The influence of oviposition status on measures of transmission potential in malaria-infected mosquitoes depends on sugar availability

Background Like other oviparous organisms, the gonotrophic cycle of mosquitoes is not complete until they have selected a suitable habitat to oviposit. In addition to the evolutionary constraints associated with selective oviposition behavior, the physiological demands relative to an organism’s oviposition status also influence their nutrient requirement from the environment. Yet, studies that measure transmission potential (vectorial capacity or competence) of mosquito-borne parasites rarely consider whether the rates of parasite replication and development could be influenced by these constraints resulting from whether mosquitoes have completed their gonotrophic cycle. Methods Anopheles stephensi mosquitoes were infected with Plasmodium berghei, the rodent analog of human malaria, and maintained on 1% or 10% dextrose and either provided oviposition sites (‘oviposited’ herein) to complete their gonotrophic cycle or forced to retain eggs (‘non-oviposited’). Transmission potential in the four groups was measured up to 27 days post-infection as the rates of (i) sporozoite appearance in the salivary glands (‘extrinsic incubation period' or EIP), (ii) vector survival and (iii) sporozoite densities. Results In the two groups of oviposited mosquitoes, rates of sporozoite appearance and densities in the salivary glands were clearly dependent on sugar availability, with shorter EIP and higher sporozoite densities in mosquitoes fed 10% dextrose. In contrast, rates of appearance and densities in the salivary glands were independent of sugar concentrations in non-oviposited mosquitoes, although both measures were slightly lower than in oviposited mosquitoes fed 10% dextrose. Vector survival was higher in non-oviposited mosquitoes. Conclusions Costs to parasite fitness and vector survival were buffered against changes in nutritional availability from the environment in non-oviposited but not oviposited mosquitoes. Taken together, these results suggest vectorial capacity for malaria parasites may be dependent on nutrient availability and oviposition/gonotrophic status and, as such, argue for more careful consideration of this interaction when estimating transmission potential. More broadly, the complex patterns resulting from physiological (nutrition) and evolutionary (egg-retention) trade-offs described here, combined with the ubiquity of selective oviposition behavior, implies the fitness of vector-borne pathogens could be shaped by selection for these traits, with implications for disease transmission and management. For instance, while reducing availability of oviposition sites and environmental sources of nutrition are key components of integrated vector management strategies, their abundance and distribution are under strong selection pressure from the patterns associated with climate change. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s13071-024-06317-2.


Background
Like other oviparous organisms, mosquitoes retain their eggs until they find a suitable habitat to oviposit and complete its gonotrophic cycle.Among oviparous organisms, however, mosquitoes occupy an important niche: most species rely exclusively on hematophagy (blood feeding) of vertebrates for producing eggs, which enables them to vector several parasites with devastating consequences for the health and fitness of humans and wildlife.Following ingestion of an infected blood meal, egg and parasite development may be initiated at the same time in the mosquito.However, by the time a mosquito has completed egg development, parasite development may or may not be complete, although transmission to the next host is dependent on the mosquito locating an oviposition site to complete its 'gonotrophic cycle' [1].While several studies suggest widespread changes in mosquito behavior and physiology associated with the gonotrophic cycle [2][3][4][5][6][7][8], few have determined if and how changes in oviposition status influences parasite development rates and transmission potential [9]; in general, measures of parasite fitness and transmission potential are performed in mosquitoes prevented from completing their gonotrophic cycle (e.g.see [10][11][12][13] and references therein).A greater understanding of how parasite fitness is shaped by oviposition status of mosquitoes is important for at least two reasons.First, integrated vector management strategies either directly or indirectly alter the availability of oviposition sites by managing water sources, for instance, to identifying specific physical, chemical and biological cues underlying selective oviposition behavior for targeting with chemical or biological means [1,[14][15][16][17][18][19].Second, the availability of oviposition sites will continue to change as a cumulative response to climatic factors such as global warming and changes in rainfall patterns to the increasing urbanization and deforestation [20][21][22][23][24].By altering the number and distribution of oviposition habitats, both scenarios could select for variation in oviposition behavior and thus modify parasite/ disease transmission in the process.
Like other oviparous organisms, adult female mosquitoes also show selective oviposition behavior [1,16].Offspring survival and reproduction are benefited by ovipositing mothers choosing sites that fulfill several criteria, including for instance habitat size, quality, risk of predation and presence of conspecifics.The ubiquity of selective oviposition behavior attests to its benefits to future reproduction, with selection for this trait imposed at the cost of reduced fecundity for instance [25][26][27][28][29][30].While several studies have shown how variation in this behavior could modify the oviparous organism's own fitness [26][27][28][30][31][32][33][34][35], few have assessed whether the associated trade-offs could have downstream effects on the fitness of other organisms, such as parasites, whose life history strategy is driven by its interaction with the oviparous host.
Organismal fitness is also dependent on how it allocates available resources from the environment to meet the costs of various life history traits [25-27, 29, 36].For vector-transmitted parasites, although their growth and reproduction are dependent on how they allocate resources from the mosquito host, the host's own fitness is also dependent on how they allocate resources from environmental sources of nutrition; in other words, nutritional availability from the environment can shape life history strategies of the vector and parasite [37,38].Among environmental sources of nutrition, the effects of sugars on life history traits of adult female mosquitoes are particularly well characterized, with their propensity for sugar feeding also exploited for controlling vector density (e.g.toxic sugar baits) [39].While the costs of parasite infection for the mosquito are more apparent at lower concentrations of sugar, in general, sugar requirements are greater in gravid/non-oviposited mosquitoes foraging for oviposition sites, carrying and/or maintaining eggs [25,38,[40][41][42].Differences in sugar concentrations and/or availability to the mosquito have also been shown to affect some parasite life history traits, especially for Anopheles mosquitoes infected with the unicellular, eukaryotic microparasites of the genus Plasmodium [38]; these parasites are the etiological agent of malaria, a disease that continues to extract significant costs in morbidity and mortality in humans [43].In malariainfected mosquitoes, Plasmodium oocyst densities in the midguts and mosquito survival were dependent on dextrose ('D-glucose') concentrations [44].Taken together, these observations suggest that differences in nutritional requirements between oviposited and non-oviposited mosquitoes could influence parasite fitness.
In principle, a single Plasmodium parasite ingested with a blood meal can invade the midgut wall of an adult female Anopheles and differentiate into an oocyst.
Within the oocyst, in a process termed 'schizogony, ' several rounds of asexual replication culminate in the generation of sporozoites.Once mature, the oocysts rupture and sporozoites are released into the mosquito hemocoel, from where they migrate to the salivary glands, ready for injection during the next blood meal.In general, transmission to the next host (i.e.future reproduction) is dependent on three traits: (i) the vector surviving past the time required for sporozoites to invade the salivary glands, (ii) rates of sporozoite migration to the salivary gland ('extrinsic incubation period' , EIP herein) [45,46] and (iii) the number of sporozoites in the salivary glands [47].Although a single oocyst can produce thousands of sporozoites, all three fitness traits (vector survival, rates of migration and the overall densities) are dependent on trade-offs with other traits in the parasite and vector's life history strategy [45,[48][49][50][51].The current study tests the hypothesis that parasite fitness would be dependent on not only whether the mosquito has oviposited or not (oviposition status) but also the amount of nutrition available.Female Anopheles stephensi mosquitoes were exposed to an infectious blood meals containing Plasmodium berghei [49,50].Vector survival, rates of sporozoite migration and densities were measured in mosquitoes that were (or not) allowed to oviposit, and whether this was affected by differences in nutritional availability (1% or 10% dextrose) was examined [40].

Methods
All reagents were purchased from Thermo Fisher unless stated otherwise.

Study design
For a visual overview of the study design, refer to Fig. 1A.Based on procedures described in detail previously [49,50], 1500 adult female An.stephensi housed in a 32.5-cm 3 cage were infected with P. berghei-infected mice [0 days post-infection (dpi herein)].At day 1 post-infection, ~ 350 mosquitoes were transferred to each of 4 cages (24.5 cm 3 ) for the respective treatments.Transfers were performed by repeatedly withdrawing 30-40 mosquitoes from the large cage and distributing them to one of the four cages at random until the total in each cage reached ~ 350 mosquitoes.Two cages were provided a low-nutrition diet (1% dextrose), while the remaining two were offered the high-nutrition diet (10% dextrose).At 6 dpi, one cage from each nutrient treatment was provided oviposition sites until 9 dpi [50].The overall study lasted 27 days and employed a fully crossed design wherein the same starting population of P. berghei-infected mosquitoes (and associated differences/biases in initial infection) was distributed across four treatment groups; in other words, in principle, all measures of transmission potential were obtained from the 'same' group of mosquitoes, after being subjected to the following four treatments: (i) oviposited mosquitoes with low nutrient availability (1% dextrose), (ii) oviposited mosquitoes with high nutrient availability (10% dextrose), (iii) non-oviposited mosquitoes with low nutrient availability and (iv) non-oviposited mosquitoes with high nutrient availability (Fig. 1A).
Vector mortality, rates of P. berghei sporozoite prevalence and sporozoite densities in the salivary glands were quantified essentially as described below ('Data collection') [49,50].While vector mortality and EIP are wellknown higher order components of the vectorial capacity model [45,46], sporozoite densities in the salivary glands are increasingly recognized as a critical fitness-conferring trait in the next vertebrate host [47].Mosquito mortality was monitored daily in all four cages from day 1 post-infection [50].To determine rates of sporozoite prevalence [50], salivary glands from 15-17 mosquitoes from each group were checked for presence or absence of sporozoites at 3-day intervals starting 9 dpi (i.e.days 9, 12, 15, 18, 21, 24 and 27 post-infection).To enumerate sporozoite densities [49], salivary glands from 10 mosquitoes from each group were sampled from days 15 to 27 post-infection, also at 3-day intervals (i.e. 15 18, 21, 24 and 27 dpi); note that due to differences in assay procedures (see below), sporozoite prevalence [50] and densities [49] were obtained from different mosquitoes in the same group.Finally, to determine whether mosquitoes had oviposited or not, ovaries of each sampled mosquito were assessed for absence (oviposited) or presence (nonoviposited) of eggs (Fig. 1B).

Mosquito husbandry
Laboratory colonies of An. stephensi mosquitoes (Walter Reed Army Institute of Research ca. 2015, "Indian" strain) were maintained in a level 2 arthropod containment laboratory at the University of Georgia, as described previously [52].Briefly, colonies were maintained at 27 °C ± 0.5 °C, 80% ± 5% relative humidity, and under a 12-h day/night photoperiod.Female mosquitoes were provided with human whole blood (Interstate Blood-Bank, Memphis, TN) with glass membrane feeders (Chemglass Life Sciences, NJ) to support egg production.Hatched larvae were dispensed at a density of 300 L1s/1000 ml of water and maintained on pelleted Hikari cichlid gold diet (Hikari USA, Hayward, CA).Adult colonies were maintained on sugar water composed of 5% dextrose (w/v) and 0.05% para-aminobenzoic acid (PABA) (w/v).

Plasmodium berghei infections of mice
All animal procedures described herein were performed as per AUP number A2020 01-013-Y3-A10, approved by Fig. 1 A Overview of the study design.Briefly, a pool of 1500, Plasmodium berghei-infected mosquitoes was generated in a 32.5-cm 3 cage (day 0 post-infection) and randomly sorted into four cages (24.5 cm 3 ) on 1 dpi, with two cages provided each of 1% or 10% dextrose.On 6 dpi, cages for each sugar concentration were provided an oviposition cup ('oviposited' herein) or prevented from ovipositing ('Non-oviposited' herein).Three key measures of transmission potential were obtained for each group at the indicated dpi, with rates of sporozoite prevalence and densities estimated from 15-17 mosquitoes and 10 mosquitoes, respectively, and survival estimated by recording the number of dead mosquitoes daily in each group.B Representative images of ovaries from a P. berghei-infected mosquito that had oviposited successfully (left pane, 'Oviposited') and another that still retained eggs (right pane, 'Non-oviposited') UGA IACUC.Four-to 6-week-old male, Hsd:ICR(CD-1) mice (Envigo, Indianapolis, IN) were infected with P. berghei GFP-LUC CON , a transgenic strain expressing GFP and Luciferase reporter genes in an ANKA background (referred to herein as P. berghei), as described previously [49].Briefly, seven mice were infected with a dose of 5 × 10 6 parasites/ml in a 500-µl inoculum.Once parasitemia reached 3-6% (4 dpi), mice were anesthetized with 1.8% tri-bromo ethanol (Avertin, Sigma-Aldrich, MO) just prior to mosquito feeds.

Plasmodium berghei infections of mosquitoes
At 3 to 7 days post-emergence, 1500 female, actively host-seeking mosquitoes were sorted into a 32.5-cm 3 cage (BugDorm, Taiwan) and transferred to a 20 °C environmental chamber (Percival Scientific, Perry, IA).On the day of infection (day 0 post-infection), anesthetized P. berghei-infected mice were placed atop the cage and mosquitoes allowed to feed on the mice for 15 min.After 24 h post-infection (or day 1), visibly blood-fed mosquitoes were distributed into four separate cages (24.5 cm 3 , BugDorm, Taiwan), as described previously [50].Briefly, 30-40 mosquitoes were repeatedly withdrawn from the large cage with pooled mosquitoes with a HEPA-filtered mouth aspirator (John W Hock Company, Gainesville, FL) and distributed to one of the four cages at random until the total in each cage reached ~ 350 mosquitoes.Mosquitoes were maintained for the duration of the experiment as described above ('Study design').

Data collection
While daily rates of mortality were recorded for all four treatments, starting 9 dpi and up to 27 dpi, the salivary glands of mosquitoes from each group were checked for the presence or absence of sporozoites as described previously [50].Briefly, mosquitoes were aspirated directly into 70% ethanol (v/v) before salivary glands were dissected from the mosquito and transferred to a 3-μl drop of PBS on a clean glass slide.Salivary glands were ruptured to release sporozoites by placing a cover glass directly on the drop of PBS; presence or absence of sporozoites was assessed on an upright DIC microscope (Leica DM2500) at 100 × or 400 × magnification.Salivary glands were sampled from 15 to 17 mosquitoes in each group at days 9, 12, 15, 18, 21, 24 and 27 post-infection.
Sporozoite densities in the salivary glands were enumerated as described previously [49].Briefly, at days 15, 12, 18, 21, 24 and 27 post-infection, ten mosquitoes from each group were terminally sampled by aspirating them directly into 70% ethanol.Salivary glands from each mosquito were resuspended in 50 μl of PBS supplemented with 0.1% bovine serum in a 0.65-ml tube.Sporozoites were released by homogenizing the salivary glands with 10 strokes of a microtube pestle.Sporozoites were quantified in a 10-μl aliquot of the homogenate as described previously [49]; while this approach can detect ≥ 125 sporozoites, the relatively laborious and time-consuming nature of this assay meant quantification of sporozoite densities was not initiated until 15 dpi when appreciable (and thus more reliable) quantities of sporozoites were available for enumeration.

Data analyses
All data analyses were performed with generalized linear models (GLM) in JMP Pro software (ver 16).In general, prevalence data were modeled with a binomial distribution, and count data were modeled with a type 2 negative binomial distribution [53].For sporozoite prevalence and densities, time was standardized and treated as a continuous fixed effect.Oviposition status (yes/no) and nutrient levels (1% vs. 10% dextrose) were treated as categorical fixed effects.The effect of time was also modeled as a quadratic effect ('humped') to account for saturation and/or decline in the dependent variable trends.Gravid status (presence or absence of eggs) was analyzed with the same fixed effects described above using a binomial distribution.Mosquito survival was analyzed using the Cox proportional hazards model with the interaction between oviposition status and nutrient levels as fixed effects.Mosquitoes that were destructively sampled for parasite quantification were considered censored in the Cox analysis.The extrinsic incubation period (EIP) for each group expressed as the time from initial infection to when 10 (EIP 10 ), 50 (EIP 50 ) and 90% (EIP 90 ) of the population became infectious, as recommended previously [45].Briefly, EIPs were estimated by interpolating from the binomial model for the dynamics of sporozoite prevalence in the salivary glands above (presence or absence of sporozoites).However, since EIP estimations are based on a logistic ('S-shaped') relationship that eventually saturates (dpi in this study), any declines in prevalence after this point in time introduces non-linearities that are difficult to model with this approach [45]; as such, predictions were restricted to 24 days post-infection (starting from 9 dpi) when our model suggested prevalence approached saturation/maximum, before beginning to decline for most of the groups.
The analyses also considered the possibility that not all mosquitoes in the respective groups may respond to the two oviposition treatments and, as such, could potentially confound analyses and comparisons between groups.Assessing the contribution of these non-responders was particularly relevant for correlations of sporozoite dynamics with the vector survival data where it was not possible to quantify oviposition status post hoc [50].For instance, while some mosquitoes may still retain eggs despite being offered oviposition sites, in the groups not offered an oviposition site, some mosquitoes may nonetheless oviposit.However, if the numbers of these nonresponders were high enough, their contribution could skew/distort the estimates of the means for the respective groups (e.g.non-oviposited individuals distorting means for the groups offered an oviposition site), which in turn could compromise statistical analyses and ultimately affect our inference and conclusions.To address this, mean sporozoite prevalence or mean sporozoites densities from all four groups at each time point were quantified from all mosquitoes irrespective of whether they responded to the treatment.Pearson's product-moment correlations were performed with the mean sporozoite prevalence and densities from the same groups but after excluding non-responder mosquitoes from the respective groups, i.e. by excluding non-oviposited (or gravid) mosquitoes from the groups offered oviposition site or oviposited mosquitoes (not gravid) from the groups not offered oviposition sites.

Data and treatment summary
For the duration of the study, a total of 616 An. stephensi mosquitoes were sampled from all four groups, with 416 sampled for EIP and 200 for sporozoite densities.To determine the effectiveness of the oviposition treatments, gravid status from each mosquito (presence or absence of eggs) was assessed at the time of dissection.In the groups that were offered oviposition sites ('oviposited' herein), 89.1% of mosquitoes successfully laid eggs (or 271 of 304 mosquitoes dissected), with 90.5% (133 of 147 mosquitoes) and 87.9% (138 of 157 mosquitoes) oviposited mosquitoes recovered from low (1% dextrose) and high (10% dextrose) nutrition groups, respectively, over the duration of the experiment (Additional file 1: Fig. S1A).While similar number of oviposited mosquitoes were recovered at each time point (Chi-squared test (χ 2 ) = 0.02, df = 1, P = 0.88), recovery was also not affected by nutritional availability over time (χ 2 = 0.22, df = 1, P = 0.64) (Additional file 1: Fig. S1B) or nutritional availability in general (χ 2 = 0.49, df = 1, P = 0.48) (Additional file 1: Fig. S1A).
In the groups that were not offered an oviposition site ('non-oviposited' herein) (Additional file 1: Fig. S1), 78.5% of the mosquitoes still retained eggs in their ovaries (or 245 of 312 mosquito ovaries assessed), with 84.8% of mosquitoes retaining eggs in the groups offered low nutrition (134 of 158 mosquitoes) higher than the 72.1% in the high nutrition group (111 of 154 mosquitoes) (χ 2 = 6.86, df = 1, P = 0.008) (Additional file 1: Fig. S1A) (Additional file 2: Table S1).However, this difference was due to the apparent decline at day 24 post-infection (right pane, Additional file 1: Fig. S1B): excluding this sampling point from the analysis suggested that while nutrient availability did not affect recovery of non-oviposited mosquitoes (χ 2 = 0, df = 1, P = 0.9686), similar numbers of non-oviposited mosquitoes were recovered at the chosen sampling times (χ 2 = 0.22, df = 1, P = 0.6322).As the reason for this discrepancy was unknown, all statistical analyses were performed with this sampling point included in the dataset.
Despite the presence of these mosquitoes in the respective groups (i.e.oviposited individuals in groups not offered an oviposition cup or non-oviposited individuals in groups offered oviposition cups), their numbers were not high enough to compromise downstream statistical analysis, with both the dynamics of sporozoite prevalence (Additional file 3: Fig. S2A) and densities (Additional file 3: Fig. S2B) generally comparable between the groups after excluding these non-responder individuals or when all individuals were included.Pearson's correlation analysis of the means before and after excluding the 'non-responders' suggested that, in general, the presence of these individuals should not confound statistical comparisons of the means among the four groups: compared to the estimations of mean sporozoite prevalence and densities from all four groups over time irrespective of oviposition status, excluding these individuals did not significantly alter the estimation of prevalence (Pearson's correlation coefficient, r (26) = 0.99, P < 0.001) or sporozoite densities (Pearson's r (18) = 0.96, P < 0.001).Thus, all analyses below were performed with datasets that consider all individuals irrespective of whether they responded to the oviposition sites as expected; furthermore, this approach also ensures compatibility with the vector survival data where it was not possible to discriminate between mosquitoes that may or may not have responded as expected to the oviposition sites.

Rates of sporozoite prevalence and time to infectiousness (EIP) in salivary glands is dependent on nutritional availability in mosquitoes that have oviposited but not those that retain eggs
Of the 616 mosquitoes, salivary glands from a total of 416 mosquitoes from all four groups were assessed for the presence or absence of sporozoites at 9, 12, 15, 18, 21, 24 and 27 dpi (Fig. 2A).Differences in nutritional availability to oviposited and non-oviposited mosquitoes affected the rates of sporozoite prevalence in the salivary glands (χ 2 = 5.45, df = 1, P = 0.02) (row 8, Table 1) (Fig. 2A) as well as the overall proportion of sporozoite-positive salivary glands (χ 2 = 10.67,df = 1, P = 0.001) (row 5, Table 1) (Fig. 2B).This effect was particularly apparent in oviposited individuals that were maintained at 1% dextrose (left pane, Fig. 2A and B).Indeed, pairwise comparisons of the marginal means predicted by the model suggested a consistently lower prevalence in this group compared to the other three groups after 15 dpi, with differences in estimated probabilities predicted to range from -0.55 to -0.23 (rows 1-21, Additional file 5: Table S2).While nutritional availability did not appear to result in clear differences between non-oviposited mosquitoes, for the groups maintained at 10% dextrose, mosquitoes were infectious earlier in the oviposited group, with higher prevalence at 15 dpi [estimated difference in probability (Est.herein) = 0.34, standard error (SE) = 0.1, P = 0.004) and to a lesser yet significant extent at 18 dpi (Est.= 0. Fig. 2 A Rates of sporozoite prevalence in the salivary glands ('Oviposition status*nutrient levels*dpi' , χ 2 = 5.45, df = 1, P = 0.02, row 8, Table 1), B mean of sporozoite prevalence over all dpi ('Oviposition status*nutrient levels' , χ 2 = 10.67,df = 1, P = 0.001, row 8, Table 1) and C estimated time to infectiousness (EIP) were clearly higher in oviposited mosquitoes maintained on 10% dextrose (blue); in the two groups of non-oviposited mosquitoes forced to retain eggs, however, while sugar availability did not influence overall sporozoite prevalence (B), the dynamics of sporozoite prevalence (right pane, A) and predicted EIP (right pane, C) indicated higher parasite fitness in mosquitoes fed 1% dextrose (yellow) in contrast to the two oviposited groups.For post hoc pairwise comparisons of the estimated marginal means of the four groups at each day post-infection, refer to Additional file 4: Table S2.EIP was expressed as time until 10% (EIP 10 ), 50% (EIP 50 ) and 90% (EIP 90 ) of the population became infectious [45], with values listed in the plotted area of C derived by interpolating from the trends predicted by the statistical model of the raw data in panes A and B (Table 1); predictions were restricted to 24 dpi to facilitate interpolations from the logistic portion of the relationship.In A and B, data points and error bars show mean and standard errors respectively (from 15-17 salivary glands/group at each dpi), while the lines and shaded areas in C depict the predicted mean and 95% confidence intervals respectively SE = 0.1, P = 0.048) (rows 38 and 39, Additional file 4: Table S2]. Estimates of the EIP from the sporozoite prevalence model above suggested EIPs were dependent on the interaction between oviposition status and nutrient availability (Fig. 2C).In the two oviposited groups for instance EIPs were clearly shorter in mosquitoes maintained on 10% dextrose, with the model suggesting 10%, 50% and 90% of the population becoming infectious by 11.8, 14.8 and 18.3 dpi, respectively (blue lines and shaded area, left pane, Fig. 2C).While this group (oviposited, 10% dextrose) also showed the shortest EIPs overall, the EIPs in the oviposited groups were shorter for the 1% group (right pane, Fig. 2C) but were generally less impacted by sugar availability compared to the non-oviposited counterparts.
The dynamics and overall sporozoite prevalence were generally higher in oviposited mosquitoes maintained at 10% dextrose compared to the other groups, albeit not as clearly as the oviposited groups maintained on 1% dextrose (Additional file 4: Table S2).In other words, the low prevalence in the 1% dextrose group was compensated by the high prevalence in the 10% dextrose group such that the mean prevalence in oviposited groups (mean = 36.8%,SE = 0.08) was comparable to the non-oviposited groups (mean = 43.4%,SE = 0.1), which could explain why oviposition status alone did not clearly influence prevalence (χ 2 = 3.14, df = 1, P = 0.076) (Table 1).In contrast, the clear effect of nutrition on prevalence (χ 2 = 12.37, df = 1, P = 0.008) (Table 1) was likely driven by the strong reduction in prevalence in the oviposited group maintained on 1% dextrose, which, despite the higher prevalence in the non-oviposited mosquitoes also maintained on 1% dextrose, resulted in mean prevalence of 32.9% (SE = 0.08) that was 14.5% lower than the mean of the groups maintained at 10% dextrose (mean = 47.4%,SE = 0.1); this difference was still greater than the 6.8% difference in mean prevalence between the two oviposition treatments.Finally, the analyses also suggested that when averaged over all four groups, the initial increase in proportion of sporozoite-positive salivary glands recovered (χ 2 = 33.94,df = 1, P < 0.001) was eventually followed by saturation and/or decline (χ 2 = 31.95,df = 1, P < 0.001) (Table 1) (Fig. 2A).Taken together, these results suggest complex consequences of interactions between nutrient availability and oviposition status of mosquitoes on the dynamics of sporozoite prevalence.

Sporozoite densities in the salivary glands are dependent on nutritional availability in mosquitoes that have oviposited, but not those that retain eggs
Of the 613 mosquitoes sampled in total, 200 salivary glands were assessed for sporozoite densities at 15, 18, 21, 24 and 27 dpi (Fig. 3); 142 individuals showed evidence of carrying ≥ 125 sporozoites (see "Methods").In general, the trends were similar to sporozoite prevalence, with differences in nutritional availability to oviposited and non-oviposited mosquitoes affected the dynamics of sporozoites densities in infected salivary glands (χ 2 = 5.36, df = 1, P = 0.021) (row 8, Table 1) (Fig. 3A) as well as overall sporozoite densities (χ 2 = 12.94, df = 1, P < 0.001) (row 5, Table 1) (Fig. 3B).Pairwise comparisons of the Table 1 Statistical modeling of sporozoite prevalence ('y = sporozoite prevalence') (Fig. 2A and B) and sporozoite densities ('y = sporozoite densities') (Fig. 3A and B) a Binomial and negative binomial (type 2) generalized linear models, respectively, were used to model sporozoite prevalence and densities.Estimates of EIP 10 , EIP 50 and EIP 90 in Fig. 2C were derived by interpolating from the temporal trends predicted for each of the four groups by the model for sporozoite prevalence (y = sporozoite prevalence) Oviposited or non-oviposited; b 1 or 10% dextrose; c salivary glands sampled at 3-day intervals from 9-27 dpi for sporozoite rates and 15-27 dpi for densities marginal means suggested lower sporozoite densities in oviposited groups maintained at 1% dextrose compared to the other three groups (rows 1-15, Additional file 6: Table S3).In general, while sporozoite densities increased more rapidly in oviposited mosquitoes maintained on 10% dextrose, overall densities in the salivary glands were also higher compared to the other groups (Additional file 5: Table S3).

B
Fig. 3 A Dynamics of sporozoite densities in the sporozoite-positive salivary glands ('Oviposition status*nutrient levels*dpi' , χ 2 = 5.36, df = 1, P = 0.021, row 8, Table 1) and B mean densities over all dpi ('Oviposition status*nutrient levels' , χ 2 = 12.94, df = 1, P < 0.001) were clearly higher in the oviposited group of mosquitoes fed 10% dextrose (blue data points), which contrasts with the non-oviposited groups where sporozoite densities appeared to be higher in mosquitoes fed 1% dextrose (yellow).Each data point represents sporozoite densities from an individual mosquito (square root transformed), with box plots to display the variability in counts between individuals in each group.Dashed line indicates the detection limit of the technique used to quantify sporozoite densities (≥ 125 sporozoites) [49].Refer to Table 1 for statistical modeling of the data set and Additional file 5: Table S3 for post hoc pairwise comparisons of the estimated marginal means of the four groups at each day post-infection.ND = not done (for rationale, refer to Study design under "Methods") at 1% dextrose (rr = 1.23,SE = 0.11, P = 0.025) (Fig. 4B).However, mosquito survival rates were not affected by the combination of nutrient concentrations and oviposition status (χ 2 = 2.36, df = 1, P = 0.124) (Fig. 4C) (Additional file 6: Table S4).

Discussion
Nutrient availability and oviposition status of mosquitoes resulted in complex consequences for Plasmodium fitness traits.Differences in nutrient availability from the environment influenced the rates at which oviposited and non-oviposited became infectious (EIP), sporozoite densities in salivary glands and vector survival.In general, however, nutritional availability had less of an impact on the three traits in mosquitoes that were prevented from ovipositing.Taken together, these results suggest that costs to parasite fitness may be 'buffered' by tradeoffs experienced by mosquitoes carrying eggs, even in a resource-poor environment.However, for assays measuring transmission potential, where mosquitoes are

C
Fig. 4 A Survival rates of Plasmodium berghei-exposed Anopheles stephensi were reduced in oviposited mosquitoes (left pane, black lines) ('Oviposition status' , χ 2 = 15.37,df = 1, P < 0.001, Additional file 6: Table S4) and B to a lesser extent in mosquitoes fed 1% dextrose (right pane, yellow data points) ('Nutrient levels' , χ 2 = 4.94, df = 1, P = 0.026, Additional file 6: Table S4), while C mean survival rates over all the time points suggested survival was independent of the interaction between oviposition status and dextrose concentrations ('Oviposition status*nutrient levels' , χ 2 = 2.36, df = 1, P = 0.124, Additional file 6: Table S4) generally not allowed to oviposit, both oviposition status and nutritional availability should receive critical consideration.For instance, at the very least, depending on oviposition status and/or nutritional availability, transmission potential may be either under-or overestimated.
The positive effect of environmental nutrition on parasite fitness (i.e.probability of transmission to the next host) was clearly evident in the mosquito groups that were allowed to oviposit: while higher sugar concentrations enhanced rates of sporozoite migration (Fig. 2A, left pane) and densities (Fig. 3A, left pane) in the salivary glands, higher vector survival (Fig. 4A, left pane), especially after 15 days, indicates a larger proportion of transmission-capable mosquitoes.In general, higher nutrient levels enhance sporozoite migration and densities by increasing the rates of replication of sporozoites within the oocysts lining the basal lamina of the mosquito's midgut [54].While low glucose concentrations adversely affected the survival of adult mosquitoes in general [40,41], Annopheles gambiae infected with P. falciparum showed increased attraction to and uptake of sugars, presumably in response to the higher energetic demands [38,55,56].Despite the potential for compensatory feeding behavior, the concentrations used in the current study (1% glucose) were likely too low to prevent the apparent decline in mosquito survival in the oviposited mosquitoes after 15-18 dpi [40].Although mounting immune defenses is energetically costly in insects, it could explain reduced sporozoite rates and densities in this group, with associated pathology resulting in reduced vector survival [28].Clearly, determining whether these costs to parasite fitness are due to a general lack of resources or how the host allocates resources or energetic budgets is critical [25,32].
Nutrition did not have a strong influence on parasite fitness in the mosquito groups that were not allowed to oviposit (right panes in Figs.2A, 3A and 4A).This effect was especially apparent when comparing the nonoviposited individuals maintained on a low nutrient diet with their oviposited counterparts.Indeed, despite the low nutrition, overall rates of sporozoite migration (Fig. 2B) and densities (Fig. 3B) were comparable to the high nutrient groups, with no clear costs to vector survival (Fig. 4B).While this indicates alternative sources of nutrition were able to support sporozoite replication and vector survival, reduced immune defenses due to reproduction-immunity tradeoffs in non-oviposited mosquitoes may also result in the within-host conditions being more conducive to parasite and vector survival [28].In the absence and/or reduced availability of environmental sources of nutrition, energetic demands in adult stages of holometabolous insects can be met by the nutrients stored in the fat bodies but also by resorption of nutrients from reproductive tissues (also referred to as 'follicular resorption, ' 'oosorption' or 'follicular atresia'), albeit at the expense of reduced fecundity [32,[57][58][59][60]. Since the current study was undertaken with mosquitoes originating from a common pool (Fig. 1), the potential contribution of the fat bodies should be similar among the four treatment groups; in other words, all else being equal, the difference in rates of parasite replication and vector survival between the non-oviposited and oviposited groups maintained under low nutrient conditions should, in principle, reflect the extent of nutrient resorption from reproductive tissues.
In mosquitoes, nutrient resorption follows apoptotic death of ovarian follicles (referred to as follicular resorption herein) [58,59,61].Follicular resorption is a trait that is shared across several orders of insects and allows the organism to adjust to physiological demands [57].While the trait incurs costs in reduced fecundity for the current generation, evolutionary theory suggests these costs are traded off with enhanced life span, which in turn allows the insect more time to locate oviposition sites that can benefit the fitness of future generations [25, 27, 30-32, 34, 35, 59].In the current study, survival was indeed higher in the non-oviposited group and particularly evident when compared to the oviposited group maintained under low nutrient conditions (Fig. 4).Increased mortality in oviposited mosquitoes maintained under low nutrient treatment, especially towards the later stages of the infection, may account for the weak but nonetheless significant negative effect of reduced dextrose on survival.In general, nutrition and oviposition status showed a more pronounced influence on the parasite traits of EIP and densities than on vector survival (Figs.2A, 3A and 4A), which could be due to the reduced costs to vector survival at the lower temperature (20 °C) [62] necessitated by this parasite-vector interaction [49].Indeed, while it is possible that the effects of nutrition and oviposition status may be accentuated with increasing temperatures [53,62], it is also worth considering that laboratory colonies of mosquitoes may not face the same constraints as their free-living counterparts [50].Taken together, while the results presented here are in line with evolutionary theory for insects in general, they also indicate that the benefits of nutrient resorption to future fitness may not only outweigh the costs of reduced fecundity to the mosquitoes but also enhance their capacity to vector disease.
Follicular resorption has been suggested to result in reduced fecundity of An. stephensi following infections with another rodent malaria species, Plasmodium yoelii nigeriensis [61], and Anopheles maculipennis infected with microsporidia [63].To the best of the authors' knowledge, whether follicular resorption directly benefits parasite replication has not been assessed empirically but is nevertheless plausible, and likely warrants confirmation.For instance, while greater availability of neutral lipids during oogenesis enhanced rates of sporozoite replication and migration to the salivary glands ( [64] and reviewed by [65]), similar lipid profiles were recovered following follicular resorption in Aedes aegypti [59].Furthermore, in previous studies with the major arbovirus vector Ae.aegypti, higher rates of follicular resorption were observed in mosquitoes provided lower concentrations of sugar, likely to compensate for the reduced nutritional availability (at the cost of reduced fecundity) [58,59].In principle, higher rates of resorption in non-oviposited An. stephensi maintained in a low nutrition environment (1% dextrose) could explain the earlier appearance of P. berghei sporozoites in the salivary glands (yellow line, right pane, Fig. 2A), while continued availability of higher nutrient concentrations (10% dextrose + follicular resorption) in the other non-oviposited group could explain the higher proportion of infected mosquitoes over time as a result of sustained sporozoite replication and/or migration (blue line, right pane, Fig. 2A) (and possibly enhanced survival, right pane, Fig. 3A) [49,50].Yet, when considered in the context of their oviposited counterparts, where a linear effect of dextrose on parasite fitness was evident in oviposited mosquitoes (shortened EIP and higher rates of infection, vector survival and sporozoite densities), it is unclear why the same tenfold higher concentration of dextrose (and additional nutrients, albeit low, available from follicle resorption) in non-oviposited mosquitoes did not enhance rates of sporozoite replication and/or migration to the salivary gland or support higher sporozoite densities.One potential reason to consider is the possibility that higher concentrations of sugar may render the mosquitoes less susceptible to infection, likely due to elevated immune responses, as show recently for Ae.aegypti infected with the arboviruses of Semliki Forest virus (SFV), Zika (ZIKV) or dengue (DENV) [66,67].Taken together, these results suggest parasite fitness may be dependent on complex, dynamic interactions between physiological (nutrition) and evolutionary (egg retention) trade-offs in the mosquito vector, albeit with the tradeoffs associated with evolutionarily conserved traits such as oviposition behavior influencing the fitness of distinct vector-borne pathogens in similar ways.

Conclusions
In summary, as with other insects, preventing mosquitoes from ovipositing (and completing their gonotrophic cycle) enhanced survival, even under nutrient limiting conditions [25,32], albeit at the cost of promoting fitness of parasite species, whose analogs in humans cause malaria [43].While increasing nutrition (% dextrose) positively impacted parasite fitness in mosquitoes that had laid eggs, the results from non-oviposited mosquitoes were consistent with the rates of follicular resorption compensating for differences in nutritional availability and in the process enhancing vectorial capacity due to higher vector survival and earlier sporozoite appearance in the salivary glands (shorter EIP) as well as sporozoite densities.By obtaining three critical measures of transmission potential [45][46][47], this study argues for careful consideration of egg retention and nutritional availability on measures of vectorial capacity and vector competence.
From the parasite's perspective, the overall patterns indicate an intrinsic ability to adjust its growth rates to the nutritional status of the host.Future studies manipulating the length of the gonotrophic cycle and duration of nutritional availability should reveal the extent of parasite's adaptive plasticity.Additionally, from the vector's perspective, whether parasite infection and low nutrition reduce fecundity and/or hatchability would be valuable to determine its effect on vector abundance, although neither trait appears to be as critical as selective oviposition behavior, which prioritizes egg retention [25, 27, 30-32, 34, 35, 59].In general, however, as these results suggest, the consequences of the interactions between physiological (nutrition) and evolutionary (egg retention) trade-offs in the vector are likely to be complex for Plasmodium fitness, albeit with the evolutionary constrained trait that is oviposition behavior potentially resulting in similar fitness consequences for other vector-borne pathogens such as the arboviruses [66,67].As such, this complexity further underscores the importance of quantifying the contribution of selective oviposition behavior to disease transmission [41,68], especially since most integrated vector control strategies [1,[14][15][16][17][18][19] and human-induced climate change either directly or indirectly target this key mosquito behavior [20][21][22][23][24]. 27, squared test (type 2); df = degrees of freedom; dpi = days post-infection; P = P-value * Statistical significance set at P ≤ 0.05