Patillas is a municipality located on the southern coast of Puerto Rico characterized by seasonal rainfall and a dry season that extends between December and April of each year (Fig. 1). The communities of Jagual and Centro, located within the municipality of Patillas, were selected as study locations based on differences in waste disposal infrastructure. Jagual relies solely on residential septic tanks as a means of facilitating disposal. Furthermore, surveillance efforts had previously confirmed the existence of Ae. aegypti adults in some septic tanks (R. Barrera, unpublished data). In contrast, Centro is a more urban community in which residential housing sections use a sewage system for waste disposal. Centro is free of any septic tank structures, but is otherwise similar to Jagual in terms of rainfall and other environmental factors. The two communities are located ≈2.5 km apart from one another.

Surveys were conducted at selected houses within each community between May and June of 2010. Well over 15 containers were sampled in each location. Within Jagual, residences were excluded if the septic tank was not accessible, lacked standing water, or if the owner could not be contacted or did not wish to participate in the study. Verbal informed consent was obtained at each house from a person aged 18 or older before conducting each survey. In Jagual, a total of three septic tanks were located that produced Ae. aegypti. Each was sampled and adults were combined for analysis. All septic tanks were within 1–2 km of each other.

After identifying candidate septic tanks, cracks were sealed in both the walls and ceiling of the structure using a commercially available polyurethane foam spray. Large openings were covered with a plastic lid and sealed with foam spray. Finally, vent pipes were screened to eliminate any potential uncontrolled entry or exit points from septic tanks. To ensure emergent adults actually developed in the septic tank and had not entered the tank before collection, we used an insecticide strip to first eliminate all adult mosquitoes residing within the structure. A commercially available 2,2-dichlorous impregnated strip (trade name: HotShot, Madison, WI) was suspended in the septic tank above the liquid and left in place for 24 h (Burke et al. 2010). After removal of the impregnated strip, previously constructed emergent traps were secured in place over the tank opening (Barrera et al. 2008). Traps were constructed from 3.8 liter plastic containers using a screen material to seal the open end and a fabric sleeve to cover a large diameter hole cut into the middle of the container wall. The fabric sleeve was placed either over the septic tank vent, or over a large inverted funnel that had been secured in place over an alternative septic tank opening. Traps were recovered and replaced daily until ≈150 adults were collected, ensuring adequate power for detecting differences during microsatellite analysis. Collection of Ae. aegypti was confirmed through visual identification. Adults were then transferred to large colony cages and fed a 10% sucrose solution. Adults were allowed to feed once through a membrane on swine blood and were kept until they laid eggs. Eggs were collected and dried for later use. Adults were killed by placing them in a −20°C freezer for 30 min. Wings were removed for measurement and carcasses were preserved in 80% ethanol for later DNA extraction.

Pupae were collected from multiple container types with standing water from within both community sites. Types of containers sampled included plant containers, tires, discarded containers, water storage containers, buckets, jars, and flower vases. Because feeding is discontinued when Ae. aegypti reach the pupal stage of development, collecting at this point minimized the developmental impacts on size that could have occurred if earlier larval stages were sampled and transferred to a laboratory setting where nutrient levels would likely not represent native habitats. Sampling of pupae also meant eliminating bias by preventing the cross contamination of sampling locations. Identification was confirmed using standard dichotomous keys (CDC, unpublished data). Pupae were placed into plastic containers filled with tap water inside large colony cages and allowed to emerge as adults. Ae. aegypti adults were fed a 10% sucrose solution, blood fed once, and allowed to deposit eggs. Eggs were collected on paper strips, dried and stored for later use. Again, adults were killed by placing them in a −20°C freezer for 30 min. Wings were removed for measurement and carcasses were preserved in 80% ethanol for later DNA extraction.

Plastic trays of uniform size were sterilized by soaking with diluted bleach, thoroughly rinsing, and placing in a UV cross-linker for 10 min. Plastic trays were housed in an insectary and the temperature was maintained at 25°C with a relative humidity of 54%. A total of 500 ml of deoxygenated tap water was added to each of the trays. Strips containing the eggs deposited by field collected Ae. aegypti adults were placed into plastic containers and allowed to hatch. Containers were loosely covered with a plastic lid to prevent any contamination by falling debris. Egg strips were removed 24 h later and first instar densities were thinned to ≈90 individual mosquitoes per plastic container. Upon hatching, 100 mg of ground fish flakes were added to each container. Additional food and water was added to all containers equally either daily or every other day as needed. Mosquitoes were allowed to develop to 2-d old adults before being killed and preserved as described previously. Wings were removed for measurement purposes.

Wing length was used as an estimate of body size (Powell et al. 2009) and was obtained by measuring from the humeral cross vein to the wing tip, excluding fringe scales, using a Zeiss (Jena, Germany) AxioCam HRc live camera attached to a computer with AxioVision software. Wing length was first measured in pixels and then converted into absolute length (millimeters) for comparison (Powell et al. 2009). A total of between 86 and 122 female and between 6 and 43 male wing measurements were taken from each of the experimental groups collected in the Jagual and Centro communities. Additionally, a total of 41–54 female and 15–28 male wing measurements were taken from each of the laboratory reared experimental groups.

Whole body genomic DNA was extracted individually from adult mosquitoes from each comparison group [Jagual septic tanks (N = 140), Jagual surface containers (N = 138), and Centro surface containers (N = 103)] using DNeasy kits (Qiagen Inc., Valencia, CA) following the protocol provided by the manufacturer. Individual genotypes were scored at 12 previously published microsatellite loci. Detailed information regarding the development and screening of these loci are provided elsewhere (Brown et al. 2011).

Microsatellite loci were paired for multiplex reactions based on nonoverlapping size ranges. Each pair was amplified in a multiplex polymerase chain reaction (PCR) using a single fluorescent M13 primer, two forward primers containing M13 tails, and two reverse primers (Boutin-Ganache et al. 2001, Brown et al. 2011). Total volume of PCR reactions was 10 μl and contained 1× Type-it Multiplex PCR Master Mix (Qiagen), 25 nM of each forward primer, 250 nM of each reverse primer, and 500 nM of a fluorescently labeled M13 primer. Thermocycling conditions were established following those outlined in Slotman et al. (2007). PCR products were subsequently run on an Applied Biosystems 3730×l DNA Genetic Analyzer with a GS 500 Rox internal size standard available from Applied Biosystems (Carlsbad, CA). Genotype calls were automated using GeneMapper software and visually checked for errors. Additional microsatellite primer information can be found in Brown et al. (2011).

A one-way analysis of variance (ANOVA) adjusted for multiple comparisons using the Tukey-Kramer method was used to compare wing length between groups for both field collected and lab reared generations.

Overall genetic differentiation among mosquito populations was assessed by calculating mean F_st values for all population pairs using Arlequin 3.5 software (Excoffier et al. 2010). Average allelic richness per locus was calculated for each population using HP-RARE (Kalinowski 2004, 2005). All microsatellite loci were tested for within-population deviations from Hardy-Weinberg equilibrium using the Web-version of the software GENEPOP (Raymond and Rousset 1995, Rousset 2008). Markov chain parameters were set at 1,000 dememorizations, 100 batches, and 1,000 iterations per batch. GENEPOP was also used to calculate allele frequencies for all loci across populations.

We evaluated patterns of population structure using the Bayesian clustering method implemented in the software program STRUCTURE v2.3 (Pritchard et al. 2000). STRUCTURE identifies genetic clusters and assigns all individuals to these clusters without any a priori information regarding sampling locations. The most likely number of clusters (K) was determined by conducting independent runs for each K = 1–5. For all runs we assumed an admixture model and correlated allele frequencies, and used a burn-in value of 100,000 iterations followed by 500,000 replications. The most likely number of clusters was determined following the guidelines in Pritchard et al. (2000). STRUCTURE results were visualized using the program DISTRUCT (Rosenberg 2004).

Finally, we assessed relatedness among the Puerto Rican populations studied here and other global populations (see Brown et al. 2011) using a nonmetric multidimensional scaling analysis based on the pairwise F_st values between populations (Hammer et al. 2001). We compared the Puerto Rican samples to those from Mexico (Pijijiapan), Texas (Houston), Dominica, two sites in Venezuela (Bolivar and Zulia), two sites in Thailand (Rayong and Prachuabkhirikan), French Polynesia (Tahiti), and three locations in Florida (Vaca Key, Conch Key, and Palm Beach County). See Brown et al. (2011) for data on these populations.

For both females and males, average wing length did not statistically differ between adult Ae. aegypti emerging from septic tanks in Jagual and those originating from surface containers in Centro (Tukey-Kramer MSD > 0.002 (nonsignificant); Tukey-Kramer MSD > 0.194 (nonsignificant)) (Fig. 2). However, average wing length did statistically differ between experimental groups in Jagual (septic tank and surface container) for both female and male adult Ae. aegypti (Tukey-Kramer MSD < 0.195 (significant); Tukey-Kramer MSD < 0.188 (significant)), with septic tank adults having longer wing lengths than surface container adults on average (Fig. 2). Finally, average wing length statistically differed between mosquitoes originating from surface containers in Jagual and Centro in both female and male adult Ae. aegypti(Tukey-Kramer MSD < 0.193 (significant); Tukey-Kramer MSD < 0.382 (significant)). Wing length of adults originating from surface containers in Centro on average was larger than wing length of adults originating from surface containers in Jagual (Fig. 2). Among the field-collected mosquitoes of both sexes, there is notably less variation in wing length within the septic tank group than in the surface container collections from both localities.

When F₁ mosquitoes were hatched and reared to adulthood in a controlled laboratory environment in which nutrient levels were kept the same between all experimental groups, both female and male differences in average wing length between groups were reduced. However, statistically significant differences still remained between some experimental groups (Fig. 2). On average, wing lengths of females descended from surface containers in Centro and males descended from septic tanks in Jagual were larger compared with other experimental groups. As demonstrated previously in this species (Schneider et al. 2011), variability in body size was greatly reduced in laboratory-reared mosquitoes as compared with those from the field (Fig. 2).

Allele frequency data for the 12 microsatellites in the Patillas collections appear online as supporting information (Supplemental Table S1 available online only). Mean pairwise F_st values showed that experimental groups differed (Table 1). However, because F_st values as high as 0.05 generally indicate negligible genetic differences (Olshen et al. 2008), and given that the average population pairwise F_st value for this subspecies in a previous study was 0.20 (Brown et al. 2011), we concluded that the differences observed in this study were not biologically meaningful and these experimental groups did not represent genetically separate Ae. aegypti populations. The Bayesian analysis of worldwide population structure indicated that the most likely number of clusters was one. All three experimental groups could not cleanly be separated into more than a single cluster (Fig. 3), indicating that mixing between groups was occurring.

Average allelic richness per locus was 5.33 in Jagual surface container mosquitoes, 5.08 in Jagual septic tanks, and 5.67 in Centro surface containers. Overall, these values are slightly lower than the average of 7.46 across the subspecies (Brown et al. 2011). Five of 36 population-by-locus specific F_IS values deviated significantly from Hardy Weinberg expectations after sequential Bonferonni correction (Supplemental Table S2 available online only).

Comparison of the Patillas populations to other Ae. aegypti populations using a nonmetric multidimensional scaling analysis based on the pairwise F_st values between populations revealed that these populations were most genetically similar to three mosquito populations in Florida that have previously been analyzed (Fig. 4).