| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
PT). Three geographically and genetically differentiated groups of populations were identified. Population substructure was only detected in the border town of Nogales. Reliable estimates of M between regions ranged from 1.02 to 3.41 and between cities within regions from 1.66 to 4.44. In general, pairwise PT were lowest between cities within regions. The observed patterns of genetic differentiation suggest infrequent migration between populations and are compatible with the idea of human transport facilitating dispersal between regions. |
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
Vector movement and transport are important factors influencing dengue transmission.9 Dengue is a health concern in Arizona due to the existence of endemic and epidemic dengue activity in the neighboring state of Sonora, Mexico and along the Texas-Mexican border.7,10 In addition to the importation of infected adult females and virus introduction from traveling viremic humans, dengue viruses could become introduced to naive Ae. aegypti populations through the importation of infected eggs. Multigenerational, transovarial passage and horizontal transmission of dengue from vertically infected progeny has been observed with Ae. aegypti.11,12 Previous work determined that the Tucson population of Ae. aegypti could transmit dengue and was most closely related to the Pacific population of Mexico, but other Arizonan populations were not analyzed.13–16
Amplified fragment length polymorphism (AFLP) has been well documented, and is aptly suited for analysis of Ae. aegypti populations.17–21 Other molecular marker systems are not well suited for use with Ae. aegypti. Microsatellites are not abundant in Ae. aegypti and the number of variable loci used in microsatellite studies has been low.22–24 Mitochondrial sequencing requires prior sequence knowledge and variation may be limited due to colonization events. Random amplified polymorphic DNA loci were not useful for comparisons of distant populations and their reliability has been questioned.15,25–27 Restriction fragment length polymorphism loci were informative, co-dominant markers, but successful digestion necessitates substantial amounts of genomic DNA and hybridization would be time-consuming.25
The aims of this study were to determine genetic differentiation and migration between Ae. aegypti populations, detect population substructure, and identify possible embarking populations that may have begun the colonization of Arizona. These aims serve to evaluate the possibility of dengue introduction in Arizona through mosquito transport and to predict the spread of insecticide resistance resulting from control measures. This can inform future policy decisions regarding mosquito control and disease prevention measures. Global positioning system coordinates of positive traps are included to aid future studies because we observed areas within infested cities consistently negative for Ae. aegypti (Merrill SA, unpublished data).
Cladistic analysis using genetic distance was performed to detect population structure within geographic populations by allocating individuals into genetic populations. Pairwise PT, an estimator of FST,28–30 was computed on genetic populations and further cladistic analysis of linearized PT/(1 – PT) was used to assign these genetic populations to genetic regions.31 Distant, well-established Ae. aegypti geographic populations corresponding to the genetic populations found within the subsequent genetic regions were identified as putative embarking cities of colonization. To analyze migration between cities in the study area, pairwise PT values were again calculated for discrete geographic populations. Effective migration (M) between geographic populations was then estimated from these pairwise PT values.30,31 Cladistic analysis of linearized PT/(1 – PT) from geographic populations was used to assign geographic populations into geographic regions. Analysis of molecular variance (AMOVA) was conducted using these geographic regions.
|
MATERIALS AND METHODS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
and Table 1). Egg papers were collected at four-day intervals, and trapping was continuous in Benson, Willcox, St. David, Douglas, Naco, Tempe, and Sierra Vista during the dates indicated. Traps were deployed for one week each month in Tucson and Nogales. Egg papers from individual traps were dried overnight, eggs were counted, and the papers were immersed in water to induce hatching. Larvae were fed a sieved homogenate of high-protein rabbit pellets and soy meal. Individual pupae were harvested and stored at –80°C prior to DNA extraction. Larval exuviae for each site were examined to confirm that the specimens were Ae. aegypti and to rule out possible contamination by other container-breeding mosquitoes known to inhabit the study area.
|
|
, and collected mosquitoes were stored at –80°C prior to identification. All collected adults were examined to ascertain species identity, females were dissected to insure that they did not have a blood meal, and spermathecae were removed from females to prevent genomic contamination from mating. Samples from Islamorada, Florida were provided from the Florida Medical Entomology Laboratory (Vero Beach, FL) as pupae reared from eggs collected in Monroe County, Florida, and were processed as above. Samples from Tempe were provided as eggs from the Arizona Department of Health Services Vector Borne and Zoonotic Disease Section (Phoenix, AZ) and were reared and processed as above. Houston field-collected adults were provided by Adilelkhidir Bala of the Harris County Health Department, Mosquito Control Division (Houston, TX). Samples from Hermosillo, Mexico were provided by Suzanne Hammer (University of Arizona) as DNAs from adults collected in 1998 using carbon dioxide traps.
Extraction of DNA. Individuals were chosen from all traps and collection dates per location and assigned a unique identification code. The extraction protocol followed that of Goldberg and others, with minor modifications.32 Individual specimens were homogenized using plastic pestles in 1.5-mL centrifuge tubes containing 350 µL of lysis buffer (100 mM Tris-HCl, pH 8.0, 2% sodium dodecyl sulfate, 100 mM NaCl, 50 mM EDTA, 100 mM sucrose) and 12 µL of proteinase K (10 mg/mL), and Phase Lock tubes (Eppendorf AG, Hamburg, Germany) were used in phase separation as per the manufacturer’s instructions. Samples were then dried under vacuum and suspended in 50 µL of TE0.1. Suspended samples were stored at –20°C. Genomic DNA was quantified with PicoGreen (Molecular Probes, Inc., Eugene, OR) as per the manufacturer’s instructions. DNA aliquots were standardized to a concentration of 15 ng/µl with TE0.1 (10 mM Tris-HCl, pH 8.0, 0.1 mM EDTA).
Amplified fragment length polymorphism analysis. The AFLP was conducted with adapter and primer sequences described by Yan and others (Qiagen, Inc., Valencia, CA).21 Digestion and ligation reactions were coupled to drive the formation of ligated genomic DNA. The 11-µL digestion/ligation reactions contained 3 µL of genomic DNA standardized at 15 ng/µl. One unit of Mse I endonuclease, five units of Eco RI endonuclease, and 0.15 Weiss units of T4 DNA ligase (New England Biolabs, Inc., Beverly, MA) were used in each reaction containing 50 µmoles of Mse I adapter, 5 µmoles of Eco RI adapter, 0.05 M NaCl, 0.55 µg of bovine serum albumin (BSA), and 10x T4 ligase buffer (1x T4 ligase buffer = 50 mM Tris-HCl, pH 7.5, 10 mM MgCl2, 10 mM dithiothreitol, 1 mM ATP, 25 µg/mL of BSA). The Eco RI and Mse I adapters were constructed by combining and annealing equimolar amounts of the corresponding forward and reverse oligonucleotides. Reactions were covered with 13 µL of mineral oil to prevent evaporation, and incubated at 37°C for 2.5 hours. Reactions were then diluted with 90 µL of TE0.1.
Primary polymerase chain reaction (PCR) amplifications were performed in a volume of 30 µL containing 3 µL of dilute digestion/ligation product, 0.5 units of Taq polymerase (Eppendorf AG), 0.2 mM deoxynucleotide triphosphates (dNTPs) (Fermentas Inc., Hanover, MD), 0.2 µM Eco-A primer, 0.2 µM Mse-C primer, and 10x PCR buffer (500 mM KCl, 100 mM Tris-HCl, pH 8.3, 20 mM MgCl2). After incubation at 72°C for 2 minutes, amplification consisted of 22 cycles at 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 1 minute. Amplifications were verified by subjecting 8 µL of each reaction to electrophoresis on a 3% agarose gel. Primary amplification products were diluted in 189 µL of TE 0.1.
Secondary PCR amplifications used two primer combinations: Eco-AGG plus Mse-CTT and Eco-ACA plus Mse-CAC. The Eco-AGG and Eco-ACA primers were 5' labeled with 6-carboxyfluorescein (6-FAM) for fluorescent detection (Applied Biosystems, Foster City, CA). The 15-µL reactions volumes contained 3 µL of dilute primary amplification product as template, 0.5 units of HotMaster Taq polymerase (Eppendorf AG), 0.2 mM dNTPs, 0.167 µM Eco primer, 0.194 µM Mse primer, and 10x HotMaster PCR buffer. A touch-down protocol (9 cycles at 94°C for 30 seconds, 65°C with –1°C /cycle, 72°C for 1 minute) was used to prevent spurious primer annealing, followed by 30 cycles of regular amplification at 94°C for 30 seconds, 56°C for 30 seconds, and 72°C for 1 minute. Amplifications were verified by subjecting 8 µL of each reaction to electrophoresis on a 3% agarose gel. Secondary amplification products were diluted with 40 µL of molecular biology grade water and submitted for capillary electrophoresis on an ABI 3100 apparatus (Applied Biosystems) at the University of Arizona Genomic Analysis and Technology Core. Negative controls, lacking only template DNA, were included in each phase of analysis to detect possible contamination.
Electropherograms were analyzed using Genotyper 3.7 software (Applied Biosystems). Amplification was verified by inspecting individual electropherograms. Successfully amplified samples for population analysis were scored at 254 polymorphic loci for the Eco-AGG plus Mse-CTT primer pair and at 254 polymorphic loci for the Eco-ACA plus Mse-CTT primer pair.
Robustness of the AFLP. Reproducibility and discriminatory power of the AFLP were first investigated by analyzing controlled outcrosses. Field-collected (Nogales, AZ) and colony mosquito eggs (F32, origin: Tucson, AZ) were reared to pupae and isolated individually before emergence. After emergence, colony and field-collected mosquitoes were individually mated and females were fed bovine blood using a feeder. Progeny from the crosses, reared to pupae, and parents were analyzed as described earlier. Using PAUP, siblings from one cross were distinguishable from siblings from other crosses on a phylogram constructed using 421 AFLP loci, minimum evolution optimality, total genetic distance, and a heuristic search using neighbor joining.33 In one randomly selected family of 42 individuals, maternal inheritance was not seen in any of the 11 loci for which the male was null (Merrill SA, unpublished data). This was consistent with prior inheritance studies of AFLP that indicated low numbers of maternal mitochondrial loci.34
Identification of genetic populations through cladistic analysis. The AFLP markers were used under the assumptions of Hardy-Weinberg equilibrium, independent assortment, and identity by descent of both fragment presence and absence. Five hundred eight polymorphic AFLP loci were used to analyze individuals and characterize the populations in the study area.
PAUP was used to generate an individual-based cladogram to detect substructure in geographic populations, thus defining genetic populations. Using minimum evolution, total character difference from 508 polymorphic AFLP loci the midpoint rooted tree was constructed using neighbor-joining and tree-bisection-reconnection.33
Cladistic and statistical analyses of genetic populations.
Pair-wise PT and analysis of molecular variance were computed with 999 permutations using GenAlEx.18,35 Linearized PT/(1 – PT) data from genetic populations were examined with PAUP through a minimum evolution, total character difference, unrooted cladogram created using a heuristic search with neighbor-joining and tree-bisection-reconnection branch swapping.33
Cladistic and statistical analyses of geographic populations.
Pairwise PT and analysis of molecular variance were computed for geographic populations, i.e., all individuals collected in one city, as described earlier. Categories of population differentiation were determined from these pairwise PT values.36 Pairwise PT values were used to estimate effective migration rates (M) between populations.19,20 Linear PT/(1 – PT) data were examined with PAUP as described earlier to define geographic population regions for use with AMOVA.
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
).
|
).
|
PT /(1 – PT) values were calculated with Nogales, Sonora divided into two genetic populations (Nogales and Nogales Minor) as illustrated in the above analysis. Three population regions were observed in the study area, and clear cladistic division of the Nogales and Nogales Minor populations was found (Figure 2). The Western region incorporated Tucson, Hermosillo, Tempe, and Nogales; the Eastern region included Willcox, Houston, El Paso, and Islamorada; the Cochise region contained Douglas, Naco, Sierra Vista, Benson, St. David, and the Nogales Minor population.
All pairwise PT comparisons were statistically significant at the P < 0.001 level as determined by a permutation test (see Materials and Methods), except for Willcox and Houston, El Paso, Hermosillo, Benson, and Nogales Minor, (Merrill SA, unpublished data). The Nogales Minor population was found to be significantly differentiated from the Nogales population at the P < 0.001 level.
Cladistic and statistical analysis of geographic populations.
Pairwise population differentiation, PT, and effective migration (M) between geographic populations are shown in Table 3. In these calculations, the genetic populations of Nogales and Nogales Minor are included as the geographic population of Nogales. Cladistic analysis of PT /(1 – PT) values indicated the same regional structure found in Figure 2, with the geographic population of Nogales in the Western region (Merrill SA, unpublished data). With the exception of the Cochise genetic region containing the Nogales Minor genetic population, each genetic region and its corresponding geographic region were found to be one in the same.
|
PT > 0.05), and none exhibited little differentiation (PT < 0.05). Tempe was greatly differentiated (0.16 < PT) from all populations except Tucson, Nogales, and Hermosillo. Great differentiation (0.20 < PT) was also observed between St. David and Tempe. El Paso was greatly differentiated (0.16 < PT) from Tempe, Willcox, and St. David. Willcox was greatly differentiated (0.16 < PT) from all populations except Houston and Benson. Permutations of PT between all populations had a P < 0.001, except for Willcox comparisons. The Willcox comparisons produced P values ranging from 0.026 to 0.001.
The lowest effective migration37 (n > 50, P < 0.001) was 1.02 individuals per generation between St. David and Tempe, which are members of two different regions (Figures 1 and 2). The highest observed rate (n > 50, P < 0.001) was 4.44 individuals per generation between Douglas and Naco, which are geographically proximal members of the Cochise region. Migration rates between populations in Arizona varied according to geography. Average pairwise migration within regions was 3.29, 2.10, and 2.42 individuals per generation in the Cochise, Eastern, and Western regions, respectively. Effective migration for Willcox (n = 2) and El Paso (n= 10) were not computed because small sample sizes are known to artificially inflate FST estimators and grossly under- estimate M.
Table 4 shows the results of AMOVA analyses. Of the total genetic diversity from the two marker sets, the majority (88.5%) can be attributed to variation within geographic populations, 8.4% to variation among geographic populations within regions, and 3.1% to variation among regions. Average statistics compiled from the two marker sets suggested that populations within regions were moderately, yet significantly, differentiated (PR = 0.09, P < 0.001), and that regions were little, yet significantly, differentiated (RT = 0.03, P < 0.001).
|
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
PT values observed for Willcox. The Eastern region encompassed the well-established populations in Islamorada, Florida and Houston, Texas; it also included the city of Willcox, Arizona. The Western region included the well-established Mexican population of Hermosillo, and extended north from Mexico incorporating Nogales (Sonora), Tucson, and Tempe. The extent of the Eastern and Western regions was in agreement with previous work by Gorrochoteguli-Escalante and others, who found that the northeast Mexico genetic population, including the geographic population of Houston, was distinct from the Pacific Mexico population, to which Tucson belonged.15 The Cochise County Arizona region appeared to represent an interface between the Western and Eastern regions, and included the Douglas, Naco, Sierra Vista, Benson, St. David, and Nogales Minor populations.
Although Nogales was found to contain two separate populations, most populations were more genetically homogenous, but had relationships that spanned regional classifications. Although Benson clustered in the Cochise region and exhibited the lowest differentiation with Douglas, the lowest differentiation observed for Houston was with Benson. Tucson demonstrated the lowest differentiation with Sierra Vista, and second lowest with Benson. These data supported the idea that the Cochise region was produced by mixing between the Western and Eastern regions.
Population substructure was only detected in Nogales, Sonora. Nogales, Sonora and Nogales, Arizona are contiguous sister cities divided by the international border. In July and August, three traps in northwestern Nogales, Sonora collected individuals that belonged to the Cochise region. Subsequent analysis showed that this Nogales Minor population was distinct from the remaining Nogales population, which belonged to the Western region. Confounding this observed substructure, the three traps that collected Cochise region individuals in July and August all collected individuals from the Western region in September. This suggests that the population substructure in Nogales, Sonora was temporary. Such temporal partitioning could have been due to recent immigration from the nearby Cochise region. The suspected partitioning of St. David was dismissed because all members composing the St. David South population were collected from one trap on one collection date, were likely to have been siblings, and were not distinct from the remainder of the St. David population.
The lowest levels of population differentiation, thus most frequent migration, were found in the Cochise region between Douglas and the cities of Naco, Benson, and Sierra Vista. The most isolated Arizona population was Tempe, which was at the northern extreme of the Western region. These data show that large geographic distances, such as the distance from Houston to Benson, were not insurmountable migratory barriers for Ae. aegypti, but that local migration was more common. High rates of local, intra-regional migration found in the Cochise region and lower local migration observed in the Western region suggest that the hypothetical spread of insecticide resistance through transport of mosquitoes is more likely to occur within the Cochise region, but that actual dengue introduction is more likely in the Western region. Compared with populations of the Cochise region, the populations of the Western region were generally more isolated from each other, but exchanged more migrants with Hermosillo, Sonora, a city with endemic dengue.
The importance of geography and human migration on Ae. aegypti population structure is likely, especially because the dry climate and geographic distances in the study region negate Ae. aegypti dispersal between cities by flight. The geography of southern Arizona has shaped Ae. aegypti population structure by influencing highway routes and human migration. While vast distances typically separate the Cochise and Eastern regions, many mountains separate the Cochise region from the Western region. All cities of the Cochise region are contained in the San Pedro Valley, which runs north from Mexico. Douglas represents a roadway hub in the Cochise region, with connections to Benson, Naco, and Sierra Vista though state highways. Tucson and Nogales reside in the Santa Cruz Valley and are connected by Interstate 19. Interstate 10 connects Tucson to the Phoenix metro area, including Tempe. Interstate 10 also connects Tucson, Benson, Willcox, El Paso, and Houston.
Roadway systems correspond to observed patterns of population differentiation. Douglas is linked to Benson, Naco, and Sierra Vista by Arizona Routes 80 and 92. Route 90 and Interstate 10 connect Tucson and Sierra Vista. Tucson and Houston are joined to Benson by Interstate 10. Although St. David lies between Douglas and Benson on Route 80, the majority of the traffic to St. David is from nearby Benson, accounting for the close genetic relationship of St. David and Benson Ae. aegypti populations.
The extent and origin of the Western region showed that Ae. aegypti from dengue-endemic areas of western Mexico have penetrated as far north as the Phoenix metro area. Migration from the Eastern region, another area with endemic dengue near the Texas-Mexico border, was detected in Willcox and Benson. The Cochise region represents the interface between the Western and Eastern regions, and received migrants from both.
We and others previously demonstrated that necessary conditions for dengue transmission, i.e., parity, population levels, vector competence, and blood feeding tendencies of Ae. aegypti, were met in Tucson.13–16 The lack of dengue cases in Arizona could be attributable to infrequent migration from dengue areas, resulting in the failure to introduce the virus to the Arizonan vector populations. Infrequent migration supports the observation of moderate differentiation between populations in the study area and could account for the initial colonization of the cities in Arizona. The absence of dengue in Arizona may not be solely attributable to infrequent migration of Ae. aegypti. Reiter and others suggested that economic factors such as air conditioning were paramount in limiting dengue transmission in Texas.38 Given the genetic history of Arizonan Ae. aegypti populations, similar factors may be integral in preventing dengue transmission in Arizona.
|
APPENDIX 1 |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
. Four different coding schemes exist. For example, "Nogales Sept 172 10903" was an individual collected in trap 172 in Nogales, Mexico during September and 10903 was a unique identifier; "Douglas 1 820h92b" is the second individual (b) harvested on September 2, 2003 from eggs collected on August 20, 2003 from the 1 trap in Douglas, AZ; "FL11" is the 11th individual obtained from Islamorada, FL; and "Houston 111a" is the first individual (a) obtained from trap 111 in Houston, TX.
|
Received May 7, 2004. Accepted for publication September 17, 2004.
Acknowledgments: We thank Adilelkhidir Bala and the staff of the Harris County Health Department Mosquito Control Division (Houston, TX), Suzanne Hammer (University of Arizona), George O’Meara (Florida Medical Entomology Laboratory, Vero Beach, FL), and T. Michael Fink and Craig Levy (Arizona Department of Health Services Vector Borne and Zoonotic Disease Section, Phoenix, AZ) for contributing field samples. We also thank Matt Kaplan, Taylor Edwards, Kathleen Walker, and Jason Wilder for helpful comments. Molecular work was conducted at the University of Arizona Genomic Analysis and Technology Core.
Note: The appendix to this manuscript appears online at www.ajtmh.org.
Financial support: This study was supported in part by Cooperative State Research, Education, and Extension Service-Integrated Pest Management funds through the College of Agriculture at the University of Arizona, and by the Undergraduate Biology Research Program to Samuel A. Merrill, funded by a grant from the Howard Hughes Medical Institute (Chevy Chase, MD).
Authors’ addresses: Samuel A. Merrill, Medical Scientist Training Program, Washington University in St. Louis, 660 S. Euclid Avenue, Box 8226, St. Louis, MO, 63108. Frank B. Ramberg, and Henry H. Hagedorn, Department of Entomology, University of Arizona, 1140 E. South Campus Drive, Forbes Room 410, PO Box 210036, Tucson, AZ, 85721-0036, E-mail: hagedorn{at}ag.arizona.edu.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIX 1REFERENCES |
|---|
1/(4Nm+1). Heredity 82: 117–125.
This article has been cited by other articles:
|
|
 |
|
  V. M. Scarpassa, T. B. Cardoza, and R. P. Cardoso Junior Population Genetics and Phylogeography of Aedes aegypti (Diptera: Culicidae) from Brazil Am J Trop Med Hyg, June 1, 2008; 78(6): 895 - 903. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. G. Clark Dengue and Dengue Hemorrhagic Fever in Northern Mexico and South Texas: Do They Really Respect the Border? Am J Trop Med Hyg, March 1, 2008; 78(3): 361 - 362. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
