aquatic and terrestrial vegetation influence … · wisconsin ecologists. members of my graduate...

AQUATIC AND TERRESTRIAL VEGETATION INFLUENCE

LACUSTRINE DRAGONFLY (ORDER ODONATA)

ASSEMBLAGES AT MULTIPLE LIFE STAGES

By

Alysa J. Remsburg

A dissertation submitted in partial fulfillment of

the requirements for the degree of

Doctor of Philosophy

(Zoology)

at the

UNIVERSITY OF WISCONSIN – MADISON

2007

iACKNOWLEDGMENTS

Reflecting on the contributions of my colleagues and friends during my graduate

studies gives me a strong sense of gratitude for the community of support that I have

enjoyed. The people who surround and support me deserve more thanks than I can

describe here. Friends and family have supported my graduate studies by generously

accommodating my tight schedule and warmly offering encouragement throughout the

process.

Monica Turner guided my graduate studies in numerous ways. It was her trust in

my abilities and willingness to learn about a new study organism that first made this

research possible. She encouraged me to pursue the research questions that most

interested and inspired me, although it meant charting territory that was new to both of us.

Monica served as the ideal mentor for me by requiring clear communication, modeling an

efficient and balanced work ethic, providing critical reviews, and listening

compassionately.

This research benefited from the expertise and generosity of outstanding

Wisconsin ecologists. Members of my graduate research committee, Steve Carpenter,

Claudio Gratton, Tony Ives, Bobbi Peckarsky, and Joy Zedler, all offered useful

suggestions and critiques on experimental design, pressing research questions, and the

manuscripts. Cecile Ane provided additional statistical advice and smiles. Bill Smith,

Bob DuBois, and Robert Bohanan answered (or reassured me that I should try to answer)

many questions about field methods, Odonata biology, and species identification.

iiThe kindness of my colleagues encouraged me to ask frequently for help or

second opinions. The graduate students and post-doctoral researchers working with

Monica have been great friends, reviewers, trouble-shooters, cooks, and sharers of table

space. Coworkers at the Center for Limnology’s Trout Lake research station introduced

me to aquatic communities and taught me how to launch a boat. Trout Lake station

graduate students and staff not only shared their field equipment, data, and study sites,

but also lured me away from the bug-picking table when I needed breaks.

My decision to research odonates was made, in part, because I hoped it would

attract and enable many undergraduate students to participate in the research. Neither the

odonates nor the students have let me down. I enjoyed and benefited from working in the

field with Jasmine Batten, Bridget Henning, Melissa Theis, and Phil Winkler. Students

who diligently counted labial palp setae in the lab with me for species identification were

Josh Estreen, Josie Iacarella, Debbie Hong, Anh Duong, and Cecilia Leugers. I thank

each of them for helping to investigate the nitty-gritty details that underlie much of

ecological research.

Anders Olson, my outstanding field assistant and coauthor for the South Africa research,

deserves many thanks. Anders helped design and conduct the field experiments on hot

South African days (including re-hanging shade cloth after the ‘Cape Doctor’ blew

through, fixing flat bike tires, and countless other efforts). On all of my research chapters,

he offered detailed reviews and patiently listened to practice presentations. In addition to

ecological contributions, the moral support and understanding he provided have

ultimately helped me to reach this stage with peace of mind. We are blessed to have each

other as coauthors in life.

iii

TABLE OF CONTENTS

Abstract of dissertation vi

I. Dissertation introduction 1

Literature Cited 4

II. Aquatic and terrestrial drivers of dragonfly (order Odonata)

assemblages within and among north-temperate lakes 7

Abstract 7

Introduction 8

Methods

Study Area 12

Aquatic and terrestrial drivers of Odonata larval density 13

Effects of riparian vegetation on adult abundance 18

Macrophyte effects on larval densities within sites 20

Results

Aquatic and terrestrial drivers of Odonata larval density 22

Effects of riparian vegetation on adult abundance 24

Macrophyte effects on larval densities within sites 25

Discussion 26

Acknowledgments 31

Literature cited 32

Tables 42

Figures 54

iv Appendix 63

III. Gomphidae densities among three life stages: Oviposition site

selection overrides post-recruitment spatial variation 66

Abstract 66

Introduction 67

Methods

Study sites 72

Odonata surveys 72

Habitat variables 74

Data analysis 75

Results 78

Discussion 79

Acknowledgments 85

Literature Cited 86

Tables 95

Figures 101

Appendix 107

IV. Shade alone from invasive riparian trees reduces dragonfly

abundance 108

Abstract 108

Introduction 109

Methods

Study area 111

v Field methods 112

Statistical methods 114

Results 116

Discussion 117

Acknowledgments 119

Literature Cited 120

Figures 125

viAQUATIC AND TERRESTRIAL VEGETATION INFLUENCE LACUSTRINE

DRAGONFLY (ORDER ODONATA) ASSEMBLAGES AT MULTIPLE

LIFE STAGES

Alysa J. Remsburg

Under the supervision of Professor Monica G. Turner

At the University of Wisconsin-Madison

Abstract

Understanding how animals respond to habitat structure is a fundamental

objective in ecology, but is particularly challenging when the animals require distinct

habitats for different life stages. Although the majority of animals have spatially

segregated life stages, research on habitat associations has generally been restricted to

only one of the life stages. The relative importance of aquatic and terrestrial habitat

structure is not well known for the order Odonata (dragonflies and damselflies).

In northern Wisconsin (USA) lakes, housing development contributes to

heterogeneity in riparian and littoral vegetation structure. I surveyed odonate larval

assemblages at 41 sites across 17 lakes. Based on mixed-effects multiple regressions,

model selection identified site-level littoral macrophyte abundance as a key driver of

larval odonate species richness, and riparian wetland plant abundance as the best

predictor for odonate density. Subsequent field experiments on larval predation and adult

site selection helped explain these patterns. Additional surveys of the most abundant

family (Gomphidae) at 22 lake sites indicated that local larval densities depend most on

viirecruitment, which I estimated from adult densities during the previous year. Densities of

emergent Gomphidae skins (exuviae) were most related to densities of the later-instar

(second-year) larvae, further suggesting that larval survivorship and movement are less

variable spatially than recruitment from the previous life stage.

Field experiments conducted at two South African lakes demonstrated how

riparian tree structures alter adult odonate abundances. Riparian shade reduced the

abundance of odonates at these potential breeding sites. Perch structures, added to

separate experimental plots, supported locally higher adult abundances, but dragonflies

were not sensitive to perch structure density or diversity. Thus shade is the critical

habitat component that should be addressed for odonate conservation in South Africa.

Collectively, this research describes the role of habitat structure during multiple

life stages. Field experiments demonstrate that generalist predators are sensitive to

vegetation structure. The results suggest that riparian habitat selection by animals with

complex life cycles can influence aquatic communities.

1CHAPTER I

Dissertation introduction

Understanding how habitat structure influences animal distributions is an

important challenge faced by both ecologists and land managers. Natural and

anthropogenic disturbances frequently alter vegetation physical structure. Fires, pest

outbreaks, and land-use changes all alter foliage cover. By contrast, invasive plant

species can add dense foliage layers having different architectures from native plant

communities. Vegetation structure directly affects animal densities by controlling

physical habitat attributes: microclimate, visibility, and permeability. Identifying which

habitat components are most infuential for animal diversity and density can facilitate

effective habitat conservation or restoration.

Researchers have long investigated how vegetation structure affects animal

diversity. Several empirical studies suggest that species richness of arthropods (Lawton

1983, Gardner et al. 1995, Halaj et al. 2000, Langellotto and Denno 2004), birds

(MacArthur and MacArthur 1961, Karr and Roth 1971), rodents (M'Closkey 1976), and

lizards (Pianka 1967) increase with greater vegetation complexity. From a theoretical

perspective, increasing diversity of habitat structure was one of several hypotheses

proposed for the accumulation of more species with increased sampling area (Connor and

McCoy 1979, McGuinness 1984).

Vegetation structure can influence trophic interactions of Odonata species

(dragonflies and damselflies) in water or on land (e.g., Crowder and Cooper 1982,

Wissinger 1988, Diehl 1992). The > 5000 Odonata species worldwide each have a

2unique spatial, temporal, and/or behavioral niche, but all members of this order are

generalist predators. Adults and larvae both detect prey visually, although larvae also

rely somewhat on mechanoreceptors (Gaino and Rebora 2001, Rebora et al. 2004, Olberg

et al. 2005). Many larvae fall prey to fish or other odonates, including conspecifics

(Wright 1946, Kennedy 1950, Larochelle 1977, Crowder and Cooper 1982, Johnson 1991,

Momot 1995, Stoks and McPeek 2003). Emerging larvae are often consumed at the

water’s edge by birds (Orians and Horn 1969, Wissinger 1988). Predators of adult

odonates include larger odonates, frogs (particularly for ovipositing females), web-

building spiders (especially on weaker, young damselflies), and the most agile bird

species (Corbet 1999).

Considerable research on the biology and behavior of odonates contributes to

their utility as model organisms for animals with complex life cycles (Corbet 1999), but

knowledge of their habitat requirements is lacking. Odonates and other animals with

spatially separated life stages may be sensitive to habitat features in either of the

ecosystems inhabited during their life history. Although 66-96 % of the lifespan of

temperate-range Odonata species is spent in the aquatic larval stage, the terrestrial adult

stage is critical for reproduction and habitat selection. Several studies on odonates have

investigated how habitat features in either their aquatic or terrestrial life stages can

influence adults (McKinnon and May 1994, Carchini et al. 2003, Foote and Hornung

2005, Ward and Mill 2005) or larvae (Diehl 1992, Sahlen 1999, Burcher and Smock

2002), but the relative importance of aquatic and terrestrial habitat structure has not been

investigated. Only a few previous authors (Meek and Herman 1991, Wildermuth 1993,

3De Marco and Resende 2004) have experimentally tested Odonata oviposition site

choices, leaving many questions about the role of riparian land cover unanswered.

In the first research chapter that follows, I address how odonate larval

assemblages vary in response to aquatic and terrestrial habitat features in and around

northern Wisconsin (USA) lakes (Chapter II, with M.G. Turner). The study was

designed to investigate the relative importance of both natural and anthropogenic drivers

of odonate assemblages. Following a survey of odonate larvae among and within 17

lakes, I conducted two field experiments to investigate why riparian wetland vegetation

and littoral macrophytes were associated with the highest density and diversity of

odonates, respectively. This research describes the role of vegetation structure during

multiple life stages that span the aquatic-terrestrial ecotone.

The next research chapter focuses on population dynamics of the odonate family

that was most abundant in northern Wisconsin lakes, Gomphidae. Using surveys of three

dragonfly life stages that I conducted in 2005 and 2006, I incorporate both recruitment

from previous life stages and habitat parameters into predictive models. This research

demonstrates how population demographics from one life stage carry over to subsequent

stages and generations, even when they occur in separate ecosystems.

In the final chapter, I test how invasive riparian trees in South African lakes affect

adult dragonfly riparian site use (Chapter IV, with A.C. Olson and M.J. Samways).

Through field experimentation, I isolated the effects of riparian shade and understory

perch structures on dragonfly abundance. The results can inform conservation efforts for

several odonate species that have been critically imperiled by the invasive trees. This

4study also contributes to the small but important body of literature investigating how

generalist predators use vegetation structure.

Literature Cited

Burcher, C. L., and L. A. Smock. 2002. Habitat distribution, dietary composition and life

history characteristics of odonate nymphs in a blackwater coastal plain stream. American Midland Naturalist 148:75-89.

Carchini, G., M. Di Domenico, T. Pacione, A. G. Solimini, and C. Tanzilli. 2003. Species distribution and habitat features in lentic Odonata. Italian Journal of Zoology 70:39-46.

Connor, E. F., and E. D. McCoy. 1979. Statistics and Biology of the Species-Area Relationship. American Naturalist 113:791-833.

Corbet, P. S. 1999. Dragonflies: Behavior and Ecology of Odonata. Cornell University Press, Ithaca, NY.

Crowder, L. B., and W. E. Cooper. 1982. Habitat structural complexity and the interaction between bluegills and their prey. Ecology 63:1802-1813.

De Marco, P., and D. C. Resende. 2004. Cues for territory choice in two tropical dragonflies. Neotropical Entomology 33:397-401.

Diehl, S. 1992. Fish predation and benthic community structure - the role of omnivory and habitat complexity. Ecology 73:1646-1661.

Foote, A. L., and C. L. R. Hornung. 2005. Odonates as biological indicators of grazing effects on Canadian prairie wetlands. Ecological Entomology 30:273-283.

Gaino, E., and M. Rebora. 2001. Apical antennal sensilla in nymphs of Libellula depressa (Odonata : Libellulidae). Invertebrate Biology 120:162-169.

Gardner, S. M., M. R. Cabido, G. R. Valladares, and S. Diaz. 1995. The Influence of Habitat Structure on Arthropod Diversity in Argentine Semiarid Chaco Forest. Journal of Vegetation Science 6:349-356.

Halaj, J., D. W. Ross, and A. R. Moldenke. 2000. Importance of habitat structure to the arthropod food-web in Douglas-fir canopies. Oikos 90:139-152.

Johnson, D. M. 1991. Behavioral ecology of larval dragonflies and damselflies. Trends in Ecology & Evolution 6:8-13.

5Karr, J. R., and R. R. Roth. 1971. Vegetation structure and avian diversity in several New

World areas. American Naturalist 105:423-&.

Kennedy, C. H. 1950. The relation of dragonfly-eating birds to their prey. Ecological Monographs 20:103-143.

Langellotto, G. A., and R. F. Denno. 2004. Responses of invertebrate natural enemies to complex-structured habitats: a meta-analytical synthesis. Oecologia 139:1-10.

Larochelle, A. 1977. Les amphibiens et reptiles comee predateurs des odonates (premiere note bibliographique). Cordulia 3:81-84, reviewed in Corbet 1999

Lawton, J. H. 1983. Plant Architecture and the Diversity of Phytophagous Insects. Annual Review of Entomology 28:23-39.

M'Closkey, R. T. 1976. Community Structure in Sympatric Rodents. Ecology 57:728-739.

MacArthur, R. H., and J. W. MacArthur. 1961. On bird species diversity. Ecology 42:594-598.

McGuinness, K. A. 1984. Equations and Explanations in the Study of Species Area Curves. Biological Reviews of the Cambridge Philosophical Society 59:423-440.

McKinnon, B., and M. L. May. 1994. Mating habitat choice and reproductive success of Pachydiplax longipennis (Burmeister) (Anisoptera: Libellulidae). Advances in Odonatology 6:59-77.

Meek, S. B., and T. B. Herman. 1991. The influence of oviposition resources on the dispersion and behavior of calopterygid damselflies. Canadian Journal of Zoology-Revue Canadienne De Zoologie 69:835-839.

Momot, W. T. 1995. Redefining the role of crayfish in aquatic systems. Reviews in Fisheries Science 3:33-63.

Olberg, R. M., A. H. Worthington, J. L. Fox, C. E. Bessette, and M. P. Loosemore. 2005. Prey size selection and distance estimation in foraging adult dragonflies. Journal of Comparative Physiology a-Neuroethology Sensory Neural and Behavioral Physiology 191:791-797.

Orians, G. H., and H. S. Horn. 1969. Overlap in foods and foraging of four species of blackbirds in the potholes of central Washington. Ecology 50:930-938.

Pianka, E. R. 1967. Lizard species diversity. Ecology 48:333-351.

Rebora, M., S. Piersanti, and E. Gaino. 2004. Visual and mechanical cues used for prey detection by the larva of Libellula depressa (Odonata Libellulidae). Ethology Ecology & Evolution 16:133-144.

6Sahlen, G. 1999. The impact of forestry on dragonfly diversity in central Sweden.

International Journal of Odonatology 2:177-186.

Stoks, R., and M. A. McPeek. 2003. Predators and life histories shape Lestes damselfly assemblages along a freshwater habitat gradient. Ecology 84:1576-1587.

Ward, L., and P. J. Mill. 2005. Habitat factors influencing the presence of adult Calopteryx splendens (Odonata : Zygoptera). European Journal of Entomology 102:47-51.

Wildermuth, H. 1993. Habitat selection and oviposition site recognition by the dragonfly Aeshna juncea (L.): an experimental approach in natural habitats (Anisoptera: Aeshnidae). Odonatologica 22:27-44.

Wissinger, S. A. 1988. Spatial-distribution, life-history and estimates of survivorship in a 14-species assemblage of larval dragonflies (Odonata, Anisoptera). Freshwater Biology 20:329-340.

Wright, M. 1946. The economic importance of dragonflies (Odonata). Journal of the Tennessee Academy of Science 21:60-71.

7CHAPTER II

Aquatic and terrestrial drivers of dragonfly (order Odonata) assemblages within

and among north-temperate lakes1

Abstract

The physical structure of vegetation influences animal interactions, movement,

and thermoregulation. Vegetation structure may be a good indicator of the habitat

requirements of generalist predators such as dragonflies and damselflies (order Odonata).

Odonates utilize both aquatic and terrestrial habitats during larval and adult life stages,

respectively, but the relative importance of vegetation in each of these habitats remained

untested prior to this study. We investigated how several aquatic and riparian habitat

variables affect odonate larvae from 41 sites (each 30 m in shoreline length) on 17 lakes

in northern Wisconsin (USA). We used principal components analyses to reduce

multiple habitat variables to two lake-level axes (lake size and amount of adjacent

wetland), two site-level littoral axes (macrophyte abundance and substrate type), and two

site-level riparian axes (foliage height diversity and length of wetland vegetation).

Multiple linear regression indicated that density of the most abundant odonate family,

Gomphidae, was positively related to the length of shorelines with tall, herbaceous

wetland plants adjacent to the water (adjusted R2 = 0.48), whereas species richness was

positively correlated with abundance of littoral macrophytes (adjusted R2 = 0.34).

To investigate whether riparian wetland plants affected larval densities via

recruitment from the adult stage, we compared adult odonate behaviors in response to

1 Coauthor Monica G. Turner

8two riparian vegetation treatments. Damselfly abundance was higher where we placed

potted wetland plants than at manicured lawns without tall vegetation.

Within littoral sites, we confirmed that larvae from the clasper and sprawler

behavioral guilds were most abundant in macrophyte patches. However, when we

tethered larvae in and outside of macrophyte patches to compare predation risk, we found

no difference. Our key findings indicate that odonate larvae may be influenced by

vegetation structure in both the aquatic and adjacent riparian habitats, demonstrating how

animals with complex life histories link aquatic and terrestrial communities.

Introduction

Understanding how vegetation structure affects animal abundance and diversity

has long been a fundamental objective in ecology (e.g., MacArthur 1965, Karr and Roth

1971). The physical structure of vegetation influences predator-prey interactions,

thermoregulation, landscape permeability (ease of movement), and mate detection.

Vegetation structure can describe habitat suitability for some predacious species better

than more time-intensive habitat descriptions such as plant species composition (e.g.,

Pianka 1967, Brose 2003, De Souza and Martins 2005, Muntifering et al. 2006). While

early studies of habitat structure focused on terrestrial systems, more recent work has

demonstrated that animals in aquatic and marine ecosystems also depend on physical

structure. For example, fish and invertebrate communities exhibit greater diversity and

abundance in habitats with structure provided by coral, macrophytes, or fallen trees (e.g.,

Werner et al. 1983, Carr 1991, Tupper and Boutilier 1995, Cheruvelil et al. 2002, Sass

2004, Milner and Gloyne-Phillips 2005).

9Animals that use multiple habitats for different behaviors or life stages may

respond to vegetation structure in each of the habitats they occupy. A few studies have

compared the importance of aquatic and terrestrial habitat features for animals that

inhabit both of these ecosystems, although most of these focused on amphibians (Marnell

1998, Scribner et al. 2001, Campbell et al. 2004, Ficetola et al. 2004). Research

investigating how Odonata (dragonfly and damselfly) species respond to both aquatic and

terrestrial habitat structure is lacking. The long aquatic phase (one to four years) of many

odonates in temperate regions may increase the importance of aquatic habitat features for

odonates relative to amphibians and other aquatic insects with shorter aquatic phases.

Land-water ecotones are often structurally complex and provide critical habitat

for many animals (Decamps et al. 2004). Structural complexity arises from the

combination of abiotic characteristics (e.g., substrate, slope), littoral vegetation, and

riparian vegetation. Human activity also interacts with natural drivers of structural

complexity. Housing development is expanding rapidly in many riparian areas (Burger

2000, Hansen et al. 2005), including the upper Midwest, USA (Schnaiberg et al. 2002,

Gonzalez-Abraham et al. 2007), where it leads to simplification of vegetation structure by

removal of the shrub layer, floating macrophytes, and coarse wood. These structural

changes may have consequences for a variety of animals in both the riparian and littoral

(near-shore aquatic) zones (Bryan and Scarnecchia 1992, Schindler et al. 2000, Lindsay

et al. 2002, Woodford and Meyer 2003, Henning and Remsburg in review). The relative

influence of natural and anthropogenic drivers on Odonata communities is not well

known.

10Odonates require both aquatic and terrestrial habitats during their complex life

cycle. However, the role of the terrestrial habitat in determining odonate abundance and

diversity remains unclear. Odonata species generally show little response to particular

plant species (terrestrial or aquatic), but diversity and abundances of many species

correlate positively with local vegetation structure (Buchwald 1992, Foote and Hornung

2005). Shoreline vegetation may provide adults with important perching structures for

thermoregulation, foraging, territory defense, mate attraction, copulation, nocturnal

roosting, or protection from adverse weather (Buchwald 1992, Wildermuth 1993,

McKinnon and May 1994, Rouquette and Thompson 2007). Several researchers in pond

and river ecosystems have documented lower odonate diversity and density at sites with

reduced riparian vegetation structure, although these studies focused mostly on the adult

stage (Table 1). To our knowledge, effects of lakeshore vegetation on Odonata larval

assemblages have remained untested. If community controls in lakes behave similarly to

other systems, we would expect lower odonate diversity and density at sites with fewer

vegetation layers and plant types in the riparian zone. Odonate populations in lakes,

however, could be controlled more by fish predation than those in pond and river systems

if fish densities in littoral zones are higher in lakes.

Vegetation and other physical structures can interact with top-down controls on

odonate larvae or on odonate tenerals (immature adults that have recently emerged from

the water) to influence population sizes. Riparian vegetation, for example, may harbor

both emerging odonates and their predators (Duffy 1994, Whitaker et al. 2000). In the

water, mesocosm experiments have shown that odonate larvae find refuge from predators

in aquatic macrophytes (Crowder and Cooper 1982, Thompson 1987, Dionne et al. 1990,

11Diehl 1992). However, unlike in enclosed experiments, local predator densities in lakes

also tend to increase where macrophytes offer protection or higher prey densities (Werner

et al. 1983, Weaver et al. 1997, Gamboa-Perez and Schmitter-Soto 1999). Littoral coarse

wood or mucky substrates may also provide refugia from predators for some larval

odonate species (Schmude 1998, Alzmann et al. 1999, Burcher and Smock 2002). In

addition to these drivers of local habitat heterogeneity, predation on odonates can vary

among water bodies as a result of water chemistry or fish communities (Johnson and

Crowley 1980, McPeek 1990, Johansson and Brodin 2003).

We surveyed odonates across a wide range of lake and site conditions to address

the following research question: (1) How do odonate abundance and community

composition vary with habitat structure both among and within lakes? We measured

habitat variables that could directly or indirectly affect Odonata mortality,

thermoregulation, perching, territory selection, or emergence. Among lakes, we expected

that odonate abundance and species richness would decline with increasing fish

abundance, especially the abundance of insectivorous fish; lake depth; and water clarity

(Aksnes and Giske 1993, Davies-Colley and Smith 2001). However, we also

hypothesized that lakes with a greater perimeter-to-area ratio and morphometry index

might be associated with more bays and shallow littoral zones that would provide

odonates with more littoral oviposition areas. Because lakes with a higher specific

conductance correlate with higher primary productivity and invertebrate density, these

lakes were also expected to support more odonates (Hrabik et al. 2005). We anticipated

that odonate density and richness would correlate negatively with lake measures of

development because residential development corresponds with reduced littoral and

12riparian habitat complexity. At the site scale (within lakes), we expected greater odonate

density and richness at sites having more abundant riparian understory vegetation, greater

macrophyte abundance, lower slopes, and wetland vegetation.

Isolating the processes that underlie patterns in community ecology calls for

experimental research (Agrawal et al. 2007). We conducted field experiments designed

to explore the potential mechanisms underpinning the relationships we observed in our

comparative study. These experiments addressed two additional research questions: (2)

Does riparian wetland vegetation influence the use of riparian sites by adult dragonflies?

Because adult odonates may use riparian vegetation as cues of littoral quality or perch on

these structures, we expected that riparian sites with tall wetland plants added would have

higher odonate abundances and oviposition occurrences than riparian lawn sites lacking

tall vegetation. Our last question was (3) Does macrophyte density affect predation on

odonate larvae in lakes? Variation in predation rates is one factor that could result in

spatially variable odonate densities. We anticipated that predation might decrease with

macrophyte density because the aquatic vegetation would reduce visual detection of

larval odonates by fish.

Methods

Study area

We conducted this study in Vilas County in northeastern Wisconsin, USA, in the

center of the Northern Highlands Lake District (see Peterson et al. (2003) for details on

NHLD). Curtis (1959) classified the vegetation as northern wet forest and boreal forest.

The forested landscape is interrupted by a high density of glacial lakes (over 1300 lakes

13in the 2639 km2 county), which cover 13 % of the NHLD surface area (Black et al. 1963).

Lakeshores in the area have experienced unprecedented lakeshore housing development

(Schnaiberg et al. 2002, Gonzalez-Abraham et al. 2007), thereby altering spatial

heterogeneity of habitat components in the lake-land ecotone.

For the first research question, we surveyed larval Odonata assemblages at 41

lake sites (Appendix 1), and collected lake- and site-level environmental information.

We defined sites as 30-m segments along the shoreline contour, including the adjacent

littoral zone to a water depth of 1 m and the riparian zone 6 m inland from the ordinary

high water mark. Sites were distributed among 17 lakes in Vilas County, Wisconsin

(Table 2); the number of sites per lake ranged from one to five. Of sites on the same

lakes, we selected sites with qualitative differences in littoral or riparian vegetation

structure. Because housing development contributed to the riparian vegetation

heterogeneity, we selected half of the sites (21) from developed areas and the rest from

undeveloped, forested areas. All sites were separated from each other by at least 120 m

(with only two pairs of sites < 500 m apart).

Aquatic and terrestrial drivers of Odonata larval density and richness

At each of the 41 sites, we collected odonate larvae samples with a D-frame net to

scoop benthic material along four 16-m contour transects that paralleled the shore. We

centered transects within the 30 m site length and placed them at 0.25-, 0.5-, 0.75-, and

1.0-m water depths; transect distance from shore varied with littoral slope at each site.

On one day per site during June – July of 2004, we performed 20 benthic scoops (4 in

14each of the two shallower transects and 6 in each of the two deeper transects) covering 12

m2 at each site. Despite differences in emergence dates, the semivoltine life history of

most odonates in northern Wisconsin lakes (except some Zygoptera and smaller

Libellulidae, R. DuBois, personal communication) suggests presence of most larval

populations throughout the year at our sites. Larvae were sorted from benthic material

by hand, preserved in 70% ethanol, and identified to species or family (for suborder

Zygoptera) using the keys of Needham et al. (2000) and Westfall and May (1996).

To characterize lake-level fish abundance, human activity, vegetation, and water

chemistry, we considered 12 variables for each of the 17 lakes in this study: catch-per-

unit-effort (CPUE) of total fish and insectivorous fish (based on eight nights of boom-

style electrofishing, 24 minnow nets, and 24 crayfish traps per lake in 2001, 2002, or

2003), number of buildings within 100 m of the lake, density of buildings (number of

buildings within 100 m divided by lake perimeter), maximum depth, perimeter, area,

specific conductance (dissolved ionic strength), secchi depth (water clarity),

morphometry (a dimensionless unit: lake perimeter divided by the perimeter of a circle

with the same area as the lake), length of lake perimeter in wetlands, and percent of

perimeter in wetlands (Table 3). S.R. Carpenter and colleagues (2006) compiled all of

the lake-level data from 2001-2004.

The nine littoral habitat variables recorded at each site were: slope to 1 m depth;

number of submerged logs > 5 cm in diameter (coarse wood) at 0.5 m depth (Marburg et

al. 2006); sand, rock, gravel, and muck substrate dominance categories at 0.5 m depth

(Marburg et al. 2006); mean macrophyte species richness, percent cover, and frequency

of emergents (Table 3). Macrophyte species and percent cover (in the littoral zone out to

152 m depth) were recorded along one transect perpendicular to the shoreline at the center

of each site (Alexander 2005).

We recorded 10 riparian variables for each site: slope, canopy cover category

(sparse, moderate, or full), length of site with shrubs or saplings (of maximum height

0.75 – 2 m) present, length of site with wetland herbaceous plants > 0.5 m tall (Carex,

Typha, and/or Iris species; hereafter termed ‘tall wetland plants’), minimum distance of

vegetation from water’s edge, mean structural diversity, mean foliage height diversity,

maximum understory plant height (up to 2.5 m), and percent cover of both shrubs and

herbaceous plants within 0.5 m of the ordinary high water mark (Table 3). We estimated

mean structural diversity and mean foliage height diversity from a point-intercept method

(Mueller-Dombois and Ellenberg 1974) at 18 random locations per site where we held

2.5 m-tall poles (2.5 cm in diameter) perpendicular to the ground. For each pole, we

calculated structural diversity by summing the number of different structural categories

(out of 11: trunk > 3 cm diameter, woody stem, dead woody steam, herbaceous stem,

moss, fungus, rock, wood chips, foliage or flower, permanent human structure, other non-

natural object ) intersected by the pole. We defined foliage height diversity as the

number of vertical layers (out of 10) on the pole intersected by plants. All vegetation

data other than shoreline percent cover variables refer to the area within 6 m of the

ordinary high water mark but outside of the water.

We used mixed-effects multiple linear regression models to compare the effects

of lake- and site-level predictors on total odonate density and richness (both square-root

transformed to meet normality assumptions) across the 41 sites. We also constructed

mixed-effects linear regressions to evaluate predictors of densities for the most common

16larval family, Gomphidae. Because all odonate families other than Gomphidae were too

rare to be distributed normally, we used mixed-effects logistic models, including a fixed-

effects term for area sampled, to compare habitat effects on presence or absence of the

Zygoptera (damselfly) suborder and the four other Anisoptera (dragonfly) families

observed. For linear and logistic models, we assigned lake as a random-effect term

because effects of specific lakes were not of interest, yet sites on the same lakes were not

independent of each other. Analyses were completed using R v.2.4.1 software (R Core

Development Team, Vienna, Austria) including the nlme package for mixed-effects

models.

To reduce the 12 lake, 9 littoral, and 10 riparian predictor variables to a smaller

set of orthogonal variables, we conducted separate principal components analyses (PCA,

using PC-ORD 4 software, Gleneden Beach, OR) on each of these variable categories.

Because many lake, littoral, and riparian variables correlated strongly within the three

categories, conducting three separate PCA’s enabled us to maximize the variation in site

predictors summarized by the PCA scores. PCA’s were based on centered, standardized

matrices of Pearson correlation coefficients. Pair-wise scatterplots within the three

categories indicated continuous distributions and linear relationships among variables.

After examining PCA eigenvalues and habitat variable correlations with the PCA scores

(Table 3), we chose to use the first two PCA axes from each category of variables in the

mixed-effects multiple linear regressions of larval densities. Scores from the three

separate PCA processes were generally not correlated with each other (Pearson

correlation coefficients < 0.3), except for the first lake axis and first riparian axis (r = -

0.41).

17We used an information theoretic approach to compare alternative linear models

containing zero to six of the six PCA scores. Unlike hypothesis testing, model selection

using information theory does not require use of arbitrary cutoff values (such as α =

0.05), nor does it involve the Bonferroni problems of multiple tests (Burnham and

Anderson 2002). The 64 candidate models (including permutations for all possible

combinations) were ranked according to second-order criterion Akaike’s Information

Criterion (AICc) values (Sugiura 1978). AICc has a bias correction factor to help prevent

over-parameterized models when sample sizes are small. Selection of the “best”

predictive models based on AICc is more closely related to the sample size than use of

hypothesis testing or Bayseian information criterion (Burnham and Anderson 2002).

Relative empirical support for a model i is indicated by ∆i, the difference between AICc

for model i and the smallest AICc of the candidate models. Strength of evidence for each

of the candidate models can be compared using Akaike weights, wi, the log likelihood of

model i divided by the sum of log likelihoods for all candidate models (Burnham and

Anderson 2002). Information theory suggests that a dataset may equivalently support

multiple predictive models (particularly if ∆i <2). We used all of the candidate models to

estimate the relative importance of each predictor by summing the Akaike weights (∑w)

over all models in which a variable appeared (Burnham and Anderson 2002).

We present adjusted R2 values for linear regressions, but the maximum likelihood

model fitting method employed for our mixed-effects models could not be used to

calculate the R2 values. We therefore calculated the adjusted R2 values from fixed-effects

models (with the parameters selected from mixed-effects models) where lake was

included as a fixed effect. We also used lake as a fixed effect to estimate pseudo-R2

18values (one minus the ratio of the model’s log-likelihood to the null model’s log-

likelihood) for the logistic regression models (Menard 2000).

Effects of riparian vegetation on adult abundance

To investigate whether adult site use helps explain larval distributions within

lakes, we recorded adult Odonata behavioral responses to riparian vegetation treatments

at two location blocks. One block had very dense aquatic macrophytes (‘dense-

macrophyte block’), while the other block had very few aquatic macrophytes (‘sparse-

macrophyte block’). Both blocks were on bays in the same lake (Little Saint Germain),

had stone seawalls separating trimmed lawns from the water, similar aspect, and riparian

slopes near zero.

The two riparian vegetation treatments applied to each block were: trimmed lawns

with 20 dwarf cattail (Typha minima) plants in pots added at the shoreline (‘Typha’

plots), and trimmed lawns without potted or other tall plants (‘no Typha’ plots). We

chose the dwarf cattails, which were 0.6 m in height, because they were readily available

at nurseries and closely resembled the height and rigidity of native sedges. Ten meters

separated the two treatment plots; each plot spanned 15 m along the shoreline and 2 m

inland. We replicated both treatments on 21 days at the dense macrophyte block and on

19 days at the sparse macrophyte block. Although this design may have allowed repeated

observations of some individuals, we believe that separate observation days could be

treated as experimental replicates because adult odonates move away from the shoreline

to roosting sites each evening (Askew 1982, Bried and Ervin 2006). Most of the genera

we observed probably do not demonstrate site attachment across multiple days (Jacobs

191955, Logan 1971). At both blocks, placement of the two treatments alternated by day

between the two plots. We thereby separated Typha treatment effects from littoral or

location-specific influences.

On each observation day, we set up the Typha treatments at least three hours prior

to recording odonate behaviors at both treatments within the same block. In order to

capture the period of greatest odonate activity, observations took place from 25 June to

18 August 2006, between 12:00 and 16:00, on days with relatively low wind and cloud

cover. A single observer (AJR) stood at one corner of the plot in order to survey

odonates in the whole plot simultaneously, including those over the water adjacent to the

riparian plot (within approximately 5 m of shore). To improve detection of the less active

odonates perched on shore, the observer also walked slowly across the plot two times per

observation period. We recorded the total number of individuals present and whether or

not each genus oviposited within the plot during ten minutes of observation. Using

binoculars to verify species would have precluded comprehensive observations across the

whole plot. The observer estimated relative odonate abundance within the plot by

counting all individuals visible at the instant when the ten-minute timer sounded.

We conducted blocked mixed-effects ANOVAs to compare Zygoptera and

Anisoptera abundance estimates among treatments, where observation day was treated as

a random effect. We pooled responses by suborder because odonate abundances were

low overall, and individuals within the same suborder seemed to behave similarly. We

tested effects of the Typha treatment on oviposition location choice only at the dense-

macrophyte block (where oviposition frequency was highest), using McNemar’s Chi-

squared tests (Argesti 2002). McNemar’s test compared the number of days when

20oviposition was observed at the Typha treatment but not at the other treatment with the

number of days that oviposition was observed at the no-Typha treatment but not at the

Typha treatment.

Macrophyte effects on larval densities within sites

To evaluate macrophyte effects on odonate larval densities at a finer scale (5 – 10

m2) than the site-level surveys, we sampled larvae at 18 microsites from nine lakes (Table

3) in July 2006. We chose nine microsites that had abundant submerged macrophytes

and nine adjacent microsites that completely lacked macrophytes. The macrophyte and

non-macrophyte microsites, paired by lake, were separated by 4 to 15 m (except for one

pair separated by 40 m across a bay). Paired microsites had similar substrates and

distances from structures such as docks. We used a D-net to collect benthos enclosed by

a cylinder (i.e., trash can with the bottom cut off) placed at five adjacent subsamples at

0.3 m depth and five subsamples at 0.5 m depth. The cylinder protocol ensured that

larvae within each subsample could not move out of the area during sampling because it

extended from the substrate to the water’s surface. By pooling the 10 subsamples, we

sampled all larvae in 7 m2 of benthos at each microsite.

We compared densities of three larval behavior guilds among the 18 microsites

with and without macrophytes using pairwise Wilcoxon tests. Behavioral guilds

provided more informative response variables than families for this experiment because

densities of some families were very low within the (relatively small) microsite sample

areas. Of the behavioral guilds, burrowers hide amid a layer of detritus or under the top

21layer of sand, while sprawlers lie flat on detritus, and claspers climb on the leaves, stems,

and roots of macrophytes (Table 4).

In August 2006 we conducted a predation experiment at 42 microsites from 21

lakes (including the 18 microsites discussed above; Table 3). Microsites were paired by

lake as in the surveys above. Macrophyte structures at the microsites varied among lakes,

so we recorded macrophyte species and one of four macrophyte biomass categories

(none, low, moderate, or high) for each microsite based on visual estimates.

At each microsite, we used Superglue® to attach monofilament fishing lines 30-40

cm in length onto the abdomens of eight live larvae (in later instars) collected from

nearby lakes or ponds. We used a variety of Anisoptera species and larval instars,

although most larvae were Cordulia shurtleffi 2 cm in length. We controlled for

variations in larval behavior and predator appeal among different odonate species by

matching larval species and sizes (range: 15 – 30 cm length) for paired microsites within

the same lake. Laboratory observations indicated that fish predation occurred on tethered

larvae, resulting in empty lines where predation had occurred. We also verified in the

laboratory that larvae never became unglued from the fishing lines. These tethered larvae

were attached just above the microsite benthos by tying fishing lines to submerged metal

stakes. We retrieved tethered larvae remaining after 3.5 - 6 hours of exposure to

predation at all microsites (timing paired by lake). All tethering experiments occurred

during the afternoon (between 12:00 and 19:00), when fish densities in macrophytes are

likely greatest (Hall et al. 1979, Gliwicz et al. 2006). We recorded larval mortality

(proportion of larvae missing after predator exposure) as an estimate of relative predation

rates.

22For comparisons of tethered larval mortality among 42 microsites with different

categories of macrophyte structure, we conducted an analysis of variance (ANOVA) with

lake added as a random effect. Because variance partitioning indicated that a large

portion of the variance occurred at the lake level, we also used mixed-effects linear

regressions to test effects of secchi depth and conductance at the lake level on larval

mortality (fish CPUE for these lakes was not available). As in the first research question,

we completed all analyses using R (version 2.4.1) software.

Results

Aquatic and terrestrial drivers of Odonata larval density and richness

We found larvae from 27 Odonata species at the 41 study sites (Table 4). Species

richness ranged from 0 to 10 species per (12 m2) site (mean = 5.2 species, SD = 3.4

species). Gomphus spicatus was the most abundant (mean = 0.45 larvae m-2, SD = 0.86

larvae m-2) and frequently-occurring species (present at 67 % of the sites). Gomphus,

which included three other species (G. exilis, lividus, and fraternus), was by far the most

prevalent genus (occurring at 86 % of the sites). The Gomphidae family occurred at 87.8

% of sites, while Zygoptera, Macrominae, Cordulinae, Libellulinae, and Aeshnidae

occurred at 44, 34, 22, 20, and 17 % of sites, respectively.

Total variance summarized by the first two axes of the PCA’s were 52.3 % from

the 12 lake-level variables (axis 1 = 30.4 %, axis 2 = 22.0 %), 53.9 % from the 9 site-

level littoral variables (axis 1 = 34.9 %, axis 2 = 19.0 %), and 56.3 % from the 10 site-

level riparian variables (axis 1 = 37.8 %, axis 2 = 18.6 %; Table 3). The third lake-level

axis represented 14.6% of the variance and correlated almost exclusively with secchi

23depth (r = -0.9). Correlations of odonate response variables with secchi depth did not

suggest significant relationships, so we did not add the third lake axis to the set of

candidate models.

The six PCA axes that we chose for inclusion in the multivariate regressions have

straightforward interpretations. The first PCA axis representing lake-level variables

(hereafter ‘lake size’) correlates positively with lake area, lake perimeter, number of

buildings, and specific conductance, while the second lake PCA axis (hereafter ‘lake

wetlands’) correlates positively with the length of wetlands bordering the lake and

negatively with insectivorous fish CPUE (Fig. 2a, Table 3).

The first PCA axis of littoral variables (hereafter ‘littoral macrophytes’) correlates

positively with macrophyte richness, macrophyte percent cover, and frequency of

emergent macrophytes and negatively with rocky substrate dominance. The second

littoral PCA axis (hereafter ‘littoral muckiness” correlates positively with mucky

substrate dominance and negatively with sandy substrate dominance (Fig. 2b, Table 3).

The first riparian axis (hereafter ‘riparian structural complexity’) correlates

positively with foliage height diversity, length of site with shrubs, structural diversity,

and canopy cover estimate. Finally, the second riparian axis (hereafter ‘riparian tall

wetland plants’) correlates most negatively with the length of sites bordered by tall

wetland herbaceous plants and shoreline percent cover of herbaceous plants (Fig. 2c;

Table 3).

More of the variance in density and richness of odonates occurred at the site

level than lake level, with all of the variance in Gomphidae densities occurring at the site

level (Table 5). The information theoretic approach for model selection allowed us to

24assess the importance of predictor variables across all possible models, rather than only

from a somewhat arbitrary “best” model. Density of all odonates combined was best

explained by increasing riparian tall wetland plants, followed by increasing littoral

macrophytes (Tables 6-7). Importance of the tall wetland plant axis was even greater for

the Gomphidae family, and the model fit was better (Tables 6-7; Fig. 3).

The strongest predictor for Odonata species richness was increasing littoral

macrophyte abundance (Table 7). This was also the best predictor for presence of both

Zygoptera and Libellulinae (Table 7). Littoral “muckiness” was the best predictor for

presence of both Aeshnidae and Cordulinae (Table 7).

Effects of riparian vegetation on adult abundance

We observed adult odonates from 15 different genera at the experimental plots,

but only Enallagma (Zygoptera: Coenagrionidae), Leucorrhinia, and Libellula spp.

(Anisoptera: Libellulinae) were present on > 50 % of the observation days (96 %, 65 %

and 56 %, respectively). Gomphidae adults were observed at these plots in much lower

abundances than Libellulidae species, possibly because they spend more time perched in

treetops and are less active by late July. Anisoptera abundance was greatest at the high-

macrophyte block and did not differ significantly with Typha treatment (p > 0.1; Fig. 4a).

Zygoptera abundance, however, was significantly higher at the riparian Typha treatments

(Table 8; Fig. 4b). We recorded oviposition for Anisoptera and Zygoptera on only 11

and 6 days, respectively. Within each observation day, oviposition for both suborders

occurred more frequently at plots with the Typha treatment than plots without Typha,

25although these differences were not significant (McNemar’s chi-squared tests p > 0.1;

Table 8).

Macrophyte effects on larval densities within sites

At the 18 paired microsites, we found most of the same species as in our 2004

survey, but also three new Libellulids: Leucorhinia hudsonica, L. proxima, and Tramea

Carolina (Table 4). Unlike the broad-scale study, the most abundant species was

Epitheca cynosura (mean = 0.45 larvae m-2, SD = 1.0 larvae m-2), even though it only

occurred at macrophyte microsites (Fig. 5). Gomphus species occurred at 67 % of the

microsites.

We found many more odonates in areas with dense macrophytes (mean = 5.2

larvae m-2, SD = 3.8 larvae m-2) than those without (mean = 0.8 larvae m-2, SD = 1.0

larvae m-2; Wilcoxon rank sum test W = 74, p = 0.0035). Clasper and sprawler guilds

were most abundant in dense macrophyte areas, while the burrower larval guild did not

differ significantly with macrophytes (Fig. 6). We also observed slightly higher species

richness at the macrophyte microsites (mean = 5.22 Anisoptera species, SD = 3.35) than

those without (mean = 2.11 Anisoptera species, SD = 1.83; Wilcoxon rank sum test W =

63, p = 0.05).

Tethered larvae mortality ranged from 0 to 100 % among the 42 microsites (mean

= 53 %, SD = 32 %). On separate occasions, we observed two bluegills and one small

painted turtle feeding on tethered larvae in the field. Mortality did not vary with

macrophyte biomass category (F1,20 = 1.19, p = 0.3), nor with macrophyte presence

(paired t-test: t20 = -1.13, p = 0.3). Microsites with macrophytes present had mean larval

26mortality of 47 % (SE = 6.9 %), and microsites with no macrophytes had a mean

mortality of 56 % larvae (SE = 7.5%). A large portion of the total variance in tethered

larval mortality (39.1 %) occurred at the lake level, but did not correlate significantly

with lake clarity or conductance (p > 0.1).

Discussion

Riparian vegetation influences larval odonate assemblages in the adjacent littoral

ecosystem. Existing literature on ecological links from riparian landscapes to aquatic

communities has focused on detrital, nutrient, and structural subsidies (Wallace et al.

1997, Nakano et al. 1999, Bastow et al. 2002, Sass 2004, Baxter et al. 2005, Roth et al.

2007). Our results (both here and in Chapter III) demonstrate that animals with complex

life histories provide another important link from terrestrial to aquatic ecosystems via

habitat selection. These findings also add an aquatic dimension to the growing body of

research, previously focused on agricultural systems, investigating how vegetation

structure in one ecosystem affects insect abundances in an adjacent system (Langellotto

and Denno 2004, Tscharntke et al. 2005, Bianchi et al. 2006). Site-level riparian

vegetation was, in fact, a better predictor of total larval Odonata density, Gomphidae

density, and Macrominae presence than site- or lake-level aquatic habitat features. In

particular, larval density of the most abundant family, Gomphidae, increased with the

amount of tall wetland plants adjacent to – but not in – the lakes (Table 7; Fig. 3). Other

studies have linked adult odonates to riparian sedge or rush abundance (Van Buskirk

1986, Clark and Samways 1996, Foote and Hornung 2005, Hofmann and Mason 2005b,

27Rouquette and Thompson 2007), but this is the first demonstration of a correlation

between odonate larvae and terrestrial herbaceous vegetation.

The larvae’s stronger relationship with tall wetland vegetation than with riparian

shrubs may reflect the role of the plant structure. Plants included in this category (Carex,

Typha, and Iris spp.) are all rigid and provide stable vertical structures on which

emerging larvae complete hardening of the wings (personal observation). However,

Gomphidae (unlike most other Anisoptera families) can also emerge on horizontal

surfaces (Eda 1963). Whether availability of suitable emergence structures significantly

influence odonate population abundances remains to be tested.

Most larvae are unlikely to move more than 20 m from the littoral site where eggs

were laid (Ubukata 1984, Schaffner and Anholt 1998, but see Alzmann et al. 1999), so

the correlations between larvae and riparian vegetation may also reflect adult habitat

selection as a consequence of larval recruitment (Chapter III). Adults may prefer the

rigid wetland plants for perching because of their height or open structure that facilitates

thermoregulation (May 1976, Pezalla 1979). Other authors’ observations of adult

odonates (from other families) have indicated strong correlations between perch locations

and presence of sedges or rushes (Van Buskirk 1986, Clark and Samways 1996, Foote

and Hornung 2005). As with all oviparous species, it is advantageous for odonates to lay

eggs at sites where they expect high survivorship of eggs and larvae (Storch and Frynta

1999). Because adults cannot assess all aspects of aquatic habitat suitability for larvae

(Wildermuth 1993), several authors have hypothesized that adult odonates use emergent

aquatic plants or riparian plants as proximate cues of macrophyte, substrate, prey

availability, or predator conditions in littoral sites (Buchwald 1992, Wildermuth 1992,

28McKinnon and May 1994). Male and female odonates split their time at riparian sites

between foraging and mating behaviors (e.g., Parr 1983), so riparian vegetation could

also affect adult abundances by correlating with prey densities (Baird and May 1997,

Whitaker et al. 2000, Garono and Kooser 2001, Henning and Remsburg in review).

These hypotheses led us to the potted Typha experiment.

Adult riparian site use generally corresponds with oviposition locations (Corbet

1999). Thus the trends recorded for higher adult abundance and oviposition at treatments

with only 20 potted Typha (Table 9, Fig. 4) suggest that site selection by adults may have

led to higher Gomphidae densities at sites with more wetland vegetation (Fig. 3). The

rarity of oviposition events indicates that it would be difficult to demonstrate oviposition

preferences without significantly longer observation periods. Previous researchers have

added artificial perches to experimental sites to increase local adult odonate densities for

other experiments or to compare perch heights (Wolf and Waltz 1988, Rehfeldt 1990,

Baird and May 1997, May and Baird 2002, De Marco and Resende 2004), but this was

the first study (that we know of) to test directly whether placement of perch structures

affects adult site use or oviposition behavior.

Aeshnidae were most likely to occur at sites with greater littoral muckiness and

riparian tall wetland plants (Table 7). Because these are endophytic species, we expected

that littoral macrophytes would be the most important predictor of their presence.

Unfortunately, our data from the paired microsites could not address this question

because we observed so few Aeshnidae larvae in those surveys. However, coarse wood,

an additional oviposition substrate for some Aeshnidae (Mead 2003), also correlated

positively with the littoral muckiness axis that explained most of the variation in

29Aeshnidae presence (Tables 6-7). In addition, odonates likely make use of floating

macrophytes that shift among lake sites through time, which the single macrophyte

record per site that we used would not reflect. As the strongest fliers of all odonates,

Aeshnidae may also respond to littoral macrophytes at a larger spatial or temporal scale

than the other odonates.

We found increasing species richness at sites with more littoral macrophytes,

which corresponds with more Zygoptera and Libellulinae occurrences at those sites

(Table 7). These results concur with previous studies of odonate assemblages in lakes:

compared to other biotic and abiotic variables, larvae showed the strongest associations

with macrophyte biomass (Weatherhead and James 2001, Michaletz et al. 2005, Osborn

2005). In addition to zygopterans’ requirement of emergent macrophytes for endophytic

oviposition, both Zygoptera and Libellulinae fall into the clasper and sprawler guilds that

use macrophytes as larvae (Fig. 6). Even though sprawlers generally do not climb

directly in macrophytes, they probably rely on macrophytes and the associated detritus

for cover. Finally, all Zygoptera and Libellulinae species are perchers as adults, so

emergent macrophytes may also serve as important perching structures for adults at

oviposition sites (McKinnon and May 1994, Switzer and Walters 1999, De Marco and

Resende 2004).

Clasper and sprawler larval densities were highest at microsites with dense

macrophytes (Fig. 6), but predation risk for tethered larvae did not help explain this

pattern. Our findings differ from mesocosm predation experiments showing the

importance of macrophytes in predator avoidance (Crowder and Cooper 1982, Thompson

1987, Dionne et al. 1990, Diehl 1992). Tethering can interfere with organisms’ predator

30avoidance behaviors (Curran and Able 1998, Kneib and Scheele 2000), but field and

aquarium observations gave us no reason to suspect that effects of tethering on larval

behavior differed among microsite treatments. Tethered larvae appeared capable of

burrowing or climbing up macrophytes, depending on their behavioral guild. Although

we used odonate larvae paired by size and genus, future tethering experiments conducted

exclusively with clasper larvae would have a higher likelihood of detecting predation risk

differences – if they exist - in the field. It is likely that larvae do indeed find refuge from

predators in dense macrophytes, but that higher predator densities within macrophytes

(Werner et al. 1983, Weaver et al. 1997, Gamboa-Perez and Schmitter-Soto 1999) and

fish community differences among lakes obscured this relationship in our field

experiment. On the other hand, competition or foraging behaviors of odonate larvae may

help explain their associations with macrophytes to a greater extent than predator

avoidance (Johansson 1991, Schmude 1998, Elkin and Baker 2000, Hofmann and Mason

2005a).

Our findings that both riparian and littoral vegetation structure influence lentic

Odonata assemblages have several conservation implications. Although some of the

habitat factors that influence local odonate assemblages (e.g., substrate, lake size) are not

influenced by anthropogenic land use, our surveys suggest that site-level littoral and

riparian plant abundances affect odonate assemblages. The largest percentages of the

variance in Gomphidae density and species richness occurred at the site rather than lake

level (Table 5). As a consequence of the vegetation simplification and removal of

macrophytes and coarse wood that often accompanies lakeshore development (Racey and

Euler 1982, Radomski and Goeman 2001, Elias and Meyer 2003, Marburg et al. 2006),

31odonate abundances and diversity could decline. Most lakeshore homeowners value

wildlife such as dragonflies (personal observation), so the result that site-level vegetation

structure affects abundance and diversity of these charismatic insects could help motivate

homeowners to maintain or restore wetland and littoral vegetation. Our results also

suggest that rusty crayfish (Orconectes rusticus) invasion, which leads to severe

macrophyte reduction in north-temperate lakes (Wilson et al. 2004), may indirectly

reduce odonate diversity and density.

As predators of smaller invertebrates throughout their complex life cycle,

Odonata abundances can influence aquatic and riparian communities. Because larval

odonates are often the dominant invertebrate predators in freshwater littoral zones, they

can significantly alter benthic communities (Thorp and Cothran 1984, Johnson et al.

1987, Van Buskirk 1988). High predation of terrestrial pollinators – including bees,

moths, and flies - by adult odonates may even alter riparian plant reproduction (Knight et

al. 2005). At critical times, odonates comprise a significant portion of the diets of fish

(Crowder and Cooper 1982, Johnson et al. 1995, Sass 2004) and birds (Orians and Horn

1969, Wissinger 1988), so habitat variables driving odonate abundances could lead to

changes in both aquatic and terrestrial food webs.

Acknowledgments

This research was supported financially by NSF Biocomplexity Program DEB-

0083545, North-temperate lakes LTER, an NSF Graduate Research Fellowship, the

Garden Club of America, and the Animal Behavior Society. Tim Kratz, M. Woodford, S.

van Egeren, and the University of Wisconsin Trout Lake station staff provided invaluable

32logistical support. C. Ane offered statistical advice. Thanks also to W. Smith, R.

DuBois, K. Tennessen, and R. Bohanan for sharing their expertise on Odonata behaviors

and taxonomy. B. Peckarsky, A. Ives, J. Zedler, C. Gratton, and S. Carpenter helped

guide and critique the manuscript. Field and laboratory assistants included M. Theis, P.

Winkler, B. Henning, J. Batten, J. Estreen, D. Hong, J. Iacarella, C. Leugers, A. Duong,

and A. Olson. Finally, the manuscript was reviewed by A. Olson.

Literature Cited

Agrawal, A. A., D. D. Ackerly, F. Adler, A. E. Arnold, C. Caceres, D. F. Doak, E. Post, P. J. Hudson, J. Maron, K. A. Mooney, M. E. Power, D. Schemske, J. Stachowicz, S. Strauss, M. G. Turner, and E. E. Werner. 2007. Filling key gaps in population and community ecology. Frontiers in Ecology and the Environment 5:145-152.

Aksnes, D. L., and J. Giske. 1993. A theoretical-model of aquatic visual feeding. Ecological Modelling 67:233-250.

Alexander, M. L. 2005. Lake ecosystem variation across landscape position and human development gradients in Vilas County, Wisconsin. University of Wisconsin, Madison.

Alzmann, N., B. Kohler, and G. Maier. 1999. Spatial distribution, food and activity of Gomphus pulchellus Selys 1840 (Insecta; Odonata; Gomphidae) from a still water habitat. International Review of Hydrobiology 84:299-313.

Argesti, A. 2002. Categorical Data Analysis, 2 edition. Wiley, New York.

Askew, R. 1982. Roosting and resting site selection by Coenagrionid damselflies. Advances in Odonatology 1:1-8.

Baird, J. M., and M. L. May. 1997. Foraging behavior of Pachydiplax longipennis (Odonata: Libellulidae). Journal of Insect Behavior 10:655-678.

Bastow, J. L., J. L. Sabo, J. C. Finlay, and M. E. Power. 2002. A basal aquatic-terrestrial trophic link in rivers: algal subsidies via shore-dwelling grasshoppers. Oecologia 131:261-268.

33Baxter, C. V., K. D. Fausch, and W. C. Saunders. 2005. Tangled webs: reciprocal flows

of invertebrate prey link streams and riparian zones. Freshwater Biology 50:201-220.

Bianchi, F., C. J. H. Booij, and T. Tscharntke. 2006. Sustainable pest regulation in agricultural landscapes: a review on landscape composition, biodiversity and natural pest control. Proceedings of the Royal Society B-Biological Sciences 273:1715-1727.

Black, J. J., L. M. Andrews, and C. W. Threinen. 1963. Surface Water Resources of Vilas County. Wisconsin Conservation Department, Madison, WI.

Bried, J. T., and G. N. Ervin. 2006. Abundance patterns of dragonflies along a wetland buffer. Wetlands 26:878-883.

Brose, U. 2003. Bottom-up control of carabid beetle communities in early successional wetlands: mediated by vegetation structure or plant diversity? Oecologia 135:407-413.

Bryan, M. D., and D. L. Scarnecchia. 1992. Species richness, composition, and abundance of fish larvae and juveniles inhabiting natural and developed shorelines of a glacial Iowa lake. Environmental Biology of Fishes 35:329-341.

Buchwald, R. 1992. Vegetation and dragonfly fauna - characteristics and examples of biocenological field studies. Vegetatio 101:99-107.

Burcher, C. L., and L. A. Smock. 2002. Habitat distribution, dietary composition and life history characteristics of odonate nymphs in a blackwater coastal plain stream. American Midland Naturalist 148:75-89.

Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach, Second edition. Springer, New York.


Carpenter, S. R., T. K. Kratz, J. Kitchell, J. J. Magnuson, and M. J. Vander Zanden. 2006. Biocomplexity; Coordinated Field Studies. http://lterquery.limnology.wisc.edu/index_new.jsp?project_id=BIOCOM1 University of Wisconsin Board of Regents.

Carr, M. H. 1991. Habitat selection and recruitment of an assemblage of temperate zone reef fishes. Journal of Experimental Marine Biology and Ecology 146:113-137.

34Cheruvelil, K. S., P. A. Soranno, J. D. Madsen, and M. J. Roberson. 2002. Plant

architecture and epiphytic macroinvertebrate communities: the role of an exotic dissected macrophyte. Journal of the North American Benthological Society 21:261-277.

Clark, T. E., and M. J. Samways. 1996. Dragonflies (Odonata) as indicators of biotope quality in the Kruger National Park, South Africa. Journal of Applied Ecology 33:1001-1012.



Curran, M. C., and K. W. Able. 1998. The value of tethering fishes (winter flounder and tautog) as a tool for assessing predation rates. Marine Ecology-Progress Series 163:45-51.

Curtis, J. T. 1959. The Vegetation of Wisconsin. University of Wisconsin Press, Madison, WI, USA.

Davies-Colley, R. J., and D. G. Smith. 2001. Turbidity, suspended sediment, and water clarity: A review. Journal of the American Water Resources Association 37:1085-1101.


De Souza, A. L. T., and R. P. Martins. 2005. Foliage density of branches and distribution of plant-dwelling spiders. Biotropica 37:416-420.

Decamps, H., G. Pinay, R. J. Naiman, G. E. Petts, M. E. McClain, A. Hillbricht-Ilkowska, T. A. Hanley, R. M. Holmes, J. Quinn, J. Gibert, A. M. P. Tabacchi, F. Schiemer, E. Tabacchi, and M. Zalewski. 2004. Riparian zones: Where biogeochemistry meets biodiversity in management practice. Polish Journal of Ecology 52:3-18.


Dionne, M., M. Butler, and C. Folt. 1990. Plant-specific expression of antipredator behavior by larval damselflies. Oecologia 83:371-377.

Duffy, W. G. 1994. Demographics of Lestes disjunctus disjunctus (Odonata, Zygoptera) in a riverine wetland. Canadian Journal of Zoology 72:910-917.

35Eda, S. 1963. Emergence of Epophlebia superstes Selys. Tombo 5:1-3, (in Japanese)

reviewed by Corbet 1999.

Elias, J. E., and M. W. Meyer. 2003. Comparisons of undeveloped and developed shorelands, northern Wisconsin, and recommendations for restoration. Wetlands 23:800-816.

Elkin, C. M., and R. Baker. 2000. Lack of preference for low-predation-risk habitats in larval damselflies explained by costs of intraspecific interactions. Animal Behaviour 60:511-521.


Gamboa-Perez, H. C., and J. J. Schmitter-Soto. 1999. Distribution of cichlid fishes in the littoral of Lake Bacalar, Yucatan Peninsula. Environmental Biology of Fishes 54:35-43.

Garono, R. J., and J. G. Kooser. 2001. The relationship between patterns in flying adult insect assemblages and vegetation structure in wetlands of Ohio and Texas. Ohio Journal of Science 101:12-21.

Gergel, S. E. 1996. Scale-dependent landscape effects on north temperate lakes and rivers, M.S. thesis. University of Wisconsin, Madison.

Gliwicz, Z. M., J. Slon, and I. Szynkarczyk. 2006. Trading safety for food: evidence from gut contents in roach and bleak captured at different distances offshore from their daytime littoral refuge. Freshwater Biology 51:823-839.

Gonzalez-Abraham, C. E., V. C. Radeloff, R. B. Hammer, T. J. Hawbaker, S. I. Steward, and M. K. Clayton. 2007. Building patterns and landscape fragmentation in northern Wisconsin, USA. Landscape Ecology 22:217-230.

Hall, D. J., E. E. Werner, J. F. Gilliam, G. G. Mittelbach, D. Howard, C. G. Doner, J. A. Dickerman, and A. J. Stewart. 1979. Diel foraging behavior and prey selection in the golden shiner (Notemigonus-crysoleucas). Journal of the Fisheries Research Board of Canada 36:1029-1039.

Henning, B. M., and A. J. Remsburg. in review. Effects of lakeshore vegetation structure on avian and amphibian abundance in northern Wisconsin. Wildlife Research.

Hofmann, T. A., and C. F. Mason. 2005a. Competition, predation and microhabitat selection of Zygoptera larvae in a lowland river. Odonatologica 34:27-36.

Hofmann, T. A., and C. F. Mason. 2005b. Habitat characteristics and the distribution of Odonata in a lowland river catchment in eastern England. Hydrobiologia 539:137-147.

36Hrabik, T. R., B. K. Greenfield, D. B. Lewis, A. I. Pollard, K. A. Wilson, and T. K.

Kratz. 2005. Landscape-scale variation in taxonomic diversity in four groups of aquatic organisms: The influence of physical, chemical, and biological properties. Ecosystems 8:301-317.

Jacobs, M. E. 1955. Studies on territorialism and sexual selection in dragonflies. Ecology 36:566-586.

Johansson, F. 1991. Foraging modes in an assemblage of odonate larvae - effects of prey and interference. Hydrobiologia 209:79-87.

Johansson, F., and T. Brodin. 2003. Effects of fish predators and abiotic factors on dragonfly community structure. Journal of Freshwater Ecology 18:415-423.

Johnson, D., and P. Crowley. 1980. Habitat and seasonal segregation among coexisting Odonate larvae. Odonatologica 9:297-308.

Johnson, D. M., T. H. Martin, M. Mahato, L. B. Crowder, and P. H. Crowley. 1995. Predation, density dependence, and life histories of dragonflies: A field experiment in a freshwater community. Journal of the North American Benthological Society 14:547-562.

Johnson, D. M., C. L. Pierce, T. H. Martin, C. N. Watson, R. E. Bohanan, and P. H. Crowley. 1987. Prey depletion by odonate larvae - combining evidence from multiple field experiments. Ecology 68:1459-1465.

Karr, J. R., and R. R. Roth. 1971. Vegetation structure and avian diversity in several New World areas. American Naturalist 105:423-&.

Kneib, R. T., and C. E. H. Scheele. 2000. Does tethering of mobile prey measure relative predation potential? An empirical test using mummichogs and grass shrimp. Marine Ecology-Progress Series 198:181-190.

Knight, T. M., M. W. McCoy, J. M. Chase, K. A. McCoy, and R. D. Holt. 2005. Trophic cascades across ecosystems. Nature 437:880-883.


Lindsay, A. R., S. S. Gillum, and M. W. Meyer. 2002. Influence of lakeshore development on breeding bird communities in a mixed northern forest. Biological Conservation 107:1-11.

Logan, E. R. 1971. A comparative ecological and behavioral study of two species of damselflies, Enallagma boreale (Selys) and Enallagma carunculatum Morse (Odonata: Coenagrionidae). Washington State University, Pullman.

37MacArthur, R. H. 1965. Patterns of species diversity. Biological Reviews 40:510-533.

Marburg, A. E., M. G. Turner, and T. K. Kratz. 2006. Natural and anthropogenic variation in coarse wood among and within lakes. Journal of Ecology 94:558-568.

May, M. L. 1976. Thermoregulation and adaptation to temperature in dragonflies (Odonata-Anisoptera). Ecological Monographs 46:1-32.

May, M. L., and J. M. Baird. 2002. A comparison of foraging behavior in two "percher" dragonflies, Pachydiplax longipennis and Erythemis simplicicollis (Odonata : Libellulidae). Journal of Insect Behavior 15:765-778.


McPeek, M. A. 1990. Determination of species composition in the Enallagma damselfly assemblages of permanent lakes. Ecology 71:83-98.

Mead, K. 2003. Dragonflies of the North Woods. Kollath-Stensaas Publishing, Duluth, MN, USA.

Menard, S. 2000. Coefficients of determination for multiple logistic regression analysis. The American Statistician 54:17-24.

Michaletz, P. H., K. E. Doisy, and C. F. Rabeni. 2005. Influences of productivity, vegetation, and fish on macroinvertebrate abundance and size in Midwestern USA impoundments. Hydrobiologia 543:147-157.

Milner, A. M., and I. T. Gloyne-Phillips. 2005. The role of riparian vegetation and woody debris in the development of macroinvertebrate assemblages in streams. River Research and Applications 21:403-420.

Mueller-Dombois, D., and H. Ellenberg. 1974. Aims and Methods of Vegetation Ecology. John Wiley & Sons, Inc., New York.

Muller, Z. 2003. Effect of sports fisherman activities on dragonfly assemblages on a Hungarian river floodplain. Biodiversity and Conservation 12:167-179.

Muntifering, J. R., A. J. Dickman, L. M. Perlow, T. Hruska, P. G. Ryan, L. L. Marker, and R. M. Jeo. 2006. Managing the matrix for large carnivores: a novel approach and perspective from cheetah (Acinonyx jubatus) habitat suitability modelling. Animal Conservation 9:103-112.

Nakano, S., H. Miyasaka, and N. Kuhara. 1999. Terrestrial-aquatic linkages: Riparian arthropod inputs alter trophic cascades in a stream food web. Ecology 80:2435-2441.

38Needham, J. G., M. J. Westfall, and M. L. May. 2000. Dragonflies of North America.

Scientific Publishers, Inc., Gainesville, FL.

Orians, G. H., and H. S. Horn. 1969. Overlap in foods and foraging of four species of blackbirds in the potholes of central Washington. Ecology 50:930-938.

Osborn, R. 2005. Odonata as indicators of habitat quality at lakes in Louisiana, United States. Odonatologica 34:259-270.

Parr, M. J. 1983. Some aspects of territoriality in Orthetrum coerulescens (Fabricius) (Anisoptera: Libellulidae). Odonatologica 12:39-57.

Peterson, G. D., T. D. Beard, B. E. Beisner, E. M. Bennet, S. R. Carpenter, G. S. Cumming, C. L. Dent, and T. D. Havlicek. 2003. Assessing future ecosystem services a case study of the Northern Highlands Lake District, Wisconsin. Conservation Ecology 7.

Pezalla, V. M. 1979. Behavioral ecology of the dragonfly Libellula pulchella Drury (Odonata:Anisoptera). American Midland Naturalist 102:1-22.

Pianka, E. R. 1967. Lizard species diversity. Ecology 48:333-351.

Racey, G. D., and D. L. Euler. 1982. Small mammal and habitat response to shoreline cottage development in central Ontario. Canadian Journal of Zoology-Revue Canadienne De Zoologie 60:865-880.

Radomski, P., and T. J. Goeman. 2001. Consequences of human lakeshore development on emergent and floating-leaf vegetation abundance. North American Journal of Fisheries Management 21:46-61.

Rehfeldt, G. E. 1990. Antipredator strategies in oviposition site selection of Pyrrhosoma-nymphula (Zygoptera, Odonata). Oecologia 85:233-237.

Rith-Najarian, J. 1998. The influence of forest vegetation variables on the distribution and diversity of dragonflies in a northern Minnesota forest landscape: a preliminary study (Anisoptera). Odonatologica 27:335-351.

Roth, B. M., I. C. Kaplan, G. G. Sass, P. T. Johnson, A. E. Marburg, A. C. Yannarell, T. D. Havlicek, T. V. Willis, M. G. Turner, and S. R. Carpenter. 2007. Linking terrestrial and aquatic ecosystems: The role of woody habitat in lake food webs. Ecological Modelling 203:439-452.

Rouquette, J. R., and D. J. Thompson. 2007. Roosting site selection in the endangered damselfly, Coenagrion mercuriale, and implications for habitat design. Journal of Insect Conservation 11:187-193.

39Sahlen, G. 1999. The impact of forestry on dragonfly diversity in central Sweden.

International Journal of Odonatology 2:177-186.

Sass, G. G. 2004. Fish community and food web responses to a whole-lake removal of coarse woody habitat. University of Wisconsin, Madison.

Schaffner, A. K., and B. R. Anholt. 1998. Influence of predator presence and prey density on behavior and growth of damselfly larvae (Ischnura elegans) (Odonata : Zygoptera). Journal of Insect Behavior 11:793-809.

Schindler, D. E., S. I. Geib, and M. R. Williams. 2000. Patterns of fish growth along a residential development gradient in north temperate lakes. Ecosystems 3:229-237.

Schmude, K. L. 1998. Effects of habitat complexity on macroinvertebrate colonization of artificial substrates in north temperate lakes. Journal of the North American Benthological Society 17:73-80.

Schnaiberg, J., J. Riera, M. G. Turner, and P. R. Voss. 2002. Explaining human settlement patterns in a recreational lake district: Vilas County, Wisconsin, USA. Environmental Management 30:24-34.

Storch, D., and D. Frynta. 1999. Evolution of habitat selection: stochastic acquisition of cognitive clues? Evolutionary Ecology 13:591-600.

Sugiura, N. 1978. Further analysis of the data by Akaike's information criterion and the finite corrections. Communications in Statistics, Theory and Methods A7:13-26.

Switzer, P. V., and W. Walters. 1999. Choice of lookout posts by territorial amberwing dragonflies, Perithemis tenera (Anisoptera : Libellulidae). Journal of Insect Behavior 12:385-398.

Thompson, D. J. 1987. Regulation of damselfly populations: the effects of weed density on larval mortality due to predation. Freshwater Biology 17:367-371.

Thorp, J., and M. Cothran. 1984. Regulation of freshwater community structure at multiple intensities of dragonfly predation. Ecology 65:1546-1555.

Tscharntke, T., T. A. Rand, and F. Bianchi. 2005. The landscape context of trophic interactions: insect spillover across the crop-noncrop interface. Annales Zoologici Fennici 42:421-432.

Tupper, M., and R. G. Boutilier. 1995. Effects of habitat on settlement, growth, and postsettlement survival of Atlantic cod (Gadus-morhua). Canadian Journal of Fisheries and Aquatic Sciences 52:1834-1841.

40Ubukata, H. 1984. Oviposition site selection and avoidance of additional mating by

females of the dragonfly, Cordulia-aenea-amurensis Selys (Corduliidae). Researches on Population Ecology 26:285-301.

Van Buskirk, J. 1986. Establishment and organization of territories in the dragonfly Sympetrum-rubicundulum (Odonata, Libellulidae). Animal Behaviour 34:1781-1790.

Van Buskirk, J. 1988. Interactive effects of dragonfly predation in experimental pond communities. Ecology 69:857-867.

Wallace, J. B., S. L. Eggert, J. L. Meyer, and J. R. Webster. 1997. Multiple trophic levels of a forest stream linked to terrestrial litter inputs. Science 277:102-104.


Weatherhead, M. A., and M. R. James. 2001. Distribution of macroinvertebrates in relation to physical and biological variables in the littoral zone of nine New Zealand lakes. Hydrobiologia 462:115-129.

Weaver, M. J., J. J. Magnuson, and M. K. Clayton. 1997. Distribution of littoral fishes in structurally complex macrophytes. Canadian Journal of Fisheries and Aquatic Sciences 54:2277-2289.

Werner, E. E., J. F. Gilliam, D. J. Hall, and G. G. Mittelbach. 1983. An experimental test of the effects of predation risk on habitat use in fish. Ecology 64:1540-1548.

Westfall, M. J., and M. L. May. 1996. Damselflies of North America. Scientific Publishers, Inc., Gainesville, FL.

Whitaker, D. M., A. L. Carroll, and W. A. Montevecchi. 2000. Elevated numbers of flying insects and insectivorous birds in riparian buffer strips. Canadian Journal of Zoology-Revue Canadienne De Zoologie 78:740-747.

Wildermuth, H. 1992. Habitate und habitatwahl der grossen moosjungfer (Leucorrhinia pectoralis) Charp. 1825 (Odonata, Libellulidae). Z. Okol. Naturschutz 1:3-21 (in German), reviewed by Corbet 1999 p.1918.


Wilson, K. A., J. J. Magnuson, D. M. Lodge, A. M. Hill, T. K. Kratz, W. L. Perry, and T. V. Willis. 2004. A long-term rusty crayfish (Orconectes rusticus) invasion:

41dispersal patterns and community change in a north temperate lake. Canadian Journal of Fisheries and Aquatic Sciences 61:2255-2266.

Wissinger, S. A. 1988. Spatial-distribution, life-history and estimates of survivorship in a 14-species assemblage of larval dragonflies (Odonata, Anisoptera). Freshwater Biology 20:329-340.

Wolf, L. L., and E. C. Waltz. 1988. Oviposition site selection and spatial predictability of female white-faced dragonflies (Leucorrhinia-intacta) (Odonata, Libellulidae). Ethology 78:306-320.

Woodford, J. E., and M. W. Meyer. 2003. Impact of lakeshore development on green frog abundance. Biological Conservation 110:277-284.

42

Table 1. Publications demonstrating correlations between Odonata species and riparian

vegetation.

Location Life stage, odonate taxa

Riparian plant variable

Shoreline distance or plant area studied

Odonate response to decreased plant variable

Reference

Italy, ponds Adults, all 50 % cover of riparian veg.

0.5 – 5 km2 ↓ richness Carchini et al. (2003)

Canada, wetlands

Adults, all Shrub and bulrush height

250 m ↓ abundance Foote and Hornung (2005)

New Jersey, pond

Adults, Pachydiplax longipennis

Grass height 5.6 m2 ↓ density McKinnon and May (1994)

Hungary, river Combined stages, all

Presence of native riparian and aquatic plants

150 m ↓ richness Muller (2003)

Pennsylvania, pond

Adults, Sympetrum rubicundulum

Grasses, sedges, rushes

% cover of 0.09 ha pond

Chose different pond section

Van Buskirk (1986)

South Africa, river

Adults, all Sedges and rushes % cover

90 m2 ↓ richness Clark and Samways (1996)

UK, river Adults, all Sedge and tall grass % cover

20 m Altered species assemblage

Hofmann and Mason (2005b)

UK, river Adults and larvae, all

Shade % cover 20 m Altered species assemblage

Hofmann and Mason (2005b)

UK, river Adults, Calopteryx splendens

Tree % cover 1.5 km ↑ abundance Ward and Mill (2005)

UK, river Adults, Calopteryx splendens

Understory stem height

1.5 km ↓ abundance Ward and Mill (2005)

Sweden, ponds Larvae, all Logging 1 km2 ↓ richness Sahlen (1999)

Note: A study on how Minnesota forest age affects Anisoptera diversity (Rith-Najarian

1998) was not included because results did not demonstrate statistical

significance.

Table 2. Physical and biological attributes1 of study lakes in northeastern Wisconsin.

Lake

Dataset2 Perim (km) 3

Area (ha)

Morpho-metry 4

Wetland perim(%)5

Houses / km 6

Cond. (µ S / cm)7

Secchi depth (m)

8

Insectiv-orous fish

CPUE9

All fish CPUE9

Arrowhead T 3.5 40.0 1.6 3.0 45.8 99.0 3.35 1095 1763 Big S 15.2 343.6 2.3 4.8 4.7 130.8 4.26 247 328 Black Oak S 12.0 230.1 2.2 15.7 18.0 50.0 5.25 174 283 Big Arbor Vitae T 16.4 441.5 NA NA NA 117.0 3.35 NA NA Big St. Germain S,M,T 12.8 665.7 1.4 0.5 24.3 90.8 2.59 411 440 Brandy S,M,T 3.5 45.1 1.5 0.0 30.1 166.6 3.06 488 501 Boulder T 11.3 215.8 NA NA NA 71.0 1.83 NA NA Carpenter S 5.6 144.5 1.3 0.9 18.0 23.0 5.50 277 294 Diamond M,T 3.6 50.4 NA NA NA 15.0 7.92 NA NA Erickson T 4.0 47.2 1.6 0.0 0.5 57.0 3.35 NA 292 Found T 6.4 139.3 1.5 2.2 16.6 49 1.52 NA 507 Jag S 5.1 68.9 1.7 14.4 1.4 15.0 3.50 219 219 Johnson T 3.6 34.7 1.7 0.0 26.2 98 1.83 NA NA Laura S 10.6 221.1 2.0 0.0 7.29 112.5 4.13 282 235 Little Arbor Vitae S,M 10.6 221.1 2.0 5.8 7.2 112.5 4.13 109 521 Little St. Germain S,M,T 23.3 402.2 3.3 2.7 19.8 76.5 3.63 238 293 Little Star S 6.4 107.1 1.7 0.0 19.5 93.3 6.01 204 284 Lost T 7.8 223.7 1.5 1.1 26.2 58 1.52 NA 1915 Lynx S 11.2 126.2 2.8 0.7 5.6 23.4 4.44 435 437 Mann T 7.7 103.0 NA NA NA 146 2.44 NA NA Pallette S,M,T 3.6 71.0 1.2 0.0 0 18.9 4.07 298 469 Papoose S 13.3 176.6 2.8 3.9 13.5 110.0 5.75 140 177 Snipe T 6.0 89.6 NA NA NA 25.0 1.22 NA NA Stella T 7.1 36.8 NA NA NA NA NA NA NA Starett T NA 26.7 NA NA NA NA NA NA NA Tomahawk T 48.6 1372.7 NA NA NA NA NA NA NA Towanda S,M,T 6.0 60.6 2.2 24.8 18.9 30.7 2.38 227 302 Trilby T 2.7 37.2 NA NA NA NA NA NA NA

Upper Buckatabon S 13.2 211.4 2.6 12.0 12.6 74.0 3.25 406 424 Upper Gresham M,T 9.3 148.1 NA NA NA NA NA NA NA Vandercook S 3.3 41.3 1.5 0.0 13.8 13.6 3.94 180 188 White Sand S 9.3 304.6 1.5 19.1 5.8 74.0 4.88 100 502 Wild Rice M,T 7.0 158.7 NA NA NA 87.0 3.05 NA NA Note: NA indicates that data are not available. 1 Data from Carpenter et al. (2006)

2 Identifies the research question included data from each lake: S = larval survey (question 1), M = comparison of larvae among paired microsites (question 3), T = predation experiment using tethered larvae (question 3). 3 Lake perimeter. 4 Lake morphometry is a unit-less index of the lake’s deviance from a circular shape. 5 Percent of lake perimeter in wetlands (Gergel 1996). 6 Number of buildings within 100 m of the shoreline (Schnaiberg et al. 2002) divided by the lake perimeter. 7 Specific conductance (dissolved ionic content) in µ Siemens cm-1. 8 Water clarity increases with greater depths of viewing a secchi plate. 9 Catch per unit effort (CPUE) based on eight nights of electro-fishing, 24 minnow traps, and 24 crayfish traps.

45Table 3. Summary statistics from 41 sites (17 lakes), abbreviations (used in Fig. 1),

hypothesized correlations with Odonata species richness, and Pearson correlation

coefficients between habitat variables and the best two axes generated by three separate

principal components analysis (PCA) procedures on the three categories of habitat

variables.

Pearson correlations

Habitat descriptors

Abbre-viation Mean

Std dev. Min Max

Expect. correl-ation

PCA Axis

1

PCA Axis

2 Lake-level1 Insectivorous fish CPUE 2 predfish 256 105 100 488 - 0.20 -0.73Total fish CPUE 2 allfish 334 114 177 521 - 0.24 -0.34Conductance (µSiemens/cm) 3 conduct. 63.60 44.69 13.64 166.59 + 0.67 -0.18Secchi depth (m) 3 secchi_m 4.36 1.36 2.38 8.05 + -0.16 0.07Morphometry 3 morpheme 1.9 0.6 1.2 3.3 + 0.60 0.54Area (ha) 3 area 200.8 166.9 41.3 665.7 - 0.75 -0.14Maximum depth (m) 3 maxdepth 13.6 6.5 4.3 25.9 - 0.49 0.22Perimeter (km) 3 perim. 9.0 5.0 3.3 23.3 + 0.87 0.30Length of wetlands (m) 4 wetlands 554.2 622.1 0.0 1917.9 + 0.15 0.85Perimeter % with wetlands 4 % wetland 0.1 0.1 0.0 0.2 + -0.15 0.78Building density (per km) 5 bldg dens 120 115 0 460 - 0.56 -0.40Number of buildings 5 bldgs 12.4 8.7 0.0 30.1 - 0.89 -0.03 Site-level littoral6 Sand (ordinal variable: 0, 1, 2)7 sand 2 1 0 2 + -0.23 -0.77Rock (ordinal variable: 0, 1, 2) 7 rock 0 1 0 2 - -0.44 0.21Gravel (ordinal variable: 0, 1, 2) 7 gravel 1 1 0 2 - -0.63 -0.10Muck (ordinal variable: 0, 1, 2) 7 muck 1 1 0 2 + 0.53 0.64Mean macrophyte richness8 macrich 4 2 0 9 + 0.90 -0.09Mean macrophyte % cover8 macperc 35.0 28.8 0.0 90.0 + 0.86 0.03Mean frequency of emergent macrophytes8 emergent 0.2 0.3 0.0 1.2 + 0.66 -0.28

Coarse wood (number per 50 m)6 num.logs 2 3 0 13 + -0.03 0.60Slope (degrees) slope 1.3 0.2 0.9 1.5 - -0.48 0.47 Site-level riparian9 Canopy (ordinal variable: 1, 2, 3) canopy 2 1 1 3 + 0.70 0.03Slope (degrees) 6 rip.slope 10.3 7.0 0.0 29.0 - 0.35 -0.10Length of shore with shrubs (m) shrublen 13.3 12.8 0.0 30.0 + 0.86 0.04Length of shore with tall wetland plants (m)10 carex.ty 9.3 11.5 0.0 30.0 + 0.17 -0.84

Vegetation distance from water (m) rip.veg.di 0.5 0.9 0.0 3.8 - 0.46 0.26

46Max. understory foliage height (m) maxveg 1.7 0.6 0.2 2.0 + 0.68 -0.25 Mean structural diversity (number of structure types intersected)

avg.veg.

2

1

0

4

+

0.73

<0.01Mean foliage height diversity (number of layers intercepted) avg.layer 4 2 1 7 + 0.92 -0.08

Shore % herbaceous herbaceo 31.0 31.4 0.0 98.0 + 0.12 -0.82Shore % shrubs shrubper 21.9 31.4 0.0 92.0 + 0.56 0.58

1 The study includes 17 lakes, all in Vilas County, Wisconsin (Table 2).

2 Catch-per-unit-effort (CPUE) based on 8 equivalent nights of electro-fishing, 24

minnow traps, and 24 crayfish traps in 2001-2004 (Carpenter et al. 2006) .

3 Data collected 2001-2003 (Carpenter et al. 2006)

4 Data from Gergel (1996).

5 Schnaiberg et al. (2002).

6Littoral sites consisted of 30 m along the lakeshore contour and the adjacent littoral zone

with water depth < 1 m.

7 Littoral substrate dominance and coarse wood abundance estimated at 0.5 m water

depth in 2001-2004 (Marburg et al. 2006).

8 Macrophyte species richness and percent cover estimated in 2001-2003 along a transect

extending 0 to 2 m depth (Alexander 2005).

9Riparian sites were 30 m along the shore by 6 m inland. However, the two shore percent

cover variables were within 0.5 m of the ordinary high water mark.

10 Carex, Typha,or Iris species.

47Table 4. Species list of all Odonata larvae collected from 59 littoral sites1 in northern

Wisconsin lakes (Table 3), including their behavioral guild classifications .

Larval behavior Adult behavior Clasper Sprawler Burrower Flier Percher Aeshnidae Basiaeschna janata X X Boyeria vinosa X X Gomphidae Arigomphus furcifer X X Dromogomphus spinosus X Gomphus spicatus X X Gomphus exilis X X Gomphus lividus X X Gomphurus fraternus X X Hagenius brevistylus X X Libellulinae Celithemis elisa X X Ladona julia X X X Leucorrhinia hudsonica X X Leucorrhinia proxima X X Somatachlora cingulata X X Tramea carolina X X Cordulinae Cordulia shurtleffIi X X Drorocordulia libera X X Epitheca cynosura X X Epitheca princeps X X Epitheca spinigera X X Macrominae Didymops transversa X X X Macromia illinoiensis X X X Coenagrionidae spp. X X 1 This includes a broad-scale survey of 41 sites (12 m2 of littoral benthos sampled per

site) in 2004 and a fine-scale survey of 18 sites (half in macrophytes; 7 m2 of littoral

benthos sampled per site) in 2006. Three Libellulinae species (L. hudsonica, L. proxima,

and T. carolina) were only observed in 2006; all other species were found in both years.

48Table 5. Variance (and percent of total variance) occurring at two nested spatial scales

for mixed-effects models of Odonata larval abundances.

Scale Odonata density Gomphidae density Odonata richness

Site 0.19 (56.5 %) 0.11 (100.0 %) 0.03 (61.6 %)

Lake 0.12 (34.6 %) 0.00 (0.0 %) 0.01 (29.4 %)

49Table 6. Relatively equivalent linear and logistic models (based on differences between

second-order Akaike scores, ∆AICc) predicting Odonata larval density or presence

among 41 Wisconsin lake sites.

Model coefficients

Model parameters ∆AICc

1 AICc

weight

Fixed-effects

model fit K2 Lake size

Lake wetlands

Littoralmacro-phytes

Littoralmuck

Riparian Struct-

ural complex

-ity

Riparian tall

wetland plants

y-intercept

Linear regressions adj. R2

Odonata density 0.00 0.14 0.43 1 -0.14 1.02

0.51 0.11 0.45 2 0.08 -0.11 1.02

1.16 0.08 0.44 1 0.11 1.03

1.59 0.05 0.44 2 -0.09 0.13 1.05

1.63 0.05 0.45 3 -0.07 0.09 -0.1 1.04

1.89 0.05 0.43 2 -0.05 -0.14 1.03

Species richness m-2 0.00 0.15 0.34 1 0.04 0.60

1.04 0.09 0.30 0 0.60

1.20 0.08 0.34 2 -0.02 0.04 0.60

1.58 0.06 0.32 2 0.04 0.02 0.60

1.99 0.05 0.32 1 -0.03 0.60

Gomphidae density

0 0.15 0.48 1 -0.15 0.88

1.11 0.09 0.48 2 -0.07 -0.14 0.90

1.15 0.08 0.49 2 0.06 -0.12 0.88

1.76 0.06 3 -0.02 0.02 0.01 0.88

Logistic regressions Pseudo

R2

Zygoptera presence 0.00 0.12 0.37 1 0.38 -2.31

0.81 0.08 0.42 2 045 0.34 -1.99

1.36 0.06 0.44 2 0.39 -0.27 -2.37

1.50 0.05 0.37 2 0.42 0.18 -2.20

1.73 0.05 0.37 0 -2.37

50Table 6, continued

Model coefficients

Model parameters ∆AICc

1 AICc

weight

Fixed-effects

model fit K2 Lake size

Lake wetlands

Littoralmacro-phytes

Littoralmuck

Riparian Struct-

ural complex

-ity

Riparian tall

wetland plants

y-intercept

Libellulinae presence 0 0.11 0.65 1 0.470 -5.91

0.68 0.08 0.56 0 -5.40

0.80 0.07 0.65 2 -0.38 0.62 -6.23

1.46 0.05 0.65 3 -0.61 0.64 -0.41 -6.84

Macrominae presence

0.00 0.09 0.63 1 0.42 -4.91

0.75 0.06 0.63 2 0.36 0.35 -4.36

1.09 0.05 0.69 2 -0.36 0.50 -5.06

1.20 0.05 0.63 1 0.43 -4.78

1.21 0.05 0.69 3 0.47 -0.50 0.43 -4.23

1.62 0.04 0.63 2 0.41 0.26 -4.75

1.63 0.04 0.63 0 -5.37

Cordulinae presence 0.00 0.22 0.92 2 0.63 0.94 -5.54

1.77 0.09 NA3 3 0.54 1.04 -0.26 -6.09

Aeshnidae presence 0.00 0.14 NA3 3 7.93 5.85 -5.09 -31.65

0.15 0.13 NA3 3 -21.36 17.03 -13.69 -51.08

0.44 0.12 NA3 3 -11.66 21.25 -35.25 -215.28

1.36 0.07 NA3 2 15.22 -17.00 -132.61

1.80 0.06 NA3 5 1.99 7.50 5.71 2.01

1 Only models with ∆AICc < 2 are shown. ∆AICi is the difference between AICc for

model i and the smallest AICc of the candidate models.

2 k = number of fixed-effects parameters in the model

3 Fixed-effects pseudo R2 for these models could not be calculated because there was no

variability within lakes.

51Table 7. Relative importance (sums of Akaike weights) of lake- and site-level

parameters1 in predicting Odonata larval densities at 41 Wisconsin lake sites.

Sums of Akaike weights from all candidate models with the predictor

Model parameters

(Lake 1) “Size”

(Lake 2) “Wetlands”

(Littoral 1) “Macrophytes”

(Littoral 2) “Muckiness”

(Riparian 1) “Structural complexity”

(Riparian 2) “Tall wetland

plants”

Linear regressions

Odonata density 0.33 0.20 0.54 0.20 0.21 0.63

Species richness 0.29 0.22 0.60 0.21 0.26 0.28

Gomphidae density 0.40 0.20 0.42 0.20 0.22 0.73

Logistic regressions

Zygoptera presence 0.23 0.22 0.73 0.33 0.31 0.34

Libellulinae presence 0.38 0.23 0.58 0.22 0.32 0.21

Macrominae presence 0.25 0.45 0.25 0.34 0.59 0.25

Cordulinae presence 0.73 0.23 0.24 0.87 0.32 0.25

Aeshnidae presence 0.34 0.20 0.35 0.95 0.45 0.65 1 The six predictors are synthetic axes derived from three separate principle components

analyses of 12 lake-level, nine site-level littoral, and 10 site-level riparian

variables.

52

Table 8. Analysis of Variance results from a mixed-effects linear

regression of Zygoptera adult abundance (n = 40).

Fixed-effects variables d.f. F p-value Typha treatment 1,51 9.2 0.004 Macrophyte block 1,51 2.8 0.10 Typha x block 1,51 2.8 0.10

53Table 9. McNemar’s Chi-squared contingency tables showing the number of days when

oviposition was observed at treatments with and without potted Typha minima at the

shoreline.

a) Anisoptera oviposition

No Typha1 Yes2 No

Yes 1 6 No 4 10

b) Zygoptera oviposition

No Typha Yes No

Yes 1 4 No 1 15

1 On all 21 days, observations were made at shoreline plots with and without the Typha

treatment.

2 A ‘Yes’ count indicates that oviposition occurred at the specified shoreline treatment.

Typh

a

Typh

a

54Fig. 1. Locations of 41 Odonata larval survey sites on lakes in Vilas County, Wisconsin.

Fig. 2. Biplots of the first two axes from Principal Components Analyses (PCA)

performed on (a) 12 lake-level, (b) 9 site-level littoral, and (c) 10 site-level riparian

habitat variables (abbreviations defined in Table 4). Direction and magnitude of the

arrows illustrate correlations of each habitat variable with the first two PCA axes. PCA

scores for the 41 sites surveyed are shown as circles.

Fig. 3. Gomphidae larval density graphed as a function of the second axis from a

principal components analysis of riparian vegetation (Table 4). As the horizontal axis

increases, the length of riparian sites having wetland herbaceous vegetation decreases.

Fig. 4. Mean adult dragonfly (a) and damselfly (b) abundance estimates (± 1 SE) within

15 x 7 m experimental plots at two lakeshore home sites in northern Wisconsin (one of

the two having dense emergent macrophytes in the littoral zone). Typha treatments,

consisting of 20 potted Typha minima placed on the sea walls, were rotated between two

plots per site on 21 days during June-August 2006.

Fig. 5. Rank-abundance curves (± 1 SE) for Odonata larval assemblages in lake

microsites with and without dense submerged macrophytes.

Fig. 6. Mean densities (± 1 SE) of three larval behavior guilds (Table 2) among paired 7

m2 lake microsites with and without dense submerged macrophytes.

Figure 1.

56Figure 2.

(a)

59Figure 3.

0

0.5

1

1.5

2

2.5

-3 -2 -1 0 1 2

PCA riparian axis 2: Tall wetland plants

sqrt

(Gom

phid

ae d

ensi

ty m

-2)

60 Figure 4.

(a)

0

0.5

1

1.5

2

2.5

Dense macrophytes Sparse macrophytes

Ani

sopt

era

abun

danc

e

No TyphaTypha

(b)

0

1

2

3

4

5

6

7

Dense macrophytes Sparse macrophytes

Zygo

pter

a ab

unda

nce

No TyphaTypha

61 Figure 5.

-2

0

2

4

6

8

10

0 5 10 15 20 25 30

Species rank

Mea

n ab

unda

nce

No Macrophytes

Macrophytes

Epitheca cynosura

62Figure 6.

0

1

2

3

4

Burrowers Claspers SprawlersOdo

nata

larv

al g

uild

den

sity

(per

m^2

)

MacrophytesNo macrophytes

63 Appendix 1. Geographic locations of 41 study sites surveyed in 2004 for Odonata

larvae.

Lake (abbreviation) and site number Latitude LongitudeBig (BG) 3 46.14771 -89.75002 Big (BG) 8 46.16294 -89.77978 Black Oak (BO) 1 46.16587 -89.31061 Black Oak (BO) 4 46.15580 -89.30875 Brandy (BR) 6 45.90317 -89.70406 Brandy (BR) 8 45.90784 -89.70465 Big St. Germain (BSG) 1 45.94415 -89.50678 Big St. Germain (BSG) 3 45.92938 -89.50267 Big St. Germain (BSG) 7 45.92993 -89.54494 Carpenter (CR) 2 45.94619 -89.13996 Carpenter (CR) 5 45.93837 -89.15183 Carpenter (CR) 7 45.94616 -89.15234 Carpenter (CR) 8 45.95045 -89.15099 Jag (JA) 1 46.09044 -89.72128 Jag (JA) 4 46.08059 -89.71660 Jag (JA) 5 46.08175 -89.72340 Jag (JA) 7 46.08747 -89.72798 Jag (JA) 8 46.08780 -89.72626 Little Arbor Vitae (LAV) 2 45.91324 -89.60908 Little Arbor Vitae (LAV)7 45.90893 -89.63440 Laura (LLA) 5 46.05154 -89.44090 Laura (LLA) 6 46.05277 -89.44571 Little St. Germain (LSG) 1 45.92841 -89.42837 Little St. Germain (LSG) 4 45.91356 -89.45675 Little Star (LST) 1 46.11925 -89.85830 Little Star (LST) 2 46.11527 -89.85170 Lynx (LX) 1 46.20764 -89.66808 Pallette (PA) 1 46.19405 -89.50623 Pallette (PA) 7 46.19823 -89.51094 Pallette (PA) 8 46.20645 -89.50517 Papoose (PP) 3 46.18217 -89.79484 Papoose (PP) 6 46.18210 -89.81336 Papoose (PP) 7 46.18678 -89.80814 Towanda (TW) 3 45.93572 -89.70187 Towanda (TW) 8 45.94200 -89.71255 Upper Buckatabon (UB) 1 46.02433 -89.33512 Upper Buckatabon (UB) 4 46.00898 -89.34746 Vandercook (VC) 5 45.97960 -89.68855 Vandercook (VC) 6 45.98056 -89.68917 White Sand (WS) 3 46.08429 -89.57942 White Sand (WS) 8 46.09515 -89.60004

Appendix 2. Densities (benthic m-2) of larval Odonata families and genera at 41 lake sites in Vilas County, Wisconsin, including

riparian vegetation data

Site1 Larval

survey summary Riparian vegetation characteristics2 Sub-

order Families3 Common genera

larv

al sa

mpl

e ar

ea

tota

l abu

ndan

ce

spec

ies r

ichn

ess

cano

py

leng

th w

ith sh

rubs

(m

)

leng

th w

ith C

arex

or

Typh

a (m

)

veg

etat

ion

dist

ance

fr

om w

ater

(m)

vege

tatio

n he

ight

(m)

mea

n st

ruct

ural

di

vers

ity

mea

n fo

liage

hei

ght

dive

rsity

Zygo

pter

a

Aes

hnid

ae

Libe

llulin

ae

Mac

rom

inae

Cor

dulin

ae

Gom

phid

ae

Gom

phus

Dro

mog

omph

us

Hag

eniu

s

Cel

ithem

is

Epith

eca

BG3 12 1 1 3 30 0 0 2.0 2.72 6.89 0.00 0.00 0.00 0.08 0.00 0.00 0.00 0.00 0.00 0.00 0.00 BG8 9 12 5 2 0 25 0 1.7 1.25 2.25 0.00 0.00 0.00 0.00 0.00 1.31 0.24 0.83 0.24 0.00 0.00 BO1 24 22 12 2 5 0 0 2.0 2.39 3.33 0.04 0.04 0.08 0.17 0.13 0.46 0.33 0.13 0.00 0.08 0.13 BO4 24 46 7 2 2 0 0 0.5 1.89 1.72 0.00 0.00 0.33 0.38 0.08 1.13 0.08 1.00 0.04 0.33 0.08 BR6 12 21 5 2 2 29 0 1.8 2.00 2.56 0.00 0.00 0.00 0.00 0.25 1.50 1.50 0.00 0.00 0.00 0.25 BR8 12 36 7 2 0 15 0 1.8 1.65 2.18 0.17 0.00 1.17 0.17 0.00 1.50 1.50 0.00 0.00 1.17 0.00 BSG1 12 2 2 2 2 0 0 2.0 1.61 2.89 0.00 0.00 0.00 0.00 0.00 0.17 0.17 0.00 0.00 0.00 0.00 BSG3 12 4 4 3 2 0 0 2.0 1.83 2.28 0.08 0.00 0.00 0.00 0.00 0.25 0.17 0.00 0.00 0.00 0.00 BSG7 12 10 6 2 2 0 0 2.0 1.71 2.29 0.17 0.00 0.00 0.00 0.00 0.58 0.17 0.00 0.08 0.00 0.00 CR2 12 50 7 2 4 6 1 2.0 1.94 3.61 0.08 0.00 0.17 0.00 0.00 3.92 3.83 0.00 0.08 0.17 0.00 CR5 12 10 4 2 25 22 0.6 2.0 3.11 5.11 0.00 0.00 0.00 0.00 0.00 0.83 0.75 0.08 0.00 0.00 0.00 CR7 12 5 2 3 0 0 1 0.2 1.29 1.86 0.00 0.00 0.00 0.00 0.00 0.42 0.25 0.17 0.00 0.00 0.00 CR8 12 65 7 2 0 6 0 0.5 1.73 2.33 0.08 0.00 0.00 0.25 0.00 5.08 4.67 0.17 0.00 0.00 0.00 JA1 24 41 10 2 30 1 2.2 2.0 3.56 6.22 0.04 0.00 0.00 0.17 0.00 1.46 1.04 0.00 0.42 0.00 0.00 JA4 16.8 12 7 3 29 2 3.8 2.0 2.61 5.67 0.06 0.00 0.06 0.06 0.00 0.54 0.42 0.12 0.00 0.06 0.00 JA5 12 4 3 3 30 15 0 2.0 2.83 5.17 0.00 0.00 0.00 0.08 0.00 0.25 0.17 0.00 0.08 0.00 0.00 JA7 12 30 9 3 25 5 3.4 2.0 2.72 6.00 0.42 0.00 0.00 0.42 0.00 1.67 0.75 0.17 0.75 0.00 0.00 JA8 12 8 6 3 15 5 2.2 2.0 2.78 5.72 0.00 0.00 0.00 0.00 0.08 0.58 0.42 0.00 0.17 0.00 0.08 LAV2 10.8 14 5 3 2 30 0 2.0 3.28 6.44 0.00 0.09 0.00 0.00 0.83 0.28 0.00 0.00 0.00 0.00 0.83

LAV7 10.8 6 3 3 30 0 0 2.0 2.72 6.50 0.19 0.00 0.00 0.09 0.00 0.19 0.19 0.00 0.00 0.00 0.00 LLA5 12 35 8 2 8 8.4 0 2.0 1.91 3.27 0.08 0.00 0.00 0.33 0.00 2.33 1.17 0.67 0.50 0.00 0.00 LLA6 12 16 5 3 8 9 0.8 2.0 2.17 4.56 0.00 0.00 0.00 0.00 0.00 1.33 0.83 0.25 0.25 0.00 0.00 LSG1 12 36 5 3 5 30 0 2.0 2.94 5.33 0.08 0.00 0.00 0.00 0.83 2.08 2.08 0.00 0.00 0.00 0.83 LSG4 12 9 4 2 0 0 0 0.2 1.41 1.82 0.00 0.00 0.00 0.00 0.25 0.50 0.42 0.00 0.00 0.00 0.17 LST1 12 1 1 1 0 0 0 0.7 1.67 1.17 0.00 0.00 0.00 0.00 0.00 0.08 0.00 0.08 0.00 0.00 0.00 LST2 12 2 2 3 30 2 0 2.0 3.17 6.44 0.00 0.00 0.00 0.00 0.00 0.17 0.17 0.00 0.00 0.00 0.00 LX1 16.8 69 16 3 25 29 1.9 2.0 2.39 4.72 0.30 0.00 0.18 0.00 0.30 3.33 1.90 1.43 0.00 0.18 0.00 PA5 9 0 0 3 30 30 0 2.0 0.00 6.17 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 PA7 12 6 1 3 30 30 0 2.0 2.28 5.17 0.00 0.00 0.00 0.00 0.00 0.50 0.00 0.00 0.50 0.00 0.00 PA8 12 3 3 3 30 10 1 2.0 2.61 4.44 0.00 0.08 0.00 0.00 0.00 0.17 0.08 0.00 0.08 0.00 0.00 PP3 12 4 3 1 0 0 0 0.6 1.72 3.33 0.00 0.00 0.00 0.00 0.00 0.33 0.17 0.00 0.17 0.00 0.00 PP6 9 0 0 1 4 2 0 1.5 1.76 2.88 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 PP7 12 9 5 3 28 0 0 2.0 2.28 4.44 0.00 0.00 0.00 0.17 0.00 0.75 0.25 0.08 0.42 0.00 0.00 TW3 12 39 9 1 7 7 0.5 2.0 1.75 1.13 0.25 0.00 0.17 0.00 0.00 2.83 2.75 0.08 0.00 0.00 0.00 TW8 12 10 4 2 2 2 0.5 1.8 0.00 2.78 0.00 0.00 0.00 0.00 0.17 0.67 0.67 0.00 0.00 0.00 0.17 UB1 12 5 3 3 20 0.5 0.1 2.0 2.83 6.78 0.00 0.08 0.00 0.00 0.00 0.33 0.25 0.00 0.08 0.00 0.00 UB4 12 2 2 1 2 2 0.2 0.9 1.50 1.50 0.08 0.08 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 VC5 12 25 10 3 20 2 0.2 2.0 1.67 3.50 0.08 0.08 0.08 0.00 0.00 1.75 1.42 0.08 0.25 0.08 0.00 VC6 12 51 7 1 2 30 1.2 1.5 2.00 4.39 0.00 0.08 0.00 0.00 0.00 4.17 3.83 0.17 0.08 0.00 0.00 WS3 12 15 6 3 30 0 0.5 2.0 2.72 6.28 0.25 0.00 0.00 0.17 0.00 0.83 0.00 0.67 0.17 0.00 0.00 WS8 12 14 6 2 30 25 0.7 2.0 2.83 5.72 0.25 0.00 0.00 0.25 0.00 0.67 0.08 0.58 0.00 0.00 0.00

1 Lake abbreviations are shown in Appendix 1. 2 Habitat variables are described in Table 3. 3 Species included in each family are listed in Table 4.

66

CHAPTER III

Gomphidae densities among three life stages: Oviposition site selection overrides

post-recruitment spatial variation

Abstract

Many animals have life stages that occupy different habitats. Although

population demographics in one stage can carry over to subsequent – even spatially

separated – life stages, most studies of habitat associations have been restricted to a

single life stage. Dragonflies (Odonata: Anisoptera) require both aquatic and terrestrial

habitats for different life stages, but habitat heterogeneity may not contribute equally to

the densities of both stages. Although recruitment via adult oviposition establishes initial

population sizes of the aquatic larvae, spatial variability in larval survivorship could

obscure the expected correlation between adult and larval densities. I used surveys of

Gomphidae larvae, exuviae, and adults from 22 lake sites in northern Wisconsin, USA, to

investigate (1) whether the density of each life stage was correlated with that of the

preceding life stage and (2) what habitat factors help explain variation in densities at each

life stage.

`I found strong positive correlations between densities of first-year larvae and

adults from the previous season, as well as between second-year larvae and exuviae later

during the same year. Adult densities did not correlate with exuvia densities from the

same year. In addition to parameters for the previous life stages, water clarity helped

predict larval densities, and riparian wetland vegetation helped predict exuvia densities.

67The results demonstrate that habitat features affecting adults or larvae can transcend the

aquatic-terrestrial boundary to influence population densities of the subsequent life stage.

Introduction

Ecologists have long described the natural processes that limit population sizes

(e.g., Lindeman 1942, Davidson and Andrewartha 1948, Cole 1954, Lack 1954), but have

only recently begun to examine how spatially explicit population stage structures affect

population dynamics (reviewed by Halpern et al. 2005). The vast majority of

multicellular animals have separate life stages that occur in different habitat types. These

include animals with stages separated by size or age structure (e.g., Ebenman and Persson

1988, Persson and De Roos 2006), as well as animals with complex life cycles, in which

animals undergo a succession of discrete life history phases (e.g., Istock 1967, Moran

1994). Despite the ubiquity of these life stage separations, population studies

incorporating habitat spatial heterogeneity have been mainly restricted to a single life

stage (reviews by Wilbur 1980, Werner et al. 2001). More recent studies have

demonstrated that spatial variation in population demographics at one life stage can affect

population sizes in subsequent stages (e.g., Roughgarden et al. 1988, Hellriegel 2000,

Hughes et al. 2000, Mumby et al. 2004, Halpern et al. 2005), but have not included

aquatic insects or other animals with longer larval than adult phases. The semi-voltine

(multi-year) life history of many temperate dragonflies (order Odonata, suborder

Anisoptera) could weaken population density correlations that are expected between

different life stages. Understanding the relative influence of different life stages on

68population demographics is critical for predicting how modifications to either habitat

could affect animals with spatially separated life stages.

The utility and attraction of land-water interfaces have resulted in human

alterations to riparian areas for several millennia, with riparian development continuing to

expand (WDNR 1996, Naiman and Decamps 1997, Burger et al. 2004, Hansen et al.

2005). Because biodiversity is often particularly high in these land-water ecotones

(Decamps et al. 2004), ecological consequences of habitat change can be

disproportionately high in riparian areas. Lakeshore housing development is expanding

rapidly in many areas and is associated with simplification of vegetation structure by

removal of the riparian shrub layer, floating macrophytes, and littoral coarse wood

(Racey and Euler 1982, Radomski and Goeman 2001, Elias and Meyer 2003, Marburg et

al. 2006). These structural changes correlate with reduced biodiversity and productivity

across multiple taxa in both the riparian and littoral (near-shore aquatic) zones (Bryan

and Scarnecchia 1992, Schindler et al. 2000, Lindsay et al. 2002, Henning and Remsburg

in review).

Although dragonflies require both suitable aquatic habitat for larvae and terrestrial

habitat for adults, the local abundances of dragonflies may not depend on each of these

habitats equally. As with many animals, dispersal occurs during the more-mobile adult

dragonfly stage, whereas survivorship during the more-vulnerable juvenile stage limits

population sizes (Johnson 1986, Wissinger 1988). The relative importance of recruitment

versus larval survival, however, remains unknown. Dragonfly adults are outstanding

fliers, enabling them largely to avoid their avian and amphibian predators and to disperse

hundreds of kilometers (Corbet 1999). Recruitment via adult oviposition thus establishes

69the initial spatial patterns of aquatic insect abundances. Dragonfly larvae, in contrast,

probably do not travel > 20 m in lentic systems because movement makes them more

susceptible to predators (Ubukata 1984, Schaffner and Anholt 1998, but see Alzmann et

al. 1999). Larval survivorship may vary among sites or lakes due to heterogeneity in

post-recruitment processes such as larval movement, competition, and predation

(Forrester 1990, Pfister 1996, Stoks and McPeek 2003). Spatial patterns of local

dragonfly population abundances therefore depend on the relative importance of adult

habitat site selectivity and the spatial heterogeneity of larval survivorship (Fig. 1).

Abundance and diversity of adult dragonflies increase in riparian areas with high

vegetation structural complexity, although the underlying reasons for these patterns

remain unclear (Van Buskirk 1986, McKinnon and May 1994, Carchini et al. 2003, Foote

and Hornung 2005, Hofmann and Mason 2005, Ward and Mill 2005). Gomphidae larvae

in northern Wisconsin lakes are associated with riparian sites having tall wetland

vegetation such as sedges (Chapter II). Among riparian and littoral habitat

characteristics, adult odonates often demonstrate strongest associations with riparian

sedges and bulrushes (Van Buskirk 1986, Clark and Samways 1996, Foote and Hornung

2005). Mature Gomphidae split their time at riparian sites between reproductive and

foraging behaviors, perching on the ground or on vegetation to watch for prey or

potential mates (May 1976, Mead 2003). The Gomphidae species considered in this

study copulate near lotic or sufficiently oxygenated lentic water and then oviposit directly

onto the surface of open water (exophytically). As with other oviparous species, it is

advantageous for adults to oviposit at sites with the greatest likelihood for survival of

their eggs and larvae (Resetarits 1996). Because adults may not be able to assess all

70aspects of aquatic habitat suitability, several authors have hypothesized that adult

odonates use emergent aquatic plants or riparian plants as proximate cues of macrophyte,

substrate, water depth, prey, or predator conditions in littoral sites (Buchwald 1992,

Wildermuth 1992 as reviewed by Corbet 1999, McKinnon and May 1994).

Alternatively, foraging odonates may be attracted to higher prey abundances in riparian

areas with high structural complexity (Baird and May 1997).

Larval odonates face high rates of predation by fish or larger odonates, and to a

lesser degree by turtles, frogs, birds, and crayfish (Wright 1946, Kennedy 1950,

Larochelle 1977, Crowder and Cooper 1982, Johnson 1991, Momot 1995, Stoks and

McPeek 2003). Mortality of larval odonates may be as high as 95-99 % (Johnson 1986,

Duffy 1994, Schutte et al. 1998). The Gomphidae larvae in this study hide from predators

by burrowing (burying themselves) in the benthic substrate, but this behavior can also

reduce their foraging efficiency (e.g., Sih 1980, Lima 1998, Alzmann et al. 1999,

Johansson 2000). Burrowing reduces their dependence on macrophytes for cover,

although burrowers may occupy macrophyte microsites (Mahato and Johnson 1991,

Alzmann et al. 1999). While moving from their typical littoral habitats (1- to 2- m water

depths [Remsburg, unpublished data], and up to 6 m depth (Thomas 1965)) onto land for

emergence, larvae may face increased risk of fish predation. The one – two hour process

of larval emergence and wing hardening is a final period of susceptibility to terrestrial

predators such as birds, spiders, and ants (Wissinger 1988, Orians and Wittenberger

1991, Duffy 1994, Jakob and Suhling 1999). High wind, rain, waves, or trampling can

also significantly reduce rates of successful emergence (Gribbin and Thompson 1990,

Jakob and Suhling 1999).

71

Heterogeneity in aquatic or riparian vegetation structure could interact with

trophic or environmental controls on Gomphidae abundances. For example, shoreline

vegetation abundance can lead to greater spider abundances (Scheidler 1990, Langellotto

and Denno 2004) and greater susceptibility of emerging odonates to damage from wind

(Jakob and Suhling 1999). Lower macrophyte abundances, higher insectivorous fish

abundances, or higher water clarity could each lead to higher predation on larvae

(Crowder and Cooper 1982, Thompson 1987, Diehl 1992, Aksnes and Giske 1993) or

reduced prey availability for larvae (Alzmann et al. 1999, Bazzanti et al. 2003). Riparian

forest understory and aquatic transition plants may provide adults with perching

structures for thermoregulation, foraging, territory defense, mate attraction, copulation,

nocturnal roosting, or protection from adverse weather (Buchwald 1992, Wildermuth

1993, McKinnon and May 1994, Rouquette and Thompson 2007).

Based on potential drivers of odonate distributions during the adult, larval, and

emergence stages, I investigated two main research questions: (1) Is there a positive

correlation between the density of one Gomphidae life stage and its previous life stage at

either lake- or within-lake- spatial scales? and (2) What habitat factors help explain

variation in Gomphidae larvae, exuviae, and adult densities? I hypothesized that if larval

densities depend most on recruitment and larvae do not move significant distances, then

first-year larval densities would correlate with adult densities from the previous year. On

the other hand, significant differences in larval survivorship among sites would obscure

the hypothesized relationship between adult abundances and their offspring. Aquatic or

terrestrial habitat features relevant to each life stage may help explain the variation in

Gomphidae densities not accounted for by the previous life stage.

72

Methods

Study sites

I conducted the study at 22 sites on 11 lakes in Vilas County, northeastern Wisconsin

(USA). This mixed-forest landscape contains over 1300 glacial lakes in the 2639 km2 county

(see Peterson et al. 2003 for more details on the region). Most of the study sites were privately

owned; 14% were public property. I selected two sites per lake, separated by at least 200 m,

with similar aspect, slope, fetch, macrophyte density, benthic substrate, and hydrologic context

(location relative to streams or bays) for each pair (Table 1). Each site spanned 30 m along the

shoreline by 5 m inland. The pair of sites on each lake represented opposite extremes of

human influence: manicured lawns and forests. Forest sites had a dense layer of shrubs

(mainly Alnus rugosa), saplings (e.g., Betula papyrifera, Thuja occidentalis, Acer rubrum), or

wetland herbaceous plants (e.g., Carex retrorsa, Juncus effusus, Sparganium spp.), and were

generally undeveloped. Manicured lawns had minimal understory vegetation and were

associated with housing inland from the study sites.

Odonata surveys

For a 30-m transect paralleling the shoreline, I collected all Odonata exuviae on

shore within approximately 0.5 m of the water’s edge. I collected exuviae on three days

per site in 2005 (May 27 - July 12) and four days per site in 2006 (June 9 – July 12). I

visited both sites from the same lake during the same day, and completed survey rounds

for nearly all sites before beginning the next round of surveys. The last two rounds of

2005 exuviae sampling on one lake (Little Arbor Vitae) took place within four days of

73each other, but all other survey rounds were separated by at least a week. Exuviae were

collected by careful inspection (on hands and knees) of the ground and vegetation for 15

– 45 minutes per survey day. Time required for searching a site exhaustively varied with

understory plant density and abundance of odonates. Exuvia and adult surveys in 2006

took place at 21 sites because one of the study sites (on Big St. Germain lake) became

inaccessible (due to owner concerns about liability).

Three adult odonate surveys per site (per year) took place between May 31 - July

27 in 2005 and June 15 – July 21 in 2006. I estimated adult odonate densities along the

same 30 m shoreline transects using a modified ‘Pollard walk’ (Pollard 1977, Walpole

and Sheldon 1999). While walking the transect over a period of two minutes, I counted

and identified all adult odonates perched or flying within 2 m around me (but not behind)

and up to 3 m above the ground. If an individual clearly flew back and forth through the

observation zone, it was only counted once. However, individuals could not be tracked if

they flew out of the observation zone and later returned, so observed abundances may be

slightly higher than actual abundances. I repeated the two-minute transect walk three or

four times on each survey day. I conducted all adult surveys on relatively calm, sunny

days between 11:00 and 16:00, when odonate flight activity is greatest (Lutz and Pittman

1970, Moore and Corbet 1990).

I surveyed larval odonates on one occasion per site during the first week of June

2006 in the littoral zone immediately adjacent to the 22 sites. At each site, I used a D-

frame net to scoop benthic material along four 16 m contour transects that paralleled the

shore. I centered transects within the 30 m site length and placed them at 0.25 m, 0.5 m,

0.75 m, and 1.0 m water-depths; transect distance from shore varied with littoral slope at

74each site. I placed a 0.95m – diameter cylinder (i.e., a trash can with the bottom cut

off) at 20 subsample locations per site (4 in each of the two shallower transects and 6 in

each of the two deeper transects) and collected four equivalent scoops of benthos within

each subsample. Because the cylinder extended from the substrate to the water’s

surface, this protocol ensured that larvae within each subsample area could not move out

of the area while I dipped the D-net (Caudill 2002).

For analysis I combined each site’s 20 subsamples, which together represent a

14m2 area surveyed per site. I separated larvae into two year classes based on head

widths (W. Smith, in prep). The Gomphidae species surveyed generally have two-year

larval phases in Wisconsin, but no sexual size dimorphism. A histogram of head widths

revealed a distribution that was approximately bimodal (Fig. 2). The range of head

widths was 0.5 to 6.0 mm; based on the distribution, I designated all individuals with

head widths less than or equal to 3.0 mm as first-year larvae and larger individuals as

second-year larvae.

Habitat variables

I estimated the relative abundances of riparian tall wetland herbaceous plants (>

0.3 m high; Carex, Typha, and Iris species), and littoral macrophytes among sites by

measuring the proportion of site shorelines (out of 30 m) having these vegetation

categories present within 4 m of the water’s edge. Lake-level variables included Secchi

depth (water clarity), insectivorous fish catch-per-unit-effort (CPUE), and rusty crayfish

(Orconectes rusticus) CPUE. Rusty crayfish density is likely to correlate inversely with

lake-level macrophyte abundance because this invasive species consumes large amounts

75of macrophytes (Wilson et al. 2004). Lake-level data were compiled by Carpenter et al.

(2006) and Olden et al. (2006).

Data analysis

All analyses focus on Gomphidae species because this was the only group with

larvae occurring at > 50 % of the study sites. Previous larval surveys suggested that

Gomphidae is by far the most abundant Odonata family in these lakes (Chapter II). In

addition, I focused on only the “burrowers,” excluding one Gomphidae species that falls

into the “hider” larval guild, Hagenius brevistylus (Corbet 1999). Co-occurrence of the 6

burrower Gomphidae species observed (Arigomphus furcifer, Dromogomphus spinosus,

Gomphus exilis, G. fraternus, G. lividus, and G. spicatus) appears to be very high

(personal obs.), which follows a pattern of minimal niche separation among many

Gomphidae species (Crowley and Johnson 1982, Burcher and Smock 2002). The

Gomphidae species considered in this study all emerge fairly synchronously in mid-June

(Mead 2003), thus restricting the number of weeks required for exuvia and adult surveys.

The main flight period lasts between two and six weeks for most Gomphidae in this

region (Mead 2003).

Adult density estimates were derived from the total number of individuals

counted during survey transects divided by the transect length (30 m). I averaged the

adult densities from transect walks on all three survey dates in 2006. Surveys in 2005,

however, took place over a longer time frame, making adult abundances difficult to

compare among lakes that were sampled during different parts of the dragonfly’s flight

season. Gomphidae adults were hardly ever present before June 10 or after July 10,

762005, so I used data from only the single survey date that took place at each site during

this time period. This required that I exclude two lakes (Lynx and Little Arbor Vitae)

from the 2005 adult analyses that were not surveyed between those optimal flight period

dates.

I used mixed-effects linear regressions to test effects of adult, larvae, and exuviae

densities on densities of the subsequent life stages among sites. I assigned lake as a

random-effect term because effects of specific lakes were not of interest, but dragonfly

densities at sites on the same lakes were not necessarily independent. To compare life

stage densities at the lake level, I averaged Gomphidae densities from both sites at each

lake and then conducted Spearman’s rank correlation tests. Gomphidae densities for all

life stages except 2006 adults were log-transformed to meet normality assumptions of the

test. Because some larval and adult densities were zero, I added the smallest non-zero

density value before taking the log and then an order-of-magnitude constant (McCune

and Grace 2002). Variance partitioning of random effects revealed what proportion of

the total variation in Gomphidae densities occurred at the lake versus site level.

To test the influence of habitat effects on each life stage, I added five habitat

variables individually to the linear regressions that also included previous life stages.

Site-level models used lake as a random-effect parameter, whereas the lake-level models

considered only fixed effects. At the site level, I tested five candidate models for first-

year larval densities: 2005 adult densities alone and combined individually with site

macrophyte abundance, lake Secchi depth, lake insectivorous fish CPUE, and lake rusty

crayfish CPUE. Models for first-year larvae averaged by lake included all of the above

lake-level habitat variables in addition to 2005 adult densities averaged by lake. To

77predict 2006 exuviae densities at the site level, I considered six candidate models:

second-year larval densities alone and combined individually with lake Secchi depth, lake

insectivorous fish CPUE, lake rusty crayfish CPUE, site development category (forest or

manicured lawn), and site abundance of riparian wetland vegetation. I also tested effects

of the three lake-level predictors (in combination with lake-level larval densities) on

exuviae densities at the lake level. Predictive models for adult abundances in both 2005

and 2006 were completed only at the site level and included four candidate models for

each year: exuviae densities of the same year alone and in combination with site

development, site riparian wetland vegetation, and site macrophytes. In addition, I tested

each of these three habitat variables alone (without the previous life stage) as predictors

of adult densities.

I used an information theoretic approach to compare alternative mixed-effects

habitat models within the set of candidate models for each life stage. In particular, I

ranked models according to Akaike’s Information Criterion (AICc) values. AICc has a

bias correction factor to help prevent over-parameterized models when sample sizes are

small (Burnham and Anderson 2002). Relative empirical support for a model i is

indicated by ∆i, the difference between AICc for model i and the smallest AICc of the

candidate models. Information theory suggests that datasets may equivalently support

multiple predictive models (particularly if ∆i < 2).

I used likelihood ratio tests to compare each habitat model with the linear model

that used only the previous life stage to predict Gomphidae densities. I also calculated

adjusted R2 values for all linear regressions, but the maximum likelihood model fitting

method employed for the mixed-effects models could not be used to calculate the R2

78values. Thus adjusted R2 values were calculated from fixed-effects models (using

identical parameters selected from mixed-effects models) where lake was added as a

fixed effect. All analyses were completed using R v.2.4.1 software (R Core Development

Team, Vienna, Austria), including the nlme package for mixed-effects models.

Results

All three life stages had high variability among the 22 sites sampled (Table 2).

Variance partitioning revealed that nearly all of the variance in first-year larval densities

occurred at the lake level, whereas second-year larvae, exuviae, and adults each had

progressively smaller proportions of the variance occurring at the lake level (Table 2).

First-year larval densities in 2006 were positively correlated with adult densities

in 2005, at both the site and lake scales (Table 3; Fig. 3). Exuviae in 2006 were

positively correlated with second-year larval densities from earlier in the same season

(Table 3; Fig. 4). Adult densities did not correlate with exuviae densities from the same

season in 2005 or 2006.

Of the three aquatic habitat variables added to the site-level model to predict first-

year larval densities, Secchi depth was the only variable that contributed significantly in

addition to previous-year adult densities (Table 4). Secchi depth correlated negatively

with larval density, indicating that larval densities were lower in lakes with clearer water

(Fig. 5). From the set of six candidate models for prediction of 2006 exuviae densities,

both macrophyte abundance and riparian tall wetland vegetation improved the model in

addition to 2006 second-year larvae (Table 4). Macrophyte abundances correlated

negatively with exuviae densities, whereas riparian wetland vegetation correlated

79positively with exuviae densities. For both larvae and exuviae densities at the lake

level, densities of the previous life stage alone served as the best predictor; models that

included habitat variables all had much higher ∆AICc The best predictor of adult

densities in 2005 was riparian wetland vegetation with 2005 exuviae density, although

this model was not significantly better than the model using only the previous life stage

(Table 4). Model selection indicated that macrophyte abundance alone was also

positively correlated with adult densities in 2005, but the model was only marginally

significant (fixed-effects adj. R2 = 0.32, mixed-effects model F1,8 = 4.4, p = 0.07). None

of the habitat variables considered helped predict adult densities in 2006 (Table 4).

Discussion

First-year larval densities varied considerably among survey sites (Table 2); most

of this variation can be attributed to the adult phase during the year that these larvae were

oviposited (Table 4, Fig. 1). Positive correlations between first-year larval densities and

adult densities in the previous year suggest that oviposition site selection controlled the

local larval distribution more than spatial variations in survivorship or movement during

the first year of the larval phase. The relationship is more significant when considering

lake-level averages of adult and larval densities (Table 3, Fig. 3), probably because

breeding adult Gomphidae cover broader areas than the 30-m shoreline sites. The

strength of this linear relationship is noteworthy because a number of other factors could

have obscured the positive linear relationship: measurement error, males present at lakes

where they did not copulate, loss of eggs to predation or water currents, larval site

selection, or heterogeneity in larval survivorship.

80

The demonstrated relationship between larvae and adults of the previous

generation does not diminish the role of population controls during the odonate larval

phase, but instead suggests that larval population controls have relatively similar effects

among lakes and sites. Thus larval movement and predation on eggs and first-year larvae

apparently did not significantly diminish the population density spatial patterns

established by adult oviposition site selection. Larval mortality or movement, however,

could influence larval densities at a finer spatial scale than the 14 m2 littoral sites I

surveyed. Finally, larval mortality among odonates may be highest during the later

instars because fish prey most heavily on larger odonates (Martin et al. 1991, Duffy 1994,

Schutte et al. 1998). Therefore, to address the complete effects of heterogeneity in larval

survivorship or movement, it would be informative to track densities of multiple larval

instars during subsequent years.

Instead of supporting relatively homogeneous pressures on odonate larvae at the

site level, results of this study may indicate that adults appropriately select sites where

larvae are most likely to survive. This process would follow the balanced distribution

model, in which discrete habitat patches function simultaneously as sources and sinks for

adjacent patches (McPeek and Holt 1992, Doncaster et al. 1997). Odonates are likely

candidates for balanced dispersal because they can easily disperse among potential

breeding sites. Further, Gomphidae species are apparently not territorial (Corbet 1999),

so site selection is largely unconstrained. It is difficult to assess whether adults

accurately perceive the heterogeneity in larval site suitability within lakes, but failure of

odonates to discern fish versus fishless ponds (McPeek 1989) makes this seem unlikely.

81

Most of the tremendous variation in exuviae densities among sites (Table 2) can

be attributed to site-level densities of the previous life stage (second-year larvae Table 3).

Unlike the relationship between first year larvae and adults, this positive correlation did

not occur at the lake level (Table 3). This finding supports the hypothesis that larvae do

not move very far during the weeks before they emerge (Ubukata 1984). Although larval

odonate mortality may be highest during later instars (Martin et al. 1991, Duffy 1994,

Schutte et al. 1998), the strength of the site-level correlation among stages also suggests

that variation in predation on final-instar larvae among sites is small relative to the

variation in recruitment of second-year larvae. I cannot make inferences about predation

on teneral dragonflies (immature adults that have recently shed their larval skins), though,

because exuviae reflect teneral densities only at the time of emergence, before tenerals

are subject to predation.

Absence of correlations between adult Gomphidae densities and corresponding

exuviae densities in 2005 or 2006 at the site or lake scales (all p > 0.1) reflects the high

mobility and temporal variability of adults. Assuming that adult density estimates were

comparable among sites (based on equivalent weather, timing, and survey effort), these

results also suggest that Gomphidae do not return to their emergence sites for

reproduction (i.e. they lack ‘homing’ or ‘philopatry’). The odonate species

demonstrating philopatry depend on temporary lentic environments for larval survival

(Utzeri et al. 1984, McPeek 1990). Studies of marked individuals (e.g., Michiels and

Dhondt 1991, Baird and May 1997, Angelibert 2003) would be necessary to assess site

attachment, including philopatry, but have not yet been conducted on adult Gomphidae.

82 Inclusion of the previous life stage in predictive models allowed me to

investigate habitat factors that affect only the life stage modeled (research question 2).

One aquatic habitat variable, Secchi depth, improved models of larval densities (Table 4).

Lower larval densities in lakes with greater water clarity (Fig. 5) may result from

improved visibility of odonates to their predators in these conditions (Aksnes and Giske

1993, Davies-Colley and Smith 2001). The negative relationship between macrophyte

abundances and exuviae densities (Table 4) probably reflects a sampling bias: I did not

sample exuviae from emergent aquatic macrophytes, which can also serve as effective

substrates for emergence (Corbet 1999). More notable was the positive relationship

between exuviae densities and abundance of riparian wetland vegetation, after densities

of second-year larvae are taken into account (Table 4). This result suggests that

emerging larvae preferentially seek areas with rigid wetland plants (such as Carex or Iris

species) on the shoreline for emergence, or that more larvae emerge successfully in these

areas. The rigid, vertical structures of wetland vegetation may offer emerging odonates

protection from predators and/or less susceptibility to wind damage than tall grasses.

However, Gomphidae can also emerge on horizontal structures, including the ground

(Corbet 1999), so it remains unclear whether riparian wetland vegetation supports greater

emergence rates. An alternative explanation for the positive relationship with wetland

vegetation is that exuviae may remain intact there longest after emergence, due to

protection provided by the wetland vegetation from weathering or trampling damage.

With or without inclusion of exuviae densities in the models, adult densities did

not relate significantly to the habitat variables measured (Table 4). Littoral macrophytes

were positively associated with higher adult densities in 2005 (Table 4), which may result

83from adults perching on emergent macrophytes or selecting those sites for oviposition.

To understand use of submerged and emergent macrophytes by adult and emerging

Gomphidae, behavior experiments in the field will be important.

The riparian development category, which distinguished forest versus lawn sites,

did not correlate with exuvia or adult odonate densities (Table 4). Previous studies

relating adult odonates to riparian vegetation indicated positive relationships between

riparian woody vegetation and odonate species richness, rather than abundance (Rith-

Najarian 1998, Sahlen 1999, Hofmann and Mason 2005). Odonate species other than the

Gomphidae considered here may therefore respond to lakeshore development. This study

suggests that Gomphidae are more sensitive to water clarity, littoral macrophytes, and

riparian wetland vegetation than to clearing of the forest understory for creation of lawns.

However, water clarity, littoral macrophyte abundance, and riparian wetland vegetation

abundance can each be reduced by the landscaping practices of riparian homeowners

(Racey and Euler 1982, Radomski and Goeman 2001, Elias and Meyer 2003, Carpenter

et al. 2007).

In addition to biological understanding, results of the first research question I

addressed provide methodological insights for surveys of odonates and other animals

with complex life cycles. Studies of only one life stage may not accurately assess

population viability (e.g., Roughgarden et al. 1988, Beck et al. 2001, De Block and Stoks

2005). Most publications on odonate habitat include surveys of only one life stage (but

see Muller 2003, D'Amico et al. 2004, Foote and Hornung 2005, Hofmann and Mason

2005), probably because sampling different life stages each have distinct advantages and

disadvantages. Larval presence indicates successful oviposition and is relatively stable

84through time, due to relatively low larval mobility and the multi-year (semivoltine)

larval periods of many temperate species. Unfortunately, density estimates for the

‘clasper’ guild of larvae, which hide among macrophytes and coarse wood, can be

underestimated without severe disturbance to the habitat (personal observation).

Sampling larvae is also more time-consuming than sampling other life stages because it

requires manual searching and separation of larvae from benthic material. Exuviae can

easily be collected by nearly anyone and later identified to species. The main drawback

with sampling these fragile exoskeletons is that they remain visible for short – but

variable – time periods before rain or trampling destroy them. In addition, phenological

differences among species make community-level surveys highly dependent on survey

dates. Higher variability observed from exuvia surveys than surveys of second-year

larvae (Table 2) likely resulted from this temporal component. As with exuviae, adults

can only be sampled during a few weeks following species emergence. Adult behavior

(and therefore conspicuity) also varies considerably with daily weather conditions. An

advantage to sampling adult males is ease of identification. Unless copulation or

oviposition are observed, though, adult abundances may not correspond with breeding

populations. The strong correlation between larvae and exuviae (Table 3) indicates that

Gomphidae exuvia surveys closely represent the densities of larvae in the adjacent littoral

zone. Densities of adults may correspond with breeding sites, but cannot serve as a proxy

for densities of successfully emerging larvae from the area.

Aquatic insects and amphibians, with both spatially segregated and complex life

histories, play a unique ecosystem role by linking their aquatic and terrestrial habitats

between life stages. They provide substantial nutrient and prey subsidies to adjacent

85ecosystems when they transition from water to land during metamorphosis (Likens

1985, Bataille and Baldassarre 1993, Polis et al. 1997, Baxter et al. 2005, Regester et al.

2006). Incorporation of these “spatial subsidies” from external ecosystems has recently

led to more complete understandings of organism diversity, abundance, and dynamics in

a variety of systems (Power and Dietrich 2002, Polis et al. 2004). Positive correlations

demonstrated here between odonate densities and those of the previous life stage suggest

an additional process by which animals with complex life histories link terrestrial and

aquatic ecosystems: habitat selection by adults affects the local abundances of larvae in

the adjacent ecosystem. More generally, the strength of odonate density correlations

across terrestrial-to-aquatic and aquatic-to-terrestrial transitions indicates that population

dynamics of one life stage carry over to spatially separated life stages and subsequent

generations.

Acknowledgments

This research was supported by NSF Biocomplexity Program DEB-0083545,

North-temperate lakes LTER, an NSF Graduate Research Fellowship, the Garden Club of

America, and the Animal Behavior Society. M. Turner and staff at the University of

Wisconsin Trout Lake station provided excellent logistical support. Thanks also to W.

Smith for sharing expertise on Gomphidae behavior. M. Turner, B. Peckarsky, A. Ives, J.

Zedler, C. Gratton, and S. Carpenter helped guide and critique the manuscript. M. Theis,

P. Winkler, B. Henning, D. Hong, J. Iacarella, C. Leugers, and A. Duong helped collect

data.

86Literature Cited Aksnes, D. L., and J. Giske. 1993. A theoretical-model of aquatic visual feeding.

Ecological Modelling 67:233-250.

Alzmann, N., B. Kohler, and G. Maier. 1999. Spatial distribution, food and activity of Gomphus pulchellus Selys 1840 (Insecta; Odonata; Gomphidae) from a still water habitat. International Review of Hydrobiology 84:299-313.

Angelibert, S. 2003. Dispersal characteristics of three odonate species in a patchy habitat. Ecography 26:13-20.


Bataille, K. J., and G. A. Baldassarre. 1993. Distribution and abundance of aquatic macroinvertebrates following drought in 3 prairie pothole wetlands. Wetlands 13:260-269.

Baxter, C. V., K. D. Fausch, and W. C. Saunders. 2005. Tangled webs: reciprocal flows of invertebrate prey link streams and riparian zones. Freshwater Biology 50:201-220.

Bazzanti, M., V. Della Bella, and M. Seminara. 2003. Factors affecting macroinvertebrate communities in astatic ponds in central Italy. Journal of Freshwater Ecology 18:537-548.

Beck, M. W., K. L. Heck, K. W. Able, D. L. Childers, D. B. Eggleston, B. M. Gillanders, B. Halpern, C. G. Hays, K. Hoshino, T. J. Minello, R. J. Orth, P. F. Sheridan, and M. R. Weinstein. 2001. The identification, conservation, and management of estuarine and marine nurseries for fish and invertebrates. BioScience 51:633-641.

Bryan, M. D., and D. L. Scarnecchia. 1992. Species richness, composition, and abundance of fish larvae and juveniles inhabiting natural and developed shorelines of a glacial Iowa lake. Environmental Biology of Fishes 35:329-341.


Burcher, C. L., and L. A. Smock. 2002. Habitat distribution, dietary composition and life history characteristics of odonate nymphs in a blackwater coastal plain stream. American Midland Naturalist 148:75-89.

Burger, J., C. Jeitner, K. Clark, and L. J. Niles. 2004. The effect of human activities on migrant shorebirds: successful adaptive management. Environmental Conservation 31:283-288.

87Burnham, K. P., and D. R. Anderson. 2002. Model Selection and Multimodel

Inference: A Practical Information-Theoretic Approach, Second edition. Springer, New York.


Carpenter, S. R., B. J. Benson, R. Biggs, J. W. Chipman, J. A. Foley, S. A. Golding, R. B. Hammer, P. C. Hanson, P. T. J. Johnson, A. M. Kamarainen, T. K. Kratz, R. C. Lathrop, K. D. McMahon, B. Provencher, J. A. Rusak, C. T. Solomon, E. H. Stanley, M. G. Turner, M. J. Vander Zanden, C. H. Wu, and H. L. Yuan. 2007. Understanding regional change: A comparison of two lake districts. BioScience 57:323-335.

Carpenter, S. R., T. K. Kratz, J. Kitchell, J. J. Magnuson, and M. J. Vander Zanden. 2006. Biocomplexity; Coordinated Field Studies. http://lterquery.limnology.wisc.edu/index_new.jsp?project_id=BIOCOM1 University of Wisconsin Board of Regents.

Caudill, C. C. 2002. The metapopulation biology of the mayfly Callibaetis ferrugineus Hageni in high elevation beaver ponds. Dissertation. Cornell University, Ithaca, NY.


Cole, L. C. 1954. The population consequences of life history phenomena. The Quarterly Review of Biology 29:103-137.



Crowley, P., and D. Johnson. 1982. Co-occurrence of Odonata in the Eastern United States. Advances in Odonatology 1:15-37.

D'Amico, F., S. Darblade, S. Avignon, S. Blanc-Manel, and S. Ormerod. 2004. Odonates as indicators of shallow lake restoration by liming: Comparing adult and larval responses. Restoration Ecology 12:439-446.

Davidson, J., and H. G. Andrewartha. 1948. The influence of rainfall, evaporation and atmospheric temperature on fluctuations in the size of a natural population of Thrips imaginis (Thysanoptera). Journal of Animal Ecology 17:200-222.

http://lterquery.limnology.wisc.edu/index_new.jsp?project_id=BIOCOM1

88Davies-Colley, R. J., and D. G. Smith. 2001. Turbidity, suspended sediment, and water

clarity: A review. Journal of the American Water Resources Association 37:1085-1101.

De Block, M., and R. Stoks. 2005. Fitness effects from egg to reproduction: Bridging the life history transition. Ecology 86:185-197.

Decamps, H., G. Pinay, R. J. Naiman, G. E. Petts, M. E. McClain, A. Hillbricht-Ilkowska, T. A. Hanley, R. M. Holmes, J. Quinn, J. Gibert, A. M. P. Tabacchi, F. Schiemer, E. Tabacchi, and M. Zalewski. 2004. Riparian zones: Where biogeochemistry meets biodiversity in management practice. Polish Journal of Ecology 52:3-18.


Doncaster, C. P., J. Clobert, B. Doligez, L. Gustafsson, and E. Danchin. 1997. Balanced dispersal between spatially varying local populations: An alternative to the source-sink model. American Naturalist 150:425-445.

Duffy, W. G. 1994. Demographics of Lestes disjunctus disjunctus (Odonata, Zygoptera) in a riverine wetland. Canadian Journal of Zoology 72:910-917.

Ebenman, B., and L. Persson. 1988. Size-structured populations. Springer, New York, USA.

Elias, J. E., and M. W. Meyer. 2003. Comparisons of undeveloped and developed shorelands, northern Wisconsin, and recommendations for restoration. Wetlands 23:800-816.


Forrester, G. E. 1990. Factors Influencing the Juvenile Demography of a Coral-Reef Fish. Ecology 71:1666-1681.

Gergel, S. E. 1996. Scale-dependent landscape effects on north temperate lakes and rivers, M.S. thesis. University of Wisconsin, Madison.

Gribbin, S. D., and D. J. Thompson. 1990. A quantitative study of mortality at emergence in the damselfly Pyrrhosoma nymphula (Sulzer) (Sygoptera: Coenagrionidae). Freshwater Biology 24:295-302.

Halpern, B. S., S. D. Gaines, and R. R. Warner. 2005. Habitat size, recruitment, and longevity as factors limiting population size in stage-structured species. American Naturalist 165:82-94.

89Hansen, A. J., R. L. Knight, J. M. Marzluff, S. Powell, K. Brown, P. H. Gude, and A.

Jones. 2005. Effects of exurban development on biodiversity: Patterns, mechanisms, and research needs. Ecological Applications 15:1893-1905.

Hellriegel, B. 2000. Single- or multistage regulation in complex life cycles: does it make a difference? Oikos 88:239-249.

Henning, B. M., and A. J. Remsburg. in review. Effects of lakeshore vegetation structure on avian and amphibian abundance in northern Wisconsin. Wildlife Research.

Hofmann, T. A., and C. F. Mason. 2005. Habitat characteristics and the distribution of Odonata in a lowland river catchment in eastern England. Hydrobiologia 539:137-147.

Hughes, T. P., A. H. Baird, E. A. Dinsdale, N. A. Moltschaniwskyj, M. S. Pratchett, J. E. Tanner, and B. L. Willis. 2000. Supply-side ecology works both ways: The link between benthic adults, fecundity, and larval recruits. Ecology 81:2241-2249.

Istock, C. A. 1967. Evolution of Complex Life Cycle Phenomena - an Ecological Perspective. Evolution 21:592-&.

Jakob, C., and F. Suhling. 1999. Risky times? Mortality during emergence in two species of dragonflies (Odonata : Gomphidae, Libellulidae). Aquatic Insects 21:1-10.

Johansson, F. 2000. The slow-fast life style characteristics in a suite of six species of odonate larvae. Freshwater Biology 43:149-159.

Johnson, D. H. 1986. The life history of Tegragoneuria cynosura (Say) in Bays Mountain Lake, Tennessee, United States (Anisoptera: Corduliidae). Odonatologica 15.

Johnson, D. M. 1991. Behavioral ecology of larval dragonflies and damselflies. Trends in Ecology & Evolution 6:8-13.

Kennedy, C. H. 1950. The relation of dragonfly-eating birds to their prey. Ecological Monographs 20:103-143.

Lack, D. 1954. The Natural Regulation of Animal Numbers. Oxford University Press, Oxford.


Larochelle, A. 1977. Les amphibiens et reptiles comee predateurs des odonates (premiere note bibliographique). Cordulia 3:81-84, reviewed in Corbet 1999

Likens, G. E. 1985. An Ecosystem approach to aquatic ecology : Mirror Lake and its environments. Springer-Verlag, New York.

90Lima, S. L. 1998. Stress and decision making under the risk of predation: Recent

developments from behavioral, reproductive, and ecological perspectives. Pages 215-290 in Stress and Behavior.

Lindeman, R. L. 1942. The trophic-dynamic aspect of ecology. Ecology 23:399-417.

Lindsay, A. R., S. S. Gillum, and M. W. Meyer. 2002. Influence of lakeshore development on breeding bird communities in a mixed northern forest. Biological Conservation 107:1-11.

Lutz, P. E., and A. R. Pittman. 1970. Some ecological factors influencing a community of adult Odonata. Ecology 51:279-284.

Mahato, M., and D. Johnson. 1991. Invasion of the Bays Mountain Lake dragonfly assemblage by Dromogomphus spinosus (Odonata, Gomphidae). Journal of the North American Benthological Society 10:165-176.

Marburg, A. E., M. G. Turner, and T. K. Kratz. 2006. Natural and anthropogenic variation in coarse wood among and within lakes. Journal of Ecology 94:558-568.

Martin, T. H., D. M. Johnson, and R. D. Moore. 1991. Fish-mediated alternative life-history strategies in the dragonfly Epitheca-cynosura. Journal of the North American Benthological Society 10:271-279.


McCune, B., and J. B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, OR.


McPeek, M. A. 1989. Differential dispersal tendencies among Enallagma damselflies (Odonata) inhabiting different habitats. Oikos 56:187-195.

McPeek, M. A. 1990. Behavioral differences between Enallagma species (Odonata) influencing differential vulnerability to predators. Ecology 71:1714-1726.

McPeek, M. A., and R. D. Holt. 1992. The evolution of dispersal in spatially and temporally varying environments. American Naturalist 140:1010-1027.

Mead, K. 2003. Dragonflies of the North Woods. Kollath-Stensaas Publishing, Duluth, MN, USA.

Michiels, N. K., and A. A. Dhondt. 1991. Characteristics of Dispersal in Sexually Mature Dragonflies. Ecological Entomology 16:449-459.

91Momot, W. T. 1995. Redefining the role of crayfish in aquatic systems. Reviews in

Fisheries Science 3:33-63.

Moore, N. W., and P. S. Corbet. 1990. Guidelines for monitoring dragonfly populations. Journal of the British Dragonfly Society 6:21-23.

Moran, N. A. 1994. Adaptation and constraint in the complex life-cycles of animals. Annual Review of Ecology and Systematics 25:573-600.

Muller, Z. 2003. Effect of sports fisherman activities on dragonfly assemblages on a Hungarian river floodplain. Biodiversity and Conservation 12:167-179.

Mumby, P. J., A. J. Edwards, J. E. Arias-Gonzalez, K. C. Lindeman, P. G. Blackwell, A. Gall, M. I. Gorczynska, A. R. Harborne, C. L. Pescod, H. Renken, C. C. C. Wabnitz, and G. Llewellyn. 2004. Mangroves enhance the biomass of coral reef fish communities in the Caribbean. Nature 427:533-536.

Naiman, R. J., and H. Decamps. 1997. The ecology of interfaces: Riparian zones. Annual Review of Ecology and Systematics 28:621-658.

Olden, J. D., J. M. McCarthy, J. T. Maxted, W. W. Fetzer, and M. J. Vander Zanden. 2006. The rapid spread of rusty crayfish (Orconectes rusticus) with observations on native crayfish declines in Wisconsin ( U.S.A.): trends over the past 130 years. Biological Invasions 8:1621-1628.

Orians, G. H., and J. F. Wittenberger. 1991. Spatial and temporal scales in habitat selection. American Naturalist 137:S29-S49.

Persson, L., and A. M. De Roos. 2006. Size-structured interactions and the dynamics of aquatic systems. Polish Journal of Ecology 54:621-632.

Pfister, C. A. 1996. The role and importance of recruitment variability to a guild of tide pool fishes. Ecology 77:1928-1941.

Polis, G. A., W. B. Anderson, and R. D. Holt. 1997. Toward an integration of landscape and food web ecology: The dynamics of spatially subsidized food webs. Annual Review of Ecology and Systematics 28:289-316.

Polis, G. A., M. E. Power, and G. R. Huxel, editors. 2004. Food webs at the landscape scale. University of Chicago Press, Chicago, IL, USA.

Pollard, E. 1977. A method for assessing changes in the abundance of butterflies. Biological Conservation 12:115-134.

Power, M. E., and W. E. Dietrich. 2002. Food webs in river networks. Ecological Research 17:451-471.

92Racey, G. D., and D. L. Euler. 1982. Small mammal and habitat response to shoreline

cottage development in central Ontario. Canadian Journal of Zoology-Revue Canadienne De Zoologie 60:865-880.

Radomski, P., and T. J. Goeman. 2001. Consequences of human lakeshore development on emergent and floating-leaf vegetation abundance. North American Journal of Fisheries Management 21:46-61.

Regester, K. J., K. R. Lips, and M. R. Whiles. 2006. Energy flow and subsidies associated with the complex life cycle of ambystomatid salamanders in ponds and adjacent forest in southern Illinois. Oecologia 147:303-314.

Resetarits, W. J. 1996. Oviposition site choice and life history evolution. American Zoologist 36:205-215.

Rith-Najarian, J. 1998. The influence of forest vegetation variables on the distribution and diversity of dragonflies in a northern Minnesota forest landscape: a preliminary study (Anisoptera). Odonatologica 27:335-351.

Roughgarden, J., S. Gaines, and H. Possingham. 1988. Recruitment Dynamics in Complex Life Cycles. Science 241:1460-1466.


Sahlen, G. 1999. The impact of forestry on dragonfly diversity in central Sweden. International Journal of Odonatology 2:177-186.

Schaffner, A. K., and B. R. Anholt. 1998. Influence of predator presence and prey density on behavior and growth of damselfly larvae (Ischnura elegans) (Odonata : Zygoptera). Journal of Insect Behavior 11:793-809.

Scheidler, M. 1990. Influence of habitat structure and vegetation architecture on spiders. Zoologischer Anzeiger 225:333-340.

Schindler, D. E., S. I. Geib, and M. R. Williams. 2000. Patterns of fish growth along a residential development gradient in north temperate lakes. Ecosystems 3:229-237.

Schutte, C., P. Schridde, and F. Suhling. 1998. Life history patterns of Onychogomphus uncatus (Charpentier) (Anisoptera : Gomphidae). Odonatologica 27:71-86.

Sih, A. 1980. Optimal Behavior - Can Foragers Balance 2 Conflicting Demands. Science 210:1041-1043.

Stoks, R., and M. A. McPeek. 2003. Predators and life histories shape Lestes damselfly assemblages along a freshwater habitat gradient. Ecology 84:1576-1587.

93Thomas, M. L. H. 1965. Fauna collected during surveys for larval lampreys in Lake

Superior in 1959, 1960 and 1961. Fisheries Research Board of Canada.

Thompson, D. J. 1987. Regulation of damselfly populations: the effects of weed density on larval mortality due to predation. Freshwater Biology 17:367-371.

Ubukata, H. 1984. Oviposition site selection and avoidance of additional mating by females of the dragonfly, Cordulia-aenea-amurensis Selys (Corduliidae). Researches on Population Ecology 26:285-301.

Utzeri, C., G. Carchini, E. Falchetti, and C. Belfiore. 1984. Philopatry, homing and dispersal in Lestes barbarus (Fabricius) (Zygoptera: Coenagrionidae). Odonatologica 13:573-584.

Van Buskirk, J. 1986. Establishment and organization of territories in the dragonfly Sympetrum-rubicundulum (Odonata, Libellulidae). Animal Behaviour 34:1781-1790.

Walpole, M. J., and I. R. Sheldon. 1999. Sampling butterflies in tropical rainforest: an evaluation of a transect walk method. Biological Conservation 87:85-91.


WDNR. 1996. Northern Wisconsin's lakes and shorelands: a report examining a resource under pressure. Wisconsin Department of Natural Resources, Madison, WI.

Werner, F. E., J. A. Quinlan, R. G. Lough, and D. R. Lynch. 2001. Spatially-explicit individual based modeling of marine populations: a review of the advances in the 1990s. Sarsia 86:411-421.

Wilbur, H. M. 1980. Complex life cycles. Annual Review of Ecology and Systematics 11:67-93.

Wildermuth, H. 1992. Habitate und habitatwahl der grossen moosjungfer (Leucorrhinia pectoralis) Charp. 1825 (Odonata, Libellulidae). Z. Okol. Naturschutz 1:3-21 (in German), reviewed by Corbet 1999 p.1918.


Wilson, K. A., J. J. Magnuson, D. M. Lodge, A. M. Hill, T. K. Kratz, W. L. Perry, and T. V. Willis. 2004. A long-term rusty crayfish (Orconectes rusticus) invasion: dispersal patterns and community change in a north temperate lake. Canadian Journal of Fisheries and Aquatic Sciences 61:2255-2266.

94Wissinger, S. A. 1988. Spatial-distribution, life-history and estimates of survivorship in

a 14-species assemblage of larval dragonflies (Odonata, Anisoptera). Freshwater Biology 20:329-340.

Wright, M. 1946. The economic importance of dragonflies (Odonata). Journal of the Tennessee Academy of Science 21:60-71.

Table 1. Lake attributes1 of the 11 study lakes in northeastern Wisconsin and three littoral variables shared by the two study sites

within each lake.

Lake Perim (km) 2

Area (ha)

Morpho-metry3

Wet-land

perim (%)4

Houses / km

Cond. (µS / cm)5

Secchi depth (m)6

Insectiv-orous

fish CPUE7

All fish

CPUE7

Cray-fish

CPUE8

Macro-phyte dom-

inance

Silt dom-

inance

Rock dom-

inance

Black Oak 12.0 230.1 2.2 0.2 18.0 50.0 5.25 174 283 6.58 1 0 1Big St. Germain 12.8

665.7 1.4 0.0 24.3 90.8 2.59 411 440 0.00 1 0 1Carpenter 5.6 144.5 1.3 0.0 18.0 23.0 5.5 277 294 0.00 1 1 0Little Arbor Vitae

10.6 221.1 2.0 0.1 7.2 112.5 4.13 109 521 12.67 1 1 1

Little St. Germain 23.3 402.2 3.3 0.0 19.8 76.5 3.63 238 293 0.00 1 1 0Little Star 6.4 107.1 1.7 0.0 19.5 93.3 6.01 204 284 3.13 0 0 1Lynx 11.2 126.2 2.8 0.0 5.6 23.4 4.44 435 437 0.00 1 0 1Papoose 13.3 176.6 2.8 0.0 13.5 110.0 5.75 140 177 28.42 0 0 1Stormy 7.6 216.0 1.5 0.0 25.0 37.0 6.25 343 345 0.17 1 0 1Towanda 6.0 60.6 2.2 0.2 18.9 30.7 2.38 227 302 0.00 1 0 1Upper Buck-atabon

13.2 211.4 2.6 0.1 12.6 74.0 3.25 406 424 0.00 0 0 1

1 Data from Carpenter et al. (2006).

2 Lake perimeter. 3 Lake morphometry was calculated as the lake perimeter divided by the perimeter of a circle with the same area. 4 Percent of lake perimeter in wetlands (Gergel 1996). 5 Specific conductance (dissolved ionic content) in µ Siemens / cm

6 A measure of water clarity 7 Catch-per-unit-effort (CPUE) 8 Catch-per-unit-effort (CPUE) of an invasive, macrophyte-eating species, rusty crayfish (Orconectes rusticus)

97Table 2. Sample size, mean, coefficient of variation (CV), and percent of total

variance in densities of three Gomphidae life stages occurring among two 22 sites

and 11 lakes.

Sample

size Mean CV (%)

Site-level variance

(%)

Lake-level variance

(%) First-year larvae / m2 22 0.54 130 0 100.0 Second-year larvae / m2 22 0.18 99 34.4 58.5 2005 Exuviae / m 22 2.63 250 34.5 58.3 2006 Exuviae / m 21 2.95 130 37.9 54.7 2005 Adults / m 18 0.05 120 64.1 27.3 2006 Adults / m 20 0.01 82 100.0 0.0

98Table 3. Results from mixed-effects linear models comparing Gomphidae life stage

densities among sites and Spearman rank correlation tests comparing life stage

densities among lakes

Adults 1st year larvae 2nd year larvae exuviae Exuviae adults

Year 2005 2006 2006 2005 2006

Site Adjusted R2 = 0.50

F1,8 = 8.9

p = 0.02

Adjusted R2 = 0.74

F1,9 = 18.1

p = 0.002

n.s.1 n.s.

Lake Spearman r = 0.91

p < 0.001

n.s. n.s. n.s.

1 n.s. = no significant relationship detected (p >0.1)

99Table 4. Mixed-effects linear models1 using population and habitat variables to

predict Gomphidae densities at three different life stages2.

Fixed-effects predictors from mixed-effects models n ∆AICc

3 log L4

p4 Adj. R2

2006 Larvae 1st year larvae2006 = adults2005 18 3.3 0.50 1st year larvae2006 = adults2005 - lake Secchi 18 0.0 8.9 <0.01 0.50 1st year larvae2006 = adults2005 + site macrophytes 18 2.5 1.5 0.22 0.11 1st year larvae2006 = adults2005 - lake crayfish 18 4.0 0.07 1st year larvae2006 = adults2005 - lake insectiv.fish 18 15.5 0.07 2006 Exuviae exuviae2006 = 2nd year larvae2006 21 3.6 0.74 exuviae2006 = 2nd year larvae2006 -site macrophytes 21 0 6.6 0.01 0.86 exuviae2006 = 2nd year larvae2006 + site wetland plants 21 2.3 4.8 0.03 0.88 exuviae2006 = 2nd year larvae2006 + site development 21 7.6 0.74 exuviae2006 = 2nd year larvae2006 - lake crayfish 21 8.4 0.74 exuviae2006 = 2nd year larvae2006 - lake Secchi 21 8.5 0.74 exuviae2006 = 2nd year larvae2006 - lake insectiv.fish 21 16.7 0.74 2005 Adults adults2005 = exuviae2005 18 3.7 0.15 adults2005 = exuviae2005 + site wetland plants 18 0.0 0.003 0.90 0.09 adults2005= macrophytes 18 1.6 0.32 adults2005 = exuvia2005+ site macrophytes 18 1.9 3.56 0.06 0.23 adults2005 = site development 18 5.7 0.16 adults2005 = exuviae2005 + site development 18 6.0 0.04 adults2005= site wetland plants 18 9.9 0.20 2006 Adults adults2006= exuviae2006 20 3.0 <0 adults2006= macrophytes 20 0.0 <0 adults2006 = site development 20 1.5 <0 adults2006= exuviae2006 + macrophytes 20 13.8 <0 adults2006 = exuviae2006 + site development 20 15.4 <0 adults2006= site wetland plants 20 24.8 <0 adults2006= exuviae2006 + site wetland plants 20 38.2 <0

1Lake was included as a random effect in each model. Only fixed-effects terms are

shown here.

1002 All life stage densities, except for 2006 adults, were log-transformed to meet

normality assumptions. Abundances of littoral macrophytes, riparian wetland

plants, and rusty crayfish were square-root-transformed.

3 ∆AICc is the difference in Akaike’s second-order criterion values between the

model with the lowest AICc and the stated model.

4 Likelihood ratio value (followed by p-value) from a likelihood ratio test

comparing the stated model with the model using density of the previous life stage

as the only predictor variable. Likelihood ratio tests were only conducted on models

with lower AICc values than the model with the previous life stage as the sole

predictor.

101Fig. 1. Diagram of the life history of temperate dragonflies in the family

Gomphidae. Dotted lines illustrate the relatively weak effects of survivorship and

site selection on Gomphidae abundances at the designated stages. Note that

transition times are not to scale.

Fig. 2. Distribution of larval Gomphidae head widths in 11 northern Wisconsin

lakes.

Fig. 3. Larval densities of Gomphidae burrowers from nine north-temperate lakes

graphed as a function of adult densities from surveys at the same sites in the

previous year (on one day per lake from June 10 – July 10, 2005).

Fig. 4. Gomphidae exuviae densities as a function of larval densities from the

second year-class sampled earlier during the same year (2006) in 11 north-

temperate lakes.

Fig. 5. First-year larval Gomphidae densities graphed in relation to lake Secchi

depth (water clarity) from 11 north-temperate lakes.

102Figure 1.

103

103

Figure 2.

104Figure 3.

2+log(0.1+adult Gomphidae m-1) in 2005

105Figure 4.

2+log(0.005+ Gomphidae larvae m-1) in early 2006

106Figure 5.

Lake Secchi depth (m)

Appendix. Site habitat data and Gomphidae densities1 at three life stages.

Site length (m) with vegetation type present 2005 Site Manicured

lawn Riparian shrubs

Riparian wetland plants

Littoral macrophytes exuviae adults larvae

first-yelarvae

BO10 0 30 NA 3 2.46 0.042 0.21 0.07BO4 1 2 0 0 0.66 0.000 0.07 0.00BSG12 0 30 5 0 31.92 0.078 0.57 0.43BSG7 1 2 0 6 5.73 0.133 0.86 0.79CR10 1 2 6 7 0.43 0.017 0.14 0.07CR5 0 25 22 1.5 0.33 0.017 0.29 0.14LAV10 1 0 0 0 0.40 NA 0.00 0.00LAV7 0 30 0 3 2.68 NA 0.50 0.21LSG1 0 5 30 7.5 2.62 0.044 2.93 2.50LSG4 1 0 0 30 0.97 0.233 2.21 2.14LST1 1 0 0 0 0.05 0.022 0.21 0.00LST2 0 30 2 0 0.60 0.025 0.57 0.43LX1 0 25 29 0 1.98 NA 1.64 1.07LX10 1 0 0 1.5 3.03 NA 1.14 0.79PP10 0 30 NA 0 0.03 0.008 0.00 0.00PP3 1 0 0 0 0.26 0.025 0.07 0.07ST1 0 0 NA 3 0.28 0.000 0.21 0.07ST8 1 0 0 0.3 0.43 0.000 0.36 0.00TW11 0 30 0 3 0.97 0.111 1.64 1.07TW8 1 2 2 3 0.25 0.011 1.57 1.50UB1 0 20 0.5 0 1.60 0.000 0.29 0.21UB4 1 2 2 0 0.18 0.078 0.21 0.21 1 Larval densities are per benthic m2; exuvia and adult densities are per shoreline m. Appendix. Site habitat data and Gomphidae densities1 at three life stages.

2006 ar exuviae adults

2.07 0.006 1.77 0.000

NA NA 6.13 0.025 0.30 0.011 1.03 0.011 0.27 0.013 1.47 0.017 4.77 0.011 0.43 0.004 4.20 0.019 2.27 NA 17.50 0.000

5.87 0.014 1.10 0.011 1.57 0.011 1.67 0.017 3.57 0.006 2.27 0.033 2.47 0.000 0.90 0.000 0.33 0.006

108CHAPTER IV

Shade alone reduces adult dragonfly abundance2

Abstract

We investigated how changes to physical habitat conditions caused by nonnative

Acacia trees in riparian areas of South Africa could influence dragonfly (Odonata:

Anisoptera) populations. In this naturally treeless region, the invasive tree canopy creates

shade and affects native vegetation. We hypothesized that most breeding odonates select

riparian areas (1) without shade, and (2) with high density and diversity of understory

perch structures. We conducted two experiments at reservoir shorelines by varying shade

and perch structures independently. Dragonfly abundances were lower at sites with high

(75%) or moderate (55%) shade cover than at sites with no shade, and also at bare sand

sites compared with sites containing stick perches. Perch stick density and diversity

(variety of stick heights and diameters), however, did not affect dragonfly abundance.

These experimental results indicate that shade alone directly affected adult odonate

habitat selection. This study isolates which aspects of habitat change affect insect site

selection and can inform conservation strategies.

Keywords: Odonata, riparian buffer, habitat structure, shade, insect behavior, habitat

selection

2 Coauthors Anders C. Olson and Michael J. Samways

109Introduction

Insects and many other animals respond behaviorally to physical habitat

conditions. Tree canopies influence these physical habitat conditions in part via light

interception, which affects local climate and visibility (Endler 1993, Thery 2001). In

addition, tree canopies may also influence insect behavior indirectly via effects on

understory vegetation and prey composition. Changes to canopy structure from riparian

development, logging, and invasion by non-native plants can thus strongly influence

insect habitat use.

In South Africa, most riparian corridors had no indigenous trees and were open

until Australian black wattle (Acacia mearnsii) and long-leaved wattle (A. longifolia)

invaded and formed dense canopies (Gorgens and van Wilgen 2004). Numerous plants

and insects are likely sensitive to the shading effects of invasive Acacia (Breytenbach

1986, Richardson et al. 1989, Samways 2006), but experimental research is needed to

isolate direct and indirect effects of shade. Our experiments independently test effects of

shade intensity and perch availability on South African adult dragonfly (suborder

Anisoptera) riparian site use. Understanding which components of habitat disturbance

are most influential for odonates can inform conservation strategies.

Observational studies suggest that many adult Odonata (dragonfly and damselfly)

species avoid shaded areas (Pezalla 1979, McKinnon and May 1994, Chwala and

Waringer 1996, Clark and Samways 1996, Painter 1998, Kinvig and Samways 2000,

Samways and Taylor 2004, Samways et al. 2005, Ward and Mill 2005). Their behavior

may reflect thermoregulation requirements. Insects and other ectothermic animals are

particularly sensitive to the variation in microclimates produced by vegetation (e.g., May

1101976, McGeoch and Samways 1991, Downes and Shine 1998, Arnan et al. 2007,

Valentine et al. 2007). Lighting can also play an important role in habitat selection (e.g.,

Davies 1978, McCall and Primack 1992, Grundel et al. 1998, Bernath et al. 2002, De

Cauwer et al. 2006, Reinhardt 2006) because insects that fly during the day orient almost

entirely based on visual cues (Lehrer 1994, Dafni et al. 1997, Egelhaaf and Kern 2002,

Olberg et al. 2005).

Adult odonates in the “percher” behavioral guild may require riparian understory

vegetation because they guard breeding territories, thermoregulate, and watch for prey

from plant perches (Corbet 1999). Previous studies indicate greater abundances of adult

dragonflies in areas with taller wetland plants (McKinnon and May 1994, Ueda 1994,

Clark and Samways 1996, De Marco and Resende 2004, Foote and Hornung 2005, Ward

and Mill 2005), perhaps because they serve as perch structures. In a few studies,

researchers added artificial stick perches to their study sites to manipulate dragonfly

abundances for other behavioral questions (Wolf and Waltz 1988, Rehfeldt 1990, Baird

and May 1997, May and Baird 2002), although no research has specifically tested effects

of perch stem density for odonate populations. We hypothesized that riparian sites with

higher densities of perch structures (sticks we erected in the sand) would have greater

odonate abundances than sites with fewer or no perch structures.

The distribution of plant heights or forms (structural diversity) may change

independently of plant density. Odonata likely respond more to structural diversity than

to the diversity of plant species (Buchwald 1992, Corbet 1999, Foote and Hornung 2005).

Taller perch structures can facilitate territory guarding, so perching height is often

proportional to body size (Corbet 1999, De Marco and Resende 2004). Odonate species

111(namely damselflies, suborder Zygoptera) also select perch structures based on stem

diameter (Askew 1982, Rouquette and Thompson 2007). Thus, we expected that sites

with a variety of stick diameters and heights (i.e. higher stem diversity) support higher

dragonfly species richness and abundance than sites with uniform sticks.

In addition, we hypothesized that perch stick location relative to the water’s edge

affects Odonata behavior. Specifically, we expected that territorial male dragonflies

would select perch sticks closest to the water, perhaps to intercept females en route to

oviposition sites (Van Buskirk 1986, Switzer and Walters 1999). Testing whether

odonates use structures farther from the water can inform land managers seeking to

establish riparian buffer widths and manage for wildlife diversity. Thus our research

questions were: 1) Do dragonflies avoid shaded zones? and 2) Do density, diversity, or

position of perch structures affect dragonfly site use?

Materials and methods

Study area

Separate shade and perch stick experiments were conducted at two small

reservoirs with little shoreline vegetation on Vergelegen Estate, Western Cape, South

Africa. Both reservoirs receive water from the Hottentots-Holland Mountains and each is

~2 ha in area. By the end of the dry summer, when field experiments began, water levels

in the reservoirs had dropped so that bare shorelines about 15 m wide were exposed

around the margins of both reservoirs. These shorelines lacked tall vegetation, providing

relatively homogeneous field sites where we could experimentally isolate the structural

112variables of interest: shade levels, perch stick density, perch stick diversity, and perch

stick distance from water.

Field methods

Although a fully crossed experimental design of shade level, perch diversity, and

perch density would have been informative, space and time constraints required us to test

these factors in two separate field experiments. We established twenty shade plots, with

four shade levels replicated randomly within five complete blocks. Distances of 8 to 20

m separated the blocks. All treatment plots (2.8 x 2.8 m each) were separated by at least

5 m, and were 1 m away from the water’s edge (although the water level receded to a

distance of about 3 m from the shade treatments during the study period). We suspended

three types of commercial plant nursery shade cloth 2 m off the ground with wooden

poles (about 10 cm in diameter) and nylon ropes. We chose this height because it was

above normal anisopteran flight routes. Shade cloths were solid green or black and

provided three levels of sunlight interference: 30 %, 55 % and 75 % shade. Control

treatments, which we refer to as 0 % shade, were set up with the same pole and rope

structures, but no cloth. Shade levels were verified using a standard light meter to

quantify percent of full sunlight available under each shade cloth treatment at

approximately the same time of day for all treatments. We measured light levels at a

typical odonate perching height of 20 cm above the ground. Nearby invasive Acacia

trees created 96-97% shade at the same height. To standardize structures and facilitate

observations in the plots, we trimmed any vegetation present, and erected 16 sticks (2 per

113m2) of equivalent sizes (6 - 10 mm diameter; 0.4 - 0.5 m tall) in a grid pattern beneath the

shade cloth.

At the other reservoir, we used regular grids of Eucalyptus sticks to mimic two

structural attributes of understory shoreline vegetation: stem diversity and stem density.

We replicated each of four stick density x stick diversity treatments plus a control

treatment at random within four blocks, giving a total of 20 plots for the perch stick

experiment. We separated each plot by 5 m and each block by 5 - 10 m. Within these

plots (3 x 2.5 m), we varied stick density by erecting two different numbers of sticks in

uniform grids. High-density stick treatments contained 30 sticks (4 per m2); low-density

treatments had 9 sticks (1.2 per m2). The third level of stick density, control plots,

contained no sticks.

Additionally, we created low stick diversity at half of the treatments by using

sticks of all the same diameter and height, and high stick diversity at the other half of

treatments by including an even mixture within plots of ‘tall thin’, ‘tall stout’, ‘short

thin’, and ‘short stout’ sticks. The ‘thin’ sticks were 2 – 6 mm in diameter, while ‘stout’

sticks were 15 – 20 mm in diameter. All sticks were clearly taller than any other

vegetation emerging in the plots: ‘tall’ sticks were 0.6 m high and ‘short’ sticks were

about 0.2 m high. Average odonate perch heights vary between 0 and 0.75 m (May 1976,

Nomakuchi 1992, McKinnon and May 1994, Reinhardt 1999). For the low perch

diversity treatments (where all sticks within a plot were of the same size), each block

contained a different stick size (from the four height and diameter combinations).

Based on average dragonfly densities and logistical constraints, we recorded

dragonfly observations in all shade treatments for ten-minute periods and perch stick

114treatments for five-minute periods. Two observers collected data only on 11 calm, sunny

days during February – April 2005 between 9h45 and 15h45, the hours when odonates

are most active. Although the shadow of shadecloths moved somewhat throughout the

course of the day, we observed odonates only within the shaded portions of each plot.

We observed the same plot portions (measurable from the regular grid of stick perches)

for the same time period within each location block. Observation periods began one

minute after the observers became situated in front of each plot. Observers sat on shore 2

m inland from the edge of each plot, at the center of each plot’s shoreline width. We saw

no evidence of altered dragonfly behaviors when observers were at least a meter away

from perch locations. Although changing weather conditions prevented us from

observing all location blocks on each of the 11 days, we completed observations on all

treatments within a block each day (for both experiments).

During each observation period, we recorded the number and species of

dragonflies perching in the plot, investigating the plot (flying slowly), and passing (flying

quickly) through the plot. We also recorded the number of dragonflies in the plot at the

instant when a timer signaled the end of the observation period, which we used as a

measure of dragonfly relative abundance per plot. Abundance estimates were mainly

comprised of individuals perching within treatments. Size of the plots enabled observers

to count all anisopterans in each plot during a single glance. Species present were easy to

distinguish on the wing. At the perch stick experiment, we also noted where perching

occurred from one of two distance-from-water categories (1-2.5 m or 2.5-4 m inland).

Statistical methods

115Single-factor randomized-block Analysis of Variance (ANOVA) tests were used

to compare mean dragonfly relative abundance (for all species combined) among shade

treatments, which was coded as an ordered variable. We also assessed habitat selectivity

for perching and investigative flight behaviors by dividing each of these variables by the

total number dragonfly entries into the given plot. Based on Bonferroni corrections for

conducting three separate tests, we assessed significance at p < 0.02. After finding

significant differences among treatments in Anisoptera perch selection, we conducted the

same ANOVAs on perch selection for the two most common species. Bonferroni

corrections for conducting these two additional tests (protected under the total Anisoptera

perch selection test) resulted in assessment of significance at p<0.025. We used the

means of response variables across all 11 observation days for each treatment (n = 20 for

both perch and shade experiments). Response variables were square-root transformed

when necessary to fulfill normality and homoscedasticity assumptions. Using Tukey’s

Honest Significant Difference (HSD) tests, we conducted post-hoc pairwise comparisons

among treatment means.

We compared Anisoptera abundance, richness, investigative flights, and perch

selection among the two perch stick diversity and the high- and low-density treatments

(excluding the control) with two-factor randomized-block ANOVAs. We subsequently

compared response variables among all three density treatments (including the control)

with one-way blocked ANOVAs. The Bonferroni-corrected rejection level was p <

0.0125 because we repeated these ANOVAs on four separate response variables. As in

the shade experiment, we conducted the same one-way blocked ANOVAs on perch

selection for the two most common species after finding significant differences among

116stick density treatments. Bonferroni corrections for conducting these two additional tests

(protected under the total Anisoptera perch selection test) resulted in assessment of

significance at p < 0.025. To test whether dragonflies perched more than expected at

random on the row of sticks closest to the water, we used a two-sided t-test (n = 20). All

analyses were conducted using R 2.2.0 software (R Development Core Team, 2005,

Vienna, Austria).

Results

Anisoptera species recorded within the plots included three Libellulidae

(Trithemis dorsalis, T. arteriosa, Pantala flavescens), one Aeshnidae (Anax imperator),

and one Gomphidae (Paragomphus sp.). One zygopteran species (Atricallagma

glaucum) was also occasionally observed in the plots, but was not included in the dataset

because their locations proved too difficult to track reliably in conjunction with

anisopterans. The two Trithemis species were by far the most abundant odonates within

the experimental plots (77 and 95 % of individuals perching in shade and perch

experiment plots, respectively).

Anisoptera mean perch selection decreased from 0.5 to 0 perches per plot entry as

shade cover increased from 0 % to 75 % shade (F3,12 = 13.7, p = 0.0004; Fig. 1).

Specifically, mean perch selection for both species tested was higher at the plots with no

or low (30 %) shade than at the 75 % shade plots (T. arteriosa: F3,12 = 12.9, p = 0.0005;

T. dorsalis: F3,12 = 14.9, p = 0.0002; pairwise differences based on Tukey’s HSD tests

with p <0.05). Mean anisopteran abundances followed the same pattern (F3,11 = 24.7, p <

0.0001), excluding abundance estimates from an outlier zero-shade plot that had no

117odonates present during the relative abundance counts. Dragonflies appear to have

avoided this particular plot more than other zero-shade plots, although we have no data to

suggest why. Mean anisopteran investigative (slow) flights did not differ by shade

treatment (p > 0.1).

Based on the perch stick experiment two-way ANOVA, mean Anisoptera

abundance, richness, investigative flights, and perch selection did not differ with perch

diversity or density. The one-way ANOVA comparing three stick densities revealed that

mean dragonfly abundance was 2.5 to 2.7 times higher at the stick treatments than at

control plots without sticks (F2,14 = 6.8, p = 0.009), but there were no significant

differences between plots with high and low stick densities (Fig. 2). Perch selection

followed the same pattern, with the mean perches per entry 1.6 times higher at stick

treatments than at the control plots (F2,14 = 6.3, p = 0.01). By species, T. dorsalis perch

selections were lower at control plots than at plots with high or low stick densities (F2,14 =

5.6, p = 0.02; pairwise differences based on Tukey’s HSD tests with p <0.05), while T.

arteriosa perch selection did not differ with stick density (p = 0.1). Based on perch

location data across all five treatments, dragonflies perched more frequently (63 % of

observed perches) in the half of the treatment closest to water (t = 90.8, p < 0.0001).

Discussion

Results from the shade experiment provide strong evidence for shade avoidance

behavior by the two most abundant anisopterans around the reservoirs. The three shade

treatments had identical structures except for cloth weave density, but the control

treatment lacked shade cloth completely. Under the assumption that presence of

118shadecloth could inhibit dragonflies from entering plots rather than their decision to perch

there, we considered whether dragonflies responded adversely to the artificial screen

structures rather than shade alone. The perch selection response adjusts for potential

alteration of behavior by screen structures because it standardizes by the total number of

entries into a plot. Additionally, we recorded lower perch selection at the 75 % shade

treatment than at the 30 % shade treatment, which had no differences in structure. After

controlling for effects of structure per se, we thus conclude that shade alone can have a

significant inhibitory effect on the behavior of these species.

Whether shade limits prey availability, mate attraction, hunting effectiveness, or

thermoregulation remains to be tested. Vegetation structure can simultaneously influence

trophic interactions through several of these mechanisms (e.g., Coll et al. 1997). Prey

availability seems less likely to control the distribution of adult odonates around water

bodies because they are mobile, generalist predators. Based on rapid flight paths

following shade boundaries (Reinhardt 2006), odonates may distinguish shade boundaries

by sight rather than solely by temperature.

Presence of perch sticks increased odonate abundances, but contrary to our

hypotheses, the density, size, height, and heterogeneity of perch sticks set up in unshaded

areas had little effect on odonate abundances. Examination of a broader range of stem

varieties and densities, though, may reveal additional odonate perch selection

preferences. The only measurable perch preferences we observed were for sticks closest

to the water. Preference for perch sites within 1 m of the water’s edge concurs with

previous odonate behavior studies (Ward and Mill 2005). Proximity to suitable

oviposition areas likely explains this perching behavior for male dragonflies (Van

119Buskirk 1986), although prey density gradients could also be investigated with additional

field studies. Our results suggest that understory perch structures only affect odonate

habitat selection when they are completely absent.

Behavioral studies are critical for effective wildlife conservation because animal

behavior is an important determinant of species’ habitat use and distributions (Anthony

and Blumstein 2000, Linklater 2004, Reed 2004). Odonata in this study exemplify how

plants can directly influence animal behavior and abundance via altered light conditions.

Further, our results suggest that creation of shaded rather than open habitats by the

introduced Acacia have a direct negative influence on adult odonate abundances.

Because body size and pigmentation likely influence thermoregulation behavior (May

1976, De Marco and Resende 2002), other Odonata species with similar morphologies

may exhibit similar responses to shade. Vulnerability and scarcity of threatened

populations often necessitates behavioral research on related species.

Experimental isolation of physical habitat conditions that affect insect behavior

and distributions can inform conservation strategies. Knowledge that shade affects

odonates directly, rather than via reductions of other habitat structures, suggests that

removal of alien tree canopies could enhance populations of indigenous dragonflies. The

South African national Working for Water Programme, devoted to removal of alien

riparian trees, has improved wildlife habitat by returning many riparian areas to sunlit

conditions (Samways and Grant 2006). Our results suggest that invasive tree growth and

other forest canopy disturbances influence riparian site use by adult dragonflies.

Acknowledgments

120G. Wright and the staff of Vergelegen offered invaluable support and access to a

large natural area currently undergoing restoration. Thanks to M.G. Turner, J.B. Zedler,

C. Gratton, A.R. Ives, S.R. Carpenter, B. Peckarsky, and two anonymous readers for

careful reviews of the manuscript. Funding was provided by a National Science

Foundation Graduate Research Fellowship awarded to the primary author. The

Stellenbosch University Department of Conservation Ecology and Entomology supplied

material and logistical support.

Literature Cited

Anthony, L. L., and D. T. Blumstein. 2000. Integrating behaviour into wildlife conservation: the multiple ways that behaviour can reduce N-e. Biological Conservation 95:303-315.

Arnan, X., A. Rodrigo, and J. Retana. 2007. Uncoupling the effects of shade and food resources of vegetation on Mediterranean ants: an experimental approach at the community level. Ecography 30:161-172.

Askew, R. 1982. Roosting and resting site selection by Coenagrionid damselflies. Advances in Odonatology 1:1-8.


Bernath, B., G. Szedenics, H. Wildermuth, and G. Horvath. 2002. How can dragonflies discern bright and dark waters from a distance? The degree of polarisation of reflected light as a possible cue for dragonfly habitat selection. Freshwater Biology 47:1707-1719.

Breytenbach, G. 1986. Impacts of alien organisms on terrestrial communities with emphasis on communities of the south-western Cape. Pages 229-238 in I. Macdonald, F. Kruger, and A. Ferrar, editors. The Ecology and Management of Biological Invasions in Southern Africa. Oxford University Press, Cape Town.


121Chwala, E., and J. Waringer. 1996. Association patterns and habitat selection of

dragonflies (Insecta: Odonata) at different ypes of Danubian backwaters at Vienna, Austria. Arch. Hydrobiol.Suppl. 115:45-60.


Coll, M., L. A. Smith, and R. L. Ridgway. 1997. Effect of plants on the searching efficiency of a generalist predator: The importance of predator-prey spatial association. Entomologia Experimentalis Et Applicata 83:1-10.


Dafni, A., M. Lehrer, and P. G. Kevan. 1997. Spatial flower parameters and insect spatial vision. Biological Reviews of the Cambridge Philosophical Society 72:239-282.

Davies, N. B. 1978. Territorial defence in speckled wood butterfly (Pararge-aegeria) - resident always wins. Animal Behaviour 26:138-147.

De Cauwer, B., D. Reheul, S. De Laethauwer, I. Nijs, and A. Milbau. 2006. Effect of light and botanical species richness on insect diversity. Agronomy for Sustainable Development 26:35-43.

De Marco, P., and D. C. Resende. 2002. Activity patterns and thermoregulation in a tropical dragonfly assemblage. Odonatologica 31:129-138.


Downes, S., and R. Shine. 1998. Heat, safety or solitude? Using habitat selection experiments to identify a lizard's priorities. Animal Behaviour 55:1387-1396.

Egelhaaf, M., and R. Kern. 2002. Vision in flying insects. Current Opinion in Neurobiology 12:699-706.

Endler, J. A. 1993. The color of light in forests and its implications. Ecological Monographs 63:1-27.


Gorgens, A. H. M., and B. W. van Wilgen. 2004. Invasive alien plants and water resources in South Africa: current understanding, predictive ability and research challenges. South African Journal of Science 100:27-33.

122Grundel, R., N. B. Pavlovic, and C. L. Sulzman. 1998. The effect of canopy cover and

seasonal change on host plant quality for the endangered Karner blue butterfly (Lycaeides melissa samuelis). Oecologia 114:243-250.

Kinvig, R. G., and M. J. Samways. 2000. Conserving dragonflies (Odonata) along streams running through commercial forestry. Odonatologica 29:195-208.

Lehrer, M. 1994. Spatial Vision in the Honeybee - the Use of Different Cues in Different Tasks. Vision Research 34:2363-2385.

Linklater, W. L. 2004. Wanted for conservation research: behavioral ecologists with a broader perspective. BioScience 54:352-360.


May, M. L., and J. M. Baird. 2002. A comparison of foraging behavior in two "percher" dragonflies, Pachydiplax longipennis and Erythemis simplicicollis (Odonata : Libellulidae). Journal of Insect Behavior 15:765-778.

McCall, C., and R. B. Primack. 1992. Influence of Flower Characteristics, Weather, Time of Day, and Season on Insect Visitation Rates in 3 Plant-Communities. American Journal of Botany 79:434-442.

McGeoch, M., and M. Samways. 1991. Dragonflies and the thermal landscape: implications for their conservation (Anisoptera). Odonatologica 20:303-320.


Nomakuchi, S. 1992. Male reproductive polymorphism and form-specific habitat utilization of the damselfly Mnais-pruinosa (Zygoptera, Calopterygidae). Ecological Research 7:87-96.

Olberg, R. M., A. H. Worthington, J. L. Fox, C. E. Bessette, and M. P. Loosemore. 2005. Prey size selection and distance estimation in foraging adult dragonflies. Journal of Comparative Physiology a-Neuroethology Sensory Neural and Behavioral Physiology 191:791-797.

Painter, D. 1998. Effects of ditch management patterns on Odonata at Wicken Fen, Cambridgeshire, UK. Biological Conservation 84:189-195.

Pezalla, V. M. 1979. Behavioral ecology of the dragonfly Libellula pulchella Drury (Odonata:Anisoptera). American Midland Naturalist 102:1-22.

123Reed, J. M. 2004. Recognition behavior based problems in species conservation. Annales

Zoologici Fennici 41:859-877.

Rehfeldt, G. E. 1990. Antipredator strategies in oviposition site selection of Pyrrhosoma-nymphula (Zygoptera, Odonata). Oecologia 85:233-237.

Reinhardt, K. 1999. The reproductive activity of two Pseudagrion species in the same habitat (Odonata : Coenagrionidae). African Entomology 7:225-232.

Reinhardt, K. 2006. Macromia illinoiensis Walsh males use shade boundaries as landmarks (Anisopter: Macromiidae). Odonatologica 35:389-393.

Richardson, D. M., I. A. W. Macdonald, and G. G. Forsyth. 1989. Reductions in plant species richness under stands of alien trees and shrubs in the fynbos biome. South African Forestry Journal 149:1-8.


Samways, M., and S. Taylor. 2004. Impacts of invasive alien plants on Red-Listed South African dragonflies (Odonata). South African Journal of Science 100:78-80.

Samways, M. J. 2006. Threat levels to odonate assemblages from invasive alien tree canopies. Pages 300 in A. Cordero, editor. Dragonflies and Forest Ecosystems. Pensoft, Sofia-Moscow.

Samways, M. J., and P. B. C. Grant. 2006. Regional response of odonata to river systems impacted by and cleared of invasive alien trees. Odonatologica 35:297-303.

Samways, M. J., S. Taylor, and W. Tarboton. 2005. Extinction reprieve following alien removal. Conservation Biology 19:1329-1330.

Switzer, P. V., and W. Walters. 1999. Choice of lookout posts by territorial amberwing dragonflies, Perithemis tenera (Anisoptera : Libellulidae). Journal of Insect Behavior 12:385-398.

Thery, M. 2001. Forest light and its influence on habitat selection. Plant Ecology 153:251-261.

Ueda, T. 1994. Spatial-distribution of mate-searching males in the damselfly, Cercion-c-calamorum (Odonata, Zygoptera). Journal of Ethology 12:97-105.

Valentine, L. E., B. Roberts, and L. Schwarzkopf. 2007. Mechanisms driving avoidance of non-native plants by lizards. Journal of Applied Ecology 44:228-237.

124Van Buskirk, J. 1986. Establishment and organization of territories in the dragonfly

Sympetrum-rubicundulum (Odonata, Libellulidae). Animal Behaviour 34:1781-1790.


Wolf, L. L., and E. C. Waltz. 1988. Oviposition site selection and spatial predictability of female white-faced dragonflies (Leucorrhinia-intacta) (Odonata, Libellulidae). Ethology 78:306-320.

125Tables

<< none>>

Figure 1. Mean count of perching Anisoptera (corrected by the total number of

anisopteran entries) within each shade treatment (n = 20). Different letters indicate

significant differences based on Tukey’s honest significant difference tests (p < 0.05).

Error bars represent two standard errors above the mean and two standard errors below

the mean.

Figure 2. Mean Anisoptera abundance associated with three types of stem treatments.

The control was without sticks, while high and low stem density treatments had 30 and 9

sticks erected, respectively. Different letters indicate significant differences based on

Tukey’s honest significant difference tests (p < 0.05); error bars show two standard errors

both above and below the mean.

126Figure 1.

0

0.2

0.4

0.6

0.8

0

Perc

hes

per

plot

ent

ry

A C AB BC

0 5 5

203
Percent shad
505
e
807

127Figure 2

0

0.4

0.8

1.2

1.6

control high low

Stem density

Abu

ndan

ce p

er tr

eatm

ent

A B B

aquatic and terrestrial vegetation influence … · wisconsin ecologists. members of my graduate...

Documents