the ecological study of the maritime ringlet butterfly (coenonympha nipisiquit mcdunnough) in daly point, bathurst, new brunswick

THE ECOLOGICAL STUDY OF THE MARITIME RINGLET BUTTERFLY COENONYMPHA NIPISIQUIT MCDUNNOUGH IN DALY POINT, BATHURST, NEW BRUNSWICK.. Porter I studied the autecology, community ecology, and

THE ECOLOGICAL STUDY OF THE MARITIME RINGLET BUTTERFLY

(COENONYMPHA NIPISIQUIT MCDUNNOUGH) IN DALY POINT, BATHURST,

NEW BRUNSWICK

A Dissertation Presented

by MAKIRI SEI

Submitted to the Graduate School of the University of Massachusetts Amherst in partial fulfillment

of the requirements for the joint degree of DOCTOR OF PHILOSOPHY

February 2006 Organismic and Evolutionary Biology

UMI Number: 3206207

3206207 2006

UMI Microform Copyright

All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, MI 48106-1346

by ProQuest Information and Learning Company

© Copyright by Makiri Sei 2006All Rights Reserved

THE ECOLOGICAL STUDY OF THE MARITIME RINGLET BUTTERFLY

(COENONYMPHA NIPISIQUIT MCDUNNOUGH) IN DALY POINT, BATHURST,

NEW BRUNSWICK

A Dissertation Presented

byMAKIRI SEI

Approved as to style and content by:

Organismic and Evolutionary Biology

To my parents, Toshio and Noriko Sei,

who wholeheartedly supported me whiled I studied abroad

I would like to express my thanks to my advisor, Adam H Porter, for his manyyears of very positive, thoughtful, and patient guidance and generosity I would like toextend my gratitude to the members of my committee, Robin H Harrington, Paul R.Sievert, and Roy G Van Driesche for their helpful discussions and suggestions on allaspects of this endeavor My dissertation has improved through communications withMitchell Baker, Thomas Crist, Christy Hentzler, Daniel P Jennings, Christopher Majka,Ben Normark, Ann Rypstra, Baiqing Wang, Reginald P Webster, Arnold D Well, andSean Werle, who generously shared their expertise The maps in Figure 1, 2, and 6 werereproduced with permission from RPW

I am indebted to the Xerces Society, the National Science Foundation, the

Graduate Program of Organismic and Evolutionary Biology, and the Department ofNatural Resources and Energy of New Brunswick for funding this research

This research was conducted under two unidentified permits issued in 1999 and

2000, #2001-06, #2002-11, and #ES03-014 issued from the Department of Natural

Resource and Energy, New Brunswick and #2003-05-15-502-11-S-F issued from

Direction de l’Aménagement de la Faune de la Gaspésie-Îles-de-la-Madeleine, Québec Iwould like to thank Maureen Toner, Dwayne Sabine, Luc Gagnon, Gilles Godin, NathalieD’Aoust of the Department of Natural Resource and Energy, New Brunswick, O’NeilPelletier and Julie Gaudet of Community College of New Brunswick, Bathurst, and StanGeorges of Société de la Faune et des Parcs du Québec for their generous support of myfieldwork in Canada

I would like to express my appreciation to Nathalie Arsenault, Christian Brideau,Christopher Drysdall, Paul Ferron, Alain Gouge, Patricia Levigne, Alexandre Haché, andMelanie Boudreau for their participation in this project Special thanks to Jeremy Houser,Christa Skow, Baiqing Wang, Elizabeth G Wells, and Clayton B Winter for the

camaraderie over the years

And lastly, a very special thank you to Chad D Hoefler whose constant supportand thoughtful advices helped me in the course of development and progress of thisproject and enabled me to persevere

ABSTRACTTHE ECOLOGICAL STUDY OF THE MARITIME RINGLET BUTTERFLY

(COENONYMPHA NIPISIQUIT MCDUNNOUGH) IN DALY POINT, BATHURST,

NEW BRUNSWICKFEBRUARY 2006MAKIRI SEI, B.S., UNIVERSITY OF TENNESSEE MARTIN

Ph.D., UNIVERSITY OF MASSACHUSETTS AMHERST

Directed by: Professor Adam H Porter

I studied the autecology, community ecology, and genetics of an endangered

butterfly, the maritime ringlet (Coenonympha nipisiquit McDunnough), that inhabits a

limited number of salt marshes in northern New Brunswick and in the Gaspé Peninsula ofQuébec I studied the survival rate of first- and second-instar larvae in various

microhabitats in a salt marsh at Daly Point Natural Reserve, Bathurst, New Brunswick I

found they survived significantly better in microhabitats dominated by Spartina patens

(Aiton) Muhl at an intermediate elevation I investigated the tolerance of the maritimeringlet larvae to tidal submergence and compared their performance to a closely related

taxon, the inornate ringlet (C tullia inornata Edwards) The experiments revealed that

the maritime ringlet possesses unique adaptations to tidal submergence I examined theflight and oviposition behaviors of adult females in response to microhabitat I found thatthey did not discriminate between microhabitats based upon the likelihood of larval

survival as long as S patens or other potential hosts were abundant I explored the

correlation between predator species richness and abundance with the larval survival rate

in microhabitats I found that predator abundance and species richness often responded

negatively to increasing tidal flooding, suggesting that high larval mortality at highelevation sites can be caused by high predation pressure Lastly, I investigated the

possibility of genetic introgression between the maritime ringlet and inornate ringlet and

reconstructed the phylogeny of the C tullia-group taxa in North America The genetic

evidence did not support the possibility of large-scale genetic introgression and raised the

taxonomic status of the maritime ringlet from a subspecies of holarctic C tullia to a full

species The phylogenetic analyses suggested that the divergence of the maritime ringletwas much earlier than previously believed My results will aid in protection and recovery

of this endangered species

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS v

ABSTRACT vii

LIST OF TABLES xi

LIST OF FIGURES xiv

CHAPTER 1 INTRODUCTION 1

Description of the salt marsh in the Daly Point Natural Reserve 4

2 MICROHABITAT-SPECIFIC EARLY LARVAL SURVIVAL OF THE MARITIME RINGLET (COENONYMPHA NIPISIQUIT) 8

Introduction 8

Materials and Methods 10

Results 15

Discussion 17

3 LARVAL ADAPTATION OF THE MARITIME RINGLET TO A SALINE WETLAND HABITAT 26

Introduction 26

Materials and Methods 29

Results 32

Discussion 34

4 FLIGHT AND OVIPOSITION BEHAVIOR OF THE ADULT MARITIME RINGLET FEMALES IN RESPONSE TO MICROHABITAT 41

Introduction 41

Materials and Methods 43

Results 48

Discussion 50

5 BETWEEN THE DEVIL AND THE DEEP BLUE SEA: PREDATION AND FLOODING MAY DELIMIT LARVAL MICRODISTRIBUTION OF THE MARITIME RINGLET 69

Introduction 69

Materials and Methods 72

Results 79

Discussion 85

6 SPECIES DELIMITATION AND CONSERVATION OF THE MARITIME RINGLET BUTTERFLY 117

Introduction 117

Materials and Methods 120

Results 126

Discussion 130

7 CONCLUSION 150

BIBLIOGRAPHY 155

LIST OF TABLES

1 The 10 microhabitat types found in the salt marsh at Daly Point, Bathurst,

New Brunswick, following Webster (1994) 6

2 The number of single and multiple release sites in microhabitats 21

3 Estimated daily death rate (d ) and finding rate (ˆ f ) along with the 2-unit ˆ

support limits (SL) from single release sites 22

4 The minimum number of larvae known to be alive (MNKA)

in multiple release sites at the end of the observation 23

5 The number of inornate and maritime ringlet larvae alive before the

experiment, 2 days, and 4 days after the end of submergence 39

6 Logistic regression coefficients (± SEM) for larval mortality prediction equations

after varying duration of submergence 40

7 Description of butterfly wing wear after Watt et al (1977) and Webster (1994) 56

8 Classifications of microhabitats used as competing hypotheses 57

9 The effects of individuals (ID), age, and microhabitats on path parameters,

with microhabitats classified according to larval survival rate (Sei &

Porter 2003) and emergence pattern of freshly eclosed adults (Webster

1994) 58

10 The effects of individuals (ID), age, and microhabitats on path parameters,

with microhabitats classified according to vegetation density 59

11 The effects of individuals (ID), age, and microhabitats on path parameters,

with microhabitats classified according to principal host availability 60

12 Percent frequency of activities recorded every minute among the three

age classes 61

13 The effects of individuals (ID), age, and microhabitats on the frequency of

various behaviors The microhabitats are classified according to larval

survival rate (Sei & Porter 2003; Webster 1994) 62

14 The effects of individuals (ID), age, and microhabitats on the frequency of

various behaviors 63

15 The effects of individuals (ID), age, and microhabitats on occurrence of

behaviors 64

16 The number of eggs oviposited per minute of stay in one location 65

17 The sequence patterns of behaviors 66

18 The predators used for the palatability tests and their collection sites 94

19 Spider species and their average densities caught by sweep-netting at nine transects 95

20 Insect species and their average densities caught by sweep-netting at nine transects 97

21 The effects of transects and collection dates on abundance of predators caught by sweep-netting 98

22 The effects of tidal flooding frequency (the number of days when flooding occurred at a transect) and log-transformed abundances of Diptera and Hemiptera on abundances of predators caught by sweep-netting 99

23 Spider species and their average densities caught by pitfall trapping at nine transects .101

24 Insect species and their average densities caught by pitfall-trapping at nine transects 103

25 The effects of transects and collection dates on abundance of spiders caught by pitfall-trapping 105

26 The effects of transects and collection dates on abundance of insect predators caught by pitfall-trapping 106

27 The effect of tidal flooding frequency as the number of days when flooding occurred at a site on abundance of spiders caught by pitfall-trapping 108

28 The effect of tidal flooding frequency as the number of days when flooding occurred at a site on abundance of insect predators caught by pitfall- trapping 109

29 The correlation between larval death rate (Table 3) and above-vegetation predator abundances in microhabitats 1, 2, 4, 6, and 8 110

30 The correlation between larval death rate (Table 3) and ground predator abundances in microhabitats 1, 2, 4, 6, and 8 111

31 The principal components and their associated predators captured with the

pitfall traps 112

32 Predators and the number of feeding trials that resulted in predation 113

33 The C tullia-group taxa, their collection sites, and abbreviations used in

39 The percentages of polymorphic AFLP loci having the lower 95% confidence

limits of FST larger than 0 and 0.25 142

LIST OF FIGURES

1 The Daly Point salt marsh and its microhabitats 7

2 Microhabitats in the Daly Point salt marsh and the larval release sites (rectangles) 24

3 The plot of daily death rate d vs standardized log likelihood values for the five microhabitats within two-unit support limits 25

4 The distribution of move lengths observed among the 71 recorded paths 67

5 The distribution of turning angles observed among the 71 recorded paths 68

6 The Daly Point salt marsh and the nine 40 m transects 114

7 Sample-based rarefaction curves for species obtained by sweep-netting for the entire study period 115

8 Sample-based rarefaction curves for predator species obtained by pitfall-trapping for the entire study period 116

9 The distribution of C nipisiquit around the Bay of the Chaleurs in northeastern North America (inset) 143

10 The majority-rule consensus of 876 trees based on COI + CR identified through maximum parsimony analysis (tree score = 704, CI = 0.8565, RI = 0.6930) 144

11 The majority-rule consensus of trees based on COI + CR identified through maximum likelihood analysis (5726.03≤ -lnL ≤ 5728.03, CI = 0.759) 145

12 The single tree based on AFLP markers identified through maximum parsimony analysis (tree score = 585, CI = 0.6154, RI = 0.5755) 146

13 The 589 polymorphic loci sorted by the mean FST with the upper and lower confidence limits estimated for C inornata and C nipisiquit (pooled populations) 147

14 The 456 polymorphic loci sorted by the mean FST with the upper and lower confidence limits estimated for C nipisiquit (NB vs QC populations) 148

15 The 261 polymorphic loci sorted by the mean FST with the upper and lower confidence limits estimated for C inornata (NB vs QC populations) 149

(Webster 1994, 1995; Godin pers comm.) Historically, maritime ringlets may have had

a wider distribution surrounding the Chaleur Bay In Quebec, four populations on the Gaspé Peninsula are known to exist (Gouge 2002, pers comm.) Two new populations

on the Acadian Peninsula have been established for several years after successful

reintroduction efforts (pers obs.) The presently limited distribution may be due to aggressive manipulation of the salt marshes, including the building of dykes and ditches, grazing, exploitation for salt hay, and burning, all of which used to be prevalent in the northern Atlantic coast (Roberts & Robertson 1986; Adam 1990) The general isolation

of the salt marshes, as well as the philopatric nature of maritime ringlets, seems to

preclude this subspecies from naturally colonizing other salt marshes Several

Coenonympha species receive conservation attention in Europe partly because of their

low vagility (Joy & Pullin 1997; Lhonore & Lagarde 1999; Cassel et al 2001) With so few extant maritime ringlet populations in such a strongly delimited and vulnerable

habitat type, there is considerable need to study the ecology of this butterfly in depth

The natural history of the maritime ringlet differs in several ways from that of the

inornate ringlet, Coenonympha tullia inornata Edwards, the sister North American

subspecies that occurs in the adjacent uplands (Webster 1998) The occurrence of

maritime ringlets coincides with salt meadow cordgrass, Spartina patens (Aiton) Muhl (Poaceae), and sea lavender, Limonium nashii Small (Plumbaginaceae) Salt meadow

cordgrass grows in the high marsh zone within a salt marsh and is known to be the

primary host plant of the larvae (Webster 1998) Sea lavender is considered to be the primary nectar source for the adult maritime ringlets and a good indicator of the maritime ringlet occurrence, although the adults can readily use other nectar sources The

maritime ringlet is univoltine, and the adults fly from late July to mid August, a month later than common ringlet populations (Wiernasz 1989) that occur within a kilometer of the salt marshes Larvae enter diapause as the second instar in late September, and they resume growth and development in late May of the following year The allele frequency data in Wiernasz (1989) suggest that the maritime ringlet may be only distantly related to all of the inornate ringlet populations sampled from the Northeastern United States and Canada It is speculated that the maritime ringlet evolved from a small coastal population

of the common ringlet Coenonympha tullia during an episode of glaciation, and the

inornate ringlet came to occupy adjacent upland in recent decades (Christie 1983;

Handfield 1989; Majka pers comm.) Thus, the maritime ringlet is a remarkable

evolutionary phenomenon that occupies a rare niche as a salt marsh butterfly: a natural heritage in need of protection

Conservation of the maritime ringlet requires relatively intensive management,

more than the protection of extant populations, because salt marshes are not free climax communities In the short term, the salt marsh vegetation is often disturbed

disturbance-by storm (Long & Mason 1983; Adam 1990) winter ice scour (Richard 1978; Roberts & Robertson 1986), and tidal erosion of the marsh itself (Jacobson & Jacobson 1987)

Because salt marshes inhabited by the maritime ringlets often occur near or within city

limits, the butterflies are at risk of habitat loss or habitat deterioration by urban and industrial sprawl and pollution (Roberts & Robertson 1986; Adam 1990) In the long term, the vegetation may change as the salt marsh hydrology changes due to sea level rise (Warren & Niering 1993), or as vegetation competition patterns are altered from

increased soil nitrogen level (Valiela et al 1975; Levine et al 1998)

Intensive management decisions need to be based on sound scientific data about the ecology of a species The condition of a butterfly population is usually monitored by assessing adult population size (New 1997) because other life stages are difficult to sample (White & Singer 1987) Thus, more often than not, the health of a population is monitored by a pattern (e.g the number of adults), not by adressing the underlying

processes that control the population size By the time one sees the sign of decline in population size, it may be too late to rescue the population by investigating the cause of decline in order to amend the problem Thus, there is a need to study the maritime ringlet population biology in detail

I chose a population of the maritime ringlet that occurs in a salt marsh within the Daly Point Natural Reserve as a model system because the vegetation composition of the microhabitats and adult natural history of this population are known from a previous fieldwork (Webster 1994) My dissertation project aimed at understanding the factors

that influence survival and persistence of the maritime ringlet I have studied early larval mortality and adult female movement and oviposition behavior to test whether

microhabitats chosen for oviposition overlaps with microhabitats suitable for the larvae

In addition, I explored the cause of larval mortality by surveying predator communities across the salt marsh and by studying the effects of tidewater submergence of larvae in laboratory Lastly, I investigated the possibility of genetic introgression between the

maritime ringlet and inornate ringlet and reconstructed the phylogeny of the C

tullia-group taxa in North America to help guide conservation efforts concordant with

evolutionary processes

Description of The Salt Marsh in the Daly Point Natural Reserve

The salt marsh is located in the Daly Point Natural Reserve in Bathurst, New Brunswick, Canada (Figure 1) The natural reserve was owned by Noranda Inc., a mining corporation, and leased to the Department of Natural Resources & Energy in Bathurst until 2002 Its ownership has now changed to the city of Bathurst

The salt marsh is about a kilometer from the nearest road A narrow sandbar and pebble beach separate most of the marsh from the tidal flat The vegetation pattern of this marsh is similar to that of the northern New England salt marsh described by

Jacobson & Jacobson (1987), with clear zonation of vegetation according to the height above sea level

The low-marsh habitat along tidal streams experiences daily tidal submergence

(Nixon, 1982), and a dense monoculture of Spartina alterniflora dominates Flooding of the high-marsh habitat is less frequent, and Spartina patens and Juncus gerardi

(Juncaceae) dominate, the latter in slightly higher sites Permanent Distichlis spicata

(Poaceae) stands are found in the high marsh where drainage is poor The eastern half of the marsh supports mixed vegetation dominated by forbs (Miller & Egler 1950; Jacobson

& Jacobson 1987; Warren & Niering 1993; Theodose & Roths 1999): a mixture of S alterniflora, S patens, Limonium nashii, Glaux maritima (Primulaceae), Plantago

maritima (Plantaginaceae), and Triglochin maritima (Juncaginaceae) The southwestern

portion of the marsh tends to be drier than the rest, and non-halophytic grasses such as

Festuca rubra grow among S patens and J gerardi Above the tidal inundation level, the salt marsh plants are replaced by Rosa virginiana and other woody plants, or by freshwater marsh plants (e.g., Typha sp.), depending on the elevation and proximity to

fresh water sources

Webster (1994) subdivided this marsh into 10 microhabitat types according to the

vegetation composition, the density of S patens and L nashii, the height of S patens, and

hydrology (Table 1, Figure 1) We follow his classification in this study

Table 1 The 10 microhabitat types found in the salt marsh at Daly Point, Bathurst, New Brunswick, following Webster (1994)

Microhabitat Description

1 Mix of Spatina patens, S alterniflora, Glaux maritima, Plantago maritima,

and Limonium nashii (50-150 plants/400m2), very wet

2 S patens (90-95% stem density, 15-20cm tall), S alterniflora, and L nashii

5 Distichlis spicata dominant, very wet

6 Juncus gerardi dominant (85-95% stem density), Argentina egedii, P

maritima, and S sempervirens

7 S alterniflora (90-100% stem density) and L nashii (<20 plants/400m2),

wet

8 Festuca rubra (80-90% stem density), G maritima, P maritima, A egedii,

S sempervirens, and Ligusticum scothicum, dry area

9 S sempervirens common to abundant, Aster sp., L scothicum, Cuscuta sp.,

J gerardi, and others

10 Beach, varied vegetation

CHAPTER 2

MICROHABITAT-SPECIFIC EARLY LARVAL SURVIVAL OF THE MARITIME

RINGLET (COENONYMPHA NIPISIQUIT)

Introduction

Maritime ringlet butterflies (Coenonympha nipisiquit McDunnough, Satyridae)

are a rare subspecies of the widespread holarctic common ringlet butterflies

(Coenonympha tullia Müller) Found only in salt marshes in maritime Canada, maritime

ringlets are subjected to periodic tidal submergence in their pre-adult stages, and they are one of the two species of salt marsh butterfly in the region The maritime ringlet has an extremely limited distribution and is listed as federally endangered in Canada Until recently only four populations in New Brunswick and four in the Gaspé Peninsula,

Québec, were known to exist (Webster 1998; Godin pers comm.; Gouge 2002, pers comm.) Two new populations have been established for several years after successful reintroduction efforts (pers obs.) Historically, maritime ringlets may have had a wider distribution surrounding the Bay of the Chaleurs, and the present distribution may be due

to alteration of the salt marshes, through the building of dikes and ditches, grazing,

exploitation for salt hay, and burning, all of which were once prevalent along the northern Atlantic coast (Roberts & Robertson 1986; Adam 1990) The general isolation of the salt marshes and the apparent philopatric nature of maritime ringlets prevent this subspecies

from readily colonizing other salt marshes Several Coenonympha species receive

conservation attention in Europe partly because of their low vagility (Joy & Pullin 1997; Lhonore & Lagarde 1999; Cassel et al 2001) With so few extant maritime ringlet

populations in such a strongly delimited and vulnerable habitat type, there is considerable need to identify the critical microhabitat requirements of this butterfly

The natural history of the maritime ringlet differs in several ways from that of the

inornate ringlet, Coenonympha tullia inornata Edwards, the sister North American

subspecies that occurs in the adjacent uplands (Webster 1998) The occurrence of the

maritime ringlet coincides with salt meadow cordgrass, Spartina patens (Aiton) Muhl (Poaceae) and sea lavender, Limonium nashii (Walter) Britton (Plumbaginaceae) Salt

meadow cordgrass grows in the high marsh zone within a salt marsh and is known to be the primary host plant of the larvae (Webster 1998) Sea lavender is considered to be the primary nectar source for the adult maritime ringlets and a good indicator of maritime ringlet occurrence, although the adults can readily use other nectar sources such as sea

goldenrod (Solidago sempervirens L., Asteraceae), sea milkwort (Glaux maritima L., Primulaceae), and Virginia rose (Rosa virginiana Mill., Rosaceae) The maritime ringlet

is univoltine, and the adults fly from late July to mid August, a month later than adults of the common ringlet (Wiernasz 1989) from populations within a kilometer of the salt marshes Larvae enter diapause during the second instar in late September, and they resume growth and development in late May of the following year (Webster 1998) The allozyme data in Wiernasz (1989) indicate that the maritime ringlet is genetically distinct from other univoltine populations of the common ringlet, and once the maritime ringlet's biology is better understood, it will probably be necessary to re-evaluate its taxonomic status Regardless, the maritime ringlet is a remarkable evolutionary phenomenon that occupies a rare niche as a salt marsh butterfly: a natural heritage in need of protection

Given the range of microhabitat types in a salt marsh (Miller & Egler 1950; Long

& Mason 1983; Adam 1990) and the vulnerability of butterfly larvae to predation

(Dempster 1983, 1984; Cornell & Hawkins 1995), I expected that a great deal of insight into the population biology of this butterfly could be gained by following larval

survivorship in their habitat I therefore placed larvae on plants in different microhabitats and re-sighted them repeatedly throughout a 47-day period This modified mark-release-recapture technique permitted me to estimate microhabitat-specific rates of larval

survivorship, which is uncommon among immature invertebrate studies because of the difficulty of marking (White & Singer 1987), high probability of mark loss, dispersal, and low recapture rate I estimated daily death rate using maximum likelihood (Lebreton

et al 1992), and I compared survival across microhabitats with Akaike's Information Criterion (Akaike 1973; Hilborn & Mangel 1997) and contingency table tests

Materials and Methods Study Site

I conducted this study on the salt marsh at Daly Point Natural Reserve in

Bathurst, New Brunswick, Canada (Figure 1) A narrow sandbar and pebble beach separate most of the marsh from the tidal flat The vegetation pattern of this marsh is similar to that of the northern New England salt marsh described by Jacobson & Jacobson (1987), with clear zonation of vegetation according to the height above sea level

The low marsh habitat along tidal streams experiences daily tidal submergence

(Nixon 1982), and a dense monoculture of Spartina alterniflora Loisel dominates

Flooding of the high marsh habitat is less frequent, and S patens and Juncus gerardi

Loisel (Juncaceae) dominate, the latter in slightly higher sites Permanent stands of the

grass Distichlis spicata (L.) E Greene (Poaceae) are found in the high marsh where

drainage is poor The eastern half of the marsh supports mixed vegetation dominated by forbs (see Miller & Egler 1950; Jacobson & Jacobson 1987; Warren & Niering 1993; Theodose & Roths 1999 for descriptions of such forbs) The vegetation in this part of my

site was a mixture of S alterniflora, S patens, Limonium nashii (Walter) Britton, Glaux maritima L (Primulaceae), Plantago maritima L (Plantaginaceae), and Triglochin maritima L (Juncaginaceae) The southwestern portion of the marsh is drier, and non- halophytic grasses such as Festuca rubra L grow there among S patens and J gerardi Above the tidal inundation level, the salt marsh plants are replaced by R virginiana and other woody plants, or by freshwater marsh plants (e.g., Typha sp.), depending on the

elevation and proximity to fresh water sources

Webster (1994) divided this marsh into 10 microhabitats according to the

vegetation composition, the density of S patens and L nashii, the height of S patens, and

hydrology (Table 1, Figure 1) I follow his classification in this study

Release of the Larval Cohorts

To obtain eggs, I captured 25 female C nipisiquit from a larger population nearby

at the Peters River in Beresford, New Brunswick, in early July, 1999 Because it was an

unusually warm year, the flight season of C nipisiquit was a few weeks earlier than usual I placed up to seven females with a clump of S patens in a 2-liter plastic container

with the top of the lid replaced by net fabric, left in half shade I collected eggs each day and kept them at room temperature until hatch

I prepared release sites in microhabitats 1- 6 and 8 (Figure 2) There was only one release site in each microhabitat except microhabitat 1, which had two release sites

on both sides of the tidal stream Microhabitats 7, 9, and 10 were excluded because they

were physically unlikely to support C nipisiquit larvae or found to be underutilized by

the Maritime Ringlets from previous observations (Webster 1994) A release site was an island of vegetation 10 cm in diameter within a clearing made by clipping the

surrounding vegetation I avoided use of cages in larval releases in order to minimize potential caging effects that influence microclimate and predation Without cages, the chances of larval escape are increased, but larvae were not likely to move from release

points because Coenonympha spp larvae, especially in the early instars, are sedentary as

long as the food is abundant (A H Porter pers obs.) Joy & Pullin (1999) studied

overwintering survival of a bog-dwelling population of C tullia in England using a

similar larval release scheme I assumed movement of larvae into or out of the release sites was negligible Release sites were searched before the release to make sure no larvae were present before the release of the cohorts Each release point contained a single first instar larva, and the clearing deterred the larva from wandering into the

surrounding vegetation by walking across grass blades Release sites were 50cm apart, and the size of the clearing was increased where vegetation was tall I also created multiple release sites in each microhabitat type because it was necessary to save time and effort in creating release sites A multiple release site was a patch 30 cm in diameter and contained 10 larvae Single and multiple release sites in each microhabitat varied in number (Table 2) depending on the number of larvae available for release Release sites

in each microhabitat occupied about 112m2

, except 1(B) which was smaller because it only contained 10 single release sites and four multiple release sites (Table 2) All the larvae were in the first instar stage, a cohort in a microhabitat was released on the same

day, and all the cohorts were released within two weeks from late July to early August After the release of the larvae, I returned every three to four days to the release sites to

search for the larvae until early September By then, the growth rate of S patens as well

as C nipisiquit decreased

Survivorship Estimation from Single-Release Sites

My estimation model is modified from that of Lebreton et al (1992) The data consist of records of sightings of larvae at various times after release into single release

sites (1 larva/site) I used these records to simultaneously estimate the finding rate f and the daily death rate d The fact that a living larva may not be found even when present necessitates the calculation of f in order not to overestimate d The parameters f and 1-d are equivalent to the capture rate p and survival rate φ in Lebreton et al (1992) I

assumed that both f and d are constant through the early larval period

I used maximum likelihood (Edwards 1992) to obtain the most likely values for f and d Each possible value of f and d can be seen as an hypothesis The likelihood of a

hypothesis being supported given the data observed is proportional to the probability of observing the data given the hypothesis (Edwards 1992), and the most likely hypothesis

(i.e., values of f and d) yields the largest likelihood value The likelihood is often

calculated in natural logarithm form for ease of computation The basic likelihood model

0

0

where T is the ending time (the time of the last sighting), S is the total number of sites, O

is a data matrix of O tj , O tj is the number observed at site j at time t, τ j is the last time

period in which the larva was seen alive at site j, ris the total number of times the larva

was observed, and C is a proportionality factor that is constant for a given data set and can be ignored for my purposes In this case, O tj is either 1 (found) or 0 (not found)

To maximize the likelihood, I used a Metropolis-Hastings algorithm (Press et al 1992), which locates a global maximum regardless of local maxima by moving upward (toward a maximum) but allowing downward movement with a certain probability The

algorithm randomly varies a value of f or d and settles on joint maximum likelihood

estimates f and ˆ d The algorithm saves 10,000 parameter combinations within 2 ˆ

likelihood units of the maximum to obtain support limits (Edwards 1992), roughly equivalent to 95% confidence limits I excluded the multiple release site data from the survivorship estimation because the uncertainty about the individual survival duration is confounded under the multiple release scheme

To compare the mortality rate across the microhabitats, I used Akaike's

Information Criterion (Akaike 1973; Hilborn & Mangel 1997) This criterion is

frequently used in conjunction with maximum likelihood to compare non-nested

alternative models (Hilborn & Mangel 1997) Specifically, I calculated the difference between the natural logarithm of maximum likelihood based on pooled observation data

from microhabitats a + b and the sum of maximum likelihoods based on separate

observation data from microhabitats a and b,

! ln L = ln L d[ a+ b , f a+ bOa+ b]" (ln L d[ a , f aOa]+ ln L d[ b , f bOb])

2! ln L follows a χ distribution with degrees of freedom equal to the difference in the number of estimated parameters (in this example, the degrees of freedom is 2) AIC can

also be applied to compare groups of microhabitats

Comparison of Larval Survival among Multiple Release Sites

I calculated minimum number known alive (MNKA; Nichols & Pollock 1983; Montgomery 1987) of the larvae at the end of the observation period for each multiple release site I compared the percentage of surviving larvae among the microhabitats that had at least one surviving larva at the end of the observation period (microhabitats 1, 2, 4, 8) using a contingency table test

Results Single Release Sites

The daily death rate of the larvae in single release sites was lowest in

microhabitats 1 (d = 0.0867) and 2 (ˆ d = 0.0986) (Table 3) Microhabitat 4 had a ˆ

moderately low death rate (d = 0.1445) Death rates of d > 0.15 were found in ˆ

microhabitats 3 and 6 The finding rates were high (f > 0.6) in microhabitats 1, 2, and 4 ˆ

(Table 3) The likely cause of the low finding rates in microhabitats 3 and 6 is the height

of vegetation Spartina patens and J gerardi in these microhabitats are taller than 30 cm, and S patens grows over matted thick thatch from previous years Searching for the

larvae among the tall and dense vegetation had proven to be difficult A maritime ringlet larva is 3mm long after hatching, and the cohorts have grown to 6 mm by the end of the observation A newly hatched larva is pale tan, and it turns cryptic bluish green once it

starts feeding on S patens The likelihood model yields estimates of survivorship that are

independent of the finding rate (Lebreton et al 1992), so this does not bias my

survivorship estimates I excluded microhabitat 5 from further analysis because I found

no larvae, and microhabitat 8 because of the small initial cohort size

Microhabitats 1 and 2 and microhabitats 3, 4, and 5 were found to be highly significantly different with regard to f and ˆ d by Akaike's Information Criterion test ˆ

(2! ln L = 51.138, d.f = 2, p < 0.0001) The plot of d against lnL (Figure 3) clearly

illustrates this result because the likelihood curves of those two groups do not overlap Microhabitats 1 and 2 were not significantly different (2! ln L = 1.514, d.f = 2, p = 0.469) Microhabitat 4 was not significantly different from 6 (2! ln L = 1.7018, d.f = 2,

p = 0.427), but it was significantly different from 1 (2! ln L = 16.968, d.f = 2, p = 0.0002) and 2 (2! ln L = 8.654, d.f = 2, p = 0.0132) Microhabitat 4 was only

marginally different from microhabitat 3 (2! ln L = 5.799, d.f = 2, p = 0.0551)

Multiple Release Sites

At the end of the observation period, 12.5 % of the larval cohort was alive in the multiple release sites in microhabitat 1, 15 % in microhabitat 2, 1.67% in microhabitat 4, and 5% in microhabitat 8 (Table 4) No larvae were found in microhabitats 3, 5, and 6 Based on minimum number known alive, the survival rates of larvae were significantly different among the microhabitats that had at least one surviving larva at the end of the observation period (χ2

= 10.77, d.f = 3, p = 0.013)

The number of surviving larvae in early September was too small to meaningfully estimate the survival rate over winter, but three larvae in microhabitat 1 and one larva each in microhabitats 2 and 4 were found alive in the following June, and the larva in microhabitat 4 pupated

Discussion The quality of microhabitats within a salt marsh differs considerably for the early instar larvae The single release site mortality data suggest that microhabitats 1 and 2 are likely to be the most suitable habitats for maritime ringlet larvae (Table 3) The results obtained from MNKA at multiple release sites also show that the death rates among microhabitats were significantly different and lowest in microhabitat 2 (Table 4) The quality of microhabitat 4 seems to be fairly good for the young maritime ringlet larvae as well; although it had a significantly higher death rate than microhabitats 1 or 2, larvae survived better in microhabitat 4 than in microhabitat 3, and a larva pupated in

microhabitat 4 Microhabitats 1, 2, and 4 contain the important host plant S patens and the nectar source L nashii, and these microhabitats coincide with the area in this marsh

where newly emerged adults were frequently captured (Webster 1994) Although the exact processes that generated the vegetation pattern and its relationship to the survival of

the maritime ringlets are still unclear, the abundance of S patens and L nashii seems to

be a fair indicator of good early-instar larval habitat for the maritime ringlets

I found microhabitats 3, 5, and 6 to be unsuitable as maritime ringlet habitats Although speculative, I present possible causes for the high mortality in those habitats The poor survival of larvae in microhabitats 3 and 5 may be due to frequent tidal

inundation Prolonged tidal submergence may have adverse effects on the survival of larvae (Joy & Pullin 1997), and the presence of fish and marine invertebrates (Pfeiffer & Wiegert 1981; Daiber 1982; Rozas & Zimmerman 2000) during submergence may have increased predation Vegetation structure (Dempster 1984) and nearby landscape

features (e.g a sand bar) (Daiber 1982; Long & Mason 1983) may influence the size of

the predator community, contributing to differential predation Although I found that microhabitat 3 is unfavorable, the same vegetation type in the southwestern part of the marsh, which experiences less frequent tidal inundation, had a high density of freshly emerged adults (Webster 1994)

The dominant vegetation in microhabitat 6, J gerardi, may not be eaten by the maritime ringlet Even if the maritime ringlet can use J gerardi, it seems likely to be a low-quality host, as it produces few new shoots in late summer, unlike S patens

(Bertness & Ellison 1987) Because it is infrequently inundated, microhabitat 6 may also support non-salt marsh specialist predators, which perhaps may result in predation

pressure higher than other microhabitats

I made two assumptions about early-instar larval movement in and out of the release sites The first assumption was that this migration rate is so slow that there is no movement across the clearing This first assumption seems to be met as I have often resighted larvae in exactly the same location within a site, or if they moved, the

displacement was a few centimeters at most over 3 or 4 days (pers obs.) Similarly,

previous studies of overwinter survival among bog-dwelling C tullia did not have

problem with larval movement (Joy & Pullin 1999) The second assumption is that the habitat quality does not affect their movement rate (i.e., they will not migrate to a high-quality habitat from a low-quality habitat) My second assumption would be violated if the larvae moved out of study sites, and it would overestimate d , but it would not affect ˆ

my conclusion that the habitat quality among the microhabitats differs significantly as perceived by the larvae At worst, I can regard d as a qualitative indicator of habitat ˆ

quality

As one of a few salt marsh butterfly species, the maritime ringlet is an

evolutionary phenomenon worth preserving Gaining knowledge of the specific survival of the maritime ringlet is necessary as none of their habitats are

microhabitat-permanently stable In the short term, the salt marsh vegetation is often disturbed by winter ice scour (Richard 1978; Roberts & Robertson 1986) and tidal erosion of the marsh itself (Jacobson & Jacobson 1987) Because salt marshes inhabited by the

maritime ringlets often occur near or within city limits, the butterflies are at risk of

habitat loss or habitat deterioration by urban and industrial sprawl and pollution (Roberts

& Robertson 1986; Adam 1990) In the long term, the vegetation may also change as the salt marsh hydrology changes due to a rise in sea level (Warren & Niering 1993) or vegetation competition patterns are altered from increased soil nitrogen level (Valiela et

al 1975; Levine et al 1998)

To conserve maritime ringlets successfully, we should recognize the importance and ephemerality of their microhabitats Microhabitats differing subtly in hydrology and vegetation have profound effects on early larval survivorship As salt marshes and microhabitats within them are not static in time and space, the preservation of this

subspecies may require relatively intensive management Because of the ringlet's low vagility and the disjunct nature of the extant salt marshes, they have a low probability of recolonizing an unoccupied but suitable salt marsh by themselves Due to this lack of metapopulation structure (Hanski & Simberloff 1997), local extinction is an extremely serious threat to the maritime ringlet

We need to provide aggressive stewardship to ensure the persistence of the maritime ringlet populations In addition to the adult population density monitoring now

being done, periodic monitoring of their habitats to assess the extent of short S patens stands with L nashii favorable to the early-larval survival (i.e., microhabitats 1, 2, and 4)

will afford more security in their management Although older larvae and adults may be

less sensitive to microhabitat differences, L nashii attracts adult maritime ringlets as a

nectar source The presence of nectar sources positively affects adult survival and

reproduction in California populations of C tullia (Weissman 1972), and maritime ringlet

females are known to nectar frequently as they age and deplete their fat reserves (Webster 1994) The extent of favorable microhabitat can be used to choose reintroduction sites and to assess habitat quality for extant populations The former will protect the maritime ringlet further from extinction, and it is essential in realizing downlisting or delisting The latter can be used to protect the extant populations The decline in favorable

microhabitat size predicts decline of the population, and the microhabitat change can be reversed if the cause is small in scale (e.g., vegetational change due to nitrogen loading) and the change is detected at an early stage

Table 2 The number of single and multiple release sites in microhabitats Only microhabitat 1 had two groups of release sites in separate areas (See Figure 2) Every multiple release site contained 10 larvae

Microhabitat Single Release Sites Multiple Release Sites

Table 3 Estimated daily death rate (d ) and finding rate (ˆ f ) along with the 2-unit support ˆ

limits (SL) from single release sites The 2-unit support limits can be considered

analogous to 95% confidence limits (Edwards 1992)

Table 4 The minimum number of larvae known to be alive (MNKA) in multiple release sites at the end of the observation I released ten larvae into each multiple release site The percentages of the larvae that had survived are also shown

Microhabitat Number of Multiple

0.10 0.15 0.20 0.25 -2.0

-1.5 -1.0 -0.5 0

d

lnLmax -lnL

Figure 3 The plot of daily death rate d vs standardized log likelihood values for the five

microhabitats within two-unit support limits