Heredity 58 (1987) 193—201 The Genetical Society of Great Britain Received 7 April 1986

Hierarchical population structure analysis of the milkweed , Tetraopes tetraophthalmus (Forster)

David E. McCauley and Department of General Biology, Vanderbilt Walter F. Eanes University, Nashville, TN 37235, U.S.A., and Department of Ecology and Evolution, SUNY at Stony Brook, Stony Brook, NY 11794, U.S.A. The population structure of the milkweed beetle Tetraopes tetraophthalmus (Forster) is tied intimately to the biology of its one host plant, Asciepias syriaca. The patchy distribution of the host plant and the limited dispersal of the beetle combine to organise the herbivore into numerous very localised populations. An analysis of hierarchical patterns of spatial variation in allozyme frequencies by means of Wright's F-statistics reveals that in some parts of the beetle's range this population structure can result in moderate genetic differentiation (F =0.03—0.06)among beetle populations inhabiting milkweed patches separated by only a few kilometres. Substantial differentiation occurs across the range of the species (F =0.154).A review of similar studies shows the values of F calculated for T. tetraophthalmus to be among the highest found for a flying . It is argued that the local differentiation results, in part, from a reduction in effective population sizes associated with the ephemeral nature of milkweed patches and that this population structure is favourable for the operation of Wright's Shifting Balance mode of evolution.

INTRODUCTION though rarely are both approaches applied together. From recent reviews that have compiled Thespatial organisation of local populations and F-statistic data for numerous species (i.e., concomitant patterns of gene flow are important Eanes and Koehn, 1978; Wright, 1978; Ayala, determinants of the potential for a species to 1982; Barrowclough, 1983) it can be seen that, not become genetically differentiated over its geo- surprisingly, there is a general agreement between graphic range. This is true whether the agent of the relative degree of population structure as differentiation is spatial variation in selection measured by F(S x T) and the perceived vagility pressures or the chance effects of genetic drift. The of the species in question. For example, among study of population structure and associated gene animal species, sedentary groups such as land flow in natural populations can take two snails and salamanders tend to have relatively high approaches, one ecological and one genetic (Slat- values of F(S x T), while most species of flying kin, 1985a). Ecological studies of the dispersal of birds and have rather low values. individuals or their gametes use various mark and A further refinement of the application of F- recapture schemes to attempt to measure gene flow statistics to the analysis of population structure directly, and can be used to make predictions about involves hierarchical analysis. Here the total geo- the population structure of a species. In contrast, graphic component of genetic variance is parti- surveys of geographic patterns of gene frequency tioned into subcomponents due to (a) variance variation that utilise electrophoretic markers can among closely spaced local populations likely to be used to make inferences about ecological par- be exchanging migrants and (b) variance among ameters such as average local population size and groups of populations defined on a greater geo- rates of migration (e.g., Larson eta!., 1984; Slatkin, graphic scale. In one such analysis Selander and 1985b). The genetic data is often summarised by Whitham (1983) argue that localised measures of the use of F-statistics as suggested by Wright F(S x T) are most likely to reflect the effects of (1978) for the analysis of population structure. population structure and gene flow. Hierarchical Both approaches have been applied to the study F-statistic analysis was developed by Wright as of the population structure in a diversity of species, part of his Shifting Balance theory of evolution in 194 D. E. McCAULEY AND W. F. EANES which some evolutionary transitions due to the Adults can be found on the foliage of the host joint effects of selection, drift, and migration are plant from June until August, depending on loca- possible in highly structured populations but can tion. Adult were collected from individual not occur in large panmictic units (Wright, 1977; milkweed patches according to a hierarchical col- 1978). lecting scheme. Milkweed patches were defined as The milkweed beetle Tetraopes tetraophthalmus nearly discrete aggregations of milkweed stalks (Forster) (Coleoptera: Cerambycidae) has been occupying from about 025 to 10 ha. All beetles the focus of several ecological studies of dispersal collected from the same patch are considered to and gene flow (McCauley et a!., 1981; McCauley, belong to the same population or random breeding 1983). These studies have concluded that the unit as evidenced by dispersal studies (McCauley, limited dispersal of the insect, combined with the 1983). Some number of populations (4-19) were patchy, island-like distribution of its host plant, sampled from each of 5 localities called regions. the common milkweed Asciepias syriaca, organise Each region consisted of an area of less than 30 x the species into thousands of local populations 30 km. Regions were defined in southern New across its range in eastern North America. Thus, Hampshire (NH), southwest Virginia (VA), middle the species would appear to have the potential for Tennessee (TN), and northern Illinois (ILL). Col- a high degree of genetic differentiation across its lections were made in New Hampshire in 1983, geographic range and should exhibit genetic Illinois in 1984, and Tennessee in 1985. Most Vir- differences between populations separated by as ginia populations were sampled in 1984, though a little as a few kilometres. The one limited genetic few localities were sampled in 1983 or 1985. All study of this species (Eanes et aL, 1977) used data presented for a particular population are polymorphic allozyme markers to document allele derived from a single generation. Because of the frequency heterogeneity among seven collections univoltine nature of this species only two gener- of T tetraophthalmus taken from individual milk- ations separate the 1983 and 1985 collections. In weed patches located in Suffolk Co., N.Y. addition, data taken from collections made on Here, we report on gene frequency variation at Long Island, New York (NY) in 1975 and three polymorphic loci derived from a more sys- described by Eanes et aL (1977) were used to tematic and hierarchical collection of beetles taken constitute a fifth region. Fig. 1 presents these across a broad geographic area. The collections regional locations against the range of the host were designed to examine whether the degree of plant as described by Chemsak (1963). local population structure documented by Eanes Beetles were stored at —70°C until they could eta!. (1977) in New York is representative of other be prepared for starch gel electrophoresis. Eanes localised regions across the species' range, and et a!. (1977) present data for leucine aminopep- also to estimate the total geographic component tidase (LAP) and phosphoglucomutase (PGM), of genetic variance. Groups of milkweed patches the only polymorphic loci they found in their were sampled on an individual patch basis within original survey. In that survey amylase, glucose-6- four 30 x 30 km regions, the regions being spaced phosphate dehydrogenase, glutamate oxaloacetate across a major portion of the range of the species. transaminase, isocitrate dehydrogenase, lactate The data, when combined with that reported by dehydrogenase, peptidase and phosphogluco- Eanes et aL (1977) are used in a hierarchical F- isomerase were resolved but found to be monomor- statistic analysis that partitions the total geo- phic (Eanes, unpubi.). Similar techniques were graphic component of genetic variance into com- followed for the resolution of LAP and PGM in ponents due to among patch variance within the the more recent samples with the following excep- local regions and a component among regions. The tions. Both enzymes could be resolved with an results are discussed in the context of the known electrode buffer described as "salamander B" by ecology of the beetle and its host plant, as well as Werth (1985) and a gel buffer consisting of a 1:9 in the context of a survey of F-statistics calculated dilution of the electrode buffer. B-mercaptoethanol for other species of insects. was added to the grinding buffer to stabilise the PGM electromorphs. A more recent enzyme survey was able to resolve two additional enzymes, malic MATERIALSAND METHODS enzyme and hexokinase. Hexokinase (HK) proved to be polymorphic when resolved with a tris-citrate Tetraopestetraophthalmus is a univoltine beetle electrode buffer (pH 8.0). The discovery of the HK whose adults and larvae feed almost exclusively polymorphism came after the on set of the Virginia on the common milkweed, Asc!epias syriaca. collecting. As a consequence not all Virginia popu- POPULATION STRUCTURE OF MILKWEED BEETLE 195

Figure1 The location of five regions from which adult Tetraopes retraophthalmus were collected. Numbers refer to the number of individual milkweed patches sampled in each region. Stippled area defines the range of the host plant, Asciepias syriaca as described by Chemsak (1963). lations were assayed for HK. Samples from various RESULTS regions were occasionally run side by side to stan- dardise allelic nomenclature. Resultsare presented as allele frequencies and Gene frequency variation was quantified by sample sizes for LAP, PGM and HK in tables 1-3 hierarchical F-statistics as suggested for the study respectively. All populations sampled expressed of population structure by Wright (1978). See Weir two electrophoretic alleles for LAP (named F and and Cockerham (1984) and Slatkin (1985a) for S), though there was considerable variation in recent discussions of the application of F-statistics frequency from region to region. PGM expressed to population structure analysis. Geographic vari- a total of three alleles (FM,S), though only NY ation in gene frequencies was computed among populations were strongly polymorphic for all populations within regions, F(P x R), among three alleles. Only the M and S alleles were found regions relative to the total F(R x T), and finally in NH and F and M in VA. In ILL the populations among populations relative to the total, F(P x T). were nearly fixed for the M allele but both F and Specific computations followed the methods of S appeared occasionally. Tennessee populations Weir and Cockerham (1984). That reference also were monomorphic for the M allele. Hexokinase describes methods used when more than two alleles expressed two common alleles (1 5) in NH, TN were found at a given locus and methods used to and VA, with the S variant being quite rare in ILL. combine information across loci. Occasionally A statistical comparison of the fit of the values of F(PxR) were computed to be very observed genotypic ratios to Hardy-Weinberg slightly negative owing to sample size corrections expectations was made for each locus for each and were set to zero. population by use of G-Goodness of Fit tests 196 D. E. McCAULEY AND W. F. EANES

Table IAllele frequencies and sample sizes for LAP gote deficiency, 5aheterozygote excess. Ignoring statistical significance, 49 of the total expressed a n n Locality 'F Locality PF heterozygote deficiency and 49 a heterozygote NH-i 84 044 VA-6 61 047 excess. Thus, there are no consistent deviations 2 72 038 7 40 038 from Hardy-Weinberg in our data set that would 3 71 043 8 69 053 indicate pervasive substructuring within our 4 74 037 9 27 046 NY-i 316 045 10 50 036 sample units. 2 199 035 ii 52 043 Values of F(P x R) calculated for each locus 3 287 0•49 12 43 047 for each region are presented in table 4. The value 4 321 031 13 84 0•42 presented for PGM in the NY sample is that 5 309 0•35 14 38 0'43 6 430 055 15 97 037 averaged across the 3 alleles by the methods of 7 259 044 16 70 040 Weir and Cockerham (1984). No value is calcu- ILL-i 51 021 17 57 037 lated for PGM in TN because the locus is 2 91 020 18 60 048 monomorphic in that region. In order to test 3 59 0•i5 19 70 044 whether estimates of F(P x R) greater than zero 4 91 016 TN-i 38 068 5 50 0•17 2 102 082 truly reflect heterogeneity in gene frequencies 6 60 0•i7 3 44 069 among populations, the association of allelic 7 69 O'i7 4 60 0•67 frequencies with populations was tested for each VA-i 124 039 5 50 063 locus in each region by G contingency analysis 2 55 030 6 46 082 (Sokal and Rohlf, 1981) with the results included 3 39 037 7 20 082 4 52 037 8 60 0'8i in table 4. Table 4 also provides an estimate of 5 69 033 9 64 069 F(P x R) calculated for each region by combining information across loci. Information combined across loci should provide the most accurate reflec- (Sokal and Rohif, 1981). (PGM in TN and PGM tion of the true population structure. and HK in ILL did not express sufficient poly- While it can be seen that there are no consistent morphism to warrant such tests.) Of 98 tests, 9 locus-specific trends, there do appear to be showed a deviation from the expected Hardy- between-region differences in local population Weinberg distribution that was significant at the structure. For example, while there is no measur- 005 or lower level. Of these, 4 showed a heterozy- able genetic differentiation among the ILL popula-

Table 2 Allele frequencies and sample sizes for POM

Locality n 'F qM rs Locality n PF q r NH-I 96 — 083 017 VA-S 72 019 081 — 2 70 — 085 015 6 65 030 070 — 3 75 — 075 025 7 40 027 073 — 4 60 — 0•59 041 8 44 018 082 — NY-i 313 0i9 0•48 0•33 10 58 0'18 082 — 2 211 0•06 068 026 11 57 024 076 — 3 287 018 061 021 13 83 028 072 — 4 342 O'il 066 0•23 14 40 028 0'72 — 5 298 001 0•65 033 15 97 018 082 — 6 435 0.03 054 043 16 70 0•15 085 — 7 235 007 064 029 17 58 014 086 — ILL-i 61 0•02 098 — 18 57 027 073 — 2 91 0•03 096 0•01 19 50 022 078 — 3 60 — 099 0•Oi TN-I 24 — 100 — 4 51 003 0•96 0•01 2 80 — 100 — 5 50 00i 099 — 3 20 — 100 — 6 60 — 098 002 4 36 — 1•00 — 7 71 0•02 0•98 — 5 40 — 1'OO — VA-I 64 0•23 077 — 6 40 — 100 — 2 55 0-18 082 — 7 60 — 100 — 3 39 0•26 074 — 8 40 — 100 — 4 60 020 080 — 9 70 — 1•00 — POPULATION STRUCTURE OF MILKWEED BEETLE 197

Table 3 Allele frequencies and sample sizes for HK the gene frequency estimate for each population by the sqllre root of the product of the regional n n Locality PF Locality F means (-.Jp).Afterperforming this transformation NH-i 57 085 VA-7 59 0•76 on the data presented in tables 1—3, the Scheffe— 2 61 057 8 68 0'74 Box test was performed across regions separately 3 35 081 10 51 0•74 for each locus. There was highly significant 4 26 08i ii 63 080 heterogeneity among regions in the standardised ILL-i 51 098 14 53 069 2 87 0•97 15 46 076 variance in LAP (pO.IO). We note 5 50 092 19 49 071 that the statistical power associated with the PGM 6 61 096 TN-2 95 044 and HK tests is less than that associated with LAP 7 70 096 3 43 051 VA-i 53 078 4 33 039 owing to a smaller number of samples. In sum, 2 42 0•75 5 45 052 there is partial support for our view that regions 4 29 086 6 41 067 vary in their local population structure. 5 71 074 8 40 045 9 58 066 Calculations of the degree of among region 6 60 072 differentiation, F(R x T), are presented for each locus and combined across loci in table 5. Since the New York samples were taken 8-10 generations Table 4 F(Px R) values as calculated for each locus at each prior to the more recent collections, two calcula- region as well as average values for each region tions are presented, one with and one without the New York data. They are in close agreement. The Region LAP PGM HK Average total geographic differentiation, F(P x T), com- bined across loci and including the New York data VA 0.005* 0.006* 0.000 0003 TN O.023*** 0.036*** 0031 is O154 ILL 0000 0001 0001 0000 0.064*** Ø.Ø95*** 0064 NH 0000 Table 5 F(R x T) values as calculated for each locus and the NY Ø.Q33*** 0.031*** 0•031 average across loci Note: Asterisks represent significance levels in G contingency LAP PGM HK tests of the independence of allele frequencies among popula- F(R x T) average tions within regions. * =p <005, ** =p <001, 'K' =p <0001 With NY 0098 0165 0164 0147 Without NY 0184 0159 0164 0172 tions, NH, NY and TN populations exhibit considerable differentiation. While Weir and Cockerham (1984) present methods for estimating DISCUSSION variance in F, we are not aware of any standard statistical methods for comparing F statistics. In The results of the earlier dispersal and life-history order to test for heterogeneity among regions in studies and the present genetic study can be com- F(Px R) values, we performed the following bined to form a comprehensive view of population analysis based on the Scheffe-Box test suggested structure in Tetraopes tetraophthalmus. Mark and by Sokal and Rohlf (1981) as a test for the recapture experiments (McCauley et a!., 1981; homogeneity of variances. This test includes com- Lawrence, 1982; McCauley, 1983) have shown that putations that require the individual variates, in most adults do not move very far during their this case allele frequencies. F(ST) is often con- reproductive lives despite having the ability to fly. sidered as the variance in gene frequencies standar- Most individuals stay in the milkweed patch in dised by the product of the mean gene frequencies which they emerged. Generation-wide gene flow (S/pq). Thus, to make a comparison of allele distance estimates derived from these dispersal frequency variances using the Scheffe-Box test studies become particularly small when net dis- relevant to a comparison of F(P x R) estimates placement distances of recaptured adults are we had to standardise the individual gene frequen- weighted by the age-specific reproductive potential cies for each population so that the variance of of the cohort at the time when each recapture was the gene frequencies within each region became made (Endler, 1979; McCauley, 1983). equal to their respective F(P x R) values calcu- In contrast, a series of laboratory tethering lated above. This was accomplished by dividing experiments and field studies (Davis, 1981; 1984) 198 D. E. McCAULEY AND W. F. EANES has shown that adults of this species are phy- lation sizes must be moderately small. Mature siologically capable of long distance flights and milkweed patches can contain several thousand might occasionally disperse several kilometres adult beetles. Average effective population sizes from their point of birth. Thus, the combined dis- are undoubtedly considerably less than this, persal studies suggest that the beetles found in a however, due to the ephemeral nature of milkweed given milkweed patch do represent a breeding unit patches. Asciepias syriaca is a disturbed habitat moderately isolated from similar populations in species primarily occupying old fields and road- nearby patches, yet patches spaced several sides. It must be subjected to frequent local extinc- kilometres apart are theoretically capable of tions and recolonisations. In the course of our exchanging rare migrants. Thus, the population studies we have seen several milkweed patches go structure viewed over a few tens of square extinct due to either human disturbance or natural kilometres is similar to that studied in "stepping successional changes. We therefore speculate that stone" models of population structure (Kimura it is unlikely that a beetle population occupying a and Weiss, 1964) in which most individuals breed particular milkweed patch would persist for more in their natal population, a small fraction move to than a maximum of 50-100 generations. In that an adjacent population, and an even smaller frac- case the genetic characteristics of a beetle popula- tion are capable of moving to any population in tion could be partially influenced by chance events the array. This view of the population structure of and associated founder effect at population initiali- the species is supported by the present within- sation and small population sizes in the gener- region F-statistic analysis which demonstrates that ations immediately after the population was gene flow must be sufficiently limited at this spatial founded. Slatkin (1977) and Maruyama and scale so as to permit some genetic differentiation Kimura (1980) derive mathematical models that of local populations. relate various modes of colonisation to the The degree of local differentiation does not accumulation of among population gene frequency appear to be uniform across the species' range, variance and show that such variance can be gener- however, and differentiation is non-existent in ated by some patterns of local extinction and rec- Illinois. This could be due to variation in the patchy olonisation. While the effects of extinction and distribution of the host plant. It was our observa- recolonisation on population structure can be com- tion that milkweed was much more abundant in plex, under some circumstances episodes of extinc- the area of Illinois where we collected (western tion and recolonisation are thought to be favour- suburbs of Chicago) and there were fewer obvious able for the operation of Wright's "Shifting Bal- barriers to gene flow compared with the other four ance" mode of evolution (Wright, 1977; Barton collecting regions. Milkweed patches appeared and Charlesworth, 1984; Lande, 1985). It is our more widely dispersed in the areas of New York, view that T tetraophthalmus displays a population New Hampshire and Tennessee where collections structure very similar to that described by Wright were made and there were more intervening pat- (1977, 1978) as favourable for evolution by the ches of forest. In southwest Virginia milkweed Shifting Balance process. patches are rather closely spaced in the agricul- It would be very useful, therfore, to document turally developed valleys but these valleys are sep- the initial colonisation and early demographic his- arated by heavily wooded ridges. An intriguing tory of a local beetle population. Several factors possibility is suggested by the work of Davis critical to the genetic consequences of population (1986). He found that in laboratory tethered-flight foundation would include number, sex, and origin experiments females collected from regions where of initial founders, rate of population growth in milkweed is relatively rare display greater average the generations immediately succeeding colonisa- flight potential than do females collected from tion, and the rate of arrival of new migrants follow- areas where milkweed is more common. Thus, ing initial colonisation. Many of these factors must regional variation in local population structure be determined, in turn, by the demography of the could be a consequence of variation in habitat host plant; specifically the dispersal distances structure, variation in the intrinsic vagility of the travelled by milkweed propagules colonising new beetles, or an interaction of the two factors. habitat, the rate of expansion of young milkweed Genetic differentiation of populations could be patches, and the rate of patch extinction. due to spatial variation in selection pressures, While values of F can theoretically range from genetic drift, or the joint effects of both processes. 0 to 1, we suggest that the F(P x R) values of In order for genetic drift to influence among popu- 003—006 measured over a very limited fraction of lation variance in gene frequencies, effective popu- the species' total range and an F(P x T) greater POPULATION STRUCTURE OF MILKWEED BEETLE 199 than 0.15 represent a surprisingly high degree of population structure that does combine ecological population structure considering that this is an and genetic information). Even fewer studies repli- abundant, flying insect. Wright (1969, 1977, 1978) cated local collections in a hierarchical fashion. suggests that population structures resulting in Table 6 presents some estimates of F(S x T) calcu- F(S x T) values as small as O05 can still favour lated from samples taken across a major portion the operation of Shifting Balance evolution. He of the respective species' ranges. These estimates notes that when a species is distributed into many would be comparable with our F(P x T). In some local populations, even a small variance in gene cases the data had been used by the authors to frequency will generate occasional populations calculate genetic identities or distances rather than whose gene frequencies represent extreme outliers. F(S x T) but an estimate of F could be calculated In order to contrast the genetic population from tabulated allelic frequencies and sample sizes structure of T tetraophthalmus to other sexually provided in those references. The list is not inten- reproducing, flying insects, we surveyed the recent ded to be an exhaustive survey but should be literature for comparable studies. We found few representative of recent work on insects. Studies that attempted to match the spatial scale of the in which the taxonomic rank of geographic popula- collecting regime to the known dispersal biology tions was in question were not included. It can be of the insect (but see work done on checkerspot seen that, when considered on a broad geographic butterflies such as that reported in McKechnie et scale, T. tetraophthalmus is one of the more a!., 1975, for an example of an approach to insect geographically structured insect species. In fact,

Table 6Values of Wright's F(st) estimated for 29 insect species

Species F(st) Reference

Diptera Musca autumnalis 0023 Bryant et aL (1981) Rhagoletis completa (native range) 0007 Berlocher (1984) R. completa (introduced range) 0054 Berlocher (1984) Prosimuliumfuscum 0003 Snyder and Linton (1984) P. mixtum 0096 Snyder and Linton (1984) Drosophilawillistoni 0022 cited in Ayala (1982) D. pseudoobscura 0028 cited in Ayala (1982) D. equinoxialis 0029 cited in Ayala (1982) D. obscura 0067 cited in Ayala (1982) D. pavani 0126 cited in Ayala (1982) Lepidoptera Danausplexippus 0009 Eanes and Koehn (1978) Spodoptera frugiperda 0084 Pashley et aL (1985) S. exempta 0006 cited in Pashley et aL (1985) Alabama argillacea 0007 cited in Pashley et aL (1985) Pierisrapae 0014 cited in Pashley et aL (1985) Anticarsia gemmatalis 0021 cited in Pashley et a!. (1985) Heliothis virescens 0048 cited in Pashley et a!. (1985) Cydia pomonella 0066 cited in Pashley et a!. (1985) Euphydryas chalcedona 0090 cited in Pashley et a!. (1985) E. editha 0118 cited in Pashley et a!. (1985) Coleoptera Dendroctonus ponderosae 0030 Stock et a!. (1984) D.frontalis 0068 Anderson et aL (1979) Ips calligraphus 0011 Anderson et aL (1983) Diabroticabarberi 0098 McDonald et a!. (1985) Leptinotarsadecemlineata 0068 Jacobson and Hsiao (1983) Pissodes strobi 0098 Phillips and Lanier (1985) Tetraopestetraophthalmus 0154 this study Hemiptera Limnoporous canalicu!atus 0082 Zera (1981) Hymenoptera Rhytidoponera confusa 0294 Ward (1980) R. chalybaea 0380 Ward (1980) 200 D. E. McCAULEY AND W. F. EANES

measures of geographic differentiation taken on a ENDLER, i. A. 1979. Gene flow and life history patterns. small spatial scale within parts of the range of Genetics, 93, 263—284. JACOBSON, J. W., AND HSIAO, T. H. 1983. Isozyme variation milkweed beetles are comparable with the degree between geographic populations of the colorado potato of differentiation found across the entire range of beetle, Leptinotarsa decemlineata (Coleoptera: Chry- many other insect species. somelidae). Ann. Entomol, Soc. Am., 76, 162—166. KIMURA, M. AND WEISS, G. H. 1964. The stepping stone model Acknowledgements We thank W. S. Lawrence for providing of population structure and the decrease of genetic correla- us with the New Hampshire samples. E. C. Lawson, L. M. tion with distance. Genetics, 49, 561—576. Reilly and R. A. Schedler provided invaluable field and labora- LANDE, R. 1985. The fixation of chromosomal rearrangements tory assistance. Much of the work was done at the Mt. Lake in a subdivided population with local extinction and Biological Station and we thank its directors and staff for their colonization. Heredity, 54, 323—332. logistical support. This project was supported financially by LARSON, A., WAKE, D. B. AND YANEV, K. 1'. 1984. Measuring grant BSR 83-20339 from the National Science Foundation and gene flow among populations having high levels of genetic an award from the Vanderbilt University Research Council, fragmentation. Genetics,106, 293—398. both awards to DEM. Contribution no. 609 in Ecology and LAWRENCE, W. S. 1982. Sexual dimorphism in between and Evolution from the State University of New York at Stony within patch movements of a monophagous insect: Brook. Tetraopes (Coleoptera: Cerambycidae). Oecologia, 53, 245—250. MARUYAMA, T. AND KIMURA, M. 1980. Genetic variability REFERENCES and effective population size when local extinction and recolonization are frequent. Proc. Nat. Acad. Sci., 77, 67 10- ANDERSON, W.W., BERISFORD, C. W. AND KIMMICH, R. H. 67 14. 1979. Genetic differences among five populations of the MCCAULEY, D. E. 1983. Gene flow distances in natural popula- southern pine beetle. Ann. EntomoL Soc. Am., 72, 323-327. tions of Tetraopes tetraophthalmus. Evolution, 37, 1239- ANDERSON, W. W., BERISFORD, C. W., TURNBOW, R. H. AND 1246. BROWN C. .i., 1983. Genetic differences among populations MCCAULEY, D. E., orr, J. R., STINE, A. AND MCGRATH, 5. 1981. of the black turpentine beetle, Dendroctonus terebrans, Limited dispersal and its effect on population structure in and an engraver beetle, Ips calligraphus (Coleoptera: the milkweed beetle Tetraopes tetraophthalmus. Oecologia, Scolytidae). Ann. Entomol. Soc. Am., 76,896—902. 51,145-150. AYALA, F. J. 1982. The genetic structure of species. Milkman, MCDONALD, IC., KRYSAN, J. L. AND JOHNSON, 0. A. 1985. R. (ed.) In Perspectives on Evolution. Sinauer, Sunderland, Genetic variation within and among geographic popula- MA, pp. 60-82. tions of Diabrotica barberi (Coleoptera: Chrysomelidae). BARROWCLOUGH, G. F. 1983. Biochemical studies of micro- Ann. Entomol. Soc. Am., 78, 271-278. evolutionary processes. Brush, A. H. and Clark, G. A. MCKECHNIE, S. W., P. R. EHRLICH AND R. R. WHITE. 1975. (eds.), In Perspectives in Ornithology,CambridgeUniv. Population genetics of Euphydrias butterflies. I. Genetic Press, Cambridge, pp. 223-261. variation and the neutrality hypothesis. Genetics, 81, 571- BARTON, N. H. AND CHARLESWORTH, B. 1984. Genetic revol- 594. utions, founder effects, and speciation. Ann. Rev. Ecol. PASHLEY,D. P., JOHNSON, S. J. AND SPARKS, A. N. 1985. Syst.,15, 133—164. Genetic population structure of migratory moths: the fall BERLOCHER, S. H. 1984. Genetic changes coinciding with the armyworm (Lepidoptera: Noctuidae). Ann. Entomol. Soc. colonization of California by the walnut husk fly, Am., 78, 756-762. Rhagoletis completa. Evolution, 38, 906-918. PHILLIPS, T. W. AND LANIER, G. N., 1985. Genetic divergence BRYANT, E. H., VAN DIJK, H. AND VAN DELDEN, w. 1981. among populations of the white pine weevil, Pissodes strobi Genetic variability of the face fly, Musca autumnalis De (Coleoptera: Curculionidae). Ann. EntomoL Soc. Am., 78, Geer, in relation to a population bottleneck, Evolution, 35, 744-750. 872—881. SELANDER, R. K. AND WHIYrAM, T. S. 1983. Protein poly- CHEMSAK, .i.A.1963. and bionomics of the genus morphisms and the genetic structure of populations. Nei, Tetraopes (Coleoptera: Cerambycidae). Univ.Calf Phb. M. and Koehn, R. K. (eds) In Evolution of Genes and in EntomoL, 30,1-90. Proteins,Sinauer,Sunderland, MA, pp. 89-114. DAVIS,M. A. 1981.The flight capacity of dispersing milkweed SLATKIN, M. 1977. Gene flow and genetic drift in a species beetles, Tetraopes tetraophthalmus. Ann. Entomot Soc. subject to frequent local extinctions. Theor. Pop. Biol., 12, Amer., 74,385—386. 253—262. DAVIS, M. A. 1984. The flight and migration ecology of the red SLATKIN, M. 1985a. Gene flow in natural populations. Ann. milkweed beetle (Tetraopes tetraophthalmus). Ecology, 65, Rev. EcoL Syst., 16, 393—430. 230—234. SLATKIN, M. 1985b. Rare alleles as indicators of gene flow. DAVIS, M. A. 1986. Geographic patterns in the flight ability of Evolution, 39, 53-65. a monophagous beetle Oecologia, 69, 407-412. SNYDER, T. P. AND LINTON, M. C. 1984. Population structure EANES, W. F., GAFFNEY, P.M., KOEHN, R. K. AND SIMON, C. M. in black flies: allozymic and morphological estimates for 1977. A study of sexual selection in natural populations Prosimuliummixtumand P. fuscum(Diptera:Simuliidae). of the milkweed beetle Tetraopes tetraophthalmus. Chris- Evolution, 38, 942-956. tiansen, F. B. and Fenchel, T. M. (eds.), In Measuring SOKAL, R. R. AND ROHLF, F. J. 1981. Biometry. W. H. Freeman, Selection in Natural Populations', Springer, New York, pp. San Francisco. 49-64. STOCK, M. W., AMMAN, 0. D. AND HIGHBY, P. K. 1984. Genetic EANES, W. F. AND KOEHN, R. K. 1978. An analysis of genetic variation among mountain pine beetle (Dendroctonus pon- structure in the monarch butterfly, Danaus plexippus L. derosae) (Coleoptera: Scolytidae) populations from seven Evolution, 32, 784—797. western states. Ann. EntomoL Soc. Am., 77, 760—764. POPULATION STRUCTURE OF MILKWEED BEETLE 201

WARD, p. S. 1980. Genetic variation and population differenti- WRIGHT, S. 1977. Evolution and the Genetics of Populations, ation in the Rhytidoponera impressa group, a species com- Vol. 3. Experimental Results and Evolutionary Deductions, plex of Ponerine ants (Hymenoptera: Formicidae). Evol- Univ. Chicago Press, Chicago. ution,34,1060-1076. WRIGHT, S. 1978. Evolution and the Genetics of Populations Vol. WEIR, B. S. AND COCKERHAM, c. c. 1984. Estimating F-statis- 4. Variability Within and Among Natural Populations, tics for the analysis of population structure. Evolution, 38, Univ. Chicago Press, Chicago. 1358—1370. ZERA, A. .i. 1981. Genetic structure of two species of water- WERTH, C. R. 1985. Implementing an isozyme laboratory at a striders (Gerridae: Hemiptera) with differing degrees of field station. Va. I Sci., 36,53-76. winglessness. Evolution, 35, 218—225. WRIGHT,S.1969.Evolution and the Genetics of Populations, Vol. 2. The Theory of Gene Frequencies, Univ. Chicago Press, Chicago.