TWO THOUSAND YEARS OF FORAGING ECOLOGY IN THE ENDANGERED
HAWAIIAN PETREL: INSIGHTS FROM STABLE ISOTOPE ANALYSIS.
By
Anne E. Wiley

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Zoology
Ecology, Evolutionary Biology and Behavior - Dual Major

2011

ABSTRACT
TWO THOUSAND YEARS OF FORAGING ECOLOGY IN THE ENDANGERED
HAWAIIAN PETREL: INSIGHTS FROM STABLE ISOTOPE ANALYSIS
By
Anne E. Wiley
Recent evidence indicates that over the last 150 years, humans may have impacted
seabird populations through modification of their marine food resources. Unfortunately, the high
mobility and large pelagic ranges of many seabirds has resulted in a dearth of information
concerning even their basic feeding habits. Here, I use stable isotope analysis to investigate the
modern and ancient foraging ecology of an endangered seabird, the Hawaiian petrel (Pterodroma
13

15

sandwichensis). Stable isotopic composition of Hawaiian petrel tissues ( C and  N values)
reflects trophic level and foraging location and can therefore be used to describe patterns of
foraging segregation or long-term temporal variation within the species.
Chapter 1 investigates isotopic variation within individual flight feathers, with the goal of
13

designing minimally-invasive and ecologically informative sampling strategies.  C values
increased from tip to base in all 52 feathers within the study, including 42 remiges from the
Hawaiian petrel and 10 from the Newell‘s Shearwater (Puffinus auricularis newelli). Such a
consistent trend, observable among different species and age classes, is unlikely to result from
15

shifts in diet or foraging location during feather synthesis. Considerable variation of  N values
was also present within feathers (average range of 1.3 ‰ within Hawaiian petrel remiges). A
sampling protocol is proposed that requires only 1.0 mg of feather and minimal preparation time.
Because it leaves the feather nearly intact, this protocol will likely facilitate obtaining isotope
values from remiges of live birds and museum specimens.

Chapter 2 explores ecological variability among modern Hawaiian petrel populations.
13

15

 C and  N values of feathers demonstrate segregation in foraging location during both the
breeding and non-breeding seasons for petrels nesting on Kauai and Hawaii. Genetic analyses
based on the mitochondrial Cytochrome b gene also reveal strong differentiation: coalescentbased analyses estimate < 1 migration event per 1,000 generations. Finally, feathers from
multiple age groups and islands show unexpected divergences in D that cannot be related to
variation in source water. Overall, these data demonstrate foraging and genetic divergence
between proximately nesting seabird populations. This divergence occurs despite high species
mobility and a lack of physical barriers between nesting sites.
Chapter 3 investigates Hawaiian petrel foraging habits and inter-colony segregation over
the course of approximately 2,000 years. The most pervasive temporal trend is a 1.4-2.6 ‰
15

decrease in average  N values, which likely reflects declining trophic level over the past 3001,000 years. Isotopic chronologies also document ca. 2,000 years of foraging segregation
15

between Hawaiian petrel colonies, observed as a long-standing divergence in average  N
values. The degree of foraging segregation between petrel colonies diminishes through time and
correlates well with genetic population structure. Shifting foraging habits of the Hawaiian petrel
may reflect relatively widespread trophic alterations in the pelagic realm of the North Pacific.
Such changes in foraging are concerning, given their implications for reproductive success and
genetic diversity.

ACKNOWLEDGEMENTS
Perhaps the biggest lesson from my Ph.D. was the extent to which scientific research
depends on a vast network of people and ideas. While I am considered the sole author of this
dissertation, the research contained herein would have been impossible without the work of
numerous colleagues and the influence of friends (I am fortunate that many people fall into both
categories). I am most grateful to my advisor, Peggy Ostrom, for her boundless energy and
imagination, infectious enthusiasm, and kind generosity with her time and ideas. I am thankful
for other colleagues who made my everyday work enjoyable and productive: Helen James, with
her amazing knowledge of Hawaiian birds and her inspiring intellectual curiosity, Hasand
Gandhi, who always had a ready smile and infinite patience for me in lab, and Andreanna Welch,
with her humor and remarkable genetic research. I am similarly grateful to Craig Stricker, Robert
Fleischer, Thomas Stafford, Jr., for their collaboration and insights. I am very grateful to my
committee members, who provided thoughtful comments and flexibility: Nathaniel Ostrom, for
his expertise in marine biochemistry, and Catherine Lindell, who provided ornithological
expertise and insights into population ecology, and Stuart Gage, who always reminded me to
think of the big picture. I am indebted to numerous other researchers and colleagues who
provided ideas, support, and inspiration: John Southon, David Ainley, Josh Adams, Sam
Rossman, Catherine Lorenz, Peter Pyle, and Greg Wiles.
On a more personal note, I am thankful for my wonderful family, who has nurtured my
curiosity for nature, from encounters with angora rabbits and Maine streams, to bird-watching
trips. Perhaps more importantly, they have always shown the utmost love and concern for my
happiness. I am grateful to my husband, Joseph Mead, who has learned more about the Hawaiian
petrel and isotopes than any attorney should have to know. He made my non-working moments

iv

happy, and he supported me in every possible way throughout my graduate career: from edits
and cheers to the occasional celebratory brownie.
Finally, I thank my colleagues in the Hawaiian Islands: Darcy Hu, Nick Holmes, Jay
Penniman, Fern Duvall, Holly Freifeld, Cathleen Bailey, Seth Judge, and Keith Swindle. Their
steadfast collection efforts in the field have made this work possible, and their daily devotion to
saving the plants and animals of the Hawaiian Islands continues to inspire me.

v

TABLE OF CONTENTS
LIST OF TABLES ....................................................................................................................... viii
LIST OF FIGURES ....................................................................................................................... ix
INTRODUCTION .......................................................................................................................... 1
LITERATURE CITED ....................................................................................................... 4
CHAPTER 1
ISOTOPIC CHARACTERIZATION OF FLIGHT FEATHERS IN TWO PELAGIC
SEABIRDS: SAMPLING STRATEGIES FOR ECOLOGICAL STUDIES ................................. 6
ABSTRACT ........................................................................................................................ 6
INTRODUCTION .............................................................................................................. 7
METHODS ......................................................................................................................... 9
Sample Acquisition and Preparation ............................................................................. 9
Seven-section protocol for Hawaiian Petrel and Newell’s Shearwater feathers ......... 10
Three-section protocol for Hawaiian Petrel and Newell’s Shearwater feathers ......... 11
Barb-sampling protocol for Hawaiian Petrel feathers ................................................ 12
Comparison of Sampling Protocols for Hawaiian Petrel Feathers ............................. 13
Stable Isotope and Elemental Analysis ........................................................................ 13
Statistical Analyses ....................................................................................................... 14
RESULTS ......................................................................................................................... 14
Hawaiian Petrel Feathers ............................................................................................ 14
Newell’s Shearwater Feathers ..................................................................................... 18
Comparison of Sampling Protocols ............................................................................. 22
DISCUSSION ................................................................................................................... 22
Sources of Isotopic Variation Within a Feather ........................................................... 23
Comparison of Sampling Protocols ............................................................................. 28
CONCLUSIONS............................................................................................................... 30
LITERATURE CITED ..................................................................................................... 31
CHAPTER 2
FORAGING SEGREGATION AND GENETIC DIVERGENCE BETWEEN
GEOGRAPHICALLY PROXIMATE COLONIES OF A HIGHLY MOBILE SEABIRD ........ 37
ABSTRACT ...................................................................................................................... 37
INTRODUCTION ............................................................................................................ 38
MATERIALS AND METHODS ...................................................................................... 42
Sampling ....................................................................................................................... 42
Genetic Analyses .......................................................................................................... 43
Stable Isotope Analyses ................................................................................................ 47
Statistical Analyses ....................................................................................................... 48
RESULTS ......................................................................................................................... 49
Genetic Analyses .......................................................................................................... 49
Stable Isotope Analyses ................................................................................................ 51
DISCUSSION ................................................................................................................... 54
vi

Genetic Analyses .......................................................................................................... 54
Stable Isotope Analyses ................................................................................................ 55
Causes and Implications of Genetic and Isotopic Variation ........................................ 63
LITERATURE CITED ..................................................................................................... 66
CHAPTER 3
TWO THOUSAND YEARS OF FORAGING ECOLOGY IN THE ENDANGERED
HAWAIIAN PETREL: INSIGHTS FROM STABLE ISOTOPE ANALYSIS ........................... 73
ABSTRACT ...................................................................................................................... 73
INTRODUCTION ............................................................................................................ 74
MATERIALS AND METHODS ...................................................................................... 77
Sample Acquisition and Feather Growth ..................................................................... 77
Stable Isotope and AMS Radiocarbon Methods ........................................................... 77
Temporal and Statistical Analysis ................................................................................ 79
RESULTS ......................................................................................................................... 80
Bone Collagen .............................................................................................................. 80
Feathers ........................................................................................................................ 81
DISCUSSION ................................................................................................................... 82
LITERATURE CITED ..................................................................................................... 90
CONCLUSION ............................................................................................................................. 95
GENERAL COMMENTS ................................................................................................ 95
QUESTIONS FOR FUTURE RESEARCH ..................................................................... 98

vii

LIST OF TABLES
Table 1. Comparison of protocols sampling Hawaiian Petrel remiges. For each feather, average
isotope values for the vanes were obtained by the seven-section, three-section, or barb-sampling
13
protocol described in the text. Values of Δ indicate the difference between protocols in δ C,
15
δ N, and δD. ................................................................................................................................ 21
Table 2. Two σ range for radiocarbon dates in each time bin and results of Tukey HSD
15
comparisons for  N. *Letters denote significant differences between groups in post hoc
analyses: values not sharing the same letter are significantly different (alpha = 0.05). ............... 81

viii

LIST OF FIGURES
Figure 1. Summary of the Hawaiian petrel annual cycle and the timing of feather growth. ........ 10
Figure 2. Feather-sampling protocols. Photographs of P1 of the (AI) Hawaiian Petrel and (BI)
Newell‘s Shearwater. Sampling schematic for (AII) Hawaiian Petrel and (BII) Newell‘s
Shearwater feathers, showing the division of the vanes into seven 1 cm sections (labeled 1–7). In
the seven-section protocol, each section was homogenized and subsampled for stable isotope
analysis. For the three-section protocol, only sections 1, 4, and 7 (shaded in gray) were
homogenized and subsampled. (AIII) Barb-sampling protocol for Hawaiian Petrel feathers, in
which barbs from the centers of sections 1, 4, and 7 (shaded in gray) were combined to form a
1.0 mg sample, labeled ―B,‖ for isotope analysis. ........................................................................ 12
13

15

Figure 3. Values of (A) δ C and (B) δ N values from five Hawaiian Petrel feathers sampled
by the seven-section sampling protocol described in the text. Individual feathers are separated by
vertical lines and identified by numbers in upper left corner. For each feather, data progress, left
to right, from section 1 (tip) to section 7 (base). Feathers 1–4 are P1; feather 5 is S10. .............. 16
Figure 4. Values of ΔM (section 4 minus section 1, or middle minus tip) and ΔL (section 7 minus
13
15
section 1, or base minus tip) for δ C, δ N, and δD in flight feathers of the Hawaiian Petrel (n =
38 P1, 1 P2, 2 S1, and 1 S10 for carbon and nitrogen; n = 21 P1 for hydrogen). Medians are
represented by thick lines, first and third quartiles by upper and lower boundaries of boxes,
respectively, ranges by whiskers, and outliers by ovals. Scaling of the y axis in each panel
varies. ............................................................................................................................................ 17
13

15

Figure 5. Values of (A) δ C and (B) δ N from three Newell‘s Shearwater feathers (all P1)
sampled by the seven-section protocol described in the text. Individual feathers are separated by
vertical lines and identified by numbers in upper left corner. For each feather, data progress, left
to right, from section 1 (tip) to section 7 (base). ........................................................................... 20
Figure 6. Values of ΔM (section 4 minus section 1, or middle minus tip) and ΔL (section 7 minus
13
15
section 1, or base minus tip) for δ C and δ N in P1 of Newell‘s Shearwater. Medians are
represented by thick lines, first and third quartiles by upper and lower boundaries of boxes,
respectively, ranges by whiskers, and outliers by ovals; n = 10. Scaling of the y axis in the two
panels varies. ................................................................................................................................. 21
Figure 7. Photographs of P1 of a Hawaiian Petrel before and after sampling for isotope analysis
by the barb protocol. As illustrated, this sampling protocol preserves the gross morphology of the
feather, with minimal disruption to the vane. ............................................................................... 30
ix

Figure 8. Marine distribution (A) and breeding distribution (B) of the Hawaiian petrel. The dark
blue area indicates the combined spring and autumn distribution as reported by Spear et al.
(1995). The light blue area approximates additional areas frequented by the Hawaiian petrel, as
inferred from Simons et al. (1998). Islands with modern breeding colonies are shaded red.
Triangles mark the origin of black-footed and laysan albatross in Figure 9. The solid line
represents the Equator, while stippled lines represent, from south to north, the Tropics of
Capricorn and Cancer, and the Arctic Circle. For interpretation of the references to color in this
and all other figures, the reader is referred to the electronic version of this dissertation. ............ 41
Figure 9. Haplotype network of Cytochrome b sequences of Hawaiian petrels from Hawaii
(black) and Kauai (grey). Each circle represents a unique sequence and the size of the circle is
proportional to the number of individuals possessing that sequence (sample sizes are indicated
for the most common haplotypes). Each line represents one mutational step, and small white
circles represent an unsampled intermediate sequence. ................................................................ 51
Figure 10. Stable carbon and nitrogen isotope values from Hawaiian petrels and adult Newell's
shearwater flight feathers and from non-breeding adult albatross muscle. Stable isotope values
from muscle are standardized to feather isotope values, as indicated in the Materials and
Methods. Sample sizes for Hawaiian petrel data are as follows: Hawaii adults (N=14), Hawaii
hatch-year birds (N=10), Kauai adults (N=13), and Kauai hatch-year birds (N=12). Remaining
sample sizes are: Newell's shearwaters (N=8), Laysan albatross (N=55), and black-footed
albatross (N=16). All albatross derive from 40-45°N and 145°W to 145°E (Transition Zone), as
shown in Figure 8.......................................................................................................................... 53
Figure 11. Stable hydrogen isotope values in Hawaiian petrel flight feathers. Sample sizes are as
follows: Hawaii adults (N=14), Hawaii hatch-year birds (N=10), Kauai adults (N=13), and Kauai
hatch-year birds (N=12). ............................................................................................................... 54
Figure 12. Fall distribution of the Hawaiian petrel, as observed by Spear et al. 1995, outlined in
black. Shades of blue represent nitrogen isotope values of a mobile consumer, yellowfin tuna
(Graham et al. 2010). Green-labeled points represent stable nitrogen isotope values of sinking
1
2
and floating sediments: Karl et al. 1997 (annual mean export pulse), Altabet and Francois
4
1994, and Voss 2001. White-labeled points represent stable nitrogen isotope values of core top
2
3
4
sediments: Altabet and Francois 1994 and Farrell et al. 1995, and Voss 2001. Latitude is
indicated on the x-axis (°N) and longitude on the y-axis (°W). All points along 140°, 137°, and
110 ° W are floating or core top sediments, as indicated by labeled arrows. ............................... 58
Figure 13. Location of sub-fossil collection sites (circles) and current breeding distribution (all
other shapes, outlined in black) for HAPE. .................................................................................. 76
x

15

Figure 14. Bone collagen  N through time in four HAPE colonies. Data points reflect average
age and isotopic composition of each time bin, plus or minus standard error. Sample sizes are
indicated next to each data point. Lines connect data points within each colony, but are intended
for visualization purposes, only. Isotopic shifts between time bins may have occurred nonlinearly. ......................................................................................................................................... 87
Figure 15. Foraging segregation among modern HAPE colonies, as observed in stable isotope
values. Data points reflect the mean plus or minus standard error for each group. Sample sizes
are as follows: Hawaii adults (n=14), Hawaii hatch-years (n=10), Kauai adults (n=13), Kauai
hatch years (n=12), Maui adults (n=13), Maui hatch years (n=9), and Lanai adults (n=18). ....... 88
Figure 16. Schematic of dissertation research. Major conclusions from each chapter are listed
(three vertical rectangles), along with questions for future research (bottom horizontal rectangle).
Thick arrows represent the major flow of ideas within chapters. Thin arrows represent the flow
of ideas between chapters and leading to future research questions. HAPE is used as an acronym
for the Hawaiian petrel.................................................................................................................. 97

xi

INTRODUCTION
Pelagic seabirds are characterized by their unusual bimodal lifestyle; they nest on land,
but forage exclusively in the marine realm. On their breeding grounds, seabirds can be influential
components of island ecosystems, contributing marine-derived nutrients in the form of guano,
eggshells, and chick mortality (Harding et al. 2004; Blais et al. 2005). Their role within marine
ecosystems, and their feeding habits in general, have been much more difficult to ascertain.
Pelagic seabirds are often distributed in low densities over large oceanic ranges, making surveys
and observational studies logistically difficult. The resulting dearth of information that exists for
many species is a challenge for conservation efforts. While seabirds constitute a highly
threatened group, managers often know little of their food resources and lack basic information
on marine range and foraging behaviors (Warham 1996).
Stable isotope analysis is one of multiple newly-emerging tools that can be used to study
formerly ―black box‖ topics in seabird ecology. Based on the premise that isotope values (such
13

15

as  C and  N) differ by a predictable amount in seabird versus prey tissues, isotope analysis
has given new insight into pelagic seabird diet (Hobson et al. 1994). In addition, isotopic
gradients exist within the foraging range of some species, allowing study of their distribution and
migration (Cherel et al. 2000). Other advances in seabird ecology derive from genetic analysis,
which allows documentation of population structure, and satellite tracking, which provides
detailed information on the flight pattern of individual birds.
Using a combination of isotopic, genetic, and tracking technologies, recent studies have
made intriguing discoveries about seabird populations and their marine habits. While species
such as the Arctic tern are famous for their long bi-annual migrations, many pelagic seabirds
travel extreme distances on a daily basis (e.g. an average of >500 km daily for tracked Hawaiian

1

petrels (Adams & Flora 2009). Population structure in such mobile birds appears to be highly
variable. Some species are distributed across multiple ocean basins with little population
structure, while others form genetically distinct populations, and even speciate, within individual
archipelagos (Friesen et al. 2007). Remarkably, population divergence may have a strong
foundation in foraging habits, as species with multiple, spatially separate, non-breeding
distributions are the most likely to display population structure (Friesen et al. 2007). Foraging
habits, in turn, may show large fluctuations as a result of human fishing practices. A small but
growing collection of isotopic studies show that within the last 150 years, some seabirds have
changed their diet as a likely result of human-induced shifts in prey availability (Becker &
Beissinger 2006; Farmer & Leonard 2011). By and large, these retrospective views have been
possible due to the availability of feathers from historically collected seabirds, now preserved as
museum study skins. Indeed, the high availability of both modern and historical feathers has
made them one of the most commonly used tissues for isotopic studies of avian ecology
(Dalerum & Angerbjörn 2005).
My dissertation research relies on stable isotope analysis to address three broad topics in
seabird ecology:
1) Methodological considerations for using isotopic variation in feathers to study seabird
ecology (Chapter 1)
2) The distribution of foraging diversity within a seabird species, especially as it relates to
breeding location and genetic population structure (Chapters 2 & 3)
3) The historical, potentially anthropogenically-driven changes that have occurred to seabird
feeding ecology (Chapter 3)

2

I address these topics primarily through focus on a single species, the Hawaiian petrel
(Pterodroma sandwichensis). One of the principal reasons for choosing the Hawaiian petrel as a
focal species is its decline from widespread abundance in the Hawaiian Islands to current
endangered status. Formerly large populations are represented by a high profusion of subfossil
bones: samples that can be used to develop isotope chronologies reaching back thousands of
years. Because the Hawaiian petrel is currently endangered, results from my research can help to
inform conservation management. The species‘ current scarcity, coupled with its high mobility
and typically nocturnal habits have resulted in a scarcity of knowledge concerning diet and
foraging distribution. The Hawaiian petrel thus represents an ideal candidate for a stable isotopic
investigation of modern and ancient foraging ecology.

Note on chapter titles, publication, and authorship: Because Chapter 3 is the
culmination of my dissertation research, incorporating data and conclusions from previous
chapters, it is given the same title as the entire dissertation. Chapter 1 is currently published in
the journal Condor with coauthors Peggy Ostrom, Craig Stricker, Helen James, and Hasand
Gandhi. Chapter 2 is accepted for publication in the journal Oecologia with coauthors Andreanna
Welch, Peggy Ostrom, Helen James, Craig Stricker, Robert Fleischer, Hasand Gandhi, Josh
Adams, David Ainley, Fern Duvall, Nick Holmes, Darcy Hu, Seth Judge, Jay Penniman, and
Keith Swindle. In Chapter 2, Andreanna Welch and Robert Fleischer had primary responsibility
for the generation and interpretation of genetic data.

3

LITERATURE CITED

4

Adams, J. and S. Flora. 2009. Correlating seabird movements with ocean winds: linking satellite
telemetry with ocean scatterometry. Marine Biology 157 (4): 915-929.
Becker, B.H., and S.R. Beissinger. 2006. Centennial Decline in the Trophic Level of an
Endangered Seabird after Fisheries Decline. Conservation Biology 20 (2): 470-479.
Blais, J. M, L. E Kimpe, D. McMahon, B. E Keatley, M. L Mallory, M. S.V Douglas, and J. P
Smol. 2005. Arctic seabirds transport marine-derived contaminants. Science 309 (5733):
445.
Cherel, Y., K. A Hobson, and H. Weimerskirch. 2000. Using stable-isotope analysis of feathers
to distinguish moulting and breeding origins of seabirds. Oecologia 122 (2): 155–162.
Dalerum, F., and A. Angerbjörn. 2005. Resolving temporal variation in vertebrate diets using
naturally occurring stable isotopes. Oecologia 144 (4): 647–658.
Farmer, R. G., and M. L. Leonard. 2011. Long-term feeding ecology of Great Black-backed
Gulls (Larus marinus) in the northwest Atlantic: 110 years of feather isotope data.
Canadian Journal of Zoology 89 (2): 123–133.
Friesen, V. L., T. M. Burg, and K. D. McCoy. 2007. Mechanisms of population differentiation in
seabirds. Molecular Ecology 16 (9): 1765–1785.
Harding, J. S, D. J Hawke, R. N Holdaway, and M. J Winterbourn. 2004. Incorporation of
marine-derived nutrients from petrel breeding colonies into stream food webs.
Freshwater Biology 49 (5): 576–586.
Hobson, K. A. 1993. Trophic relationships among high Arctic seabirds: insights from tissuedependent stable-isotope models. Marine Ecology Progress Series 95: 7–7.
Hobson, K. A, J. F Piatt, and J. Pitocchelli. 1994. Using stable isotopes to determine seabird
trophic relationships. Journal of Animal Ecology 63 (4): 786–798.
Warham, John. 1996. The behaviour, population biology and physiology of the petrels.
Academic Press.

5

CHAPTER 1
ISOTOPIC CHARACTERIZATION OF FLIGHT FEATHERS IN TWO PELAGIC
SEABIRDS: SAMPLING STRATEGIES FOR ECOLOGICAL STUDIES
ABSTRACT
We wish to use stable isotope analysis of flight feathers to understand the feeding behavior of
pelagic seabirds, such as the Hawaiian Petrel (Pterodroma sandwichensis) and Newell‘s
Shearwater (Puffinus auricularis newelli). Analysis of remiges is particularly informative
because the sequence and timing of remex molt are often known. The initial step, reported here,
is to obtain accurate isotope values from whole remiges by means of a minimally invasive
13

protocol appropriate for live birds or museum specimens. The high variability observed in δ C
15

and δ N values within a feather precludes the use of a small section of vane. We found the
13

15

average range within 42 Hawaiian Petrel remiges to be 1.3‰ for both δ C and δ N and that
13

15

within 10 Newell‘s Shearwater remiges to be 1.3‰ and 0.7‰ for δ C and δ N, respectively.
13

The δ C of all 52 feathers increased from tip to base, and the majority of Hawaiian Petrel
15

feathers showed an analogous trend in δ N. Although the average range of δD in 21 Hawaiian
Petrel remiges was 11‰, we found no longitudinal trend. We discuss influences of trophic level,
foraging location, metabolism, and pigmentation on isotope values and compare three methods
of obtaining isotope averages of whole feathers. Our novel barb-sampling protocol requires only
1.0 mg of feather and minimal preparation time. Because it leaves the feather nearly intact, this
protocol will likely facilitate obtaining isotope values from remiges of live birds and museum
specimens. As a consequence, it will help expand the understanding of historical trends in
foraging behavior.

6

INTRODUCTION
From elucidating the diet of historical populations to clarifying the migratory patterns of highly
mobile species, stable isotope analysis has become an indispensable tool in the study of animal
movement and feeding ecology (Cherel et al. 2006, Norris et al. 2007). Carbon, nitrogen, and
hydrogen isotopes (δ

13

C, δ

15

N and δ D) act as intrinsic markers of foraging behavior and

geographic origin because they are passed from diet or water source to consumer either
conservatively or with a predictable increase (Cormie & Schwarcz 1994, Martinez del Rio et al.
2008). Feathers are frequently used in avian isotopic ecology, due in part to their availability
from live individuals, museum specimens, and salvaged carcasses (Dalerum & Angerbjorn
2005). Because they are metabolically inactive following synthesis, feathers record isotopic
signals during molt. For example, stable isotope analysis of feathers provides information on the
foraging ecology of seabirds that molt during the nonbreeding season, when it is logistically
difficult to observe birds or obtain stomach contents (Cherel et al. 2006). Isotopic analysis of
remiges is particularly informative because the sequence and timing of remex molt are often
known.
Reliance on stable isotope analysis in avian foraging ecology derives from controlled
13

15

experiments that showed the δ C and δ N values of diet or water source are recorded in
feathers (Hobson & Clark 1992a & b, Hobson et al. 1999). Additional investigations provided
information on turnover rate, fractionation, and the influence of nutritional status, stress, and a
variety of life-history traits on isotope values (Bearhop et al. 2002, Cherel et al. 2005, Sears et
al. 2009).
Most isotope-based studies of wild birds either use a small section of a feather or
completely homogenize the vanes or entire feather. The first approach is time-efficient, less

7

destructive, and useful for sampling feathers from museum specimens. Because feathers
represent a temporal sequence of tissue synthesis over days to weeks, this approach also provides
a short-term record of diet. Homogenization is destructive but provides the average isotope
values over the entire period of feather growth. Alternatively, multiple samples can be taken
along the length of a feather. These longitudinal samples provide both long-term isotopic
information and a record of dietary, geographic, or physiological changes during feather
synthesis (Hobson & Clark 1992a, Knoff et al. 2002). Such information is of interest for species
that change their diet or water source during the period of feather growth or in studies of niche
width (Thompson & Furness 1995, Newsome et al. 2007). To date, few investigations have
evaluated isotopic heterogeneity within feathers (Thompson & Furness 1995, Wassenaar &
Hobson 2006, Smith et al. 2008).
We are interested in using stable isotope analysis of flight feathers to study the feeding
ecology of the endangered Hawaiian Petrel (Pterodroma sandwichensis). Prior to making
13

15

ecological interpretations, we wished to (1) quantify the degree of variation of δ C, δ N, and
δD within individual flight feathers and (2) design sampling strategies to accurately estimate,
with minimal destruction, the average isotope value of the vanes. In the Hawaiian Petrel, the
13

15

degree of heterogeneity of δ C and δ N within a feather could be high. Specifically, the birds
make foraging trips that extend several thousand kilometers, traversing a spatial gradient in the
13

North Pacific Ocean over which δ C values decrease with latitude (Goericke & Fry 1994,
13

15

Adams et al. 2006). They also consume a variety of prey that vary in δ C and δ N as a
13

function of trophic level (Simons 1985, Michener & Schell 1994). Therefore, we studied δ C
15

and δ N values from samples taken along the length of Hawaiian Petrel remiges. We also
explored variation in δD within a subset of these feathers. Although δD values of the Hawaiian
8

Petrel‘s water source do not vary (Lecuyer et al. 1997), they may fluctuate as a function of
evaporative water loss and other physiological factors (McKechnie et al. 2004). To determine if
our findings were species-specific or more broadly characteristic of seabird remiges, we
13

15

examined longitudinal variation of δ C and δ N in primaries of Newell‘s Shearwater (Puffinus
auricularis newelli). This threatened species is endemic to the Hawaiian Islands, feeds primarily
on squid, and has a marine range more restricted than the Hawaiian Petrel‘s (Spear et al. 1995,
Ainley et al. 1997, Simons et al. 1998).

METHODS
Sample Acquisition and Preparation
We obtained Hawaiian Petrel feathers from carcasses salvaged on the islands of Hawaii, Maui,
Lanai, and Kauai between 1990 and 2008. Newell‘s Shearwater feathers were also collected
from salvaged carcasses, found on Kauai between 1999 and 2006. Following departure from
their colonies on the Hawaiian Islands, adult Hawaiian Petrels spend 3.5–6 months at sea,
depending on breeding status, while they grow their flight feathers (Figure 1; Simons 1985). In
Pterodroma, the molt of remiges typically begins with primary 1 (P1, the feather most frequently
analyzed in our study, where primaries are numbered distally) and ends near secondary 10 (S10,
where secondaries are numbered proximally) (Warham 1996, Pyle 2008). Adult Newell‘s
Shearwaters also grow their remiges at sea following departure from their breeding colonies
(Jehl 1982). As in other shearwaters, molt of primaries likely proceeds distally (Warham 1996).
From estimates of rates of feather growth in juvenile Hawaiian Petrels and Newell‘s Shearwaters
(Sincock and Swedberg 1969, Simons 1985) and in other Procellariiformes (Ainley et al. 1976,
Langston & Rohwer 1996), P1 and each secondary require 12 to 35 days to grow. We compare

9

13

15

intrafeather variation in δ C and δ N in juveniles to that of adults from the island of Hawaii,
from which sufficient age-group identification and isotope data were available.

Figure 1. Summary of the Hawaiian petrel annual cycle and the timing of feather growth.
*The period of feather growth refers to the presumed timing of growth for the primary 1 feather.

Prior to being sampled, all feathers were washed in solvent (87:13 chloroform:methanol
by volume), rinsed with ultrapure distilled water (E-Pure, Barnstead), and dried in a vacuum
oven at 25°C. Both vanes of the cleaned and dried feathers were cut into sections according to
one of three sampling protocols described below. The rachis of each feather was excluded from
analysis.

Seven-section protocol for Hawaiian Petrel and Newell’s Shearwater feathers
We sampled the vanes from each of five Hawaiian Petrel remiges (four P1 and one S10) and
three Newell‘s Shearwater remiges (all P1) longitudinally at 1 cm intervals (Figure 2AII and

10

BII). About 7 cm long, the vanes provided 7 sections which we labeled 1–7 in order from tip to
base. We accounted for any small variation in vane length (≤0.3 cm) by increasing or decreasing
the size of section 7. We weighed each section, cut the barbs into fragments 3 mm long, and
homogenized the fragments by mixing them thoroughly with forceps. Finally, we took a 1.0 mg
aliquot for analysis of carbon and nitrogen isotopes.

Three-section protocol for Hawaiian Petrel and Newell’s Shearwater feathers
We sampled 37 additional Hawaiian Petrel remiges (33 P1, one P2, and two S1) and 7 additional
Newell‘s Shearwater remiges (P1) in an abbreviated protocol. We cut sections of vane analogous
13

15

to 1, 4, and 7 above (tip, middle, and base of vane) and took an aliquot for δ C and δ N
analysis as described above. The remaining mass of section 1 was insufficient for δD analysis,
but we analyzed 0.5 mg aliquots of sections 2, 4 and 7 from 21 of the Hawaiian Petrel P1
feathers for isotopic composition of hydrogen (nonexchangeable)

11

Figure 2. Feather-sampling protocols. Photographs of P1 of the (AI) Hawaiian Petrel and (BI)
Newell‘s Shearwater. Sampling schematic for (AII) Hawaiian Petrel and (BII) Newell‘s
Shearwater feathers, showing the division of the vanes into seven 1 cm sections (labeled 1–7). In
the seven-section protocol, each section was homogenized and subsampled for stable isotope
analysis. For the three-section protocol, only sections 1, 4, and 7 (shaded in gray) were
homogenized and subsampled. (AIII) Barb-sampling protocol for Hawaiian Petrel feathers, in
which barbs from the centers of sections 1, 4, and 7 (shaded in gray) were combined to form a
1.0 mg sample, labeled ―B,‖ for isotope analysis.

Barb-sampling protocol for Hawaiian Petrel feathers
From eight additional Hawaiian Petrel remiges (all P1), we plucked individual barbs from the
middle of sections 1, 4, and 7 and combined them into a single 1.0 mg composite sample, labeled
―B‖ (Figure 2 AIII). The number of barbs taken from each section (two each from sections 1 and
4 and eight from section 7) was based on the average distribution of mass found in other
13

15

Hawaiian Petrel P1 feathers. Using this method, we analyzed five feathers for δ C and δ N
values and three for δD.

12

Comparison of Sampling Protocols for Hawaiian Petrel Feathers
We used mass-weighted averages of the isotope values to evaluate whether the different
sampling protocols captured the same average isotope values for whole feathers. For the five
13

feathers analyzed comprehensively by the seven-section protocol, we calculated average δ C
15

and δ N values from all seven sections of vane and compared them with weighted averages
recalculated from only three of the seven sections: 1 (tip), 4 (middle), and 7 (base). We used a
different set of feathers to compare the barb-sampling and three-section protocols (three feathers
13

15

for δ C and δ N and three for hydrogen). After barbs were removed for sample ―B,‖ we cut
and homogenized the remnants of sections 1, 4, and 7 and took an aliquot for isotope analysis.
We then compared the weighted average isotope values from sections 1, 4, and 7 (representing
the three-section protocol) to the values obtained from sample ―B‖ (barb protocol).

Stable Isotope and Elemental Analysis
13

Aliquots (0.8–1.0 mg) of feathers for δ C, δ

15

N, and elemental analyses were weighed into tin

capsules and analyzed with an elemental analyzer (Eurovector) interfaced to an Isoprime mass
spectrometer (Elementar; Wong et al. 1992). Aliquots (0.5 mg) for δD analysis were weighed
into silver capsules and pyrolyzed at 1425 ºC in a high-temperature elemental analyzer (TC/EA,
Thermo-Finnigan) interfaced to a mass spectrometer (Thermo-Finnigan DeltaPlus XL). Prior to
δD analysis, samples were allowed to air-equilibrate with ambient laboratory conditions for a
minimum of 1 week (Wassenaar & Hobson 2003).
Stable isotope values are reported in δ-notation, expressed in parts per thousand (‰)
according to the equation δX = ([Rsample/Rstandard] – 1) × 1000, where X is
is the corresponding ratio

13

12

C/ C,

15

14

13

C,

15

N, or D and R

N/ N, or D/H. Rstandard is V-PDB, air, and V-SMOW for
13

13

15

13

15

δ C, δ N, and δD, respectively. For δ C and δ N, we analyzed laboratory standards between
13

15

every five unknowns, with a precision of ≤0.2‰ for both δ C and δ N. We first normalized
deuterium data to V-SMOW with benzoic acid (–60 ‰) and polyethylene foil (IAEA-CH-7, –
100 ‰), followed by a second normalization procedure using in-house keratin standards (–78‰
and –172‰, respectively) calibrated to CFS-CHS-BWB (Wassenaar & Hobson 2003) to account
for exchangeable hydrogen. Accuracy and precision for δD was ≤4‰. A standard of known
elemental composition, alanine, was used to determine C/N. The accuracy and precision of a
second amino acid analyzed against this standard was <0.1

Statistical Analyses
To compare the isotope values generated by different sampling protocols, we performed twotailed paired t-tests (ta where a = df) with the statistical program SPSS version 10.0 (SPSS, Inc.,
Chicago, IL). We report values for ΔL, ΔM, and range as means ± SD where ΔL is the isotope
value of the base of a feather minus that of the tip (section 7 minus 1), ΔM is the isotope value of
the middle of a feather minus that of the tip (section 4 minus 1), and range is the maximum
minus minimum isotope value within a feather.

RESULTS
Hawaiian Petrel Feathers
The five feathers sampled comprehensively by the seven-section protocol had a longitudinal
13

trend in their isotope values (Figure 3). In every feather, the δ C of the oldest material at the tip
of the vane (section 1) was lower than that of the youngest material at the base (section 7). Four
15

of the five feathers showed a similar trend in δ N. The increase in isotope values from tip to
base, which we designate as a positive ΔL (the isotope value of section 7 minus section 1),

14

13

15

averaged 0.7‰ and 1.2‰ for δ C and δ N, respectively, and so could not be explained by
analytical error (≤0.2 ‰).

15

13

15

Figure 3. Values of (A) δ C and (B) δ N values from five Hawaiian Petrel feathers sampled
by the seven-section sampling protocol described in the text. Individual feathers are separated by
vertical lines and identified by numbers in upper left corner. For each feather, data progress, left
to right, from section 1 (tip) to section 7 (base). Feathers 1–4 are P1; feather 5 is S10.

16

Figure 4. Values of ΔM (section 4 minus section 1, or middle minus tip) and ΔL (section 7 minus
13
15
section 1, or base minus tip) for δ C, δ N, and δD in flight feathers of the Hawaiian Petrel (n =
38 P1, 1 P2, 2 S1, and 1 S10 for carbon and nitrogen; n = 21 P1 for hydrogen). Medians are
represented by thick lines, first and third quartiles by upper and lower boundaries of boxes,
respectively, ranges by whiskers, and outliers by ovals. Scaling of the y axis in each panel
varies.

We analyzed an additional 37 feathers by the three-section protocol. We report isotope
values of section 4 relative to those of section 1 and define this as ΔM (Figure 4A and B). For
13

13

δ C, ΔL was uniformly positive, with all 37 feathers showing an increase in δ C from tip
(section 1) to base (section 7). Additionally, ΔM was positive in 31 of the 37 feathers. When
13

considered with trends in ΔL, this pattern indicated a consistent longitudinal increase in δ C
from tip to middle to base of the vane. Although the median value of ΔM was only slightly
15

15

greater than zero for δ N, ΔL values for δ N were positive in the majority of feathers (25 out of
15

37), and a longitudinal increase in δ N was consistent in slightly over half the feathers (20).

17

Data from all 42 Hawaiian Petrel feathers combined, the average ΔL was 1.3 ±
13

0.6‰ for δ C (equivalent to an increase of approximately one trophic level). The average range
13

(highest minus lowest isotope value within a feather) of δ C values was an equivalent 1.3 ±
0.5‰, suggesting that longitudinal enrichment accounted for the majority of variation in our
13

15

δ C data. Conversely, the average range in δ N values was 1.3 ± 1.0‰, larger than the average
ΔL of 0.6 ± 1.3‰. Thus longitudinal enrichment accounted for only a portion of variation in
15

δ N.
A comparison of eight known adults to seven juveniles from Hawaii showed no
13

15

difference in ΔL or range for δ C or δ N. Average values for adults and juveniles,
13

respectively, were 0.9 ± 0.4‰ and 1.0 ± 0.3‰ for ΔL of δ C, 1.0 ± 0.3 ‰ and 1.0 ± 0.2‰ for
13

15

range of δ C, 0.8 ± 0.8‰ and 0.8 ± 0.6‰ for ΔL of δ N, and 1.1 ± 0.5‰ and 0.9 ± 0.6‰ for
15

range of δ N.
With respect to δD, our data showed little evidence of longitudinal enrichment, although
generally section 7 was enriched compared to section 1 (Figure 4C). Of the 21 feathers analyzed,
ΔL was positive for 11 feathers and the average intrafeather range of δD values was 11 ± 7‰. In
20 Hawaiian Petrel remiges C/N averaged 3.8 ± 0.1. The average ΔM, ΔL, and range for C/N
were all –0.1 ± 0.1.
Newell’s Shearwater Feathers
Like those for the Hawaiian Petrel, results from the seven-section protocol for all three Newell‘s
13

Shearwater feathers showed clear increases in δ C from tip to base (Figure 5). Only one feather
15

showed a longitudinal trend in δ N (ΔL = 1.5‰).

18

We used the three-section protocol on seven Newell‘s Shearwater feathers and compiled
13

data from all ten feathers (Figure 6). As for the Hawaiian Petrel, for δ C the value of ΔL in
every feather was positive. Additionally, values of ΔM in all feathers were positive. When
13

considered with those of ΔL, values of ΔM indicated a consistent increase in δ C from tip to
13

middle to base of the vanes in every individual. For δ C, both ΔL and the range averaged 1.3 ±
15

0.2‰, remarkably similar to results for the Hawaiian Petrel. Longitudinal trends in δ N were
less consistent. In only four Newell‘s Shearwater feathers were positive values of ΔL greater than
our analytical error, and in only three were values of ΔM positive. The average ΔL and range for
15

δ N were 0.4 ± 0.6‰ and 0.7 ± 0.5‰, respectively. C/N was measured in seven Newell‘s
Shearwater remiges and averaged 3.8 ± <0.1. The average ΔM for C/N was –0.1 ± < 0.1, and the
average ΔL and range were both –0.2 ± <0.1.

19

13

15

Figure 5. Values of (A) δ C and (B) δ N from three Newell‘s Shearwater feathers (all P1)
sampled by the seven-section protocol described in the text. Individual feathers are separated by
vertical lines and identified by numbers in upper left corner. For each feather, data progress, left
to right, from section 1 (tip) to section 7 (base).

A

B

20

Figure 6. Values of ΔM (section 4 minus section 1, or middle minus tip) and ΔL (section 7 minus
13
15
section 1, or base minus tip) for δ C and δ N in P1 of Newell‘s Shearwater. Medians are
represented by thick lines, first and third quartiles by upper and lower boundaries of boxes,
respectively, ranges by whiskers, and outliers by ovals; n = 10. Scaling of the y axis in the two
panels varies.

Table 1. Comparison of protocols sampling Hawaiian Petrel remiges. For each feather, average
isotope values for the vanes were obtained by the seven-section, three-section, or barb-sampling
13
protocol described in the text. Values of Δ indicate the difference between protocols in δ C,
15
δ N, and δD.
Comparison
Δ13C (‰)
Δ15N (‰)
ΔD (‰)
Seven-section vs. three-section
Individual 1
0.0
0.0
—
Individual 2
0.0
0.1
—
Individual 3
0.2
0.0
—
Individual 4
0.0
0.0
—
Individual 5
0.1
0.2
—
Average (SD)
0.1 (0.1)
0.1 (0.1)
—
Three-section vs. barb
Individual 6
0.0
0.2
3
Individual 7
0.1
0.1
6
Individual 8
0.0
0.1
1
Individual 9
0.1
0.0
5
Individual 10
0.0
0.3
0
Average (SD)
0.0 (0.1)
0.0 (0.1)
3 (2)
21

Comparison of Sampling Protocols
Because the seven-section protocol was comprehensive, we assumed the whole-feather average
isotope values it yielded to be accurate. We assessed the accuracy of the three-section protocol
by comparing the results from this method to those of the seven-section protocol. The difference
of 0.1 ± 0.1‰ between the two protocols was analytically undetectable and statistically
13

15

insignificant (Table 1; t4, n = 5, P = 0.48 for δ C, P = 0.70 for δ N). We assessed our
minimally invasive barb-sampling protocol by comparing its results with those of the three13

15

section protocol. For δ C, δ N, and δD, the average difference between these two protocols
13

15

was analytically undetectable (0.0 ± 0.1‰ for δ C, 0.1 ± 0.1‰ for δ N, 3 ± 3 ‰ for δD) and
13

15

statistically insignificant (Table 1; t4, n = 5, P = 0.18 for δ C, P = 0.91 for δ N, P = 0.51 for
δD).

DISCUSSION
Our objectives were to characterize the degree of isotopic heterogeneity within Hawaiian Petrel
and Newell‘s Shearwater feathers and to design sampling protocols that account for this
variability. Analysis of the entirety of the vanes was particularly important because our goal was
to obtain the longest-term dietary signals available from a feather: a measurement that is clearly
not represented by a single section of vane. In addition, we wished to develop a sampling
protocol appropriate for museum specimens that accurately represents the average isotope value
of the vane of an entire feather with minimal destruction. For Hawaiian Petrels and Newell‘s
Shearwater remiges, isotope values of small sections of vane were not representative of the
whole feather. Although heterogeneity within a feather in δD generally exceeded analytical error,
longitudinal trends were not obvious. In contrast, we observed large directional trends along the
13

15

length of the vane in δ C and δ N.
22

In pelagic seabirds, unlike terrestrial birds, δD is not expected to vary with latitude
because pelagic birds rely on an isotopically homogeneous water source of 0‰ (Craig 1961,
Lecuyer et al. 1997). A few studies have considered the influence of trophic level on δD, but no
consensus has been reached (Estep & Dabrowski 1980, Birchall et al. 2005, Wolf et al. 2009).
However, temporal variation in the amount of deuterium-depleted water lost through evaporation
might contribute to the observed variation of δD within feathers (Schoeller et al. 1986,
McKechnie et al. 2004).
13

The most salient feature of our data, a positive ΔL for δ C values, was evident in every
feather: 42 Hawaiian Petrel and 10 Newell‘s Shearwater remiges. These positive values indicate
13

that the δ C value of the oldest material at the tip of the feather vane was lower than the
15

youngest material at the base. In the majority of individuals the longitudinal trend in δ N was
similar. Although our study was not designed to constrain causes of variation within a feather
experimentally, we believe it important to explore factors that might contribute to the prevailing
trends.

Sources of Isotopic Variation Within a Feather
Variation in the

13

C content of avian tissues is often attributed to differences in the location

where a bird forages. However, location of foraging is not a parsimonious explanation for our
13

results. In the North Pacific Ocean, there is a negative relationship between latitude and the δ C
value of phytoplankton that permeates through the food web to organisms such as squid and
13

seabirds (Goericke & Fry 1994, Kelly 2000, Takai et al. 2000). Therefore, the δ C Hawaiian
Petrel and Newell‘s Shearwater feathers are expected to increase as birds spend more time
13

foraging in southern latitudes. We observed a positive ΔL, or increase in δ C from tip to base,

23

in every feather. For this trend to have resulted from a change in foraging location, every adult
must have traveled ~10° south (Kelly 2000) while growing its remiges. In addition, chicks must
have been provisioned with prey from increasingly southern latitudes over the ~25 days their P1
was growing. Because we observed no difference in average ΔL between adult and hatch-year
Hawaiian Petrels from the island of Hawaii, there must also be no difference in the degree to
which adults move south as they provision chicks and the degree to which they move south later
in the year, when they grow their own remiges. Finally, because average ΔL values in Hawaiian
Petrels and Newell‘s Shearwaters did not differ, these two species must have moved south by the
same distance during growth of P1, despite known differences in foraging range (Spear et al.
1995). In addition to uniformly positive values of ΔL, we observed that multiple feathers within
13

an individual begin their growth with roughly the same low δ C value at the tip of the vanes.
For this observation to be explained by foraging location, individuals must grow feathers only
while traveling south and always begin remex growth in the same location. Therefore,
individuals must make repeated trips south and north, growing their feathers only while moving
south. Because these scenarios seem improbable, we considered alternative explanations for our
data.
13

15

In addition to foraging location, trophic level is known to influence δ C and δ N
15

13

consumers. Specifically, within marine food webs, δ N shifts by ~3‰ and δ C shifts by ~1‰
(Michener & Schell 1994). If trophic level controls the longitudinal trends in our data, ΔL values
must reflect changes in trophic level during feather growth. However, in the Hawaiian Petrel, ΔL
13

15

for δ C (1.3‰) represents an increase of at least one full trophic level, while the ΔL for δ N
(0.6‰) signifies a shift of 0.2 trophic levels. This disparity is even larger for the Newell‘s
13

15

Shearwater, in which the average ΔL for δ C is 1.3‰, though only 0.4‰ for δ N. Therefore,
24

while changes in trophic level may contribute to intrafeather variation, they are unlikely to be the
predominant control.
As an alternative to trophic level and foraging location, we considered the possible role
of metabolism in controlling ΔL. There are two problems in formulating hypotheses related to
metabolism: little is known about avian intermediary metabolism, and few fractionation factors
(α) associated with metabolic transformations exist in birds. Because estimates of reaction rate
constants for the isotopically light and heavy molecules (required for determining α) are difficult
to estimate, isotope shifts are often described as a net isotope effects (NIE). A NIE is the
difference in isotope value between the substrate and product for a reaction consisting of
multiple steps (O‘Leary 1981; Farquhar et al. 1982). In fact, a NIE describes complicated
processes such as trophic fractionation despite uncertainties in intermediary metabolism and
associated fractionation factors. Thus, though speculative, it seems reasonable to consider
potential metabolic explanations for ΔL (a NIE) and differences in the magnitude of ΔL between
13

15

δ C and δ N.
13

15

Although the difference in ΔL between δ C and δ N might be explained as a difference
between carbon and nitrogen in the NIE for feather synthesis, this would not account for our
longitudinal trends. Instead, because the NIE is influenced by reservoir size, differences in the
sizes of carbon and nitrogen reservoirs may control ΔL. This reasoning requires an understanding
12

of fractionation. At the onset of the reaction, the light isotope ( C or

14

N) is transferred to the

product rapidly, yielding a product with an isotope value lower than that of the substrate. As the
substrate is converted to product, the ratio of

13

C to

12

C or

15

N to

14

N in the product increases

(O‘Leary 1981; Farquhar et al. 1982). Therefore, as the substrate reservoir diminishes, the
isotope value of the product will increase. As the reaction nears completion, the isotope value of
25

the product becomes similar to that of the initial substrate and an estimate of NIE based on the
difference between the isotope values of the initial substrate and final product approaches zero.
13

15

Consequently, an increase in δ C or δ N of the product is observed only if the reservoir is
large and depleted slowly. If the reservoir of a substrate is small or depleted quickly, a change in
the isotope value of the product and NIE is more difficult to detect.
In birds, carbon and nitrogen reservoirs differ in terms of their storage, potential size, and
utilization. The carbon in the feather protein keratin is derived from dietary protein or glucose.
Nitrogen, however, is ultimately derived from dietary protein. Whereas carbon can be stored in
the form of glycogen (the metabolic precursor to glucose), protein cannot be stored (Blem 1976,
Stevens 1996). Because molt is sometimes brief and requires synthesis of an amount of protein
equal to one fourth of a bird‘s total protein mass, nitrogen balance during molt can be difficult
(Myrcha & Pinowski 1970, Stevens 1996). Some herbivorous species may increase nitrogen
absorption and decrease nitrogen excretion during molt (Fox & Kahlert 1999), and others
develop feathers at the expense of muscle (Piersma 1988). Owing to their dependence on flight,
carnivorous seabirds, such as those we studied, are unlikely to rely heavily on muscle as a
protein reservoir but place a high demand on exogenous sources to meet nitrogen needs, such as
molt. Consequently, nitrogen is likely derived primarily or exclusively from the diet and is
rapidly depleted, making it difficult to observe isotopic variation in keratin as it is produced. This
15

is consistent with the low ΔL we observed for δ N. Because carbon used for keratin synthesis
can be derived from an endogenous carbon store (glycogen), there is greater potential for the
13

15

trend (e.g., large ΔL) in δ C to be larger than that in δ N. Although our data comparing
13

feathers are sparse, we did not find a continuous increase in the δ C from one feather to another
13

that grew at a later time (e.g., the δ C of the tips of secondary and primary feathers were
26

similar). Thus, if our explanation is correct, glycogen reservoirs are depleted and replenished
intermittently. Clearly, the potential for metabolic influences on isotope values feathers is an
important topic of future investigation.
The last factor that we considered as a control for isotopic variation within feathers was
pigmentation. In the species we studied, coloration is derived predominantly from eumelanin,
whose concentration varied substantially from the dark tip to the white base of each feather (see
Figure 2, AI and BI). Eumelanin is synthesized from the amino acid tyrosine, which is typically
depleted in

13

13

C relative to bulk tissue (McCullagh et al. 2005, 2006). If the δ C of eumelanin

mirrors its precursor tyrosine, isotope values of dark eumelanin-rich feather material should be
lower than in white eumelanin-free material. Indeed, we observed this pattern within all 52
15

Hawaiian Petrel and Newell‘s Shearwater feathers. Although the pattern of δ N in 29 of 42
15

Hawaiian Petrel feathers was parallel, eumelanin may alter δ N values to a lesser extent than it
13

13

15

does δ C values. Like that of δ C, the δ N value of tyrosine can be low relative to bulk tissue
(McClelland et al. 2003). However, the high C/N ratio of eumelanin (8 to 9) relative to that in
pigment-free feathers (3 to 4) indicates that eumelanin makes a smaller contribution to the
nitrogen reservoir than to the carbon reservoir (Jimbow et al. 1984, Tiquia et al. 2005, McGraw
15

et al. 2007). Any trend in δ N resulting from eumelanin concentration should therefore be
13

weaker than the analogous trend in δ C and more easily offset by other sources of isotopic
–1

variation. Given the low concentration of melanin within feathers (<1 to 60 mg g ), its isotope
values would need to be substantively different from those of pigment-free feather to cause the
observed longitudinal trends (McGraw 2006). While a longitudinal gradient in melanin
concentration is visible within every feather, we explored the possibility that C/N data might also
reflect this trend. The absence of clear longitudinal trends in C/N within Hawaiian Petrel feathers
27

likely reflects the difficulty in observing a small change in the contribution of melanin when
keratin predominates. Newell‘s Shearwater feathers showed small but observable longitudinal
trends. We had seven-section data on C/N for four feathers (one Hawaiian Petrel and three
Newell‘s Shearwater P1). The clear linear decrease in C/N observed in the three Newell‘s
Shearwater feathers is consistent with the interpretation that the negative ΔM and ΔL observed in
all Newell‘s Shearwater feathers reflects a longitudinal decrease in the contribution of melanin.
We are unable to assess the influence of melanin further because the isotope values of melanins
and their carbon contribution to feathers are unknown. Because melanin pigments are nearly
ubiquitous in birds, their influence on isotope values deserves further attention.

Comparison of Sampling Protocols
Among our approaches, the seven-section protocol sacrifices time and money for high resolution
and is therefore appropriate for studies of isotope variation within feathers. The three-section
protocol is less expensive and more practical for studies requiring longitudinal isotope data from
a large number of individuals. In contrast, the barb-sampling protocol is ideal when the average
isotope value of a feather is desired but longitudinal data are unnecessary. This protocol requires
minimal preparation time and expense. Because it also preserves the integrity of the feather, barb
sampling is a critical innovation for studies using museum specimens or other scenarios where
maintenance of feather integrity is desirable (Figure 7). For example, barb sampling may prove
useful for investigations of live individuals of threatened or endangered species. Barb sampling
of flight feathers is far less invasive than collection of entire remiges and provides a record of
diet over a longer term than a small section of vane. While collection of body contour feathers
from live birds is often an attractive option, isotope values from remiges are preferable in species

28

such as the Hawaiian Petrel in which the timing of body-feather molt cannot be well constrained
(Warham 1996).
Average isotope values for feathers may also be obtained by homogenizing entire
feathers, that is, reducing them to powder and taking an aliquot. Although this protocol requires
the isotopic analysis of only one sample, the barb protocol accomplishes the same task while
preserving most of the feather in its original form, avoiding any difficulties associated with
homogenization and leaving open the possibility of longitudinal sampling.
The barb protocol provided a reliable estimate of average isotope values in feathers of the
Hawaiian Petrel because it took into account both intrafeather isotope variation and mass
distribution, factors that may vary by species and feather type. Clearly, the barb protocol may
require modification for use in future studies, in which barb and whole-feather isotope values
should be compared to ensure accuracy. Even with modification, the barb protocol will
compromise the feather minimally because its premise is to take only the mass of barbs required
for isotope analysis (ca. 1.0 mg) from strategic locations along the feather. Owing to limitations
of sample size, the barb protocol offers the greatest advantage for analysis of large feathers in
which the mass of the vanes greatly exceeds that required for stable isotope analysis.

29

Figure 7. Photographs of P1 of a Hawaiian Petrel before and after sampling for isotope analysis
by the barb protocol. As illustrated, this sampling protocol preserves the gross morphology of the
feather, with minimal disruption to the vane.

AFTER

BEFORE

CONCLUSIONS
13

15

Variation in δ C and δ N within Hawaiian Petrel and Newell‘s Shearwater remiges has
15

important ecological implications. Because the range of 1.3‰ for δ N indicates that trophic
level varies by 33% within a feather, there is appeal in obtaining time-integrated average isotope
values of whole remiges. Establishing a time-integrated average isotope value is an important
objective in light of recent studies documenting historical declines in the trophic level of several
marine species (Jennings & Warr 2003, Norris et al. 2007). Moreover, we envision that increased
knowledge of variation within feathers may yield new insights into the metabolism and ecology
of birds.

30

LITERATURE CITED

31

Adams, J., D. Ainley, H. Freifeld, J. Penniman, F. Duvall, J. Tamayose, C. Bailey, N. Holmes,
M. Laut, and G. Spencer. 2006. Summer movements of ‗Ua‗u (Hawaiian Dark-rumped
Petrel Pterodroma phyaeopygia sandwichensis) nesting on Haleakala and Lanai‗i: can we
use satellite tracking to gain new information and advise conservation management?,
2006 Seabird conservation and management in the Hawaiian Islands, Kaneohe, Hawaii.
Ainley, D.G., T.J. Lewis, and S. Morrell. 1976. Molt in Leach's and Ashy Storm-Petrels. Wilson
Bulletin 88:76–95.
Ainley, D.G., T.C. Telfer, and M.H. Reynolds. 1997. Townsend's and Newell's Shearwater
(Puffinus auricularis), no. 297. In A. Poole and F. Gill [eds.], The birds of North
America. Academy of Natural Sciences, Philadelphia.
Bearhop, S., S. Waldron, S.C. Votier, and R.W. Furness. 2002. Factors that influence
assimilation rates and fractionation of nitrogen and carbon stable isotopes in avian blood
and feathers. Physiological and Biochemical Zoology 75:451–458.
Birchall, J., T.C. O'Connell, T.H.E. Heaton, and R.E.M. Hedges. 2005. Hydrogen isotope ratios
in animal body protein reflect trophic level. Journal of Animal Ecology 74:877–881.
Blem, C. R. 1976. Patterns of lipid storage and utilization in birds. American Zoologist 16:671–
684.
Cherel, Y., K. Hobson, F. Bailleul, and R. Groscolas. 2005. Nutrition, physiology, and stable
isotopes: new information from fasting and molting penguins. Ecology 86:2881–2888.
Cherel, Y., R.A. Phillips, K.A. Hobson, and R. McGill. 2006. Stable isotope evidence of diverse
species-specific and individual wintering strategies in seabirds. Biology Letters 2:301–
303.
Cormie, A.B., and H.P. Schwarcz. 1994. Relation between hydrogen isotopic ratios of bone
collagen and rain. Geochimica et Cosmochimica Acta 58:377–391.
Craig, H. 1961. Standard for reporting concentrations of deuterium and oxygen-18 in natural
waters. Science 133:1833–1834.
Dalerum, F., and A. Angerbjorn. 2005. Resolving temporal variation in vertebrate diets using
naturally occurring stable isotopes. Oecologia 144:647–658.
Estep, M. F., and H. Dabrowski. 1980. Tracing food webs with stable hydrogen isotopes. Science
209:1537–1538.
Farquhar, G.D., M.H. O‘Leary, and J.A. Berry. 1982. On the Relationship Between Carbon
Isotope Discrimination and the Intercellular Carbon Dioxide Concentration in Leaves.
Functional Plant Biology 9 (2): 121-137.
32

Fox, A.D., and J. Kahlert. 1999. Adjustments to nitrogen metabolism during wing moult in
Greylag Geese, Anser anser. Functional Ecology 13:661–669.
Goericke, R., and B. Fry. 1994. Variations of marine plankton δ13C with latitude, temperature,
and dissolved CO2 in the world ocean. Global Biogeochemical Cycles 8:85–90.
Hobson, K.A., and R.G. Clark. 1992a. Assessing avian diets using stable isotopes I: turnover of
13C in tissues. Condor 94:181–188.
Hobson, K.A., and R.G. Clark. 1992b. Assessing avian diets using stable isotopes II: factors
influencing diet–tissue fractionation. Condor 94:189–197.
Hobson, K.A., L. Atwell, and L.I. Wassenaar. 1999. Influence of drinking water and diet on the
stable-hydrogen isotope ratios of animal tissues. Proceedings of the National Academy of
Sciences of the United States of America 96:8003–8006.
Jehl, J.R. Jr. 1982. The biology and taxonomy of Townsend's Shearwater. Gerfaut 72:121–135.
Jennings, S., and K.J. Warr. 2003. Environmental correlates of large-scale spatial variation in the
15
δ N of marine animals. Marine Biology 142:1131–1140.
Jimbow, K., Y. Miyake, K. Homma, K. Yasuda, Y. Izumi, A. Tsutsumi, and S. Ito. 1984.
Characterization of melanogenesis and morphogenesis of melanosomes by
physicochemical properties of melanin and melanosomes in malignant melanoma.
Cancer Research 44:1128–1134.
Kelly, J.F. 2000. Stable isotopes of carbon and nitrogen in the study of avian and mammalian
trophic ecology. Canadian Journal of Zoology 78:1–27.
Knoff, A.J., S.A. Macko, R.M. Erwin, and K.M. Brown. 2002. Stable isotope analysis of
temporal variation in the diets of pre-fledged Laughing Gulls. Waterbirds 25:142–148.
Langston, N.E., and S. Rohwer. 1996. Molt–breeding tradeoffs in albatrosses: life history
implications for big birds. Oikos 76:498–510.
Lecuyer, C., P. Gillet, and F. Robert. 1997. The hydrogen isotope composition of seawater and
the global water cycle. Chemical Geology 145:249–261.
Martínez del Rio, C., N. Wolf, S.A. Carleton, and L.Z. Gannes. 2008. Isotopic ecology ten years
after a call for more laboratory experiments. Biological Reviews 84:91–111.
McClelland, J.W., C.M. Holl, and J.P. Montoya. 2003. Relating low δ15N values of zooplankton
to N2-fixation in the tropical North Atlantic: insights provided by stable isotope ratios of
amino acids. Deep-Sea Research I 50:849–861.
33

McCullagh, J.S.O., J.A. Tripp, and R.E.M. Hedges. 2005. Carbon isotope analysis of bulk
keratin and single amino acids from British and North American hair. Rapid
Communications in Mass Spectrometry 19:3227–3231.
McCullagh, J.S.O., D. Juchelka, and R.E.M. Hedges. 2006. Analysis of amino acid 13C
abundance from human and faunal bone collagen using liquid chromatography/isotope
ratio mass spectrometry. Rapid Communications in Mass Spectrometry 20:2761–2768.
McGraw, K.J. 2006. Mechanics of melanin-based coloration, p. 243–294. In G. E. Hill and K. J.
McGraw [eds.], Bird coloration. Harvard University Press, London.
McGraw, K.J., M.B. Toomey, P.M. Nolan, N.I. Morehouse, M. Massaro, and P. Jouventin. 2007.
A description of unique fluorescent yellow pigments in penguin feathers. Pigment Cell
Research 20:301–304.
McKechnie, A.E., B.O. Wolf, and C. Martínez del Rio. 2004. Deuterium stable isotope ratios as
tracers of water resource use: an experimental test with Rock Doves. Oecologia 140:191–
200.
Michener, R.H., and D.M. Schell. 1994. Stable isotope ratios as tracers in marine aquatic food
webs, p. 138–157. In K. Lajtha and R. Michener [eds.], Stable isotopes in ecology and
environmental science. Blackwell Scientific, Oxford, England.
Myrcha, A., and J. Pinowski. 1970. Weights, body composition, and caloric value of postjuvenal
molting European Tree Sparrows (Passer montanus). Condor 72:175–181.
Newsome, S., C. Martínez del Rio, S. Bearhop, and D.L. Phillips. 2007. A niche for isotopic
ecology. Frontiers in Ecology and the Environment 5:429–436.
Norris, D.R., P. Arcese, D. Preikshot, D.F. Bertram, and T.K. Kyser. 2007. Diet reconstruction
and historic population dynamics in a threatened seabird. Journal of Applied Ecology
44:875–884.
O'Leary, M.H. 1981. Carbon isotope fractionation in plants. Phytochemistry 20:553.
Piersma, T. 1988. The annual moult cycle of Great Crested Grebes. Ardea 76:82–95.
Pyle, P. 2008. Identification guide to North American Birds, part 2. Slate Creek Press, Point
Reyes Station, CA.
Schoeller, D., C. Leitch, and C. Brown. 1986. Doubly labeled water method: in vivo oxygen and
hydrogen isotope fractionation. American Journal of Physiology 251:R1137–R1143.
Sears, J., S. Hatch, and D.M. O'Brien. 2009. Disentangling effects of growth and nutritional
status on seabird stable isotope ratios. Oecologia 159:41–48.

34

Simons, T.R. 1985. Biology and behavior of the endangered Hawaiian Dark-rumped Petrel.
Condor 87:229–245.
Simons, T.R., and C.N. Hodges. 1998. Dark-rumped Petrel (Pterodroma phaeopygia), no. 345. In
A. Poole and F. Gill [eds.], The birds of North America. Birds of North America, Inc.,
Philadelphia.
Sincock, J.L., and G.E. Swedberg. 1969. Rediscovery of the nesting grounds of Newell's Manx
Shearwater (Puffinus puffinus newelli) with initial observations. Condor 71:69–71.
Smith, A.D., K. Donohue, and A.M. Dufty. 2008. Intrafeather and intraindividual variation in the
stable-hydrogen isotope (δD) content of raptor feathers. Condor 110:500–506.
Spear, L.B., D.G. Ainley, N. Nur, and S.N.G. Howell. 1995. Population size and factors affecting
at-sea distributions of four endangered procellariids in the tropical Pacific. Condor
97:613–638.
Stevens, L. 1996. Avian biochemistry and molecular biology. Cambridge University Press, New
York.
Takai, N., S. Onaka, Y. Ikeda, A. Yatsu, H. Kidokoro, and W. Sakamoto. 2000. Geographical
variations in carbon and nitrogen stable isotope ratios in squid. Journal of Marine
Biological Association of the United Kingdom 80:675–684.
Thompson, D.R., and R.W. Furness. 1995. Stable-isotope ratios of carbon and nitrogen in
feathers indicate seasonal dietary shifts in Northern Fulmars. Auk 112:493–498.
Tiquia, S.M., J.M. Ichida, H.M. Keener, D.L. Elwell, E.H. Burtt Jr., and F.C. Michel Jr. 2005.
Bacterial community profiles on feathers during composting as determined by terminal
restriction fragment length polymorphism analysis of 16S rDNA genes. Applied
Microbiology and Biotechnology 67:412–419.
Warham, J. 1996. The behaviour, population biology, and physiology of the petrels. Academic
Press, Christchurch, New Zealand.
Wassenaar, L.I., and K.A. Hobson. 2003. Comparative equilibration and online technique for
determination of non-exchangeable hydrogen of keratins for use in animal migration
studies. Isotopes in Environmental Health Studies 39:211–217.
Wassenaar, L.I., and K.A. Hobson. 2006. Stable-hydrogen isotope heterogeneity in keratinous
materials: mass spectrometry and migratory wildlife tissue subsampling strategies. Rapid
Communications in Mass Spectrometry 20:2505–2510.
Wolf, N., S.A. Carleton, and C. Martínez del Rio. 2009. Ten years of experimental animal
isotopic ecology. Functional Ecology 23:17–26.

35

Wong, W.W., L.L. Clarke, G.A. Johnson, M. Llaurador, and P.D. Klein. 1992. Comparison of
two elemental-analyzer gas-isotope-ratio mass spectrometer systems in the simultaneous
measurement of carbon-13/carbon-12 ratios and carbon content in organic samples.
Analytical Chemistry 64 (4): 354-358.

36

CHAPTER 2
FORAGING SEGREGATION AND GENETIC DIVERGENCE BETWEEN
GEOGRAPHICALLY PROXIMATE COLONIES OF A HIGHLY MOBILE SEABIRD
ABSTRACT
Foraging segregation may play an important role in the maintenance of animal diversity and is a
proposed mechanism for promoting genetic divergence within seabird species. However, little
information exists regarding its presence among seabird populations. We investigated genetic
and foraging divergence between two colonies of endangered Hawaiian petrels (Pterodroma
sandwichensis) nesting on the islands of Hawaii and Kauai using the mitochondrial Cytochrome
13

15

b gene and carbon, nitrogen and hydrogen isotope values (δ C, δ N and δD, respectively) of
feathers. Genetic analyses revealed strong differentiation between colonies on Hawaii and Kauai,
with ST = 0.50 (p < 0.0001). Coalescent-based analyses gave estimates of < 1 migration event
13

15

per 1,000 generations. Hatch-year birds from Kauai had significantly lower δ C and δ N
values than those from Hawaii. This is consistent with Kauai birds provisioning chicks with prey
derived from near or north of the Hawaiian Islands, and Hawaii birds provisioning young with
15

prey from regions of the equatorial Pacific characterized by elevated δ N values at the food web
15

base.  N values of Kauai and Hawaii adults differed significantly, indicating additional
foraging segregation during molt. Feather δD varied from -69 to 53 ‰. This variation cannot be
related solely to an isotopically homogeneous ocean water source or evaporative water loss.
Instead, we propose the involvement of salt gland excretion. Our data demonstrate the presence
of foraging segregation between proximately nesting seabird populations, despite high species

37

mobility. This ecological diversity may facilitate population coexistence and its preservation
should be a focus of conservation strategies.

INTRODUCTION
As consumers of widely dispersed marine prey, pelagic seabirds are noted for their remarkably
long foraging trips that may extend thousands of kilometers (Warham 1996). However, the same
birds that travel great distances to forage may be reluctant to disperse among nearby breeding
sites, an apparent inconsistency referred to as the seabird paradox (Milot et al. 2008). For
example, Laysan albatross (Phoebastria immutabilis) regularly travel >1500 km away from
breeding sites in search of food, but typically breed within 20 km of their hatch site (Fisher 1967,
1976). In association with natal and breeding philopatry, some species show genetic
differentiation between breeding colonies (Friesen et al. 2007). Contact between seabird colonies
may be further reduced by foraging segregation (Friesen et al. 2007).
Foraging segregation is thought to play an important role in the maintenance of species
diversity by reducing competition and allowing seabirds with otherwise similar niches to coexist
(Wemerskirch et al. 1986; Ainley et al. 1992; Spear et al. 2007). Sympatric seabird species are
known to segregate by feeding niche in a variety of ways, including feeding location, diving
depth, and size of targeted prey (Ashmole & Ashmole 1967; Harris 1970; Bocher et al. 2000;
Mori and Boyd 2004; Finkelstein et al. 2006). Less is known about foraging segregation between
populations, partly due to the difficulty of differentiating seabird populations (Abbott & Double
2003; Friesen et al. 2006). However, foraging segregation has been found to occur within some
species: between sexes and in limited cases, between breeding colonies (Phillips et al. 2004;
Zavalaga et al. 2007). There is evidence that intra-specific competition for food among pelagic
38

seabirds can be intense, and this competition may drive foraging segregation among colonies
(Lewis et al. 2001). Foraging segregation, in turn, may play a central role in producing and
maintaining both distinct populations and ecologically specialized morphotypes within a species
(Gilardi 1992; Zavalaga et al. 2007).
We investigated both foraging segregation and population divergence between breeding
colonies of the Hawaiian petrel (Pterodroma sandwichensis, or U‘au). This endangered species
exemplifies the seabird paradox, sometimes traveling over 10,000 km on individual foraging
trips, but breeding no more than 500 km apart on the islands of Hawaii and Kauai in the
Hawaiian Archipelago (Adams & Flora 2009; Simons 1985). Hawaiian petrel foraging range
extends throughout most of the North Pacific, from the Aleutian Islands to the Equator (Bourne
1965; Bourne & Dixon 1975; Simons et al. 1998; Spear et al. 1995). However, there is only one
recent study providing colony-level information on foraging range (J. Adams and D. Ainley et
al. unpublished data). Hawaiian petrels from different breeding colonies are morphologically
indistinguishable, and little is known about population structure or genetic variation within the
species (Browne et al. 1997). Because Hawaiian petrels are endangered and logistically difficult
to study, a better understanding of their foraging and genetic diversity is particularly valuable, as
it can inform future conservation management decisions.
In the current study, we examined genetic and ecological differentiation among Hawaiian
petrels breeding on the islands of Kauai and Hawaii (Figure 8). We focused on colonies at the
endpoints of the Hawaiian petrel breeding range because we considered them most likely to
display genetic and ecological differentiation. To investigate genetic population structure, we
sequenced the mitochondrial Cytochrome b gene, which has been widely used in genetic studies
of procellariiforms (Austin et al. 2004; Bretagnolle et al. 1998; Nunn & Stanley 1998). To study
39

potential variation in foraging ecology, we examined the stable isotope composition of flight
feathers from hatch-year and adult petrels. Specifically, we assembled information on trophic
dynamics and foraging location using stable carbon and nitrogen isotopes (Michener & Schell
1994).We also examined the hydrogen isotope composition of Hawaiian petrel feathers. In North
American terrestrial environments, hydrogen isotope values of water fluctuate with latitude, and
this variation is transferred to organisms such as birds (Hobson et al. 1999). Such variation is not
expected in marine organisms that derive their hydrogen from the isotopically homogenous
ocean (Craig 1961; Lecuyer et al. 1997). Thus, an investigation of hydrogen isotopes in the
Hawaiian petrel can offer insight into factors other than latitude that may contribute to variation
within free-ranging birds, in general. Together, we used isotope and genetic techniques to gain a
better understanding of Hawaiian petrel ecological and genetic diversity.

40

Figure 8. Marine distribution (A) and breeding distribution (B) of the Hawaiian petrel. The dark blue area indicates the combined
spring and autumn distribution as reported by Spear et al. (1995). The light blue area approximates additional areas frequented by the
Hawaiian petrel, as inferred from Simons et al. (1998). Islands with modern breeding colonies are shaded red. Triangles mark the
origin of black-footed and laysan albatross in Figure 9. The solid line represents the Equator, while stippled lines represent, from south
to north, the Tropics of Capricorn and Cancer, and the Arctic Circle. For interpretation of the references to color in this and all other
figures, the reader is referred to the electronic version of this dissertation.

41

MATERIALS AND METHODS
Sampling
We sampled a total of 80 Hawaiian petrels, 40 from the island of Kauai and 40 from Hawaii. All
individuals were included in genetic analyses, but samples for isotope analysis were only
available from 24 Hawaii and 25 Kauai birds. We collected samples from carcasses of birds that
were depredated by introduced mammals or that, upon attraction to lights, had collided with
human-constructed objects between 1989 and 2009. We also obtained samples from birds
rehabilitated after grounding, as well as from two individuals that were collected and prepared as
museum skins in 1980 and 1995.
For genetic analyses, we extracted DNA from blood, tissue, feather, toe pad, or bone. For
stable isotope analyses, we sampled the same flight feather, primary 1 (P1), from salvaged
carcasses of both adult and hatch-year birds (<one year in age). Flight feathers in hatch-year
birds are grown during the late growth stages, from September to December (Ainley et al. 1997;
Simons 1985). As in other Pterodroma, adult Hawaiian petrel presumably begin molt
immediately after cessation of nest attendance (December for breeding adults), with primary
molt proceeding distally from P1 (Pyle 2008; Warham 1996). We also sampled P1 from salvaged
carcasses of eight adult Newell‘s shearwaters (Puffinus auricularis newelli) collected between
2002 and 2006. We included this species in our analyses for the sake of comparison, as molt
location can be constrained to the eastern tropical North Pacific (Spear et al. 1995). Similar to
the Hawaiian petrel, Newell‘s shearwaters molt their remiges at sea following departure from
breeding colonies (Jehl 1982). Molt likely proceeds distally from P1, as in other shearwaters
(Pyle 2008).
To enable a comparison of isotope data from Hawaiian petrels and birds foraging in the
Transition Zone of the North Pacific (the region bordered by the subarctic and subtropical frontal
42

zones), we reference black-footed albatross (Ph. nigripes) and Laysan albatross muscle data
taken from birds killed in driftnet fisheries (Gould et al. 1997). We included only breeding age
adult albatross in our data set, all of which died between 40-45°N and 145°E and 145°W (Figure
8). Specifically, we selected individuals that were believed to have obtained their isotope values
at sea during the non-breeding season, as opposed to birds that recently migrated or provisioned
chicks. To this end, we reference data from adults that died between September and November,
at least three months after departure from their breeding colonies. Because different tissues may
vary in their isotopic enrichment relative to diet, we standardized stable isotope values from
muscle to those of feather by correcting for the average offset between muscle and flight feather
in hatch year Hawaiian petrels. Isotope values from both tissues reflect diet during the four
months of chick growth, most likely the latter two months (Hobson & Clark 1992; Simons 1985;
Tieszen et al. 1983). We felt this comparison was the best approximation of a difference in
tissue-specific fractionation factors for our species.

Genetic Analyses
Genetic analyses assessed levels of divergence and gene flow between petrels breeding on
Hawaii and Kauai. We extracted DNA from blood and tissue samples via the Qiagen DNEasy
Blood and Tissue kit, or through a standard phenol/chloroform procedure with centrifugal
dialysis for toe pad, feather, and bone samples (Fleischer et al. 2000). A 524 bp region of the
mitochondrial Cytochrome b gene was amplified using the primers CytbL and CytbR9 (Welch et
al. In Press), which were designed in Primer3 (Rozen & Skaletsky 2000) from petrel sequences
available on Genbank. However, some samples had low DNA concentration and/or quality, so
short overlapping primers were also developed to amplify the same region (Welch et al. In
Press). Polymerase chain reaction was performed in 15 L total reaction volumes containing 1X
43

PCR Gold Buffer (Applied Biosystems), 0.2 mM each dNTP, 2.5 to 3.5 mM MgCl2, 1.2 mg/mL
BSA (New England Biolabs), 0.5 M of each primer, 1 unit AmpliTaqGold DNA Polymerase
(Applied Biosystems) and 1 to 2 L DNA extract. DNA was initially denatured for 8 minutes at
94oC, followed by 35 to 40 cycles of 30 sec at 92oC, 30 sec at a primer specific annealing
temperature between 50 and 55oC, and 30 to 60 sec extension at 72oC, with a final extension step
of 7 min at 72oC. PCR products were cleaned up using a 1/10 dilution of ExoSAP-IT (USB)
before cycle sequencing in both directions using the Big Dye Terminator v3.1 Cycle-Sequencing
kit from Applied Biosystems. Cycle-sequencing reactions were purified using Sephadex G-50
fine columns prior to being electrophoresed in a 3130xl Genetic Analyzer (Applied Biosystems).
Sequences were aligned and visually inspected in Sequencher version (v) 4.8 (GeneCodes
Corporation).
We examined Cytochrome b sequences to confirm that they had a mitochondrial origin
and were not nuclear copies (Sorenson & Quinn 1998). Amplifications of the same individual
with overlapping fragments resulted in identical sequences, and no insertions, deletions, or
ambiguous bases were found. After translation in DAMBE 5.1.2, no gaps, non-sense or stop
codons were detected and most substitutions occurred in the third codon position (4, 0, and 10
substitutions for the first, second, and third codon position, respectively). Additionally, 13 of 14
observed substitutions were transitions and only two substitutions resulted in non-synonymous
changes. Therefore, it is unlikely that any of these sequences represent nuclear copies.
Haplotype diversity and nucleotide diversity were calculated in Arlequin v3.1 (Excoffier
et al. 2005). An Analysis of Molecular Variance (AMOVA) was also performed with
significance determined through 1000 permutations. The Hasegawa-Kishino-Yano model plus a
proportion of invariant sites (HKY + I) was selected as the best-fit substitution model under the
44

Akaike information criterion in jModelTest (Posada 2008; Posada & Buckley 2004). However
this model was not available in Arlequin so the Kimura two-parameter (K80) model, which is
very similar, was used instead. A 95% statistical parsimony haplotype network was created in
TCS v1.21 (Clement et al. 2000). Networks graphically depict relationships among haplotypes in
closely related populations where ancestral alleles may still be present (Posada & Crandall
2001). Sequence divergence within each population as well as the net divergence between
populations was calculated in Mega4 (Tamura et al. 2007) using a K80 distance matrix. Standard
error was estimated through 500 bootstrap replicates.
Migration rates were estimated in the coalescent-based program MIGRATE-N v3.0.3
(Beerli & Felsenstein 1999). As opposed to summary statistics, coalescent-based analyses take
into account the genealogical relationships of the samples to estimate demographic parameters
(Kuhner 2009). In MIGRATE, estimates are compounded by the mutation rate per generation, 
(i.e. m = M, where M is the estimate of migration). The mutation rate was assumed to be 1.89 x
-8

10

substitutions/year, as estimated for the Cytochrome b gene in procellariiform seabirds (Weir

& Schluter 2008). The generation time for Hawaiian petrels is unknown, however age at first
breeding is correlated with body size in seabirds of this order (Warham 1990). Petrels similar in
size to the Hawaiian petrel begin to breed at approximately 5 to 6 years of age (Simons 1984)
and albatrosses begin to breed regularly by about 12 years of age (Warham 1996). The age at
first breeding may not be equivalent to generation time, so demographic estimates were
calculated using both 6 and 12 years. Interpretations were consistent regardless of the generation
time used, so results are presented for a conservative generation time estimate of 12 years.
For all MIGRATE analyses we employed the Bayesian inference method with uniform
priors (Holder & Lewis 2003). The analyses utilized a single long chain and three heated chains
45

of 10 million steps each with the first 10% discarded as burn-in. The static heating scheme was
used with temperatures set to 1.0, 1.5, 3.0 and 6.0. This analysis was conducted concurrently
with two additional replicates, and repeated independently so that runs could be compared to
assess convergence. In all runs the effective sample size (ESS) was greater than 7,000. While
two intervening Hawaiian petrel populations are not represented in the current study, estimates of
migration obtained from MIGRATE have been found to be robust as long as migration rates from
the unsampled populations into the sampled populations are of approximately the same
magnitude or lower than migration rates between the sampled populations (Beerli 2004).
The program Bayesian Evolutionary Analysis by Sampling Trees, or BEAST, v1.5.4
(Drummond & Rambaut 2007) was employed to estimate the divergence time between the Kauai
and Hawaii petrel populations. The strict molecular clock model (Arbogast et al. 2002) was
implemented with the substitution rate fixed at 1.89 x 10

-8

as above. The HKY + I substitution

model was implemented as selected by the AIC. A corresponding 524 bp Cytochrome b
sequence (Genbank accession number HQ420337) from the Galapagos petrel (P. phaeopygia),
the sister taxon of the Hawaiian petrel, was aligned to Hawaiian petrel sequences to root the tree.
Therefore, a Yule tree prior was utilized for dating analyses. Analyses were conducted for 10
million generations with the first 10% excluded as burn-in. Stationarity was assessed in Tracer
v1.5 (Drummond & Rambaut 2007) through plots of the –lnL across generations and
examination of ESS, which were greater than 500 for all parameters. Multiple independent runs
were compared to assess convergence.

46

Stable Isotope Analyses
All feathers were washed in solvent (87:13 chloroform:methanol by volume), rinsed with
ultrapure distilled water (E-Pure, Barnstead), and dried at 25°C in a vacuum oven. We then took
samples representative of the entire feather vanes using the barb or 3-section protocols described
13

15

in Wiley et al. (2010). For δ C and δ N analysis, we weighed 0.8-1.0 mg aliquots of feather
into tin capsules and analyzed them with an elemental analyzer (Eurovector) interfaced to an
Isoprime mass spectrometer (Elementar; Wong et al. 1992). For δD analysis, we weighed 0.5 mg
aliquots of feather vane into silver capsules and allowed them to air-equilibrate with ambient
laboratory conditions for a minimum of 2 weeks prior to δD analysis (Wassenaar & Hobson
2003). Following equilibration to local conditions, samples were pyrolyzed at 1425°C in a high
temperature elemental analyzer (TC/EA, Thermo-Finnigan) interfaced to an isotope ratio mass
spectrometer (Thermo-Finnigan DeltaPlus XL).
Stable isotope values are expressed in per mil (‰) as:
δX = ([Rsample/Rstandard] – 1) × 1000
where X is

13

C,

15

N, or D and R is the corresponding ratio
13

15

13

12

C/ C,

15

14

N/ N, or D/H. Rstandard
13

15

is V-PDB, air, and V-SMOW for δ C, δ N, and δD, respectively. For δ C and δ N, we
analyzed laboratory standards between every five unknowns, with a precision of ≤ 0.2 ‰ for
13

15

both δ C and δ N. Non-exchangeable δD values were normalized to in-house keratin
standards (LA bear hair, -78.1 ‰, AK bear hair, -171.5 ‰) calibrated to BWB-CHS-CFS
(Wassenaar & Hobson 2003). Analytical precision and accuracy were ≤ 4 ‰. Because many of
our unknown samples were outside of the calibration range, we performed several experiments in
an effort to constrain extrapolation error (Kelly et al. 2009). Specifically, water standards (VSMOW, 0 ‰; V-SLAP, -428 ‰) were dispensed (~0.25 µL) into silver capsules, crimp sealed,

47

and analyzed against solid inorganic (IAEA-CH7, -100 ‰; benzoic acid, -61 ‰) standards. The
experiments confirmed linearity when data are extrapolated 100 ‰ beyond our calibration range,
and as a consequence of this extrapolation our error increases to ± 8 ‰. We recognize that
matrix differences confound the interpretation of this test, but complex organic standards with
high δD values are currently lacking, though desperately needed (Kelly et al. 2009). We also
analyzed the δD composition of muscle from a variety of species from the Pacific Ocean.
Whereas hydrogen exchangeability in muscle has received little study compared to keratin, the
magnitude of exchange appears to be similar (Wassenaar and Hobson 2000; Chesson et al.
2009). To this end, we analyzed lipid-extracted muscle tissue in the same fashion as feather
samples, including normalization to keratin standards (Doucett et al. 2007; Finlay et al. 2010).

Statistical Analyses
We assessed differences in isotope values between island and age groups through a multiple
analysis of variance (MANOVA) using the general linear model procedure (GLM) in SAS 9.2
(SAS Institute Inc. 2009). We assessed temporal variation in isotope values in the same manner,
with year of feather growth, El Niño-Southern Oscillation (ENSO; cold, neutral, or warm phase
as indicated by the National Weather Service Climate Prediction Center), and century (before or
after 2001) designated as independent variables in the MANOVA. For the evaluation of ENSO,
2006 and 2009 were considered warm phase years, and 1989, 1998, and 2000 were categorized
as cold phase. We performed separate MANOVA‘s for individuals from Hawaii, adults from
Kauai, and hatch-year birds from Kauai. Age groups from Hawaii were combined for this
temporal analysis because they showed no significant differences in any isotope values (see
Results). Among Kauai adults, only the effect of year could be incorporated into our statistical

48

model due to the lack of variation in century and ENSO phase. For all statistical models, normal
quantile-quantile plots and Levene‘s tests were used to check assumptions of normality and
homogeneity of variances. In figures and text, isotope values for island and age groups are
reported as averages ± standard errors.

RESULTS
Genetic Analyses
We sequenced 524 bp of the mitochondrial Cytochrome b gene in 80 Hawaiian petrels, 40 from
each island (Genbank accession numbers JF264905-72, HQ420355-60, HQ420366-68,
HQ420372-74). Eleven haplotypes were discovered from 14 polymorphic sites. Six haplotypes
occurred in individuals from Hawaii and eight occurred in individuals from Kauai, with three
shared between the two islands. However, shared haplotypes were found at different frequencies
(0.05 vs. 0.70, 0.28 vs. 0.05, and 0.15 vs. 0.03 for Hawaii and Kauai populations, respectively;
Figure 9). The majority of Kauai birds shared a single haplotype, whereas two haplotypes were
common in Hawaii individuals. Three to four mutational steps separated the most common
haplotypes in each population. Gene diversity was significantly higher (p < 0.05) in petrels from
Hawaii (0.74  0.04) than Kauai (0.51  0.09). Nucleotide diversity was similar between the two
populations (0.0037  0.0024 for Hawaii, and 0.0031  0.0021 for Kauai).
AMOVA results revealed high levels of differentiation between petrels from Kauai and
Hawaii with ST = 0.50 (p < 0.0001). Results from MIGRATE concurred. All MIGRATE runs
produced distinct, unimodal posterior distributions which overlapped between independent runs.
We found migration to be low and roughly symmetrical between populations of petrels on Kauai
and Hawaii. Migration of Hawaii individuals into the Kauai population was estimated at 1.2 x
-4

-6

-4

10 migrants per generation (95% Highest Posterior Density (HPD): 4.5 x 10 to 4.8 x 10 )
49

-5

while migration rate from Kauai into the Hawaii population was 6.6 x 10 migrants per
-4

generation (95% HPD: 0.0 to 3.9 x 10 ). Analyses in BEAST indicated that the mean population
divergence time was 118,000 years ago (95% HPD: 61,800 to 185,000 years ago). Corrected
pairwise sequence divergence between the two populations was 0.22%  0.40%.

50

Figure 9. Haplotype network of Cytochrome b sequences of Hawaiian petrels from Hawaii
(black) and Kauai (grey). Each circle represents a unique sequence and the size of the circle is
proportional to the number of individuals possessing that sequence (sample sizes are indicated
for the most common haplotypes). Each line represents one mutational step, and small white
circles represent an unsampled intermediate sequence.
Hawaii
Kauai

N =7

N = 16

N = 30

N = 13

Stable Isotope Analyses
13

Among 49 Hawaiian petrels, δ C values varied between -17.7 ‰ and -14.6 ‰ and the range in
15

δ N values was 10.6 ‰ to 17.6 ‰. While adults from the islands of Hawaii and Kauai did not
13

15

differ in δ C (p=0.7243), Hawaii adults had significantly higher δ N values as compared with
those from Kauai (Figure 10, p<0.0001). Among Hawaii individuals, hatch-year birds and adults
13

15

had similar isotope values and showed no significant difference in either δ C or δ N
13

15

(p=0.5721 for δ C and p=0.9581 for δ N). In contrast, hatch year birds from Kauai had
51

15

13

significantly lower δ N and δ C values than adults from the same island (p<0.0001). Kauai
13

hatch-year birds were also lower in δ C than adults and hatch-year birds from the island of
15

Hawaii (p≤ 0.0001). Eight Newell‘s shearwaters from Kauai showed a broad range of δ N
values between 8.5 ‰ and 17.6 ‰, while their δ13C values ranged from -16.1 ‰ to -15.3 ‰.
Stable hydrogen isotope values varied extensively among Hawaiian petrels, from -69 ‰
to 53 ‰. Hatch-year birds from Kauai had an average δD value of -60 ± 3 (SE) ‰, at least 79 ‰
lower than any other group (Figure 11, p<0.0001 for all comparisons). Kauai adults, Hawaii
adults, and Hawaii hatch-year birds had average δD values of 20 ± 6 ‰, 31 ± 4 ‰, and 19 ± 5
‰, respectively.
We found no significant effect of year, century, or ENSO phase on stable isotope values.
Wilks‘ Lambda test for the effects of year and ENSO had values > 0.67 (p > 0.25) and
Hotelling‘s Trace test values for the effect of century were < 0.49 (p > 0.34).
Among 12 hatch year Hawaiian petrels, flight feathers were 1.5 ± 0.2 ‰ and 0.2 ±
13

15

0.4 ‰ higher in average  C and  N, respectively, than muscle.

52

Figure 10. Stable carbon and nitrogen isotope values from Hawaiian petrels and adult Newell's
shearwater flight feathers and from non-breeding adult albatross muscle. Stable isotope values
from muscle are standardized to feather isotope values, as indicated in the Materials and
Methods. Sample sizes for Hawaiian petrel data are as follows: Hawaii adults (N=14), Hawaii
hatch-year birds (N=10), Kauai adults (N=13), and Kauai hatch-year birds (N=12). Remaining
sample sizes are: Newell's shearwaters (N=8), Laysan albatross (N=55), and black-footed
albatross (N=16). All albatross derive from 40-45°N and 145°W to 145°E (Transition Zone), as
shown in Figure 8.

17
Black-footed
albatross

16

Hawaii hatch-years &
adults

albatross
spp.

δ15N (‰)

15
14
13

Kauai adults

Laysan albatross

Hawaiian
petrels

12
11
10
-18.0

Newell's
shearwaters

Kauai hatch-years

-17.5

-17.0

-16.5

-16.0

δ13C (‰)

53

-15.5

-15.0

Figure 11. Stable hydrogen isotope values in Hawaiian petrel flight feathers. Sample sizes are as
follows: Hawaii adults (N=14), Hawaii hatch-year birds (N=10), Kauai adults (N=13), and Kauai
hatch-year birds (N=12).

DISCUSSION
Genetic Analyses
Genetic analyses indicate that Hawaiian petrels nesting on the islands of Hawaii and Kauai are
significantly differentiated. While Hawaiian petrels are known to make foraging trips to the Gulf
of Alaska during the breeding season, it appears that birds rarely disperse between these two
islands, which lie just 500 km apart. Three haplotypes were found to be shared between islands,
however, major differences in their frequencies, as demonstrated by the high level of ST,
suggest low migration. Sophisticated coalescent-based analyses also indicate that gene flow is

54

very low, with less than one migrant per 1,000 generations. While Hawaiian petrels from other
islands were not included in these analyses, migration estimates from MIGRATE, such as those
presented here, have been found to be robust as long as gene flow is low to moderate from the
missing populations (Beerli 2004). Even if gene flow from missing populations was high, the
migration rates obtained here would represent overestimates (Beerli 2004). Therefore dispersal
between Kauai and Hawaii appears to be low. While these results are based on mitochondrial
DNA sequences, and therefore only reflect female dispersal, male dispersal may also be low.
Previous analyses of three nuclear intron loci from a subset of Hawaiian petrels from Kauai and
Hawaii show similarly low levels of gene flow (Welch et al. In Press). Similarly, using a
microsatellite data set Friesen et al. (2006) found that populations of the Hawaiian petrel‘s sistertaxon, the Galapagos petrel, are also highly differentiated (FST between 0.07 and 0.26, with p <
0.01 for all comparisons).
BEAST analyses indicate that populations on Hawaii and Kauai may have diverged as
recently as 100,000 years ago. Therefore, it is likely that the shared haplotypes between Hawaii
and Kauai are a result of common ancestry. Corrected pairwise sequence divergence between the
two populations of 0.22% is low, but similar to levels of divergence between some petrel and
shearwater subspecies (Austin et al. 2004; Techow et al. 2009).

Stable Isotope Analyses
Consistent with our genetic analyses, stable carbon and nitrogen isotopes reveal differentiation
between Hawaiian petrels nesting on the islands of Kauai and Hawaii. Specifically, there is a
15

15

disparity in average δ N values between adults from the two islands, and a difference in δ N
13

and δ C between hatch-year birds. To help explain factors that contribute to isotope variation,

55

we present data from other non-breeding, pelagic seabirds in the North Pacific (Figure 10). From
stable isotope and stomach content data, Gould et al. (1997) showed that black-footed albatross
associated with drift net fisheries fed approximately one trophic level higher than Laysan
15

13

albatross. While the albatross species show a trophic shift in both δ N and δ C, a similar
13

difference in δ C is not observed between adult Hawaiian petrels from Kauai and Hawaii. Thus,
trophic level, alone, does not account for the isotopic variation between Hawaiian petrel
15

populations. The likelihood that δ N is not a simple function of trophic level in the Hawaiian
petrel is further suggested by a comparison of Hawaiian petrels from the island of Hawaii and
15

black-footed albatross. While these two groups have similarly high δ N values, they are
unlikely to feed at the same trophic level for several reasons. First, the two species differ greatly
in size (e.g. culmen or bill length of 97 cm vs. 33 cm), and are therefore likely to target prey of
different size and dissimilar trophic levels (Adams & Brown 1989; Ashmole & Ashmole 1967;
Frings & Frings 1961; Simons 1985; Spear et al. 2007). Second, these particular black-footed
albatross fed heavily on large neon flying squid (Ommastrephes bartrami, >40 cm in mantle
length; Gould et al. 1997). Unless obtained as a fishery subsidy, the large size of neon flying
squid makes it an unlikely food source for Hawaiian petrels. The use of fishery subsidies is also
improbable because Hawaiian petrels have not been reported near fishing vessels, have not been
reported to feed on fishing offal, and do not closely associate with fisheries (as evidenced by
records from the Burke Museum showing only three Hawaiian petrels associated with the North
Pacific driftnet fishery, and P. Gould, personal communication). It is thus improbable that the
two species forage at similar trophic levels.
15

Variation in δ N between Hawaiian petrel populations is most likely associated with
differences in foraging location and spatial variation in biogeochemical cycling at the base of the
56

food web. A well-characterized nitrogen isotopic gradient in the eastern tropical North Pacific
occurs within the foraging range of the Hawaiian petrel, with the most elevated sedimentary
15

δ N values observed at 4-10° N and 135-140°W (Figure 12, location approximated by red
circle). This isotopic pattern arises from spatial variation in nitrogen cycling. In upwelling areas
off the coast of Peru, high nitrate concentrations fuel denitrification, which produces a pool of
residual nitrate that is markedly enriched in

15

N (Liu & Kaplan 1989; Farrell et al. 1995). A

15

further increase in the δ N of nitrate results from phytoplankton uptake (Altabet et al. 1991;
Altabet & Francois 1994). Thus, phytoplankton uptake in the eastern tropical North Pacific
15

results in a conspicuous meridional pattern whereby the δ N of nitrate increases with distance
15

from the locus of upwelling (Farrell et al. 1995). Such variation in δ N of nitrate is mirrored in
15

phytoplankton. Consequently, high δ N values for nitrate and sediments, near 12-15 ‰,
characterize areas where phytoplankton utilize a

15

N enriched reservoir of residual nitrate, as

opposed to values of ≤10 ‰ for sediments derived from phytoplankton dependent on average
marine nitrate (6 ‰) or fixed N2 (0 ‰) (Altabet & Francois 1994).

57

Figure 12. Fall distribution of the Hawaiian petrel, as observed by Spear et al. 1995, outlined in black. Shades of blue represent
nitrogen isotope values of a mobile consumer, yellowfin tuna (Graham et al. 2010). Green-labeled points represent stable nitrogen
1
2
4
isotope values of sinking and floating sediments: Karl et al. 1997 (annual mean export pulse), Altabet and Francois 1994, and Voss
2
3
2001. White-labeled points represent stable nitrogen isotope values of core top sediments: Altabet and Francois 1994 and Farrell et
4
al. 1995, and Voss 2001. Latitude is indicated on the x-axis (°N) and longitude on the y-axis (°W). All points along 140°, 137°, and
110 ° W are floating or core top sediments, as indicated by labeled arrows.

58

30

59

15

Because isotope values at the base of the food web are passed along to consumers,  N
values for yellowfin tuna (Thunnus albacores) also exhibit these spatial trends (Figure 12;
Graham et al. 2010; Olson et al. 2010). Like tuna, highly mobile seabirds have the potential to
incorporate isotope values unique to their foraging location. Newell‘s shearwaters grow their
feathers in autumn when birds are commonly found southeast of the Hawaiian Islands (highest
15

densities at 4-10° N, 130-165° W) in areas characterized by highly variable δ N values (Figure
15

12; Spear et al. 1995). The high variability in the δ N among Newell‘s shearwaters (Figure 10)
may therefore be accounted for by individual variation in foraging location. Similarly, the
15

disparity between average  N values of adult Hawaiian petrels from Hawaii and Kauai is
likely related to differences in the location where these two populations concentrate their
foraging efforts. If this is the case, breeding and molting Hawaii birds use a foraging area
15

characterized by higher δ N values than that used by molting Kauai adults. In this respect, the
two populations exhibit segregation in at least one aspect of their marine niche.
13

Although δ C values increase with trophic level within a food web from a specific
13

location, there is an inverse relationship between δ C and latitude for phytoplankton and upper
level consumers, including squid and seabirds (Goericke & Fry 1994; Kelly 2000; Takai et al.
13

2000). The low δ C values of Transition Zone albatross (from 40 to 45°N) relative to those of
Newell‘s shearwaters, adult Hawaiian petrels, and hatch-year Hawaiian petrels from Hawaii
(Figure 10) are therefore likely related to the albatross having foraged at more northerly latitudes
than the other groups. Indeed, Newell‘s shearwaters concentrate their foraging between 4 and
13

10ºN near the period of molt (Spear et al. 1995). Similarly, δ C values in hatch-year Hawaiian
petrels from Kauai are significantly lower than those found in other Hawaiian petrels, and likely
derive from food resources located farther north.
60

Our interpretations are consistent with at-sea observations of the Hawaiian petrel. During
the breeding season, individuals have been seen between the Hawaiian and Aleutian Islands
(Bourne 1965; Bourne & Dixon 1975). However, Spear et al. (1995) also observed Hawaiian
petrels to the southeast of the Hawaiian Islands during the breeding season and early nonbreeding season (Figures 8 and 10). More recently, satellite telemetry revealed that chickprovisioning adults from Maui and Lanai forage extensively throughout the northeastern Pacific,
including the Transition Zone and the Subarctic Frontal Domain near the Aleutian Islands (J.
13

15

Adams and D. Ainley et al., unpublished data). Our δ C and δ N data suggest that breeding
adults from Kauai also forage north of the Hawaiian Islands. Thus, chick-provisioning adults
from Kauai may be among the individuals observed by Bourne and Dixon between the Hawaiian
and Aleutian Islands. Our data further suggest that individuals observed southeast of Hawaii by
Spear et al. (1995) included chick-provisioning adults from Hawaii, and both Hawaii and Kauai
adults during the period of molt.
Surprisingly, Hawaiian petrels showed a very broad range of δD values (-69 to 53 ‰),
despite the uniform δD of their ultimate water source (Pacific Ocean: 0 ‰; Figure 11; Lecuyer et
al. 1997). The variation in D between Hawaiian petrel groups mirrors the separation observed
13

in  C values (Figure 10). In general, marine sediments and organisms exhibit δD values near
or below -100 ‰, reflecting fractionation between plants and their water source during
photosynthesis (Estep & Hoering 1980; Hoering 1974; Stuermer et al. 1978). The prevalence of
low δD values in marine ecosystems is further supported by our measurements of neon flying
squid, lancetfish (Alepisaurus spp.), Pacific herring (Clupea pallasii), eulachon (Thaleichthys
pacificus), and Pacific ocean perch (Sebastes alutus), which range from -93 to -130 ‰. Because
seabirds typically derive >80% of their water from prey, low δD values are also expected in
61

Hawaiian petrels (Goldstein 2001). However, this was not the case for adult Hawaiian petrels
from Kauai or any of the Hawaiian petrels from Hawaii.
Water lost through evaporatory processes has a δD value ~50 ‰ lower than body water
in captive rock pigeons and results in an enrichment in body water deuterium (McKechnie et al.
2004). While evaporatory processes may contribute to the D variation observed among
Hawaiian petrels, it is unlikely to result in the 120 ‰ range of D values in our dataset.
Furthermore, variation in evaporative water loss is most likely to impose a difference in δD
values between hatch-year birds (feathers grown while sitting in cool, high-elevation burrows)
and adults (feathers grown while flying at sea, where birds are more likely to employ evaporative
cooling). However, the difference in δD between the two age classes of Hawaiian petrels from
Hawaii was small (12 ‰). Similar to water lost through evaporation, we expect water lost
through salt glands to have low δD values relative to body water and consider it to be a more
probable source of elevated δD values in the Hawaiian petrel. Consumption of prey such as squid
and crustaceans that are isosmotic with seawater versus hyposmotic teleost fish can result in a
doubling of salt load for seabirds (Goldstein 2001). Increasing salt load, in turn, results in
increased salt gland excretion by petrels (Goldstein 2001; Warham 1996). Thus, our data may
reflect a greater reliance on fish by Kauai hatch-year birds and a heavier reliance on squid and
crustaceans by molting adults and Hawaii hatch-year birds. While Hawaiian petrels are known to
feed on both fish and squid (Simons 1985), further research is clearly needed to delineate the
causes of δD variation in pelagic seabirds. In avian research, stable hydrogen isotopes are
primarily used to study the migration of terrestrial birds. However, variation in our data
demonstrates that factors other than source water can have a large influence on δD values, at
least in some species. The clear segregation of δD values among Hawaiian petrel age and island
62

groups also suggests stable hydrogen isotopes may provide useful information about seabird
physiology and ecology.

Causes and Implications of Genetic and Isotopic Variation
The genetic and ecological variation we observed between Hawaiian petrels from Kauai and
Hawaii reveal unexpectedly complex intra-specific dynamics. Movement between islands is
clearly not limited by dispersal ability, as evidenced by the extreme length of foraging trips
(often >10,000 km; Adams & Flora 2009; J. Adams and D. Ainley et al. unpublished data).
Furthermore, populations on Hawaii and Kauai appear to breed synchronously (S. Judge,
unpublished data), and there are no clear morphological differences that separate them. Given the
many similarities between Hawaiian petrels from Kauai and Hawaii, the inter-colony variation
we observed appears to be a compelling example of cryptic ecological and genetic diversity.
Several factors may have contributed to the genetic differentiation between Hawaiian
petrel colonies. Simons (1985) found evidence of strong nest-site and mate fidelity among
Hawaiian petrels, and this philopatry likely decreases dispersal between colonies and contributes
to genetic isolation (Friesen et al. 2007). Divergence in foraging locations may have further
reduced contact between colonies and imposed distinct selective pressures. Adaptation to local
breeding environment is another possible source of divergence. In Kauai, Hawaiian petrels breed
in humid areas characterized by dense vegetation receiving over 1,200 cm of precipitation per
year. In contrast, Hawaiian petrels on Hawaii breed in open, xeric environments that are 2,000 to
3,000 m higher in elevation than breeding areas on Kauai (Ainley et al. 1997; Hu et al. 2001).
Differences in breeding habitat could result in different selective pressures between islands and
reduced fitness among immigrants. However, Hawaiian petrels may have nested in a wide range
63

of habitats on both Kauai and Hawaii in the past, before they were confined to their current range
by habitat loss, human harvesting, and introduced mammalian predators (Olson & James 1982;
Harrison 1990).
The significant isotopic differences between both adult and hatch-year birds from Hawaii
and Kauai result from their unique foraging strategies. Although reminiscent of habitat
segregation between closely related seabird species (Fasola et al. 1989; Finkelstein et al. 2006),
such differentiation is rarely observed between seabird populations (Navarro et al. 2009).
Competition-driven niche segregation, or niche partitioning, may be at work in the Hawaiian
petrel, and may have developed as a mechanism to reduce competition and allow coexistence of
geographically proximate colonies. Competition among seabirds is thought to be especially
strong during the breeding season when adults must consistently return to similar nesting
locations, and it may account for the particularly large divergence between hatch-year birds from
Kauai and Hawaii (Forero et al. 2002; Lewis et al. 2001).
Genetic and ecological variation between Hawaiian petrel populations has important
conservation implications. Procellariiform seabirds are a highly endangered group and face
numerous threats from fishing, habitat loss, and introduced predators (Bartle et al. 1993; Bell &
Merton 2002). While further studies of the Lanai and Maui populations are required, it appears
that distinct management units exist within the Hawaiian petrel (Mortiz 1994 & 2002).
Conservation measures should be considered on an island-to-island basis, as opposed to working
under the assumption that populations are homogeneous. Inter-colony diversification in
additional seabird species may be unrecognized due to the logistical difficulties of studying
cryptic and highly mobile marine organisms, particularly when they are rare. Genetic and

64

isotopic methods, such as those reported here, can be used to help overcome this barrier and
increase knowledge of the diversity and ecology of marine vertebrates.

65

LITERATURE CITED

66

Abbott, C.L. and M.C. Double. 2003. Genetic structure, conservation genetics and evidence of
speciation by range expansion in shy and white-capped albatrosses. Molecular Ecology
12:2953-2962.
Adams, J. and F. Flora. 2009. Correlating seabird movements with ocean winds: linking satellite
telemetry with ocean scatterometry. Marine Biology 157: 915-929.
Adams, N.J. and C.R. Brown. 1989. Dietary differentiation and trophic relationships in the subAntarctic penguin community at Marion Island. Marine Ecology Progress Series 57:249258.
Ainley, D.G., R. Podolsky, L. DeForest, and G. Spencer. 1997. New insights into the status of
the Hawaiian Petrel on Kauai. Colonial Waterbirds 20:24-30.
Altabet, M. and R. Francois. 1994. Sedimentary Nitrogen Isotopic Ratio as a Recorder for
Surface Ocean Nitrate Utilization. Global Biogeochemical Cycles 8:103-116.
Altabet, M.A., W.G. Deuser, S. Honjo, and C. Stienen. 1991. Seasonal and depth-related changes
in the source of sinking particles in the North Atlantic. Nature 354:136-139.
Arbogast, B.S., S.V. Edwards, J. Wakeley, P. Beerli, and J.B. Slowinksi. 2002. Estimating
divergence times from molecular data on phylogenetic and population genetic timescales.
Annual Review of Ecology and Systematics 33:707-740.
Ashmole, N.P. and M.J. Ashmole. 1967. Comparative feeding ecology of seabirds of a tropical
oceanic island. Bulletin of the Peabody Museum of Natural History 24:1-131.
Austin, J.J., V. Bretagnolle, and E. Pasquet. 2004. A global molecular phylogeny of the small
Puffinus shearwaters and implications for systematics of the Little-Audobon's shearwater
complex. Auk 121:847-864.
Bartle, J.A., Hu, D., Stahl, J.C. et al., 1993. Status and ecology of gadfly petrels in the temperate
North Pacific In: The status, ecology, and conservation of marine birds in the North
Pacific (eds. Vermeer K, Briggs KT, Morgan KH, Siegal-Causey D). Canadian Wildlife
Service, Ottawa, Canada.
Beerli, P., 2004. Effect of unsampled populations on the estimation of population sizes and
migration rates between sampled populations. Molecular Ecology 13:827-836.
Beerli, P. and J. Felsenstein. 1999. Maximum-likelihood estimation of migration rates and
effective population numbers in two populations using a colaescent approach. Genetics
152:763-773.

67

Bell, B.D., Merton, D.V., 2002. Critically endangered bird populations and their management.
In: Conserving bird biodiversity: general principles and their application (eds. Norris K,
Pain DJ). Cambridge University Press, Cambridge.
Bocher, P., Y. Cherel, and K.A. Hobson. 2000. Complete trophic segregation between South
Georgian and common diving petrels during breeding at Iles Kerguelen. Marine Ecology
Progress Series 208:249-264.
Bourne, W.P.R. 1965. The missing petrels. Bulletin of the British Ornithologists' Club 85:97105.
Bourne, W.R.P. and T.J. Dixon. 1975. Observations of seabirds 1970-1972. Sea Swallow 24:6588.
Bretagnolle, V., C. Attié, and E. Pasquet. 1998. Cytochrome-B Evidence for validity and
phylogenetic relationships of Pseudobulweria and Bulweria (Procellariidae). Auk
115:188-195.
Browne, R.A., D.J. Anderson, and J.N. Houser, et al. 1997. Genetic diversity and divergence of
endangered Galapagos and Hawaiian Petrel populations. Condor 99, 812-815.
Clement, M., Posada, D., Crandall, K. 2000. TCS: a computer program to estimate gene
genealogies. Molecular Ecology 9:1657-1660.
Craig, H. 1961. Standard for Reporting Concentrations of Deuterium and Oxygen-18 in Natural
Waters. Science 133:1833-1834.
Drummond, A.J. and A. Rambaut. 2007. BEAST: Bayesian evolutionary analysis by sampling
trees. Bmc Evolutionary Biology 7:214.
Estep, M.F. and T.C. Hoering. 1980. Biogeochemistry of the stable hydrogen isotopes.
Geochimica et Cosmochimica Acta 44:1197-1206.
Excoffier, L., G. Laval, and S. Schneider. 2005. Arlequin ver. 3.0: An integrated software
package for population genetics data analysis. Evolutionary Bioinformatics Online 1:4750.
Farrell, J.W., T.F. Pederson, S.E. Calvert, and B. Nielsen. 1995. Glacial-interglacial changes in
nutrient utilization in the equatorial Pacific Ocean. Nature 377:514-516.
Fasola, M., G. Bogliani, N. Saino, and L. Canova. 1989. Foraging, feeding, and time-activity
niches of eight species of breeding seabirds in the coastal wetlands of the Adriatic Sea.
Italian Journal of Zoology 56:61-72.
Finkelstein, M., B.S. Keitt, and D.A. Croll, et al. 2006. Albatross species demonstrate regional
differences in north Pacific marine contamination. Ecological Applications 16:678-686.
68

Fisher, H.I. 1967. Body weights in Laysan Albatrosses Diomedea immutabilis. Ibis 109:373-382.
Fisher, H.I. 1976. Some dynamics of a breeding colony of Laysan Albatrosses. Wilson Bulletin
88:121-142.
Fleischer, R.C., S.L. Olson, S.L., H.F. James, and A.C. Cooper. 2000. Identification of the
extinct Hawaiian Eagle (Haliaeetus) by mtDNA sequence analysis. Auk 117:1051-1056.
Forero, M.G., J.L. Tella, K.A. Hobson, M. Bertellotti, and G. Blanco. 2002. Conspecific food
competition explains variability in colony size: a test in Magellanic Penguins. Ecology
83:3466-3475.
Friesen, V.L., J.A. González, and F. Cruz-Delgado. 2006. Population genetic structure and
conservation of the Galapagos petrel (Pterodroma phaeopygia). Conservation Genetics
7:105-115.
Frings, H. and M. Frings. 1961. Some biometric studies on the albatrosses of Midway Atoll.
Condor 63:304-312.
Goericke, R. and B. Fry. 1994. Variations of Marine Plankton Delta-C-13 with Latitude,
Temperature, and Dissolved Co2 in the World Ocean. Global Biogeochemical Cycles
8:85-90.
Goldstein, D.L. 2001. Water and Salt Balance in Seabirds. In: Biology of Marine Birds (eds.
Schreiber EA, Burger J), pp. 467-480. CRC Press.
Gould, P., P. Ostrom, and W. Walker. 1997. Trophic relationships of albatrosses associated with
squid and large-mesh drift-net fisheries in the North Pacific Ocean. Canadian Journal of
Zoology-Revue Canadienne De Zoologie 75:549-562.
Graham, B.S., P.L. Koch, S.D. Newsome, K.W. McMahon, D. Aurioles. 2010. Using Isoscapes
to Trace the Movements and Foraging Behavior of Top Predators in Oceanic Ecosystems.
In: Isoscapes: Understanding Movement, Pattern, and Process on Earth Through Isotope
Mapping (ed. West JB). Springer Science.
Harris, M.P. 1970. Differences in the diet of British auks. Ibis 112:540-541.
Hobson, K.A. and R.G. Clark. 1992. Assessing Avian Diets Using Stable Isotopes .1. Turnover
of C-13 in Tissues. Condor 94:181-188.
Hoering, T.C. 1974. The isotopic composition of the carbon and hydrogen in organic matter of
Recent sediments. In: Annual Report of the Director, 1973-1974, pp. 590-595. Carnegie
Institution, Washington, D.C.

69

Holder, M. and P.O. Lewis. 2003. Phylogeny estimation: traditional and Bayesian approaches.
Nature Reviews Genetics 4:275-284.
Hu, D., C. Glidden and J.S. Lippert, et al. 2001. Habitat use and limiting factors in a population
of Hawaiian dark-rumped petrels on Mauna Loa, Hawai'i. Studies in Avian Biology
22:234-242.
Jehl, J., J. R. 1982. The biology and taxonomy of Townsend's Shearwater. Le Gerfaut 72:121135.
Kelly, J.F. 2000. Stable isotopes of carbon and nitrogen in the study of avian and mammalian
trophic ecology. Canadian Journal of Zoology-Revue Canadienne De Zoologie 78:1-27.
Kuhner, M.K. 2009. Coalescent genealogy samplers: windows into population history. Trends in
Ecology & Evolution 24:86-93.
Lecuyer, C., P. Gillet, and F. Robert. 1997. The hydrogen isotope composition of seawater and
the global water cycle. Chemical Geology 145:249-261.
Lewis, S., T.N. Sherratt, K.C. Hamer, and S. Wanless. 2001. Evidence of intra-specific
competition for food in a pelagic seabird. Nature 412:816-819.
15

Liu, K., and I.R. Kaplan. 1989. The Eastern Tropical Pacific as a source of  N-enriched nitrate
in seawater off southern California.‖ Limnology and Oceanography 34 (5): 820-830.
McKechnie, A.E., B.O. Wolf, and C.Md. Rio. 2004. Deuterium stable isotope ratios as tracers of
water resource use: an experimental test with rock doves. Oecologia 140:191-200.
Michener, R.H. and D.M. Schell. 1994. Stable isotope ratios as tracers in marine aquatic food
webs. In: Stable isotopes in ecology and environmental science (eds. Lajtha K, Michener
R), pp. 138-157. Blackwell Scientific.
Milot, E., H. Weimerskirch, and L. Bernatchez. 2008. The seabird paradox: dispersal, genetic
structure and population dynamics in a highly mobile, but philopatric albatross species.
Molecular Ecology 17:1658-1673.
Navarro, J., M. Forero, and J. Gonzalez-Solis, et al. 2009. Foraging segregation between two
closely related shearwaters breeding in sympatry. Biology Letters 5:54-548.
Nunn, G.B. and S.E. Stanley. 1998. Body size effects and rates of cytochrome b evolution in
tube-nosed seabirds. Molecular Biology and Evolution 15:1360-1371.
Olson, S.L. and H.F. James. 1982. Prodromus of the fossil avifauna of the Hawaiian islands.
Smithsonian Contributions to Zoology 365:1-59.

70

Phillips, R.A., J.R.D. Silk, B. Phalan, P. Catry, and J.P. Croxall. 2004. Seasonal sexual
segregation in two Thalassarche albatross species: competitive exclusion, reproductive
role specialization or foraging niche divergence? Proceedings of the Royal Society B
271:1283-1291.
Posada, D. 2008. jModelTest: Phylogenetic model averaging. Molecular Biology and Evolution
25:1253-1256.
Posada, D. and T.R. Buckley. 2004. Model selection and model averaging in phylogenetics:
Advantages of akaike information criterion and Bayesian approaches over likelihood
ratio tests. Systematic Biology 53:793-808.
Posada, D. and K.A. Crandall. 2001. Intraspecific gene genealogies: trees grafting into networks.
Trends in Ecology & Evolution 16:37-45.
Pyle, P. 2008. Molt and age determination in Procellariiformes. In: Identification Guide to North
American Birds, Part 2. Slate Creek Press, Point Reyes Station, CA, USA.
Rozen, S., and H.J. Skaletsky. 2000. Primer3 on the WWW for general users and biologist
programmers. In: Bioinformatics Methods and Protocols: Methods in Molecular Biology
(eds. Krawetz S, Misener S), pp. 365-386. Humana Press, Totowa, NJ.
Simons, T.R. 1984. A population model of the endangered Hawaiian dark-rumped petrel.
Journal of Wildlife Management 48:1065-1076.
Simons, T.R. 1985. Biology and Behavior of the Endangered Hawaiian Dark-Rumped Petrel.
Condor 87:229-245.
Simons, T.R. and C.N. Hodges, American Ornithologists' Union., Academy of Natural Sciences
of Philadelphia. (1998) Dark-rumped petrel : Pterodroma phaeopygia. In: The Birds of
North America ; no. 345, p. 24 p. American Ornithologists' Union ; The Academy of
Natural Sciences, [Washington, D.C.] Philadelphia, PA.
Sorenson, M.D. and T.W. Quinn. 1998. Numts: A challenge for avian systematics and population
biology. Auk 115:214-221.
Spear, L., D. Ainley, and W. Walker. 2007. Foraging dynamics of seabirds in the eastern tropical
pacific ocean. Studies in Avian Biology 35:1-99.
Spear, L.B., D.G. Ainley, N. Nur, and S.N.G. Howell. 1995. Population-Size and Factors
Affecting at-Sea Distributions of 4 Endangered Procellariids in the Tropical Pacific.
Condor 97:613-638.
Stuermer, D.H., K.E. Peters, and I.R. Kaplan. 1978. Source Indicators of Humic Substances and
Proto-Kerogen - Stable Isotope Ratios, Elemental Compositions and Electron-Spin
Resonance-Spectra. Geochimica et Cosmochimica Acta 42:989-997.
71

Takai, N., S. Onaka, and Y. Ikeda, et al. 2000. Geographical variations in carbon and nitrogen
stable isotope ratios in squid. Journal of Marine Biological Association of the United
Kingdom 80:675-684.
Tamura, K., J. Dudley, M. Nei, and S. Kumar. 2007. MEGA4: Molecular Evolutionary Genetics
Analysis (MEGA) software version 4.0. Molecular Biology and Evolution 24:1596-1599.
Techow, N.M.S.M., P.G. Ryan, and C. O'Ryan. 2009. Phylogeography and taxonomy of Whitechinned and Spectacled Petrels. Molecular Phylogenetics and Evolution 52:25-33.
Tieszen, L.L., T.W. Boutton, K.G. Tesdahl, and N.A. Slade. 1983. Fractionation and Turnover of
Stable Carbon Isotopes in Animal-Tissues - Implications for Delta-C-13 Analysis of Diet.
Oecologia 57:32-37.
Warham, J. 1990. The Petrels: Their Ecology and Breeding Systems Academic Press, New York.
Warham, J. 1996. The Behaviour, Population Biology, and Physiology of the Petrels Academic
Press, Christchurch, New Zealand.
Wassenaar, L.I. and K.A. Hobson. 2003. Comparative equilibration and online technique for
determination of non-exchangeable hydrogen of keratins for use in animal migration
studies Isotopes in Environmental Health Studies 39:211-217.
Weir, J.T. and D. Schluter. 2008. Calibrating the avian molecular clock. Molecular Ecology
17:2321-2328.
Welch, A.J., and A.A. Yoshida, and R.C. Fleischer. (Submitted) Mitochondrial and nuclear DNA
sequences reveal recent divergence in morphologically indistiguishable petrels.
Wiley, A.E., P.H. Ostrom, C.A. Stricker, H.F. James, and H. Gandhi. 2010. Isotopic
characterization of flight feathers in two pelagic seabirds: Sampling strategies for
ecological studies. Condor 112:337-346.
Wong, W.W., L.L. Clarke, G.A. Johnson, M. Llaurador, and P.D. Klein. 1992. Comparison of
two elemental-analyzer gas-isotope-ratio mass spectrometer systems in the simultaneous
measurement of carbon-13/carbon-12 ratios and carbon content in organic samples.
Analytical Chemistry 64 (4): 354-358.
Zavalaga, C.B., S. Benvenuti, L. Dall'Antonia, and S.D. Emslie. 2007. Diving behavior of bluefooted boobies Sula nebouxii in northern Peru in relation to sex, body size and prey type.
Marine Ecology Progress Series 336:291-303.

72

CHAPTER 3
TWO THOUSAND YEARS OF FORAGING ECOLOGY IN THE ENDANGERED
HAWAIIAN PETREL: INSIGHTS FROM STABLE ISOTOPE ANALYSIS
ABSTRACT
As a consequence of their extreme isolation in the North Pacific Ocean, the Hawaiian Islands
provide a unique setting in which to study anthropogenic impacts to biodiversity. Here, we
present a biomolecular record for a Hawaiian seabird, the Hawaiian petrel (Pterodroma
sandwichensis), in order to study shifting ecology within the Pacific Ocean. Chronologies of
15

13

 N and  C are used for a retrospective view of petrel trophic level and foraging location,
extending back approximately 1,000 to 4,000 years for four colonies. The most pervasive
15

temporal trend we observe is a 1.4-2.6 ‰ decrease in average  N values during the past 1,000
years. This species-wide shift likely reflects declining trophic level associated with modern
13

15

fishing practices. It stands in sharp contrast to temporally consistent  C values, as well as  N
values prior to 1,000 years BP. Our chronologies also document 2,000 years of foraging
segregation between Hawaiian petrel colonies, observed as a long-standing divergence in
15

average  N values. Differences in foraging location appear to separate modern Hawaiian petrel
colonies and likely explain at least a portion of the isotopic segregation between ancient
colonies. The degree of foraging segregation between petrel colonies diminishes through time
and correlates well with genetic population structure. We suspect that shifting foraging habits of
the Hawaiian petrel reflect relatively widespread trophic alterations in pelagic food webs of the
North Pacific. Such changes in foraging are concerning, given their implications for reproductive
success and genetic diversity in the Hawaiian petrel, and given the potential for similar shifts in
other seabird species.

73

INTRODUCTION
Nearly four thousand kilometers from any continental shore, the Hawaiian Islands were among
the last landmasses on Earth to be colonized by humans (ca. 800-1000 years ago; Kirch 1990;
Wilmshurst et al. 2011). As a consequence, they provide a unique setting in which to study
anthropogenic impacts on island biodiversity. The Hawaiian subfossil record affords a startling
perspective, showing the extinction of approximately two thirds of endemic vertebrates
following human arrival. Among birds, most remaining species are now in danger of extinction,
and many occupy only a fraction of their pre-historic ranges (Olson & James 1982). Recently, a
biomolecular record from Hawaiian subfossils has begun to elucidate the origins and fates of
terrestrial species. This record exposed previously unknown adaptive radiations, as well as
striking population bottlenecks (Paxinos, James, et al. 2002a; Paxinos, James, et al. 2002b).
However, Hawaii‘s rich fossil record has yet to be exploited to understand ecological shifts
taking place in the marine realm. Here we explore the history of human impacts on the Pacific
Ocean as evidenced in the stable isotopic chronology of Hawaii‘s most common subfossil
seabird, the Hawaiian petrel (Pterodroma sandwichensis, or HAPE).
HAPE bones predominate many archaeological and paleontological collections
throughout the main Hawaiian Islands, indicating large and widespread prehistoric populations
(Olson & James 1982b). Despite the species‘ decline to its current endangered status, genetically
divergent populations persist on four islands: Hawaii, Maui, Lanai, and Kauai (Figure 13; Welch
In Prep.). While HAPE nesting is confined to a single archipelago, the species travels great
distances at sea, from the equator to waters near the Aleutian Islands (Spear et al. 1995; Adams
& Flora 2009). Indeed, breeding adults can travel in excess of 10,000 km in a period of two
74

weeks, as shown by recent satellite tracking results (Adams & Flora 2009). HAPE prey items
include a wide variety of fish, squid, and crustaceans from pelagic waters (Simons 1985; Spear et
al. 1995). HAPE foraging ecology can be further elucidated using the stable isotopic content of
15

tissues such as feather and bone collagen. Specifically,  N values increase ca. 3 ‰ with each
15

13

trophic level, and can thus be used to study an aspect of diet. In addition, both  N and  C
values vary spatially within the foraging range of HAPE, providing a measure of foraging
location (Wiley et al. In Press).
15

13

In the present study, we use  N and  C values of 0-4,000 year old bone collagen to
develop temporal records of HAPE foraging habits (here defined as trophic level and foraging
location). Our chronology extends back to a period before human influence in the Hawaiian
Islands: roughly 3,000 years prior to colonization. In addition, our record likely predates human
impact on HAPE foraging grounds in the pelagic N Pacific (Torenson & Rick 2008). Our study
is therefore able to juxtapose current HAPE foraging habits against a pre-human baseline. This
view is particularly useful for gauging anthropogenic impact to the marine realm. Human
harvesting of marine organisms has precipitated major shifts in the abundance of targeted species
(Myers & Worm 2003). Food web restructuring is a predicted result, but such trends have proven
difficult to observe, particularly for the vast pelagic realm. As apex predators feeding on diverse
and relatively widespread prey, HAPE have the ability to echo large-scale shifts in pelagic food
webs. We therefore use this species as a sentinel of change for the North Pacific Ocean and as an
indicator for the potentially wide-reaching effects of human fishing practices.
In addition to a record of pelagic food webs, HAPE bones provide the unique opportunity
to study inter-population dynamics over the course of hundreds to thousands of years. Our
previous isotopic work on modern HAPE documented divergent foraging habits in the most
75

geographically distinct colonies on the islands of Hawaii and Kauai. Here, we expand our
investigation of foraging segregation to colonies on all four islands known to comprise the
13

15

HAPE nesting distribution. The  C and  N composition of modern flight feathers, which
reflects foraging during the non-breeding and breeding seasons for adult and hatch-year birds,
respectively, forms the basis of our inquiry. Because subfossil remains are available from
multiple colonies, we further study the persistence of foraging segregation through time using the
isotopic composition of bone collagen. The long-term perspective afforded by our subfossil
chronology provides a rare glimpse of potential anthropogenic alterations to within-species
ecological diversity at the scale of centuries.

Figure 13. Location of sub-fossil collection sites (circles) and current breeding distribution (all
other shapes, outlined in black) for HAPE.

76

MATERIALS AND METHODS
Sample Acquisition and Feather Growth
We collected 68 primary 1 (P1) feathers and 55 bones from HAPE carcasses recovered between
1989 and 2009. Carcasses were salvaged from birds killed by introduced mammals or killed after
collision with man-made objects (e.g. wind mills). We also sampled two P1 feathers from birds
prepared as museum study skins in 1980 and 1995. 135 sub-fossil bones were collected from
sites across four of the Hawaiian Islands (Figure 13). In hatch-year HAPE (<1 year in age), P1
and other flight feathers are grown during the late growth stages, from September to December
(Simons 1985; Ainley et al. 1997). As in other Pterodroma, adult HAPE are presumed to begin
primary molt, beginning with P1, following cessation of nest attendance (November to
Decmeber for breeders; Simons 1985; Warham 1996).
Stable Isotope and AMS Radiocarbon Methods
Prior to stable isotope analysis, feathers were washed in solvent (87:13 chloroform:methanol by
volume), rinsed with ultrapure distilled water (E-Pure, Barnstead), and dried at 25°C in a vacuum
oven. Stable isotope data were obtained from samples representative of the entire feather vanes
(Wiley et al. 2010).
Collagen was isolated and purified using a method modified from (Stafford et al. 1988).
Bones were decalcified with distilled hydrochloric acid (0.2 to 0.5 M) and soaked in 0.05 M
potassium hydroxide overnight to remove humate contaminants. The resulting collagen was
gelatinized with 0.05 M hydrochloric acid (110C, 1-3 hrs), passed through a 0.45 μm Millipore
HV filter and lyophilized. One aliquot of gelatinized collagen was used for stable isotope
analysis. For ancient samples, a second aliquot was hydrolyzed in hydrochloric acid (6 M, 22
hrs) and passed through a column containing XAD-2 resin to remove fulvic acids. The resulting

77

hydrolysate was combusted to CO2 and graphitized for AMS dating (W. M. Keck Carbon Cycle
AMS laboratory, University of CA, Irvine). Background contamination from

14

C-depleted and

14

C-enriched carbon during the preparation of each sample set was evaluated by dating

hydrolyzed gelatin of known age:

14

C-dead whale (ca. 70,000 years BP; Stafford et al. 1991) and

young Bison bison (mean pooled radiocarbon age: 1794  5.8 yrs BP, n=9).
For ten ancient samples, collagen was extracted at the Keck facility using techniques
modified from Longin (1971), followed by ultrafiltration (Brown et al. 1988). We demonstrated
the comparability of dates obtained using XAD-2 purification versus Longin-ultrafiltration
methods. First, we compared dates obtained from the Bison bison sample (median probabilites of
1740-1820 yrs BP (n=10) by XAD purification; 1750-1785 yrs BP (n=2) for Longinultrafiltration). Second, we compared dates from HAPE bones found in Fireplough Cave,
Hawaii, which were expected to yield a small range of ages based on the particular
archaeological context in which they were discovered (median probabilities of 459-525 yrs BP
(n=4) for XAD purification; 473-482 yrs BP (n=3) for Longin-ultrafiltration; James et al. In
Prep.). In both cases, dates for the Longin-ultrafiltration methods fell with in the range of those
prepared using XAD purification.
We calibrated our conventional radiocarbon ages using the program CALIB 6.0 and
applied a marine reservoir correction to account for incorporation of

14

C-depleted marine carbon

(Gordon & Harkness 1992). Specifically, we included a global model of the marine reservoir
effect (Marine09 model; Reimer et al. 2009), along with a regional correction, or ΔR, of 54 ± 20
years, calculated specifically for HAPE (James et al. In Prep.). All radiocarbon dates referred to
in the text are median probabilities. Similarly, median probability dates were used for all
graphing and statistical analysis.
78

13

15

δ C and δ N values of gelatinized collagen (ca. 1.0 mg) were determined using an
elemental analyzer (Eurovector) interfaced to an Isoprime mass spectrometer (Elementar; Wong
et al. 1992). Stable isotope values are expressed in per mil (‰) as: δX = ([Rsample/Rstandard] – 1)
× 1000, where X is

13

C or

13

15

N, R is the corresponding ratio

13

12

C/ C or

15

15

14

N/ N, and Rstandard is
13

15

V-PDB and air for δ C and δ N respectively. Precision was ≤ 0.2 ‰ for both δ C and δ N.
13

We corrected for the Suess Effect using an ice-core based estimate of the rate of  C
decrease in the atmosphere: 0.22 ‰ per decade since 1960, and 0.05 ‰ per decade between
1860 and 1960 (Francey et al. 1999; Chamberlain et al. 2005).

Temporal and Statistical Analysis
Isotope data for gelatinized collagen was binned based on periods of human history in the
Hawaiian Islands: Modern Period (1950 CE-2011), Historic Period (1800-1950 CE; 0-150 yrs
BP), Late Expansion Period (1400-1800 CE; 150-550 yrs BP), Foundation and Early Expansion
Period (1100-1400 CE; 550-850 yrs BP), and Pre-Human Period (<1100 CE; 850 yrs BP)(Kirch
1990; Wilmshurst et al. 2011). We subdivided the Pre-human time bin for the island of Maui in
half along a natural gap in the data of >850 years, due to the exceptionally long period of ca.
3500 years covered by these samples. We combined all ancient samples (>100 years old) from
the island of Lanai into one time bin, due to their relatively narrow range of dates (899-1088 yrs
BP) and our small sample size (n=5).
The effects of island (breeding location) and time on isotope values were evaluated
15

through multiple analysis of variance (ANOVA) models. For  N only (where both island and
time had significant effects), Tukey HSD post hoc tests were used to make all possible pair-wise
comparisons between island-time bin groups. ANOVA and Tukey HSD tests were similarly used
79

to evaluate isotopic variation among modern feathers. Normal quantile-quantile plots and
Levene‘s tests were used to check assumptions of normality and homogeneity of variance. All
statistical tests were conducted using R statistical software (version 2.12.1, R Foundation for
Statistical Computing, 2010).

RESULTS
Bone Collagen
15

We observed a significant decrease in  N through time for HAPE on the islands of Hawaii,
Maui, and Lanai (Figure 14; see Table 2 for statistical results of all paired comparisons and for
15

calibrated radiocarbon ages within each time bin). For all three islands, the majority of the  N
decline (1.4-2.6 ‰) was observed between the two most recent data points, with no significant
15

changes in  N occurring beforehand. Our most complete chronology derived from the island
15

of Hawaii, where  N decline consisted of a significant decrease of 1.9 ‰ between the Late
Expansion and Modern periods (p<0.01, Tukey HSD) and a non-significant decrease of 0.8 ‰
between the Pre-human and Late Expansion periods (p=0.311). HAPE from the island of Oahu
15

showed no change in  N through time (p=1.000 for the Pre-human vs. Foundation & Early
Expansion periods), up until their apparent extirpation around 600 BP (youngest radiocarbon
15

date = 615 BP). HAPE from the island of Maui maintained lower average  N values than birds
from Hawaii for the duration of our chronologies. However, the disparity between islands
diminished from 1.5 ‰ in the Pre-human period (average difference between Hawaii and both
Pre-human periods for Maui) to 0.5 ‰ in the Modern period. No significant relationship was
13

13

found between  C and time, or between  C and island (F≤0.725 and p≥0.541 in ANOVA
model).

80

Table 2. Two σ range for radiocarbon dates in each time bin and results of Tukey HSD
15
comparisons for  N. *Letters denote significant differences between groups in post hoc
analyses: values not sharing the same letter are significantly different (alpha = 0.05).
15

Island

Time Bin

2σ Range (Yrs BP)

 N Comparisons*

Hawaii
Hawaii

Modern
Late Expansion I

NA
48-562

A, B, F
C,E,F,G,H

Hawaii

Pre-human

1390-3342

G,H

Maui

Modern

NA

A, B, F

Maui

Foundation & Early Expansion

497-1007

A,B,D,E,F,H

Maui

Pre-human I

875-2316

H,F

Maui

Pre-human II

2947-4513

H

Oahu

Foundation & Early Expansion

534-934

B,D,E,F,H

Oahu

Pre-human

949-2730

H,F

Lanai

Modern
Pre-human/Foundation & Early
Expansion

NA

A

781-1189

E,F,G,H

Lanai

Feathers
15

13

Among adults, we observed significant variation in  N, but not  C (Figure 15). On average,
15

adult HAPE from the island of Kauai had significantly lower  N values than adults from Lanai
15

or Hawaii (difference of 2.8-3.0 ‰, p<0.0001, Tukey HSD). Maui adults had intermediate  N
values, 1.5 and 1.8 ‰ lower than Hawaii and Lanai adults, respectively (p=0.071 and 0.009) and
1.2 ‰ higher than those of Kauai adults (1.2 ‰; p=0.247). Hatch-year HAPE from Hawaii were
nearly identical in isotopic composition to adults from the same colony (average difference of
15

13

0.0 ‰ in  N and 0.2 ‰ in  C). In contrast, hatch-year birds from Maui and Kauai were
13

significantly lower in average  C than any other group of HAPE (difference ≥ 1.2 ‰; p≤

81

15

0.001). Maui and Kauai hatch-years were also significantly lower in average  N than any other
group of petrels, with the exception of Kauai adults (difference ≥ 2.2 ‰; p≤0.006 for all groups
except Kauai adults).

DISCUSSION
Our data show considerable temporal and spatial variation in HAPE foraging ecology over the
13

past 2,000 to 3,000 years. Consistent with some other studies, we find no evidence that  C or
15

 N values of the HAPE are correlated with climatic variation (Farrell et al. 1995; Norris et al.
13

2007). There is no temporal variation in  C values or influence of El Niño Southern
15

Oscillation on  N values (Wiley et al. In Press). Furthermore, there is no difference in average
isotope values for HAPE on Maui or Oahu between 750 and 1550 years BP (Figure 14): a period
that encompassed considerable climatic variation around the Pacific basin (e.g. Medieval
Climate Anomaly vs. cooling at ca. 1500 BP; Grove 2004; Nunn 2007; Gregory Wiles, personal
communication). These observations suggest that large-scale climate variation has a negligible
influence our isotopic record. We therefore consider alternative explanations for our data.
15

Temporally, the most pervasive trend we observed was a decline in  N values. This
shift is apparent during the past 1,000 years for HAPE colonies on the islands of Maui and Lanai.
However, our chronology from Hawaii suggests that the majority of the decline occurred after
15

300 yrs BP (Figure 14). One interpretation for  N decline is that HAPE have shifted their
foraging location. Recent studies argue that in the Eastern tropical North Pacific (ETNP), spatial
15

variability in  N at the base of food webs produces corresponding isotopic deviations in tuna
(Thunnus spp.) and HAPE (Graham et al. 2010; Wiley et al. In Press). If such spatial variation
15

were to account for declining  N in our data, colonies from multiple islands must have
82

15

dispersed away from a region of the ENTP where relatively high  N values permeate local
food webs (ca. 4-10N,130-140W; Altabet & Francois 1994; Graham et al. 2010). However,
13

there is no evidence for a latitudinal shift in HAPE foraging location.  C values vary inversely
with latitude throughout Pacific food webs (Goericke & Fry 1994; Kelly 2000; Takai 2000), but
13

HAPE  C does not change through time. While a predominantly longitudinal shift in foraging
15

location could account for declining  N, a consideration of the timing and impacts of human
fisheries suggests that trophic decline is a more parsimonious explanation for our data.
Prior to European contact with North America (during the Late Expansion Period),
fishing in the Pacific Ocean was conducted almost exclusively near shore and at relatively small
scales (Rick & Erlandson 2008). In contrast, large-scale and industrialized fishing during the
th

20

century is associated with exponential declines of major fish stocks (Myers & Worm 2003;

Worm 2006). In the last 50 years, alone, nearly one third of open sea fisheries have declined to
less than 10 % of their original yields (Worm et al. 2003). Declining trophic level (i.e. vertical
position in a food chain) is a hallmark of worldwide fishery harvest and is evident in multiple
marine ecosystems and species, including haddock from Georges Bank and coastal seabirds such
as the Marbled murrelet and Great black-backed gull (Becker & Beissinger 2006; Farmer &
Leonard 2011). These multiple lines of evidence suggest that the temporal decline in HAPE
15

 N values reflects a transition to lower trophic prey. Specifically, the observed isotopic shifts
are consistent with a decline of 1/2 to 4/5 of a trophic level in Maui and Lanai colonies over the
past 1,000 years, and a decline of approximately 2/3 of a trophic level for the island of Hawaii
over the past 300 years.
Changes in HAPE foraging habits have broad implications. Given that reproduction can
suffer when seabirds shift to nutritionally inferior, lower trophic prey, trophic decline may pose a
83

threat to the conservation of endangered HAPE (Forero et al. 2002; Norris et al. 2007). A shift in
the diet of Hawaiian petrels also implies a shift in the trophic structure of pelagic food webs. As
conspicuous apex predators, seabirds integrate multiple trophic levels and can thus echo changes
extending throughout their food webs (Piatt et al. 2007). Since the foraging habitat of HAPE
includes a large portion of the NE Pacific Ocean, extending from the equator to the Aleutian
Islands (Spear et al. 1995), declining trophic level in HAPE likely indicates shifting trophic
structure at a broad spatial-scale. Widespread shifts in trophic structure are generally very
difficult to document for the pelagic realm. Catch statistics can be used to study exploited pelagic
species, but most of these records date back less than 100 years and reflect efficiency of catch in
addition to shifting ecology (Watson & Pauly 2001; Myers & Worm 2003). Outside of catch
statistics, long-term isotopic records can be used to investigate food web alterations. However,
15

historical studies of  N have focused largely on species from near-shore regions. Near-shore
dynamics may be irreflective of the pelagic realm, given the relatively high accessibility of
coastal regions to human influence, their longer history of fishing exploitation, and their distinct
food web structure (Steele 1998). The few isotopic records available from pelagic seabirds likely
reflect a declining reliance of Northern fulmars on Atlantic whaling offal over the past 100 years
(Thompson et al. 1995), or else the increasing availability of krill to penguins in Antarctic and
sub-Antarctic waters (Hilton et al. 2006; Emslie & Patterson 2007). We are the first to document
15

declining  N in the pelagic realm of the North Pacific.
Prior to the Late Expansion Period, there was relatively little temporal change in HAPE
isotope values (Figure 14). The island of Hawaii, however, shows a notable albeit statistically
15

insignificant  N decline of 0.8 ‰ between the Pre-human and the Late Expansion periods.
15

This shift comprises nearly one third of the total  N decline observed on Hawaii. It is unlikely
84

to result from the impacts of human fisheries, which were restricted to near-shore environments
during the Pre-human and Late Expansion periods (Rick & Erlandson 2008). In contrast, humans
had made extensive impacts to the Hawaiian Islands by the late Expansion Period. Widespread
burning and habitat change at low elevations led, in part, to the extinction of dozens of bird
species, and the extirpation of at least one HAPE colony at Barber‘s Point, Oahu (Olson & James
1982b). The loss of the Oahu population, the largest known subfossil HAPE colony, and the
potential loss of other colonies during a period of rapid ecosystem change may have had
considerable influence on remaining HAPE. Growing evidence suggests that competition for
food within seabird species can be intense, and that this competition can shape foraging
behaviors (Furness & Birkhead 1984; Lewis et al. 2001). Given their similar isotopic
compositions, HAPE from Oahu likely relied on prey from a similar trophic level and location as
ancient HAPE from Maui. In contrast, ancient HAPE from the island of Hawaii differed in
foraging habits from those on Oahu. Once the Oahu colony vanished, ca. 600 yrs BP,
15

competition would have declined for birds exploiting prey or locations conferring lower  N
values, similar to those exploited by Oahu birds. A shift in foraging habits would therefore have
been beneficial for HAPE from Hawaii, but not for HAPE from Maui. Indeed, this is the pattern
of temporal change suggested by our data.
In addition to temporal shifts in feeding habits, our chronologies show persistent isotopic
disparities among HAPE colonies. Most strikingly, our data suggest a 2,000 year-long foraging
segregation between HAPE from Maui and Hawaii (Figure 14). To explore the causes of
foraging segregation among colonies, we use previously published interpretations of isotope data
from modern HAPE feathers (Wiley et al. In Press). This approach provides a higher resolution
analysis than can be conducted using bone collagen. Whereas the isotopic composition of
85

13

15

collagen may reflect foraging over a period of years (Rucklidge et al. 1992),  C and  N
values of flight feathers elucidate foraging habits during discrete periods within the annual cycle
HAPE. Specifically, adult feathers record foraging habits during the winter non-breeding season
and hatch-year feathers reflect the fall breeding season.
Previously, isotopic variation among HAPE feathers was interpreted as segregation in
foraging location. During both the breeding and non-breeding seasons, HAPE from Hawaii
15

appeared to spend considerable time feeding in waters of the ENTP characterized by high  N
13

values (e.g. SE of the Hawaiian Islands 4-10N,130-140W). Because  C values show an
inverse trend with latitude, the similarity in 13C values of Hawaii and Kauai adults suggest that
these individuals grew their feathers at the same latitude. However, relative to Hawaii birds,
Kauai individuals apparently spent less time in regions characterized by high 15N. Kauai HAPE
also differed from Hawaii birds in that Kauai hatch-years were likely fed prey from relatively
high latitudes, where both 13C and 15N values are low throughout local food webs. Our current
data suggest that Lanai and Hawaii adults have similar foraging habits, while the foraging habits
of Maui adults are intermediate to those of Hawaii and Kauai (Figure 15). Finally, Maui hatchyear diet appears to derive from relatively high latitudes, similar to Kauai hatch-year diet.

86

15

Figure 14. Bone collagen  N through time in four HAPE colonies. Data points reflect average age and isotopic composition of each
time bin, plus or minus standard error. Sample sizes are indicated next to each data point. Lines connect data points within each
colony, but are intended for visualization purposes, only. Isotopic shifts between time bins may have occurred non-linearly.

87

Figure 15. Foraging segregation among modern HAPE colonies, as observed in stable isotope
values. Data points reflect the mean plus or minus standard error for each group. Sample sizes
are as follows: Hawaii adults (n=14), Hawaii hatch-years (n=10), Kauai adults (n=13), Kauai
hatch years (n=12), Maui adults (n=13), Maui hatch years (n=9), and Lanai adults (n=18).

Owing to its predominant role in differentiating modern HAPE colonies, foraging
location is likely an important factor contributing to the 2,000 year-long distinction observed
between Maui and Hawaii and to the segregation between HAPE from the lost colony of Oahu
15

and ancient HAPE from Hawaii. If the extirpation of Oahu birds resulted in the coincident  N
decline in Hawaii, as we posit, Hawaii birds were most likely to have changed their foraging
range between ca. 2,100 and 300 years BP. Such a shift may have begun to obscure the isotopic
distinction between Hawaii and Maui individuals. Later in time, as fisheries likely forced all
HAPE colonies to a lower trophic level, the distinction between Hawaii and Maui colonies grew

88

noticeably smaller (Figure 14). This waning distinction may signal convergence on a single, low
trophic level, as human fishing diminished the diversity of prey available to seabirds.
When our spatial and temporal data are taken together, they suggest that HAPE foraging
habits have changed over the course of hundreds, if not thousands of years, and that a divergence
in foraging habits has developed between colonies from different islands. Spatial foraging
segregation, in particular, may have important evolutionary implications. According to Friesen et
al. (2007), variation in foraging location during the non-breeding season may help drive
population divergence by limiting contact between seabird colonies. While genetic divergence of
HAPE populations is undoubtedly influenced by multiple factors, our data support Friesen‘s
hypothesis. Foraging segregation among modern colonies during the non-breeding season, as
measured through isotopic composition of adult feathers, closely matches a pattern of genetic
divergence observed in nuclear introns and microsatellites (Welch et al. In Prep). Notably, this
divergence does not correspond to geographic distance between breeding locations, variation in
breeding phenology (S. Judge, personal communication), or breeding habitat type (Welch et al.
In Prep). Our data from colonies on Maui and Hawaii further afford the ability to look back in
time at foraging and genetic diversity. Not only did prehistoric colonies have a greater distinction
in foraging habits than modern colonies, ancient DNA shows they were more genetically
divergent (Welch et al. In Prep.). Such temporal shifts in HAPE ecology may have important
influences on the species‘ ability to respond to future environmental change and its long-term
survival. In addition, our perspective of ancient populations suggests intricate connections
between seabird foraging ecology, genetic structure, and the increasingly dominant role of
humans in the Pacific realm

89

LITERATURE CITED

90

Adams, J. and F. Flora. 2009. Correlating seabird movements with ocean winds: linking satellite
telemetry with ocean scatterometry. Marine Biology 157: 915-929.
Ainley, D.G., et al. 1997. New insights into the status of the Hawaiian Petrel on Kauai. Colonial
Waterbirds 20 (1): 24–30.
Altabet, M.A. and R. Francois. 1994. Sedimentary nitrogen isotopic ratio as a recorder for
surface ocean nitrate utilization. Global Biogeochemical Cycles 8 (1): 103-116.
Becker, B.H. and S.R. Beissinger. 2006. Centennial Decline in the Trophic Level of an
Endangered Seabird after Fisheries Decline. Conservation Biology 20 (2): 470-479.
Brown, T.A., D.E. Nelson, J.S. Vogel, and J.R. Southon. 1988. Improved collagen extraction by
modified Longin method. Radiocarbon 30 (2): 171-177.
Chamberlain, C.P., et al. 2005. Pleistocene to recent dietary shifts in California condors.
Proceedings of the National Academy of Sciences 102 (46): 16707-16711.
Emslie, S.D. and W.P. Patterson. 2007. Abrupt recent shift in 13 C and 15 N values in Adelie
penguin eggshell in Antarctica. Proceedings of the National Academy of Sciences 104
(28): 11666-11669.
Farrell, J.W., T. F. Pedersen, S. E. Calvert, and B. Nielsen. 1995. Glacial-interglacial changes in
nutrient utilization in the equatorial Pacific Ocean. Nature 377: 514-517.
Forero, M. G, K. A Hobson, G. R Bortolotti, J. A Donázar, M. Bertellotti, and G. Blanco. 2002.
Food resource utilisation by the Magellanic penguin evaluated through stable-isotope
analysis: segregation by sex and age and influence on offspring quality. Marine Ecology
Progress Series 234: 289–299.
13

Francey, R.J., et al. 1999. A 1000-year high precision record of δ C in atmospheric CO2. Tellus
B 51(2): 170-193.
Friesen, V.L., T.M. Burg, and K. D. McCoy. 2007. Mechanisms of population differentiation in
seabirds. Molecular Ecology 16(9): 1765–1785.
Furness, R.W. and T.R. Birkhead. 1984. Seabird colony distributions suggest competition for
food supplies during the breeding season. Nature 311: 655-656.
Goericke, R. and B. Fry. 1994. Variations of Marine Plankton Delta-C-13 with Latitude,
Temperature, and Dissolved Co2 in the World Ocean. Global Biogeochemical Cycles 8:
85-90.

91

Gordon, J.E. and D.D. Harkness. 1992. Magnitude and geographic variation of the radiocarbon
content in Antarctic marine life: Implications for reservoir corrections in radiocarbon
dating. Quaternary Science Reviews 11 (7-8): 697-708.
Graham, B.S., P.L. Koch, S.D. Newsome, K.W. McMahon, D. Aurioles. 2010. Using Isoscapes
to Trace the Movements and Foraging Behavior of Top Predators in Oceanic Ecosystems.
In: Isoscapes: Understanding Movement, Pattern, and Process on Earth Through Isotope
Mapping (ed. West JB). Springer Science.
Grove, J.M. 2004. Little ice ages: ancient and modern, Second Edition. Routledge. London,
United Kingdom.
Hilton, G.M., et al. 2006. A stable isotopic investigation into the causes of decline in a subAntarctic predator, the rockhopper penguin Eudyptes chrysocome. Global Change
Biology 12(4): 611–625.
James, H., A. Wiley, P. Ostrom, T. Stafford, and J. Southon. A marine reservoir correction for
the Hawaiian petrel, a far-ranging pelagic seabird. In preparation.
Kelly, J. F. 2000. Stable isotopes of carbon and nitrogen in the study of avian and mammalian
trophic ecology. Canadian Journal of Zoology 78: 1–27.
Kirch, P.V. 1990. The evolution of sociopolitical complexity in prehistoric Hawaii: An
assessment of the archaeological evidence. Journal of World Prehistory 4 (3): 311-345.
Lewis, S., et al. 2001. Evidence of intra-specific competition for food in a pelagic seabird.
Nature 412 (6849): 816–819.
Longin, R. 1971. New method of collagen extraction for radiocarbon dating. Nature 230: 241242.
Michener, R.H. and D.M. Schell. 1994. Stable isotope ratios as tracers in marine aquatic food
webs. Stable Isotopes in Ecology and Environmental Science 7: 138–157.
Myers, R.A. and B. Worm. 2003. Rapid worldwide depletion of predatory fish communities.
Nature 423 (6937): 280–283.
Norris, D.R., P. Arcese, D. Preikshot, D.F. Bertram, and T.K. Kyser. 2007. Diet reconstruction
and historic population dynamics in a threatened seabird. Journal of Applied Ecology 44
(4): 875-884.
Nunn, P. 2007. Climate, environment and society in the Pacific during the last millennium.
Developments in Earth and Environmental Sciences 6: 1-302.
Olson, S.L. and H.F. James. 1982a. Fossil birds from the Hawaiian Islands: evidence for
wholesale extinction by man before western contact. Science 217 (4560): 633.
92

Olson, S.L. and H.F. James. 1982b. Prodromus of the fossil avifauna of the Hawaiian Islands,
Smithsonian Institution Press.
Myers, R.A. and B. Worm. 2003. Rapid worldwide depletion of predatory fish communities.
Nature 423 (6937): 280–283.
Pauly, D., C. Villy, J. Dalsgaard, R. Froese, and F. Torres. 1998. Fishing Down Marine Food
Webs. Science 279 (5352): 860 -863.
Paxinos, E.E., H.F. James, et al. 2002a. Prehistoric decline of genetic diversity in the nene.
Science 296 (5574): 1827.
Paxinos, E.E., H.F. James. 2002b. mtDNA from fossils reveals a radiation of Hawaiian geese
recently derived from the Canada goose (Branta canadensis). Proceedings of the
National Academy of Sciences 99 (3): 1399.
Piatt, J.F., et al. 2007. Introduction: A Modern Role for Seabirds as Indicators. Marine EcologyProgress Series, 352: 199-204.
Pyle, P. 2008. Identification Guide to North American Birds, Part 2. Slate Creek Press, Point
Reyes Station, California.
Reimer, P.J. et al., 2009. IntCal09 and Marine09 radiocarbon age calibration curves, 0-50,000
years cal BP. Radiocarbon 51 (4): 1111-1150.
Rick, C. and J.M. Erlandson. 2008. Human Impacts on Ancient Marine Ecosystems: A Global
Perspective. University of California Press, Berkeley.
Rucklidge, G.J., G. Milne, B.A. McGaw, E. Milne, and S.P. Robins. 1992. Turnover rates of
different collagen types measured by isotope ratio mass spectrometry. Biochimica et
Biophysica Acta (BBA) - General Subjects 1156 (1): 57-61.
Simons, T.R. 1985. Biology and behavior of the endangered Hawaiian Dark-rumped Petrel. The
Condor 87 (2): 229–245.
Spear, L.B., D.G. Ainley and N. Nur. 1995. Population size and factors affecting at-sea
distributions of four endangered procellariids in the tropical Pacific. The Condor 97 (3):
613–638.
Stafford Jr., T.W., K. Brendel and R.C. Duhamel. 1988. Radiocarbon, 13C and 15N analysis of
fossil bone: Removal of humates with XAD-2 resin. Geochimica et Cosmochimica Acta,
52 (9): 2257-2267.
Stafford, T.W., et al. 1991. Accelerator radiocarbon dating at the molecular level. Journal of
Archaeological Science 18 (1): 35-72.
93

Steele, J.H. 1998. Regime Shifts in marine ecosystems. Ecological Applications 8 (sp1): S33S36.
Takai, N., S. Onaka, Y. Ikeda, A. Yatsu, H. Kidokoro, and W. Sakamoto. 2000. Geographical
variations in carbon and nitrogen stable isotope ratios in squid. Journal of Marine
Biological Association of the United Kingdom 80: 675–684.
Thompson, D.R., R. W. Furness, and S.A. Lewis. 1995. Diets and long-term changes in delta15N
and delta13C values in northern fulmars Fulmarus glacialis from two northeast Atlantic
colonies. Marine Ecology Progress Series 125: 3–11
Wainright, S. C., M. J. Fogarty, R. C. Greenfield, and B. Fry. 1993. Long-term changes in the
Georges Bank food web: trends in stable isotopic compositions of fish scales. Marine
Biology 115 (3): 481-493.
Warham, J., 1996. The behaviour, population biology and physiology of the petrels. Academic
Press. Christchurch, New Zealand.
Watson, R. and D. Pauly. 2001. Systematic distortions in world fisheries catch trends. Nature
414 (6863): 534-536.
Welch, A.J., A.E. Wiley, H.F. James, P. Ostrom, J. Adams, F. Duvall, N. Holmes, D. Hu, J.
Penniman, K. Swindle, and R.C. Fleischer. Population divergence and gene flow in an
endangered and highly mobile seabird. In Review.
Worm B., et al. 2003. Impacts of Biodiversity Loss on Ocean Ecosystem Services. Science 314
(5800): 787-790.
Worm, B., et al. 2006. Impacts of biodiversity loss on ocean ecosystem services. Science 314
(5800): 787-790.
Wiley, A., P.H. Ostrom, C.A. Stricker, H. James, and H. Gandhi. 2010. Isotopic
Characterization of Flight Feathers in Two Pelagic Seabirds: Sampling Strategies for
Ecological Studies. The Condor 112(2): 337-346.
Wilmshurst, J.M. et al., 2011. High-precision radiocarbon dating shows recent and rapid initial
human colonization of East Polynesia. Proceedings of the National Academy of Sciences
108 (5): 1815-1820.
Wong, W.W., L.L. Clarke, G.A. Johnson, M. Llaurador, and P.D. Klein. 1992. Comparison of
two elemental-analyzer gas-isotope-ratio mass spectrometer systems in the simultaneous
measurement of carbon-13/carbon-12 ratios and carbon content in organic samples.
Analytical Chemistry 64 (4): 354-358.

94

CONCLUSION
GENERAL COMMENTS
This dissertation research highlights many of the challenges and utilities of using stable isotopes
to investigate seabird foraging ecology. Sampling feathers for isotope analysis is a seemingly
simple portion of many studies. Yet, as Chapter 1 demonstrates, non-ecologically based isotopic
variation likely exists within at least some feathers. Stable hydrogen isotope data from Hawaiian
petrels show that feathers also contain previously uncharacterized sources of D variation.
Chapter 2 advanced the first hypothesis to explain this variation: differences in salt load leading
to variable rates of water excretion through the salt glad. If the salt load hypothesis can be tested
and supported in controlled settings, D variation may prove highly useful in future studies of
seabird diet. For example, multiple seabirds are thought to have shifted to high-salt, krill-based
diets during the past 200 years. D values may provide a new approach for testing this
interpretation: another line of evidence that can be used to demonstrate temporal shifts in diet
composition.
Changes to the foraging habits of the Hawaiian petrel through time caution that humans
may play a substantive role in modifying seabird ecology. Isotopic results presented in Chapter 3
are consistent with a decline in approximately two thirds of a trophic level. This dietary shift has
largely unknown consequences, but previous studies suggest that given trophic decline, the
nutritional quality of Hawaiian petrel diet is likely lower now than in the past. Chapter 3 also
presents evidence that the foraging segregation between Hawaiian petrel colonies has diminished
through time. While such a shift may or may not prove detrimental to population viability, it will
likely have complex effects on genetic structure within the species. Overall, the results of the
present research argue for a careful consideration of seabird foraging habits. While most seabird

95

management plans focus on terrestrial-based threats, recent shifts in prey availability could
potentially undermine long-term population sustainability.

96

Figure 16. Schematic of dissertation with major conclusions from each chapter. Thick arrows represent the major flow of ideas within
chapters. Thin arrows represent the flow of ideas between chapters and leading to future research questions.

97

QUESTIONS FOR FUTURE RESEARCH
As highlighted in Figure 16, the research contained in this dissertation raises several questions
for future study.
What is the effect of melanin on feather isotope values? Chapter 1 provides evidence
13

for a consistent difference in the  C composition of black versus white feather material. If the
pigment melanin (the source of black coloration in the feathers analyzed) does indeed cause
isotopic variation in feathers, this effect must be supported with further study, quantified, and
13

recognized by the field of researchers who currently use  C values of feathers to make
ecological interpretations about avian biology. Studies focused on a single feather type from a
single species can avoid any confounding effect from melanin by choosing stable isotope
samples that do not vary in their melanin content (as in Chapters 2 and 3). Unfortunately, such
13

precautions are not always possible for studies that seek to compare  C values among multiple
species or among feathers with different coloration.
The isotopic effect of melanin pigments will be most successfully characterized by two
methods: 1) measuring the isotopic composition of melanin extracted from feathers and, 2)
analyzing melanin-rich and melanin-free feather grown by birds with a controlled diet. The
former method will require extraction of melanin using a protocol that does not alter isotopic
composition. Such a protocol may take substantive time to develop, but would result in the most
direct assessment. Controlled diet experiments will be most useful if they focus on multiple
species with variable melanin content in their feathers, and if melanin-rich and melanin-free
feathers are grown simultaneously by the same individuals.
Does increasing salt load increase D? Chapter 2 shows unpredicted levels of D
variation within seabird feathers: variation that is hypothesized to result from differences in salt

98

load and resulting salt gland excretion in adult versus some hatch-year petrels. The ―salt load
hypothesis‖ deserves testing in additional seabird systems. If verified and quantified, salt loadinduced variability in D would provide a new tool for use in the study of modern and ancient
seabird ecology. Most notably, seabirds with anthropogenically-driven dietary shifts could be
tested for an increase in salt load: a change that could pose a physiological challenge, especially
to chicks with underdeveloped salt glands.
The effect of salt load and salt gland excretion on the D composition of seabird tissues
will be best tested in a controlled setting, where treatments of low versus high salt load are
applied using diet. For example, captive gulls could be fed exclusively marine invertebrates (a
high salt load diet) or exclusively teleost fish (a lower salt load diet) during a period of feather
growth. Importantly, the magnitude of isotope effects in a single captive bird species cannot
necessarily be extrapolated to other species. The efficiency of the salt gland differs between
species, meaning that some species will lose more water through their salt glands than other,
even when salt load is similar. As a result, the impact of a given salt load on the D composition
of growing tissues may also differ between species.
What drives variation in foraging habits among seabird colonies? Chapter 2
documents substantial foraging segregation between closely-spaced seabird colonies. While the
presence of this intra-specific diversity has important implications for seabird conservation (at
least in the Hawaiian petrel), the mechanism by which foraging segregation arises within seabird
species is unclear. Understanding the forces that drive variation between colonies may be critical
if we, as managers, wish to preserve the differences between seabird colonies that live and feed
in increasingly human-modified environments. Currently, it is unknown whether pelagic seabirds
such as the Hawaiian petrel rely heavily on social cues from other individuals in their breeding
99

colonies in order to locate and choose prey items. Such interactions would explain the observed
similarity of foraging habits within island colonies, but they are logistically difficult to study
given the vast foraging habitat of pelagic species. A strong genetic basis for foraging habits
could also explain inter-colony variation, and may prove easier to elucidate. Conservation efforts
have resulted in new breeding colonies, formed artificially by human-rearing of chicks and
placement on a formerly-unoccupied site. Comparison of the foraging habits between newly
formed colonies and their ―parent colonies‖, especially when colonies are thought to have no
contact at sea, may help to elucidate the degree to which seabird foraging habits are genetically
programmed.
How widespread is trophic decline? Chapter 3 provides evidence for a fishery-induced
trophic decline in a wide-ranging pelagic seabird, a phenomenon that has been previously
observed in seabirds feeding in more human-dominated, coastal food webs. These results imply
that fisheries have had a drastic impact on pelagic food webs, even in temperate and tropical
waters, where targeted species are typically considered more resilient. However, the mechanisms
by which trophic decline occurs, whether through reduction in size of traditionally-targeted
seabird prey, or through a change in the species-composition of seabird prey, remains unclear.
Future study of trophic decline in pelagic waters should focus on documenting trophic decline in
other organisms. Such research will clarify the extent to which trophic decline widespread:
spatially, at different trophic levels, and for generalist versus specialist predators. Understanding
the distribution of trophic shifts, in turn, will provide rare insight into the changing interactions
of pelagic food webs. While long-term dietary records are not always readily available, isotope
studies have successfully used, and can continue to focus on historical museum collections.
Subfossil bones from paleontological and archaeological sites, such as those used to construct
100

isotope chronologies in Chapter 3, may be especially useful for documenting dietary shifts in
additional seabird species.
Are we losing ecological variation within seabird species? Unlike previous studies of
shifting seabird diet, Chapter 3 provides evidence for a waning distinction between breeding
colonies (both genetically and ecologically). Coalescence of foraging habits is particularly
interesting, as it occurs simultaneously with declining trophic level. Human fisheries may
therefore have a complex effect on the intra-specific dynamics of seabirds. It remains unclear
whether or not a loss of diversity between seabird colonies will result in an overall decline in
diversity within a species. This uncertainty can be at least partially addressed by studies of
modern seabird colonies and the distribution of dietary diversity within and between individuals
from different breeding sites. Just as importantly, research should focus on the extent to which
ecological variability is shifting within other species of seabirds. Previous work has documented
a change in the average diet of seabird species through time, but declining variation within a
species may also have critical consequences. As large-scale industrial fishing continues, seabird
species are more likely to adapt when they employ a variety of foraging strategies.

101