Plasmodium relictum (lineage SGS1) is the most common and prevalent parasite lineage causing avian malaria in Europe and is known to induce a long latent phase marked by sudden events of recurrences (Valkiunas, 2004). The parasite strain used in the first and second experiment was isolated from two infected House sparrows (Passer domesticus) captured in January 2019 and December 2020. The fourth experiment was carried out with a P. relictum strain isolated from a House sparrow captured in August 2018.

The first three experiments were carried out with domestic canaries (Serinus canaria) while the fourth experiment was performed with juvenile house sparrows. Both bird species are known to be naturally infected with Plasmodium relictum (lineage SGS1). Indeed, House sparrows infected by this lineage have been described all over Europe, and canaries, when kept in outdoor aviaries, can commonly acquire this infection. House sparrows used in this experiment were captured with mist nets in on the campus of the University of Lausanne, Switzerland (46°31′25.607″N 6°34′40.714″E, altitude: 380 m). Prior to the experiments, a small amount (ca. 3–5 µL) of blood was collected from the medial metatarsal vein of each bird (i.e. canary and house sparrow) to ensure that they were free from any hemosporidian infection (Hellgren et al., 2004, for further details see Supplementary Material, Section 1).

When experiments required the use of mosquitoes, these were conducted with a line of Culex pipiens collected in Lausanne (Switzerland, 46°31′25.607″N 6°34′40.714″E) in August 2017, and maintained in our insectarium since. Mosquitoes were reared using standard protocols (25°C ± 1°C, 70 ± 5% RH and with 12L:12D photoperiod). We used females 7–13 days after emergence, which had not previously access to blood meals. Mosquitoes were maintained on glucose solution (10%) since their emergence and were starved (but provided with water to prevent dehydration) for 24h before the experiments.

Eighteen canaries were infected by intraperitoneal injection of 150µL of an infected blood pool (day 0). The blood pool was made of a 1:1 mixture of PBS and mixed blood sample from eight canaries that had been similarly infected two weeks before the experiment (between 200 and 300 µL of blood were taken from each of the eight canaries from their brachial veins, Pigeault et al., 2015). After experimental infection, blood from the medial metatarsal vein (2–3 µL) was sampled in all birds to monitor parasitemia during the acute (10 dpi, i.e. day post-infection) and the latent pre-experimental stage of infection (35, 47 dpi). One individual died day 25 post-infection. The remaining were assigned to two different experimental groups: “CORT” (n=9) and “Control” (n=8). To that end, we ranked the individuals by decreasing parasitemia level [measured by quantitative PCR (qPCR) during the acute stage of infection, i.e. 10 dpi]. We randomly assigned one of the two individuals with the highest parasitemia levels to each experimental group. We repeated this for the two individuals with the next highest ranks until all individuals were assigned to a treatment group.

Corticosterone is the main glucocorticoid in birds and is an essential component of their stress response (Cockrem, 2007). To study the impact of an artificial increase of this stress hormone on Plasmodium replication rate during the latent infection stage, we carried out an experiment consisting of four experimental days separated by three days interval. Due to the high number of manipulations, all experimental days were split into two blocks corresponding to two consecutive days. Eight birds were randomly selected and assigned to the first experimental block (experimental days: 52, 55, 58 and 61 dpi). The remaining nine individuals were assigned to the second experimental block (experimental days: 53, 56, 59 and 62 dpi). At 5:45 PM of each experimental day, birds ingested either 20 μL of corticosterone solution (CORT group, 20 μg of corticosterone diluted in 20 μL of dimethyl sulfoxide, DMSO, dose per body mass ≈ 1 μg/g, (Breuner et al., 1998) or 20 μL of DMSO (control group). A pipette with a 200 µL tip was used to inject the solutions directly into the birds’ beaks. The ingestion of 20 μg of corticosterone solution induced a rapid and strong increase in corticosterone concentration in birds’ blood (average increase of 46.4 ng/mL ± 16 ng/mL compared to an increase of 2.17 ng/mL ± 0.45 ng/mL in the control group, see Supplementary Material, Section 2). At 6:00 PM, birds were immobilized in PVC tubes (length 18 cm, diameter 4.5 cm) covered by a mesh at both ends and placed in individual experimental cages (L40 × W40 × H40 cm) in the dark for two hours. At 8:00 PM, blood samples from the medial metatarsal vein (ca. 2–3 µL) were taken from all individuals to quantify parasitemia by qPCR. A red lamp was used to capture the birds and collect the blood samples to avoid stressing other individuals in the room.

To explore the impact of corticosterone ingestion on Plasmodium transmission from vertebrate host to mosquitoes, on the fourth experimental day (61/62 dpi), 60 ± 5 uninfected female mosquitoes were added in each cage at 7:00 PM (i.e. one hour after corticosterone or DMSO ingestion). After the blood sample has been taken from all birds at 8:00 PM, blood-fed mosquitoes were counted and individually placed in numbered plastic tubes (30 mL) covered with a mesh and with a cotton pad soaked in a 10% glucose solution. Tubes were kept in standard insectary conditions for one week. Females were then taken out of the tubes and the amount of hematin excreted at the bottom of each tube was quantified as an estimate of the blood meal size (Vézilier et al., 2010). Females were dissected, and the number of Plasmodium oocysts in their midgut counted with the aid of a binocular microscope.

In addition to the effect associated with the artificial increase of host corticosterone concentration on Plasmodium replication rate, the first experiment also suggested an additional effect associated to the stress induced by host handling (see results section). In the second experiment we therefore explored this issue. For this purpose, eighteen canaries were infected and their parasitemia was monitored during the acute and latent stage of infection as described above. However, to minimize the hosts’ stress induced by handling before the experiment, only one blood sample was taken during the latent pre-experimental stage of infection (35 dpi). Individuals were then assigned to two different experimental groups: “handling” (n=9) and “control” (n=9) using the same procedure as described above.

While control individuals were not manipulated, birds belonging to the “handling” group were captured at days 52 or 53, 55 or 56 and 58 or 59 post-infection (experimental days were split into two blocks as described above) and handled and restrained in the same way as in Experiment 1 to generate stress. To avoid a confounding effect between handling stress and blood loss, no blood sample was taken during this period. Then, to explore the impact of handling stress on the dynamics of Plasmodium replication rate and transmission to mosquitoes, on day 61 or 62 dpi, all birds were captured, immobilized in PVC tubes and placed, at 6:00 PM, in the dark in individual experimental cages. One hour later, all individuals were exposed for one hour to 60 ± 5 uninfected female mosquitoes. At the end of the exposure session, a blood sample (ca. 2–3 µL) was taken from the medial metatarsal vein of all individuals to quantify parasitemia by qPCR. Blood-fed female mosquitoes were counted, isolated, and dissected as described above.

Finally, for both corticosterone and handling experiments, blood samples were taken on all birds at day 65 and 75 post-infection to measure their parasitemia by qPCR. The latent infection period was therefore divided into three different stages: the latent pre-experimental stage (35–47 dpi), the latent experimental stage (52–62 dpi) and the latent post-experimental stage (65–75 dpi).

Eighteen uninfected canaries were used for this experiment, from which nine were exposed to uninfected mosquito bites (“exposed”) and the remaining nine were used as “control”. To obtain baseline or near-baseline corticosterone levels prior to mosquito exposure, 90–110 µL of blood was taken from the brachial vein of all birds within three minutes of capture (Wingfield and Romero, 2011). Immediately after the blood sampling, birds were immobilized in PVC tube covered by a mesh at both ends and placed in individual experimental cages in the dark at 6:00 PM. After one hour of acclimatization, 60 ± 5 mosquitoes were released in each cage containing an “exposed” bird. One hour later (i.e. 8:00 PM), a new blood sample was taken from the brachial vein of all birds to quantify corticosterone concentrations. Total corticosterone from plasma samples was quantified using enzyme immunoassay kit from Enzo Life Sciences (Corticosterone ELISA Kit, cat. No. ADI-900-097, Lausen, Switzerland) following the manufacturer’s instructions. Plasma samples were obtained by centrifugation of bird blood, which was kept on ice. After centrifugation, plasma was collected and stored at −20 °C. Samples were run in duplicate and randomized across plates. Concentration values were extrapolated from standard curves and all samples felt within the standard curves. Inter-plate variation was 1.3%.

On October 11, 2018, seven uninfected juvenile House sparrows were infected by intraperitoneal injection of an infected blood pool (day 0) and placed together in an outdoor aviary (4.6 m × 4.6 m, 2.7 m high (complete material and methods are provided in the Supplementary Material, Section 3). Blood samples were taken regularly – i.e. between monthly and weekly, depending on the stage of infection, see Figure S3 – during one year from the medial metatarsal vein (ca. 2–3 µL) of all individuals to quantify parasitemia by qPCR. One bird died for unknown reason day 104 post-infection.

Day 295 post-infection birds were captured and housed in an individual cage (100 cm × 30 cm × 40 cm high) that was placed in the outdoor aviary. They were isolated to facilitate their capture for the rest of the experiment and thus minimize their stress. Thirteen days later (i.e. 308 dpi) each bird was captured and transported in holding bags to the laboratory, 250 m apart. A t 6:00 PM, individuals were randomly assigned to the “exposed” (n=3) or to the “control” group (n=3), immobilized in PVC tubes covered by a mesh at both extremities and placed in individual experimental cages for two hours. After one hour, 60 ± 5 mosquitoes were released in each cage containing an “exposed” bird. One hour later (i.e. 8:00 PM), a blood sample was taken from the medial metatarsal vein of all individuals to quantify parasitemia by qPCR and birds were release outside in their individual cage. Parasitemia was then monitored every 2–3 days during two weeks and birds were then released in the large aviary. Thereafter, birds’ parasitemia was monitored regularly until the end of the experiment (i.e. 365 dpi). At the end of the experiment all House sparrows were treated to clear infection and kept in the outdoor aviaries.

The quantification of blood parasitemia was carried out using qPCR with a protocol adapted from Christe et al., 2011. Briefly, DNA was extracted from the blood using a standard protocol (Qiagen DNeasy 96 blood and tissue kit). For each individual, we conducted a multiplex qPCR. We targeted the nuclear cyt b gene of Plasmodium (Primers: L4050Plasmo: 5′-GCTTTATGTATTGTATTTATAC-3′, H4121Plasmo: 5′-GACTTAAAAGATTTGGATAG-3′, TaqMan probe: CY3-CYTb-BHQ2: 5′-CCTTTAGGGTATGATACAGC-3′) and the 18s rRNA gene of the bird (Primers: Plasmo18S-f: 5′-GGCAGCTTTGGTGACTCTAGA-3′, Plasmo18S-r: 5′-AGTTGATAGGGCAGACATTCG-3′, TaqMan probe: FAM-18S-BHQ1: 5′-AACCTCGAGCCGATCGCACG-3′). All samples were run in triplicate (Bio-Rad CFX96TM Real-Time System). Parasite number was calculated with relative quantification values (RQ). RQ can be interpreted as the fold‐amount of target gene (Plasmodium 18s rDNA) with respect to the amount of the reference gene (Bird18s rDNA) and are calculated as 2^{−(Ct18s Plasmodium – Ct18s Bird)}. For convenience, RQ values were standardized by ×10⁴ factor and log-transformed (log(x, base = exp(1))).

Analyses were carried out using the R statistical package (v. 4.0.2). Data were analyzed separately for each experiment. The different statistical models built to analyse the data are described in the Supplementary Material ( Table S1 ). Briefly, analyses where a same individual bird was sampled repeatedly, such as corticosterone quantification or fluctuation of blood parasitemia throughout time, were analysed fitting bird as a random factor into the models (to account for the temporal pseudoreplication), using a mixed model procedure (lmer, package: lme4, Bates et al., 2015). When experimental days were split into two blocks, bird random factor was nested in experimental blocks. Bird treatment and day post-infection (experiment 1, 2 & 4) or the time of blood sampling (experiment 3) were used as fixed factors. When only one parasitemia measurement was taken per bird and experimental days were not split into two blocks, the analysis was performed using Mann-Whitney-Wilcoxon Test (i.e. comparison of bird latent pre-experimental parasitemia according to bird treatment in the handling experiment). Mosquito-centred traits, such as blood meal size (hematin quantity, ng), infection prevalence (proportion of mosquitoes containing at least one oocyst) and oocyst burden (taking account of mosquitoes with one or more oocysts in the midgut), which may depend on which bird mosquitoes fed on, were analysed fitting bird as a random factor into the models, using lmer, glmer.nb or glmer (package: lme4) according to whether the errors were normally (hematin quantity), binomially (infection prevalence) or negative binomially distributed (oocyst burden). Bird treatment and, when necessary, blood meal size (haematin) were used as fixed factors.

Maximal models, including all higher-order interactions, were simplified by sequentially eliminating non-significant terms and interactions to establish a minimal model (Crawley, 2012). The significance of the explanatory variables was established using a likelihood ratio test (which is approximately distributed as a Chi-square distribution or an F test (Bolker, 2008). The significant Chi-square or F values given in the text are for the minimal model, whereas non-significant values correspond to those obtained before the deletion of the variable from the model. A posteriori contrasts were carried out by aggregating factor levels together and by testing the fit of the simplified model using a likelihood ratio test.

Bird parasitemia decreased drastically between the acute infection stage (10 dpi) and the first blood sample of the latent infection stage (35 dpi, model 1: χ² = 61.09 p < 0.0001). Then, a significant increase in host parasitemia over time was observed during the latent infection period (from 35 to 75 dpi, model 2: χ² = 42.13 p < 0.0001, Figure 1A ).

We further divided the latent stage of infection into three different stages: the latent pre-experimental stage (35–47 dpi), the latent experimental stage (52–62 dpi) and the latent post-experimental stage (65–75 dpi). During the latent pre-experimental stage, parasitemia did not vary between the different blood sampling days (model 3, 35 and 47 dpi, χ² = 1.29, p = 0.255, Figure 1A ) and was not different between control individuals and birds that would later ingest corticosterone (model 3, χ² = 0.65, p = 0.422). During the latent experimental stage, an interaction between dpi and bird treatment was observed (model 4, χ² = 3.95, p = 0.046, Figure 1A ). While the parasitemia of the control individuals did not vary significantly over time (model 5, χ² = 1.98, p = 0.158), the parasite burden of CORT birds increased (model 6, χ² = 16.40, p < 0.0001). Then, during the post-experimental stage, parasitemia did not vary between the different blood sampling days (65 and 75 dpi, model 7, χ² = 0.96, p = 0. 326, Figure 1A ) and was not different between control and CORT birds (model 7, χ² = 1.70, p = 0.191).

On the last day of the latent experimental stage (61/62 dpi), we evaluated the impact of the oral hormone treatment on Plasmodium transmission rate to mosquito vector. No effect of bird treatment was observed on mosquito blood meal size (mean ± s.e., CORT: 23.15 ng ± 8.26, control: 23.94 ng ± 8.07, model 8, χ²₁ = 0.029 p = 0.865), mosquito infection prevalence (CORT: 0.66, 95% CI: 0.06, control: 0.53, 95% CI: 0.06, model 9, χ²₁ = 1.140 p = 0.286) and oocyst burden (geometric mean ± s.d., CORT: 2.74 ± 2.40, control: 4.07 ± 3.11, model 10, χ²₁ = 0.033 p = 0.857).

Bird parasitemia decreased drastically between the acute infection stage (10 dpi) and the first blood sample of the latent infection stage (35 dpi, model 11: χ² = 54.39 p < 0.0001). Then, a significant interaction between bird treatment and dpi was observed during the latent infection period (model 12: χ² = 4.92 p = 0.026, Figure 1B ). While the parasitemia of control birds did not vary, the parasite burden of handling individuals increased over time ( Figure 1B ).

During the latent pre-experimental stage, parasitemia was not different between control individuals and birds in the “handling” group (model 13, W = 40, p = 0.99, Figure 1B ). After the four days of bird manipulation (i.e. latent experimental stage), the parasitemia of handled individuals, measured day 61 or 62 post-infection, was significantly higher than the parasitemia of control birds (mean RQ10 ± s.e., control = 12.65 ± 4.62, handling = 39.54 ± 14.95, model 14, χ² = 5.32 p = 0.021, Figure 1B ). This significant difference was maintained throughout the latent post-experiment stage (control = 8.4 ± 2.55, handling = 32.32 ± 9.82, model 15, χ² = 4.91 p = 0.027, Figure 1B ), but a slight decrease of host parasitemia was observed according to blood sampling days (65 versus 75 dpi, model 15, χ² = 6.206, p = 0.013, Figure 1B ).

As in the first experiment, on the last day of the latent experimental stage (61/62 dpi), no effect of bird treatment was observed on mosquito blood meal size (mean ± s.e., Control: 31.83 ng ± 9.04, Handling: 28.86 ng ± 9.98, model 16, χ²₁ = 1.97 p = 0.160), infection prevalence (Control: 0.53, CI: 0.06, Handling: 0.55, CI: 0.07, model 17, χ²₁ = 0.08 p = 0.772) and oocyst burden (geometric mean ± s.d., control: 3.87 ± 2.70, handling: 5.38 ± 3.24, model 18, χ²₁ = 0.69 p = 0.407).

The mean baseline corticosterone of birds was 2.50 ± 0.66 ng.mL⁻¹ (mean ± s.e.). Corticosterone level increased significantly after the handling and isolation period in both “control” and “exposed” groups (model 19, time of blood sampling: χ² = 19.21, p < 0.0001, contrast analysis: unexposed χ² = 4.61, p = 0.032, exposed χ² = 21.41, p < 0.0001, Figure 2 ). The increase was however significantly higher in birds exposed to mosquito bites (significant interaction between bird treatment and time of blood sampling, model 19, χ² = 4.56, p = 0.032). On average, the corticosterone level increased by more than 7.5 times in the “exposed” group (first blood sample: 1.79 ng.mL⁻¹ ± 0.31, second blood sample: 13.57 ng.mL⁻¹ ± 2.57) and by 2.5 times in the “control” group (first blood sample: 3.30 ng.mL⁻¹ ± 1.35, second blood sample: 8.28 ng.mL⁻¹ ± 1.90).

Three hundred and eight days after their infections (i.e., dpi), we exposed latently infected birds to a stressful event. Handling stress alone and handling stress following by exposure to mosquito bites influenced Plasmodium replication rate in house sparrows. We recorded a significant variation in parasitemia overtime after bird exposure to the stressful events (model 20, χ² = 4.29, p = 0.038, Figure 3 ). Peak parasitemia was reached approximately one week after bird manipulation and then decreased thereafter. Parasitemia in birds exposed to mosquito bites appeared to increase more strongly over time than parasitemia of birds exposed to handling stress only. However, we did not observe a significant interaction between bird treatment and dpi (model 20, χ² = 0.01, p = 0.927, Figure 3 ). The description and analysis of the overall infection dynamics is presented in Supplementary Material (Section 3 & Figure S2 ).