wMel remains at a high frequency in Cairns following releases

We performed ovitrapping across suburbs in Cairns in 2016 and 2018 to monitor Wolbachia infection frequencies in the Ae. aegypti population. In 2016, all suburbs where wMel releases had taken place had a wMel infection frequency above 0.96 (S1A Fig). Some pre-release suburbs had low infection frequencies (0.05–0.21), indicating spread of wMel to adjacent suburbs (e.g. to Mt Sheridan and Holloways Beach). By 2018, releases had occurred in additional locations (e.g. Trinity Beach and Redlynch) and all suburbs had wMel infection frequencies greater than 0.93 (S1B Fig), except for Redlynch where wMel was not released until 2019. Our data are broadly consistent with Wolbachia infection frequencies from an independent study which also shows that wMel has maintained high frequencies in most release locations [16].

wMel density remains stable across suburbs and laboratory generations

We measured wMel density in 4^th instar larvae reared in the laboratory from ovitraps collected across suburbs in 2018. We found no clear effect of release year (GLM: F_4,269 = 2.274, P = 0.062) or suburb (nested within release year) (F_6,269 = 2.017, P = 0.064) on Wolbachia density (Fig 1A). These results suggest that the wMel infection has remained stable after being established in the field for different periods of time (from 1–7 years), with no clear effects of local environmental conditions on wMel in the next generation of mosquitoes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. wMel density remains stable across suburbs and laboratory generations.

(A) Wolbachia density in 4^th instar larvae hatched from ovitraps collected across Cairns suburbs in February-March 2018. The year where releases were undertaken in each suburb is shown in parentheses. (B-C) Wolbachia density in adult (B) females and (C) males in populations that had been reared in the laboratory for different numbers of generations. Dots represent data from three technical replicates of an individual mosquito. Vertical lines and error bars represent medians and 95% confidence intervals.

https://doi.org/10.1371/journal.ppat.1010256.g001

In June 2019, we measured wMel density in adults from laboratory populations that were established from field collections at different times. We found no significant differences between populations for females (GLM: F_4,65 = 1.767, P = 0.146, Fig 1B) or males (F_4,70 = 1.942, P = 0.113, Fig 1C), suggesting that whole-adult wMel density has not changed across different durations of laboratory rearing.

Population origin and wMel infection influence mosquito fitness

In September 2018 we performed phenotypic comparisons between long-term laboratory and near-field populations that were wMel-infected or cleared of wMel through antibiotic treatment (Fig 2). We found significant effects following Bonferroni correction (adjusted α: 0.008) of population origin on larval development time (females: F_1,44 = 19.656, P < 0.001, males: F_1,44 = 18.266, P < 0.001) and female fecundity (F_1,150 = 11.640, P = 0.001) but not wing length (females: F_1,140 = 6.107, P = 0.015, males: F_1,141 = 0.303, P = 0.583) or (logit transformed) egg hatch proportions (F_1,150 = 0.305, P = 0.581). wMel infection had no significant effect on any trait (all P > 0.089) except for male development time (F_1,44 = 9.296, P = 0.004). For this trait, we also found an interaction between wMel infection and population origin (F_1,44 = 7.958, P = 0.007), where wMel infection increased male development time in the lab populations but not the YK (Yorkeys Knob) populations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phenotypic effects of wMel in laboratory and field Aedes aegypti backgrounds.

wMel-infected populations were compared to tetracycline-cured counterparts to control for genetic background between infected and uninfected mosquitoes. Populations were measured for (A) female and (B) male development time, (C) female and (D) male wing length, (E) fecundity and (F) egg hatch. Data from two replicate populations were pooled for visualization. Dots represent data from replicate trays (A-B) or individual mosquitoes (C-F). Vertical lines and error bars represent medians and 95% confidence intervals.

https://doi.org/10.1371/journal.ppat.1010256.g002

No clear effect of wMel origin in a common host background

In the first experiment, we found that effects on fitness were driven mainly by population origin. Due to potential interactions between Wolbachia infection and background in the first experiment, we performed a second set of experiments in August 2020 to evaluate the effects of lab and field-derived wMel infections in a common host background. We found no significant differences following Bonferroni correction (adjusted α: 0.006) between populations for most traits tested, including development time (GLM: females: F_3,43 = 2.841, P = 0.049, males: F_3,43 = 3.558, P = 0.022), female wing length (F_3,74 = 1.826, P = 0.150), fecundity (F_3,433 = 1.950, P = 0.121) and adult longevity (Log-rank: females: χ² = 5.700, P = 0.127, males: χ² = 3.428, P = 0.330, Fig 3). There was also no significant effect of population on fecundity and (logit transformed) egg hatch (all P > 0.056), except for fecundity in gonotrophic cycle 1 (F_3,107 = 5.847, P = 0.001), where wMel YK females laid more eggs (Fig 3C), and (logit transformed) hatch proportion in gonotrophic cycle 3 (F_3,108 = 5.449, P = 0.002), with wMel GV having reduced egg hatch compared to the other populations (Fig 3F). Population effects were also observed for male wing length (GLM: F_3,73 = 5.293, P = 0.002), with wMel YK males having larger wings (Fig 3E). In the quiescent egg viability experiment, we found no significant effect of population on (logit transformed) hatch proportion after 1 week (GLM: F_4,55 = 1.110, P = 0.361). Egg viability declined more rapidly for wMel-infected populations (Fig 3I), with substantial differences between populations by week 22 (F_4,47 = 35.563, P < 0.001). However, there were no significant differences among wMel-infected populations at the same time point (F_2,27 = 1.156, P = 0.330). Overall, our results demonstrate few consistent and strong effects of wMel origin on mosquito fitness.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Phenotypic effects of wMel from lab and field origins in a common Aedes aegypti background.

wMel-infections originating from Gordonvale (GV), Yorkeys Knob (YK) or the laboratory were introduced to a common background through backcrossing. Populations were measured for (A) female and (B) male development time, (C) fecundity, (D) female and (E) male wing length, (F) egg hatch, (G) female and (H) male adult longevity and (I) quiescent egg viability. Dots represent data from replicate trays (A-B) or individual mosquitoes (D-E). Lines and error bars represent medians and 95% confidence intervals in all panels.

https://doi.org/10.1371/journal.ppat.1010256.g003

Limited effects of antibiotic treatment on fitness

To test whether tetracycline treatment could influence mosquito fitness, we performed phenotypic comparisons of wMel-infected and uninfected mosquitoes following parental tetracycline treatment (Fig 4). We found no significant effect of tetracycline treatment on any trait (all P > 0.007, adjusted α: 0.006) except for (logit transformed) egg hatch proportion where we found an interaction between treatment and Wolbachia infection type (F_1,70 = 9.261, P = 0.003). Decreased egg hatch due to tetracycline treatment was apparent only for the wMel-infected population, likely due to incomplete curing resulting in partial self-incompatibility. In this experiment, we also found no significant effect of Wolbachia infection type for any trait (all P > 0.017, adjusted α: 0.006) except for male wing length, where wMel-infected males had larger wings (F_1,69 = 12.387, P = 0.001). These results provide confidence that observed effects in comparisons between Wolbachia-infected and cured mosquitoes are due to removal of the Wolbachia infection and not off-target tetracycline-related effects on the gut microbiome or mitochondria.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Phenotypic effects of parental tetracycline treatment in Aedes aegypti.

wMel-infected mosquitoes or uninfected mosquitoes were left untreated or fed 2 mg/mL tetracycline. Their offspring were measured for (A) female and (B) male development time, (C) survival to pupa, (D) pupal sex ratio, (E) female and (F) male wing length, (G) fecundity and (H) egg hatch. Dots represent data from replicate trays (A-D) or individual mosquitoes (E-H). Lines and error bars represent medians and 95% confidence intervals in all panels.

https://doi.org/10.1371/journal.ppat.1010256.g004

wMel induces complete cytoplasmic incompatibility after eight years in the field

In June 2019, we tested the ability of wMel-infected males from different origins (Yorkeys Knob, Gordonvale and laboratory) to induce cytoplasmic incompatibility with uninfected females. Males from all wMel-infected populations induced complete cytoplasmic incompatibility, with no eggs hatching in crosses with uninfected females (Table 1). All other crosses yielded egg hatch proportions above 90% (Table 1), consistent with data from the previous experiments showing limited effects of wMel infection on egg hatch. These results show that wMel has retained complete cytoplasmic incompatibility after > 8 years in the field and one generation of laboratory rearing.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Reciprocal crosses between wMel-infected Aedes aegypti from laboratory or field origins and uninfected Ae. aegypti.

Crosses with “-”were not tested.

https://doi.org/10.1371/journal.ppat.1010256.t001

Limited changes in wMel and mitochondrial genomes across a decade in a novel Aedes aegypti host

Previous wMel genome comparisons suggest limited changes in the Wolbachia genome have occurred since field releases [39,40]. We extend this timeframe by including pre-release material from 2010 and field collections from Gordonvale and Yorkeys Knob in 2020. We also include a dataset from a previous whole genome sequencing study of individuals collected from Gordonvale in 2013 and 2018 and Yorkey’s Knob in 2018 [38]. Little evidence of change was observed among the analysed wMel genomes, for which we achieved ≥ 99.99% coverage. Within populations, each of which was represented by a sample of pooled individuals, seven loci were found to have one or more alternate alleles present at a frequency greater than our threshold value of 25% in at least one sample, all of which were SNPs or small indels. At all other positions within the genome, the wC45 F₁₀ (pre-release) consensus sequence was identical to that of the Gordonvale and Yorkeys Knob field collections from 2011, 2018 and 2020, and the wMel lab population (see also Huang et al. [39] and Dainty et al. [40]). Relative to these genomes, one single nucleotide deletion was observed in the wC45 F₉ pre-release genome sequence and seven single nucleotide indels were detected in the 2013 Gordonvale field collection genome sequence (S1 Table). It is probable that these eight indels represent sequencing artifacts, as they involve non-parsimonious changes in populations sampled over multiple time points, were often associated with a marked localized decrease in read depth in the variant samples and mostly represent frameshift mutations.

Of the seven loci that displayed within-sample polymorphism, two were within non-coding regions; three were within three different copies of the group II intron-encoded protein gene LtrA, each at the same position within the gene and corresponding to an Asp > Asn substitution; one was within a gene encoding a hypothetical protein and corresponds to a silent change; and one was within a tRNA-Arg gene (S2 Table). It is possible that the apparent polymorphism within the LtrA genes is due to sequence variation between gene copies, rather than the presence of alternate alleles. Multiple alleles were observed in all samples for these three loci and for the polymorphic locus within the tRNA-Arg gene. At the other three loci, allele frequencies were more variable, with only some samples displaying polymorphism, most of which were from the field populations at Gordonvale and Yorkeys Knob. Two of these loci, at positions 587,862 and 1,174,712 relative to the wMel reference genome, were previously noted by Huang et al. [39] and Dainty et al. [40]. No clear pattern was observed across different populations or time points. While it is possible that this variation is attributable to underlying fluctuation in allele frequencies within populations, or methodological differences between studies, it is more likely to be due to stochastic sampling effects, given the relatively small number of individuals pooled in some samples.

Similar to the wMel genomes, we detected very few changes between the mitochondrial genomes sequenced in this study, for which we obtained ≥ 99.31% coverage. The genomes shared 100% sequence identity across all gene-encoding regions. Nucleotide differences were observed among the consensus sequences at 24 positions within an approximately 1.8Kb region of the genome that does not contain any predicted gene sequences, is very AT rich and displayed greatly reduced read mapping depths (S1 Table). It is therefore likely that many or all of these differences represent sequencing errors. A further 16 positions within this region were found to show within-sample polymorphism, with alternate alleles present at a frequency greater than the threshold value in at least one sample (S2 Table). The lack of variation observed among the wMel genomes and among the mitochondrial genomes is consistent with a high fidelity of maternal co-transmission of both Wolbachia and mitochondria.

Ethics statement

Blood feeding of female mosquitoes on human volunteers for this research was approved by the University of Melbourne Human Ethics Committee (approval 0723847). All adult subjects provided informed written consent (no children were involved).

Field collections and colony establishment

Aedes aegypti mosquitoes were collected as eggs from suburban Cairns in 2016, 2018, 2019 and 2020 from ovitraps. Felt strips from ovitraps were collected and processed identically to previously described methods [24]. Ovitrapping was performed throughout suburban Cairns in September-October 2016 (19 suburbs, 50–100 traps per suburb) and February-March 2018 (12 suburbs, 40–100 traps per suburb), many of which were targeted by wMel release programs from 2011 to 2017 [14,16,49]. The 2019 and 2020 field collections targeted two suburbs with 20 ovitraps each, Gordonvale and Yorkeys Knob, where wMel-infected Ae. aegypti were released in 2011 [14]. Subsets of Ae. aegypti larvae from central Cairns, Gordonvale and Yorkeys Knob were pooled from all traps within a suburb to establish laboratory populations for phenotypic comparisons. Thirty individuals from the F₁ and F₂ generations were screened for Wolbachia infection (see below) to confirm fixation of wMel within the laboratory populations. All populations were maintained at census size of ~450 individuals per generation at 26°C and a 12:12 light:dark cycle as described previously [50]. Female mosquitoes were fed on the forearm of a single human volunteer for egg production.

Wolbachia infection frequency and density

Wolbachia infection frequencies were estimated from the 2016 and 2018 field collections by screening subsets of individuals hatching from ovitraps for Wolbachia. For the 2016 field collections, life stage was uncontrolled, with a mix of adults and larvae tested, thus data were only suitable for Wolbachia frequency estimates. Up to 10 individuals were screened per trap, with between 10 and 243 individuals screened per suburb. For the 2018 field collections, 30 4^th-instar larvae from 15 ovitraps (2 per trap) per suburb were screened for Wolbachia infection. Larvae were reared at a controlled density (50 larvae per tray) and stored at the same age (5 d post-hatching), allowing for a comparison of Wolbachia density across suburbs. In June 2019, we measured wMel density in adults (15 females and 15 males) from laboratory populations that were established from field collections at different times (Gordonvale at F₁, Yorkeys Knob at F₁ and F₁₆, central Cairns at F₉ and the wMel lab population). Aedes aegypti from field collections and laboratory populations were screened for Wolbachia infection status and density using a Roche Lightcycler 480 according to previously described methods [51]. Genomic DNA was extracted with 250 μL of 5% Chelex 100 Resin (Bio-Rad laboratories, Hercules CA) and 3 μL of Proteinase K (20 mg/mL) (Roche Diagnostics Australia Pty. Ltd., Castle Hill New South Wales, Australia). Tubes were incubated for 60 min at 65°C then 10 min at 90°C. Three primer sets were used to amplify markers specific to mosquitoes (mRpS6_F 5’AGTTGAACGTATCGTTTCCCGCTAC3’ and mRpS6_R 5’ GAAGTGACGCAGCTTGTGGTCGTCC3’), Ae. aegypti (aRpS6_F 5’ATCAAGAAGCGCCGTGTCG3’ and aRpS6_R 5’CAGGTGCAGGATCTTCATGTATTCG3’), and wMel (w1_F 5’AAAATCTTTGTGAAGAGGTGATCTGC3’ and w1_R 5’ GCACTGGGATGACAGGAAAAGG3’). Relative Wolbachia densities were determined by subtracting the crossing point (Cp) value of the wMel-specific marker from the Cp value of the Ae. aegypti-specific marker. Differences in Cp were averaged across 3 consistent replicate runs, then transformed by 2ⁿ.

Phenotypic comparisons

We performed two sets of experiments to evaluate the phenotypic effects of wMel derived from field and laboratory populations. In the first set, populations were cured with tetracycline to remove the wMel infection, maintaining similar genetic backgrounds between infected and uninfected lines. In the second set, we used backcrossing to introduce the wMel infection from different origins to a common genetic background, then compared populations to uninfected lines that had been crossed to the same background. Both sets of experiments involved the wMel Lab population which was collected from Cairns in 2014 and had spent at least 60 generations in the laboratory before the first set of experiments commenced. All experiments were performed at 26°C and a 12:12 light:dark cycle.

Experiment 1 was performed in September 2018 using wMel-infected populations collected from Yorkeys Knob in February 2018 (wMel YK) and Cairns in 2014 (wMel Lab). wMel YK and wMel Lab were divided into four population cages each. Two replicate populations from each line were treated for three consecutive generations with tetracycline hydrochloride (2 mg/mL) provided to adults in 10% sucrose solution to cure the wMel infection. Females were blood fed at 10 d old to ensure that they had fed on the antibiotic solution prior to blood feeding. The other two replicate populations from each line were left untreated but reared synchronously. All populations were reared in the absence of antibiotics for two generations before experiments commenced, with thirty adults from each population screened for Wolbachia infection to ensure complete removal of wMel in the treated lines (wMel YK.tet and wMel Lab.tet) and fixation of wMel in the untreated lines (wMel YK and wMel Lab). The wMel YK and wMel YK.tet populations were at F₈ in the laboratory when experiments commenced.

Experiment 2 was performed in August 2020 using wMel-infected populations collected from Yorkeys Knob in February 2020 (wMel YK), Gordonvale in February 2020 (wMel GV), Cairns in 2014 (wMel Lab), as well as an uninfected population that had been cured of wMel in the previous experiment (wMel Lab.tet). Two hundred females from each population were crossed to 200 males from a natively uninfected population collected from locations in Cairns prior to wMel releases. This process was repeated for two further generations to produce a similar nuclear background between populations. The backcrossed wMel YK and wMel GV populations were at F₅ in the laboratory when experiments commenced.

In both experiments, we measured larval development time, adult wing length, female fecundity and egg hatch proportions. Eggs (<1 week old) from each population were hatched in reverse osmosis (RO) water and 100 larvae (<1 d old) were counted into plastic trays filled with 500 mL RO water (with 6 replicate trays per population in experiment 1 and 12 replicate trays per population in experiment 2). Larvae were fed TetraMin Tropical Fish Food tablets (Tetra, Melle, Germany) ad libitum (with daily monitoring and removal of excess food to prevent overfeeding) until pupation. Development time was scored by counting and sexing pupae twice per day. Adults were pooled across replicate trays and released into cages. Wings of adults (20 males and 20 females per population) were dissected and measured for their length (from the alular notch to the wing tip). Females (5–7 d old, sugar-starved for 1 d) were blood fed and isolated in 70 mL specimen cups filled with 15 mL of larval rearing water, lined with a strip of sandpaper (Norton Master Painters P80; Saint-Gobain Abrasives Pty. Ltd., Thomastown, Victoria, Australia) and covered with a mesh lid. Twenty females were isolated per population in experiment 1. In experiment 2, fecundity and egg hatch proportions were tracked across four consecutive gonotrophic cycles by isolating 30 engorged females per population, returning females that laid eggs to cages, then blood feeding and isolating 30 engorged females again every 4–5 d. Eggs were collected 4 days after blood feeding, partially dried, then hatched 3 d after collection in trays filled with RO water and a few grains of yeast. Fecundity and egg hatch proportions were determined by counting the total number of eggs and the number of hatched eggs (with the egg cap clearly detached) under a dissecting microscope. In experiment 2, we also measured adult longevity (8 replicate cages of 25 males and 25 females per population) by maintaining adults in 3 L cages with cups of 10% sucrose and scoring mortality 3 times per week. Quiescent egg viability was measured in experiment 2 by storing eggs on sandpaper strips in a sealed container with a saturated solution of potassium chloride to maintain a constant humidity of ~80%. Twelve replicate batches of eggs per population (median 68 eggs) were hatched on week 1 and 2, then every two weeks until week 22. Egg hatch proportions were scored as above for individual females.

Effects of tetracycline treatment on fitness

In the above experiments we used populations that had been cleared of Wolbachia infections through tetracycline treatment. Despite allowing for at least two generations of recovery, these treatments could potentially disrupt the microbiome (including the mitochondria), leading to fitness differences between lines that are independent of Wolbachia infection. To test whether antibiotic treatment has any effect on fitness, we fed wMel-infected (wMel Lab) and natively uninfected (F₂₉ in the laboratory) adults with 2 mg/mL tetracycline hydrochloride for 10 d before blood feeding, then measured fitness in the subsequent generation. We scored larval development time, survival to pupa and sex ratio (6 replicate trays per treatment) as well as female and male wing length, female fecundity and egg hatch proportions (20 individuals each). Experiments were performed identically to the phenotypic comparisons above.

Cytoplasmic incompatibility

We tested the ability of wMel-infected Ae. aegypti to induce cytoplasmic incompatibility in near-field and laboratory populations. Reciprocal crosses were performed in June 2019 between wMel-infected populations established from Gordonvale at F₁ (wMel GV F₁), Yorkeys Knob at F₁ (wMel YK F₁) or Cairns at F₆₀₊ (wMel Lab), and a natively uninfected population (F₃₅ in the laboratory). Pupae from each population were sexed and released into separate 3 L cages to confirm accurate sex sorting. Groups of 40 females and 40 males (1 d old) were then aspirated into cages together and left for 5 d to mate. Females (starved of sugar for 24 hr) were blood-fed, isolated for oviposition and scored for fecundity and egg hatch proportions according to the methods described above for the phenotypic comparisons.

Wolbachia and mitochondrial whole genome sequencing

Previously, we sequenced the wMel and mitochondrial genomes of Ae. aegypti collected up to eight years after field releases [39]. Here, we extend these findings by sequencing the Wolbachia and mitochondrial genomes of wMel-infected Ae. aegypti sampled a decade apart from pre-release and post-release populations. The pre-release wMel-infected populations (wC45 F₉ and F₁₀) were sampled in 2010 and stored at -20°C, while three wMel-infected populations were sampled in 2020: wMel GV F₂ collected from Gordonvale in 2020, wMel YK F₂ collected from Yorkeys Knob in 2020, and wMel Lab, collected from Cairns in 2014 and maintained under laboratory conditions until sampling. Genomic DNA was extracted from pooled samples containing five adult females. Sequencing libraries were then prepared as described previously [52].

Reference genome assembly

Sequencing reads were quality filtered using Trimmomatic [53]. The samples were trimmed in paired-end mode with the following parameter settings: leading = 20; trailing = 20; slidingwindow = 4:20; minlen = 70; adapter sequences were removed using the ILLUMINACLIP option, with maximum seed mismatches = 2 and the palindrome clip threshold = 30. Reads were aligned to a wMel reference genome (GenBank accession: NC_002978.6) and an Ae. aegypti mitochondrial reference genome (GenBank accession: MH348177.1) with the Burrows-Wheeler Aligner (BWA; [54]) using the bwa mem algorithm and default parameter settings. Quality filtering of alignments and variant calling was performed with SAMtools and BCFtools [55,56]. PCR duplicates were excluded from subsequent analyses by soft masking. Reads with a MAPQ score < 25 were removed from the alignment, except for reads with MAPQ = 0, which were permitted to allow for mapping to repetitive regions. Genotype likelihoods were calculated using a maximum of 2000 reads per position. For variant calling, ploidy was set to haploid. The variant call output was used to create a consensus nucleotide sequence, wherein genome positions with coverage < 5 were masked as ‘N’. Loci were considered to be polymorphic within a sample if they had coverage ≥ 30 and two or more alleles with a frequency of ≥ 25%. Any such loci occurring within the 16S rRNA or 23S rRNA genes were excluded from analysis, as we observed a relatively high level of contaminant reads mapping to these regions. Genome sequences were inspected and aligned with Geneious v 9.1.8 (https://www.geneious.com).

Kraken2 [57] and the Standard-8 precompiled reference database (https://benlangmead.github.io/aws-indexes/k2; downloaded 17/9/21) were used to search for sequence contamination within the Wolbachia genomes. The sequencing reads mapped by bwa to the wMel reference genome were filtered to remove reads matching taxa other than Wolbachia, and genome assemblies were then repeated with the filtered datasets, using the above pipeline. The original pre-filtration genome sequences were edited to correct erroneous positions after comparison with the corresponding post-filtration genome sequences.

Statistical analysis

Most experimental data (including Wolbachia density, development time, wing length, fecundity and egg hatch) were analyzed with general linear models (GLMs) while adult longevity data were analysed with log-rank tests. All analyses were carried out with IBM SPSS Statistics 26. Data were transformed where appropriate (with all proportional data being logit transformed). The first experiment had two replicate populations and data were initially analysed with replicate population (nested within population origin x Wolbachia infection status) included as a factor. Replicate populations were then pooled for a second analysis due to a lack of significant effect of replicate population (P > 0.1) for any trait. Wolbachia infection status and population origin were included as factors. In the second experiment, fecundity and egg hatch proportions were tracked across gonotrophic cycles with the same mosquitoes, so we ran separate analyses for each cycle. For quiescent egg viability, wMel-infected and uninfected populations were analysed both together and separately at the first (1 week) and last time (22 weeks) points. Field Wolbachia density data were analysed with release year and suburb (nested within release year) as factors. We performed Bonferroni corrections where multiple traits were evaluated in the same cohort of mosquitoes.