120 563 THS (hhszf-T, llllll'lllfllllllllIllllllllllll'lll'llllllll 3 1293 01389 3601 This is to certify that the thesis entitled The geographic distribution of Silene latifolia and its anther-smut pathogen Microbotryum violaceum in the eastern United States presented by Emily.J;‘Lyons has been accepted towards fulfillment of the requirements for M.S. degree in Botany and Plant ‘Path. Qua/fl / Major professor/ Date/ 9(2/19L1/ fé’ 0.7639 MS U i: an Affirmative Action/Equal Opportunity Institution _._ H-—_~ ._—__———.—.._ LIBRARY Mlchlgan State Unlvorelty PLACE ll RENEW BOX to remove We checkout from your record. TO AVOID F INES return on or betore dete due. DATE DUE DATE DUE DATE DUE MSU le An Affirmative Action/Ewe! Opportuntty lnetltulon Wanna-a1 m ,wfi THE GEOGRAPHIC DISTRIBUTION OF SILENE LATIFQLIA AND ITS ANTHER- SMUT PATHOGEN MICROBQTRYUM VIQLAQEQM IN THE EASTERN UNITED STATES By Emily J. Lyons A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Botany and Plant Pathology 1996 ABSTRACT THE GEOGRAPHIC DISTRIBUTION OF SILENE LATIFQLIA AND ITS ANTHER— SMUT PATHOGEN MIQRQBQIR! QM VIQLAQEUM IN THE EASTERN UNITED STATES By Emily J. Lyons The eastern United States was censused to map the geographic distribution of 5M Ma and its obligate anther-smut pathogen MW m. The smut fungus is found primarily in western Virginia though the natural range of S, latiiolia extends throughout most of the eastern US. Exceptions included three infected populations on Nantucket Island and one infected plant in northwestern New York. Northeastern states were characterized by large populations of i M that were randomly distributed across the landscape, while those in Virginia were smaller and more clumped. This may reduce the pathogen’s ability to invade northern host populations. A two year field experiment conducted at seven sites, composing two south-north transects in the eastern United States indicated that fl violagum is able to infect and overwinter in S, Latifglja plants at northern latitudes even though it is rare in these areas. Disease rarity at northern latitudes may be due to: 1) historical accident of disease introduction, 2) difi‘erences in plant population size and resistance, or 3) variation in pollinator foraging strategies. ACKNOWLEDGMENTS A three year study such as this one involved a great deal of traveling and manual labor and is not the sort of thing that one can do alone. Not surprisingly then, I am grateful and indebted to many people for their assistance and support. First, many thanks to my advisor Andy who provided me with the opportunity to tackle this project. Over the past three years, he has taught me a great deal about the rigors of field biology; patience, physical labor, and critical thought. Perhaps more importantly though, he never forgot the importance of good laughs, scenic views and cheesy tourist spots. Love and special thanks to my mom, whose skepticism made me all the more confident that this was what I wanted to do. To Anita Davelos, my first and favorite ofiicemate, thank you for taking me in and teaching me how to be a graduate student and in the process, a kinder, gentler person. Many thanks to my committee, Tom Getty, Susan Kalisz, and Stephen Tonsor for their academic advice and support and to Don Hall and members of the Ecology and Evolutionary Biology program for fostering a simultaneously challenging and social learning environment. Additionally, I am grateful to the following fi'iends and colleagues for their assistance, advice, support and friendship: I. Antonovics, H. Alexander, P. Batra, E. Boydston, J. Byme, E. Capaldi, B. Cardinale, V. Cooper, K. Desmarais, J. Dudycha, S. Emery, R Fuller, M. Hatch, A. Hernandez, L. Leege, M. Loveless, S. Paderewski, J. iii Rosinski, J. Secki, J. Staser, M. Szykrnan, D. Taylor, P. Thrall, E. Wagner, and H. Wilbur. Finally, and perhaps most importantly, I want to thank A. South, R. Jenna, J. Kilgore, and B. Calen for their assistance in the field and on the road and J. Guyton, J. Nobis and M. McLouth for bailing me out of office troubles more times than I can count. That said, Peanut Barrel anyone? iv TABLE OF CONTENTS LIST OF TABLES .......................................................................................................... vi LIST OF FIGURES ....................................................................................................... vii CHAPTER 1 REVIEW OF LITERATURE ........................................................................................... l The Pathosystem .................................................................................................. 1 History of the Pathosystem ................................................................................... 3 Silene-Microbotgum Population Studies .............................................................. 4 Silene-Microbotgmm Metapopulation Studies ...................................................... 8 Wm: A Model System .............................................................. l 1 CHAPTER 2 DESCRIPTIVE AND EXPERIMENTAL STUDIES ..................................................... 13 Introduction ........................................................................................................ 13 Methods ............................................................................................................. 19 Results ............................................................................................................... 27 Discussion .......................................................................................................... 38 CHAPTER 3 CONCLUSIONS ........................................................................................................... 44 APPENDIX Appendix I: Counties in the eastern United States where S, latifglia is present. Presence of populations infected with M_. violaceum is denoted with a (D) ........ 47 BIBLIOGRAPHY .......................................................................................................... 48 LIST OF TABLES Table 1: Latitude and longitude coordinates for the seven sites used in the field experiment .......................................................................................................... 22 Table 2: Non- parametric, post-data pairwise comparisons of $_. latifolia distance of populations from census starting point and population size for the three regions ............................................................................................................... 33 Table 3: Weighted mean distance to populations and population size of S, latifolia for three regions of the eastern United States ........................................................... 34 Table 4: 1994 survival and infection of S_. latifolia from the seven field sites .................... 36 Table 5: 1995 survival and infection of i lgifglia fiom the seven field sites .................... 36 LIST OF FIGURES Figure 1: Geographic Distribution of Silene latifolia and Microbotryum violaceum in the eastern United States .......................................................................................... 29 Figure 2: Size and distribution of S, latifolia populations in the three regions of the eastern United States ...................................................................................................... 32 vii Chapter 1 REVIEW OF LITERATURE The Pathosystem $i_le_n_e latifolia (Poiret) (= Silene alba [Miller] Krause) (Caryophyllaceae), white campion, is a short-lived, dioecious, perennial that is native to Europe though it has been found in the eastern United States since the mid-1800’s (McNeill, 1977). Flowers of 5, Elm open in the evening and remain open until mid-morning the following day. The primary pollinators are bumble bees, butterflies, sphingid moths, and other Lepidoptera (Shykoff & Bucheli, 1995; Roche et al., 1995; Altizer, unpub.). i latifoha grows primarily along roadsides and in other ruderal habitats. Miggbotmgm violaceum (=Lliilggg Mace; Pers. [Deml & OberwinklerD (U stilaginales, Basidiomycotina), the anther-smut pathogen of S, l_aergl_ia, is a basidiomycete fungus that is spread by vectors during the process that would normally be pollination, thus it is often analogized to a sexually transmitted disease. Diploid teliospores of the fungus are deposited on healthy, susceptible S_. latijoha plants where they germinate and undergo meiosis. The resulting haploid sporidia of opposite mating type then conjugate and form an infection hypha that can penetrate the plant tissue. Mycelium grow throughout the adjacent stems and into the rootstock where the firngus perenniates. Newly diseased flowers appear after the firngus has grown into the host and entered newly developing flower ‘buds (Batcho & Audran, 1980), a period that ranges between three weeks and two months (Alexander, 1990b). Successful infection by M, violaceum is manifested in the production of spore producing starninate flowers in both male and female plants. In males, the normal pollen producing stamens are replaced with spore sacs; in infected female flowers, the ovary becomes rudimentary and spore sacs develop on stamen-like structures. Plants often become systemically infected so that the following year all flowers are infected, resulting in complete sterility. Vertical transmission through the seed does not occur (Baker, 1947; E. Lyons, unpub.) though transmission can occur passively by spores falling on seedlings growing close to infected plants (Alexander and Maltby, 1991). M_. violaceum has been recorded on 21 species of the Caryophyllaceae in North America, including §_. Ma and _S_. 111m and 92 species in Europe (Thrall et al., 1993). The distribution of M .violageum among species in the Caryophyllaceae is related to the life span of its host species. More specifically, & violmm is more commonly found on perennial species than on annual species. This is not surprising given the life cycle of M, violaceum. Recall that fungus may take as long as two months from the time its spores are deposited, a large portion of the growing season, to invade newly developing flowers and cause spore production. Complete systemic infection usually occurs in the growing season following initial infection and the fungus is therefore reliant upon the perennial life history of its host plant to insure further transmission. Given this, it is plausible to suggest that, if the ability of Mg violaceum to survive and persist is, in part, dependent on the life history of its host, then the factors that influence host life histories may also affect the distribution of M violagm History of the Pathosystem M violaceum occurs on S, latifolia and other members of the Caryophyllaceae in both North America and Europe. It is hypothesized that S latifolia was introduced to North America in the mid-1800’s when agricultural soil, used as ballast on European trade ships, was deposited on the shores in port areas. S, latifolia seeds contained in the soil germinated and the range of the plant subsequently spread throughout most of the eastern United States and Canada. It is unclear how M violaceum was introduced to North America though it infects both the introduced S, latifolia and the native fire-pink, Silene virginica. Two hypotheses have been explored experimentally (Antonovics et al., 1996): l) The presence of infected S, m in close proximity to infected S, lgthQLia represents a host-shift from the endemic S, W to S, 1% or vice versa; 2) The anther-smut 9_f S, latitblia is related to isolates found in Europe and thus may have been introduced with S_. latfiolra. Results of cross-infection studies between the two Silene species reveal that the isolates of M violaceum fi'om both species are very host specific. Very little (1.5%) cross infection occurred when S, “Lifglja plants were inoculated with fungal isolates from S, mginjga and vice versa (Antonovics et. al, 1996). These results do not support hypothesis 1 that the infection of both Sil_er;e species by M Mm is due to a host shift from one species to the other. With respect to hypothesis 2, electrophoretic studies indicate that M _vjdmm isolates fi'om S latifolia in the southeastern US. resemble isolates fi'om an infected population in England (Antonovics, et al., in press). These data suggest that the anther-smut on S latifolia in the US. may have been introduced from Europe with the host plant. The origin of M violaceum isolates infecting S virginica is still unclear. SilgngMigmbotgyum Population Studies The detrimental effect that pathogens often have on their hosts has led to the suggestion that disease-causing organisms can regulate host population size, afl'ect genetic variability in host populations and influence species coexistence in communities (e. g. Burdon, 1987; May & Anderson, 1983 a,b; Alexander & Antonovics, 1988). Given the severe fitness effect, sterilization, that M violaceum has on S MM, it is likely that the fungus is influencing the population dynamics of its host. Since the mid-nineteen eighties, a great deal of research has been conducted on the Silene-Micrgbgtggm host-pathogen system in an attempt to understand the population dynamics of both host and pathogen. A deterministic model, by Alexander and Antonovics (1988) simulated the dynamics in the Silene-Microbotggm system and explored the effect of plant recruitment and disease spread on the fate of infected S M populations. The parameters for the model were estimated using demographic data from infected populations of S RIM: in southwestern Virginia. Three outcomes were possible, depending on the initial model parameters. At low host recruitment rates and high infection rates, the outcome was total infection of the population resulting local extinction of both the host and pathogen. When infection was low and recruitment rates were high, the pathogen was expunged fiom the plant population. When both recruitment and infection rates were high, the host and pathogen could co-exist at equilibrium and regulate each other’s populations (Alexander & Antonovics, 1988). These results are highly dependent on the transmission mode of the pathogen and the probability of successful systemic infection of the host. As mentioned above, M violaceum is a vector transmitted disease and therefore successful transmission from infected to healthy, susceptible hosts is highly dependent on pollinator behavior. Hand pollination studies performed in populations of S lgtifglLa, indicate that the populations are pollinator limited. Plants that were hand pollinated produced approximately twice the number of flowers with maturing fruits and seeds as those that were naturally vector pollinated (Alexander, 1987). Inadequate pollinator service influences not only plant fecundity but successfiil transmission of M W as well. Investigations of pollinator visitation patterns in S M populations reveal that pollinators typically deposit pollen and/or spores on plants that are in close proximity to the host plant (Roche et al, 1995; S. Altizer, unpublished). Vectors prefer plants with large floral displays and preferentially visit males over females and healthy plants over infected plants (Shykofi‘ & Bucheli, 1995). The pollinator preference of males over females or infected plants corresponds with the production of higher quality nectar in males. Relative to females and infected plants, males consistently produce nectar with a higher sugar concentration (Shykofl' & Bucheli, 1995). Also related to the preferential visitation of male plants by pollinators is the fact that males almost always produce many more flowers than females. Thus they are usually the plants with the largest floral displays as well as the highest quality nectar. These factors may put males at a higher risk of infection by M violaceum. This is consistent with the findings of Thrall and Jarosz (1994a) that males that begin flowering earlier, produce more flowers and remain in flower longer are two times more likely to become infected than females with shorter flowering periods and fewer flowers overall. Regardless of sex, plants that produce more flowers are, in general, more likely to become infected. Pollinator behavior is not the only factor influencing the transmission of M viglaggim in populations of S l_a_tileia. The pollinator must not only visit an infected plant and pick up spores, but it must also deposit those spores on a healthy, susceptible host. An experimental inoculation study performed by Alexander et al. (1993) indicates that S 141M; plants vary in their susceptibility to isolates of M violaceum. This variation in susceptibility is not only present, but heritable and consistent in both the greenhouse and field environments (Alexander & Antonovics, 1995). Given that heritable variation in resistance exists in natural populations of S Mtifgjia, it is essential to understand how the host genetics influence the population dynamics of both plant and pathogen. In order to further investigate the transmission dynamics of M viglgeum in the context of host resistance, Thrall and Jarosz (1994a,b) established experimental populations of S latijofia containing both resistant and susceptible plants with difl‘erent total densities. In addition, the frequency and density of plants infected with M violaggrm was also varied. The actual numbers of transitions fi'om healthy to infected in S l_a_ti_fg;li3 were compared to predictions generated from computer simulations and it is clear that susceptible and resistant populations have very different dynamics. Host and pathogen are able to coexist in susceptible populations while the fungus is expunged form resistant populations (Thrall & Jarosz, 1994b). However, knowledge of population resistance structure alone is not adequate to explain the natural population variation in disease levels. More long-term, across-season processes are also important. These include: host recruitment, recovery from infection and over-wintering mortality (Thrall & Jarosz, 1994a) Variation in host recruitment and adult mortality may alter population age structure and subsequently affect disease spread (Burdon, 1987). First year plants are more likely to delay flowering until later in the growing season relative to older plants. Since time spent in flower is positively correlated with the probability of infection by M violaceum, then pathogen transmission dynamics should vary with the relative abundance of first year and older plants (Thrall & Jarosz, 1994a). Additionally, the pathogen is expected to be rapidly lost fi'om plant populations where host overwintering mortality is high (Thrall & Jarosz, 1994b). Even if M violagggm is not lost, these across-season processes reduce the fi'equency of infected plants in the population at the start of the next growing season thereby decreasing the transmission rates within the next growing season (Thrall & Jarosz, 1994a). What is clear from both of the above studies (Thrall & Jarosz, 1994a,b) is that in the Silfle-Mimbotrmm system, host and pathogen within population dynamics are unstable at the scale of the population. The combined effects of plant resistance, pollinator avoidance of infected individuals, and the occasional year where overwintering mortality is high, lead to the conclusion that M violaceum cannot persist indefinitely within a single plant population. This suggests that stable coexistence of host and pathogen may be reliant upon long distance pollinator dispersal to affect new colonizations to counteract local extinction events of both the host and pathogen. This possibility requires the investigation of the Silene-Micrgbgtflum system in a metapopulation context. SilgngMicrobotgum Metapopulation Studies The concept of the metapopulation, defined here as a system of interconnected populations, has recently been used to address population processes on a larger scale (Hanski, 1991a,b). For example, the problem of increasing habitat fi'agmentation has prompted researchers to investigate systems in the context of the metapopulation where genetic structure is influenced by colonization and extinction processes (Hanski et al., 1994; Antonovics et al., 1994). The idea of a metapopulation is particularly amenable to the Silene-Microbotmm system because the host-pathogen dynamics at the population level are unstable and coexistence may be reliant upon continual colonization of the pathogen or resistant or susceptible plants. Metapopulation simulation studies support this idea showing that coexistence of host and pathogen can occur even when within- population dynamics lead to local extinction (Antonovics, et al., 1994). In order to better determine the applicability of the metapopulation concept to the Silene-Micrgbgtgmm system, a roadside census of approximately ninety miles of roadside in southwestern Virginia was initiated in 1988. This study recorded every S Lati_fol_ia population that was encountered and counted the number of healthy and infected individuals in order to determine when colonizations or extinctions occurred for either the plant or pathogen (Antonovics, et al., 1994). The data from the first three years of this study reveal that there is a high turnover rate of populations. In addition, the colonization rate of the fungus exceeds its extinction rate thereby suggesting that the disease is increasing in that area of Virginia (Antonovics, et al., 1994). More generally, populations are more likely to be infected if they are large though among infected populations, smaller populations tend to have a higher percentage of disease. Additionally, colonization and establishment of both the host and pathogen is dependent upon their proximity to the nearest source population as most new populations and infections are found close to existing ones. However, some long-distance dispersal may occur because new populations do arise a considerable distance (> 1 mile) from any existing populations. It is plausible, from these data, to suggest that populations of _S_. latifolia and M m are interconnected even over long distances and are characterized by some long distance dispersal of pollen, seed and spores (Antonovics, et al., 1994). For the purpose of further exploring the consequences of interconnected populations on disease spread, Thrall and Antonovics (1996) established an experimental metapopulation consisting of replicate sets of experimental populations separated by increasing distances (5, 10, 20, 40 and 80m). The results of this study indicate that disease spread is greater within the more isolated populations. This suggests that vectors may forage difl‘erently in small isolated populations than they do in those that are large resulting in the larger, or more closely distributed populations receiving a disproportionate 10 number of pollinator visits (Thrall & Antonovics, 1996). This phenomenon could have profound implications for the spread of M violaceum on a larger geographic scale. The transmission and subsequent establishment of M viglacgm in regions that are far fi'om a source population may be hindered by the movements of vectors. Ifthe distribution of S LatifMia across a landscape is characterized by small populations that are far apart, the data presented above (Thrall & Antonovics, 1996) suggest it is likely that pollinators will forage less selectively within a population and seldom move among populations. This may limit the spread of M violaceum to localized populations and ultimately result in local extinction of both host and pathogen. Conversely, if populations of _S_. Latif_oli_a are large and continuously or randomly distributed across a landscape, it is possible that pollinators are more likely to move among populations and M W may be more easily transmitted among populations to invade new areas. While the above speculations seem plausible, it is necessary to recall that the transmission of M violaceum is frequency dependent and therefore reliant upon the probability of a vector visiting an infected plant and subsequently landing on a healthy, susceptible host (Thrall & Jarosz, 1994b). This fact may reduce the advantage of large populations for fungal transmission. For example, consider one infected plant in a population of one thousand plants. The frequency of the infected plant in that population is 0.001 and thus the probability that a vector will land on that plant is relatively low. Conversely, in a population of ten plants, one infected plant has a frequency of 0. l, and the probability of vector contact is greatly increased. It is therefore necessary to consider 11 the importance of both population size and distribution when investigating the spread of M viglaggim on a geographic scale and beyond the level of the metapopulation. Silene-Miggghotflum: A Model System As evidenced in the review of literature above, there has been a great deal of research conducted on the Silene-Micrgbotmm host-pathogen system. One reason for this is the ease of use. The plant and fungus in this system are easily manipulated in both the field and greenhouse. In addition, the fact that the transmission of M violaceum is fiequency dependent and reliant upon vectors makes this system useful for understanding the dynamics of other vector transmitted diseases that are less manipulable. The systemic and sterilizing effects of the fungus on the host in conjunction with the mode of transmission makes it possible to liken the Silene-Migrobotrmm system to a sexually transmitted disease. This analogy can be taken one step firrther in the context of metapopulation dynamics. For example, May and Anderson (1990) assert, based on the levels of sequence divergence between simian and human lentivinrses, that HIV, the virus that causes AIDS (Acquired Immune Deficiency Syndrome) has persisted in human populations for over a century. If this is indeed the case, they then assert that the persistence of the virus can be explained by the spatial models that underly sexual contacts within and among rural villages. In a metapopulation context, the villages are individual populations where the virus might be present. Any among village sexual contact potentially serves to maintain the virus in the overall geographic region. Increases in inter-village contact may account 12 for the recent increase in prevalence and spread of HIV. This is not unlike the way in which M violaceum is maintained in populations of S latifolia. In addition to the applicability of the Silene-Micrgbgtggm system to the understanding of sexually transmitted diseases, research on the system contributes to an emerging body of literature on the influence of metapopulation dynamics in plant- pathogen interactions (Frank, 1992; Burdon & Jarosz, 1992). For example, in the LiMum marginale-Melampsora li_ni host-pathogen system, Burdon and Jarosz determined that the resistance structure of individual host populations is likely to be influenced by differences in the rates of local extinction of both host and pathogen populations. Given this, the dynamics of both the host and pathogen may best be understood in the context of a metapopulation characterized by frequent colonization and extinction events. (Burdon & Jarosz, 1992). Additionally, metapopulations have been recognized as important in influencing the coevolution of hosts and pathogens (Frank, 1992). The study of the Sligng-Microbgtggm system not only contributes to this emerging body of knowledge but provides an example of a situation where the concept of the metapopulation is particularly useful for the understanding of the dynamics of both host and pathogen. Chapter 2 DESCRIPTIVE AND EXPERIJVIENTAL STUDIES Introduction The relationships of a species with its environment are reflected in the distribution of its abundance in both time and space. For almost a century, however, when ecologists have studied abundance, they have nearly always studied either population dynamics, (ie, fluctuations in the numbers of individuals in a single local population over time) or species distributions. The history of p0pulation dynamics extends fiom the theoretical contributions of Pearl (1925) and Verhulst (193 8) and the classical empirical studies of Elton (1924, 1942), to recent attempts to understand the complex fluctuations within populations revealed in long time series (May .1974, 1987, Pirnm and Redfeam 1988, 1989, Antonovics et al. 1994) In contrast, when ecologists have studied species distributions, they have focused primarily on territoriality, foraging movements, habitat selection, and other processes that influence the spatial dispersion of individual organisms within populations or among habitats (e.g., Skellam 1951, Fretwell 1972). When biogeographers have studied distributions, they have been concerned primarily with the influence of contemporary processes and historical events on the size, location, and limits of geographic ranges of species as they appear on maps (e.g. Root 1988b, Meyers and Giller 1989, and Hengeveld 1990). 13 14 Similarly, biogeographers have equally rarely studied the abundance and distribution of individuals within a geographic range. There have been some notable exceptions (e.g. Whittaker 1967, MacArthur 1972, Hengeveld and Haeck 1981, 1982, Bock and Ricklefs 1983, Bock, 1984, Brown 1984, Shoener 1987, 1990, Root 1988 a,b Maurer 1994). These large scale geographic studies focused primarily on the influence of abiotic factors such as rainfall and temperature on species distributions. In particular, MacArthur refers directly to what he calls the “environmental control of community structure” or the concept that the members of an ecological community are determined by their ability to survive and reproduce in a particular environment (MacArthur, 1972). Still largely missing, however, is any concerted theoretical or empirical research on the magnitude and pattern of interspecific geographic variation in local population density, particularly in the context of host-pathogen interactions. Host-pathogen interactions are particularly interesting because pathogen distributions are likely to be strongly affected by both abiotic factors (temperature, rainfall, and humidity) and biotic factors (host resistance and availability). A few of the host-pathogen geographic distribution studies that have been done, investigated either powdery mildew resistance in wild barley (Nevo et. al, 1979, 1983, 1985a, 1985b, 1986) or crown rust resistance in wild oat species (Wahl, 1970). Because the host and pathogen were investigated separately in these studies, they relied heavily on correlations between host resistance and ecological factors to determine pathogen distribution. In this paper we discuss both a descriptive and empirical approach to studying the distribution and abundance of Silene latifolia (=Sileng alba) and its anther- 15 smut pathogen Mjcgojzommm Mm (=Lfitj1agg W) in the eastern United States. Anecdotal observations made by Dr. Janis Antonovics and others indicated the possibility that the range of M violaceum was restricted to only a small fraction of S Mia’s range. The Silene-Micrgbgtmm host-pathogen system in particularly interesting because we can readily identify a series of highly probable abiotic and biotic reasons why the pathogen might be limited to only certain parts of the range of its host. Below I provide a description of the Sflmhdiggbgtmm system, as some knowledge of the life histories of these organisms will make our hypotheses more easily understood. Si1e_n_e latif_olia is a short-lived dioecious perennial found along roadsides and in other ruderal habitats. The plant was introduced from Europe in the mid-1800's and has since spread throughout eastern North America (McNeill, 197 7). The presence of the plant provided suitable open niches for the colonization of M violamm and not surprisingly, the anther-smut is present in populations of S laflfiofia in North America. The pathogen has most likely migrated from Europe, since it is not closely related to M Mm found on other Sim species that are native to the United States (Antonovics et al. 1996). Prior to the initiation of our study, M. violaceum had only been reported in northwestern Virginia in the region surrounding Mountain Lake Biological Station where it is common, and once by Janis Antonovics and Stephen Tonsor (pers. comm.) in 1988 in southwestern Michigan near the WK. Kellogg Biological Station (KBS), though it is no longer present in this area. 16 The Silene-Microbotgrum plant-pathogen system is characterized by the production of teliospores on anther-like structures in male flowers. In female plants, M violaceum causes the production of anther-like structures. In addition, the ovary becomes rudimentary and sterile. The teliospores are transmitted to new hosts by insect pollinators (Baker, 1947; Hassan and MacDonald, 1971; Lee, 1981; Jennersten, 1983, 1988: Alexander & Antonovics 1988; Thrall et al., 1993b). Newly diseased flowers appear after the fungus has grown into the host and entered newly developing flower buds (Batcho & Audran, 1980), a period that ranges between three weeks and two months (Alexander, 1990b). The fungus over-winters in the root crown of the plant resulting in systemic infection and complete sterility of the plant. We propose five hypotheses for the rarity of M. violaceum in northern latitudes. They are: l) The distribution of M. violaceum is the result of an historical accident i.e., infection foci initiated at MLBS and KBS have not yet spread to other parts of S M’s range. The introduction of S 133% from Europe occurred only recently in the early 1800’s and there is evidence to suggest that the firngus had a restricted introduction and then spread from that point. This is fiirther corroborated by the fact that it is not the same anther-smut that infects members of the Caryophyllaceae (e. g. Sm mg) native to the United States (Antonovics et al., 1996). 2) There is a lack of pollinator service such that pollinators are not visiting infected plants with a high enough frequency to insure fungal transmission within and among metapopulations. Unpublished work of Ms. Sonia Altizer and others 17 indicates that bumblebees can discriminate between healthy and infected S latifioLia flowers, and preferentially visit those that are healthy. The way in which pollinators forage has implications for fungal transmission both within and among S Mtifofla populations. For example, it has been suggested by Altizer and others that in small plant populations bumblebees are less discriminatory in the flowers that they visit than they are in large populations, often foraging on all of the flowers in a small population (Altizer et al., unpublished). In contrast, hawk moths, the primary nocturnal pollinator of S lgti_le_i_a,, are trapliners and regularly move from p0pulation to population carrying spores and pollen. The relative abundance of these two types of pollinators as well as S latifolia population structure is likely to influence pollinator foraging behavior and subsequently, transmission of M violaceum. 3) High levels of resistance among northern S. MOI—ii. populations prevent the successful infection and subsequent persistence of M violaceum at northern latitudes. Highly resistant plants of S latifolia are known to occur in southwestern Virginia (Alexander et al. 1993). In addition, computer simulations indicate that in populations where resistance is high, M W is readily expunged (Thrall and Jarosz 1994b). Thus, the rarity of infected populations at northern latitudes might be due to high levels of resistance in S latifolia prohibiting the persistence of the firngus. 4) The inability of the pathogen to establish perennial infection limits further transmission. There are three sub-hypotheses within this hypothesis. 1) Mthin the Caryophyllaceae, the distribution of M violacegm is closely related to the life span of its host species with the proportion of perennial species on which anther-smuts have been 18 reported being five times higher than the proportion of annuals (Thrall et. al, 1993b). At northern latitudes, the life history of S latjf_oli_a may change from perennial to annual habit due to harsh over-wintering conditions. It may be that the temperatures in the north are too cold during the winter for the plants to survive to flower again in the spring. 2) The growing season at northern latitudes is shorter than in more southern regions resulting in a shorter and perhaps insufficient amount of time for the flrngus to grow from its point of deposition on a flower down into the root crown where it must be to over-winter. 3) It is possible that even if the fungus does successfully infect a plant and reach the root crown, it is unable to survive the cold winter temperatures in the north. 5) Plant population structure in areas outside Virginia is not conducive to invasion by the pathogen. By this we mean to suggest that the conditions are not optimal for pollinator movement of spores and transmission is subsequently reduced as a result. The transmission of M violaceum is fiequency dependent, and thus reliant on the probability of a pollinator landing on an infected flower and depositing spores on to healthy, susceptible flowers (Thrall and J arosz, 1994b). Therefore, there are two problems associated with fungal invasion into a region: 1) The fi'equency of disease in the colonized population. In larger populations, the fi'equency of disease is low and therefore less spread is expected. 2) The movement of the fungus to another population must occur before it becomes locally extinct. The facility of movement of the pathogen may be determined by the dispersion (uniform, random, or clumped) of the neighboring populations (A.M. Jarosz, P.H. Thrall, and E]. Lyons, unpublished). It is realistic to expect that the size and distribution of S latifolia populations would be important in 19 determining pollinator foraging behavior and firngal transmission. We investigated these five hypotheses using a combination of the roadside census and field transplant studies described below. Methods Roadside Census Data Collection In the summer of 1993, 1994 and 1995 twelve states in the eastern United States were censused for the presence of Siege latifolia and mm m. 1993: Virginia, Tennessee, Ohio, and Michigan; 1994: New York, Connecticut, Vermont, New Hampshire, Massachusetts, New Jersey, and Maryland; 1995: Pennsylvania. In addition, data were gathered on population sizes and distances between populations to test our hypothesis regarding the role of metapopulation structure in the successful colonization of M Mohegan. States were censused according to the following protocol. To expedite the process of covering large areas (ie, states) in a reasonable amount of time, the following census protocol was developed. A starting point for a slow segment was haphazardly chosen on the map (De Lorme’s detailed topographic atlases were used) and biased by concentrating on disturbed or open farm areas where S latifolia was likely to be present. From that starting point we drove slowly (20 mph) until we found S latifolia plants or until 5 miles elapsed. If S latifolia plants were found within 5 miles then we drove 0.3 miles to the end of the population and estimated the number of plants present. Three tenths of a mile was chosen as the distance over which the population size would be estimated. This was the size of the largest population of S lathngia that we found in our first week of censusing. Population sizes were put into the 20 following categories : 1-10, 10-100. 100-1000 or >100. The presence or absence of infected plants was also recorded for each population. Ifno S latifolia plants were found within 5 miles of the starting point or no nearest neighbor could be found within 5 miles of the previous population, we drove 5 miles at the speed limit without looking for plants and then resumed the slow census according to the rules above. Roadside Census Analysis The information gained fi'om the census is plotted on a United States map divided by county (Figure l). Counties shaded in red indicate that S latifolia populations are present but no M violaceum was found. Counties shaded in blue denote the presence of infected plants, while counties shaded in yellow are those in which censuses were done but S latjfiolia was not found. For the purpose of analysis the map is divided into three regions based on latitude. Region 1 includes those populations south of 39°N. Region 2 includes those populations between 39°N and 42°N. Region 3 includes populations located at latitudes greater than 42°N. Counties that straddle a latitudinal boundary were considered to be in the more southerly region. These regions were defined in an attempt to make distinctions that might be more ecologically meaningful in terms of fungal growth and establishment than arbitrary state boundaries. Factors used in making this decision include temperature, rainfall and overall weather conditions. The data are categorized by the distance (in miles) of populations from the starting point of a slow segment (0-1, 1-2, 2-3, 3-4, 4-5, >5) and by population size (number of plants) (1-10, 10-100, 100-1000, >1000). The mean (weighted) distances are calculated using 0.5, 1.5, 2.5, 3.5, 4.5, and 5.5 to weight each of the above distance categories 21 respectively. The mean (weighted) population sizes were calculated using 5, 50, 500, and 1500 to weight each of the size categories. These weights are somewhat arbitrary though with the exception of 1,500 they represent the mid-point of each category. To further explore the differences between the three regions with respect to population size and distance between populations, a non-parametric, one-way analysis of variance (AN OVA) was performed using the NPARlWAY procedure in SAS (SAS Institute, 1995). The rank sums resulting from this procedure were then used to hand calculate post-data pair-wise comparisons among the three regions with respect to both size and distance (Conover, 1988). Field Experiment In an attempt to test our hypotheses regarding the rarity of M. violaceum at latitudes above 39°N we initiated a two year field transplant experiment. Seven sites were chosen forming two south-north transects in the eastern United States. The western transects included: Dunbar Forest Experiment Station (Dunbar), Sault St. Marie, MI; WK. Kellogg Biological Station (KBS), Hickory Comers, MI; and the Ohio Agricultural Research and Development Center (OARDC), Wooster, OH. While the eastern transect included: University of Maryland Agricultural Experiment Station (UMD), Beltsville, MD; University of Connecticut Agricultural Experiment Station (U CONN), South Haven, CT; and the University of Vermont Horticultural Farm (UVM), South Burlington, VT. Experimental populations at Mountain Lake Biological Station (MLBS), Pembroke, VA served as the southernmost location for both transects. The latitude and longitude of each site is given in Table l. 22 cam .80? 582 .303. SS: cam .xofi 582 as; 228: «am .32 582 .802 92: Em .58 532 .802” m3: sum .80; £82 .308 ooe1000 0-1 1-2 2-3 3-4 4-5 >5 Size Category Distance in Miles Region 2 c 0.5 0.8 0.7 0.6 >1 2 E 0 5 9 o = a 0.4 ’5' o- 3 2 03 h an 0.2 0.1 0.0 1-10 10-100 100-1000 >1000 0-1 1-2 2-3 3-4 4-5 Size Category Distance in Miles Region 3 e f 0.7 0.8 0.6 L 0.7 0.5 _q 0.6 E. 04 E 0'5 3 a 0.4 0.3 __ E E 0.3 0.2 .— 0 2 0.1 — 0.1 0.0 .3 0.0 1-10 10-100 100-1000 >1000 0-1 1-2 2-3 3-4 4-5 Size Category Distance in Miles Figure 2 33 $3.: + .5: agziai as «B a A _axm - a3 A32 .3958 £258 mama: 28v 083 0:38» 68 05 mo «cote—3.8 =e ”a 36 v a on Sgomiwmm 86:2. .1. 2.. mm 5.8 .32 an «no A Smog .9 N 8m. am .32 2:. a; : 34 ”no: m 83% .9 _ Seam 2.. SN :32 1.... SN :62 N Saga .9 _ See: :5 2.3 coca—sack a... a «£23m SP... 3:5...3 u... a Eaten—.30 :2on .meofio. ooh: 05 new one cone—anon ecu Eon 955m 383 Bob macaw—anon no 3563 128:2 1m: we macaw—3&3 8333 33-82. 650883-52 ”a 933—. 34 353 S .5. e. :1 m _S_flal can t a: e. a: a Save 353 3.: a. a2 _ Snug cum ooze—=9..— 532 2:...— “528m Seam 8:83: :82 :2 um 3.. E 39 .385 32.5 5068 05 «o 2532 8:: com «.33 d me one noun—sac: can macaw—anon 3 353:. :88 gimme? ”n 03:. 35 mile and > 5 miles from the starting point (Fig. 2a). In contrast, populations in region 2 and 3 were characterized by populations that are <1 mile from the starting point and seem to be more random across the landscape (Figs. 2c & 2e). Field Experiment The results of the 1994 census indicated severe mortality of all three plant types. Overall plant mortality was above 45% at all of the sites with the exception of MLBS (40%) and Dunbar (2%) (Table 4). In general, the class that had the highest mortality was the susceptible class (50% or greater at all sites except Dunbar). Despite high mortality, there were 6 transitions at MLB S, 7 at the OARDC, 2 at both KBS and Dunbar, and 1 at UVM (Table 4). Two of the transitions at MLBS were the result of hand inoculation as was the one transition at UVM. The remainder of the transitions were due to spore deposition on healthy flowers by pollinators. The low number of transitions that occurred during 1994 prohibited statistical analyses. In 1995 transitions occurred at the five sites that were successfirlly censused (Dunbar, KBS, MLBS, UCONN and UVM) Specifically, there were 14 infections at Dunbar, 13 at UCONN, 12 at both KBS and MLBS and 10 at UVM (Table 5). In addition, mortality was greatly reduced to 30% or less at all sites except UMD and OARDC where it was 100% (Table 5). All of the infections that ocurred in 1994 survived overwinter (Table 5). A non-parametric categorical ANOVA on the 1995 data revealed no differences in the number of transitions between plots within sites so the data were pooled for comparisons among sites. Again, there were no differences among sites with respect to the number of transitions from healthy to diseased. 36 : 2 o S o N— v— . . f. . e no t _Soh. NNOOOOF' ,.. I. e e 3.53.35 2 m2 2 E.— o o o m: o o 3 G— N— o: 8.8.5 as»: o\coM o\om o\ooo _ o\oma A$00 ~ ..\..2 o\ov nu.) ZZOUD mmflz 0955 mg 5355 see. :8 =28 2: see 83 e “=an duo 8:82 as Feta "m 03:. _ 42 $8 :5: o NN $3 228: o t. see 92: e e: :2. 842 N E :3 ego N 8 $8 m5. N a: .\.N 555 .3— 5 8.8.5 :52 a .3.— 3 83:6 .2 £382 .\. ea 2...... as 23 =26... 2: Swag e salsa due 8:82 as 335 s. as: 37 The chi-square tests revealed a significant difl‘erence (X2(3,m)=4.274, P<0.04) between native and susceptible plants indicating that the susceptible plants were more likely to become infected than native plants. In addition, plants that were hand inoculated were more 15% more likely to become infected than plants that were subject to only natural vector transmission (Xza,m,=49.425, P<0.001). This phenomenon did not follow any detectable latitudinal gradient (X2(3,m,=0. 12, P