Recent Evolutionary History of American Onchocerca voZvuZus, Based on Analysis of a Tandemly Repeated DNA Sequence Family

Peter A. Zimmerman, * $ Charles R. Katholi,? Michael C. Wooten,§ Naomi Lang-Unnasch, * and Thomas R. Unnasch* *Division of Geographic Medicine, Department of Medicine, and TDepartment of Biostatistics and Biomathematics, University of Alabama at Birmingham; $Department of Biology, Case Western Reserve University; and $Department of Zoology, Auburn University

Polymerase chain reaction (PCR) products were characterized for a repeated sequence family (designated “O- 150”) of the human filarial parasite Onchocerca volvulus. In phylogenetic inferences, the 0- 150 sequences clustered into closely related groups, suggesting that concerted evolution maintains sequence homology in this family. Using a novel mathematical model based on a nested application of an analysis of variance, we demonstrated that African rainforest and Savannah strain parasite populations are significantly different. In contrast, parasites collected in the New World are indistinguishable from African Savannah strains of 0. volvulus. This finding supports the hypothesis that onchocerciasis was recently introduced into the New World, possibly as a result of the slave trade.

Introduction

Onchocerca volvulus is the cause of human oncho- cerciasis, which is the second leading cause of infectious blindness in humans (Thylefors 1978 ) . Onchocerca vol- vulus is endemic throughout subsaharan Africa and is also found in several foci in South and Central America. Entomological, immunological, and genetic studies have suggested that different strains of the parasite are en- demic in the rainforest and Savannah bioclimes of west- ern Africa (Duke et al. 1966; Lobos and Weiss 1985; Flockhart et al. 1986; Erttman et al. 1987, 1990) and that the Savannah strain of the parasite causes a more virulent form of the disease (Dadzie et al. 1989; Remme et al. 1989).

The origin of American onchocerciasis is unclear (Nelson 199 1). Biochemical differences in the parasite populations in the two hemispheres (Yarzabal et al. 1983), together with the inability of American parasites to develop in African blackflies (Duke 198 1 ), suggest that American and African 0. volvulus populations have been isolated for a long period of time. In contrast, inefficient transmission of American 0. volvulus by American Simulium species ( De Leon and Duke 1966 ) , together with cytotaxonomic similarities between Amer- ican and African parasites (Hirai et al. 1987) supports the argument that parasite populations on the two con- tinents share a recent common ancestor.

Key words: Onchocerca volvulus, African river blindness, tandem repeat family, cluster algorithm, phenetics.

Address for correspondence and reprints: Thomas R. Unnasch, Division of Geographic Medicine, BBRB 206, University of Alabama at Birmingham, Birmingham, Alabama 35294.

Mol. Biol. Evol. 11(3):384-392. 1994. 0 1994 by The University of Chicago. All rights reserved. 0737-4038/94/l 103-0006$02.00

Recently genus-, species-, and strain-specific DNA sequences present in the genome of 0. volvulus have been identified (Erttmann et al. 1987, 1990; Shah et al. 1987; Harnett et al. 1989; Meredith et al. 1989). These sequences are all members of a variable tandemly re- peated DNA sequence family of - 150 bp, designated “O- 150.”

Rapidly evolving sequences can be used to infer evolutionary relationships among closely related organ- isms. However, repeated sequences have generally not been used for this purpose, for two reasons. First, in order to ensure that an accurate representation of the repeat population is obtained and that the results are not biased by stochastic errors, it is necessary to examine a relatively large number of individual examples of a given repeat population (Pascale et al. 1993 ) . Second, only recently have mathematical and statistical models necessary for the analysis of these unique data sets been widely considered (Sawyer 1989; Hein 1993).

In the work reported below, the O-l 50 family has been used to deduce relationships among different 0. volvulus populations. Application of a novel mathe- matical model to sequence data derived from the 0- 150 family has been used to examine the relationship be- tween Old and New World populations of 0. volvulus.

Material and Methods Biological Material

Onchocerca volvulus were obtained by nodulectomy from infected individuals and were freed from sur- rounding host tissue by collagenase digestion (Engel- brecht and Schulz-Key 1984). Samples generally con- sisted of the four to six adult female parasites, associated males, and microfilariae present in an individual oncho-

Origin of American Onchocerciasis 385

cercomata. The one exception was the Brazilian isolate, testing, which uses permutations of the data under the where an individual adult female was obtained. Biogeo- null hypothesis to calculate significance levels (Petrondas graphic, epidemiological, and DNA probe-based tech- and Gabriel 1983; Manley 199 1) . To apply the random- niques were used to classify the African parasites as be- ization procedure to the sequence data, the ratio of sums longing to either the forest or Savannah strains. of squares described previously was first calculated for Specimens of 0. ochengi, a closely related cattle parasite, each identified cluster. This test statistic will be denoted were obtained for use as an outgroup. by “F” by way of analogy with the usual ANOVA

methods, but we emphasize that this statistic probably Sequencing Procedures does not have an F distribution. The sequences were

Genomic DNA preparation and polymerase chain reaction (PCR) amplification of the 0- 150 repeat family were performed as described elsewhere (Meredith et al. 199 1). Synthetic Hind111 sites were incorporated into the PCR primers to facilitate cloning. This restriction site was not present within known examples of O-l 50, nor were motifs present which are easily mutable into Hind111 sites. PCR products were digested with HindIII, and O-l 50 monomers were purified by electroelution after separation on a 10% polyacrylamide gel. The pu- rified DNA was ligated into M 13mp19, and the DNA sequence of individual clones was determined by stan- dard methods (Sanger et al. 1977).

Overview of the Statistical Analysis

In order to use the 0- 150 sequence data to study the relationship among the 0. volvuZus populations, it was necessary to determine whether the observed clusters were meaningful and whether there were significant dif- ferences in the clusters found in the parasite populations examined. When the data analyzed are numeric, these questions could be addressed by analysis of variance (ANOVA). Since DNA sequences are not numeric, this technique was not directly applicable. However, the ANOVA approach could be extended to develop a method that could be applied to sequence data. To ac- complish this, it was noted that the F-tests associated with an ANOVA are found by considering ratios of sums of squares of deviations from appropriate means. To apply this method to DNA sequence data, a measure of the common behavior of the sequences in a cluster was thus required. The cluster consensus sequence was cho- sen for this purpose. Next, a measure of distance of one sequence from another was required. The obvious choice for this measure was the Hamming distance, i.e., a count of the number of places where the two sequences differ. In light of these analogues of the mean and the distance measure, it was possible to find ratios of sums of squares of deviations in a manner analogous to the usual ANOVA.

Classical ANOVA relies on distributional assump- tions concerning the observed data. Those assumptions were unjustified when applied to DNA sequence data, and so an alternative method for deciding significance was required. The alternative chosen was randomization

then permuted among the clusters, and a set {F, , F2, . . . , FN} of values of the test statistic were generated. These were computed in order to calculate the distri- bution of the test statistic under the null hypothesis. Fi- nally, each of these Fi was compared with the original F. The number of times that these Fi equaled or exceeded F was counted, and the ratio of that number to the total number of permutations used was taken as the proba- bility of observing a test statistic as large as or larger than the one obtained by visual clustering under the null hy- pothesis, i.e., the P-value for the test.

When the data set was small, it was possible to ex- amine all possible rearrangements and thus obtain an exact measure of the significance level. However, the number of possible arrangements rapidly increases as the size of the data set grows. In such cases, the signifi- cance level was estimated by considering a large random subsample of the possible assignments of sequences to clusters. A detailed description of this approach may be found in Manley ( 199 1).

Formulas Used to Test Significance of Clusters

In the formulas provided below, weighted sums of squares are used because of the unequal numbers of se- quences in the clusters. Let there be k groups with each group having nk observations. Let Yi,j be thejth sequence in the ith group, let Yi,* denote the consensus sequence of the ith group, let Y*,* the grand consensus sequence and let n = nl + n2 + . . . + nk be the total number of observations. Let the Hamming distance operator d( A, B) be defined as the count of the number of places that two sequences A and B differ. Finally, define each of the following sums,

T= i rti[d(Yi,*, Y*,*)2] (1) i=l

and

k nj

E = C C [d(K,j, Yi,*)2] -

i=l j=l

(2)

The analysis table was produced using equation ( 1) for the among-groups sum of squared distance and equation

386 Zimmerman et al.

(2) for the within-groups sum of squared distance. The associated test statistic was then

F = P+m-1 ww(~-w *

Formulas for Testing Isolate Differences

Let Yi,j,k be the kth sequence in the jth cluster of the ith isolate, where i = 1, . . . , Z; j = 1, . . . , ni; k = 1 the jth

* * 7 ni, j. Let Yi, j, + be the consensus sequence of cluster in the ith isolate, Yi,*, * the consensus se-

quence of the ith isolate, and let Y*, *, * be the grand con- sensus sequence. Let d( A, B) be the Hamming distance between the two sequences A and B and define each of the following sums,

T=;: i=l

E=;: i=l

s=;: i=l

Ni[d(K,*,*, y*,*,*>21

5 [dtYi,j,*9 yi,*,*121 j=l

5 5 [d(Yijk Y”* 9.9 I,),

j=l k=l

where

n1

Ni = 2 nij j=l

N=iNi. i= 1

Also define the constants cl and c2 as follows,

I nj

- 2 2 rig/N [Z-l] i=l j=l II

(3)

(4)

(5)

(6)

(7)

(8)

C9)

Finally define,

Life= t: (rti-1) (10) i=l

df, = i 5 (nij--1) (11) i=l j=l

andL = (c2/c1)[E/dfe] + (1 - c2/c1)[S/df,]. The final analysis table was produced using equation ( 6) for the

between-isolates sum of squared distance, equation (7) for the between-clusters/within-groups sum of squared distance, and equation ( 8) for the within-clusters/ within-isolates sum of squared distance. The overall test statistic was then [ T/( I- 1 )] /L. Note that the test sta- tistic was constructed in the spirit of Satterthwaite’s ( 1946) approximation and represented an attempt to adjust for the unequal sizes of the clusters in the isolates and for the unequal numbers of clusters in each isolate.

Phylogenetic Analyses

Hypothetical evolutionary topologies were gener- ated under the principle of parsimony using the pro- grams SEQBOOT, DNAPARS, and CONSENSE from the Phylogenetic Inference Package (PHYLIP, version 3.5~; Felsenstein 1993). The initial data set, which con- sisted of five operational taxonomic units (OTUs) and 26 characters, was subsampled, and 1,000 randomized equal-sized data sets were generated (bootstrap). Par- simony-based topologies were generated for each set us- ing DNAPARS (all possible trees searched), with the results summarized as a strict-consensus tree (CON- SENSE).

Results The O-150 PCR Product Population Is Representative of the Template DNA

Relationships among Onchocerca VOZVUZUS popu- lations can be inferred from comparisons of the 0- 150 family, if variations among repeats are defined within a population. The PCR products of the 0- 150 family from each parasite population were cloned into M 13mp19, and 460 individual templates were sequenced. This ex- perimental strategy provided information about the de- gree of variability among 0- 150 repeats. The validity of this approach depends on unbiased representation of repeat units in the PCR products. A control experiment was devised which utilized the plasmid pOVS 134, which contains 12 tandemly linked copies of 0- 150 (Meredith et al. 1989). Of the 10 repeat units that should amplify under our experimental conditions, 9 sequences can be distinguished by unique bases at one or more positions. PCR products produced using pOVS 134 as a template were cloned into M 13mp 19, and the DNA sequences of 20 randomly selected clones were determined. Only one spurious nucleotide substitution was identified. The fre- quency of misincorporation ( <4.8X 10 -5 [ Gelfand and White 19901) was far lower than the variability observed between individual samples of 0- 150 (see figs. 1 and 2). Furthermore, x2 analysis demonstrated that the fre- quency of the individual monomers in the population sampled for DNA sequence analysis was not different from the expected distribution (-0.36). Thus, the PCR products were qualitatively and quantitatively represen- tative of the template DNA.

Origin of American Onchocerciasis 387

0 A CONS ATTTTGCAAAATTGATTATTAACAGATGATGACATA CGGGTACGTACCTTCAAACTGGGTCCTAAGAUAATATTCGACCATATTTTTT

BMl ........................................................................ ..A.......A....G.T......T ......... BM2 .A...T........CC..A....T............-CCC....C...............A...A...G..T................................G G

BM3 .A...T........CC..A....T............-CCC....C...............A...G...G..T................................G G BM4 ... ..CG...................................................................A.......A....G.T......T ......... BM5 .A...T........CC..A....T............-CCC....C...........................A.......................T ......... BM6 ................................................................................... ..G..G ................ BM7 ............................... ..C..............................................................T ......... BM9 .............................................................................................. ..T ......... BMll G ......................................................................... A ....... A .... G.T......T .........

BM12 G...............................................................................................T ......... BM13 ........................................................................ ..A.......A....G.T......T ......... BM14 .A...T........CC..A....T............-CCC....C...............A...A...G..T................................G G BMIS ............................. ..ATA.....G.............................................G..G...............G G

BM16 .............................................................................................. ..T ......... BM17 G . ..T........CC..A...TT............-CCC....C...................................................T ......... BM19 ................................................................................ ..A....G.T......T ......... BM20 .A...T........CC..A....T............-CCC....C...............A...A...G..T ................................. BM22 .A...T........CC..A.:..T............-CCC....C...............A...A...G..T ................................. BM23 G . ..T........C...A....T........G...-CCC....C...............A...A.......................................G G BM24 .A...T........CC..A....T............-CCC....C...............A...A...G..T .................................

cl B

CONS ATTTTGCAAAATTGATTATTAACAGATGACCTATGACATATAATTTCAAAAJQI CGGGTACGTACCTTCAAACTGGGTCCTAAGAUAATATTCGACCATATTTTW

BM2 .A...T........CC..A....T............-CCC....C...............A...A...G..T................................G G BM3

r-l BM;: ~~~~~~~lllllll~~~~l~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ BCOMN BM20 ........................................................................................

BM23 G...T........C...A....T........G...-CCC....C...............A...A.......................................G G

I Bl BM5 .A...T........CC..A....T............-CCC....C...........................A.......................T .........

BM17 G . ..T........CC..A...TT............-CCC....C...................................................T .........

BM9 .............................................................................................. ..T .........

I B2 BM16 .............................................................................................. ..T .........

BM7 ............................... ..C..............................................................T ......... BM12 G...............................................................................................T .........

I B3 BM6 ................................................................................... ..G..G ................ BM15 ............................. ..ATA.....G.............................................G..G ................

BMl ........................................................................ ..A.......A....G.T......T .........

I

BM13 ........................................................................ ..A.......A....G.T......T .........

B4 BM19 ................................................................................ ..A....G.T......T ......... BM4 ... ..CG..........................................................._.......A.......A....G.T......T ......... BMll G.........................................................................A.......A....G.T......T .........

FIG. 1 .-Clustering of 20 examples of 0- 150 from the Brazilian isolate of Onchocerca volvulus. BM I -BM24 represent the DNA sequence of individual clones containing 0- 150 monomers originally derived from the Brazilian isolate of 0. volvulus. Four of the 24 originally selected clones were found to lack inserts and were not characterized further. The DNA sequence presented represents the 106 nucleotides that fall between the 3’ ends of the PCR primers used to amplify the 0- 150 repeat. A, Sequences in the order in which they were obtained. B, Sequences rearranged into clusters based on recognition of short common sequence motifs along the sequence. Clusters are labeled corresponding to the designations in table 1 and fig. 2. In each panel, “CONS” represents the consensus sequence derived from all 20 individual monomers. Dots indicate identities with the base present in the consensus sequence. Dashes represent gaps introduced to maximize the sequence alignment.

Additional evidence suggested that the PCR prod- ucts were also representative of the repeat family when genomic DNA was the template. If this was not the case, then it would be expected that multimers of the 150-bp repeat would be the major PCR product or that no PCR product would be produced. However, in >200 parasite samples analyzed by this technique, PCR products have been produced, and the monomeric unit has been the major product (e.g., see fig. 3A). In contrast, when re- peat-specific primers derived from an 0. VOZVUZUS forest- specific 0- 150 sequence (Erttmann et al. 1987) were used in the PCR at a higher annealing temperature, mo- nomeric fragments were nearly absent from the ampli- fication products (fig. 3 B) .

Collection and Initial Analysis of Sequence Data

The O-l 50 family was amplified from parasites collected in two foci endemic for the Savannah strain 0. voZvulus (Mali and Zaire), two foci endemic for the rainforest strain (Liberia and southern Ivory Coast), and two foci in the Americas (Brazil and Guatemala). A total of 150 cloned monomers were subjected to DNA sequence analysis. Because degenerate amplification primers were annealed at low temperatures, the analysis was restricted to the 106 nucleotides between the 3’ ends of the primers. The sequences clustered into subsets on the basis of the presence of specific sequence motifs. Within these subsets, the monomeric sequences were found to be identical or nearly identical (fig. 1) . Between

390 Zimmerman et al.

A Sav Bra Gua F

977

r out

11000

FIG. 4.-Summary of the relationship between New World and African Onchocerca volvulus, and its relationship to the tram+Atlantic African slave trade. A, Strict-consensus cladogram summarizing the results obtained from analysis of the 0- 150 repeat family. B, Summary of the historical record documenting the major sources of African slaves imported into the Caribbean Islands and into South and Central America, as compiled from published sources (Davidson 196 1, pp. 41-56; Herring 1961, pp. 89-113; Povoas 1989, pp. 17-18). Areas from which the majority of the slaves were obtained are included within the hatched region. The final destinations of the slaves relevant to this study are indicated by the arrows. The sources of the parasite samples utilized in this study are indicated by the symbols. 0 = Rainforest isolates (Liberia and Ivory Coast); a = savannah isolates (Mali and Zaire); and A = New World isolates (Guatemala and Brazil).

equally manifold. Aside from the practical problems as- sociated with phylogenetic analyses of moderate-sized matrices, there existed both a lack of knowledge about the duplication history of the repeats sampled and the likelihood that the underlying connectivity structure was nonhierarchical. In addition, no estimates of the intensity of multiple hits or gene conversion were available. Fi- nally, there existed the possibility of strain hybridization in the New World populations. In phylogenetic terms, a priori identification of orthologous and paralogous characters was impossible, the potential for homoplasy

was substantial, and the appropriateness of using repeats as terminal units was questionable (Davis and Nixon 1992; Hein 1993). The confounding effects of these conditions on traditional phylogenetic reconstruction algorithms have been considered for multiple-copy se- quences (Sanderson and Doyle 1992). For these reasons, and because our initial use of parsimony-based algo- rithms resulted in extensive groups of unstable or un- resolved topologies despite repeated attempts to remove noninformative autapomorphies ( Li and Graur 199 1, pp. 1 1 l- 113 ) , we developed a method for subdividing the data into statistically homogeneous clusters. Com- monalities uniting sequences within clusters were used as the initial character states for traditional phylogenetic reconstruction. The resulting topologies served as start- ing points for refining hypotheses regarding the evolu- tionary history of both the strains of the parasite and the repeats themselves. The algorithm described here does carry its own encumbrances, such as the phenetic nature of the approach and information compression inherent in the consensus process. But overall, in light of the complex nature of the input data, the method proved successful as a pattern identification and testing technique.

Highly ordered clusters of individual repeat se- quences were found in all of the 0- 150 populations ex- amined. The existence of such clusters suggests that these sequences have been subject to mechanisms of concerted evolution. The fact that O-l 50 is organized in tandem arrays (Meredith et al. 1989) may promote interactions between members of O-l 50, which helps to maintain such clusters through recombination.

The results presented above, together with those described elsewhere (Erttmann et al. 1987; Zimmerman et al. 1993)) suggest that a unique form of O-l 50, which represents a significant proportion of the 0- 150 repeat population, has developed in the rainforest strain of 0. voZvulus. For this to occur, some barrier inhibiting ho- mogenization between the African rainforest and savan- nah strains must have developed. It is interesting that the vector for 0. volvulus in western Africa, Simulium damnosum sensu lato, is actually a complex of at least six sibling species. Experimental cross-transmission studies have shown that the different sibling species are inefficient vectors for parasites derived from outside their endemic region. This has led to the hypothesis that spe- cific vector-parasite strain-transmission complexes exist (Duke et al. 1966). The existence of such transmission complexes may inhibit homogenization between rain- forest and Savannah 0. voZvulus. Another possibility is that the presence of this unique cluster may reflect an ongoing operation of a molecular speciation mechanism. In light of their frequency of occurrence, these elements would seem to meet the theoretical expectations for a

“mechanical genome incompatibility” mechanism ( Rose and Doolittle 1983 ) .

The results from all of our analyses demonstrate a close relationship between New World and African sa- vannah populations of 0. VOZVUZUS. This suggests a recent introduction of 0. VOZVUZUS into the New World that could have resulted from African slave trade in the Americas. Nineteenth-century medical records docu- ment that Africans with symptoms consistent with onchocerciasis were seen by physicians in Rio de Janeiro, one of the most significant Portuguese slave ports in Brazil (Karasch 1987, p. 163). Because slaves imported into the Americas originated from both the rainforest and Savannah of Africa, it is interesting that parasites from the New World are so closely related to 0. VOZVUZUS from the African Savannah. This might have resulted from a founder effect ( Provine 1989 ) . As indicated in figure 4 B, the African Savannah is thought to have been the primary source of slaves imported into the New World during the peak of the slave trade in the 18th and 19th centuries. The fact that African Savannah and American 0. VOZVUZUS populations are indistinguishable, together with the fact that the majority of the slaves im- ported into the Americas originated in savannah regions, supports the hypothesis that onchocerciasis was intro- duced into the New World as a result of the slave trade.

It is possible that the statistical model presented above may be useful in studies of different parasitic or- ganisms. Sequences similar to 0- 150 have been identified in several other human parasites (Barker et al. 1986; McReynolds et al. 1986; Sim et al. 1986; Majiwa and Webster 1987; Rogers and Wirth 1987; Dissanayake and Piessens 1990). When infectious agents of the same spe- cies show important phenotypic differences resulting from single gene mutations, concurrent analysis of re- peated sequences may be useful to describe more com- pletely the overall context within which these phenotypes have diverged. To this end, we are continuing to refine our statistical algorithm, with particular attention being given to ways of evaluating individual character-state variation.

Sequence Availability

The DNA sequences have been deposited in GenBank under accession numbers UO2590 to U02594 and U0273 1 to U02875.

Acknowledgments

The authors would like to thank Drs. S. E. 0. Mer- edith, T. B. Nutman, L. A. McReynolds, J. Remme, and G. DeSole for providing parasite material and the UAB oligonucleotide synthesis facility for providing the PCR primers used in this study. Dr. J. Felsenstein pro- vided the program PHYLIP. We would also like to ac-

GELFAND, D. H., and T. J. WHITE. 1990. Thermostable DNA polvmerases. Pp. 129- 14 1 in M. A. INNIS, D. H. GELFAND, _ _

Origin of American Onchocerciasis 39 1

knowledge the helpful comments of Drs. R. Losick and R. Blanton and two anonymous reviewers. This research received support from the WHO Special Programme in Research and Training in Tropical Diseases (project 890307), the National Institutes of Health (NIAID AI33008), and the Alabama Agricultural Experiment Station.

LITERATURE CITED

BARKER, R. H., L. SUEBAENG, W. ROONEY, G. C. ALECRIM, H. V. DOURADO, and D. F. WIRTH. 1986. Specific DNA probe for the diagnosis of Plasmodium falciparum malaria. Science 231: 1434- 1436.

DADZIE, K. Y., J. REMME, A. ROLLAND, and B. THYLEFORS. 1989. Ocular onchocerciasis and intensity of infection in the community. II. West African rainforest foci of the vector Simulium yuhense. Trop. Med. Parasitol. 40:348-354.

DAVIDSON, B. 196 1. Black mother. Victor Gollancz, London. DAVIS, J. I., and K. C. NIXON. 1992. Populations, genetic

variation, and the delimitation of phylogenetic species. Syst. Biol. 41:42 l-435.

DE LEON, R. J., and B. 0. L. DUKE. 1966. Experimental studies in the transmission of Guatemalan and West African strains of Onchocerca VOIVUIUS by Simulium ochraceum, S. metal- icum and S. callidum. Trans. R. Sot. Trop. Med. Hyg. 60: 735-752.

DISSANAYAKE, S., and W. F. PIESSENS. 1990. Cloning and characterization of a Wuchereria bancrofti-specific DNA sequence. Mol. Biochem. Parasitol. 39: 147- 150.

DUKE, B. 0. L. 198 1. Geographical aspects of onchocerciasis. Ann. Belg. Sot. Trop. Med. 61: 179-186.

DUKE, B. 0. L., D. J. LEWIS, and P. J. MOORE. 1966. Oncho- cerca-Simulium complexes. I. Transmission of forest and Sudan-Savannah strains of Onchocerca volvulus, from Ca- meroon, by Simulium dumnosum from various West Af- rican bioclimatic zones. Ann. Trop. Med. Parasitol. 60:3 1% 336.

ENGELBRECHT, F., and H. SCHULZ-KEY. 1984. Observations on adult Onchocercu volvulus maintained in vitro. Trans. R. Sot. Trop. Med. Hyg. 78:2 12-2 15.

ERTTMANN, K. D., S. E. 0. MEREDITH, B. M. GREENE, and T. R. UNNASCH. 1990. Isolation and characterization of form specific DNA sequences of Onchocerca volvulus. Acta Leiden 59:253-260.

ERTTMANN, IS. D., T. R. UNNASCH, B. M. GREENE, E. J. AL- BIEZ, J. BOATENG, A. M. DENKE, J. J. FERRARONI, M. KARAM, H. SCHULZ-KEY, and P. N. WILLIAMS. 1987. A DNA sequence specific for forest form Onchocerca volvulus. Nature 327:415-417.

FELSENSTEIN, J. 1993. PHYLIP: phylogenetic inference pack- age, version 3.5~. Department of Genetics, University of Washington, Seattle.

FLOCKHART, H. A., R. E. CIBULSKIS, M. KARAM, and E. J. ALBIEZ. 1986. Onchocerca volvulus: enzyme polymorphism in relation to the differentiation of the forest and savannah strains of this parasite. Trans. R. Sot. Trop. Med. Hyg. 80: 285-292.

392 Zimmerman et al.

J. J. SNINSKY, and T. J. WHITE, eds. PCR protocols: a guide to methods and applications. Academic Press, San Diego.

HARNETT, A. E. CHAMBERS, A. RENZ, and R. M. E. PARK- HOUSE. 1989. An oligonucleotide probe specific for Oncho- cerca volvulus. Mol. Biochem. Parasitol. 35: 119- 126.

HEIN, J. 1993. A heuristic method to reconstruct the history of sequences subject to recombination. J. Mol. Evol. 36: 396-405.

HERRING, H. 196 1. A history of Latin America. Alfred A. Knopf, New York.

HIRAI, H., I. TADA, H. TAKAHASHI, E. B. NWOKE, and G. 0. UFOMADU . 1987. Chromosomes of Onchocerca volvulus (Spirurida: Onchocercidae ) : a comparative study between Nigeria and Guatemala. J. Helminthol. 61:43-46.

KARASCH, M. C. 1987. Slave life in Rio de Janeiro: 1808- 1850. Princeton University Press, Princeton.

LI, W., and D. GRAUR. 199 1. Fundamentals of molecular evolution. Sinauer, Sunderland, Mass.

LOBOS, E., and N. WEISS. 1985. Immunochemical comparison between worm extracts of Onchocerca volvulus from sa- vanna and rain forest. Parasite Immunol. 7:333-347.

MCREYNOLDS, L. A., S. M. DESIMEONE, and S. A. WILLIAMS. 1986. Cloning and comparison of repeated DNA sequences from the human filarial parasite Brugia malayi and the animal parasite Burgia pahangi. Proc. Natl. Acad. Sci. USA 83:797-80 1.

MAJIWA, P. A. O., and P. WEBSTER. 1987. A repetitive de- oxyribonucleic acid sequence distinguishes Trypanosoma simiae from T. congolense. Parasitology 95:543-558.

MANLEY, B. F. 199 1. Randomization and Monte Carlo meth- ods in biology. Chapman & Hall, London.

MEREDITH, S. E. O., G. LANDO, A. A. GBAKIMA, P. A. ZIM- MERMAN, and T. R. UNNASCH . 199 1. Onchocerca volvulus: application of the polymerase chain reaction to identifica- tion and strain differentiation of the parasite. Exp. Parasitol. 73:335-344.

MEREDITH, S. E. O., T. R. UNNASCH, M. KARAM, W. F. PIES- SENS, and D. F. WIRTH . 1989. Cloning and characterization of an Onchocerca volvulus specific DNA sequence. Mol. Biochem. Parasitol. 36: 1 - 10.

NEI, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York.

NELSON, G. S. 199 1. Human onchocerciasis: notes on the his- tory, the parasite and the life cycle. Ann. Trop. Med. Par- asitol. 85:83-95.

PASCALE, E., C. LIU, E. VALLE, K. USD~N, and A. V. FURANO. 1993. The evolution of long interspersed repeated DNA (L 1, LINE 1) as revealed by the analysis of an ancient rodent Ll DNA family. J. Mol. Evol. 36:9-20.

PETRONDAS, D. A., and K. R. GABRIEL. 1983. Multiple com-

P~VOAS, R. DE C. 1989. The language of CandomblC: levels of Afro-Portuguese sociolinguistic integration. Jose Olym- pio, Rio de Janeiro.

PROVINE, W. B. 1989. Founder effects and genetic revolutions in microevolution and speciation: an historical perspective. Pp. 43-78 in L. V. GIDDINGS, K. Y. KANESHIRO, and W. W. ANDERSON, eds. Genetics, speciation, and the founder principle. Oxford University Press, New York.

REMME, J., K. Y. DADZIE, A. ROLLAND, and B. THYLEFORS. 1989. Ocular onchocerciasis and intensity of infection in the community. I. West African savanna. Trop. Med. Par- asitol. 40:340-347.

ROGERS, W. O., and D. F. WIRTH. 1987. Kinetoplast DNA minicircles: regions of extensive sequence divergence. Proc. Natl. Acad. Sci. USA 84:565-569.

ROSE, M. R., and W. F. D~~LITTLE. 1983. Molecular biological methods of speciation. Science 220: 157- 16 1.

SANDERSON, M. J., and J. L. DOYLE. 1992. Reconstruction of organismal and gene phylogenies from data on multigene families: concerted evolution, homoplasy, and confidence. Syst. Biol. 41:4-17.

SANGER, F., S. NICKLEN, and A. R. COULSON. 1977. DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 12:5463-5467.

SATTERTHWAITE, F. E. 1946. An approximate distribution of estimates of variance components. Biometrics 2: 1 lo- 114.

SAWYER, S. 1989. Statistical tests for detecting gene conversion. Mol. Biol. Evol. 6:526-539.

SHAH, J. S., M. KARAM, W. F. PIESSENS, and D. F. WIRTH. 1987. Characterization of an Onchocerca specific DNA clone from Onchocerca volvulus. Am. J. Trop. Med. Hyg. 37~376-384.

SIM, B. K. L., J. W. MAK, W. H. CHEONG, I. SUTANTO, L. KURNIAWAN, H. A. MARWOTY, E. FRANK, J. R. CAMBELL, D. F. WIRTH, and W. F. PIESSENS. 1986. Identification of Brugia malayi in vectors with a species specific probe. Am. J. Trop. Med. Hyg. 35:559-564.

THYLEFORS, B. 1978. Ocular onchocerciasis. Bull. World Health Organ. 56:63-72.

YARZABAL, L., I. PETRALANDA, M. ARANGO, L. LOBO, and C. BOTTO . 1983. Acid phosphatase patterns in microfilariae of Onchocerca VOIVUIUS s.1. from the Upper Orinoco Basin, Venezuela. Trop. Med. Parasitol. 34: 109- 112.

ZIMMERMAN, P. A., L. TOE, and T. R. UNNASCH. 1993. Design of Onchocerca DNA probes based upon analysis of a re- peated sequence family. Mol. Biochem. Parasitol. 58:259- 269.

MITCHELL L. SOGIN, reviewing editor

Received July 23, 1993 parisons by rerandomization tests. J. Am. Stat. Assoc. 78: 949-957. Accepted January 10, 1994