COMPARING THE DIVERSITY, GEOGRAPHIC DISTRIBUTION, AND INTRASPECIFIC VARIATION OF SUBTERRANEAN TERMITES (RETICULITERMES: ISOPTERA: RHINOTERMITIDAE) OCCURRING IN WOODLANDS AND URBAN ENVIRONMENTS OF MISSOURI USING MORPHOLOGY AND 16S mtDNA

A Dissertation

Presented to

The Faculty of the Graduate School

University of Missouri-Columbia

By

OLGA PATRICIA PINZON FLORIAN Dr. Richard Houseman, Dissertation Supervisor

DECEMBER 2007

The undersigned, appointed by the Dean of the Graduate School,

Have examined the dissertation entitled:

COMPARING THE DIVERSITY, GEOGRAPHIC DISTRIBUTION, AND INTRASPECIFIC VARIATION OF SUBTERRANEAN TERMITES (RETICULITERMES: ISOPTERA: RHINOTERMITIDAE) OCCURRING IN WOODLANDS AND URBAN ENVIRONMENTS OF MISSOURI USING MORPHOLOGY AND 16S mtDNA

Presented by Olga Patricia Pinzon Florian

A candidate for the degree of Doctor of Philosophy

And hereby certify that in their opinion it is worthy of acceptance.

Dr. Richard M. Houseman, Dissertation Supervisor, Div. of Plant Sciences

Dr. Qisheng Song, Division of Plant Sciences

Dr. Marc Linit, Division of Plant Sciences

Dr. Mark Cowell, Department of Geograhy

Dr. Mark Ellersieck, Department of Statistics

ACKNOWLEDGEMENTS

During the development of my graduate studies and research at the

University of Missouri I received input, support and cooperation from

many people. Particularly, I want to acknowledge the important

contributions to my scientific career in the field of entomology to: Dr.

Richard Houseman, Academic Adviser; Committee members: Mark Linit,

Quisheng Song, Mark Ellersieck, Mark Cowell and former committee

member Dr. Tom Clark; Professional Pest Companies and home owner’s

from Missouri; Missouri Conservation Department; Dr. Shigen Ann and

graduate students: Tongtong Xu, Yaning Sun, Songje Wang, from the

MU Laboratory of Insect Molecular Physiology; my classmates: Sarah

Phipps, Akekawat Vithepraadit, Carlos Angel; Universidad Distrital

Francisco José de Caldas, Fulbright Program; Victor Nieto, Andres Nieto,

Olga Florian, family and friends.

iii TABLE OF CONTENTS

Acknowledgements...... ii

Abstract ...... xii

Overview...... 1

Abstract ...... 9

Chapter 1 Diversity and genetic variation of Reticulitermes species occurring in woodlands and urban environments of Missouri

Introduction ...... 12

Methods ...... 15

Results...... 24

Discussion...... 32

Chapter 2 Morphometric variation of soldiers of Reticulitermes populations of Missouri

Abstract ...... 72

Introduction ...... 74

Methods ...... 77

Results...... 80

Discussion...... 83

References cited ...... 108

VITA...... 118

iii iii

LIST OF TABLES

Tables Page

Chapter 1 Diversity and genetic variation of Reticulitermes species occurring in woodlands and urban environments of Missouri

Table 1. Location and general information for nine Missouri conservation areas sampled for subterranean termites (2004-2005)...... 41

Table 2. Number of samples collected for Reticulitermes species within Missouri according to habitat. Number of samples occurring in woodlands only included those collected during spring 2004...... 42

Table 3. Total number of samples of each Reticulitermes species collected from home infestations, urban environments and woodlands by county.43

Table 4. Number of samples of Reticulitermes species collected from home infestations in the central, southern and northern regions of Missouri during 2004 and 2005 listed by city...... 44

Table 5. Number of Reticulitermes species samples collected from urban environments in the central, southern and northern regions of Missouri during 2004 and 2005...... 46

Table 6. Number of samples of each Reticulitermes species collected from woodlands in the central, southern, and northern regions of Missouri during spring 2004...... 47

Table 7. Species composition and number of samples of Reticulitermes species collected from wooded areas in the central (Rudolf Bennit),

iv iv southern (Mark Twain) and northern (Union Ridge) regions of Missouri during spring and fall of 2004...... 48

Table 8. Haplotypes of Reticulitermes species found in Missouri based on nucleotide variation using the species Coptotermes formosanus as the outgroup species. Dashes indicate same nucleotide occur at that base pair position...... 49

Table 9. Diversity and abundance of Reticulitermes species haplotypes for each habitat sampled in Missouri...... 50

Table 10. Number of samples of each Reticulitermes haplotype collected from home infestations, urban environments and woodlands of Missouri 2004-2005...... 51

Table 11. Location and number of samples of Reticulitermes flavipes haplotypes collected within Missouri...... 52

Table 12. Location and number of samples of Reticulitermes hageni, R. tibialis and R. virginicus haplotypes collected within Missouri...... 53

Table 13. Genetic variability parameters of Reticulitermes flavipes from Missouri, obtained following alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus...... 54

Table 14. Genetic variability of Reticulitermes hageni for each habitat obtained by alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus ...... 55

Table 15. Genetic variability of Reticulitermes virginicus for each habitat sampled obtained by alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus...... 56

Table 16. Genetic variability of Reticulitermes tibialis for each habitat sampled obtained by alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus...... 57

v v Chapter 2 Morphometric variation of soldiers of Reticulitermes populations of Missouri

Table 1. Means and standard deviation of morphometric characters associated with the head, pronotum, labrum and gula of Reticulitermes soldiers collected in urban and woodlands of Missouri...... 88

Table 2. Results of ANOVA comparing the sizes of nine individual metric characters among four Reticulitermes species found in Missouri. (P value≤ 0.005 indicate mean difference is significant at the 0.05 level) . 89

Table 3. Means and standard deviation for soldier morphometric characters of Reticulitermes collected in urban environments, home infestations, and woodlands of Missouri whose identity was confirmed using 16S mtDNA gene sequencing...... 90

Table 4. Summary of statistical differences among the means of nine soldier morphometric characters from four Reticulitermes species. HW: head width, HL: Head length, THL: Total length of head, LW: Labrum width, PW: Pronotum width, PL: Pronotum length, Min. G.: Minimum width of gula, Max. G. Maximum width of gula. Bold characters indicate mean differences at P<0.000 and no bold characters indicate P<0.005. Numbers in parenthesis indicate the number of characters that differentiate between two species...... 91

Table 5. Structure matrix containing the pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions of three functions obtained after conducting stepwise discriminant function analysis of nine morphometric characters of four Reticulitermes species found in Missouri. Variables ordered by absolute size of correlation within function...... 92

vi vi Table 6. Number of samples analyzed and percentage of correct species identifications after performing stepwise discriminant function analysis of nine metric characters from Reticulitermes soldiers in Missouri...... 93

Table 7. Results of ANOVA comparing soldier morphometric characters of Reticulitermes flavipes from urban environments (n=84), home infestations (n=38) and woodlands (n=38) of Missouri. (P value ≤ 0.005 indicate mean difference is significant at the 0.05 level)...... 94

Table 8. Mean and standard deviation of nine metric characters of soldiers of Reticulitermes flavipes occurring around homes (urban) (n=84), home infestations (n=38) and woodlands (n=38) of Missouri...... 95

Table 9. Results of ANOVA comparing soldier morphometric characters in Reticulitermes flavipes among haplotypes RFM1 (n=35) and RFMF (n=60) from Missouri...... 96

Table 10. Mean and standard deviation of soldier morphometric characters from Reticulitermes flavipes haplotypes RFM1 (n=35) and RFM5 (n=60) in Missouri samples. (P value ≤ 0.005 indicate mean difference is significant at the 0.05 level)...... 97

Table 11. The size of soldier Morphometric characters recorded in the literature for four Reticulitermes species found in the United States ..... 98

vii vii LIST OF FIGURES

Figures Page

Chapter 1 Diversity and genetic variation of Reticulitermes species occurring in woodlands and urban environments of Missouri

Figure 1. General scheme of location and distribution of transects and individual colonies sampled in woodlands (showing Mark Twain N.F.). Current scales of the aerial images are not indicated...... 58

Figure 2. Location of samples collected around homes (showing Salem, Mo)...... 59

Figure 3. Approximate location of the collection points of Reticulitermes within Missouri...... 60

Figure 4. Relative abundance of Reticulitermes species found in woodlands, urban environments and home infestations within the state. Total samples: home infestations (188); urban environments (163); home infestations (188)...... 61

Figure 5. Relative abundance of Reticulitermes species in woodlands, and urban environments of Missouri. Wooded areas and associated cities are listed in the same order side by side with locations are geographically paired with each other...... 62

Figure 6. Relative abundance of Reticulitermes species in woodlands, and urban environments of Missouri, according to the geographic position of the sampling site within the state...... 63

viii viii Figure 7. Frequency of Reticulitermes flavipes haplotypes in nine urban environments and nine woodlands of Missouri...... 64

Figure 8. Approximate geographical distribution of Reticulitermes flavipes within the counties where samples were collected from woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequences are available. Insert: previous published Reticulitermes flavipes distribution (Nutting, 1990). Numbers correspond to the haplotype code given for the species in this study...... 65

Figure 9. Approximate geographical distribution of Reticulitermes hageni within the counties where samples were collected from woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequences are available. Insert: previous register regarding R. hageni distribution (Nutting, 1990). The numbers correspond to the haplotype code given for the species...... 66

Figure 10. Approximate geographical distribution of Reticulitermes virginicus within the counties where samples were collected on woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequences are available. Insert: previous register regarding R. virginicus distribution (Nutting, 1990). The numbers correspond to the haplotype code given for the species...... 67

Figure 11. Approximate geographical distribution of Reticulitermes tibialis within the counties where samples were collected on woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties

ix ix from which the species is known to occur but no molecular sequenceialiss are available. Insert: previous register regarding R. ti distribution (Nutting, 1990). The numbers correspond to the haplotype code given for the species...... 68

Figure 12. Neighbor –joining tree depicting phylogenetic relationships of Reticulitermes from Missouri, inferred from 16srDNA. Condensed>50%)...... 69

Figure 13. Networks assembled using TCS algorithm in TCS 1.21 based on the number of mutational steps between haplotypes. a: Reticulitermes flavipes. The size of the oval is not proportional to the frequency of a haplotype. Rectangular boxes indicate which haplotype has the highest outgroup probability for its network. Based on its frequency and number of mutational connections, this haplotype is likely to be most closely related to the ancestral haplotype. The numbers of mutational steps between haplotypes are indicated next to branches when the values are > 1...... 70

Figure 14. Networks assembled using TCS algorithm in TCS 1.21 based on the number of mutational steps between haplotypes. a: Reticulitermes hageni, b: Reticulitermes virginicus, c: Reticulitermes tibialis. The size of the oval is not proportional to the frequency of a haplotype. Rectangular boxes indicate which haplotype has the highest outgroup probability for its network. Based on its frequency and number of mutational connections, this haplotype is likely to be most closely related to the ancestral haplotype. The numbers of mutational steps between haplotypes are indicated next to branches when the values are > 1. .... 71

x x Chapter 2 Morphometric variation of soldiers of Reticulitermes populations of Missouri

Figure 1. Soldier morphometric characters measured and used to describe and make comparisons among Reticulitermes species found in Missouri. 1: Head length, 2: Total head length, 3: head width, 4: Pronotum wide, 5: Pronotum length, 6: Maximum gular width, 7: Minimum gular width, 8: Labral width, 9: Labral length...... 99

Figure 2. a-i. Box plots illustrating and comparing distribution and dispersion of nine metrics of the four Reticulitermes species found in Missouri. A: Head width, b: head length, c: total head length, d: Labrum length, e: Labrum width, f: Pronotum width, g: Pronotum length, h: Minimum gula, i: Maximum gula...... 100

Figure 3. Bi-dimensional plot indicating intraspecific dispersion of individuals based on soldier Morphometric characters. Four centroids each corresponding to one Reticulitermes species are also shown...... 105

Figure 4. Three-dimensional plot after performing discriminant function analysis of morphometric characters from soldiers of four Reticulitermes species. Intraspecific dispersion of individuals by species is also shown. R. flavipes (black), R. hageni (red), R. virginicus (blue), R. tibialis (green) ...... 106

Figure 5. Comparison of the size of nine morphometric characters measured from soldiers of the 16S mtDNA molecular haplotypes RFM5 and RFM1 of Reticulitermes flavipes sampled in Missouri. Means of RFM1 individual characters are smaller ( df: 1, P<0.005) for all individual morphometric characters...... 107

xi xi

Abstract

Subterranean termite species of the Neartic genus Reticulitermes are ecologically and economically important in the United States. At least six species of the genus Reticulitermes are native from North American forests where they feed on cellulose materials. In urban environments they may feed upon dead wood, and therefore infest and destroy man made wooden structures.

Missouri is considered to have a moderate risk for termite infestations when compared with the states of the southeast part of the country. Despite their ecological and economical importance, Missouri’s subterranean termite faunal composition and geographical distribution in not well known, since some of the records are more than 50 years old and others provide conflicting information regarding occurrence and distribution of species. In addition, most of the current available data for Reticulitermes distribution in the United States is based on samples gathered from home infestations, but little is known about populations inhabiting more natural environments where native genotypes and phenotypes are expected to be found.

To determine Reticulitermes species diversity within Missouri and to explore diversity and abundance of species among different habitats, I gathered approximately 600 samples of termite colonies across Missouri during 2004 and 2005 in nine conservation areas, nine cities located nearby, and from home infestations occurring in different locations within the state.

xii xii To explore the possible influence of season on sampling the diversity and abundance of each species, I sampled twice in three conservation areas, once during the spring season and once during the fall season of 2004.

To explore the intraspecific genetic variability of individual Reticulitermes species and to explore inter and intraspecific phylogenetic relationships, I extracted and amplified a portion of aprox. 428 bp of DNA of the 16S mtDNA gene from one individual termite pseudergate per sample. The samples were sequenced in both directions, contigs formed, manually edited, aligned and analyzed to determine: nucleotide diversity per site, number of identical haplotypes, haplotype diversity, and Tajima’s test to confirm the hypothesis of neutrality of 16S rDNA. The haplotypes were subjected to phylogenetic analysis of interspecific relationships using parsimonious analysis and network analysis using TCS software.

To morphometrically characterize and compare soldiers of Reticulitermes

species from Missouri, I measured nine morphometrc characters including

width and length of the head, pronotum, gula and labrum from 393 soldiers.

Moreover, since the identity of a number of samples was confirmed using

molecular sequencing, I compared metrics among the more common

haplotypes of R. flavipes.

Reticulitermes flavipes (Kollar), Reticulitermes virginicus (Banks),

Reticulitermes tibialis Banks and Reticulitermes hageni Banks were found occurring in Missouri.

Reticulitermes flavipes and R. hageni were the most abundant species found within Missouri. Reticulitermes flavipes was found to be more abundant

xiii xiii in urban environments and home infestations than in woodlands.

Reticulitermes hageni also occurs throughout the state. The geographic distribution of R. hageni, was further north east and west within the state than previously known and its populations are more abundant in woodlands.

Although R. virginicus was found scarcely within the sampling, it was also found occurring at the furthest northwest point of sampling within the state and correspond to a new record of the specie within the state. Unexpectedly,

Reticulitermes tibialis was only found scarcely in urban and home infestations

and not in the woodlands sampled in Missouri.

Intraspecific genetic variability based on a portion of the 16S mtDNA gene produced thirty-three haplotypes from 364 samples sequenced of all four

Reticulitermes species. Reticulitermes flavipes genetic variability based on haplotype number was the highest among the Missouri species having 21 of the 33 haplotypes found; 12 of which also occur in neighboring states. The nine remaining haplotypes are new records.

Reticulitermes hageni was the less diverse species in terms of haplotypes, since only three haplotypes were found from one hundred and sixteen R. hageni samples sequenced. Two of those haplotypes are also known to occur in other states of the United States.

Reticulitermes virginicus and R. tibialis were more diverse species in terms of haplotypes since four haplotypes were differentiated from seven R. virginicus samples sequenced and five from seven R. tibilalis samples.

The six most abundant haplotypes of Reticulitermes flavipes: RFM1-6 were found in all the three environments sampled within Missouri and also occurr

xiv xiv in neighboring states. Haplotype diversity of R. flavipes was highest in urban environments; R. hageni haplotype diversity was highest in woodlands and R.

virginicus and R. tibialis haplotype diversity was highest in home infestations.

Phylogenetic relationships among the Reticulitermes species depicted on the Neigbor-Joining tree and the most parsimonious tree were congruent; correct species grouping of haplotypes of individual species is well supported by bootstrap values (>88). Bootstrap values of nodes supporting relationships among Reticulitermes species had lower support (~60) leading to unresolved interrelationships among species through the use of this portion of the 16S mtDNA gene.

No single morphometric character completely separates among the soldiers of the four species of Reticulitermes from Missouri due to overlapping size among species. However, at least two characters may separate between any group of two species.

Morphometric analyses suggest differences in size among soldiers of R.

flavipes collected from woodlands and home infestations. Moreover, our

results suggest there are also differences in size among the soldiers of the

most common haplotypes of R. flavipes. These findings provide evidence for

the effect of environment and genotype as underlying causes of size variation

among soldiers of Reticulitermes.

xv xv

Overview

Termites are a diverse group of social insects having approximately 2600

species worldwide distributed in seven families, 14 subfamilies and 281

genera (Kambhampati and Eggleton 2000).Termites are primarily a group of

exceptional ecological importance. Their diversity of feeding habits enhances

soil chemical, biological and physical processes (Holt and Lepage 2000).

Termites are considered “ecosystem engineers”; they are one of the macro

organisms present in the soil that may control available residual material

available to organisms in other functional categories. Construction of feeding

galleries of subterranean termites may increase soil porosity and improve soil

water transmission (Holt and Lepage 2000).Termite salivary secretions and

soil translocation may modify soil chemical properties in soils with high

termite activity (Holt and Lepage 2000).

Termites feed on plant material at various levels of decomposition, thus recycling nutrients via feces, salivary secretions and corpses as well as returning energy and nutrients to the ecosystem by becoming easy prey to invertebrates and vertebrates (Wood and Sands 1978). Termite assemblages are known to influence litter consumption and C fluxes, affecting the establishment of stable pools of complex organic material, nitrogen fixation and denitrification, all important determinants of plant growth (Bignell and Eggleton 2000).

1 Termites may be divided in four ecological types: subterranean termites, dry wood termites, damp wood termites and harvester termites, of which subterranean termites contains the more economically damaging species of the United States (Nutting 1990). About 183 termite species may cause damage in buildings, with subterranean termites containing 80% of the economically important species (Su and Scheffrah 2000). Nine species of subterranean termites from the genera Coptotermes and Reticulitermes are responsible for major termite problems in buildings in the United States.

Although estimates varies by region, annual costs of chemical treatment continually increase (Su and Scheffrah 1990) and exceeded $2 billion dollars in 1999 not including the repair cost (Su and Scheffrah 1990, Lewis 1997,

Kard 2003) and may reach $11 billion when including building repairs (Su

2002, Kard 2003).

Reticulitermes (Holmgren) is a neartic genus that originally inhabited the forest in which it plays an important role in wood decomposition. At least six different species have been recorded until now in the continental United

States including:, Reticulitermes arenincola Goellner, Reticulitermes flavipes

(Kollar), Reticulitermes hageni Banks, Reticulitermes hesperus Banks,

Reticulitermes tibialis Banks, Reticulitermes virginicus Banks (Haverty et al.

1996).

Reticulitermes colonies are composed of castes: reproductives, soldiers and workers. They exhibit a complex life cycle in which polymorphism plasticity and regressive molts may occur (Laine and Wright 2003). In addition, they may reproduce from primary or secondary reproductives and

2 2 some species also may reproduce by parthenogenesis. Differences among life cycles of most of the Reticulitermes species are still unknown and most of the available information is based on the European species R. lucifugus and

R. santonensis (Laine and Wright 2003).

Geographical distribution of the Reticulitermes species known to occur in the United States are mainly based on records from home infestations with few records from forested regions, therefore its distribution in the United

States is not completely known (Emerson 1952, Weesner 1970, Messenger et al. 2002, Messenger 2003a, Austin et al. 2005).

The objective of subterranean termites control is to protect a structure and its contents; however detection and prediction of potential damage is difficult due to the cryptic habits of these termites. Until recently, prevention of termite attacks was based on soil insecticide barriers used as pre- construction treatments with liquid insecticides. These treatments have long- lasting effects applied in volumes of 5-10 Kg per 185 m2 of construction (Su

and Scheffrah 1998). A recent development for control of subterranean

termites involves the use of a monitoring-baiting system in which the use of

feeding toxicants is targeted at previously detected colonies. The advantages

to this strategy are more specific and low volumes of insecticides such as

chitin synthesis inhibitors may be used to eliminate colonies that locate and

feed on the bait. The monitoring-baiting system is advantageous for the

environment but their implementation still relies on abundant hand labor that

makes it expensive. Research to reduce time, improve techniques, along with

studies of termite population dynamics of each species involved in the

3 3 attacks are still needed to improve efficacy of monitoring, reduce costs of baiting and improve the use of insecticide barriers (Su and Scheffrah 1998).

A better understanding of physical, chemical and biological factors

affecting foraging and food finding behaviors of individual subterranean

termite species will significantly improve the monitoring baiting program so it

becomes more acceptable to termite control industry (Su and Scheffrah

2000). Therefore, accurate differentiation among Reticulitermes species is a key factor since biological, ecological and behavioral studies are dependent of correct species identification.

Reproductives of Reticulitermes species may be differentiated based on size, color and ocelli characters (Scheffrahn and Su 1994). Unfortunately this caste is only seasonal, and not always available. There are a number of morphological keys to differentiate among soldiers of North American

Reticulitermes species (Snyder 1954, Weesner 1965a, Nutting 1990,

Scheffrahn and Su 1994, Hostettler et al. 1995) based primarily on populations occurring in the southernmost part of the United States. The worker caste is not useful for the purpose of identification. Some authors claim the necessity of new approaches since trying to identify colonies using available morphological and morphometric characters of Reticulitermes species, may be confusing. Novel identification techniques comprise the use of specific behavior (Haverty, 2003 #18; Delphia, 2003 #126); analytical chemistry of cuticular hydrocarbons (Page et al. 2002) and soldier defense secretions (Nelson et al. 2001). More recently, molecular techniques such us

PCR-RFLP, and sequencing of portions of nuclear and mitochondrial genes

4 4 have been incorporated as useful techniques that allow separation of species as well as provide meaningful information regarding phylogenetic relationships of Reticulitermes from the United States (Jenkins et al. 1998,

Szalanski et al. 2003, Foster et al. 2004, Austin et al. 2005)

It is known that environmental factors such as temperature and humidity affect geographical distribution of subterranean termites (Kofoid 1934a).

Temperature and moisture are the main environmental factors affecting both geographical and local distribution of termite genera and species (Kofoid

1934a, Collins 1991). Critical temperatures define ecological or behavioral tolerance (Hu and Appel 2004) and temperature changes may influence seasonal, local and diurnal distribution. Local occurrence of termites is determined by moisture (atmospheric and soil), temperature that regulates relative humidity, and moisture that accelerates, restricts, or inhibits vital processes in termites(Kofoid 1934a, b, Collins 1991).

Reticulitermes flavipes critical thermal limits are broad; this species may survive in either in severe winter conditions (Ontario) or high summer temperatures (Texas and south states) but in general higher humidity than other Reticulitermes species. Seasonal variation in soil temperatures may for example, affect R. flavipes (Hu and Appel 2004). Critical temperatures and humidity preferences of the other Reticulitermes species known to occur in the United States remain unknown. Reticulitermes virginicus is known to occur in overlapping regions with R. flavipes in northern limits with the average annual minimum temperature no less than -10o F (Emerson 1952,

Weesner 1970, Messenger et al. 2002, Messenger 2003a, Austin et al. 2005).

5 5 Reticulitermes hageni on the other hand overlaps R. virginicus but extents

further to the west into Kansas and Oklahoma (Weesner 1970).

Despite the economic importance of Reticulitermes species little information is known regarding their diversity, habitat preference, and genetic variability of populations from Missouri. In this study, I will examine which species occur within the state and approximate their geographical distribution. Since the differentiation of sympatric Reticulitermes species is difficult in the absence of reproductives or almost impossible in the absence of soldiers and adults, I will use traditional as well as molecular approaches to differentiate the species with certainty.

Also, since Reticulitermes populations occurring in forested environments are rarely included in studies, I examined species composition and compared abundance among different environments that are conducive for the occurrence of subterranean termites i.e. urban vs woodlands. To better understand the structure of Reticulitermes populations I used sequenced a portion of 16S mtDNA to better characterize intraspecific genetic variability of individual species from populations occurring in both urban environments and woodlands of Missouri.

Lastly, I characterized the morphology of Reticulitermes occurring in

Missouri using a morphological approach supported by descriptive as well as by multivariate statistical analyses in an attempt to better differentiate among species. A combination of morphology supported by molecular identification was used to explore genetic and environmental influence on the size of soldiers of Reticulitermes flavipes.

6 6 This study suggests that four Reticulitermes species occur in Missouri:

Reticulitermes flavipes (Kollar), Reticulitermes virginicus (Banks),

Reticulitermes tibialis Banks, and Reticulitermes hageni Banks.

Reticulitermes flavipes and R. hageni, the more abundant species found in the sampling, exhibited contrasting preferences of habitat. Reticulitermes flavipes was found throughout the state but more abundantly in urban environments, while R. hageni predominates in woodlands. Reticulitermes virginicus and R. tibialis conversely were scarce in the sampling and mostly occurred in urban infestations of Missouri. This study provides evidence of the occurrence of R. virginicus and R. hageni further north than known before for the state. Reticulitermes virginicus was found in Atchinson Co, and

R. hageni was found in Atchinson and Adair Co.

Intraspecific genetic variability revealed that Missouri have the highest diversity in terms of number of haplotypes for R. flavipes when compared

with neighboring states. Twenty one haplotypes were identified for this

species in Missouri. Twelve of these haplotypes are already reported in the

47 haplotypes known for R. flavipes in the United States, while the rest nine

haplotypes have not yet been registered.

This study provides novel information regarding the morphology and genetic variation of R. hageni native populations from Missouri since most of the specimens were collected in woodlands. This specie presented the smallest variation in terms of haplotype number, with the two more common haplotypes shared in both urban and woodlands environments. Urban and woodlands of Missouri share a number of haplotypes of R. hageni and R.

7 7 flavipes suggesting that most of the urban infestations may have originated from native forest environments.

Finally, my study provides characterization, comparison, and statistical inferences regarding the morphology of the four Reticulitermes species in

Missouri and from populations of R. flavipes of Missouri not previously studied. Morphometric data gathered from Missouri will help to improve the certainty of identification for samples collected in the state. Differences in size of individual metric characters from this study where observed with metric parameters from populations from the southeast states from whom the current available diagnostic keys were developed. In addition, this study provides preliminary information suggesting environmental or genetically induced phenotypic variation of R. flavipes, a phenomenon known to occur but scarcely documented.

8 9 Chapter 1

Diversity and genetic variation of Reticulitermes

species occurring in woodlands and urban

environments of Missouri

Abstract

In order to examine diversity, habitat preferences and intraspecific variation of the populations of Reticulitermes (Isoptera: Rhinotermitidae) from Missouri, approximately 600 colonies were sampled through the state during 2004 and 2005.

Sampling included populations from nine conservation areas, nine cities located nearby and home infestations occurring in different locations within the state. Samples containing either adults or soldiers were first identified to the species level using morphology and morphometric characters while the samples containing only soldiers were identified using sequencing of mtDNA.

A portion of approximately 428 bp from the 16S mtDNA from three hundred and sixty five colony samples were used to confirm morphology based species identification, to identify samples not containing diagnostic castes and to describe intraspecific genetic variability of populations of

Reticulitermes species from Missouri.

9 10 Reticulitermes flavipes (Kollar), Reticulitermes virginicus (Banks),

Reticulitermes tibialis Banks and Reticulitermes hageni Banks were found occurring in Missouri.

Reticulitermes flavipes occurs through the state, predominating in urban

environments and home infestations rather than in woodlands, while R.

hageni predominated in the woodlands of Missouri. Reticulitermes virginicus

and R. tibialis were found scarcely and scattered within the state.

Reticulitermes hageni and R. virginicus were found in a further north range

than previously known within the state. Conversely, R. tibialis wide range of

occurrence within Missouri was not confirmed in this study; on the contrary,

it was found scarcely and only in urban environments and home infestations.

Twenty one haplotypes of Reticulitermes flavipes were found in Missouri, the highest diversity registered to date for a single state. Moreover, eight of those 21 haplotypes have not been registered before in the United States.

The more abundant haplotypes of R. flavipes and R. hageni occur in woodlands, urban environments and home infestations, suggesting the probable local origin of most of the infestations occurring in the state.

10 11 Introduction

Subterranean termites of the genus Reticulitermes play an important role in both natural and urban environments of the neartic region. Reticulitermes are important decomposers of fallen trees and other dead wood accumulations in many native forest of North America (La Fage and Nutting

1978), and thus have an important role in tunneling and wood removal

(Gentry and Whitford 1982). In urban environments, Reticulitermes species are some of the most economically important wood-destroying pests in the

United States; they are responsible for an estimated cost of 2.5 million dollars a year in chemical treatment and may reach 11 billion annually when damage repair is included (Nutting 1990, Su and Scheffrah 1990, Lewis

1997, Su and Scheffrah 1998, Su 2002).

Despite the ecological and economic importance of subterranean termites

in the genus Reticulitermes (Rhinotermitidae: Heterotermitinae), the

diversity, geographical distribution and habitat preferences of individual

species occurring in the United States remain only partially understood

(Haverty and Nelson 1997, Jenkins et al. 2000, Nelson et al. 2001, Austin et

al. 2004c, Austin et al. 2004b, Austin et al. 2004a, Szalanski et al. 2006).

Until recently, Reticulitermes flavipes (Kollar), Reticulitermes virginicus

(Banks), Reticulitermes hesperus Banks, Reticulitermes tibialis Banks and

Reticulitermes hageni Banks and Reticulitermes arenincola Goellner were the

six species of Reticulitermes recognized in the Neartic (Weesner 1970). More

recent findings based on chemical and molecular characters have suggested

11 12 the occurrence of at least one additional Reticulitermes species occurring in the United States (Jenkins et al. 2000, Page et al. 2002, Szalanski et al.

2006).

Most of the current available data of Reticulitermes distribution in the

United States is based on samples gathered from home infestations, but little

is known about populations inhabiting forested environments. The need to

have a better understanding of biology and infestation dynamics of the

Reticulitermes species that become pests on urban environments are calling

more attention to more intense local sampling efforts in urban habitats and

undisturbed habitats (Howard et al. 1982, Vargo 2003, Austin et al. 2005,

Vargo and Carlson 2006) and more attention is needed to human activities

extending into previously undisturbed woodlands (Behr 1973).

Missouri’s subterranean termite faunal composition and geographical

distribution needs to be updated and clarified. Many of the records are more

than 50 years old and others provide conflicting information regarding

occurrence and distribution of species. The available data also do not

represent all the environments of the state that are favorable for termite

activity.

Early records of termite species occurring in Missouri (Snyder 1954), included R. flavipes, R. tibialis and R. virginicus. Later, however, R. virginicus occurrence was not recognized (Weesner 1970). Nutting (1990) mapped R. virginicus and R. hageni as occurring in an overlapping range in the southern area of the state (Nutting 1990), and R. tibialis was reported as occurring widespread in the state. Recent data confirms the occurrence of R. flavipes

12 13 throughout Missouri, but there is still a lack of information regarding the occurrence or distribution of other species (Messenger 2003b, Austin et al.

2005).

Missouri’s climatic regime and natural conditions are well-suited for the establishment of Reticulitermes populations. Missouri occupies a central continental position between the great grassland biome to the west and the great forest biome to the east, leading to a complex mixture of grassland, savanna, woodland, and forest with a continental type of climate marked by strong seasonality. The state exhibits regional differences in climate with climatic changes following a basic gradient along a line crossing the state from northwest to southeast (Nigh and Schroeder 2002).

Biological, ecological and pest management research is necessary to

improve the understanding and management of Reticulitermes species in

Missouri. However, some problems rely on accurate species identification.

Differentiation among the North American Reticulitermes species is

challenging since the most reliable morphological and morphometrical

characters occur on the adult stage that is only found during one season of

the year. Characters of the soldier caste may be used but they often do not

exhibit enough interspecific variability, so (Weesner 1965b) identification

based only on soldiers is considered ambiguous. Besides morphological and

morphometric characters, chemical analysis of cuticular hydrocarbons (Page

et al. 2002), soldier secretions (Nelson et al. 2001), and behavior (Delphia et

al. 2003, Haverty et al. 2003) have been analyzed to establish differences

among the most common Reticulitermes species from North America.

13 14 Recently, molecular techniques such as PCR-RFLP, and sequencing of portions of nuclear and mitochondrial genes have been incorporated as useful techniques that allow separation of species as well as provide meaningful information regarding phylogenetic relationships of Reticulitermes from the

United States (Jenkins et al. 1998, Szalanski et al. 2003, Foster et al. 2004,

Austin et al. 2005).

The objectives of this research were to update the species diversity and geographical distribution of Reticulitermes species occurring in Missouri, to compare their habitat preferences, and to characterize and compare intraspecific genetic diversity of populations occurring in urban and forested habitats.

14 15 Methods

Study area

Current land cover in Missouri has resulted from a combination of natural

processes and human influence resulting in an overall pattern of intense

fragmentation into small tracts. As a result of the historical processes of

intensive farming on prairies followed by reduction of agriculture activities,

Northern Missouri exhibits a complex mixture of cropland on smoother

surfaces and better soils, pastures on irregular surfaces and eroded soils, and

woodlands on steeper slopes and rougher tracts (Nigh and Schroeder 2002).

The Osage Plains of west central Missouri possess more land dedicated to

cropland, pastures, and native prairies. In the Ozark region, croplands and

pasture predominate in the western Ozarks while the eastern Ozarks are the

most thoroughly forested regions of Missouri and are composed mostly of

second-growth forest with different types of management (Nigh and

Schroeder 2002).

Samples of termite colonies were collected during 2004 and 2005 from nine conservation areas, nine nearby cities, and from home infestations throughout the state of Missouri. Conservation areas having predominantly woody vegetation administered by the Missouri Department of Conservation

(collection permit 12414) were chosen in the northern, central and southern regions of the state (Table 1). Sampling in these areas was done during spring (May - June 2004) in all conservation area woodlands. Sampling was repeated during the fall (September 2004) for three conservation areas, one

15 16 in each of the northern, central, and southern regions (Rudolf Bennit C.A.,

Mark Twain N.F., Union Ridge C.A.). With the exception of Mark Twain

National Forest, on which vegetation was predominantly short leaf pine

(Pinus echinata P. Mill), forest vegetation in natural areas varied in

composition, age, and density but mainly consisted of different species of

oaks (Quercus spp.). Other species common in the areas were cottonwood

(Populus sp.), willow (Salix sp.), hickory (Carya sp.) and American elm

(Ulmus sp.).

Each conservation area was sampled using linear transects running perpendicular to roadways. Transects were 2 m wide and varied in length according to the distance needed to collect at least five samples. In order to avoid collecting multiple samples from the same colony, at least 15 m separated individual samples from one another. Samples were obtained along each transect by examining the soil underneath logs, branches and stumps that were in contact with the soil for the presence of live termites.

When living termites where observed, all available castes where collected in

80-90% ethanol using an aspirator.

Approximately, 100 m separated one transect from another (Figure 1).

The direction for sampling the first transect was chosen randomly by flipping a coin and the next transects alternated from one side of the road to the other. A total of four or five transects were used for sampling in each conservation area in order to get 20 – 25 samples.

Nine cities located near the wooded conservation areas were chosen to obtain termite samples from the environment around homes, but not from

16 17 the homes themselves. Wood piles and other wood fragments in contact with the soil surface in urban areas were examined for termite infestations in

Maryville, Kirksville, Platte City, Columbia, O’Fallon, Joplin, Springfield,

Salem, and Cape Girardeau. We call the samples collected in this manner

“urban samples” to distinguish them from samples obtained from infested homes. Termite collection in urban areas was done during May of 2004 and

2005. Geographic coordinates were recorded for each sample we collected in woodland and urban environments (Figure 1, 2) using a Garmin TM GPS IV.

Samples from infested homes were obtained through cooperation with pest management companies across Missouri. We searched the Missouri Pest

Management Association’s database in 2004 and sent a letter to all listed companies outlining the nature of the study and requesting their cooperation.

Of the letters sent, 33 companies agreed to cooperate.

I assembled kits containing materials for collecting samples and mailed a kit to each of the 33 interested companies. Kits included numbered plastic vials containing 80% ethyl alcohol, paint brushes, data sheets, detailed collecting instructions, and prepaid mailing envelopes to return collected samples to our laboratory. Fifteen of the original 33 (from 200 initially invited to participate) pest management companies that agreed to cooperate returned samples to our laboratory. Besides exact address, zip code and type of infestation, each sample included the time and date of swarming when known.

A total of 611 samples were collected: 260 samples were collected in woodlands, 188 from home infestations, and 163 from urban areas. Figure 3

17 18 summarizes all the counties and sources from which samples where collected within the state.

All termite samples were collected and preserved in vials containing 80-

90% alcohol. In the laboratory, samples were examined to determine the number of specimens and castes available for identification, cleaned of debris, and the alcohol replaced with 95% ethyl alcohol. Samples are kept at the Household Insect Laboratory Collection at the University of Missouri.

Forty-six of these samples remain unidentified and were not included in the analysis presented on the results section.

Species Identification

Morphology and Morphometrics. Identification of samples that contained

alates was done following (Scheffrahn and Su 1994). Identification of

samples that did not contain alates but contained soldiers was done using

morphological diagnostics keys for soldiers (Scheffrahn and Su 1994,

Hostettler et al. 1995). In addition, nine anatomical features of soldiers were

measured morphometrically in order to compare and identify size

characteristics that could be helpful to separate species. The anatomical

features measured included: labral length and width, pronotal length and

width, length of head with and without mandibles, width of head at the

widest point and gular length and width.

Molecular characters. Identification of samples that did not include the alate or soldier castes was done by matching the DNA sequences obtained from a portion of the 16S rRNA gene against previously published DNA

18 19 sequences of Reticulitermes from the United States and then reconfirmed by grouping them on separate clades using phylogenic analysis. Samples containing soldiers were also sequenced to confirm accuracy of the tentative identifications that were made using morphology and to establish a reference for morphometric measures of anatomical features. Most of the species were corroborated using NEB V2.0 (Vincze 2003) to cut the sequences with the enzymes DraI and TspRI then compare the correspondence of the restriction fragments obtained with previously published sites and fragments of

Reticulitermes species from the same gene and amplicon (Szalanski et al.

2003).

In order to generate molecular data for Reticulitermes species occurring in

Missouri that are comparable with available information from other states, I followed extraction and amplification protocols with slight modifications from previous studies (Austin et al. 2004c, Austin et al. 2004b, Austin et al.

2004a, Austin et al. 2005).

To prevent masking of population variation, genomic DNA was extracted

from one individual termite per colony using the complete body (Jenkins et

al. 1998) following the protocol described by (Austin et al. 2005). We used

the PurigeneTM DNA purification system D-5000A from Gentra® and DNA pellets obtained during extraction were rehydrated in 25 microliters of TE

(Tris: EDTA).

An amplicon of approximately 428 bp from the 16S rRNA gene was targeted for sequencing by polymerase chain reaction (PCR) using the previously established forward primer: LR-J-13007 5’-

19 20 TTACGCTGTTATCCCTAA-3’ (Kambhampati and Smith 1995) and reverse primer LR-N-133989 5’-CGCCTGTTTATCAAAAACAT-3’ (Simon et al. 1994).

PCR conditions were 94o C for 60 s, 35 cycles of 94oC for 45 s, 46oC for 45 s and 72oC for 45 s and 72o C 5’. PCR reactions were conducted in an

Eppendorf® Mastercycler Personal using 50 microliters of a master mix (PCR buffer 10 mM; DNTP’s 0.2 mM each; MgCl2 2 mM; primers 0.5 µM each, and

Taq polymerase 1 unit) and 1 or 2 microliters of the hydrated DNA solution

as a template. PCR amplified products were purified using the QIAquick®

PCR purification kit. Amplified PCR products were submitted for direct

sequencing in both directions to the DNA Core facility of the University of

Missouri-Columbia.

Contigs of individual sequences were obtained using Sequencher®4.7 and/or Bioedit® 7 (Hall 1999). Multiple alignments of all sequences were done using Clustal X® and then manually refined using Bioedit®7 (Hall

1999).

Data analysis

Frequency of occurrence of Reticulitermes species was compared among habitats, localities, and geographical locations within the state using the non- parametric Chi-square test. Small P values are indicative that the data were not sampled from the same distribution. Geographical location within the state was assigned based on a general Missouri map of ecoregions.

20 21 Intraspecific variability

Several intraspecific genetic variability parameters were calculated using the software DnaSP® version 4 (Rozas et al. 2003):

1. Nucleotide diversity per site (Pi): the average number of nucleotide

differences per site between two sequences.

2. Nucleotide diversity per site (Pi): mean sequence divergence among

haplotypes.

3. Haplotype diversity (Hd): number of different forms of a sequence.

4. Tajima’s test of neutrality compares the number of segregating sites

per site with nucleotide diversity. (A site is considered segregating if,

in a comparison of m sequences, there are two or more nucleotides at

that site). If all the alleles are selectively neutral, then the product 4Nv

(where N is the effective population size and v is the mutation rate per

site) can be estimated in two ways, and the difference in the estimate

obtained provides an indication of non-neutral evolution. This test is

based on the fact that under the neutral model, estimates of the

number of segregating sites and of the average number of nucleotide

differences are correlated (Rozas et al. 2003).

Phylogenetic Inferences

In order to reduce computing time and to improve the chances of recovering fewer optimal trees, haplotypes were used to establish phylogenetic relationships among Reticulitermes species. Coptotermes

21 22 formosanus (Shiraki) was chosen as the outgroup taxon. The most parsimonious tree was constructed conducting a heuristic search and bootstrapped 1000 times using Winclada (Nixon 2000).

A neighbor-joining tree was also constructed to search for the optimal tree. The optimal tree was bootstrapped 1000 times using the software Mega version 3.1 (Kumar et al. 2004).

To explore intraspecific relationships between haplotypes and identify relationships between ancestral haplotypes and their descendent haplotypes

(that may not be properly inferred from traditional phylogenetic methods), a statistical approach was used to construct genealogies of haplotype networks for individual species using TCS software version 2.1 (Posada and Crandall

2000). The TCS algorithm calculates the probability of a parsimonious connection between a pair of haplotypes. Connections supported with 95% probability are joined into networks as cladograms having a 95% confidence

(Templeton et al. 1992).

22 23 Results

Diversity

Reticulitermes flavipes, R. hageni, R. virginicus and R. tibialis were found throughout the state of Missouri. Reticulitermes flavipes, R. hageni, and R. virginicus were identified using morphological characters of alates and/or soldiers, and most of these identifications were confirmed with molecular data. Reticulitermes tibialis was identified only by molecular data obtained from workers.

Significantly greater proportion of R. flavipes (70%, 358/511) than R. hageni (24.7%, 126/511) (χ2=11.207, P=0, df=1), R. virginicus (3.9%,

20/511) (χ2=302.333, P=0, df=1) and R. tibialis (1.4%, 7/511) (χ2=337.54,

P=0, df=1) was found in the total number of samples identified (Table 2).

Habitat Preferences

Reticulitermes flavipes, R. hageni and R. virginicus were found in home

infestations, urban environments and woodlands while R. tibialis was found only in home infestations and urban environments and not in woodland areas

(Table 2, Figures 4, 5).

Individual Reticulitermes species frequency of occurrence differed among habitats of Missouri. I found greater proportion of R. flavipes (87.4%,

159/182) than R. hageni (3.8%, 7/182), R. virginicus (6.6%, 12 /182) and

R. tibialis (2.2%, 4 /182) in home infestations (χ2=139.181, P=0, df=1;

χ2=126.368, P=0, df=1; χ2=147.393, P=0, df=1).

23 24 Reticulitermes flavipes was proportionally (82.4%, 122/148), greater than

R. hageni (11.5%, 17/148), R. virginicus (4.1%, 6/148) and R. tibialis (2%,

3 /148) in urban environments (χ2=79.317, P=0, df=1; χ2=105.125, P=0,

df=1; χ2= 113.288, P=0, df=1). Greater proportion of R. hageni than R.

virginicus and R. tibialis (χ2=5.261, P=0.03, df=1; χ2=, P=0.001, df=1) and

no difference between R. virginicus and R. tibialis.

Reticulitermes hageni relative frequency in woodlands (56.4%, 102/181) was greater than R. flavipes (42.5%, 77/181) and R. virginicus (1.1%,

2/181) (χ2=3.492, P=0.061, df=1; χ2=96.16, P=0, df=1). R. flavipes proportion was greater than R. virginicus in woodlands (χ2=71.203, P=0,

df=1). Reticulitermes tibialis was not found in woodlands (Figure 6).

Species Distribution

A total of 57 cities and 34 counties in Missouri were sampled during this study when adding sampling from home infestations, urban environments and woodlands. Reticulitermes flavipes was found in 100% of the counties

from which samples were collected; R. hageni was found in 44.4 %, (15/34);

R. virginicus in 32.4% (11/34); and R. tibialis in 6% (2/34) (Table 3).

Samples from home infestations were collected from 52 cities grouped in

25 counties (Table 4). R. flavipes was collected in greater number of cities

(86.5%, 45/52) compared with R. hageni (7.7%, 4/52), R. virginicus

(21.2%, 11/52), and R. tibialis (3.8 %, 2/52) (χ2= 34.06, P=0, df=1;

χ2=20.643, P≤0.005, df=1; χ2=39.34, P=0, df=1).

24 25 Reticulitermes flavipes occurred in 100% (9), R. hageni was found in 66%

(3/9), R. virginicus 44.4 % (4/9) and R. tibialis in 11.1% (1/9) of the nine

cities sampled around homes (Table 5)

Reticulitermes flavipes, R. hageni and R. virginicus were all found in woodland areas of Missouri (Table 6). Reticulitermes tibialis was not found in

any samples taken from woodlands. Reticulitermes flavipes was found in

100% (9), R. hageni in 100% (9), and R. virginicus in 2.2% (2/9) of all nine woodland areas.

Most samples from home infestations were collected from the central region of the state: 71.9% (131/182). Reticulitermes flavipes was identified

in 87.8% (115/131) of these samples; R. hageni in 5.3% (7/131) and R.

tibialis in 0.76% (1/131). Samples received from the southern regions of the

state (25.8%, 47/182) were also composed mostly of R. flavipes (85.1%,

40/47), while R. tibialis occurred in 6.4% (4/47) and R. virginicus in 8.5%

(4/47). Only 2.2% (4/182) of the total samples from home infestations were

collected in the north region and all corresponded to R. flavipes.

Reticulitermes flavipes and R. hageni frequency of occurrence in woodlands did not change significantly among geographical locations when comparing woodlands from the north (Brickyard Hill C.A., Union Ridge C.A. and Platte Falls C.A.), central (Rudolf Bennit C.A., Weldon Spring C.A., Mark

Twain N.F.) and southern regions of the state (Fort Crowder C.A., Compton

Hollow C.A. and Apple Creek C.A.). Reticulitermes virginicus was not found in woodland areas sampled from the central region of the state.

25 26 Reticulitermes hageni and R. virginicus were found farther north than previously known within the state. Reticulitermes hageni was found in the counties of Atchison (Brickyard Hill C.A.) and Nodaway (Maryville) in the northwestern and in Adair (Union Ridge C.A.) in the northeastern part of the state. Reticulitermes virginicus was found in one sample from the northwestern part of Missouri in the Brickyard Hill conservation area in

Atchison County.

Although the purpose of this study was not to measure the influence of

wood availability on foraging frequency, it was apparent that foraging was

scarce in conservation areas where mulch accumulation or pieces of wood

bigger than 5 cm in diameter were scarce. That situation was observed in

Platte Falls C.A. and some steep areas of Brickyard Hill C.A. In contrast, in

areas with abundant pieces of wood, such as Mark Twain N. F., termite

colonies were relatively easy to find. Termites were more commonly found

foraging above ground in pieces of wood showing medium to advanced

decomposition. In urban environments most samples were collected from

woodpiles around homes under the logs in contact with the soil surface, in

mulches, and in wooden fences.

Seasonality of Reticulitermes species diversity and frequency in woodlands.

Species composition did not vary between spring and fall of 2004 in the three conservation areas that were sampled twice during the same year

26 27 (Union Ridge C.A., Rudolf Bennit C.A., and Mark Twain N.F.). Reticulitermes flavipes, and R. hageni were found in both sampling seasons (Table 7).

Reticulitermes hageni frequency (90.7%, 49/54) was 26.2% higher during the fall compared with spring (64.5%, 40/62) (adding frequencies of the three woodland areas); the opposite occurred with R. flavipes during the fall, its frequency of occurrence (9.3%, 5/54) decreased by 26.2% compared with the spring (35.5%, 22/62).

Reticulitermes hageni frequency was 47.2 % higher than R. flavipes in

Union Ridge C.A. (χ2=38.354, P≤0.05, df=1), 61.9% higher than R. flavipes

in Mark Twain N. F. (χ2=5.261, P≤0.05, df=1) and similar than R. flavipes in

Rudolf Bennit C.A. during spring season (χ2=2, P≤0.16, df=1) . However,

during the fall season, greater proportion of R. hageni than R. flavipes was

found in samples from in all three woodlands: Union Ridge C.A. (χ2=12.25,

df=1), Mark Twain N.F. (χ2=12.8, P≤0.001, df=1) and Rudolf Bennit C.A.

(χ2=10.889, P≤0.001, df=1).

Intraspecific variability of Reticulitermes species in Missouri Based on 16S

mtDNA

Genetic Variability. The average nucleotides from all sequences were T:

22.9, C: 22.8, A: 41.3, G: 13.0. A total of 33 haplotypes were identified from

the 364 sequences of Reticulitermes species included in this analysis.

Nucleotide variability and number of samples for the 33 Reticulitermes

haplotypes is showed in Table 8.

27 28 Twenty one haplotypes were found of R. flavipes, three of R. hageni, four of R. virginicus, and five of R. tibialis. R. flavipes highest haplotype diversity was sampled in urban environments; R. hageni highest diversity was sampled in woodlands; and R. virginicus and R. tibialis highest diversity

occurred in home infestations (Table 9).

Three unique haplotypes were found in R. flavipes (two occurring in home

infestations and one from the urban environment); four unique haplotypes

were found in R. tibialis (three in home infestations and one from the urban

environment). No unique haplotypes were found for R. virginicus (Table

10). Unique haplotypes are haplotypes identified from only one sample.

A summary of absolute frequency of occurrence of all 21 R. flavipes haplotypes in the nine woodlands and nearby cities is presented in Figure 7.

The most abundant R. flavipes haplotypes were RFM5, RFM2 and RFM1

(33%; 16.2% and 15.3% of the total samples of R. flavipes sequenced).

These haplotypes were abundant in all three environments sampled.

Haplotype RFM5 occurred in 64%, haplotype RFM1 occurred in 47.2% and haplotype RFM2 in 44.4% of all counties from which we analyzed samples.

Specific locations of Reticulitermes haplotypes are showed in Table 11, 12 and Figure 8.

Reticulitermes hageni displayed the lowest diversity in terms of number of haplotypes. Haplotype RHM2 was the most abundant around homes and woodlands and was found in 82% of the counties in which R. hageni was

sampled (Figure 9).

28 29 Despite the small number of samples of R. virginicus and R. tibialis, these species were very diverse in terms of the number of haplotypes.

Reticulitermes virginicus haplotypes RVM1 and RVM2 occurred in two distant counties once, while haplotypes RVM1 and RVM3 occurred in the same county. Reticulitermes tibialis occurred in two distant localities, one of which had four different haplotypes (Greene county), while the other location had only one haplotype, which was different from the four haplotypes in Greene county (Figures 10, 11).

Specific genetic variability parameters including haplotype (gene) diversity

(Hd), nucleotide diversity (Pi), number of variable sites (S), number of mutations (Eta) and Tajima’s D test of neutrality are summarized in tables

13-16. Tajima’s test (P<0.05) confirmed the hypothesis of neutrality of 16s rRNA. Tajima-Nei distances ranged from 0.003 to 0.008 within R. flavipes,

0.003 to 0.005 within R. hageni, 0.003 to 0.041 within R. virginicus and

0.003 - 0.011 within R. tibialis.

Phylogenetic analysis

The matrix including all Reticulitermes species and the outgroup taxon included 432 characters (385 total number of sites excluding sites with gaps/missing data). A total of 311 sites were invariable (monomorphic), 38 sites were variable (polymorphic; total number of mutations: 41), and of these, 32 were parsimony informative. Results from the neighbor- joining tree and the most parsimonious tree were both consistent in the monophyletic nature of each Reticulitermes species. Phylogenetic

29 30 relationships among the Reticulitermes species depicted on the neigbor- joining tree and the most parsimonious tree are congruent, therefore only the neighbor – joining tree is shown (Figure 12). Correct species grouping of haplotypes is well supported by boostrap values (>88). However, bootstrapp values of the nodes supporting relationships among Reticulitermes species had lower support (~60) leading to unresolved interrelationships among species through the use of this portion of the 16S mtDNA gene.

The networks constructed using TCS 2.1 are shown in Figure 13.

Pairwise connections of up to 8 mutational steps are resolved with 95% confidence. based on abundance and mutational steps haplotypes RFM5,

RH3, RV4 and RT2 are suggested as the more ancestral haplotypes of R.

flavipes, R. hageni, R. virginicus and R. tibialis respectively. There are few

loops in the R. flavipes network, those rings which may correspond to

homoplasy (convergent nucleotide substitutions) or are derived from

sequencing error.

30 31 Discussion

Reticulitermes flavipes, R. hageni, R. virginicus and R. tibialis were found

in all three environments studied within the state of Missouri: home

infestations, urban environments occurring around homes, and woodlands.

This finding contrasts with previous studies that either exclude R. virginicus or R. hageni (Snyder 1954, Weesner 1970) from the Missouri termite fauna.

In the maps depicting the geographical distribution of R. virginicus and R. hageni in the United States presented by Nutting (1990) the northernmost limit of occurrence of these species appears in the south of Missouri.

However, in our study we found both species occurring in a wider range that includes Atchison County, the most distant county in the northwest corner of

Missouri. In contrast, R. tibialis occurred over most of the state of Missouri

(excluding the southeast corner) in our study was found scarcely and was never found in woodlands.

Reticulitermes diversity and abundance in Missouri is similar to the neighboring states Oklahoma and Arkansas where R. flavipes is the main home infesting species and R. hageni and R. virginicus occur less frequently

(Austin et al. 2004a, Brown et al. 2004, Uva et al. 2004).To date, R. tibialis is not recorded in Arkansas (Austin et al. 2004a).

Reticulitermes flavipes was found occurring in all locations and

environments from which samples were collected within Missouri. Its wide

geographic distribution and abundance in urban environments of Missouri,

agreed with the known wide distribution of this species from home

31 32 infestations in the United States (Nutting 1990, Messenger 2003b).

Reticulitermes flavipes was also the dominant species occurring around homes; however, it was usually less frequent than R. hageni in woodland areas.

Reticulitermes hageni was the second most abundant species found in the state, occurring most frequently in woodlands. In fact, it was found in all the wooded areas that we sampled regardless of the location within the state.

This study shows the first records of a more northeast and northwestern occurrence of this species within Missouri than before published (Atchison and Adair counties).

Reticulitermes virginicus and R. tibialis were the scarcest species found within the state. The distribution of R. virginicus reaches a more northwestern area than formerly published for Missouri (Atchison County).

Whether populations of R. virginicus from this part of the state are native or introduced remains unknown. The fact that our R. virginicus sample was collected in one forested area of Brickyard Hill C.A. and that it belongs to the more ancestral 16s RNA haplotype proposed for this species supports the probable native nature of their population in this part of the state. However, given that previously this species was unknown from northern locations including the Nebraska-Missouri border and Nebraska, the possible anthropogenic origin is also possible. The anthropogenic origins of this and other Reticulitermes species in different regions of the United States is discussed in the recent literature (Austin et al. 2006). Given that the haplotypes of R. virginicus collected from the Missouri-Nebraska border at

32 33 Brickyard Hill C.A. (V4) and the haplotypes from Nebraska (V2) are different, it is also possible that each one may have a different origin.

Reticulitermes tibialis was not found in the woodlands studied in Missouri.

This result strongly contradicts the wide distribution of R. tibialis within the

state and instead suggests introduction associated with home infestations.

There are only two previous records of R. tibialis occurring in or near

Missouri, one in Quincy (Illinois) and another in the north west border of

Missouri and Kansas (Messenger 2003b, Austin et al. 2006). Both of these

records were associated with home infestations.

Flight records in Missouri have suggested that R. hageni and R. virginicus

overlap in the southern region of the state (Nutting 1990). According to our

study, these two species occur in overlapping regions along with R. flavipes

(with the exception of the northeast limit of the state in which R. virginicus

was not found). Reticulitermes virginicus was found less frequently within our

samples but its geographic range in Missouri still appears to overlap with R.

hageni and R. flavipes.

There are no previous records of the relative frequency of Missouri

Reticulitermes species in woody areas or forests of Missouri and neighboring

states and there are few comparable records from any other states. The

closest records are from the Delta of Mississippi and South Carolina forests

where R. flavipes and R. virginicus are the predominant species (Gentry and

Whitford 1982, Howard et al. 1982), unlike Missouri where R. flavipes and R. hageni predominate woodlands. Reticulitermes hageni and R. virginicus were found more frequently in woody areas than in urban areas and this agrees

33 34 with observations from states where these species are more common in undisturbed habitats or areas where forests are prevalent (Gentry and

Whitford 1982, Howard et al. 1982, Austin et al. 2004b, Vargo et al. 2006).

It is known that abiotic factors may affect the frequency of surface foraging in sympatric species of subterranean termites (Kofoid 1934a,

Houseman et al., Houseman and Gold 2003). Individual Reticulitermes species respond differently to moisture and temperature. For example; changes in soil temperature and humidity, as a result of autumn rain, stimulate foraging activity of R. tibialis but diminish foraging of Amitermes sp. (Kofoid 1934). Reticulitermes flavipes tolerates higher levels of moisture than R. hageni or R. virginicus and may survive better in environments favoring damp conditions (Forschler and Henderson 1995). In particular, R. flavipes and R. hageni are known to coexist and share food resources by their differences in adaptive responses to changes in temperature and moisture in the soil profile. Reticulitermes hageni seems to be more active in dry and warmer periods while the opposite is observed for R. flavipes

(Houseman et al. 2001).

Although in our study fall sampling was done in slightly different locations within the same wooded areas than spring sampling, the higher relative frequency of R. flavipes in spring samples followed by a higher frequency of

R. hageni in fall samples supports previous proposed relationships between seasonal climatic variation and termite abundance documented for these species. This result may indicate that the relative frequency of occurrence

34 35 reported here for the nine conservation areas may be biased toward species more suited to environmental conditions that occur in late spring in Missouri.

Intraspecific genetic variability of Reticulitermes from Missouri

Twelve of the 21 R. flavipes haplotypes found in Missouri are already

registered in the 47 haplotypes of R. flavipes known to occur in North

America (Austin et al. 2005). The number of haplotypes of R. flavipes found in Missouri seems high when compared with the number of haplotypes known to occur in nearby states. For example 13, 10, 11, and six haplotypes had been reported for Texas, Oklahoma, Arkansas and Louisiana respectively.

However, the number of locations analyzed in those states (Austin et al.

2004a) are less than 30% of the number we analyzed from Missouri, and most of them correspond to home infestations. Haplotypes RFM1, RFM2,

RFM3, RFM4, RFM5, RFM6, RFM7, RFM12, RFM13, RFM15, RFM18 and RFM20 are equivalent with haplotypes: P, Q, M, S, F, R, Z, RR, U, V, J, and JJ respectively. Six of those haplotypes: P, Q, M, F, R, J and U also occur in

Oklahoma; haplotypes M, F,J, R, Q also occur in Arkansas; haplotypes P and

R occur in Louisiana; haplotypes P, M, F and U are found in Kansas; haplotypes P and F in Nebraska; and haplotypes M and F in Tennessee

(Austin et al. 2004b, Austin et al. 2004a, Austin et al. 2005). Unfortunately, only one haplotype is known from Illinois and therefore not enough information to compare the extent of R. flavipes haplotypes from Missouri farther toward the east (Austin et al. 2005).

35 36 The low intraspecific variability observed within R. hageni at the level of

haplotype when analyzing 16s rRNA from samples of urban environments is

also valid for populations from woodlands. In our study we found the same

two haplotypes occurring in woody areas and urban areas, corresponding

with two of the three R. hageni haplotypes that have been recorded from

North America. Haplotype RHM2 from Missouri is equivalent with haplotype

H1 and haplotype RHM3 is equivalent with H2 while no equivalent was found

in Missouri for H3 (Austin et al. 2004b).

Two of the four R. virginicus haplotypes found in Missouri are known to

occur in other states. Our RVM2 is equivalent to V1 found in Arkansas (Austin

et al. 2004a), Texas (Austin et al. 2004c), Oklahoma (Austin et al. 2004b),

and was also recently recorded in Nebraska (Austin et al. 2004c). Our RVM4

is equivalent to V3 found in Arkansas (Austin et al. 2004a). The level of

nucleotide and haplotype diversity observed for R. virginicus is similar to R.

flavipes.

Reticulitermes tibialis exhibited the highest haplotype and nucleotide diversity relative to the number of samples. Each sample collected in this study represented a different haplotype. Haplotype RTM3 correspond with haplotype T7 found in Texas, Oklahoma, and Kansas (Austin et al. 2004c,

Austin et al. 2004b, Austin et al. 2006).

The more frequent haplotypes of R. flavipes or R. hageni found in

Missouri urban environments and woodlands did not exhibit geographic or

habitat isolation. This finding is consistent with previous studies in which no

geographic structure of 16S rRNA based haplotypes have been reported for

36 37 R. flavipes involving either regional areas like nearby states or extensive areas such as populations from across the United States (Austin et al. 2004c,

Austin et al. 2004b, Austin et al. 2004a, Austin et al. 2005). Lack of geographic structure of R. flavipes haplotypes in the United States has been discussed by Austin (2004a,2004d) and may be attributed to several factors, including the probable effects of anthropogenic activities, lack of nest mate agonistic behavior (Bulmer and Traniello 2002), ability to hybridize and fuse colonies (Clement 1986), coexistence through resource partitioning

(Houseman et al. 2001), life histories promoting dispersal, sharing interconnected history through gene flow (Avise et al. 1987, Jenkins et al.

1998), and recent population bottlenecks.

Few studies have studied the genetic variation of populations of

Reticulitermes occurring in both urban and forested environments. In our study, R. flavipes and R. hageni are the predominant species in forested areas with apparent differences related to preferences of foraging season.

However, in general they share the same geographic locations and resources year long. Reticulitermes flavipes has the highest number of haplotypes in both urban and forested habitats when compared with R. hageni. Even at the transect level in woodlands, R. flavipes is comprised of several different haplotypes, while one R. hageni predominant haplotype was typically found

(H2). It is known that independent colonies of Reticulitermes formed by primary reproductive pairs can also divide by budding after some time.

Isolation mechanisms may differentiate these colonies from their ancestor and give rise to different haplotypes at that location (Nobre et al. 2006).

37 38 Additional explanations may be found when examining colony level diversity, but our genetic analysis is unable to determine whether each sample belonged to a different colony or the origin of the colony. However, characteristics of the breeding systems of these two species as revealed by genotype analysis at the microsatelite level may help to explain the nature of haplotype variability observed within R. flavipes and R. hageni (Vargo 2003,

Vargo and Carlson 2006). Most colonies of R. hageni and R. flavipes from

urban and forest environments are single families founded by pairs of

primary reproductives. There are some extended families headed by

neotenic reproductives and few mixed families originating from colony fusion.

It may be important to consider the relatively smaller distances over which

R. hageni reproductive pairs will disperse. This may lead to a greater level of

inbreeding in simple family colonies, smaller population sizes and significant

structuring of its local populations (Vargo 2003, Vargo and Carlson 2006).

Our study suggests that most of the Reticulitermes populations inhabiting Missouri’s urban environments likely originated from populations occurring in forests of Missouri given that the more frequent haplotypes of

Reticulitermes in our study were found in woodlands and the urban habitats sampled (Figure 10). For example, the most abundant haplotype of R. flavipes (RFM5) has a wide distribution within the state as well as in neighboring states and corresponds to one of the most ancestral haplotypes of R. flavipes based on the haplotype network analysis (Crandall and

Templeton 1993, Posada and Crandall 2001) we conducted. Haplotype RHM3 from R. hageni and haplotype RVM4 from R. virginicus that were found only

38 39 in Missouri’s woodlands were also considered to be as the most ancestral relative to other haplotypes identified. However, it is interesting to note that there are several R. flavipes haplotypes that only occur in home infestations or around homes. These may correspond to introductions or anthropogenic origins.

40

Table 1. Location and general information for nine Missouri conservation areas sampled for subterranean termites (2004-2005).

Forest land cover / Conservation Area County total (acres)

Brickyard Hill C.A. 1,460 / 2,262 Atchison

Union Ridge C.A. 5,194 / 7,981 Adair, Putman, Sullivan

Rudolf Bennit C.A. 2,088 / 3,515 Boone, Howard, Randolph 41

Weldon Spring C.A. 6,241/ 7,356 Saint Charles

Platte Falls 890/2,360 Platte

Fort Crowder C.A. 1,753 / 2,212 Newton

Compton Hollow C.A. 721.7 / 840 Webster

Apple Creek C.A. 1,517 / 2,100 Cape Girardeau

Mark Twain N.F. Salem Dent Ranger District

Table 2. Number of samples collected for Reticulitermes species within Missouri according to habitat. Number of samples occurring in woodlands only included those collected during spring 2004.

Species Source R. flavipes R. hageni R. tibialis R. virginicus Total

Home infestations 159 7 4 12 182 42 Urban environment 122 17 3 6 148

Woodlands 77 102 0 2 181

Total 358 126 7 20 511

Table 3. Total number of samples of each Reticulitermes species collected from home infestations, urban environments and woodlands by county.

Species Counties Total R. flavipes R. hageni R. tibialis R. virginicus Adair 4 32 0 0 36 Atchison 13 8 0 1 22 Boone 40 3 0 2 45 Caldwell 110 02 Camdem 12 1 0 0 13 Cape Girardeau 21 2 0 3 26 Cass 1 0 0 0 1 Christian 4 0 0 0 4 Clay 11 0 0 0 11 Cole 20 4 0 2 26 Dade 1 0 0 0 1 Dent 23 36 0 1 60 Dunklin 3 0 0 0 3 Franklin 1 0 0 0 1 Greene 45 4 6 4 59 Henry 100 01 Holden 1 0 0 0 1 Howard 14 22 0 0 36 Jackson 1 0 0 0 1 Jasper 19 0 0 0 19 Jhonson 10 0 0 1 11 Knox 6 0 0 0 6 Livingston 2 0 0 0 2 Morgan 2 0 0 2 4 Newton 8 10 0 1 19 Nodaway 11 1 0 0 12 Osage10001 Platte 22 12 0 0 34 Polk 1 0 0 0 1 Randolph10001 Scott 10 13 0 0 23 St. Charles 20 16 0 2 38 St. Louis 17 0 1 1 19 Stone 1 0 0 0 1 Warrenton 1 0 0 0 1 Webster 14 10 0 0 24 Total 363 175 7 20 565

43

Table 4. Number of samples of Reticulitermes species collected from home infestations in the central, southern and northern regions of Missouri during 2004 and 2005 listed by city.

Species Geographical City Total region R. R. R. R. flavipes hageni tibialis virginicus Ashland 1 0 0 0 1 Baldwin 0 0 0 1 1 Braymer 1 0 0 0 1 Brentwood 6 0 0 0 6 Camdenton 1 0 0 0 1 Centertown 0 0 0 1 1 Chesterfield 2 0 0 0 2 Clinton 1 0 0 0 1 Columbia 25 0 0 1 26 Eureka 0 0 1 0 1 Florissant 1 0 0 0 1 Gladstone 1 0 0 0 1 Grandview 1 0 0 0 1 Gravers Mills 1 0 0 0 1 Holden 1 0 0 0 1 Jefferson City 20 4 0 1 25 Johnson 1 0 0 0 1 Kansas City 13 0 0 0 13 Central Kirksville 1 0 0 0 1 Laurie 2 0 0 2 4 Liberty 10001 Odessa 1 0 0 0 1 O'Fallon 1 0 0 0 1 Pleasant 1 0 0 0 1 Polo 0 1 0 0 1 Shrewsberry 20002 St. Charles 4 0 0 1 5 St. Louis 6 1 0 0 7 Sunrise Beach 9 1 0 0 10 Town and Country 10001 Warrensburg 70018 Warrenton 1 0 0 0 1 Webster Grooves 10001 Westphalia 1 0 0 0 1 Total 115 7 1 8 131

43 44

Table 4. Continued

Species Geographical City R. R. Total region R. hageni R. tibialis flavipes virginicus Chillicothe 2 0 0 0 2 Moberly 100 01 North Parkville 1 0 0 0 1 Total 4 0 0 0 4 Battlefield 1 0 0 0 1 Billings00011 Caldwell 2 0 0 0 2 Greenfield 1 0 0 0 1 Kimberling 000 11 Nixa 2 0 0 0 2 Ozark 2 0 0 0 2 South Pardy 100 01 Republic 8 0 0 1 9 Rogersville 0 0 0 1 1 Senath 1 0 0 0 1 Springfield 19 0 3 0 22 Stratford 1 0 0 0 1 Walnut Grove 1 0 0 0 1 Total 40 0 3 4 47 Total 159 7 4 12 182

45

Table 5. Number of Reticulitermes species samples collected from urban environments in the central, southern and northern regions of Missouri during 2004 and 2005.

Geographical Species region of City R. R. R. R. Total Missouri flavipes hageni tibialis virginicus Columbia 15 3 0 1 19 O'Fallon 6 0 0 1 7 Central Salem 15 1 0 1 17

46 Total 36 4 0 3 43 Kirksville 6 0 0 0 6 Maryville 11 1 0 0 12 North Platte City 15 6 0 0 21 Total 32 7 0 0 39 Cape Girardeau 21 2 0 3 26 Joplin1900 019 South Springfield 14 4 3 0 21 Total 54 6 3 3 66

Total 122 17 3 6 148

Table 6. Number of samples of each Reticulitermes species collected from woodlands in the central, southern, and northern regions of Missouri during spring 2004.

Species Geographical region of Area name R. R. R. R. Total Missouri flavipes hageni tibialis virginicus

Weldon Spring C.A. 7150 022 Rudolf Bennit C.A. 12 6 0 0 18 Central Mark Twain N. F. 6170 023 47 Total 25 38 0 0 63 Union Ridge C.A. 4170 021 Platte Falls C.A. 360 09 North Brickyard Hill C.A. 13 8 0 1 22 Total 20 31 0 1 52 Fort Crowder C.A. 8100 119 Compton Hollow C.A. 14 10 0 0 24 South Apple Creek C.A. 10 13 0 0 23 Total 32 33 1 66 Total 77 102 0 2 181

Table 7. Species composition and number of samples of Reticulitermes species collected from wooded areas in the central (Rudolf Bennit), southern (Mark Twain) and northern (Union Ridge) regions of Missouri during spring and fall of 2004.

Sampling Species Woodland Total season R. flavipes R. hageni Rudolf Bennit C.A. 12 6 18 Mark Twain N. F. 6 17 23 Spring Union Ridge C.A. 4 17 21

48 Total 22 40 62 Rudolf Bennit C.A. 2 16 18 Mark Twain N. F. 2 18 20 Fall Union Ridge C.A. 1 15 16 Total 5 49 54

Table 8. Haplotypes of Reticulitermes species found in Missouri based on nucleotide variation using the species Coptotermes formosanus as the outgroup species. Dashes indicate same nucleotide occur at that base pair position.

Number of base pair position ( 5' to 3' end) Haplotype 128 131 133 134 139 141 146 154 157 158 162 163 166 168 169 170 171 172 179 180 250 256 271 272 275 285 364 365 366 367 369 374 375 402 407 Species 55 56 75 79 80 96 97 99 samples code

R. flavipes 36 RFM1 GCCACTATTACATGCAAGGACAT__ATCACTCGGTTCTGGTGG R. flavipes 38 RFM2 ...... G.G_..T...T......

R. flavipes 15 RFM3 ...... __...... T......

R. flavipes 5RFM4...... A....__...... T.A......

R. flavipes 76 RFM5 ...... G.A_...... T......

R. flavipes 23 RFM6 ...... __...... T.A......

R. flavipes 2RFM7....TC...... __..A......

R. flavipes 1RFM8...... T...... G.__...... T......

R. flavipes 2RFM9...... A....__......

R. flavipes 1RFM10...... A...... __......

49 R. flavipes 2RFM11...... G.__.C....T......

R. flavipes 2RFM12...... T....G.__...... T......

R. flavipes 2RFM13...... G.__.....CT......

R. flavipes 3RFM14...... G...... G.__...... T......

R. flavipes 4RFM15...... A..G.__...... T......

R. flavipes 4RFM16...... __...... TA......

R. flavipes 2RFM17...... G...... __......

R. flavipes 9RFM18...... G...... __...... T......

R. flavipes 3RFM19...... __...... T...... A

R. flavipes 3RFM20...... G.__...... TA......

R. flavipes 1RFM21...... G...... __...... T.A......

R. hageni 3RHM1A...... C.TG_...GA....C_A.....C....CA_A.C.A

R. hageni 112 RHM2 A.....T.C.TG_...GA....._A.....C....CATA.C.A

R. hageni 3RHM3A...... C.TG_...GA....._A.....C....CA_A.C.A

R. tibialis 1RTM1.T.CT. ._C. .G.. .G..AG.. .__. .T. .C. . .._A.A. ..A

R. tibialis 1RTM2.T.CT.._C..G...... AG...__..T..C...._A.A...A

R. tibialis 1RTM3.T.CT.._C...... AGA..__..T..C...._A.A...A

R. tibialis 2RTM4.T.CT.._CT.G...... AG...__..T..C...._A.A...A

R. tibialis 1RTM5.T.CTC._C..G...... AG...__T.T..C...._ .A...A

R. virginicus 2RVM1A.....T.C.TG...... G...__...GTC...C_..A.CAA

R. virginicus 3RVM2A.....T...TG...... __...GTC...C_..A.CAA

R. virginicus 1RVM3A.A...T...TG...... __...G.C...C_..A.CAA R. virginicus 1 RVM4 A.....T...TG...... G...__...GTC...C_..A.CAA

48

Table 9. Diversity and abundance of Reticulitermes species haplotypes for each habitat sampled in Missouri.

Number of haplotypes Home Urban Woodland Species Total n infestations areas areas R. flavipes 235 13 19 10 21 R. hageni 116 2 2 3 3 R. virginicus 72 1 14

50 R. tibialis 63 2 05 Total 364 20 24 14 33

Table 10. Number of samples of each Reticulitermes haplotype collected from home infestations, urban environments and woodlands of Missouri 2004-2005.

Source (number of samples) Haplotype Species Home Code Urban Woodlands Total infestations RFM1 15 8 13 36 RFM2 19 14 5 38 RFM3 7 3 5 15 RFM4 2 1 2 5 RFM5 22 33 22 77 RFM6 910423 RFM7 1 1 - 2 RFM8 1 - - 1 RFM9 -2 -2 RFM10 -1 -1 R. flavipes RFM11 -2 -2 RFM12 -2 -2 RFM13 -2 -2 RFM14 1 1 1 3 RFM15 1 3 - 4 RFM16 -4 -4 RFM17 -1 1 2 RFM18 3 5 1 9 RFM19 -1 2 3 RFM20 2 1 - 3 RFM21 1 - -1 RHM1 1 1 2 4 R. hageni RHM2 1 10 92 103 RHM3 --99 RTM1 1 - - 1 RTM2 -1 -1 R. tibialis RTM3 2 - - 2 RTM4 1 - - 1 RTM5 -1- 1 RVM1 2 - -2 RVM2 -3- 3 R. virginicus RVM3 -- 1 1 RVM4 1 - -1

51 Table 11. Location and number of samples of Reticulitermes flavipes haplotypes collected within Missouri. Haplotype code Location (number of samples)

RFM1 Apple Creek C.A. (2), Ashland (1), Bennit C.A. (1), Brentwood (1), Brickyard Hill C.A. (1), Compton Hollow C.A.(3), Fort Crowder C.A. (1), Jefferson City (4), Joplin (4), Kansas City (1), Mark Twain N. F. (3), O'Fallon (1), Pardy (1), Platte City (1), Pleasant (1), Salem (1), Springfield (3), St. Louis (2), Sunrise Beach (2), Union Ridge C.A.(2) RFM2 Bennit C.A.(1), Brentwood (1), Cape Girardeau (2), Columbia (2), Compton Hollow C.A.(3), Fort Crowder C.A.(1), Grandview (1), Jefferson City (4), Joplin (7), Kansas City (1), Kirksville (1), Maryville (2), Nixa (1), Ozark(1), Platte City (2), Republic (1), Shrewsberry (1), Springfield (3), St. Louis (1), Sunrise Beach (1), Warrensburg (1) RFM3 Apple Creek C.A.(3), Brentwood (1), Cape Girardeau (3), Chesterfield (1), Clinton (1), Sunrise Beach (1), Town and Country (1), Union Ridge C.A. (1), Warrensburg (2), Weldon Spring C.A. (1) RFM4 Apple Creek C.A.(1), Columbia (1), Fort Crowder C.A.(1), Jefferson City (1),Springfield (1) RFM5 Apple Creek C.A. (2), Battlefield (1), Bennit C.A.(4), Brentwood (1), Brickyard Hill C.A.(5), Cape Girardeau(5), Columbia (3), Compton Hollow C.A. (1), Fort Crowder C.A.(5), Holden (1), Jefferson City 6 Joplin (2), Kirksville (1), Laurie (1), Liberty (1), Mark Twain N. F. (2), Maryville 6 Nixa (1), O'Fallon (3), Platte City (4), Salem(5), Senath (1), Springfield 9 St. Louis (1), Union Ridge C.A. (1), Warrensburg (2), Weldon Spring C.A. (2), Westphalia (1) RFM6 Columbia (2), Compton Hollow C.A. (2), Jefferson City (1), Johnson (1), Joplin (2), Kansas City (2), Mark Twain N. F.(1), O'Fallon (2), Republic (1), Salem (3), Springfield (4), Warrensburg (1), Weldon Spring C.A.(1) RFM7 Cape Girardeau (1), Florissant (1) RFM8 Cardwell (1) RFM9 Columbia (2) RFM10 Cape Girardeau (1) RFM11 Cape Girardeau (1), Springfield (1) RFM12 Cape Girardeau (2) RFM13 Cape Girardeau (1), Salem (1) RFM14 Jefferson City (1), Joplin (1), Weldon Spring C.A. (1) RFM15 Jefferson City (1), Kirksville (1), Salem (2) RFM16 Columbia (1), Kirksville (3) RFM17 Apple Creek C.A. (1), Bennit C.A. (1), Compton Hollow C.A. (2), Fort Crowder C.A. (2), Mark Twain N. F. (2), Maryville (1) RFM18 Brickyard Hill C.A. (1) RFM19 Brickyard Hill C.A.(1), Columbia (2), Gladstone(1), Kansas City (2), Maryville (1), Platte City (2) RFM20 Brickyard Hill C.A.(1), Mark Twain N. F.(1), Platte City(1) RFM21 Springfield (1)

51 52 Table 12. Location and number of samples of Reticulitermes hageni, R. tibialis and R. virginicus haplotypes collected within Missouri.

Haplotype Location (number of samples) code RHM1 Mark Twain N. F.(1), Springfield (1), St. Louis (1), Hollow Creek (1) RHM2 Weldon Spring C.A. (12), Union Ridge C.A.(21), Platte Falls C.A. (2), Mark Twain N. F. (21), Compton Hollow C.A. (5), Fort Crowder C.A.(8) Columbia (2), Brickyard Hill C.A.(5), Bennit C.A. (11), Apple Creek C.A.(7) Jefferson City (1), Springfield (2), Platte City(5), Cape Girardeau (1) RHM3 Mark Twain N. F. (1), Compton Hollow C.A. (2), Fort Crowder (2), Rudolf Bennit (1), Apple Creek (2), Mark Twain N. F. (1) RTM1 Eureka (1) RTM2 Springfield (1) RTM3 Springfield (1) RTM4 Springfield (1) RTM5 Springfield (2) RVM1 Kimberling (1), St. Charles (1) RVM2 Cape Girardeau (2), Columbia (1) RVM3 Rogersville (1) RVM4 Brickyard Hill C.A.(1)

53

Table 13. Genetic variability parameters of Reticulitermes flavipes from Missouri, obtained following alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus.

Sampling source Parameter Home Urban Woodlands Total infestations Number of samples 84 94 56 234 Number of haplotypes 13 20 10 21 Number of variable sites (S) 10 14 19 25

54 Number of unique haplotypes 2 7 0 9 Haplotype (gene) diversity (Hd) 0.837 0.842 0.801 0.837 Nucleotide diversity (per site)(Pi) 0.004 0.005 0.005 0.006 Neutrality Test: Tajima's D (P-value*) -0.564 -1.111 -1.622 -1.43 ns ns ns ns * P-value: Statistical significance: Not significant, P > 0.10

Table 14. Genetic variability of Reticulitermes hageni for each habitat obtained by alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus

Sampling source Parameter Home Urban Woodlands Total infestations Number of samples 2 11 103 116

Number of haplotypes 2 2 3 3 Number of variable sites(S) 2235

55 Number of unique haplotypes 0 0 1 1 Haplotype (gene) diversity (Hd) 1 0.1818 0.217 0.225 Nucleotide diversity (per site)(Pi) 0.005 0.001 0.001 0.001 Neutrality Test: Tajima's D (P-value*) -1.43 -1.478 -1.368 ns* ns* ns* ns* * P-value: Statistical significance: Not significant, P > 0.10

Table 15. Genetic variability of Reticulitermes virginicus for each habitat sampled obtained by alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus.

Sampling source Parameter Home Urban Woodlands Total infestations Number of samples 3 3 1 7 Number of haplotypes 2 1 1 4 Number of variable sites (S) 44 Total number of mutations (Eta) 44 Number of unique haplotypes 1 0 1 2 56 Haplotype (gene) diversity (Hd) 0.667 0.81 Nucleotide diversity (per site)(Pi) 0.007 0.004 Neutrality Test: Tajima's D(P-value*) -0.04 ns* * P-value: Statistical significance: Not significant, P > 0.10

Table 16. Genetic variability of Reticulitermes tibialis for each habitat sampled obtained by alignment of the 16S mtDNA sequences from each sample with the outgroup species Coptotermes formosanus.

Sampling source Parameter Home Urban Woodlands Total infestations Number of samples 3 3 1 7 Number of haplotypes 2 1 1 4 Number of variable sites (S) 4004 Total number of mutations (Eta) 4004

57 Number of unique haplotypes 1 0 1 2 Haplotype (gene) diversity (Hd) 0.667 0.81 Nucleotide diversity (per site)(Pi) 0.007 0.004 Neutrality Test: Tajima's D(P-value*) -0.04 ns* * P-value: Statistical significance: Not significant, P > 0.10

Figure 1. General scheme of location and distribution of transects and individual colonies sampled in woodlands (showing Mark Twain N.F.). Current scales of the aerial images are not indicated.

15 m 100 m

58 Figure 2. Location of samples collected around homes (showing Salem, Mo).

58 59 Figure 3. Approximate location of the collection points of Reticulitermes within Missouri.

Woodlands Urban centers Home infestations

60

Figure 4. Relative abundance of Reticulitermes species found in woodlands, urban environments and home infestations within the state. Total samples: home infestations (188); urban environments (163); home infestations (188).

R. t ibialis R. R. R. t ibialis virginicus 2% virginic us 2% R. 1% R. hageni 7% R. hageni virginicus 3% 11% 5%

R. flavipes 33% R. hageni R. flavipes 88% R. flavipes 61 66% 82%

Woodlands Home infestations Urban environments

Figure 5. Relative abundance of Reticulitermes species in woodlands, and urban environments of Missouri. Wooded areas and associated cities are listed in the same order side by side with locations are geographically paired with each other.

R. flavipes R. hageni R. flavipes R. hageni R. virginicus R. virginicus R. tibialis Cape Girardeau Apple Creek Springfield Hollow Creek Joplin Fort Crowder Weldon Spring O'Fallon

Mark Twain Salem 62 Platte Falls Platte City

Bennit Columbia

Union Ridge Kirksville

Bricyard Hill Maryville

0 20406080100 020406080100

% samples % samples

61

Figure 6. Relative abundance of Reticulitermes species in woodlands, and urban environments of Missouri, according to the geographic position of the sampling site within the state.

North

North North Central es s l 100

e Central p l 100 Central

le100 s p South m 90 p m 90 a sa 90 South South 80 f s 80

f 80 o 70

70 o 70 e e of sam e of e j j 60 j 60 60 a ta t 50 50 n n 50

e 40 40 c

ce 40 r 30 r 30 Percenta

e 30 Pe

20 P 20 20 10 10 10 0 0 0 63

R. flavipes R. hageni R. tibialis R. virginicus R. flavipes R. hageni R. tibialis R. virginicus R. flavipes R. hageni R. tibialis R. virginicus Woodlands Home infestations Urban environments

Figure 7. Frequency of Reticulitermes flavipes haplotypes in nine urban environments and nine woodlands of Missouri.

Springfield Rfm5 Rfm1 Rfm2 Salem Platte City O'Fallon Rfm3 Rfm10 Rfm11 Maryville Kirksiville Joplin Rfm12 Rfm13 Rfm14 64 Columbia Cape Girardeau Weldon Spring C.A. Rfm15 Rfm16 Rfm17 Union Ridge C.A. Rudolf Bennit C.A. Platte Falls C.A. Mark Twain N. F. Rfm18 Rfm19 Rfm20 Fort Crowder C.A. Compton Hollow C.A. Brickyard Hill C.A. Rfm4 Rfm6 Rfm7 Apple Creek C.A.

02468101214 Rfm9 number of samples

Figure 8. Approximate geographical distribution of Reticulitermes flavipes within the counties where samples were collected from woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequences are available. Insert: previous published Reticulitermes flavipes distribution (Nutting, 1990). Numbers correspond to the haplotype code given for the species in this study.

17,18,2 1,3 15,16,5

1,5,17,18,19

1,2,5 1,2,4,5,6,9,16,18

1,18,19,2,20,5 1,3,5,6,14 1

2 2,3,5,6 1,2,3,5,7, 1,18,2,5,6 1,2,4,5,6, 14,15 3 5 5 5

1,5,6,13,15, 19 21 10,13, 2,3,5,7 1,2,5,6,14, 1,3,4,5 1,2,4,5 20

2,5 8,5

2,4,5,6,11 1,2,5,6

65 Figure 9. Approximate geographical distribution of Reticulitermes hageni within the counties where samples were collected from woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequences are available. Insert: previous register regarding R. hageni distribution (Nutting, 1990). The numbers correspond to the haplotype code given for the species.

2 2

2 2 2 1,2

1,2 2 2 1,2 2 2

65 66

Figure 10. Approximate geographical distribution of Reticulitermes virginicus within the counties where samples were collected on woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequences are available. Insert: previous register regarding R. virginicus distribution (Nutting, 1990). The numbers correspond to the haplotype code given for the species.

4

4

1 1 1 1

2 1,3 2 1,3

66 67 Figure 11. Approximate geographical distribution of Reticulitermes tibialis within the counties where samples were collected on woodlands, urban environments or cities. Light shadow: counties from where molecular sequences and haplotype diversity is available. Dark shadow: counties from which the species is known to occur but no molecular sequenceialiss are available. Insert: previous register regarding R. ti distribution (Nutting, 1990). The numbers correspond to the haplotype code given for the species.

1

2,3 4,5

67 68 Figure 12. Neighbor –joining tree depicting phylogenetic relationships of Reticulitermes from Missouri, inferred from 16srDNA. Condensed>50%).

R. flavipes RFM11

R. flavipes RFM12

R. flavipes RFM8

R. flavipes RFM2

R. flavipes RFM5

R. flavipes RFM13

R. flavipes RFM15

R. flavipes RFM14

R. flavipes RFM16

50 55 R. flavipes RFM20 R. flavipes RFM3

R. flavipes RFM19

R. flavipes RFM21

R. flavipes RFM4

R. flavipes RFM6 88 R. flavipes RFM18

R. flavipes RFM17

R. flavipes RFM1

59 R. flavipes RFM10

R. flavipes RFM9

R. flavipes RFM7

R. tibialis RTM5

R. tibialis RTM3 95 R. tibialis RTM4 50 54 R. tibialis RTM1 65 R. tibialis RTM2

66 R. hageni RHM1 92 R. hageniRHM2

R. hageni RHM3

R. virginicus RVM3

R. virginicus RVM2 92 80 R. virginicus RVM1 76 R. virginicus RVM4 Coptotermes formosanus

69

Figure 13. Networks assembled using TCS algorithm in TCS 1.21 based on the number of mutational steps between haplotypes. a: Reticulitermes flavipes. The size of the oval is not proportional to the frequency of a haplotype. Rectangular boxes indicate which haplotype has the highest outgroup probability for its network. Based on its frequency and number of mutational connections, this haplotype is likely to be most closely related to the ancestral haplotype. The numbers of mutational steps between haplotypes are indicated next to branches when the values are > 1.

70

Figure 14. Networks assembled using TCS algorithm in TCS 1.21 based on the number of mutational steps between haplotypes. a: Reticulitermes hageni, b: Reticulitermes virginicus, c: Reticulitermes tibialis. The size of the oval is not proportional to the frequency of a haplotype. Rectangular boxes indicate which haplotype has the highest outgroup probability for its network. Based on its frequency and number of mutational connections, this haplotype is likely to be most closely related to the ancestral haplotype. The numbers of mutational steps between haplotypes are indicated next to branches when the values are > 1.

71

a. .b c.

Chapter 2

Morphometric variation in the soldier caste of

Reticulitermes populations from Missouri

Abstract

Subterranean termites of the genus Reticulitermes are native to North

American forests, where they feed on cellulosic materials. Some of the six

Reticulitermes species recorded in North America have become pests in urban environments by feeding upon man–made wooden structures. Because of its climatic regime and geographic position within the continent Missouri is at moderate risk for the occurrence of subterranean termites in urban environments when compared with the southeast part of the country.

No previous studies have addressed the morphological characterization of the species in the genus Reticulitermes that occur within Missouri.

Moreover, no previous studies have measured or compared populations occurring in home infestations, urban environments and woodlands.

To morphologically characterize soldier populations of Reticulitermes inhabiting Missouri, approximately 600 colonies were sampled from urban environments, home infestations and woodlands. Nine morphological features were measured from the soldiers collected. These measurements were used to characterize individual species as well as to perform interspecific comparisons using univariate statistical analyses. In addition, multivariate

72 techniques were used to try to enhance the probability of effectively separate among Reticulitermes species soldiers from Missouri.

Although no single morphometric character studied fully separates among all four Reticulitermes species from Missouri, multivariate techniques improved the separation of R. tibialis from each of the other three species.

This study provides morphometric information from R. flavipes, R. hageni.

R. virginicus and R. tibialis found in three different habitats of Missouri. This information will improve the reliability of species identification of samples from this state. Particularly, for the species R. hageni, this study provides novel information regarding morphometric characters of its populations in forested environments.

Our study suggests that R. flavipes populations occurring in urban environments are larger when compared with populations from woodlands.

In addition, differences in size observed among soldiers of R. flavipes may be related to genetic differences at the haplotype level.

72 73

Introduction

Subterranean termites of the genus Reticulitermes are ecologically important in North American forests, where they play a major role in wood degradation (Holt and Lepage 2000). Some species are serious pests of man-made wood structures (Su and Scheffrah 1990, Lewis 1997, FAO-UNEP

2000) and therefore, have become the focus of research to improve current control strategies.

Until recently, six species of subterranean termites in the genus

Reticulitermes were known to occur in the United States (Austin 2007), of which at least three sympatric species are known to occur in the urban environments of Missouri (Nutting 1990, Messenger 2003a).

Reticulitermes species have very similar morphological characters despite differences in behavior, genetic structure, colony structure, and response to chemical treatments.(Uva et al., Bagneres et al. 1990, Nelson et al. 2001, Page et al. 2002, Haverty et al. 2005, Vargo et al. 2006) Effective differentiation between species within this genus is needed in order to support the development of species specific treatments leading to improved methods of control.

Morphological identification of sympatric species of Reticulitermes is

difficult and depends greatly on the castes that are available. The alate stage

provides the most distinctive morphological characters to separate species;

however, this stage is only seasonal, is difficult to find even during the

73 74 season when it is present and occurs only in mature colonies. Soldiers can also be useful for distinguishing among species of Reticulitermes. The soldier caste is found in the colony throughout the year and displays some interspecific variation in morphological characters. However, they also display some intraspecific and interspecific overlapping that can obscure differentiation (Scheffrahn and Su 1994). Until recent molecular advances in termite taxonomy, soldier morphology remained as the traditional and least expensive method to separate species of Reticulitermes since worker homogeneity in morphology from species to species makes them useless for morphological species differentiation.

Available diagnostic keys to separate among R. flavipes, R. virginicus, and R. hageni collected from home infestations in the southeastern United

States are mainly based on pronotal and gular metrics along with labral morphology (Snyder 1954, Scheffrahn and Su 1994, Hostettler et al. 1995,

Messenger 2002). Morphometric characterizations of Reticulitermes from other geographic regions of the Unites States such as Missouri are scarce.

However, it is important to explore regional variations (Brown, 2004) to support molecular studies (Heintschel 2006, Vargo 2003, Whitney - King,

2007).

In addition to morphology and morphometric diagnostic characters, chemical taxonomy (Clement, 1988; Haverty, 1996; Haverty, 1997; Jenkins,

2000, Bagneres, 1990; Clement, 1998) and molecular characters have been utilized to differentiate among Reticulitermes species. DNA sequencing to separate species and study intraspecific variability has been done using both

74 75 the nuclear and mitochondrial genome (Austin 2004, Austin 2004, Austin

2004, Austin 2005, Austin 2005, Szalanski 2006, Szalanski 2003, Szalanski

2003).

Several factors may influence phenotypic expression by Reticulitermes within their range of distribution in the United States. Geographical location, food quality, and genotype (Hostettler et al. 1995) have all been shown to influence morphology. This phenotypic plasticity may obscure relationships between phenotypes and genotypes (Freeland 2005). Recent studies are analyzing the relationships between phenotypic polymorphisms like cuticular hydrocarbons or size variations with the underlying molecular structure

(Jenkins et al. 2000, Whitney - King et al. 2007).

The primary objective of this study was to morphometrically characterize and compare populations of Reticulitermes species occurring in urban environments, home infestations, and woodlands of Missouri.

75 76

Methods

Three hundred and ninety three soldiers gathered from nine urban areas, nine woodland areas, and from home infestations during 2004 and 2005, were used to morphologically characterize the four species of Reticulitermes found in Missouri: R. flavipes, R. virginicus, R. hageni, and R. tibialis.

A total of nine characters were measured for every specimen (Figure

1). Seven of the nine characters were measured from soldiers kept in 90% alcohol using an Olympus SZ-ST dissecting scope with an ocular micrometer at 40X. The characters illustrated in Figure 1 mostly followed those identified by Roonwal (1969):

1. Length of head capsule: Head length (HL)

2. Length of head including mandibles: Total head length (THL)

3. Maximum width of head: Head width (HW)

4. Maximum width of Pronotum: Pronotum width (PW)

5. Median length of Pronotum: Pronotum Length (PL)

6. Maximum Width of gula: (Max. G)

7. Minimum Width of gula: (Min. G)

Two remaining characters: Labrum width (LW) and labrum length (LL) were measured at 200X using an Olympus BH-2 microscope. Labra were previously dissected from the soldiers’s heads and mounted for microscopic observation and measurement on glass slides using euparal®.

76 77 One-way analysis of variance (ANOVA) was implemented using the software SPSS to compare individual measurements among the four

Reticulitermes species. Only soldiers from whom molecular identification was available were included in this comparison to ensure that the measurements were associated with the correct species.

Discriminant function analysis was implemented using actual measurements and log-transformed values to simultaneously minimize within group differences and maximize among species differences by generating a linear combination of characters to separate species groups. The classification and cross-tabulation proportions of individuals correctly classified were used as indicators of the ability of the complete set of characters to successfully differentiate a species or to assign a species name/ grouping to an unidentified individual. The cross-tabulation test uses characters from all samples to classify individual species groups, excluding their own characters from the calculation.

Wilks’ Lambda was used to test significant differences among means of

morphometric characters used in the multivariate analysis. Levene’s statistic

tested the hypothesis of homogeneity of variances. Kaiser-Meyer-Olkin

(KMO) was used to establish the factorability of the morphometric sets using

multivariate techniques and Bartlett’s tested the sphericity to confirm that

the degree of common variance among variables was appropriate for

conducting a factor analysis and accounts for substantial amounts of

variance.

77 78 Since R. flavipes was abundant in both urban and woodland environments, one-way ANOVA was used to test the hypothesis that significant differences exist among R. flavipes populations collected from

urban and woodland environments. Post-hoc comparisons were performed

using Bonferroni or Dunnet’s correction. Once the haplotype identity of R.

flavipes was established by molecular sequencing a portion of the 16s rRNA

mithochondrial gene, one-way ANOVA was conducted to test the hypothesis

that the more abundant haplotypes differ in size based on the set of nine

morphometrics.

78 79

Results

The complete study included measurements of 231 soldiers of R. flavipes,

140 of R. hageni, 14 of R. virginicus and 8 of R. tibialis (Table 1). However,

since only 165 samples of R. flavipes, 89 of R. hageni, 15 of R. virginicus and

8 of R. tibialis were sequenced using mtDNA 16s rRNA, only metrics

associated with these 165 samples where used to compare species, illustrate

the dispersion of data for each character measured (figures 2 a-i), and to

perform DFA analysis. No significant differences were observed among the

means or ranges of the characters that included all the samples from the

table that include only the soldiers having molecular identification, supporting

the reliability of species identification using soldier morphology

Soldiers of R. flavipes, R. virginicus, R. hageni and R. tibialis found in

Missouri differed in the mean size of at least four of the nine characters we measured (Table 2, 3). There is overlap among species in the range of individual characters that makes it difficult to establish a purely unique range for each species. For example, the size of morphological characters of some of the smaller individuals of R. flavipes may overlap with the largest individuals of R. hageni. However, in at least ninety five percent of comparisons the complete set of nine metrics was able to differentiate among

R. flavipes and R. hageni.

Mean pronotum width (PW) and mean maximum gular width (Max. G) differentiated R. flavipes from the other three species (P<0.05). Mean total

79 80 head length (THL) differentiated R. hageni from the other three species

(Tables 3, 4). As summarized in Table 6 at least two characters differentiated (P<0.005) between two of the four species. Reticulitermes virginicus and R. tibialis were intermediate in size between R. flavipes and R.

hageni so fewer characters are useful for separating them.

The evaluation of the complete set of morphometrics using a stepwise discriminant function analysis of the complete set of soldier data (DFA) to maximize the ability of the nine metrics to separate among the four

Reticulitermes species resulted in exclusion of pronotum width and maximum gula as predictors of species groups. Three functions assigned the biggest value of correlation to minimum width of the gula, head width, total head length, and pronotum length. A second function assigned the largest correlation values to head length, and a third function included both labrum length and width (Table 5).

A plot of the first three DFA canonical axes shows the species groups resulting from the analysis (Figures 3,4). In the cross validation test this discriminant analysis successfully assigned 84.1% all samples to the correct species, with 100% of R. tibialis, 88.97% of R. flavipes, and 84.15% of R. hageni samples being assigned correctly (Table 6).

Reticulitermes flavipes was the only species found abundantly in both

urban and woodlands of Missouri. Due to its widespread distribution and

haplotype diversity, R. flavipes was used to compare size differences among populations occupying different habitats (Table 7). Means of the nine metrics

80 81 of R. flavipes soldiers from woodlands tend to be smaller than those collected from home infestations and urban environments (P<0.005) (Table 7, 8).

In addition, R. flavipes exhibited phenotypic differences in size among the more abundant molecular mtDNA haplotypes. Means of the nine morphometrics from individuals sharing the molecular haplotype RFM1 tend to be smaller than those sharing the molecular haplotype RFM5 (P<0.05)

(Table 9, 10, Figure 5).

81 82

Discussion

This study describes the variation in nine morphological characters and

their usefulness for identifying soldiers of the sympatric species

Reticulitermes flavipes, R. virginicus, R. hageni and R. tibialis found in

Missouri whose molecular haplotype was confirmed by sequencing of a

portion of 16S rRNA. This study also provides data examining factors such

as habitat and genotype that are related to observed variations in the size of

certain morphological features of R. flavipes.

Intraspecific variability in the characters measured on the head and pronotum of Reticulitermes soldiers in this study was large and lead to interspecific overlapping among species. This interspecific overlap has been observed in previous studies of Reticulitermes from the southeastern U.S.

(Hostettler et al. 1995, Whitney - King et al. 2007). No single morphometric character measured on the soldier caste in this study provides absolute certainty to separate all four Reticulitermes species found in Missouri. The morphometrics tested in this study best differentiate between R. flavipes and

R. hageni (P<0.000). Four other morphometric characters can differentiate among at least two of the other species (P<0.000), however interspecific overlap reduces the certainty of the identification in the absence of the alate caste. Moreover, not enough specimens of R. virginicus and R. tibialis were collected in this study to provide an accurate representation of the variability

82 83 within these species, thus limiting the use of morphometric characters as diagnostics for these species.

The use of a DFA analysis has been previously suggested as a useful method to improve the ability of correctly separating insect species and populations (Sites and Willig 1994a, Petrarca et al. 1998a, Sites and Willig

2000a, Cazorla and Acosta 2003a, Radloff et al. 2005a) and to improve separation of Reticulitermes species (Brown et al. 2004). In our study, only

R. tibialis is completely separable from the other three species occurring in

Missouri with a 100% of certainty using DFA.

Geographic location, habitat and genotype are the main factors influencing phenotypic expressions of organisms, including termites (Freeland

2005). Differences related with geographic location can be observed in the morphometrics of Reticulitermes species found in Missouri when compared with Reticulitermes from Florida, Texas, Oklahoma and the Delmarva

Peninsula (Scheffrahn and Su 1994, Hostettler et al. 1995, Jenkins et al.

2000, Brown et al. 2004, Heintschel et al. 2006, Whitney- King et al. 2007).

Measurements of soldier characters from R. flavipes home infestations in

Missouri fit very close within the range of means (plus or less one standard deviation) (Table 11) of the same characters measured from populations in

Oklahoma. Comparisons of the means using an unpaired t-test show that populations from Missouri are slightly (5-10%) bigger (Brown et al. 2005) than those in Oklahoma. Similarly, the range of (plus or less one standard deviation) mean labral width of R. flavipes soldiers occurring in home infestations from Missouri are smaller than soldiers from Texas (Heintschel et

83 84 al. 2006). No comparison was available for labrum length since we used a different reference point for measurement of this character than that used in other available studies. We found R. flavipes having pronotum width below the range found in Florida (0.90 mm) by Scheffrahn and Su (1994). The range of our data for pronotum width also is below the other studies from

Florida (Hostettler et al. 1995). Missouri soldier morphometrics fall within

the range observed on soldiers from the Delmarva peninsula (Delaware,

Maryland, and Virginia USA)(Whitney- King et al. 2007).

Developmental plasticity observed in Reticulitermes also may account

for the intraspecific size variability observed in soldier morphology. Given

that soldiers may originate from different sizes and instars of workers

(usually W5-W6) or nymphs (N3-N6)(Roisin 2000, Laine and Wright 2003), it

is expected that their final size would be partially dependent on the pathway

from which the soldier originated. In addition, sexual dimorphism expressed

as differences in the size of soldiers may also be present as has been

reported for groups of workers in R. flavipes (Zimet and Stuart 1982). In

urban environments for example, chemical treatments using juvenile

hormone analogs to control subterranean termite infestations, aim to

increase the number of soldiers by inducing the incidence of nymphs molting

into soldiers (Hrdy et al. 2001, 2004, Scharf et al. 2005). Our study provides

some evidence supporting the influence of habitat on phenotypic character

size among populations from woodland and urban environments. We would

hypothesize that the length of the foraging season, which is related to

available soil moisture and suitable soil temperature (Kofoid 1934, Collins

84 85 1991, Houseman et al. 2001, Hu and Appel 2004), is more limiting in woodlands than urban environments or home infestations (Myles and Grace

1991, Davis and Kamble 1994, Grace 1996).

Phenotypic plasticity may obscure the genotypic variation in traits observed in some characters such as growth rate, gender, coloration, and feeding-related morphology (Freeland, 2005). In Reticulitermes there is evidence of genotype underlying some of the phenotypic variation such us cuticular hydrocarbons (Haverty et al. 1996, Haverty and Nelson 1997,

Jenkins et al. 2000). In our study, differences in phenotypic characters related to the size of soldiers were observed between two of the common haplotypes of R. flavipes found in Missouri (RFM1 and RFM 5) that correspond to the more common haplotypes found in neighboring states. Although preliminary, our data show some support for a genotypic basis for some of the phenotypic variation.

Few studies describe morphometric differences of Reticulitermes populations in natural environments where the resident genotypes and phenotypes can be found. In particular, little is known about the species R. hageni since its populations are scarce in urban environments.

Morphometrics from R. hageni in our study represent populations from woodlands (81%). When comparing R. hageni populations from Missouri woodlands with samples from urban environments in Oklahoma (Brown et al.

2005), the size of Missouri R. hageni exceeded Oklahoma’s mean values for eight of the nine metrics we measured by at least 12%. When compared using an unpaired t-test, the range of the means were greater than one

85 86 standard deviation. Similarly, the width of the pronotum exceeded the mean plus one standard deviation when compared with Florida populations

(Hostettler et al. 1995).

The fact that Missouri R. hageni populations appear to be larger than

other regions of the U.S. may limit the certainty of species determinations of

Missouri specimens based on keys that were developed based on R. hageni in

Florida (Scheffrahn and Su 1994, Hostettler et al. 1995). Moreover, the

similarity of morphometrics in our study between R. tibialis and R. hageni,

raises the possibility that the broad distribution of R. tibialis previously cited

in Missouri but not confirmed in our surveys may have been the result of

confusion with R. hageni (Nutting 1990).

Reticulitermes tibialis and R. virginicus were uncommon in both urban environments and woodlands, so there may not be enough morphometric data from our study to draw conclusions regarding phenotypic size ranges of

Missouri populations. Seven morphometric characters of R. virginicus are at least 5% smaller in Oklahoma than Missouri. We should note that identification of R. virginicus using soldier morphology always agreed with identification based on molecular characters. This is likely due to the distinctive shape of the labrum of R. virginicus when compared to other

Reticulitermes. The unique shape of these characters may not have been detected by simple length-width measurements made in our study.

86 87 Table 1. Means and standard deviation of morphometric characters associated with the head, pronotum, labrum and gula of Reticulitermes soldiers collected in urban and woodlands of Missouri.

Std. Character Species N Mean (mm) Minimum Maximum Deviation

R. flavipes 231 1,169 0.088 0.900 1,400 R. hageni 140 1,018 0.077 0.900 1,250 R. virginicus 14 1,114 0.064 0.975 1,225 Head width R. tibialis 8 1,067 0.040 1,025 1,125 R. flavipes 231 2,656 0.186 2,200 3,200 R. hageni 140 2,375 0.148 2,050 2,875 Total head R. virginicus 14 2,647 0.133 2,325 2,875 lenght R. tibialis 8 2,506 0.095 2,420 2,650 R. flavipes 231 1,670 0.151 1,350 2,075 R. hageni 140 1,516 0.097 1,270 1,825 R. virginicus 14 1,804 0.118 1,550 2,000 Head lenght R. tibialis 8 1,643 0.089 1,500 1,800 R. flavipes 231 0.344 0.024 0.287 0.424 R. hageni 140 0.329 0.021 0.260 0.386 R. virginicus 14 0.360 0.022 0.323 0.399 Labrum width R. tibialis 8 0.305 0.016 0.286 0.332 R. flavipes 231 0.477 0.031 0.378 0.546 R. hageni 140 0.432 0.028 0.370 0.529 R. virginicus 14 0.461 0.024 0.428 0.504 Labrum lenght R. tibialis 8 0.388 0.021 0.353 0.412

R. flavipes 231 0.890 0.072 0.675 1,075 R. hageni 140 0.790 0.064 0.675 1,025 R. virginicus 14 0.833 0.074 0.750 0.975 Pronotum width R. tibialis 8 0.789 0.019 0.771 0.825

R. flavipes 231 0.478 0.036 0.350 0.600 R. hageni 140 0.445 0.032 0.375 0.525 R. virginicus 14 0.438 0.030 0.400 0.500 Pronotum lenght R. tibialis 8 0.459 0.013 0.450 0.475

R. flavipes 231 0.235 0.026 0.173 0.300 R. hageni 140 0.190 0.026 0.150 0.275 R. virginicus 14 0.179 0.019 0.150 0.225 Minimum gula R. tibialis 8 0.228 0.013 0.213 0.250

R. flavipes 231 0.505 0.033 0.400 0.600 R. hageni 140 0.439 0.032 0.375 0.525 R. virginicus 14 0.459 0.025 0.425 0.525 Maximum gula R. tibialis 8 0.463 0.030 0.425 0.525

87 88 Table 2. Results of ANOVA comparing the sizes of nine individual metric characters among four Reticulitermes species found in Missouri. (P value≤ 0.005 indicate mean difference is significant at the 0.05 level) Mean Source df F P Square Head width Between Groups 3 0.434 55.618 0.000 Within Groups 273 0.008 Total 276 Total head lenght Between Groups 3 1.423 42.100 0.000 Within Groups 273 0.034 Total 276 Head lenght Between Groups 3 0.656 31.359 0.000 Within Groups 273 0.021 Total 276 Labrum width Between Groups 3 0.008 15.386 0.000 Within Groups 273 0.001 Total 276 Labrum lenght Between Groups 3 0.045 45.735 0.000 Within Groups 273 0.001 Total 276 Pronotum width Between Groups 3 0.212 39.766 0.000 Within Groups 273 0.005 Total 276 Pronotum lenght Between Groups 3 0.026 18.417 0.000 Within Groups 273 0.001 Total 276 Minimum gula Between Groups 3 0.052 77.273 0.000 Within Groups 273 0.001 Total 276 Maximum gula Between Groups 3 0.090 78.591 0.000 Within Groups 273 0.001 Total 276

88 89 Table 3. Means and standard deviation for soldier morphometric characters of Reticulitermes collected in urban environments, home infestations, and woodlands of Missouri whose identity was confirmed using 16S mtDNA gene sequencing.

Std. Observed values Character Species N Mean* Std. Error Deviation Min. Max. R. flavipes 165 1.168 a 0.097 0.008 0.913 1.400 R. hageni 89 1.020 b 0.076 0.008 0.900 1.250 Head width R. virginicus 14 1.125 a,c 0.061 0.016 0.975 1.225 R. tibialis 8 1.067 b,c,d 0.040 0.014 1.025 1.125 R. flavipes 160 2.656 a 0.207 0.016 2.200 3.200 Total head R. hageni 89 2.385 b 0.149 0.016 2.050 2.775 length R. virginicus 14 2.622 a 0.130 0.034 2.325 2.875 R. tibialis 8 2.506 a 0.095 0.034 2.420 2.650 R. flavipes 164 1.679 a 0.168 0.013 1.275 2.075 R. hageni 88 1.514 b,d 0.099 0.011 1.300 1.825 Head length R. virginicus 14 1.800 c 0.114 0.031 1.550 2.000 R. tibialis 8 1.643 a,b,c 0.089 0.031 1.500 1.800 R. flavipes 165 0.343 a 0.025 0.002 0.287 0.424 R. hageni 89 0.330 b 0.020 0.002 0.277 0.378 Labrum width R. virginicus 14 0.358 a 0.027 0.007 0.298 0.399 R. tibialis 8 0.305 c 0.016 0.006 0.286 0.332 R. flavipes 164 0.474 a 0.034 0.003 0.378 0.546 R. hageni 90 0.433 b 0.028 0.003 0.370 0.508 Labrum length R. virginicus 14 0.455 a 0.026 0.007 0.399 0.504 R. tibialis 8 0.388 c 0.021 0.008 0.353 0.412 R. flavipes 167 0.888 a 0.080 0.006 0.675 1.075 Pronotum R. hageni 90 0.788 b 0.064 0.007 0.675 1.025 width R. virginicus 14 0.819 b 0.062 0.016 0.750 0.950 R. tibialis 8 0.789 b 0.019 0.007 0.771 0.825 R. flavipes 167 0.479 a 0.041 0.003 0.350 0.600 Pronotum R. hageni 90 0.447 b 0.032 0.003 0.375 0.525 length R. virginicus 14 0.433 b,c 0.026 0.007 0.400 0.488 R. tibialis 8 0.459 b,d 0.013 0.005 0.450 0.475 R. flavipes 162 0.234 a 0.027 0.002 0.173 0.300 R. hageni 86 0.186 b 0.026 0.003 0.150 0.275 Minimum gul a R. virginicus 14 0.175 b,c 0.014 0.004 0.150 0.200 R. tibialis 8 0.228 a 0.013 0.005 0.213 0.250 R. flavipes 161 0.504 a 0.036 0.003 0.400 0.600 R. hageni 86 0.436 b 0.031 0.003 0.375 0.525 Maximum gula R. virginicus 14 0.450 b 0.021 0.005 0.400 0.475 R. tibialis 8 0.463 b 0.030 0.011 0.425 0.525

* Means having the same letter are not significantly different at the 0.5 level

90

Table 4. Summary of statistical differences among the means of nine soldier morphometric characters from four Reticulitermes species. HW: head width, HL: Head length, THL: Total length of head, LW: Labrum width, PW: Pronotum width, PL: Pronotum length, Min. G.: Minimum width of gula, Max. G. Maximum width of gula. Bold characters indicate mean differences at P<0.000 and no bold characters indicate P<0.005. Numbers in parenthesis indicate the number of characters that differentiate between two species.

Species R. hageni R. virginicus R. tibialis

R. flavipes HW,HL,THL, LW, PL, PW, Min. G., HW,LW, LL, PW, LL, PW, PL, Min. G., Max. G., HL (5) Max. G., THL (6) Max. G. (9) 91

R. hageni HW,HL,THL,LW, THL, HL, Min. LL (5) G.,LL,LW (5)

R. virginicus LW, LL, Min. G., HL(4)

Table 5. Structure matrix containing the pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions of three functions obtained after conducting stepwise discriminant function analysis of nine morphometric characters of four Reticulitermes species found in Missouri. Variables ordered by absolute size of correlation within function.

Function 123 Minimum gula 0.768 (*) -0.039 -0.233 Head width 0.765 (*) 0.355 0.193 Total head length 0.659 (*) 0.441 0.154 92 Maximum gula (a) 0.646 (*) 0.182 0.032 Pronotum width (a) 0.567 (*) 0.244 0.128 Pronotum length 0.412 (*) -0.229 (*) -0.017 Head length 0.406 0.709 0.141 Labrum length 0.627 0.107 0.676 (*) Labrum width 0.239 0.263 0.516 (*)

* Largest absolute correlation between each variable and any discriminant function a This variable not used in the analysis.

Table 6. Number of samples analyzed and percentage of correct species identifications after performing stepwise discriminant function analysis of nine metric characters from Reticulitermes soldiers in Missouri.

Species / % correct R. R. R. R. Total identification flavipes hageni virginicus tibialis R. flavipes 129 14 0 2 145 89.0 9.7 0.0 1.4 100 R. hageni 8694 182 9.8 84.2 4.9 1.2 100 R. virginicus 01 12013 93 0.0 7.7 92.3 0.0 100 R. tibialis 00 0 88 00 0100 100

Table 7. Results of ANOVA comparing soldier morphometric characters of Reticulitermes flavipes from urban environments (n=84), home infestations (n=38) and woodlands (n=38) of Missouri. (P value ≤ 0.005 indicate mean difference is significant at the 0.05 level).

Mean Source df FP Square

Between Groups 2 0.231 31.699 0.0000 Head width Within Groups 158 0.007 Total 160 Between Groups 2 0.957 27.269 0.0000 Total head Within Groups 158 0.035 length Total 160 Between Groups 2 0.581 25.340 0.0000 Head Within Groups 158 0.023 length Total 160 0.004 7.001 0.0012 Labrum Between Groups 2 Within Groups 158 0.001 width Total 160 0.021 18.888 0.0000 Labrum Between Groups 2 length Within Groups 158 0.001 Total 160 Between Groups 2 0.096 17.144 0.0000 Pronotum Within Groups 158 0.006 width Total 160 Between Groups 2 0.016 10.260 0.0001 Pronotum Within Groups 158 0.002 length Total 160 0.005 7.424 0.0008 Minimum Between Groups 2 Within Groups 158 0.001 gula Total 160 0.026 18.195 0.0000 Maximum Between Groups 2 gula Within Groups 158 0.001 Total 160

94 Table 8. Mean and standard deviation of nine metric characters of soldiers of Reticulitermes flavipes occurring around homes (urban) (n=84), home infestations (n=38) and woodlands (n=38) of Missouri.

Mean Std. Std. observed values Character Habitat (mm) * Deviation Error Minimum Maximum

Head width Urban 1.195 a 0.084 0.009 1.000 1.400 Homes 1.191 b 0.078 0.012 1.023 1.350 Woodlands 1.082 b 0.099 0.016 0.913 1.250 Total head length Urban 2.702 a 0.195 0.022 2.250 3.200 Homes 2.718 b 0.163 0.025 2.325 3.125 Woodlands 2.470 b 0.177 0.030 2.200 2.800 Head length Urban 1.720 a 0.170 0.018 1.275 2.075 Homes 1.717 b 0.145 0.022 1.425 2.025 Woodlands 1.535 b 0.092 0.015 1.375 1.750 Labrum width Urban 0.347 a 0.028 0.003 0.290 0.424 Homes 0.346 b 0.022 0.003 0.287 0.386 Woodlands 0.332 b 0.018 0.003 0.294 0.370 Labrum length Urban 0.485 a 0.029 0.003 0.416 0.546 Homes 0.475 b 0.025 0.004 0.420 0.521 Woodlands 0.448 b 0.039 0.006 0.378 0.504 Pronotum width Urban 0.903 a 0.076 0.008 0.700 1.075 Homes 0.907 b 0.060 0.009 0.800 1.025 Woodlands 0.833 b 0.083 0.013 0.675 0.975 Pronotum length Urban 0.482 a 0.045 0.005 0.350 0.600 Homes 0.491 b 0.027 0.004 0.425 0.550 Woodlands 0.457 b 0.036 0.006 0.350 0.525 Minimum gula Urban 0.236 a 0.027 0.003 0.173 0.300 Homes 0.242 b 0.024 0.004 0.200 0.300 Woodlands 0.223 b 0.029 0.005 0.175 0.275 Maximum gula Urban 0.509 a 0.032 0.003 0.425 0.600 Homes 0.517 b 0.026 0.004 0.450 0.575 Woodlands 0.476 b 0.044 0.007 0.400 0.550

* Habitats having different letters indicate their means are statistically different. (P<0.05, df: 2)

94 95 Table 9. Results of ANOVA comparing soldier morphometric characters in Reticulitermes flavipes among haplotypes RFM1 (n=35) and RFMF (n=60) from Missouri.

Mean Source df F Prob >F Square Between Groups 1 0.619 12.969 0.001 Total head Within Groups 95 0.048 length Total 96 Between Groups 1 0.262 7.942 0.006 Head length Within Groups 95 0.033 Total 96 Between Groups 1 0.164 16.181 0.000 Head width Within Groups 95 0.010 Total 96 Between Groups 1 0.004 6.816 0.010 Labrum width Within Groups 95 0.001 Total 96 Between Groups 1 0.019 15.460 0.000 Labrum Within Groups 95 0.001 length Total 96 Between Groups 1 0.064 9.432 0.003 Pronotum width Within Groups 95 0.007 Total 96 Between Groups 1 0.031 18.622 0.000 Pronotum Within Groups 95 0.002 length Total 96 Between Groups 1 0.007 8.785 0.004 Minimum gula Within Groups 95 0.001 Total 96 Between Groups 1 0.014 7.473 0.007 Maximum Within Groups 95 0.002 gula Total 96

95 96 Table 10. Mean and standard deviation of soldier morphometric characters from Reticulitermes flavipes haplotypes RFM1 (n=35) and RFM5 (n=60) in Missouri samples. (P value ≤ 0.005 indicate mean difference is significant at the 0.05 level)

Observed range Mean Std. Std. Character Haplotype * (mm) (mm) Deviation Error Min. Max.

RFM1 1.115 a 0.101 0.017 0.913 1.300 Head width RFM5 1.196 b 0.099 0.013 0.925 1.400 RFM1 Total head 2.530 a 0.185 0.031 2.200 2.875 length RFM5 2.689 b 0.235 0.031 2.250 3.200 RFM1 1.602 a 0.149 0.025 1.375 2.000 Head length RFM5 1.705 b 0.198 0.026 1.275 2.075 RFM1 0.333 a 0.024 0.004 0.287 0.382 Labrum width RFM5 0.345 b 0.024 0.003 0.290 0.403 RFM1 0.451 a 0.038 0.006 0.386 0.512 Labrum length RFM5 0.479 b 0.034 0.004 0.378 0.546 RFM1 0.857 a 0.078 0.013 0.700 1.025 Pronotum width RFM5 0.906 b 0.084 0.011 0.675 1.075 RFM1 0.456 a 0.028 0.005 0.400 0.500 Pronotum length RFM5 0.492 b 0.047 0.006 0.350 0.600 RFM1 0.222 a 0.027 0.005 0.173 0.275 Minimum gula RFM5 0.238 b 0.028 0.004 0.175 0.300 RFM1 0.488 a 0.043 0.007 0.400 0.575 Maximum gula RFM5 0.509 b 0.038 0.005 0.400 0.575

97

Table 11. The size of soldier Morphometric characters recorded in the literature for four Reticulitermes species found in the United States

References

Metric Hostleter et al. 1994 Scheffrah et al. 1995 Brown et al. 2005 Heintshel et al. 2006 Whitney et al. 2007 Current study Species character Mean - Mean Mean + Mean - Mean Mean + Mean - Mean Mean + Mean - Mean + Mean - Mean Mean - Mean Mean + Mean Mean SD (mm) SD SD (mm) SD SD (mm) SD SD SD SD + SD SD (mm) SD R. flavipes 0.371 0.341 0.311 0.426 0.396 0.366 0.374 0.346 0.318 Labrum width R. virginicus (mm) 0.345 0.31 0.275 0.43 0.42 0.41 0.385 0.358 0.331 R. hageni 0.303 0.285 0.267 0.35 0.33 0.31 R. flavipes Labrum 0.421 0.374 0.327 0.489 0.462 0.435 0.514 0.485 0.456 length (mm) R. virginicus 0.436 0.425 0.414 0.715 0.455 0.195 R. hageni 0.329 0.286 0.243 0.461 0.433 0.405 R. flavipes >0.9 >0.9 0.918 0.869 0.82 0.99 0.94 0.89 0.979 0.903 0.827 Pronotum R. virginicus width (mm) >0.7-0.8 >0.7 0.853 0.792 0.731 0.77 0.73 0.69 0.881 0.819 0.757 R. hageni <0.7 ≤ 0.7 0.632 0.614 0.596 0.852 0.788 0.724

98 R. flavipes 0.527 0.482 0.437 R. virginicus Pronotum 0.459 0.433 0.407 length (mm) R. hageni 0.479 0.447 0.415 R. flavipes >2.8 2.885 2.739 2.593 2.897 2.702 2.507 R. virginicus 2.855 2.59 2.325 2.752 2.622 2.492 Total Head R. hageni 2.139 2.086 2.033 2.534 2.385 2.236 length (mm) R. flavipes 1.932 1.827 1.722 1.89 1.72 1.55

R. virginicus 1.988 1.78 1.572 1.914 1.8 1.686 Head length R. hageni 1.48 1.455 1.43 1.613 1.514 1.415 (mm) R. flavipes 1.219 1.152 1.085 1.279 1.195 1.111

R. virginicus 1.143 1.067 0.991 1.186 1.125 1.064 R. hageni 0.846 0.834 0.822 1.096 1.02 0.944 Head Width R. flavipes 0.51 0.487 0.464 0.541 0.509 0.477 (mm) R. virginicus 0.473 0.446 0.419 0.471 0.45 0.429 R. hageni 0.368 0.364 0.36 0.467 0.436 0.405 R. flavipes GularWitdh 0.254 0.232 0.21 0.263 0.236 0.209 max (mm) R. virginicus 0.22 0.205 0.19 0.189 0.175 0.161 R. hageni 0.184 0.178 0.172 0.212 0.186 0.16

Gular Witdh min (mm)

Figure 1. Soldier morphometric characters measured and used to describe and make comparisons among Reticulitermes species found in Missouri. 1: Head length, 2: Total head length, 3: head width, 4: Pronotum wide, 5: Pronotum length, 6: Maximum gular width, 7: Minimum gular width, 8: Labral width, 9: Labral length

9

8

99 Figure 2. a-i. Box plots illustrating and comparing distribution and dispersion of nine metrics of the four Reticulitermes species found in Missouri. A: Head width, b: head length, c: total head length, d: Labrum length, e: Labrum width, f: Pronotum width, g: Pronotum length, h: Minimum gula, i: Maximum gula

1.400

1.300

1.200 a.

1.100 Head width width Head (mm)

1.000

0.900

R. flavipes R. hageni R. virginicus R. tibialis Species

2.2

2.0

1.8 b.

1.6 Head lenght (mm)

1.4

1.2

R. flavipes R. hageni R. virginicus R. tibialis Species

99 100 Figure 2. Continued

3.200

3.000

2.800

2.600 c.

2.400 Total head lenght (mm) Total

2.200

2.000

R. flavipes R. hageni R. virginicus R. tibialis Species

100 101 Figure 2. Continued

0.5500

0.5000

d. 0.4500

0.4000 Labrum lenght

0.3500

0.3000

R. flavipes R. hageni R. virginicus R. tibialis Species

0.4500

0.4000

e.

0.3500 Labrum width (mm) width Labrum

0.3000

0.2500

R. flavipes R. hageni R. virginicus R. tibilalis Species

101 102 Figure 2. Continued

1.100

1.000

0.900

f. 0.800 Pronotum width (mm) Pronotum width

0.700

0.600

R. flavipes R. hageni R. virginicus R. tibilais Species

0.6

0.55

0.5

g.

0.45 Pronotum lenght (mm) lenght Pronotum

0.4

0.35

R. flavipes R. hageni R. virginicus R. tibilais Species

102 103 Figure 2. Continued

0.300

0.280

0.260 h.

0.240

0.220

0.200 Minimum gula (mm)

0.180

0.160

R. flavipes R. hageni R. virginicus R. tibialis Species

0.600

0.550

0.500

i. 0.450 Maximum gula Maximum

0.400

0.350

R. flavipes R. hageni R. virginicus R. tibialis Species

104

Figure 3. Bi-dimensional plot indicating intraspecific dispersion of individuals based on soldier Morphometric characters. Four centroids each corresponding to one Reticulitermes species are also shown.

Canonical Discriminant Functions

4 Species R. flavipes R. hageni R. virginicus 3 R. tibialis Group Centroid 2 105

4

1 0 2 Function 2

-2

-4 -2 0 2 4 Function 1

Figure 4. Three-dimensional plot after performing discriminant function analysis of morphometric characters from soldiers of four Reticulitermes species. Intraspecific dispersion of individuals by species is also shown. R. flavipes (black), R. hageni (red), R. virginicus (blue), R. tibialis (green)

106

105

Figure 5. Comparison of the size of nine morphometric characters measured from soldiers of the 16S mtDNA molecular haplotypes RFM5 and RFM1 of Reticulitermes flavipes sampled in Missouri. Means of RFM1 individual characters are smaller ( df: 1, P<0.005) for all individual morphometric characters.

2.80

107 2.40 RFM5

2.00 RFM1 1.60 m m

1.20

0.80

0.40

0.00

Head width Total head Head length Labrum Labrum Pronotum Pronotum Minimum Maximum length width length width length gula gula

References cited

Austin, J., A. Szalanski, M. T. Messenger, J. McKern, and R. E. Gold. 2006. Genetic variation and Phylogenetics of Reticulitermes (Isoptera: Rhinotermitidae) from the American Great Plains. Sociobiology 48: 427-445.

Austin, J. W., A. L. Szalanski, and M. T. Messenger. 2004a. Mitochondrial DNA variation and distribution of the subterranean termite genus Reticulitermes (Isoptera: Rhinotermitidae) in Arkansas and Louisiana. Florida Entomologist 87(4): 473 - 480.

Austin, J. W., A. L. Szalanski, and B. M. Kard. 2004b. Distribution and genetic variation of Reticulitermes (Isoptera: Rhinotermitidae) in Oklahoma. Florida Entomologist 87(2): 152-158.

Austin, J. W., A. L. Szalanski, R. E. Gold, and B. T. Foster. 2004c. Genetic variation and geographical distribution of the subterranean termite genus Reticulitermes in Texas. Southwestern Entomologist 29: 1-11.

Austin, J. W., A. L. Szalanski, R. H. Scheffrahn, and M. T. Messenger. 2005. Genetic variation of Reticulitermes flavipes (Isoptera : Rhinotermitidae) in North America applying the mitochondrial rRNA 16S gene. Annals of the Entomological Society of America 98: 980- 988.

Avise, J., A. Ball, T. Lamb, J. Neigel, C. Reeb, and N. Saunders. 1987. Intraspecific phylogeography: the mithochondrial DNA bridge between

108107 population genetics and systematics. Annual Review of Ecology and Systematics 18: 489-522. Bagneres, A., J. Clement, M. Blum, R. Severson, C. Joulie, and C. Lange. 1990. Cuticular hydrocarbons and defensive compounds of Reticulitermes flavipes (Kollar) and R. santonensis (Feytaud) : polymorphism and chemotaxonomy. Journal of Chemical Ecology 16: 3213-3244.

Behr, E. A. 1973. Termite distribution in Michigan. The Great Lakes Entomologist 6: 107-119.

Bignell, D., and P. Eggleton. 2000. Termites in ecosystems, pp. 363 - 388. In T. Abe, D. Bignell and M. Higashi [eds.], Termites: Evolution, Sociality, Symbioses, Ecology. Kluwer Academic Publishers

Brown, K. B., M. Kard, and M. Doss. 2004. 2002 Oklahoma termite survey (Isoptera). Journal of the Kansas Entomological Society 77: 1- 9.

Bulmer, M. S., and J. F. A. Traniello. 2002. Lack of aggression and spatial association of colony members in Reticulitermes flavipes. Journal of Insect Behavior: 121-126.

Clement, J. L. 1986. Open and close societies in Reticulitermes termites (Isoptera, Rhinotermitidae): geographic and seasonal variations. Sociobiology 11: 311-323.

Collins, M. 1991. Physical factors affecting termite distribution. Sociobiology 19: 283-285.

Crandall, K., and A. Templeton. 1993. Empirical test of some predictions from coalescent theory with applications to intraspecific phylogeny reconstruction. Genetics: 959-969.

108 109 Delphia, C., K. Copren, and M. Haverty. 2003. Agonistic behavior between individual worker termites from three cuticular hydrocarbon phenotypes of Reticulitermes (Isoptera: Rhinotermitidae) from northern California. Annals of the Entomological Society of America 96: 585-593.

Emerson, A. E. 1952. The biogeography of termites. Bulletin American Museum of Natural History 99: 217-225.

FAO-UNEP. 2000. Termite Biology and Management Workshop Geneva, Switzertland.

Forschler, B. T., and G. Henderson. 1995. Subterranean termites behavioral reactions to water and survival of inundation: implications for field populations. Environmental Entomology 24: 1592-1597.

Foster, B. T., A. I. Cognato, and R. E. Gold. 2004. DNA-based identification of the eastern subterranean termite, Reticulitermes flavipes (Isoptera: Rhinotermitidae). Journal of Economic Entomology 97: 95-101.

Freeland, J. 2005. Molecular Ecology. John Wiley & Sons, Ltd.

Gentry, J., and G. Whitford. 1982. The relationship between wood litter infall and relative abundance and feeding activity of subterranean termites Reticulitermes spp. in three subterranean coastal plain habitats. Oecologia 54: 63-67.

Hall, T. A. 1999. Bioedit: a user friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41: 95-98.

Haverty, M., B. Forshler, and J. Nelson. 1996. An assessment of the taxonomy of Reticulitermes (Isoptera: Rhinotermitidae) from the

109 110 Southeastern United States based on cuticular hydrocarbons. Sociobiology 28: 287-318. Haverty, M., G. Getty, K. Copren, and V. Lewis. 2003. Flight phenology of sympatric populations of Reticulitermes (Isoptera: Rhinotermitidae) in Northern California: disparate flight intervals indicate reproductive isolation among cuticular hydrocarbon phenotypes. Annals of the Entomological Society of America 96: 828-833.

Haverty, M. I., and L. J. Nelson. 1997. Cuticular hydrocarbons of Reticulitermes (Isoptera: Rhinotermitidae) from Northern California indicate undescribed species. Comparative Biochemistry & Physiology B 118: 869-880.

Haverty, M. I., R. J. Woodrow, L. J. Nelson, and J. K. Grace. 2005. Identification of termite species by the hydrocarbons in their feces. Journal of Chemical Ecology 31: 2119-2151.

Holt, J., and M. Lepage. 2000. Termites and soil properties, pp. 389 - 408. In T. Abe, D. Bignell and M. Higashi [eds.], Termites: Evolution, Sociality, Symbioses, Ecology. Kluwer Academic Publishers.

Hostettler, N., D. Hall, and R. Scheffrahn. 1995. Intracolony morphometric variation and labral shape in Florida Reticulitermes (Isoptera:Rhinotermitidae) soldiers: Significance for identification. Florida Entomologist 78: 119 -129.

Houseman, R. M., and R. E. Gold. 2003. Factors that influence tunneling in the eastern subterranean termite, Reticulitermes flavipes (Kollar) (Isoptera: Rhinotermitidae). Journal of Agricultural & Urban Entomology 20: 69-81.

Houseman, R. M., R. E. Gold, and B. M. Pawson. 2001. Resource partitioning in two sympatric species of subterranean termites,

110 111 Reticulitermes flavipes and Reticulitermes hageni (Isoptera: Rhinotermitidae). Environmental Entomology 30: 673-685. Howard, R., S. Jones, J. Mauldin, and R. Beal. 1982. Abundance, distribution and colony size estimates for Reticulitermes spp. (Isoptera: Rhinotermitidae) in Southern Mississippi. Environmental Entomology 11: 1290-1293.

Hu, X. P., and A. G. Appel. 2004. Seasonal variation of critical thermal limits and temperature tolerance in Formosan and eastern subterranean termites (Isoptera: Rhinotermitidae). Environmental Entomology 33: 197-205.

Jenkins, T. M., C. J. Basten, R. Dean, S. Mitchell, S. Kresovich, and B. T. Forschler. 1998. Matriachal genetic structure of Reticulitermes (Isoptera: Rhinotermitidae) Populations. Sociobiology 33: 239-263.

Jenkins, T. M., M. I. Haverty, C. J. Basten, L. J. Nelson, M. Page, and B. T. Forschler. 2000. Correlation of Mitochondrial Haplotypes with Cuticular Hydrocarbon Phenotypes of Sympatric Reticulitermes Species from the Southeastern United States. Journal of Chemical Ecology 26: 1525-1542.

Kambhampati, S., and P. T. Smith. 1995. PCR primers for the amplification of four insect mitochondrial gene fragments. Insect Molecular Biology 4: 233-236.

Kambhampati, S., and P. Eggleton. 2000. Taxonomy and phylogeny of termites. In D. Bignell and M. Higashi [eds.], Termites; evolution, sociality, symbioses, ecology. Kluwer Academic Publishers.

Kard, B. M. 2003. Integrated pest management of subterranean termites (Isoptera). Journal of Entomological Sciences 38: 200-224.

111 112 Kofoid, C. A. 1934a. Termites and their control. University of California Press Association, London.

Kofoid, C. A. 1934b. Climatic factors affecting the local occurrence of termites and their geographical distribution, pp. 13-21. In C. A. Kofoid [ed.], Termites and their control University of California Press Association, London.

Kumar, S., K. Tamura, and M. Nei. 2004. Mega 3.1: Integrated software for molecular evolutionary genetics analysis and sequence alignment Briefings in Bioinformatics 5: 150-163.

La Fage, J. P., and W. Nutting. 1978. Nutrient dynamics of termites, pp. 165-232. In M. V. Brian [ed.], Production ecology of ants and termites. University Press, Cambridge.

Laine, L. V., and D. J. Wright. 2003. The life cycle of Reticulitermes spp. (Isoptera: Rhinotermitidae): What do we know? Bulletin of Entomological Research 93: 267-278.

Lewis, V. R. 1997. Alternative control strategies for termites. Journal of Agricultural Entomology 14: 291-307.

Messenger, M. 2003a. National Termite Survey, http://wwwtermitesurveycom/.

Messenger, M., N. Su, and R. Scheffrahn. 2002. Current distribution of the formosan subterranean termite and other termite species (Isoptera: Rhinotermitidae, Kalotermitidae) in Louisiana. Florida Entomologist 85: 580-587.

Messenger, M. T. 2003b. National termite survey, City of New Orleans Mosquito and Termite Control Board.

112 113 Nelson, L. J., L. G. Cool, B. T. Forschler, and M. I. Haverty. 2001. Correspondence of soldier defense secretion mixtures with cuticular hydrocarbon phenotypes for chemotaxonomy of the termite genus Reticulitermes in North America. Journal of Chemical Ecology 27: 1449-1479.

Nigh, T., and W. Schroeder. 2002. Atlas of Missouri Ecoregions. Missouri Department of Conservation

Nobre, T., L. Nunes, P. Eggleton, and D. E. Bignell. 2006. Distribution and genetic variation of Reticulitermes (Isoptera : Rhinotermitidae) in Portugal. Heredity 96: 403-409.

Nutting, W. L. 1990. Insecta: Isoptera, pp. 997-1032. In D. D. L. [ed.], Soil biology guide. Wiley- Interscience, New York.

Page, M., L. J. Nelson, B. T. Forschler, and M. I. Haverty. 2002. Cuticular hydrocarbons suggest three lineages in Reticulitermes (Isoptera: Rhinotermitidae) from North America. Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology 131: 305-324.

Posada, C., and K. Crandall. 2000. TCS: a computer program to estimate gene genealogies. Molecular Ecology 9: 1657-1660.

Posada, C., and K. Crandall. 2001. Intraspecific gene genealogies: trees grafting into networks. Trends Ecological Evolution: 37-45.

Roisin, I. 2000. Diversity and evolution of caste patterns, pp. 95-120. In T. Abe, D. Bignell and M. Higashi [eds.], Termites: Evolution, Sociality, Symbioses, Ecology. Kluwer Academic Publishers.

Rozas, J., J. Sanchez, X. Messeguer, and R. Rozas. 2003. DnaSP, DNA polymorphism analyses by the coalescent and other methods. Bioinformatics 19: 2496-2497.

113 114 Scheffrahn, R., and N. Su. 1994. Keys to soldier and winged adult termites (Isoptera) of Florida. Florida Entomologist 77: 460 - 474.

Simon, C., F. Frati, A. Beckenbach, B. Crespi, H. Liu, and P. Flook. 1994. Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Annals of the Entomological Society of America 87: 651-701.

Snyder, T. 1954. Order Isoptera. The termites of the United States and Canada, New York.

SPSS for Windows. 1997. Versión 8.0.0. Chicago: SPSS Inc.

Su, N. Y. 2002. Novel technologies for subterranean termite control. Sociobiology 40: 95-101.

Su, N. Y., and R. H. Scheffrah. 1990. Economically important termites in the United States and their control. Sociobiology 17: 77-94.

Su, N. Y., and R. H. Scheffrah. 1998. A review of subterranean termite control practices and prospects for integrated management programs. Integrated pest management reviews 3: 1-13.

Su, N. Y., and R. H. Scheffrah. 2000. Termites as pest of buildings, pp. 437-454. In T. Abe, D. Bignell and M. Higashi [eds.], Termites: Evolution, Sociality, Symbioses, Ecology. Kluwer Academic Publishers.

Szalanski, A. L., J. W. Austin, and C. B. Owens. 2003. Identification of Reticulitermes spp. (Isoptera: Reticulitermatidae) from south central United States by PCR-RFLP. Journal of Economic Entomology 96: 1514 -1519.

Szalanski, A. L., J. W. Austin, J. McKern, and M. T. Messenger. 2006. Genetic evidence for a new subterranean termite species (Isoptera :

114 115 Rhinotermitidae) from western United States and Canada. Florida Entomologist 89: 299-304. Templeton, A., K. Crandall, and F. Sing. 1992. A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. cladogram estimation. Genetics 132: 619-633.

Uva, P., J. L. Clement, and A. G. Bagneres. Colonial and geographic variations in agonistic behaviour, cuticular hydrocarbons and mtDNA of Italian populations of Reticulitermes lucifugus (Isoptera, Rhinotermitidae). SO - Insectes Sociaux-Social Insects 51(2) May 2004 163-170.

Uva, P., J.-L. Clement, J. W. Austin, J. Aubert, V. Zaffagnini, A. Quintana, and A.-G. Bagneres. 2004. Origin of a new Reticulitermes termite (Isoptera, Rhinotermitidae) inferred from mitochondrial and nuclear DNA data. Molecular Phylogenetics & Evolution 30: 344-353.

Vargo, E. L. 2003. Genetic structure of Reticulitermes flavipes and R. virginicus (Isoptera: Rhinotermitidae) colonies in an urban habitat and tracking of colonies following treatment with hexaflumuron bait. Environmental Entomology 32: 1271-1282.

Vargo, E. L., and J. R. Carlson. 2006. Comparative study of breeding systems of sympatric subterranean termites (Reticulitermes flavipes and R-hageni) in Central North Carolina using two classes of molecular genetic markers. Environmental Entomology 35: 173-187.

Vargo, E. L., T. R. Juba, and C. J. Deheer. 2006. Relative abundance and comparative breeding structure of subterranean termite colonies (Reticulitermes flavipes, Reticulitermes hageni, Reticulitermes virginicus, and Coptotermes formosanus) in a South Carolina low

115 116 country site as revealed by molecular markers. Annals of the Entomological Society of America 99: 1101-1109. Vincze, T., Posfai, J. and Roberts, R.J. . 2003. NEBcutter: a program to cleave DNA with restriction enzymes. Nucleic Acids Res 31: 3688-3691

Weesner, F. 1965a. The Termites of the United States a Handbook, pp. 64 The National Pest Control Association, NY.

Weesner, F. 1965b. The Termites of the United States a Handbook. The National Pest Control Association.

Weesner, F. 1970. Termites of the Neartic region, pp. 477 - 525. In W. F. a. K. Khrisna [ed.], Biology of termites. Academic Press, New York

Whitney - King, S., J. W. Austin, and A. Szalanski. 2007. Use of soldier pronotal width and mitochondrial DNA sequencing to distinguish the subterranean termites Reticulitermes flavipes (Kollar) and R. virginicus ( Banks) (Isoptera : Rhinotermitidae), on the Delmarva Peninsula: Delaware, Maryland and Virginia U.S.A. Entomological news 118: 41- 48.

Wood, T., and W. Sands. 1978. The role of termites in ecosystems., pp. 245 - 292. In M. Brian [ed.], Production ecology of ants and termites. Cambridge University Press.

116 117

VITA

Olga Patricia Pinzon was born in Bogota, Colombia, South America. After attending Universidad Distrital “Francisco Jose de Caldas” –Bogota (1987) where she received her B.S. in Forest Engineering, she finished her M.S. in

Agricultural Sciences-Entomology from the Universidad Nacional de Colombia

(2001). She won a scholarship with the Fulbright Program to join the Ph D program in Entomology at the University of Missouri in the fall of 2002.

She has been researcher in forest entomology for various private and public institutions in Colombia and taught in public institutions. She is married to

Victor Manuel Nieto (Forest Engineer M.S.) from Bogota, and has a son named Manuel Andres Nieto. She is presently professor at the Universidad

Distrital Francisco Jose de Caldas in Bogota.

117 118