability will

affect age and size at metamorphosis predictably, based on

allocation patterns. An increase in food supply that occurs

prior to initiation of metamorphosis will make more energy

available for both growth and development, resulting in

earlier metamorphosis at a larger size. Decreases in food

supply prior to initiation of metamorphosis will reduce the

total amount of energy allocated to both growth and

development, and will decrease the proportion of that energy

allocated to growth versus development, resulting in longer

larval period and reduced size at metamorphosis.

After initiation of metamorphosis, however, changes in

food supply will have different effects (Figure 4-3E).

According to Harris (in press), at initiation of

metamorphosis both the developmental trajectory (timing of

metamorphosis), and the energy required to complete

development are fixed. This energy demand is dependent on a

tadpole's previous growth and is larger for tadpoles from

environments with higher food availability. Therefore, when

tadpoles experience a food increase after initiation of

metamorphosis, a large fraction of the newly available energy

is allocated to growth and results in increased size at

metamorphosis. For tadpoles that experience a decline in

food late in the larval period, the proportional demands of

development increase and result in reduced, and possibly

negative, growth.

The foregoing model (Harris in press) distinguishes

between total energy expenditure and proportional energy

expenditure, but also lacks an energy storage component.

Harris states that the energy demands of the metamorphic

period (from initiation through metamorphic climax) are fixed

at the time of initiation and are dependent on a tadpole's

recent food availability. Though not explicitly stated, this

dependence probably represents the combined effects of body

size and developmental stage. This period of fixed

allocation and expenditure is based on observations that

development rate becomes insensitive to changes in growth

rate (Hensley 1993, Leips and Travis 1994). Analysis of

developmental trajectories (Chapter 3) indicates that a lack

of differences in timing of metamorphosis is not necessarily

indicative of a lack of developmental plasticity, and thus

may not represent fixed allocation to development.

Crump (1981) first suggested that a minimal energy

reserve was probably a prerequisite for successful

metamorphosis, and proposed that fat stores were a likely

source of such energy. I suggest that ontogenetic changes in

allocation to fat storage are complementary to Harris's (in

press) model and can explain the patterns of lipid storage

seen in Pseudacris crucifer (Chaptei 2) and B. terrestris.

The results from B. terrestris indicate that changes in food

supply at different developmental stages produce differences

in allocation to growth versus fat storage. There is a

significant trend for allocation to lipids to be greatest if

food availability increases at intermediate developmental

stages, compared to both earlier and later increases (Figure

4-4). The non-linear response of lipid storage suggests that

for tadpoles on low food, lipid storage is an increasing

priority relative to body size, up to a point. Later food

increases seem to be allocated toward increased body size

rather than proportionally larger fat reserves (Figure 4-4).

This pattern is consistent with the hypothesis that tadpoles

must store some minimum amount of fat for successful

initiation of metamorphosis near stage 35-37, but that for

tadpoles on low food any excess energy is allocated to


In contrast to the effects of food increases, decreases

in food supply showed no evidence of stage-specific effects

on lipid storage (Figure 4-4). Regardless of how long

tadpoles experienced high food availability, they

metamorphosed with similar fat content. The fat content of

tadpoles that experienced food declines was also not

different from either the constant high or constant low food


Unlike B. terrestris, P. crucifer showed a significant

tendency for tadpoles with high food availability to

metamorphose with significantly higher lipid stores on a

percentage basis (Chapter 2). A general model of fat storage

in tadpoles must account for the observed patterns in both of

these species.

I propose a model of allocation that can explain these

results. This model, an extension of Harris's model of

proportional allocation (Harris in press), emphasizes the

importance of size-specific and stage-specific allocation

(Figure 4-5). Assimilated energy that is not allocated to

maintenance or activity is referred to as production energy,

and may be allocated to growth, development, reproduction, or

storage. Both body size and developmental stage influence

how energy is allocated. Generally body size has the

strongest influence, and larger metamorphs have higher fat

storage in terms of absolute energy and on a percentage

basis, as in P. crucifer (Chapter 2). In some cases, such as

B. terrestris tadpoles in the present study, stage-specific

effects may dominate allocation.

Central to this model is the relationship between lipid

storage and development rates (Figure 4-5 arrow A). This is

the relationship originally proposed by Crump (1981), to be a

positive correlation. Some support for this proposed

relationship is seen in the tendency for B. terrestris

tadpoles with high size-adjusted fat storage to metamorphose

earlier (Figure 4-2). Under some conditions, however, a

negative relationship might be predicted. For example, in

extremely ephemeral ponds tadpoles might be predicted to

sacrifice lipid storage in favor of rapid development to

avoid death by desiccation.

The influence of developmental stage on the relationship

between allocation to growth versus allocation to development

(Figure 4-5 arrow B), is central to the dynamic allocation

model (Leips and Travis 1994) and the proportional allocation

model (Harris in press). Developmental stage influences the

relationship between growth and storage (Figure 4-5 arrow C),

and how this relationship changes when energy income changes.

This is seen in the non-allometric responses of size-adjusted

lipid reserves to changes in food supply in P. crucifer

(Chapter 2), and in stage-specific responses in B. terrestris

(Figure 4-4).

Most anurans metamorphose with undifferentiated gonads,

and thus allocation to reproduction is inconsequential. In

some species, however, gonads differentiate during the final

larval stages (Nodzenski et al. 1989, Hsu et al. 1991,

Hensley and Anderson unpublished), and stage-specific effects

on reproductive allocation may become important.

In this model, body-size effects generally dominate the

allocation patterns seen in tadpoles, with larger tadpoles

metamorphosing with relatively larger fat reserves. Tadpoles

raised on high food are predicted to store a greater fraction

of their total lipid reserves at earlier developmental stages

than do tadpoles on low food, but this stage-specific pattern

is overshadowed by body size effects. When tadpoles raised

on high and low food levels metamorphose with similar size-

adjusted lipid reserves, such as B. terrestris in this study,

stage-specific effects are predicted to be more apparent. In

this case, lipid storage might proceed at similar rates,

rather than occurring earlier for tadpoles on high food

(Figure 4-6A,B).

The model includes a hypothesis that changes in food

supply are potentially associated with changes in allocation,

and this potential depends on the stage when food supplies

change (Figure 4-6C). This hypothesis leads to predictions of

stage-specific allocation responses to changes in food

availability (Figure 4-6D,E,F). Tadpoles that experience food

increases at intermediate stages are predicted to be able to

allocate a greater fraction of their energy income to fat


In contrast, tadpoles that experience declines in food

supply midway through development might be predicted to

metamorphose with significantly reduced fat compared to

controls. They may not, however, if most lipid storage

occurs at earlier stages when food availability is higher. A

period on high food in the early part of larval development

could result in both rapid growth and a high rate of lipid

storage. A decrease in food supply would affect both growth

and development, but these tadpoles might experience both the

peak in fat storage associated with high food supply, and

then the later peak on low food supply. Such stage-specific

allocation could result in similar size-adjusted reserves at

metamorphosis for tadpole on high, low, and decreasing food

supplies. The final lipid reserves would depend on the

actual shape of the stage-specific allocation curves.

The pattern observed in B. terrestris might be the

product of programmed stage-specific responses to changes in

food supply, or to a gradient of changing plasticity in

allocation. One interpretation of the pattern is that food

increases at intermediate developmental stages generally

result in programmed increases in lipid storage, and that

this program is fixed. An alternative interpretation is that

the pattern is the product of a gradient in plasticity, and

that tadpoles that experience fluctuating food supplies at

intermediate stages are more plastic in their allocation

response than are tadpoles that experience such changes at

earlier or later stages. This ontogenetic change might

represent gradual increases and decreases in plasticity, or

might be due to abrupt changes between fixed and plastic

allocation patterns. Further tests of stage-specific

responses are necessary to quantify the changes in plasticity

and characterize which developmental stages show plasticity

in allocation.

This model of stage-specific allocation is designed to

be general, accommodating the patterns seen in P. crucifer

(Chapter 2) and in B. terrestris. In neither of these two

studies did feeding treatments generate much difference in

timing of metamorphosis, in contrast to several previous

studies (Wilbur and Collins 1973, Travis 1984, Alford and

Harris 1988, Hensley 1990, 1993). Whether the proposed model

is adequate under conditions where feeding treatments

strongly affect timing of metamorphosis remains to be seen.
In B. terrestris there is evidence that rapid

development is associated with high size-adjusted lipid

storage, as suggested by Crump (1981). Under conditions of

extreme food limitation or of time constraints on

development, however, one might predict a trade-off between

rapid development and greater lipid storage. Pfennig (1992)

found evidence for such a trade-off between omnivore and

carnivore morphs of the spadefoot toad Scaphiopus couchii.

Rapid growth and development of carnivore morphs was

associated with low fat storage. Omnivores metamorphosed

later and smaller, but with absolutely larger fat bodies.

Carnivores were more successful in short duration ponds,

where developmental timing was constrained; omnivores,

however, had higher post-metamorphic survival. The tendency

for rapidly growing carnivores to have smaller fat bodies

than slowly growing omnivores is not consistent with Crump's

(1981) hypothesis, but is accommodated by the model proposed

above. Pfennig (1992) indicated that this relationship was

part of an evolutionarily stable strategy that allows

spadefoots to persist in a desert environment where pond

permanence is highly unpredictable. The model proposed above

appears adequate for highly specialized tadpoles with

polymorphisms that reflect unique adaptations.

The diversity of patterns in larval fat storage seen in

P. crucifer, B. terrestris, and S. couchii points to the


importance of studies of energy allocation in complex life

cycles. Simple growth-based or size dependent models are

common, but may often be inadequate for predicting the

dynamics of developmental responses to fluctuating

environments. Models for predicting life history transitions

such as maturity, metamorphosis, or diapause are stronger if

they include stage-specific or age-specific patterns of

energy allocation and responses to changing environments.

Table 4-1. Analysis of covariance for loglo lipid storage in
B. terrestris tadpoles. Type I sums of squares are used to
test sequential hypotheses explained in the text.
Probabilities significant at the a = 0.05 level are marked
with *.

Source df Type I SS MS F P
dry mass j1 1.09 x 10-5 1.09 x 10-5 371.26 .0001*
family 1 4.41 x 10-7 4.41 x 10-7 15.04 .0002*
treatment 7 3.78 x 10-7 5.40 x 10-8 1.84 .0852
day of
metamorphosis 1 4.43 x 10-7 4.43 x 10-7 15.10 .0002*
Residual .127 3.73 x 10-6 2.94 x 10-8



0.002- A D1
0 0 D2
S-- D3
r II
< 0.001--- 12
U.; 13



o o 0 o
o o 0 0
d d o o


Figure 4-1. Mean mass of stored lipids versus mass at
metamorphosis for each treatment group. For clarity, only
one side of some 95% confidence intervals is shown. The
upper diagonal represents a 10% fat content, by mass; the
lower diagonal represents 5% fat.



S.00040- *
S* *
SE .00020- *
S- 0.00000 *0. -
=* V
S-.00020- *
0-. *; 0*
=E -.00040

2 -.00060

-.00100 *
30 40 50 60 70 80 90

Figure 4-2 Regression plot for the relationship between
adjusted lipid storage and day of metamorphosis. This plot
represents the final step in the analysis in Table 4-1.
After correcting for the effects of body size, sibship, and
feeding treatment, there is a decline in allocation to lipids
as length of the larval period increases. The slope of the
regression (-5.39 x 10-6) is significantly different from zero
(P = 0.0002) but the relationship is not strong (R2 = 0.087).

M 0
I23 2

o m o. ........ o *..
S.........- j


....... Low Food
-- High Food
- Food Level Switch


Figure 4-3. A graphical presentation of a model for dynamic
allocation of energy to growth and development in tadpoles
(modified from Harris in press). Each rectangle indicates
the period when development rate and developmental
expenditure are fixed and insensitive to changes in food
A. The proportion of energy allocated to development.
B. Absolute energy (Joules) allocated to development.
C. Proportional allocation to growth.
D. Absolute allocation to growth.
E. Absolute allocation to growth when food availability
changes (arrow) after development rate is fixed.


15- 0

0 0

S 0




0 0

0- I I I


Figure 4-4. The effect of developmental stage at which food
supply changed on lipid reserves at metamorphosis. The low
food control group is equivalent to a food decrease at stage
25 or a food increase at stage 46 (i.e., never). The high
food control is equivalent to an increase in food at stage 25
or a decrease at 46. Stage when food increased affected
percent lipid at metamorphosis, and both linear and quadratic
terms were significant (P < 0.001). Stage at food decrease
had no effect on lipid reserve at metamorphosis.

production PRODUCTION I
gonads production" PRO ON growth body size
gonads ENERGY



Figure 4-5. A graphical presentation of allocation in
tadpoles. Energy transfer functions are shown by wide arrows
and energy sinks by boxes. Dashed arrows indicate
relationships between energy transfer functions.
A. The relationship between development and energy storage
was first proposed by Crump (1981).
B. The trade-off between growth and development is central to
the models of Leips and Travis (1994) and Harris (in press).
C. Growth and storage are competing functions, but the trade-
off is often not detected (van Noordwijk and de Jong 1986,
Houle 1991).
Solid black arrows show the influence of body size
allometryy) and developmental stage on energy transfers;
either may dominate the relationship between growth and
storage. For some species, allocation to reproduction may
begin in later larval stages.

A.High Food

C. Stage-Specific Allocation

D. Early Increase

E. Intermediate Increase

F. Late Increase

Figure 4-6. In this model, total energy allocated to growth
declines over time (Harris in press), and is indicated by the
area under the upper curve. Allocation to lipid storage is
shown as the lower, shaded region. Size-adjusted fat reserves
at metamorphosis can be equivalent on high and low food
(shaded areas in A and B represent the same proportion of
allocation to growth). Panel C presents the hypothesis, that
tadpoles at intermediate developmental stages have greater
plasticity in allocation. The checkered region indicates the
additional lipid storage that can occur when intermediate
stage tadpoles experience food increases. Panels D-F are
predictions from C. Tadpoles that experience food increases
early (D) or late (F) in development, maintain the same
proportional allocation as in constant food environments
(A,B). A food increase at an intermediate stage (E), when
allocation patterns are more plastic, results in
proportionally more fat storage (checkered region) To
estimate a tadpole's energy allocation requires knowledge of
both its current stage and food supply, but also its
historical food supply.

000015-- ofi-,M_

A. Low Food



All organisms face the challenge of allocating

assimilated energy among the competing functions of growth,

maintenance, and reproduction. Energy allocated to one

function is unavailable for the others, but assimilated

energy can be stored for future allocation to one of these

three competing functions. The influence of energy storage

on allocation strategies is poorly understood (Meffe and

Snelson 1993) in spite of the fitness consequences of

allocation strategies (reviewed in Perrin and Sibly 1993).

Organisms with complex life cycles undergo a

morphological metamorphosis, usually accompanied by a habitat

shift (Wilbur 1980). An energy reserve may be critical for

surviving the metamorphic period (Crump 1981) and dispersal

into the new habitat. For amphibian larvae, age and size at

metamorphosis are phenotypically plastic traits that respond

to environmental conditions. Models have been developed to

predict age and size at metamorphosis, but these models do

not predict how energy is allocated between growth and

storage when conditions in the environment change


Models have been proposed to explain the phenotypic

responses of amphibian larvae to environmental variables such

as habitat duration, predation, food supply, and seasonality

(reviewed in Alford and Harris 1988, Hensley 1993, Harris in

press). One model, proposed by Wilbur and Collins (1973),

has been central to understanding how larval growth and

development respond to environmental variables. The major

tenets of the Wilbur-Collins model are that metamorphosis is

not possible below a certain minimum body size (b) but is

obligate at some maximum size (b + c). Within this size

range, metamorphosis is initiated when growth rate falls

below some size-dependent threshold (g) (Wilbur and Collins

1973). This model, with some refinements, has withstood

several empirical tests (reviewed in Alford and Harris 1988,

Hensley 1993, Harris in press) and has provided the best

framework for understanding the ecology of amphibian larval

growth and development.

In a study of the Wilbur-Collins model, Crump (1981)

found that tadpoles raised under crowded conditions

metamorphosed with less energy per unit body mass than did

tadpoles raised at low density. Crump suggested that

initiation of metamorphosis may require not simply attaining

a minimum body size (b), but also a minimum energy reserve,

and that successful models of amphibian metamorphosis should

include energy storage in addition to growth and development.

The dynamics of energy allocation in tadpoles have only

recently been studied. In a previous study (Chapter 2) I

found that fat storage in tadpoles is strongly allometric,

but that tadpoles also exhibit phenotypic plasticity in fat

storage independent of body size. Fat storage can also be a

significant predictor of timing of metamorphosis (Chapter 4),

which confirms Crump's (1981) suggestion that models for

predicting metamorphosis can be improved when energy

allocation is considered.

Previous studies of energy allocation in tadpoles

(Chapters 2-4, Leips and Travis (1994)) have been limited,

however, to studies of individually raised tadpoles and their

responses to changes in food availability. Many other

factors such as pond drying, temperature, and predation risk

can influence age and size at metamorphosis, and perhaps

energy allocation. The dynamics of energy allocation in

response to these factors have not been studied.

I proposed a model of energy allocation in tadpoles that

emphasizes the importance of allometric and stage-specific

allocation to fat storage (Chapter 4). A key element of this

model is the relationship between development rate and energy

storage rate, which Crump (1981) asserted should be positive.

I predicted, however that under conditions where time

constraints on development are extreme, such as when ponds

dry unpredictably, lipid storage may be traded against rapid

development. Pfennig (1992) provided evidence that this

trade-off occurs in spadefoot toads, Scaphiopus couchii, in

very ephemeral ponds.

I chose to examine the response of tadpoles to early

pond drying because several species are known to accelerate

development facultatively in response to habitat desiccation.

This developmental acceleration has been demonstrated to have

a genetic component and is considered to be adaptive (Newman

1988a, Semlitsch et al. 1990, Newman 1992). The mechanism of

this acceleration of development is unknown. One possible

factor in this acceleration may be a shift in allocation of

energy away from storage and toward increased growth and

rapid development.

Two questions are central to this study. First, do

tadpoles adjust their energy allocation in response to early

pond drying? I predicted that tadpoles from short duration

ponds would metamorphose early, with reduced fat storage.

Second, do tadpoles from different genetic backgrounds

(sibships) respond to early pond drying in different ways? I

predicted that sibships that tend to develop more rapidly

would be less influenced by short duration ponds than would

sibships with slower development.


Previous work has demonstrated genetic variation for

amphibian larval growth and development, including genetic

variation for responses to early pond drying (Berven et al.

Full Text
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 EAG93MHF9_5KSKGE INGEST_TIME 2012-02-17T16:59:07Z PACKAGE AA00003625_00001