Project 1: mesocosm field experiments
Mesocosm experiments took place at Lockwood Farm located in Hamden, Connecticut. Individual mesocosms were composed of black 20 L cylindrical plastic containers filled with 12 L tap water and seeded with 10 mg of a 3:2 ratio liver powder/brewer’s yeast mixture and 1 g of grass hay. Drain-holes were drilled into the sides of each container 5 mm from the 12 L surface to allow flooding for Aedes spp. egg emergence and to allow overflow beyond this level due to precipitation. Four experimental mesocosm clusters were dispersed throughout the Lockwood Farm in microhabitats previously sampled in Eastwood et al.22. Clusters contained 4 mesocosms spaced 3 m apart in a 2 × 2 grid. We utilized four L. sphaericus treatment levels in each cluster: no L. sphaericus, the LC50 (0.053 ITU/ml) and LC95 (1.0 ITU/ml) for Culex pipiens derived from Burtis et al.3, and the label rate of L. sphaericus (~ 1.2 ITU/ml). All treatments were derived from VectoLex WDG. Prior to insecticide application, we prepared 1 L of a 1000 ITU/ml stock solution. To inoculate each mesocosm, we measured the depth of the container’s water column, calculated water volume, and applied the appropriate amount of stock to achieve the target LC value. Replicate insecticide treatments were randomized within each cluster, and insecticides were applied 30-days post mesocosm seeding with nutrients. All mesocosms in each cluster were rotated within the 2 × 2 grid each week. Two clusters were then randomly chosen for a second application of L. sphaericus 30-days post initial insecticide application.
To sample the larval habitat of each mesocosm, we performed a figure-8 sweep with an aquarium fish net (4 × 3-in. opening, Penn-Plax) each Monday and Thursday of the week for each week of the experiment. Sweep contents were washed from the net into a white photo development pan, and pupae were removed for in-lab identification after eclosion following a dichotomous key23. All larvae were then returned to the mesocosm. This sampling protocol minimized destruction of larval habitats and influence of interspecific interactions due to removal sampling.
In addition to sampling containers for pupae, we collected water samples from each container for an in-lab bioassay to determine the realized mortality of the larval environment. Due to time constraints of the field crew, a 50% randomized sample of containers were sampled on Monday with the remaining 50% sampled on Thursday of each sampling week. Bioassay procedures followed McMillan et al.24 for Cx. pipiens with the addition of screening mortality in CAES’ Ae. albopictus colonies. We finally performed in-lab susceptibility trials to L. sphaericus with larvae from CAES’ Cx. pipiens and Ae. albopictus colonies to confirm each species’ colony varied in their sensitivity to the product. Briefly, 15 3rd to 4th instar larvae of each species per replicate dose were exposed to a wide range of L. sphaericus concentrations and mortality was recorded 24-h post-exposure. Lethal concentrations were then estimated from a generalized linear model with mortality (corrected for mortality in untreated control replicates) as the response term and the log10-dose as the predictor term.
Primary endpoints from the field experiment included the number and species identity of pupae collected from each mesocosm. We compared total weekly pupal collections per mesocosm using a generalized linear mixed model (GLMM) framework with treatment level and cluster ID as fixed effects, species ID and week of collection as a random effect, and a Poisson-error distribution. We repeated this analysis excluding all collected Culex spp. to examine how the L. sphaericus treatments impacted the more tolerant Aedes spp. The primary endpoint for the mortality assays was the corrected larval mortality. We initially compared mortality using a species-specific GLMM with L. sphaericus treatment concentration and treatment period as fixed effects, week of collection as a random effect, and a binomial-error distribution. Preliminary analyses revealed negligible variance attributed to week of collection, so all subsequent models were a GLM. All analyses were performed in R V4.1.325 using the following packages: tidyverse26, gridExtra27, ggplot228, ggeffects29, and glmmTMB30.
Project 2: laboratory competition assays
Competition assays took place at CAES’ main facility in New Haven, CT. This facility contains an Ae. albopictus colony (founded circa 2014 from Stratford, CT) and a Cx. pipiens colony (founded circa 2018 from New Haven, CT;). Colony maintenance for each species was similar: larval rearing pans consisted of approx. 200 eggs (on papers, Ae. albopictus, or as egg rafts, Cx. pipiens) in ~ 2 L RO water and initiated with ~ 20 ml of a 1% 3:2 liver powder/brewer’s yeast slurry. Pans were held at 25.5 °C and 80% humidity and fed ~ 20 ml of the 1% slurry every other day. Pupae were removed to an eclosion chamber and adults were allowed access to 10% sucrose solution ad libitum. Aedes albopictus females were given access to defibrinated sheep’s blood (HemoStat©) through a Hemotek membrane feeder for 1 h every 2–3 weeks and moistened, fluted filter paper was provided to collect eggs. Culex pipiens females were given access to a live, restrained buttonquail overnight once per week and a small cup seeded with 5 ml 1% slurry and 15 RO ml water was provided to collect egg rafts. The use of buttonquail was reviewed and approved in accordance with CAES Institutional Animal Care and Use Committee.
We performed two experiments. All experiments consisted of the following treatments: variable ratios of Ae:Cx larvae and two L. sphaericus treatments (no treatment and 0.01 ITU/ml). Larval density (40 per container) remained constant across all replicate treatments, but Ae:Cx ratios varied from 40/0, 30/10, 20/20, 10/30, and 0/40. Nutrients supplied were a low concentration (3 mg larva−1) of a 3:2 liver powder/brewer’s yeast mix applied at the beginning of the experiment. Temperature was held constant at the colony maintenance level. Assays took place in 300 ml disposable plastic cups filled with 100 ml of RO water. The first experiments consisted of the addition of the 40 larvae as newly hatched individuals (+/− 1 day between species’ hatch) at the appropriate ratios, the larval diet, and the 0.01 ITU/ml concentration (diluted from a lab stock of 1000 ITU/ml). Assays were monitored daily until all larvae were dead and/or all larvae pupated. Experiment 2 consisted of the addition of only the Cx. pipiens larvae and the larval diet. After all Cx. pipiens had pupated, containers were treated with L. sphaericus and then the Ae. Albopictus larvae were added.
Primary endpoints included species-specific pupation success. Preliminary analyses in a GLMM framework revealed negligible variance attributed to a replicate ID random effect; replicate as a random term also interfered with model convergence. Preliminary analyses further revealed there was neither a significant interaction nor an improvement in the Akaike Information Criterion between the L. sphaericus treatment and initial starting condition terms. Thus, we adopted a GLM rather than a GLMM framework in all further analyses, and species-specific mortality was analyzed as a binomial response term with treatment and initial starting conditions included as fixed effects All analyses were performed in R V4.1.325 using the following packages: tidyverse26, gridExtra27, and ggplot228.
Source: Ecology - nature.com
