Expression of fluctuating asymmetry in primate teeth: Analyzing the role of growth duration.
Dissertation
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of the Ohio State University
By
Sarah Abigail Martin, M.A.
Graduate Program in Anthropology
The Ohio State University
2013
Dissertation Committee:
Dr. Guatelli-Steinberg, Advisor
Dr. Scott McGraw
Dr. Paul Sciulli
Dr. Dawn Kitchen
i
Copyright by
Sarah Abigail Martin
2013
ii
Abstract
This dissertation furthers our understanding of the association between growth duration and developmental noise (DN) by examining fluctuating asymmetry (FA) in a non-sexually selected and a sexually selected structure. FA occurs as small, random deviations from bilateral symmetry without any directionality between sides. Extended periods of growth may provide a ‘window of opportunity’ for growing body structures to accumulate perturbations resulting in elevated DN, manifested as FA. In comparison to other mammalian species, primates exhibit prolonged growth. Within the primate order, growth periods lengthen from prosimians to apes and humans. Although prolonged growth periods can be advantageous, lengthening of the growth period may provide the opportunity to accumulate deviations from symmetry. An association between growth duration and DN has yet to be studied across the primate order. This dissertation tested if and to what extent growth duration was associated with dental FA in primates.
This potential association between growth duration and FA was first examined in the non-sexually selected first molar. First molar FA was compared between species based on crown formation times (CFTs) and life history (LH) schedules to test the hypothesis that species with prolonged CFTs or LH schedules would express greater first molar FA relative to species with shorter CFTs or LH schedules. The results generally lend support to the hypothesis; however, not all comparisons are statistically significant. ii
An additional aim was to elucidate the mechanism(s) which underlie the
association between FA and sexually selected structures. Sexually selected structures are
prone to exhibiting elevated FA because they may be destabilized by directional
selection. The primate canine has been given as one example of a sexually selected
structure prone to expressing elevated FA. It is possible that the mechanisms believed to
explain elevated canine FA in primates may not be associated with the destabilization of
developmental processes but rather with the fact that males of sexually dimorphic species
grow their canines for a longer period of time relative to their female counterparts. Two
approaches were used in testing for an association between FA and sexual selection.
First, canine FA was compared between males and females of a species. Generally, males of sexually dimorphic species expressed greater canine FA but not all comparisons were statistically significant. To further test this association, male canine FA was evaluated through linear regressions and controlled comparisons among males of different species.
Linear regression results indicated that canine growth duration more closely predicted canine FA than variables representing sexual selection. Controlled comparisons that included great apes support the hypothesis that prolongation of growth is associated with elevated FA. An association between growth duration and FA is further supported in some but not all comparisons between males of different monkey species.
Because anthropologists, and other scholars, use FA as an indicator of developmental stress, understanding what factors are associated with FA is essential to accurately intrepreting FA data. The results of this dissertation lend support to the hypothesis that growth duration contributes to dental FA in both a non-sexually selected and a sexually selected structure. iii
DEDICATION
I dedicate this dissertation to my mother, Christine, my father, Andrew, and my brother, Daniel. Without your love, support, and encouragement I would not have made it this far.
iv
ACKNOWLEDGEMENTS
I wish to thank my advisor, Dr. Debbie Guatelli-Steinberg, for her intellectual support, encouragement, and enthusiasm. I am also grateful to my other dissertation committee members, Dr. Paul W. Sciulli, Dr. W. Scott McGraw, and Dr. Dawn Kitchen, for their guidance and support, as well as Dr. Gary Schwartz for providing unpublished data on dental growth of great apes. I would also like to extend my gratitude to my fellow graduate students who served as a source of support through our shared experiences.
I thank everyone who provided access to the primate collections used in this dissertation: Judith Chupasko (Museum of Comparative Zoology, Harvard University),
Lyman Jellema (Cleveland Museum of Natural History), Darrin Lunde (National
Museum of Natural History), Bill Stanley (Field Museum), Martha Tappen (Tappen
Collection, University of Minneapolis), and Eileen Westwig (American Museum of
Natural History). This dissertation would not have been possible without the support from the Wenner-Gren Foundation (Grant #Gr8395), National Sigma Xi, and The Ohio
State University’s Sigma Si Chapter, Critical Difference for Women Program, Graduate
School, and Department of Anthropology.
Finally, I am indebted to my family and friends for their love and support. I especially thank Daniel Martin, Bradley Vile, Denise Fickes, Christina Leach, Krystle
Gnatz, Whitney and Matthew Senn, Jennifer Spence, Jennifer Everhart, and Jesse
v
Fuhrman for sharing in the joys and growing pains of life. Lastly, I thank my parents,
Andrew and Christine, for always being a source of love, guidance, and support.
vi
VITA
June 2000 ...... Lower Cape May Regional High School
2004...... B.S. Biological Sciences, Rowan University
2006...... M.A. Anthropology, The Ohio State University
2007-2011 ...... Graduate Teaching Associate, Department of Anthropology, The Ohio State University
2011-present ...... Lecturer, Department of Anthropology, The Ohio State University
Publications
Martin SA, Guatelli-Steinberg D Sciulli PW. (2011). Relationships among crown formation time, fluctuating asymmetry, and linear enamel hypoplasia in gibbons and gorillas. Presented at the 80th Annual Meeting of the American Association of Physical Anthropologists, Minneapolis, Minnesota. American Journal of Physical Anthropology 144 (S52): 207.
Martin SA, Guatelli-Steinberg, D, Sciulli PW, Walker, PL. (2008). Brief Communication: Comparison of Methods for Estimating Chronological Age at Linear Enamel Formation on Anterior Dentition. American Journal of Physical Anthropology 135(3): 362-365.
Fields of Study
Major Field: Anthropology
vii
TABLE OF CONTENTS
Abstract ...... ii
Dedication ...... iv
Acknowledgements ...... v
Vita ...... vii
List of Tables ...... xiv
List of Figures ...... xvii
Chapter 1: Introduction ...... 1
Objectives ...... 5
Hypotheses ...... 7
Hypothesis #1...... 7
Hypothesis #2...... 8
Hypothesis #3...... 10
Significance...... 11
Chapter 2: Literature Review – Theoretical Context ...... 14
Description and Causes of Fluctuating Asymmetry ...... 18
Elevated Fluctuating Asymmetry: A Measure of Increased Developmental Noise
...... 20
Directional Selection, Fluctuating Asymmetry, and Sexually-Selected Structures ...26
Growth Processes and Fluctuating Asymmetry ...... 31 viii
Growth Duration and Fluctuating Asymmetry ...... 35
Filling in the Gaps ...... 38
Chapter 3: Literature Review – Estimating Fluctuating Asymmetry ...... 41
Validating the Data Prior to Calculating Fluctuating Asymmetry...... 43
Measurement Error ...... 44
Forms of Asymmetry ...... 46
Type and Number of Structures ...... 49
Wear and Damage ...... 52
Structure Size ...... 53
Indices and Analysis: Fluctuating Asymmetry ...... 56
Fluctuating Asymmetry Indices ...... 56
Recommendations Given in the Literature for Statistical Analysis of Fluctuating
Asymmetry ...... 63
Chapter 4: Material and Methods ...... 69
Materials ...... 69
Sample Selection of Non-human Primate Species ...... 69
Taxonomy Considerations ...... 70
Sample Size ...... 71
Methods...... 74
Variables ...... 74
Data Collection ...... 83
Analytical Phase ...... 87
Summary ...... 93 ix
Chapter 5: Fluctuating Asymmetry of Primate First Molars: Association with Growth
Duration ...... 94
Introduction ...... 94
Stage One Description ...... 95
Stage One Description ...... 96
Results ...... 99
Stage 1 ...... 99
Discussion ...... 109
Stage 1 ...... 109
Results II ...... 115
Stage 2 ...... 115
Discussion ...... 121
Stage 2 ...... 121
Summary ...... 122
Chapter 6: Extending the Test of the Association Between Growth Duration and Dental
Fluctuating Asymmetry ...... 124
Introduction ...... 124
Predictions of Hypothesis 1b ...... 127
First Set of Comparisons ...... 127
Second Set of Comparisons ...... 130
Results ...... 131
Maxillary First Molars ...... 132
Mandibular First Molars ...... 140 x
Discussion ...... 144
Great Apes: Life In the Slow Lane ...... 147
Folivorus and Frugivorous: Old World Monkeys ...... 149
Linear Enamel Hypoplasia, Fluctuating Asymmetry, and the WOV Hypothesis
...... 150
Conclusions ...... 152
Chapter 7: Canine Fluctuating Asymmetry: A Comparisons Between Males and Females
...... 154
Introduction ...... 154
Objectives and Hypothesis ...... 156
Canine Dimorphism Index ...... 157
Testing Hypothesis 2...... 159
Stage 1 ...... 159
Stage 2 ...... 163
Results ...... 163
Stage 1 ...... 164
Stage 2 ...... 169
Discussion ...... 181
Conclusion ...... 185
Chapter 8: Fluctuating Asymmetry of the Male Primate Canine ...... 187
Hypothesis 3...... 189
Linear Enamel Formation Times (EFTs) ...... 189
Competition Levels ...... 192 xi
Regression Analysis ...... 193
Linear Regression: Results and Discussion ...... 195
Controlled Comparisons ...... 206
Controlled Comparison Results ...... 209
First Series of Controlled Comparisons: Mean Lateral EFTs of 80% Crown
Height ...... 211
Second Series of Controlled Comparisons: Mean Lateral EFTS of 80% Crown
Height ...... 217
First Series of Controlled Comparisons: Mean Lateral EFTs of 90% Crown
Height ...... 221
Second Series of Controlled Comparisons: Mean Lateral EFTs of 90% Crown
Height ...... 227
Discussion ...... 231
Reprised: Vulnerability to Asymmetry ...... 233
Cebus apella: The Odd Species Out? ...... 236
Size Matters ...... 240
Conclusion ...... 242
Chapter 9: Discussion and Conclusions ...... 245
Summary of Objectives ...... 245
Summary of Hypotheses ...... 247
‘Window of Vulnerability’ ...... 253
So Little Time, Such Great Symmetry ...... 253
Sex Differences in Mandibular Canine FA ...... 255 xii
Growth Processes ...... 262
Dental Indicators of Stress ...... 264
General Overview of Patterns Observed Across Primate Species ...... 264
A Second Look: Dental FA in Great Apes ...... 268
Nutritional Stress ...... 268
Parasites and FA ...... 271
Limitations of Comparisons ...... 273
Conclusions ...... 274
Contribution of this Dissertation to Understanding Dental FA Patterns ...... 276
References ...... 279
Appendix A: Specimen Listings ...... 311
Appendix B: Hypothesis 1a: ANOVA Tables ...... 317
Appendix C: Hypothesis 1b: ANOVA Tables ...... 325
Appendix D: Hypothesis 2: ANOVA Tables...... 329
xiii
LIST OF TABLES
Table 1: Factors Influencing the Presence of Developmental Stability and Developmental
Instability within a Developmental System ...... 16
Table 2: Indices for Fluctuating Asymmetry ...... 58
Table 3: Indices for Measurement Error ...... 66
Table 4: Taxonomic Classification – Changes from 1990s to Present ...... 71
Table 5: Primates Included in the Dissertation Sample ...... 72
Table 6: References and First Molar Crown Formation Times for Testing H1a ...... 75
Table 7: Weaning Ages of Primate Species for Testing H1b ...... 77
Table 8: Mean Lateral Enamel Formation Times for 80% and 90% Crown Height ...... 79
Table 9: Competition Levels of Primate Species Used to Test H3 ...... 82
Table 10: Sample Size and CFTs of Primate Species Used to Test H1a ...... 97
Table 11: Maxillary and Mandibular First Molar FA Results ...... 99
Table 12: F-tests Stage 1, H1a ...... 100
1 Table 13: F-test for M and M1 FA Comparisons...... 113
Table 14: FA Estimates, H1a Stage 2 ...... 115
Table 15: F-test Stage 2, H1a ...... 116
Table 16: Sample Sizes of Primate Species Used to Test H1b ...... 128
Table 17: Weaning Ages of Primate Species Used to Test H1b ...... 131
Table 18: FA Estimates of Maxillary and Mandibular First Molar Crown Area ...... 132 xiv
Table 19: Maxillary and Mandibular F-tests ...... 133
Table 20: ANOVA Tables for Linear Regression Analysis (N=12)...... 143
Table 21: Sample Sizes for Stage 1 of Testing H2 ...... 158
Table 22: Sample Sizes for Stage 2 of Testing H2 ...... 158
Table 23: Maxillary and Mandibular Canine Dimorphism Indices ...... 160
Table 24: Canine FA Estimates – H2, Stage 1 ...... 165
Table 25: F-tests, Stage 1...... 166
Table 26: Results of Comparisons Between Males and Females of Sexually Dimorphic
Species – H2, Stage 1 ...... 169
Table 27: Canine FA Estimates, H2, Stage 2 ...... 170
Table 28: F-tests, Stage 2...... 172
Table 29: Results of Comparisons Between Males and Females of Sexually Dimorphic
Species – H2, Stage 2 ...... 179
Table 30: Mean Lateral EFTs for 80% and 90% Crown Height – New World Monkeys,
Old World Monkeys, and Lesser Apes ...... 190
Table 31: Mean Lateral EFTs for 80% and 90% Crown Height for Great Apes ...... 192
Table 32: Data Used for Linear Regressions (LR) ...... 196
Table 33: Set 1, Linear Regression Results ...... 200
Table 34: Set 2, Linear Regression Results ...... 201
Table 35: Set 3, Linear Regression Results ...... 201
Table 36: Species Used for Controlled Comparisons ...... 207
Table 37: First Series of Controlled Comparisons ...... 208
Table 38: Second Series of Controlled Comparisons ...... 208 xv
Table 39: F-tests for Controlled Comparisons Using Mean Lateral EFTs at 80% Crown
Height ...... 212
Table 40: F-tests for Controlled Comparisons Using Mean Lateral EFTs at 90% Crown
Height ...... 222
Table 41: Results of Comparisons: Same CLs, Different C1 Lateral EFTs ...... 232
Table 42: Results of Comparisons: Different CLs, Similar C1 Lateral EFTs ...... 233
Table 43: Mean Lateral EFTs for 80% Crown Height ...... 257
Table 44: Mean Lateral EFTs for 90% Crown Height ...... 258
Table 45: F-tests for Cebus C1 FA ...... 267
Table 46: First Molar Crown Formation Times, Sample Sizes, and FA Estimates ...... 269
Table 47: F-tests for First Molar FA of Great Apes ...... 270
xvi
LIST OF FIGURES
Figure 1: Relationship Among Developmental Processes ...... 17
Figure 2: Tooth Section Depicting the Relationship of Striae of Retizus to Perikymata
...... 78
Figure 3: Predicted Fluctuating Asymmetry Magnitudes Among Platyrrhines and
Catarrhines ...... 98
Figure 4: Mandibular First Molar FA Estimates of Apes, H1a Stage 1 ...... 103
Figure 5: Maxillary First Molar FA Estimates, H1a Stage 1 ...... 105
Figure 6: Mandibular First Molar FA Estimates, H1a Stage 1 ...... 106
Figure 7: Maxillary and Mandibular FA Estimates, H1a Stage 2 ...... 119
Figure 8: Maxillary First Molar Crown Area FA Results...... 141
Figure 9: Regression Graph for Maxillary Molars ...... 142
Figure 10: Mandibular First Molar Crown Area FA Results ...... 145
Figure 11: Regression Graph for Mandibular Molars ...... 146
Figure 12: Maxillary Canine Dimorphism Index ...... 161
Figure 13: Mandibular Canine Dimorphism Index ...... 162
Figure 14: C1 FA Results for Predictions 2.1 and 2.2, Stage 1 ...... 167
Figure 15: C1 FA Results for Predictions 2.1 and 2.2, Stage 1 ...... 168
Figure 16: C1 BL FA Results for Prediction 2.1, Stage 2 ...... 175
Figure 17: C1MD FA Results for Prediction 2.1, Stage 2 ...... 176 xvii
Figure 18: C1 BL FA Results for Prediction 2.1, Stage 2 ...... 177
Figure 19: C1 MD FA Results for Prediction 2.1, Stage 2 ...... 178
1 Figure 20: C1 and C FA Results for Prediction 2.2, Stage 2 ...... 180
Figure 21: Graph of C1 BL FA (asy) and Lateral EFTs (LEFTS) ...... 198
Figure 22: Graph of C1 MD FA (asy) and Lateral EFTs (LEFTS) ...... 199
Figure 23: Graphs for C1 FA (asy) and Competition Level (cl), Set 1 ...... 202
Figure 24: Graphs for C1 FA (asy) and Canine Dimorphism (cd), Set 1 ...... 203
Figure 25: Graphs for C1 FA (asy) and Competition Level (cl), Set 2 ...... 204
Figure 26: Graphs for C1 FA (asy) and Canine Dimorphism (cd), Set 2 ...... 205
Figure 27: First Series of Controlled Comparisons for 80% Crown Height ...... 209
Figure 28: First Series of Controlled Comparisons H3a, 90% Crown Height...... 210
Figure 29: Second Series of Controlled Comparisons 80% Crown Height ...... 210
Figure 30: Second Series of Controlled Comparisons for 90% Crown Height ...... 211
Figure 31: Results for the First Series of Controlled Comparisons – Mean Lateral EFTs at
80% Crown Height – C1 BL ...... 214
Figure 32: Results for the First Series of Controlled Comparisons – Mean Lateral EFTs at
80% Crown Height – C1 MD ...... 215
Figure 33: Results for the Second Series of Controlled Comparisons – Mean Lateral
EFTs at 80% Crown Height – C1 BL ...... 219
Figure 34: Results for the Second Series of Controlled Comparisons – Mean Lateral
EFTs at 80% Crown Height – C1 MD ...... 220
Figure 35: Results for the First Series of Controlled Comparisons – Mean Lateral EFTs at
90% Crown Height – C1 BL ...... 224 xviii
Figure 36: Results for the First Series of Controlled Comparisons – Mean Lateral EFTs at
90% Crown Height – C1 MD ...... 225
Figure 37: Results for the Second Series of Controlled Comparisons – Mean Lateral
EFTs at 90% Crown Height – C1 BL ...... 229
Figure 38: Results for the Second Series of Controlled Comparisons – Mean Lateral
EFTs at 90% Crown Height – C1 MD ...... 230
xix
CHAPTER 1: INTRODUCTION
Across the primate order, periods of growth lengthen from strepsirrhines, which
erupt their teeth and achieve adult body proportions over short periods of time (Smith
1989; Godfrey et al. 2001), to apes and humans, which require several years of growth to
achieve adult form (Leigh and Shea 1995; Schwartz and Dean 2001). While prolonged
growth periods may be advantageous in some respects (e.g., affording time for learning), they may represent an opportunity for growing body structures to experience
developmental stress and sustain injury to developing systems. Various researchers
(Vrijenhoek 1985; Hallgrimsson 1993, 1995) have hypothesized that longer periods of growth provide the opportunity for growing body structures to accumulate developmental perturbations resulting in deviations from perfect bilateral symmetry. Van Valen (1962)
coined the term ‘fluctuating asymmetry’ (FA) to describe small deviations from bilateral
symmetry that are random with respect to the side that is larger. Despite efforts to test the
association between growth duration and FA in primates (e.g. Nass 1982; Hallgrimsson
1999), this potential association has yet to be tested systematically across the primate
order. This dissertation aims to do so, providing a foundation for interpreting what FA
means in terms of stress, as it is used as an indicator of stress experienced during
development (e.g. Mal et al. 2002; Trotta et al. 2005; Dongen et al. 2009).
1
Sexually selected structures are suggested to be more prone to expressing elevated
FA relative to non-sexually selected structures because directional selection relaxes the
control mechanisms of development in sexually selected structures so that an increase in
trait expression can occur (Moller 1990; Moller and Pomiankowski, 1993; Watson and
Thornhill 1994). FA in sexually selected structures has therefore been proposed as a signal of quality which can be used by in conflict assessment by competing males and in mate selection by choosy females (Moller, 1993; Moller and Pomiankowski, 1993). The primate canine has been given as one example: males of species with sexually dimorphic
canines tend to exhibit greater FA than males of species with sexually monomorphic
canines (Manning and Chamberlain, 1993).
However, not all studies have found that sexually selected structures exhibit greater
FA than non-sexually selected structures (e.g. Tomkins and Simmons, 1995; Bjorksten et al., 2000; Lens et al., 2002). Furthermore, since the proposal of the FA-sexual selection hypothesis in the 1990s (e.g. Moller, 1990; Moller and Hoglund, 1991; Moller, 1992) there has been a dramatic decrease in the amount of support for the hypothesis (Tomkins and Simmons, 2003). The inconsistent results raise the question as to why some but not all studies find a relationship between sexually selected structures and elevated FA. As such, an additional aim of this study is to determine whether growth duration can explain the association between FA and sexually selected primate canine. The explanation for the association between sexually selected structures and FA may not be limited to the destabilization of developmental processes associated with sexual selection (Moller 1990;
Moller and Pomiankowski, 1993; Watson and Thornhill 1994). Rather, prolonged growth
2
duration may contribute to the observation that sexually selected structures exhibit elevated FA.
Vrijenhoek (1985) describes the opportunity to accumulate deviations during development as a “window of vulnerability” (WOV). Relative to body structures that take
a short time to form, those structures that take longer to form have a greater WOV to
experience stress and accumulate developmental perturbations, which reduce bilateral
symmetry. If sexually selected structures have prolonged growth periods relative to non- sexually selected structures, then they may also have a greater WOV to accumulate perturbations and express developmental noise, as measured through FA. Therefore, the mechanism that explains the association between sexually selected structures and FA might not be the destabilization of the developmental process associated with sexual selection, but rather a “window of vulnerability” caused by growth prolongation.
In sexually dimorphic primate species, male canines take longer to form than those of females (Swindler et al., 1982; Sirianni and Swindler, 1985; Swindler, 1985;
Schwartz et al. 1999, 2001; Schwartz and Dean 2001; Guatelli-Steinberg et al. 2009). For example, on average Gorilla males grow their canines for almost three years longer than
Gorilla females (Schwartz and Dean, 2001). Thus, male canines may simply have a greater opportunity to accumulate deviations from symmetry resulting in elevated canine
FA relative to those of females. This dissertation aims to elucidate the mechanism(s) that underlie the association between sexually selected structures and FA. However, answering the question of if, and to what degree, duration of canine growth, as assessed
3
by canine crown formation times, explains the variation observed in canine FA
expression across the primate taxa is complicated.
Directional selection, or selection that tends to favor phenotypes at one extreme of
the phenotypic range, is believed to relax developmental control mechanisms resulting in
a reduction in efficiency and, ultimately, the stability of the developmental system
(Parsons 1992; Palmer 1994; Watson and Thornhill 1994). The relaxation of control mechanisms in structures under intense directional selection is hypothesized as the reason for why sexually selected structures are more prone to developmental perturbations and exhibit elevated FA than non-sexually selected structures (Moller 1990, 1992; Watson and Thornhill 1994; Moller and Swaddle 1997). Sexual selection has not only been
hypothesized to bring about ‘developmental destabilization’ leading to increased FA but
is also associated with prolonged canine crown formation times in sexually dimorphic
primate species (Schwartz et al. 1999, 2001; Schwartz and Dean 2001; Guatelli-Steinberg
et al. 2009). Because ‘developmental destabilization’ and prolonged canine crown
formation are both associated with sexual selection, it is difficult to separate them from
each other in order to adequately address which variable appears to better explain
variation in dental FA.
This dissertation was designed to determine if growth duration is associated with
dental FA in a non-sexually selected structure and to disentangle the potential effects of
‘developmental destabilization’ and prolonged canine crown formation times in male FA,
both of which are associated with sexual selection. Before examining the sexually
selected primate canine, it must first be determined if growth duration is associated with
4
FA in teeth that are not sexually selected. If a non-sexually selected structure follows the
pattern of FA predicted by WOV (Vrijenhoek, 1985), then it is reasonable to hypothesize
that the same FA pattern would be observed in sexually selected structures.
Objectives
The purpose of this dissertation is to determine the extent to which growth duration is associated with developmental noise in primate teeth. There are important
reasons to investigate developmental instability and examine dental FA.
Developmental instability (DI) is “the inherent noisiness of a developmental
pathway,” (Waddington, 1957, p. 40). The developmental noise (DN) that Waddington
(1957) refers to is the result of all the random, independent events occurring over
developmental time, which have the potential to disrupt the normal trajectory of
development. When the developmental system’s trajectory is disrupted, DN develops,
causing bilateral structures to express non-directional subtle differences referred to as FA.
Branches of biology as well as other disciplines have commonly used departures
from symmetry to estimate the occurrence of processes that disrupt development (e.g.
Parsons, 1990; Livshits and Kobyliansky, 1991; Zakharov, 1994; Hallgrimmson, 1995;
Moller and Swaddle, 1997). A great deal of focus has been given to the association
between FA and fitness (e.g. Moller, 1990; Parsons, 1990; Zakharov, 1992; Palmer,
1996). The general assumption is that individuals who are more fit are better able to
successfully buffer developmental perturbations and achieve the target phenotype relative
to individuals who have reduced fitness (Maynard Smith et al. 1985; Gibson and Wagner,
5
2000). If an individual is unable to buffer the effects of perturbations, the developmental
trajectory is disrupted resulting in noise that is expressed as FA.
There are advantages to investigating FA in dentition over that of the skeleton
(Goodman and Armelagos, 1989; Goodman and Rose, 1991; Hillson, 1996; Wood, 1996;
Hillson and Bond, 1997). First, unless affected by wear, the size and shape of the tooth does not change after crown formation ceases. Relative to FA in the skeleton, teeth record perturbations experienced throughout the developmental period only rather than during the most recent stage of development or since the last time bone was remodeled.
Additionally, teeth are durable, leading them to be commonly found within paleontological and archaeological assemblages.
In this study, FA is evaluated in two tooth types of non-human primates: the permanent first molar and permanent canine. In the first step of this dissertation, the objective is to determine if growth duration is associated with variation in FA expression in the primate first molar. Since ‘developmental destabilization’ and growth duration are both associated with sexual selection, the sexually dimorphic primate canine cannot be used to initially determine if prolonged growth provides an opportunity for the accumulation of perturbations leading to greater FA. Because the first molar of non- human primates is a non-sexually selected tooth, it can be used to determine if FA expression is influenced by prolonged growth since it is not influenced by
‘developmental destabilization’. First molar FA will be compared across primate species
with varying crown formation times (CFTs) of first molars obtained from the literature
(e.g. Macho, 2001; Smith et al., 2007; Kelley and Schwartz, 2009). In this first step,
6
males and females of a species are combined because first molar CFTs do not differ between the sexes of a species (Macho, 2001). If growth duration is shown to be
associated with elevated first molar FA, it is hypothesized that growth duration may also
be associated with canine FA.
Canine FA is evaluated through two separate hypotheses. First, canine FA is
compared between males and females of a species in order to examine if males of
sexually dimorphic species express greater canine FA relative to females. However,
comparing male and female canine FA will not provide a clear indication of if canine FA
is associated with ‘developmental destabilization’ or growth duration, or both. To
evaluate ‘growth duration’ and ‘developmental destabilization’ independently, canine FA
is compared among males of different primate species.
Hypotheses
Hypothesis #1
The first hypothesis (H1) is that length of growth duration influences the
expression of FA such that longer periods of growth are associated with elevated FA in a
non-sexually selected structure. H1 is tested in two different phases: H1a and H1b. Both
phases test the hypothesis that primate species with longer growth periods will exhibit a
greater WOV resulting in elevated FA relative to primate species with shorter growth
periods. In the first phase, H1a, first molar CFTs represent the growth variation through
which FA of primate species are compared. The second phase, H1b, uses a continuum of
7
‘slow’ to ‘fast’ life history (LH) schedules of primates to evaluate if developmental timing is associated with elevated first molar FA.
Primate first molar FA has been examined only in Macaca (Hallgrimsson 1999).
This dissertation examines FA in the first molar of ten catarrhine and three platyrrhine species to explicitly test the hypothesis that variation in growth duration is associated with elevated FA. Using both first molar CFTs and a continuum of LH schedules allows for H1 to be tested through two different but related variables. Dental development has been shown to be closely associated with LH variables across primates (Smith, 1989,
1992, 1994; Macho, 2001; Dean, 2006). Moreover, Macho (2001) demonstrated that the timing of weaning is strongly related to first molar CFTs in anthropoid primates. H1a will be tested using CFTs of primate species. In order to increase the number of species included in the comparison, H1b will be tested using LH schedules and weaning age as proxies for developmental timing.
Hypothesis #2
Hypothesis 2 (H2) is that: 1) within sexually dimorphic anthropoid primate species, FA will be greater in male canines relative to female canines; and 2) within anthropoid primate species with minimal or no canine sexual dimorphism, male and female canines will exhibit similar FA. H2 represents the first step in evaluating dental
FA in a sexually selected structure.
The primate canine is a sexually selected structure whose opportunity to experience developmental perturbations may be influenced by ‘developmental
8
destabilization’ associated with sexual selection and/or sex differences in canine growth durations also associated with sexual selection. Therefore, H2 does not tease apart the potential effects of ‘developmental destabilization’ and/or the ‘window of vulnerability’
(WOV). Rather, the results of H2 will determine 1) if males of sexually dimorphic species exhibit canine FA that is statistically significantly greater than their female counterparts; and 2) if similar canine FA is expressed by males and females of primate species with minimal or no canine sexual dimorphism. If male canines of sexually dimorphic species do express FA that is statistically significantly greater than female canine FA, then the question becomes: Do canine FA estimates better support the
‘developmental destabilization’ or WOV hypothesis?
If some, but not all, males of sexually dimorphic primate species express statistically significantly greater FA than their female counterparts, an additional question arises: What factor(s) may influence one sexually dimorphic primate species to express canine FA estimates that differ significantly between the sexes, while another sexually dimorphic primate species does not?
An additional component of testing H2 is the use of canine dimensions not influenced by use-related wear. In a previous canine FA study (Manning and
Chamberlain, 1993), FA was determined from the dimension of canine height. Canine height is a dimension susceptible to use-related wear. Because use-related wear does not necessarily occur bilaterally, canine height wear could skew FA estimates. Moreover, significant methodological and statistical advances in the study of FA have occurred since the publication of Manning and Chamberlain’s (1993) paper. FA is greatly
9
influenced by measurement error (ME) because the distribution of signed FA and ME
display similar statistical properties and are often comparable in magnitude (Smith et al.,
1982; Greene, 1984; Palmer and Strobeck, 1986; Palmer, 1994; Swaddle et al., 1995).
This study will measure canine dimensions that are not affected by use-related wear and includes a step in the methodological protocol to examine wear prior to measurement. The most recent methodological and statistical approaches for FA studies are utilized in this study including performing a two-way mixed model ANOVA and factoring out other forms of asymmetry prior to calculating FA.
Hypothesis #3
Hypothesis 3 (H3) is that males of primate species with similar durations of
canine growth will exhibit similar FA, while males with different durations of canine growth will express different FA magnitudes. Specifically, males with longer durations of
canine growth, as determined by lateral enamel formation times (EFTs) of canines, will
exhibit greater FA relative to males of primate species with shorter periods of growth.
To test this hypothesis a series of controlled comparisons meant to disentangle the
variables of ‘developmental destabilization’ and growth duration are employed. The
strength of sexual selection is represented by competition levels (CL) among males taken
from the work of Plavcan and colleagues (Kay et al., 1988; Plavcan and van Schaik,
1992). Growth duration is represented by canine lateral enamel formation times
(Schwartz and Dean, 2001; Guatelli-Steinberg et al., 2009). The first series of controlled
comparisons is between primate species with different durations of canine growth, but
10
assigned to the same CL. Under the WOV hypothesis, species with a longer duration of
growth will exhibit greater canine FA; while under the ‘developmental destabilization’
hypothesis, species with the same CL will exhibit similar canine FA. The second set of
controlled comparisons will compare primate species with similar growth durations, but
different CLs. Under the WOV hypothesis, canine FA is expected to be similar when
growth duration is the similar; while under the alternative ‘developmental destabilization’
hypothesis, canine FA should be greater in species assigned to a higher ranked CL.
Controlled comparisons between males of different primate species will allow
‘developmental destabilization’ and WOV to be evaluated separately in the primate
canine. The results of H3 will elucidate if either or both “developmental destabilization”
and WOV play role in canine FA.
Significance
This dissertation contributes to anthropological theory and to the discipline by
attempting to resolve critical questions surrounding what underlies the expression of FA
in both a non-sexually selected and a sexually-selected structure. Because FA is
commonly used as an indicator of stress experienced during development, knowing what
influences or is associated with FA of a morphological structure is essential to accurately
interpreting FA data (e.g.Townsend and Brown, 1980; Manning and Chamberlain, 1994;
Leung and Forbes, 1996, 1997; Roy and Stanton, 1999).
Human anterior teeth are known to vary in CFTs across populations (Reid and
Dean, 2006) and differences in CFTs may exist between some modern humans and
11
Neanderthals (Guatelli-Steinberg et al., 2005, 2009; Guatelli-Steinberg and Reid, 2008;
Smith et al., 2007; Smith et al., 2010). If growth duration is associated with FA, elevated
FA in one human population relative to another human population may not be due to greater environmental and/or genetic stress as suggested in past studies (e.g. Suarez,
1974; Doyle and Johnston, 1977; Perzigian, 1977; Corruccini et al., 1982). Rather, a longer duration of growth might have permitted a greater window of time to experience developmental perturbations, deviate from bilateral symmetry and express FA.
Associations between FA and growth duration have not been adequately explored within mammalian species and, in particular, among primate species. Through this study, examining a non-sexually selected and sexually selected structure in separate hypotheses, a framework that elucidates what underlies FA expression will be established.
If the variation of FA observed in the non-sexually selected primate first molar is found to be associated with growth duration this result suggests that prolonged periods of growth entail costs to the developmental system, as proposed by Van Valen (1985; 2003).
Furthermore, through determining if there is an association between prolonged growth periods and elevated FA, researchers will be better equipped to interpret FA data when making comparsion within and between populations or species.
Examining FA of the sexually dimorphic primate canine will determine if males of sexually dimorphic species exhibit greater canine FA than their female counterparts.
This project goes a step further and attempts to disentangle the potential effects of
‘developmental destablization’ and prolonged canine growth, both associated with sexual
12
selection. If developmental timing is found to be associated with DI of the primate canine, then these results could help explain why some studies do not find evidence for the ‘developmental destabilization’ hypothesis (e.g. Bjorksten et al., 2000; Lens et al.,
2002). By offering an opportunity to experience a greater number of perturbations, prolonged periods of growth of some, but not all, sexually-selected structures could account for the inconsistent results in studies of the relationship between FA and sexual selection. This is because some sexually selected structures may result from sex differences in growth rates as opposed to sex differeneces in growth duration.
Finally, primate species represent a strong sample in which to test these hypotheses because of the variation of CFTs, lateral EFTs, and life history schedules observed across the primate order. The intense focus on canine dimorphism over the last few decades has generated a wealth of information on the achievement of canine sexual dimorphism and on the variation that exists across primate species (Leutenegger and
Kelly, 1977; Harvey et al., 1978; Kay et al., 1988; Plavcan and van Schaik, 1992;
Greenfield, 1992; Plavcan, 1993; Plavcan et al., 1995; Thoren et al., 2006; Guatelli-
Steinberg et al., 2009). Such information permits the use of controlled comparisons in order to disentangle ‘development destabilization’ and WOV when examining canine FA.
Overall, the testing of these hypotheses will not only provide a greater understanding of the effects of growth duration on developmental instability, but also produce a more detailed picture of dental FA variation in primate species.
13
CHAPTER 2: LITERATURE REVIEW – THEORETICAL CONTEXT
Symmetry has long been a topic of interest for biologists, since body plans of many organisms’ exhibit one form or another of symmetry. Structures of bilaterally symmetrical organisms have the possibility of exhibiting three different forms of asymmetry, or the absence of symmetry (Ludwig, 1932; Van Valen, 1962; Leary and
Allendorf, 1988). Directional asymmetry (DA) occurs when a particular character has
greater development on one side of the bilateral plane than on the other (Van Valen,
1962). Several examples of DA are found in bilateral organisms, such as the mammalian
heart. Antisymmetry (AS) occurs in a population of individuals where most are
asymmetric, but it is random as to which side of the organism shows greater development
(Van Valen, 1962). The randomness of which side exhibits greater development results in
a bimodal distribution of (R-L) variation (Van Dongen et al., 1999). The most frequently cited example of AS are the claws of male fiddler crabs (genus Uca). One claw of the fiddler crab is always much larger than the other, but which claw (right or left) varies from male to male. The third form of biological asymmetry is fluctuating asymmetry
(FA), which occurs as small deviations from perfect bilateral symmetry that are random with respect to direction (Van Valen, 1962). Unlike the other two forms of asymmetry,
14
FA can be detected in a population when a structure exhibits left-minus-right differences, which are centered around a mean of zero (Van Valen, 1962).
A relationship between stability and symmetry is not a novel idea, but it is a
relationship that was not given significant attention until the mid-1900s when terms to
describe developmental processes were proposed (citations). Waddington (1942, 1953,
1961) introduced the terms developmental stability (DS) and developmental noise (DN)
to describe various processes associated with development. DS was originally defined by
Waddington (1942) as referring to the lack of disruption within the developmental system. Later researchers (Zakharov 1989, 1992; Palmer, 1994; Clarke, 1998; Auffray et al., 1999; Van Dongen and Lens, 2000) expanded on DS, providing slightly different definitions, but all agreeing that a buffering process deflecting developmental perturbations in order to maintain development along an ideal trajectory is part of achieving DS. Thus, morphological structures continue to develop along the ideal
developmental trajectory since DS minimizes the effects of developmental perturbations
on the developmental system through a variety of avenues including feedback regulation,
gene interactions, and biochemical interactions (Rutherford, 2000; Hartman et al., 2001).
However, when developmental perturbations overload the system, an organism may be
unable to buffer developmental perturbations resulting in a deviation or several
deviations from the ideal form.
The counterpart to DS is developmental instability (DI), which describes
processes that disrupt development (Waddington, 1942; Polak and Trivers, 1994;
Markow, 1995; Nijhout and Davidowitz, 2003). The outcome of developmental
15
perturbations displacing developmental off of an ideal trajectory is known as development noise (DN) (Table 1; Figure 1).
Developmental Process Factors Result 1. Buffering processes (other than Ideal developmental feedback mechanisms) trajectory is maintained or 2. Feedback mechanisms ‐ 'correct' ‘restored’; greater presence Developmental Stability or 'restore' development of developmental stability 3. Genetics ‐ increased relative to developmental heterozygosity, epigenetics (normal instability interaction) 1. Sensitivity to stress ‐ directional selection Ideal developmental 2. Genetics ‐ decreased trajectory is disrupted; heterozygosity, epigenetics (non‐ greater presence of Developmental Instability normal interactions) developmental instability 3. Environmental ‐ temperature relative to developmental variation, chemical pollutants, stability parasite load
Table 1: Factors Influencing the Presence of Developmental Stability and Developmental Instability within a Developmental System
Bilateral asymmetry has emerged as the most convenient variable to evaluate the integrity of developmental systems. All three forms of asymmetry have been proposed as indicators of DS being disrupted during development (e.g. Graham et al., 1993; Moller and Sawddle, 1997; Graham et al., 1998). Directional asymmetry has been argued not to be an indicator of DS because this form of asymmetry has an unknown amount of genetic variance that does not provide information about whether, and to what extent, development has been disrupted (Palmer and Strobeck, 1986, 1992, 2003; Palmer, 1994;
Klingerberg, 2003). Scholars have debated whether or not AS represents an indicator of
16
Resulting Intensity of Developmental Noise: Assessed through measuring a structure’s fluctuating asymmetry
Developmental noise decreases Developmental noise increases
Development does not deviate Development deviates from or is from or is restored to the ideal not restored to the ideal trajectory trajectory
Developmental instability Developmental instability decreases increases Developmental stability increases Developmental stability decreases
Able to buffer disruptions or Unable to buffer disruptions or development has been have development
‘corrected’/‘restored’ through ‘corrected’/‘restored’ through feedback mechanisms feedback mechanisms
Experiences developmental perturbation(s)
Figure 1: Relationships Among Developmental Processes
the processes that work to maintain the ideal developmental trajectory. Some scholars argue that AS does provide information processes that work to buffer development 17
because it is sensitive to perturbations experienced during development (Leary and
Allendorf, 1989; Graham et al., 1993; Moller and Swaddle, 1997; Rowe et al., 1997);
while others claim that AS has a genetic origin that is not known to be sensitive to
developmental perturbations (Palmer and Strobeck, 1986, 1992, 2003; Palmer, 1994;
Klingerberg, 2003). Since both DA and AS contain a genetic component that is believed to prevent these forms of asymmetry from communicating information on the integrity of
developmental processes, the only form of asymmetry left to provide such information is
FA. FA is the most commonly used indicator of whether or not developmental
perturbations have affected developmental trajectories. Fluctuating asymmetry is argued
to be affected by both DS and DI. Buffering processes that resist or minimize the effect
of developmental perturbations maintain the integrity of the developmental process of
DS. If DS is maintained, only a low degree of DN is present in the developmental system
leading FA of a morphological structure to be low or absent (Figure 1). Factors that
disrupt development (Figure 1) lead to an increase in DI because the integrity of DS is be
maintained. An increase in DI results in an increase in intensity of DN within the
developmental system, which is measured as elevated FA. FA estimates are therefore
measuring the amount of underlying DN since DN represents the combination of the
processes of DS and DI.
Description and Causes of Fluctuating Asymmetry
Because FA can be examined in an individual as well as in a population, two
definitions have emerged. Fluctuating asymmetry is defined as either the average
18
deviation of multiple structures within a single individual (Van Valen, 1962) or the deviation of a single structure within a population (Palmer and Strobeck, 1986, 2003;
Parsons, 1992). Since FA is commonly used measure to evaluate DN, it is important to discuss how FA relates to the definitions of developmental stability, developmental instability, and DN (Figure 1). When there is an absence of DN (e.g. when DS is acting to
buffer developmental perturbations) in a morphological structure, FA will be low or
absent. When there is an increase in DN (e.g. when DS fails to buffer developmental
perturbations and processes associated with DI disrupt development), FA will increase in
a morphological structure.
Fluctuating asymmetry values are used to assess the accumulation of
developmental perturbations that acted upon the developmental system causing a
deviation from the ideal developmental trajectory. The core idea behind FA as a measure
of DN is that, because the left and right body sides of a bilateral symmetric organism are
replicates of the same structure, they share the same genome and react to external effects
in development in similar ways within a given environment. In a system which is
deterministic, ideally, left and right body sides should be mirror images of each other because they develop under the same conditions both genetically and environmentally.
However, developing organisms do not develop within a deterministic system. Rather, because developmental perturbations act locally, perturbations do not occur bilaterally.
Left and right sides of a body structure separately experience and accumulate developmental perturbations. The resulting bilateral asymmetry, assessed through FA, is the manifestation of DN.
19
Elevated Fluctuating Asymmetry: A Measure of Increased Developmental Noise
When processes work to buffer disruptions, development is predicted to continue along an ideal trajectory. Stress acting upon the developmental system can cause disruptions resulting in physiological consequences including decreased DS resulting in increased DN. A handful of environmental and genetic stressors have been suggested as forms of developmental perturbations which have the potential to cause a deviation from ideal developmental trajectories.
Environmental Stressors
Studies investigating the relationship between elevated FA and environmental stress have focused on a wide range of taxa including: amphibians (e.g. Fox et al., 1961), insect (e.g. Beardmore, 1960; Parsons, 1962; McKenzie and Yen, 1995), and fish (e.g.
Leary et al., 1992). A variety of environmental stressors have been claimed to cause developmental disruption of morphological structures, resulting in increased DN (e.g.
Palmer and Strobeck, 1986; Parsons, 1990, 1992; Markow, 1995; Moller and Swaddle,
1997). Studies on temperature extremes, parasitic loads, pollutants, and environmental change comprise the majority of the literature addressing the relationship between FA magnitude and environmental stress.
Bilateral structures exposed to temperatures that deviate from a species’ specific temperature range have been reported to exhibit increased FA (Doyle 1975; Sciulli et al.,
1979; Zakharov, 1987, 1989, 1993; Fox et al., 1961; Clarke and McKenzie, 1992; Leary et al., 1992; Imasheva et al., 1997). For example, sand lizards were found to exhibit low
20
FA within a specific temperature range, but as soon as temperatures shifted outside of the
range, FA increased (Zakharov, 1987, 1989). However, not all species exhibit greater FA
when temperature fluctuations occur (Brana and Ji, 2000: Chapman and Goulson, 2000).
Parasite loads are another environmental factor that has received attention in the FA
literature (e.g. Hamilton and Zuk, 1982; Parsons, 1990; Zakharov et al., 1991; Polak and
Trivers, 1994). Hamilton and Zuk (1982) found that individuals able to resist parasites not only produced the most extravagant sexually selected structures, but also exhibited low FA relative to individuals not able to resist the parasites. Chemical pollutants have
been linked to elevated FA in some, but not all, studies (e.g. Valentine and Soule, 1973;
Zakharov and Yablokou, 1990; Fox et al., 1961; Graham et al., 1993; Mpho et al., 2001).
Several researchers (Ames et al., 1979; Zakharov, 1981; Jagoe and Haines, 1985) have found that fish species located in bodies of water with high concentrations of mercury and/or low pH exhibit increased FA . Studies investigating the link between asymmetry
and chemical pollution were motivated by ecological considerations and FA has since
emerged as important research tool for conservation.
Habitat destruction and changes to the environmental context are considered to be stressors capable of increasing FA in morphological structures (e.g. Manning and
Chamberlain, 1994). Moreover, measuring DN through FA has emerged as an assessment
tool in conservation studies because such information can be used to detect stressors
before a population is seriously affected (Leary and Allendorf, 1989; Parsons, 1992;
Clarke, 1995; Lens et al., 1998; Lens et al., 2002). In particular, Lens and colleagues
(2002) suggest that FA of a single-trait may be an effective early warning system for
21
conservationists. Finally, habitat location including optimal versus suboptimal and/or
non-optimal sites has also been connected to FA in some Passeriformes (e.g. perching
birds) (e.g. Swaddle and Witter, 1994; Carbonell and Telleria, 1998; Gonzalez-Guzman
and Mehlman, 2000) as well plants (e.g. Siikamaki and Lammi, 1998) and lizards (e.g.
Zakharov, 1981).
When environmental stress acts on the developing system, the organism is
believed to be more susceptible to developmental perturbations because the
developmental system is under pressure and does not perform at optimal levels. A
decrease in the system’s efficiency of buffering developmental perturbations (e.g. a
decrease in DS) coupled with an increase in DI (e.g. processes that disrupt development),
generates high DN as assessed through FA.
Genetic Stressors
When studying Drosophila, Mather (1953) and Thoday (1958) proposed that the
general characteristics of the genotype influence FA. Waddington (1957) argued that DS, or the processes that act to buffer developmental disruptions and maintain developmental along an ideal trajectory, could be influenced by genetic factors. In later studies investigating the influence of genes on FA, overall genotype characteristics were linked to the levels of homozygosity and genetic co-adaptation (e.g. Zakharov, 1987, 1989;
Clarke, 1993; Zakharov and Yablokov, 1997). Genetic co-adaptation describes the
beneficial interaction among genes located on different loci (Ridley, 1996).
22
Processes that buffer developmental perturbations (e.g. DS) are argued to be
positively correlated with levels of genetic heterozygosity (Lerner, 1954; Palmer and
Strobeck, 1986; Clarke & McKenzie, 1987; Leary and Allendorf, 1989; Zakharov, 1989;
Moller, 1992; Moller and Pomiankowski, 1993; Hutchinson and Cheverud, 1995; Roldan et al., 1998; Waldmann, 2001; Radwan, 2003). When homozygosity is high or co- adaptation is disrupted, FA is expected to be elevated (Palmer and Strobeck, 1986). The heterozygote is believed to be more capable of buffering developmental perturbations and remaining on the ideal developmental trajectory because of allelic dominance results in the masking of recessive alleles that might be deleterious to the developmental system
(Soule, 1979; Vrijenhoek and Lerman, 1982; Leary et al., 1984 1992; Mitton, 1993).
Thus, the heterozygote buffers perturbations occurring during development, which
reduces DN as assessed through FA.
When different genotypes are exposed to similar forms of stress, varying levels of
FA are produced suggesting that certain genotypes might be better suited to buffering
developmental stress (Zakharov, 1989). Researchers have suggested that heterozygosity
represents one way in which an organism buffers developmental perturbations (Clarke et
al., 1986; Ben-David et al., 1992; Roldan et al., 1998). Hutchinson and Cheverud (1995),
for example, found that Saguinus populations exhibiting less heterozygosity in cranial
morphology traits expressed elevated FA in comparison with populations with greater
heterozygosity. The effects of inbreeding on FA have also been investigated as a way to
investigate if heterozygosity is associated with buffering developmental perturbations
(Clarke et al., 1986; Ben-David et al., 1992; Roldan et al., 1998). Inbreeding has been
23
suggested as leading to an increase in homozygosity which reveals the presence of
deleterious recessive alleles (Clarke et al., 1986). Increased homozygosity is proposed to decrease an organism’s ability to buffer developmental perturbations leading to increased
DN as assessed through FA (Clarke et al., 1986; Moller and Swaddle, 1997).
An analysis conducted by Vollestad and colleagues (1999) demonstrated a weak association between heterozygosity and FA. This analysis, coupled with the fact that the relationship between DS and heterozygosity is supported mainly through correlational evidence, called into question whether or not heterozygosity contributes to the
maintenance of processes which buffer developmental perturbations (Clarke et al., 1992;
Clarke, 1993; Markow, 1995; Kruuk et al., 2003; Woolf and Markow, 2003). Because of conflicting evidence, the relationship between heterozygosity and DS continues to be investigated in the FA literature.
The field of epigenetics studies heritable changes in gene expression that do not involve a change to underlying DNA sequences, while epistasis describes the
“interactions between different genes,” (Cordell, 2002, p.2463). With respect to FA studies, epigenetics has been discussed as the “interactions of gene products,” (Evans and
Marshall, 1996, p.718) that cause a disruption to the developmental trajectory resulting in increased DN as observed through FA. FA has been described as representing an epigenetic measure of the organism’s ability of the organism to buffer genetic and environmental stressors during development (Leary and Allendorf, 1989, Parsons, 1990;
Hallgrimsson, 1999; Leamy et al., 2002). Leamy and Klingenberg (2005) hypothesize
“that FA has a pre-dominantly nonadditive genetic basis with substantial dominance and
24
especially epistasis,” (p. 13). Thus, the genetic basis of FA rests on several different
genes, which interact with each other. Furthermore, these authors postulate that genes in
these interactions are likely character-specific and are probably involved in the development of the character, but do not code directly for FA (Leamy and Klingenberg,
2005). Untangling the relationship between epigenetics and FA is still in its adolescence, but as the field of genetics expands so will our understanding of if FA represents a valid reflection of the outcome of “interaction of gene products” (Evans and Marshall, 1996, p.718).
Finally, feedback mechanisms that potentially restore bilateral symmetry by correcting deviations that caused one side of a bilateral trait to deviate from the ideal developmental trajectory during development have been researched (Emlen et al., 1993;
Kellner and Alford, 2003). Emlen and colleagues (1993) argue that bilateral structures experiencing developmental perturbations are likely to randomly drift during growth and that evolved ‘checking mechanisms’ should naturally hold any resulting FA within bounds. Since the work of these authors, others have expanded the idea leading to what is now known as the compensatory growth hypothesis (Swaddle and Witter, 1997).
Swaddle and Witter (1997) hypothesized that the interaction between sides of a bilateral structure should compensate for any asymmetries that arise during development, which are not part of the developmental plan. Two hypothetical mechanisms have been proposed for how unplanned asymmetries are compensated. First, the side of the bilateral character which is deviating from the ideal developmental trajectory is buffered in an attempt to suppress asymmetrical development (Emlen et al., 1993; Kellner and Alford,
25
2003, Dongen, 2006). Second, catch-up ‘growth’ occurs on the side of the bilateral
character that could not buffer developmental perturbations (Emlen et al., 1993; Kellner
and Alford, 2003, Dongen, 2006).
If researchers want to use FA as one way to assess the role of genetics in sustaining DS, then it is important to understand the genetic basis for FA in order to accurately utilized FA. Presently, the connection between genetic mechanisms and FA remains unclear. From what is known, though, little evidence supports a specific gene or
genes influencing FA. Rather, evidence lends support to the hypotheses that FA
magnitude is influenced by dominance and epistatic interactions among genes
(Klingenberg and Nijhourt, 1999; Leamy and Klingenberg, 2005).
Directional Selection, Fluctuating Asymmetry, and Sexually Selected Structures
Directional selection is a form of genomic stress proposed to have an association
with elevated FA in morphological structures. The homeostatic mechanisms of
morphological structures under strong directional selection, or selection that tends to
favor phenotypes at one extreme of the phenotypic range, are believed to be more
susceptible to stress because these developmental control mechanisms are relaxed in
order to expedite the response to selection (Waston and Thornhill, 1994). Relaxation of
homeostatic mechanisms results in reduced buffering during development and an increase
in DI (Parsons 1992; Palmer 1994; Watson and Thornhill 1994).
For a long period of time, all morphological structures were considered to be
equally sensitive to developmental perturbations. Over the last several decades, however,
26
some structures have been shown to be more sensitive to developmental perturbations as
well as exhibiting greater FA when compared to structures that are not as sensitive to
developmental perturbations (e.g. Soule and Cuzin-Roudy, 1982; Emlen et al., 1993;
Clarke, 1995, 2003; Bowyer et al., 2001). Clarke (2003) has described a structure’s
sensitivity to stress as ‘stress sensitivity’. Even though a procedure for accurately detecting a structure’s ‘stress sensitivity’ has yet to be designed, structures under the influence of sexual selection do appear to be more sensitive to stress than non-sexually selected structures (Moller and Pomiankowski, 1993; Swaddle et al., 1994). Directional
selection is argued to be the reason for why sexually selected structures are more
sensitive to stress and exhibit greater FA than non-sexually selected structures (Bubenik
and Bubenik, 1990; Moller, 1990, 1992, 1997; Moller and Pomiankowski, 1993; Watson
and Thornhill, 1994; Moller and Swaddle, 1997; Bjorksten et al., 2000). Sexually
selected structures are therefore considered more susceptible to ‘developmental
destabilization’ resulting from an increase in ‘stress sensitivity’ because of reduced
effectiveness of developmental processes to buffer developmental disruptions (Watson
and Thornhill, 1994). Thus, the relaxation of control mechanisms in structures under
intense directional selection is believed to account for sexually selected structures being
more prone to developmental perturbations and exhibiting greater FA (Moller 1990,
1992; Watson and Thornhill 1994; Moller and Swaddle 1997).
FA of sexually selected structures is proposed as providing reliable information to
females in mate choice and for males while assessing weaponry of potential opponents
(e.g. Moller, 1992; Thornhill et al., 1998). Moller (1990) found that, within a population
27
of barn swallows, the sexually selected structure of tail length exhibited greater FA than did the non-sexually selected structures. Additionally, Moller (1990) found that males with the largest sexually selected structure expressed the lowest FA, therefore signaling the highest quality among males. Several studies following Moller’s (1990) work found that females select males with lower FA, or more symmetric structures, than those males with greater asymmetry (e.g. Thornhill, 1992; Hunt and Simmons, 1997). Two years later, Moller (1992) investigated weaponry and reported that weapons (e.g. bird spurs and
beetle horns) exhibited greater FA than structures considered non-weapons. Moller’s
(1992) work also demonstrated that the individuals with the largest weapons had the lowest FA. Other researchers (Manning and Harley, 1991; Ditchkoff et al., 2000) have confirmed Moller’s (1990, 1992) findings that sexually selected characters exhibit greater
FA than do non-sexually selected characters and that there is a relationship between FA and the size of a structure.
With respect to structure size, condition-dependency has been proposed as the reason sexually selected structures show a negative relationship between FA and structure size (Moller and Pomiankowski, 1993). Individuals capable of producing large sexually dimorphic structures are hypothesized to be in prime condition with superior DS and, thus, capable of maintaining symmetrical sexually dimorphic structures. Individuals possessing the largest sexually selected structures have been reported as exhibiting the lowest FA because they successfully buffer developmental perturbations due to their prime condition (Moller, 1992; Moller and Swaddle, 1997). Therefore, FA of a sexually selected structure has been proposed as representing a phenotypic cue for females during
28
mate choice and males in conflict assessment (Moller, 1992; Thornhill, 1992; Hunt and
Simmons, 1997; Moller and Swaddle, 1997).
Evans and Hatchwell (1993) suggest that the symmetry of larger ornaments in some species (e.g. birds) is not only connected to an organism’s quality, but may be associated with the functional importance of the structure. In their paper, Evans and
Hatchwell (1993) argue that as ornaments become exaggerated, they are constrained into development, which is more symmetrical in order to maintain their function. Swaddle and
Witter (1997) further expanded on the relationship between FA, size, and structures of mechanical and aerodynamic importance (e.g. horns, wings, antlers). These authors argue that a trade-off exists between the life history benefits of fast growth (e.g. earlier age at maturity) and the possible consequences of developing asymmetry in a morphological structure such as lower reproductive fitness and/or mechanical problems (Swaddle and
Witter, 1997). Swaddle and Witter (1997) propose that in some species a developmental mechanism restores symmetry to a structure of functional importance as growth is ending. The organism, therefore, gains the benefits of fast growth such as an earlier age at
maturity as well as symmetry of the functional structure. In support of their assertion,
Swaddle and Witter (1997) studied the development of primary feathers in starlings and
demonstrated that the level of FA varies during development and, by the end of growth, the primary feathers exhibited low FA.
Because some sexually selected structures are also of functional importance, it is
important to consider what increased FA of these structures might be conveying. If a developmental mechanism that restores a structure’s symmetry does exist but the
29
structure which is sexually selected and of functional importance expresses elevated FA at the end of growth, the question becomes: “what does this mean?” Swaddle and Witter
(1997) among others (e.g. Moller, 1992) suggest that FA of some sexually selected structures might be part of an organism-wide signaling system designed to convey information to choosy females and competing males. If this is the case, elevated FA in sexually selected structures relative to non-sexually selected structures might be conveying information about quality. For example, Moller (1992) suggests that males of a species with the lowest FA but of the ‘highest’ quality might have a developmental mechanism that is able to restore symmetry or effectively buffer developmentally perturbations better than among the males of the species with elevated FA.
Although the above argument by Moller and fellow researchers (e.g. Moller,
1990, 1992; Manning and Harley, 1991; Moller and Hoglund, 1991; Moller and
Pomiankowski, 1993; Moller and Swaddle, 1997; Swaddle and Witter, 1997; Ditchkoff et al., 2000) has been supported in the literature, not all studies agree. Some studies have not found that sexually selected structures exhibit elevated FA relative to non-sexually selected structures or that a relationship between FA and size of sexually selected structures is true for all species (e.g. Balmford et al., 1993; Evans et al., 1995; David et al., 1998; Bjorksten et al., 2000; Hosken, 2001; Ditchkoff and deFreeze, 2010).
Furthermore, other studies claim that not only is it hard to test if choosy females and competing males are using a structure’s FA to assess quality, but that even when tested, not all studies have found that FA is used for quality assessment (e.g. Evans, 1993; Fiske et al., 1994; Tomkins and Simmons, 1995, 1996; Moller et al., 1996; Hunt and Simmons,
30
1997; Ligon et al., 1998). Due to these varying results, the question emerges as why some, but not all, studies find a relationship between sexually selected structures and elevated FA.
Growth Processes and Fluctuating Asymmetry
Size can be achieved through sex differences in growth rates or growth duration
(bimaturism) or a combination of both. In the literature, particularly on avian species,
focus has been placed on the tentative association between FA and growth rates. Little
attention, however, has been directed at an association between bimaturism and FA.
The skewed focus on growth rates may be a result of the type of species that
permeate the FA literature (e.g. birds, plants, and insects). Not only can structures of
these species (e.g. wings) be manipulated much easier than structures of mammalian
species, but rates of development are readily available for avian species. Birds, for
example, tend to experience rapid growth rates early in life (Ricklefs et al., 1968; Schew
and Ricklefs, 1998; Searcy et al., 2004). In addition to birds, growth rates of certain
plants are also well known leading to a handful of studies investigating the relationship
between growth rates and FA in plants (e.g. Martel et al., 1999; Lempo et al., 2000;
Kozlov et al., 2003).
In reviewing developmental stability and fitness, Moller (1997) states that
evidence from FA studies support an association between fast growth and symmetrical
phenotypes, but does not go into detail about why this association exists. Some later
studies (e.g. Lemp et al., 2000; Shakarad et al., 2001) also lend support for an association
31
between low FA and fast growth rates; while other studies have not found an association
between FA and growth rates (Searcy et al., 2004).
Although conflicting results complicate understanding the potential association
between growth rates and FA, the term ‘growth rate’ further complicates an evaluation of
this potential association. A distinction between growth rate and growth duration is not always made clear. For instance, it is unclear if the studies Moller (1997) cites as evidence for an association between fast growth rates and symmetrical phenotypes evaluated growth rate or growth duration. Also, phrases such “shifts in developmental timing,” (Kegley and Hemingway, 2007, p.47) and “rapidly growing” (Martel et al.,
1999, p.212) are commonly used to describe growth rather than distinguishing between growth rates and bimaturism or providing a clear definition of ‘growth’.
Furthermore, ‘growth rate’ has also been applied as a measure of fitness (Mitton and Grant, 1984; Thornhill, 1992). ‘Growth rate’ being viewed as a measure of fitness is not surprising because rapid growth entails ‘costs’ to the organism (Rose et al., 2009).
Some researchers (Perrin and Sibly, 1993; Iwasa, 2000; Rose et al., 2009) suggest that
‘costs’ due to rapid growth are ‘assumed in theoretical studies’ (Rose et al., 2009, p.
1379) and that quantifying such ‘costs’ is a challenging feat (Arendt 1997, 2003; Rose, et
al., 2009). With respect to FA studies, one can only assume that when growth rates are used to represent a measure of fitness they are referring to how well an organism maintained its growth rate while experiencing developmental disruptions; however, such an explanation is not supplied in these studies (Mitton and Grant, 1984; Thornhill, 1992).
32
Vrijenhoek (1985) suggested that relative to body structures that form over short periods of time, those that take longer to form have a greater “widow of vulnerability”
(WOV) to accumulate disruptions during development. Prolonged periods of growth could be providing an opportunity for developmental perturbations to accumulate and steer development off an ideal trajectory resulting in increased DN observed through elevated FA.
Studies addressing the association between FA and structures not under sexual selection have suggested that growth duration may be influencing FA expression in humans and non-human primates (Garn et al. 1967; Nass 1982; Saunders and Mayhall
1982; Kuswandari and Nishino, 2004; Guatelli-Steinberg et al., 2006). These studies
mention growth duration as a possible explanation for their results, but they do not
directly test the hypothesis that growth duration influences FA. For instance, Nass (1982)
suggested that growth patterns, specifically eruption times, may explain the pattern of FA
observed across the dentition of Japanese macaques while Hallgrimsson (1999) reported that bone FA of Macaca mulatta and Homo sapiens increased over the growth period.
Thus, early on in the growth period it appears that FA is low or absent but over the
duration of the growth period FA of bone increased (Hallgrimsson, 1999). Hallgrimsson
(1999) further states that his results demonstrate that because of their longer growth
period, slow growing mammals express greater FA (Hallgrimsson, 1999). From the
research performed thus far, it is possible that duration of growth could be providing a
WOV (Vrijenhoek, 1985) for structures forming over longer periods of time to
33 experience more developmental perturbations relative to those structures forming over shorter periods of time.
From a developmental perspective, sexual dimorphism or “any consistent difference between males and females beyond the basic functional portions of the sex organs,” (Wilson, 1975, p322), can be achieved through either differences in growth rates, bimaturism, or through a combination of both. In primates, body size dimorphism is achieved through both sex differences in growth rates and bimaturism. For example, body size dimorphism in Cebus apella is achieved through bimaturism, or a mode of growth in which males and females grow for a different amount of time (Leigh, 1992). A large component of body size for Pan troglodytes is achieved through differences in the rate of growth between males and females (Leigh, 1992). Canine dimorphism in primates is primarily achieved through bimaturism (e.g. Schwartz and Dean, 2001; Guatelli-
Steinberg et al., 2009).
Within the FA literature, sexually selected structures have, in some studies, been found to express elevated FA relative to non-sexually selected structures (e.g. Moller,
1990, 1992; Uetz et al., 1996; Hunt and Simmons, 1997; Koshio et al., 2007). However, this pattern does not hold across all studies investigating the relationship between FA and sexually selected structures (e.g. Balmford et al., 1993; Evans et al., 1995; David et al.,
1998; Bjorksten et al., 2000; Hosken, 2001; Ditchkoff and deFreeze, 2010).
The different modes of dimorphic growth – bimaturism and growth rate – might explain why some studies lend support to the hypothesis that sexually selected structures express elevated FA more frequently than non-sexually selected structures, while other
34
studies do not find evidence to support the hypothesis. Bimaturism, in particular, is
connected to Vrijenhoek’s (1985) WOV since it describes a mode of growth in which
males and females grow for a different amount of time permitting the opportunity for one
sex to accumulate more perturbations during development. Bimaturism might be a better
approach to examining if and to what extent growth duration influences FA since the
relationship between growth rate and FA is unclear and because growth rates could be
hard to quantify in species other than birds and plants. Therefore, bimaturism provides a
strong framework from which to examine if a relationship between developmental timing
and FA exists.
Growth Duration and Fluctuating Asymmetry
The general trend across the primate order is that strepsirrhines have shorter
periods of growth relative to the growth periods of apes and humans (e.g. Smith, 1989;
Godfrey et al. 2001). Moreover, within anthropoids, large-bodied species tend to grow for
longer periods of time than small-bodied species (e.g. Leigh and Shea, 1995). While prolonged growth periods may be advantageous in some respects (e.g., affording time for learning), they may also influence a structure’s developmental stability resulting in a
decreased ability to buffer developmental perturbations. Hallgrimmson (1995) has
demonstrated that within mammals, species of Macaca and humans tend to show
elevated FA in osteometric traits in comparison to mammals with shorter periods of
growth. Thus, prolonged growth periods for some mammalian species, such as slow-
35
growing primates, could entail a cost to the developmental system in which sources of developmental stress affect body structures resulting in elevated DN.
Although some of studies on FA and non-sexually selected teeth mention growth duration as a possible explanation for their results the hypothesis that growth duration influences FA expression has not been directly tested (e.g. Harris and Nweeia, 1980;
Saunders and Mayhall, 1982; Kuswandari and Nishino, 2004).
Neandertal dental FA has been of interest to researchers for several decades (e.g.
Suarez, 1974; Doyle and Johnston, 1977; Frederick and Gallup, 2007; Barrett et al.,
2012). The study by Frederick and Gallup (2007) found that modern humans and
Neandertals exhibited greater average dental FA than the great apes. These authors argue that, because dental growth is not strongly influenced by environmental factors and that growth of the dentition is dependent on the genome, genetic differences between the
species in their study is the primary reason for the observed FA values (Frederick and
Gallup, 2007). Frederick and Gallup (2007) appeared to have grouped all tooth types
together when calculating dental FA, not taking into account that growth durations might
vary among the tooth types (Macho, 2001; Schwartz and Dean, 2001; Reid and Dean,
2006), that sex differences exist in some tooth types, such as the canine (e.g. Schwartz
and Dean, 2001), or that differences in growth duration could exist among hominin
species (e.g. Dean et al 2001; Guatelli-Steinberg, 2007; Smith et al., 2011). Furthermore,
it is unclear if modern human and Neandertal dental FA differ. A more recent study
found evidence for Neandertals experiencing greater developmental stress relative
prehistoric modern human populations (Barrett et al., 2012). Despite the data on
36
Neandertal (e.g. Guatelli-Steinberg, 2007; Smith et al., 2011), human (e.g. Reid and
Dean, 2006), and great ape (e.g. Reid et al., 1998; Schwartz and Dean, 2001) dental
growth available in the literature, both of these studies (Frederick and Gallup, 2007;
Barrett et al., 2012) did not address dental growth data in relation to their FA results,
leaving several questions unanswered.
In studies addressing non-human primate and dental FA, Nass (1982) suggested
that eruption times may explain the pattern of FA observed across the dentition of rhesus
macaques, but did not directly state how or why eruption time explained the observed
FA. In skeletal series of rhesus macaques, Hallgrimsson (1999) found an ontogenetic pattern of asymmetry in which FA of bone was shown to increase over the growth period.
If, and to what extent, growth duration is associated with elevated FA in primates has yet to be tested systematically across the primate order or even formally tested within a primate genus. Because FA is used to infer information on developmental precision, individual quality and stress experienced during development and since crown formation times are known to vary both across and within species, it is important to understand what underlying variables might affect the expression of FA in morphological structures.
In addition to explaining the pattern observed in non-sexually selected structures and FA, growth duration could explain the association between FA and sexually selected structures. Sexually selected structures may be destabilized by directional selection. They are particularly prone to exhibiting elevated FA leading to an enhanced ability to convey
information to choosy females and competing males about male fitness/quality (Moller
1990; Watson and Thornhill 1994). The primate canine has been given as one example
37
(Manning and Chamberlain 1993): males of species with sexually dimorphic canines tend to exhibit greater FA than males of species with sexually monomorphic canines.
However, the mechanism which explains the association between canine sexual dimorphism and FA may not be the destabilization of developmental processes associated with sexual selection. Research on adult canine sexual dimorphism in primate species demonstrates that canine dimorphism is achieved through prolonged periods of canine growth in males relative to females (Swindler et al., 1982; Sirianni and Swindler, 1985;
Swindler, 1985; Schwartz et al. 1999; Schwartz and Dean 2001; Schwartz et al., 2001;
Guatelli-Steinberg et al. 2009). Thus, instead of ‘developmental destabilization’ being the reason for an increase in FA magnitude observed in primate canines, male canines of sexually dimorphic primate species may simply have a greater opportunity to accumulate deviations from symmetry because their canines take longer to form than canines of primate males of monomorphic species (Schwartz et al. 1999; Schwartz and Dean 2001;
Schwartz et al., 2001; Guatelli-Steinberg et al. 2009). This opportunity to accumulate more developmental perturbations because of prolonged growth periods was initially described by Vrijenhoek (1985) as the “window of vulnerability”.
Filling in the Gaps
FA of a morphological structure has been of interest to researchers for decades because of the hypothesis that FA conveys information about quality, fitness, and stress experienced during development. The research interest in FA has also generated several unanswered questions including if duration of growth affects the magnitude of FA.
38
Information on how canine dimorphism is achieved in primate species has expanded through studies on a variety of developmental factors, including but not limited to, histological markers of tooth growth. Schwartz and Dean (2001) found that large- bodied apes achieve canine dimorphism through bimaturism. Using the spacing and number of perikymata on the mandibular canine crown, Guatelli-Steinberg and colleagues (2009) found that males in three catarrhine genera and two platyrrhine genera form their large canines over longer periods of time relative to females of the same genera. Recent advances in the study of FA coupled with the information on crown formation times in primate species provide an opportunity to examine the potential association between growth duration and FA.
The mechanism(s) which underlie the association FA and structures which are not sexually selected are of interest because FA is commonly used as an indicator of stress placed upon the system during development. This dissertation aims to elucidate the mechanism(s) which underlie the association between FA and structures that are and are not influenced by sexual selection through hypotheses that are designed to test the possible effects of growth duration on FA of morphological structures. Moreover, this dissertation aims to tease apart effects of ‘developmental destabilization’ and growth duration by evaluating FA expression in a sexually selected structure. This study will examine catarrhine and platyrrhine species and make comparisons within and between species in order to gain a full view of if, and to what extent, growth duration influences
FA expression. Although the results of study are important for anthropology, particularly since FA is used to assess stress experienced during development of fossil specimens
39
(e.g. Neandertals and Australopithecus), the results of this study will be applicable
beyond the discipline of anthropology. Disciplines such as evolutionary biology, botany,
genetics, ornithology, and entomology could investigate if and to what extent growth
duration is influencing FA in structures in species typically exmained in their disciplines.
All of these disciplines commonly use FA as an indicator of stress or as a signal of
quality in both extant and extinct species (e.g. Roy and Stanton, 1999; Bateman, 2000;
Carchini et al. 2000; Goddard and Lawes, 2000; Hogg et al., 2001; Prentice et al., 2008;
Dongen et al., 2009). Typically, it is assumed that a population expressing a greater FA
relative to another was under greater stress during development. However, if growth
duration is shown to influence the expression FA, then greater FA in one population
relative to another may not be due to greater stress experienced during development but
rather to a longer period of growth in which to experience stress. Therefore, reearchers
can apply of results of this study when formulating hypotheses on species within their
disciplines when comparing between species or within a species.
Overall, the results of this study will provide critical information towards not only understanding FA in primate dentition but also the potential influence of growth duration on FA expression. If a relationship is found, more than likely this relationship is not
constrained to the a single mammalian order or even within the mammalian taxa. Thus,
researchers in other disciplines will be able to apply the results of this study to their own investigations of FA.
40
CHAPTER 3: LITERATURE REVIEW – ESTIMATING FLUCTUATING
ASYMMETRY
Fluctuating asymmetry (FA) or subtle, random, differences between the left and
right sides of a bilateral structure (Ludwig, 1932; Van Valen, 1962; Palmer and Strobeck,
1986) represents the outcome of developmental noise (DN). FA estimates are argued to
provide insight into the amount of stress placed upon the developing system and are applied by researchers to infer information about developmental precision of an individual or population and the ability of the developmental system to resist stress experienced during development.
There are several issues involved in quantifying FA including procedures to validate FA data, indices (or equations) to calculation of FA, and analyses to test for differences within and between individuals, populations, or species. This chapter reviews key issues associated with quantifying FA.
Generally, FA measurements are non-invasive and easily attainable (e.g. from field studies or museum collections). Despite the ease of taking FA measurements, FA studies require careful and accurate experimental design including a comprehensive measurement protocol and appropriate statistical analyses. A primary reason why much forethought is necessary for FA studies is because as a biological signal, FA is
41 exceedingly small, on the order of approximately 1-2% of trait size or less (Palmer and
Strobeck, 1986; Palmer, 1994, 1996; Lens et al., 2002). For this reason, repeated measures must be done and correctly analyzed to ensure that the observed asymmetry represents ‘real’ FA and not errors that occurred during measurement or analysis (Palmer,
1994; Palmer and Strobeck, 2003). Measurement error (ME) has the potential to greatly affect the final calculation of FA if measurement and analysis protocols are not established. In addition to ME, other forms of asymmetry and the morphological structure being examined including structure type and size are primary issues affecting both measurement and analysis of FA.
Several steps are involved in quantifying FA beginning with validating the data and concluding with using the appropriate index to calculate FA (e.g. Palmer and
Strobeck, 1986; Palmer, 1994). Depending on the organism under investigation as well as the structures being measured for FA, the methodological protocol might differ slightly from study to study. The discussion that follows outlines the essential recommendations put forward by Palmer (1994) and colleagues (Palmer and Strobeck, 1986; Parsons, 1990;
Palmer and Strobeck, 1992; Swaddle et al., 1994; Merila and Bjorklund, 1995; Leung and
Forbes, 1996; Swaddle and Witter, 1997; Whitlock, 1998; David et al., 1999; Van
Dongen, 1999; Kellner and Alford, 2003; Palmer and Strobeck, 2003; Knierim et al.,
2007) for a methodological and statistical protocol that, if adhered to, will result in FA estimations that most closely represent ‘real’ FA.
42
Validating the Data Prior to Calculating Fluctuating Asymmetry
The following discussion addresses five issues viewed as essential considerations in collecting data for FA studies and in assessing as clearly as possible ‘real’ FA of either
individuals or populations. Palmer (1994) and other scholars (Palmer and Strobeck, 1986;
Parsons, 1990; Palmer and Strobeck, 1992; Swaddle et al., 1994; Merila and Bjorklund,
1995; Leung and Forbes, 1996; Swaddle and Witter, 1997; Whitlock, 1998; David et al.,
1999; Van Dongen, 1999; Kellner and Alford, 2003; Palmer and Strobeck, 2003; Knierim et al., 2007) have addressed these issues in great detail because not all studies have been performed in a rigorous manner or accounted for the sensitivity of FA to several factors
(e.g. Siegel and Doyle, 1975; Felley, 1980; Balmford et al., 1993; Tomkins and
Simmons, 1995; Veiga et al., 1997; Anne et al., 1998; David et al., 1998; Bjorksten et al.,
2000). After Palmer’s (1994) significant paper on FA methods, a renewed interest in
employing both measurement and analytical procedures appropriate for the study of FA
was generated. Despite the focus on appropriate procedures, inconsistencies still persist
in the literature leading several researchers to suggest that a strict methodological
protocol be established in order for both inter- and intra-disciplinary comparisons of FA estimates (e.g. Green, 1984; Palmer and Strobeck, 1986; Palmer, 1994; van Dongen et al.,
1999).
The following discussion describes the influence that factors such as ME, other forms of asymmetry, trait size, and trait shape can have on data collection and FA calculation. These descriptions are summaries from the work of Palmer and colleagues
(Smith et al., 1982; Green, 1984; Palmer and Strobeck, 1986; Palmer, 1992, 1994; van
43
Dongen et al., 1999; Palmer and Strobeck, 2003; Knierim et al., 2007). Additionally, the
discussion that follows addresses methodological recommendations of Palmer and
colleagues (Smith et al., 1982; Green, 1984; Palmer and Strobeck, 1986; Palmer, 1992,
1994; van Dongen et al., 1999; Palmer and Strobeck, 2003; Knierim et al., 2007) for
studies investigating FA.
Measurement Error
FA estimations tend to be small (Palmer and Strobeck, 1986; Palmer, 1994, 1996;
Lens et al., 2002) resulting in a similar magnitude to ME (Palmer, 1996). The sensitivity of FA to ME results in inflated FA estimations in both individual and population level between-sides variance (Green, 1984; Palmer and Strobeck, 1986; Palmer, 1994; Merila and Bjorklund, 1995; Swaddle et al., 1995; Bjorklund and Merila, 1997; Van Dongen et al., 1999; Van Dongen, 2000). For instance, scholars have reported ME as accounting for high percentages of the total between-sides variation that is commonly considered to be
representative of FA (Fields et al., 1995; Van Dongen and Lens, 2000). If ME is not controlled for, it is difficult to determine if FA estimates represent biological asymmetry or a combination of biological asymmetry and the variance resulting from ME.
The most frequently recommended step for minimizing ME is to repeatedly
measure right and left structures (Lundstrom, 1960; Green, 1984; Palmer and Strobeck,
1986; Knierim et al., 2007; Palmer and Strobeck, 2003). The recommended minimum
number of repeated measures is two (Moller and Swaddle, 1997). If ME is suspected to
be high for a particular morphological structure, the suggested number of repeated
44
measures increases to 4 to 6 (Swaddle et al., 1994; Van Dongen et al., 1999). Repeated
measures should be taken at different times to ensure that measurement replication is
‘blindly’ performed. When taking measurements ‘blindly’, researchers should measure
their samples at random so that specimens of similar species or sex are not being
measured in order (Knierim et al., 2007). Taking the challenge of sample access into
considation, David and colleagues (1999) suggest that if repeated measures need to be
taken during the same measuring session that methodological protocols are established to
reduce ‘session bias’.
Although repeated measures are argued to be the best procedure for decreasing
the risk of ME being incorporated into FA by separating out ME from ‘real’ FA (Green,
1984; Merila and Bjorklund, 1995; Palmer and Strobeck, 1986; Palmer, 1994; Swaddle et
al., 1994; Van Dongen et al., 1999; Van Dongen, 2000; Palmer and Strobeck, 2003),
applying a measuring technique with a low probability of ME should be built into the
measurement protocol (Knierim et al., 2007). To achieve this recommendation, proper
training in the project’s equipment is needed and the same person should perform both
initial and repeated measurements (Palmer and Strobeck, 1986, 2003; Knierim et al.,
2007). Moreover, the equipment should be appropriate for the traits being measured in
order to ensure that measurements are taken accurately. Accounting for precision in measurements through training of researchers and equipment selection will reduce the potential for ME incorporation during the data collection phase. This is an essential recommendation for all studies, but particularly those studies where structures may differ in size (Palmer and Strobeck, 2003). Taking this recommendation into account when
45 designing a methodological protocol will decrease the likelihood of ME being part of the
FA estimation.
Forms of Asymmetry
Bilateral organisms possess asymmetries which are adaptive and as such are considered to have a genetic basis (Palmer, 1994; Klingerberg, 2003). These adaptive asymmetries, which include directional asymmetry (DA) and antisymmetry (AS), are argued to contain little to no information which would be useful for interpretation of a morphological structure’s DN because of their genetic origin (Palmer and Strobeck,
1992; Palmer, 1994; Klingerberg, 2003; Palmer and Strobeck, 2003).
Accounting for different forms of asymmetry will avoid skewed FA estimates since both DA and AS have the potential to alter FA estimates (Palmer and Strobeck,
1986, 1992; Palmer, 1994; Van Dongen et al., 1999; Klingenberg, 2003; Palmer and
Strobeck, 2003). For instance, a morphological structure may be characterized by DA resulting in an inflated FA value. A distribution of the right minus left differences (R-L) that appears normal could have underlying AS resulting in a decrease in FA expression
(Palmer and Strobeck, 1986, 1992, 2003). Therefore morphological structures exhibiting
DA or AS will not yield reliable information on the underlying DN since FA expression is distorted by the presence of another form of biological asymmetry (Palmer and
Strobeck, 1992; Palmer, 1996).
Some researchers, however, argue that FA, DA, and AS are interrelated (Graham et al., 1993, 1998; Leamy, 1999) and that all three types of asymmetry can provide
46 information about DN (Leary and Allendorf, 1989; Graham et al., 1993; Moller and
Swaddle, 1997; Rowe et al., 1997). DA has been found in several studies on wing asymmetry (Simmons and Ritchie, 1996; Smith et al., 1997; Klingenberg et al., 1998;
Goulson et al., 1999; Windig and Nylin, 1999; Pither and Taylor, 2000) and recorded in mouse jaws (Leamy et al., 1997; Klingenberg et al., 2001). Studies on insect wings suggest that when DA is present it is not consistent in direction (e.g. left or ride side), between species, or within species among populations (Simmons and Ritchie, 1996;
Klingenberg et al., 1998; Goulson et al., 1999). The results of these studies imply that DA does not convey information about DN (Simmons and Ritchie, 1996; Klingenberg et al.,
1998; Goulson et al., 1999). After the random variation representative of DN is factored out DA provides information “only for the systematic difference between sides,”
(Klingenberg, 2003, p.17). DA cannot provide information on developmental disruptions resulting from developmental perturbations because it represents the mean asymmetry of a sample (Klingenberg, 2003).
With respect to AS, the resulting pattern is such that approximately half the population has a larger right side while the other half develops a larger left side (Palmer,
1994). Because AS has a genetic origin (Klingerberg, 2003) it is believed that AS is not associated with processes that maintain or disrupt the stability of the developmental system.
Since these two types of asymmetry are believed to represent a genetic component of the developmental trajectory, their statistical significance should be carefully evaluated in order to avoid misrepresentation of FA. The influence of DA on FA has received
47
considerable attention (e.g. Graham et al., 1993; Leung and Forbes, 1997; Klingenberg et
al., 1998; Pither and Taylor, 2000; Kark et al., 2004). Morphological structures exhibiting
DA complicate both analysis and interpretation of FA (Palmer and Strobeck, 1986, 2003;
Klingenberg, 2003; Van Dongen, 2006). As noted, analyses are complicated since a number of FA indices are inflated by DA (Palmer, 1994; Palmer and Strobeck, 1986).
Also because left and right of structures associated with DA develop differently, the influence of developmental perturbations may not be identical (Klingenberg, 2003). To avoid such problems, either structures expressing significant DA should be removed from the study (Palmer and Strobeck, 1992; Palmer, 1994; Klingenberg, 2003; Van Dongen,
2006) or the asymmetry of interest should be shifted to DA (Klingenberg et al., 1998;
Lens and Van Dongen, 2000; Pither and Taylor, 2000; Van Dongen, 2006). Cases where
very slight DA is present could also result in statistical problems during analysis
particularly in large sample sizes (Palmer and Strobeck, 2003). In such cases, factoring
out DA before FA analysis is favored over removing the morphological structure(s) from
the study sample. It is possible to statistically correct for mean DA by subtracting the
average amount of asymmetry from individual signed asymmetry values (Palmer and
Strobeck, 1986, 2003; Van Dongen, 2006). Statistically correcting for DA has emerged as
an essential step in any FA analysis where DA is suspected to be a factor or in studies
where it is unclear if DA could be present (Palmer, 1994; Palmer and Strobeck, 1986;
Palmer and Strobeck, 2003; Knierim et al., 2007).
To account for potential confounding effects prior to FA calculation, it is
necessary to test for AS (Palmer and Strobeck, 2003; Knierim et al., 2007). AS can cause
48 platykurtosis, or a distribution of right minus left differences (R-L) which is flatter than a normal distribution, in the data output (Van Dongen et al., 1999; Van Dongen, 2006).
The frequency distribution associated with platykurtosis exhibits a valley between two, equally sized peaks which are the same distance from zero (otherwise known as bimodal distribution). The flat valley part of the distribution which is centered on zero does not contain observations. Both bimodal and multimodal distributions are examples of platykurtosis. In a multimodal distribution, there are several valleys as well as equally sized peaks on either side of the valley. Testing for kurtosis in the data set will detect the presence of AS.
Type and Number of Structures
Generally there are four important factors to consider when determining if a structure is appropriate for an FA study. The structure should be: 1) accessible; 2) have identifying landmarks or dimensions; 3) permit repeated measurements; and 4) not exhibit DA or AS. Accessibility of a structure is important because pieces of equipment, such as calipers, need to be correctly positioned on the structure to obtain an accurate measurement. If a structure is easily accessible, the task of taking repeated measurements will not be difficult or time consuming since the structure can be accurately positioned each time a measurement is taken. Structures should have identifiable landmarks (e.g. tip of a bird’s feather) or dimensions (e.g. buccal-lingual and basal circumference).
Landmarks or dimensions are helpful because they identify where a measurement needs to be taken permitting not only repeated measurements to be accurately taken but for
49
comparisons to be made between different species. For example, antler landmarks are
similar across several species (citation).
Taking repeated measurements is easier on structures that are accessed without
difficulty and present identifiable landmarks or dimensions. Successfully repositioning
equipment to conduct repeated measurements is important because the inclusion of
repeated measurements in the data set factors out ME (Green, 1984; Palmer and
Strobeck, 1986; Palmer, 1994; Swaddle et al., 1995; Van Dongen et al., 1999; Van
Dongen, 2000; Palmer and Strobeck, 2003; Knierim et al., 2007). Repeated measurements are also useful in assessing intra-observer error (e.g. Smith et al., 1982).
The type of structure selected is extremely important in designing a FA study
(Soule and Cuzin-Roudy, 1982; Palmer and Strobeck, 1986; Palmer, 1994; Leung and
Forbes, 1996; Palmer and Strobeck, 2003; Knierim et al., 2007). Although every
specimen has a number of structures available for examination, some structures may be
better suited for examining underlying DN than others.
In the early FA literature, all morphological structures were considered to be
equally sensitive to developmental perturbations. Over the last several decades, research
has demonstrated that some structures are not equally sensitive to developmental
perturbations. Rather, certain structures may be more sensitive to developmental
perturbations resulting in those structures exhibiting elevated FA relative to structures
believed not to be sensitive to developmental perturbations (e.g. Soule and Cuzin-Roudy,
1982; Emlen et al., 1993; Clarke, 1995; Bowyer et al., 2001; Clarke, 2003). Clarke
(2003) has termed a structure’s sensitivity to stress as ‘stress sensitivity’. Structures of
50
functional importance and those under intense directional selection are proposed as
having a high ‘stress sensitivity’ to developmental perturbations, suggesting that these
structures are more likely to accumulate perturbations, deviate from the ideal developmental trajectory, and express DN.
In the initial steps of formulating a methodological protocol, whether a structure
is, or could be, sensitive to stress should be taken into account (Moller, 1990, 1992;
Moller and Pomiankowski, 1993; Swaddle et al., 1994; Clarke, 1995, 2003). A procedure for accurately detecting a structure’s ‘stress sensitivity’ has yet to be developed; however, there does appear to be a pattern for both structures of functional importance (Clarke,
1995, 2003) and those under intense directional selection, such as sexually selected structures (Moller and Pomiankowski, 1993; Swaddle et al., 1994).
Structures of functional importance (e.g. tail feathers, wings of birds and insects) might be more sensitive to stress during early stages of development because growth rate is accelerated leaving the processes that buffer developmental perturbations without sufficient time to react to developmental perturbations (Swaddle eta l., 1995; Swaddle and Witter, 1997). These structures are therefore more likely to exhibit greater FA early in development relative to later in development. Although there are studies that lend support to avian tail feathers exhibiting greater FA during early development and lowered
FA as development proceeds (e.g. Swaddle and Witter, 1997) the evidence to support a relationship between growth rates and elevated FA is not well supported outside of avian
FA studies (Shakarad et al., 2001; Searcy et al., 2004).
51
In addition to structures of functional importance, sexually selected structures are
also believed to be more prone to developmental perturbations resulting in greater FA in
sexually selected structures relative to non-sexually selected structures (Moller, 1990,
1992; Moller and Hoglund, 1991; Moller and Pomiankowski, 1993; Tomkins and
Simmons, 2003). The relaxation of control mechanisms in structures under intense
directional selection may account for some sexually selected structures exhibiting greater
FA than non-sexually selected structures (Thoday, 1958; Watson and Thornhill, 1994).
Wear and Damage
It has long been recognized that wear or damage to a structure can skew FA estimates (e.g. Doyle and Johnston, 1977; Townsend and Garcia-Godoy, 1984; Palmer and Strobeck, 1986; Moller, 1991; Swaddle and Witter, 1994, 1995; Swaddle, et al.,
1996). FA estimates based on data collected from morphological structures exhibiting wear and/or damage do not represent differences between R-L sides that are due to underlying DN. Rather, these FA estimates represent either asymmetry resulting from wear or a combination of asymmetry partly from the underlying DN as well as asymmetry resulting from wear and/or damage. Asymmetry that results from wear and/or damage is not associated with stress upon the developmental system and thus does not reflect underlying DN. Furthermore, because equivalent degrees of wear do not necessarily occur bilaterally in several morphological structures, such as teeth and antlers, it is possible that FA of a morphological structure with wear does not reflect FA resulting from underlying DN.
52
Because the amount of asymmetry due to wear and/or damage is difficult to estimate and factor in after measuring is complete, including an examination for wear and/or damage in the methodological protocol is an important consideration (Cuthill et al., 1993; Moller, 1993; Swaddle et al., 1994). Standards for assessing wear and/or damage of a morphological structure are generally discipline-specific. Regardless, the methodological studies on FA (e.g. Palmer, 1994; Palmer and Strobeck, 2003) recommend visually inspecting morphological structures for wear or damage prior to measuring. If wear or damage is found to be extensive or would prevent accurate measurement, then the structure should not be measured. Approximal wear is an example of extensive wear because it causes the measurement point of the mesio-distal diameter to become undefined (Kieser, 1990; Hillson, 1996).
Several studies have incorporated an assessment of wear and damage into their methodological protocol. For example, teeth exhibiting damage including caries and attrition have been removed from the sample (Angelopoulou et al., 2009). Scholars examining antlers for FA remove damaged or extremely worn antlers from their studies to avoid skewed asymmetry results (e.g. Ditchkoff and deFreese, 2010). Similar to antlers, damaged avian wings are removed from the sample prior to measurement
(Swaddle and Witter, 1997).
Structure Size
The relationship between FA and structure size has received considerable attention in the literature because FA is argued to vary with trait size leading to the
53
possibility of FA analysis and interpretation being complicated if size variation is not
considered (Palmer and Strobeck, 1986; Palmer, 1994; Palmer and Strobeck, 2003). One of the most important reasons to take size variation into account is because when the
same ‘level’ of DN is being experienced by morphological structures of different sizes
the larger morphological structure is expected to exhibit greater asymmetry on average
compared to smaller structures (Palmer and Strobeck, 2003; Knierim et al., 2007). For a
multiple-structure FA study, if structure size is not corrected for prior to FA analysis the
resulting average FA reflects not only the underlying DN but also any size-dependent
variation (Palmer and Strobeck, 1986; Palmer, 1994; Rowe et al., 1997; Palmer and
Strobeck, 2003).
A positive relationship between FA and structure size has been noted (e.g. Moller
and Swaddle, 1997); however, the relationship of FA and structure size is not always
positive. Sexually selected structures have exhibited a negative FA-structure size
relationship which differs from the positive relationship typically expressed by non-
sexually selected structures (Moller, 1990; Manning and Hartley, 1991; Moller and
Hoglund, 1991; Moller, 2000). To further complicate the relationship between FA-
structure size, not all sexually selected structures show a negative relationship between
FA and structure size (e.g. Balmford et al., 1993; Tomkins and Simmons, 1995; Jennions,
1996; Bjorksten et al., 2000). These opposing results might be connected to the size of
the sexually selected structure; however, a definitive explanation for why some but not all
sexually selected structures exhibit a negative relationship between FA and structure size
has yet to be presented.
54
Scholars recommend correcting for trait size effects through appropriate models
to ensure that differences in FA are preserved in the measurements (Palmer and Strobeck,
2003; Leung, 1998; Graham et al. 2003; Van Dongen et al., 2005; Van Dongen and
Moller, 2007). Testing for size dependence prior to testing for FA is recommended because the influence of size-dependence in asymmetry calculations could result in a leptokurtosis frequency distribution as well as obscuring any presence of AS in the distribution. Several FA indices – originally constructed by Palmer and Strobeck (1986) and Palmer (1994) – correct for structure size effects by expressing deviations as a proportion of structure size.
Unlike biological variation, ME is unaffected by the average differences between sides of a morphological structure (Palmer and Strobeck, 2003). Therefore, the presence or absence of ME needs to be determined prior to making size-adjusted estimates of FA.
If ME is minimal, applying |ln / | to scale out size variation is recommended
(Palmer, 1994). Leung (1998) suggests that relative FA be used when the coefficient of variation in structure size is small. If, however, ME exceeds 10% of the estimated average of the difference between the sides, Palmer and Strobeck (2003) suggest either 1) using a t-test if the slope of the proportional FA versus structure size is steeper than the slope of the proportional ME versus structure size; or 2) dividing the size range into three
or more size categories and then applying the appropriate index to factor out ME from
each category.
55
Indices and Analysis: Fluctuating Asymmetry
Choosing the appropriate FA index is one part of the methodological protocol and
involves having adequate information concerning sample size, presence of measurement
error, other forms of asymmetry, and the possible influence of trait size. Understanding the limitations of one’s study is important in selecting a FA index since many factors
such as ME can influence the resulting FA estimate. Therefore, selecting a FA index that best suits the needs of the study being conducted is an essential component to any FA study. Furthermore, the choice of index determines the choice of analysis. For example, several indices are generated from a sides x individual ANOVA while other indices cannot be generated from an ANOVA. In this section, FA indices proposed by Palmer and colleagues (Palmer and Strobeck, 1986; Palmer, 1994; Klingerberg and McIntyel,
1998; Leung et al., 2000; Palmer and Strobeck, 2003) will be discussed followed by recommended statistical analyzes for FA.
FA Indices
Before discussing specific FA indices, two terms need to be addressed. Signed asymmetry (R – L) is the difference between the right (R) and left (L) sides of a bilateral trait of an individual. Information on the direction of asymmetry is provided through the use of signed asymmetry. Absolute asymmetry, or unsigned asymmetry (|R – L|), is the absolute value of the difference between R and L sides of a bilateral trait for an individual.
56
Palmer and Strobeck (1986) provided a summary of nine fundamental indices. FA indices were grouped into categories based on absolute asymmetry and signed asymmetry, type of size correction, and whether the index can be applied to single or multiple traits per individual. Palmer (1994) added three additional indices to address multiple traits per individual. Other scholars (Klingerberg and McIntyel, 1998; Leung et al., 2000) have contributed additional indices including a landmark method index. Palmer and Strobeck (2003) published an updated summary of FA indices (Table 2) which expanded on their original summary (Palmer and Strobeck,1986) as well as indexes proposed by other scholars (e.g. Palmer, 1994; Klingerberg and McIntyel, 1998; Leung et al., 2000). A brief summary of the most widely used FA indices, as discussed by Palmer and Strobeck (2003), is provided below. FA indices are labeled according to Palmer and
Strobeck (2003). After discussing these FA indices, a summary of all FA indices is provided in Table 2.
FA1 is the most commonly used index for FA studies (Palmer, 1994; Palmer and
Strobeck, 2003) because it is easily used in the ANOVA procedure for testing differences among samples. This index, however, does not include a size correction. Because FA2
| | ( ) includes a size correction at the individual level it is widely used in / studies on FA variation. Tests for significant of FA variation are restricted for FA2 if ME is not accounted for because the replicate measurements are averaged prior to applying the index (Palmer and Strobeck, 2003). Palmer (1994) and colleagues (Palmer and
Strobeck, 1986, 2003; Leung et al., 2000; Palmer and Strobeck, 2003), recommend using
FA2 only when it is clear that a size dependence of | | among individuals within a
57
Indices Equation Pros and Cons of Index (As explained by Palmer and Strobeck (1986, 2003) and Palmer (1994))
1 | | Not very sensitive to outliers Size correction not included Should not be used if DA or AS are present
| | 2 Not very sensitive to outliers / Size correction at individual level Replicate measures are averaged prior to calculation Sensitive to DA and AS Recommended only for use when a size dependence of | |among individuals is present