Seasonal dynamics in mosquito abundance and temperature do not influence avian malaria prevalence in the Himalayan foothills

Abstract We examined seasonal prevalence in avian haemosporidians (Plasmodium and Haemoproteus) in migrant and resident birds in western Himalaya, India. We investigated how infection with haemosporidians in avian hosts is associated with temporal changes in temperature and mosquito abundance along with host abundance and life‐history traits (body mass). Using molecular methods for parasite detection and sequencing partial cytochrome b gene, 12 Plasmodium and 27 Haemoproteus lineages were isolated. Our 1‐year study from December 2008 to December 2009 in tropical Himalayan foothills revealed a lack of seasonal variation in Plasmodium spp. prevalence in birds despite a strong correlation between mosquito abundance and temperature. The probability of infection with Plasmodium decreased with increase in temperature. Total parasite prevalence and specifically Plasmodium prevalence showed an increase with average avian body mass. In addition, total prevalence exhibited a U‐shaped relationship with avian host abundance. There was no difference in prevalence of Plasmodium spp. or Haemoproteus spp. across altitudes; parasite prevalence in high‐altitude locations was mainly driven by the seasonal migrants. One Haemoproteus lineage showed cross‐species infections between migrant and resident birds. This is the first molecular study in the tropical Himalayan bird community that emphasizes the importance of studying seasonal variation in parasite prevalence. Our study provides a basis for further evolutionary study on the epidemiology of avian malaria and spread of disease across Himalayan bird communities, which may not have been exposed to vectors and parasites throughout the year, with consequential implications to the risk of infection to naïve resident birds in high altitude.


Avian haemosporidians in the genera Plasmodium and
Haemoproteus (Phylum Apicomplexa, Order Haemosporida) are globally distributed vector-mediated parasites found in a broad range of birds (Valkinũas, 2005). Plasmodium and Haemoproteus parasites reproduce sexually in dipteran vectors-culicid mosquitoes and ceratopogonid midges, respectively (Valkinũas, 2005)-and use birds as intermediate hosts in which they undergo asexual reproduction.
The incubation period of malaria parasites within mosquitoes is exquisitely temperature sensitive (Paaijmans, Read, & Thomas, 2009); therefore, temperature is a major determinant of malaria transmission and persistence in host populations (LaPointe et al. 2005). Low ambient temperatures prevent sporogonic development of malaria parasites within vectors, influencing their altitudinal distributions (LaPointe, Goff, & Atkinson, 2010). Similarly, change in habitat composition from wet to dry due to climatic shift may alter both the geographic range and local abundance of malaria pathogens because their vectors (mosquitoes, Ceratopogonid midges, and Simuliid flies) require wet habitat to complete their life cycle (Valkinũas, 2005). Furthermore, owing to this vector dependency, both human (Hay et al., 2002) and avian (Cosgrove, Wood, & Sheldon, 2008) Plasmodium show a marked seasonality in transmission which appear to result in skewed prevalence estimates in winters. However, such stark variation in seasonal patterns ought to be more prominent in temperate regions where a peak in malaria prevalence occurs in late summer and autumn when the proportion of juveniles coincides with a rise in vector populations (see Beaudoin, Applegate, David, & McLean, 1971;Cranston, Ramsdale, Snow, & White, 1987;Marshall, 1938). In tropical and subtropical climates, malaria parasites transmission can occur yearround (Valkinũas, 2005). In Hawaii, for example, malaria transmission occurs after the breeding season as chronically infected native birds serve as year-round reservoir of disease (LaPointe, Atkinson, & Samuel, 2012). However, within tropical and subtropical areas, transmission will not occur at very high altitudes or during the cold season.
There is a lack of quantitative studies which limits our capacity to understand and predict such changes in other threatened ecosystems and has largely remained unexplored in the tropics and in particular in the Indian subcontinent, despite it being the major staging and wintering ground of the Central Asian Flyway populations.
The Himalayan mountain range is one of the most species-rich areas in the world, harboring about 8% of the world's bird species (Price et al. 2003). The high species diversity is due to species turnover associated with altitudinal variation in habitat, as well as variation in species composition along the range (Martens & Eck 1995;Price et al. 2003) and the possibility that such regions are buffered from climatic extremes, for example, because habitats and species can shift altitudinally in response to climate change (Fjeldså 1995;Fjeldså & Rahbek 2006). The Indian western Himalaya is species rich but remains a relatively understudied biogeographic region with only a handful of studies on birds (e.g., Price 1991), plants (Oommen and Shankar 2005), and mosquitoes (Devi and Jauhari 2004), with no previous studies on the dynamics of avian diseases and their vectors. Given the effect of climate and host demography which play a crucial role in determining parasite prevalence, as well as the dynamics of parasite transmission and host migration patterns both of which can increase the risk of infection. Many migrants move between altitudes or to the plains, and thereby encounter diverse parasite and vector faunas compared with the resident counterparts which remain at high altitudes throughout.
Given that suitable vectors are present to transmit and maintain the infection, migrants can form an effective bridge for parasites between wintering and breeding grounds, hence increasing the risk of infection to naïve resident birds at the higher altitudes.
In this first molecular study on Plasmodium and Haemoproteus prevalence across resident and migrant birds, we aimed to explore the seasonal dynamics in parasite prevalence and its association with seasonal environmental drivers and host life-history traits. Specifically, we examine (1) how mosquito abundance and temperature relate with Plasmodium spp. prevalence in foothills birds; (2) the effect of altitude on haemosporidian prevalence and diversity; and (3) the extent of cross-species infections between resident and migrant birds. At each location, we set up 10-12 mist nets in high bird activity locations, often along forest edges, footpaths, or off-road nature trails.

| Study site and bird blood sampling
Mist nets were 38-mm gauge, 2.6 m tall, and 6, 9, or 12 m long. Birds were sampled between 0540 and 1230, and nets were checked every 5-10 min. Birds were identified to species according to Rasmussen and Anderton (2005) and ringed. Wing length was measured (maximum wing-cord) with a steel rule (±1 mm) and body mass recorded with an electronic balance (±0.1 g). All birds were measured by F.I. Captured individuals were released at the site immediately after processing. We sampled 20-40 μl of bird blood from the sub-brachial wing vein (never exceeding 1% of the individual's body weight). All samples were stored in SET Buffer (20-40 μl in 500 μl buffer 0.15 mol/L NaCl, 0.05 mol/L Tris, 0.001 mol/L EDTA, pH 8.0) at room temperature and subsequently transferred to −20°C.

| Mosquito sampling and temperature data across seasons in foothills
Mosquito sampling was conducted during January-December 2009 four times in each month using standard Centers for Disease Control and Prevention (CDC) miniature black light (UV) traps without dry ice baits. For all sampling, the traps were suspended within the forest canopy, no more than 2 m from the ground on trees (DiMenna et al., 2006;Okanga et al., 2013) around avian blood sampling (mist-netting) sites. Traps were operated overnight starting at 1,800 hr and picked up next morning at 0700 hr. Mosquitoes were collected from traps in the morning and stored at -20°C before identification. The species and sex of all mosquito samples were determined using a morphological mosquito-identification key (Barraud, 1934;Christophers, 1933), and sorted by species and date followed by DNA barcoding technique (Kumar, Rajavel, Natarajan, & Jambulingam, 2007). As only female mosquitoes are responsible for parasite transmission, we provide estimates of abundance for female mosquitoes. Daily temperature records were obtained from the on-site WII weather station in Dehradun.

| Molecular methods
DNA extractions were performed using phenol chloroform extraction method (Sambrook, Fritsch, & Maniatis, 1987) or ammonium acetate protocol (Nicholls, Double, Rowell, & Magrath, 2000). We screened all bird samples for the combined presence of parasites of the genera Plasmodium and Haemoproteus using a polymerase chain reaction (PCR) protocol designed to amplify a 160-bp fragment of mitochondrial ribosomal RNA gene (rRNA) of avian haemosporidians (213F/372R; Beadell & Fleischer, 2005) followed by a restriction enzyme-based assay.
For samples that screened positive for the 16S rRNA gene fragment of the parasites, we amplified the cytochrome b (cyt-b) gene fragments ranging from 533, 477, or 351 bp following Beadell et al. (2004), Hellgren, Waldenström, and Bensch (2004), and Ishtiaq et al. (2006), respectively. Each plate accompanied a parasite positive control and also a negative control to examine for any potential contamination. We screened all parasite-negative samples for bird DNA (for cyt-b gene) following Dumbacher, Pratt, and Fleischer (2003).
The resulting PCR products were then sequenced in both directions.
Sequences were assembled, aligned, and edited using SEQUENCHER version 5.2. We then identified sequences to genus using their closest sequence matches in GenBank or MalAvi database (Bensch, Hellgren, & Pérez-Tris, 2009). Novel lineages were defined as lineages that F I G U R E 1 Bird sampling sites in Uttarakhand state, western Himalaya, India differed by one or more nucleotides from any lineage deposited in GenBank prior to this study. All sequences are deposited in GenBank accession numbers: MF565807-MF565836 (Table S1).

| Phylogenetic analysis
To explore ecological and evolutionary relationships, a model-based approach was used following phylogenetic reconstruction using the maximum-likelihood analysis on 27 Haemoproteus sequences isolated from western Himalayan birds as well as 46 reference sequences downloaded from MalAvi database (Bensch et al., 2009).
The Plasmodium phylogeny included 12 sequences isolated from western Himalayan birds and 27 reference sequences downloaded from MalAvi database (Bensch et al., 2009). Leucocytozoon majoris (GenBank accession number: AY393804) served as an outgroup in both phylogenies. The maximum-likelihood tree was constructed using Bayesian phylogenetics as implemented in BEAST version 1.4.3 (Drummond & Rambaut, 2007) using the most appropriate substitution model (GTR + G) according to the Akaike Information Criterion implemented MEGA version 5.2 (Tamura et al., 2011). We present a maximum clade credibility tree using a relaxed molecular clock approach (Drummond & Rambaut, 2007). Rates of substitution were drawn from a lognormal distribution, and Yule prior was used for branching rates. We conducted two runs of 20 million generations, each with sampling conducted every 1,000 generations. Tracer (Rambaut & Drummond, 2003) was used to assess convergence, and whether two chains were mixing and whether the estimated sample size (ESS) for each parameter was sufficient (ESS > 200) to obtain robust parameter estimates. Four million generations were discarded as burn-in from each run, leaving a posterior distribution of 32,000 trees.

| Statistical analysis
We used contingency table analyses using G tests for the heterogeneity in parasite prevalence across host species and families followed by partitioned analyses (Sokal & Rohlf, 1995).

| Individual-level traits
We used generalized linear mixed models (GLMMs, function glmer in lme4; Bates, Maechler, Bolker, & Walker, 2015) to assess whether individual infection was influenced by individual-level trait such as body condition, as fixed effects and bird taxonomy (species nested in genus and genus nested in family) as a random effect. Body condition (or size-corrected body mass) is measured as the residuals of a speciesspecific log (mass) by log (wing length) regression, which is often used as a proxy for overall condition (Schulte-Hostedde, Zinner, Millar, & Hickling, 2005).

| Species-level traits
We also used GLMMs to assess whether infection prevalence in each host species was influenced by species-specific traits, including avian abundance, mean species body mass (g), and migratory status as fixed effect and avian families with ≥5 individuals with bird taxonomy (species nested in genus and genus nested in family). We derived abundance estimates for each species by a single experienced observer (CB) recording the maximum number of birds seen on a given fixed circuit on dates within each week. We ensured that for any given week, at least one circuit of the nature trail within the WII campus was completed, but in some weeks, it may have been as many as 3-4 times. For any species known to be migratory or simply scarce, the exact date was noted, but for the remainder, simply the maximum figure seen during the week was noted. This may sound obvious, but by doing this, on a weekly (or even monthly) basis, it not only flags up when migrants are arriving, but in some cases, it identifies species that are not generally known to be migrants, but clearly do undergo at least localized movements. This presence/absence information can be a first step to detecting such localized seasonal movements. The averaged weekly maximum count for two visits per month was used as an index of abundance of each species. Abundance estimates (which are effectively an index for many of the more cryptic species) represent the predicted number of individuals of a given species detected in ~2.5 hr of surveying.

| Seasonal effects on parasite prevalence, mosquito abundance, and temperature
Samples were ordered by the season collected (spring, summer, monsoon, autumn, and winter). We used GLMMs to assess whether individual infection was influenced by season, mosquito abundance, or average monthly temperature as fixed effects, and bird taxonomy (species nested in genus and genus nested in family) as a random effect. Mosquito abundance was the mean number of mosquitoes caught, and this was calculated to provide comparability with months where mosquitoes were trapped less than four nights in a month. Pearson's product-moment correlations (r) were run between mosquito abundance and temperature. To explore variation in parasite prevalence, we plotted correlation between prevalence in winter against prevalence in spring-summer, including the bird species for which at least four individuals were sampled in both seasons.

| Altitudinal effects on parasite prevalence and diversity
We used GLMMs to assess whether individual infection was influenced by altitude and migratory status (resident, seasonal migrant, and long-distance migrant) as fixed effects, and bird taxonomy (species nested in genus and genus nested in family) as a random effect.
All models were specified with a binomial error distribution and logit link function. The significance of fixed effects was evaluated with Wald's chi-square tests (Bolker et al., 2009). Analyses were conducted in R v. 3.0.1 (R Development Core Team, 2010; R: A language and environment for statistical computing).
However, resident birds were marginally less infected (Wald's χ² = 6.01, df = 2, p < .49; Table 2) than seasonal migrants. There was no relationship detected between prevalence of Plasmodium spp. and Haemoproteus spp. when accounting for host taxonomy.

| Cross-species infections between migrant and resident birds
Using PCR-based detection methods, 37 cyt-b sequences of parasite lineages were isolated, with a high proportion of infections of Haemoproteus spp., representing 27 lineages whereas Plasmodium spp.
had just 10 lineages (GenBank accession numbers are listed in Table   S1). Plasmodium and Haemoproteus mitochondrial lineage relationships are presented independently in Figures 7 and 8

| DISCUSSION
Our study is the first seasonal study in the tropics which captures the absence of seasonal pattern in prevalence of avian Plasmodium spp. year molecular study in tropical Himalayan foothills revealed a strong correlation between mosquito abundance and temperature; however, there was a lack of seasonal variation in Plasmodium spp. prevalence.
There was no effect of body condition on parasite prevalence; however, log avian abundance contributed to variation in total haemosporidian prevalence. Haemoproteus spp. prevalence showed no variation across seasons and life-history traits. Migrant birds were slightly more infected than resident species across altitudes.

| Seasonal variation in parasite prevalence
Our year-round study shows the lack of seasonal variation in parasite prevalence and its association with tropical climate which allows for the presence of mosquitoes and other hematophagous arthropod vectors and therefore facilitates avian malaria transmission throughout the year. Human malaria, however, shows a seasonal pattern in the Himalayan foothills with peak transmission from July-October (Dhiman, Chavan, Pant, & Pahwa, 2011). We did not find any peak in prevalence in spring due to a relapse of the previous year's infections (Beaudoin et al., 1971), which coincides with a peak in vector abundance (Atkinson & van Riper, 1991). In temperate regions, most Haemoproteus spp. show peak prevalence in the spring months when vectors (e.g., Culicoides spp.) become abundant or in the summer with acute infections in newly recruited immunologically naive juveniles (Cranston et al., 1987;Klei & DeGiusti, 1975;Marshall, 1938). It is very difficult to differentiate recrudescence and relapse in parasitemia in the tropics as the optimal conditions without any thermal constraints facilitate parasite transmission throughout the year.
Furthermore, our use of sensitive PCR-based screening methods provides a better estimate of parasite prevalence (Ishtiaq et al., 2017) and suggests that Plasmodium spp. and Haemoproteus spp. were indeed present in the blood throughout the year without undergoing a latent (dormant) phase where parasitemia survive in the tissues of avian hosts (Atkinson & van Riper, 1991). In temperate regions, dormancy is thought to minimize clearance from the host and to seasonally reduce virulence, thereby enhancing overwinter survival of the parasite (Worms, 1972 T A B L E 2 (Continued)

| Effect of host life-history traits and environmental drivers on parasite prevalence
Prevalence of parasites varied significantly across host species and families. Of the three species-level traits, we found no significant relationship between host body condition and infection status.
However, both host abundance and mean body mass influenced the parasite prevalence in the foothills Himalayan bird community.
These findings are similar to patterns in prevalence observed in temperate regions where species with the large body mass and at greater abundance are known to facilitate transmission and increase parasite prevalence (Matthews, Ellis, Roberts, Ricklefs, & Collins, 2015;Scheuerlein & Ricklefs, 2004). Large birds' surface area provides more biting opportunities (Atkinson & van Riper, 1991) and emits more carbon dioxide which attracts host-seeking vectors (Scheuerlein & Ricklefs, 2004). Epidemiological theory predicts that host density plays a central role in transmission of directly transmitted parasites (Anderson and May 1978). The positive relationship between host abundance and parasite prevalence is consistent with results from Dobson (2004) and Brown et al. (2001). In particular, the U-shaped relationship between host abundance and parasite prevalence was similar to the temperate study by Ricklefs et al. (2005) F I G U R E 3 Relationship between total haemosporidian prevalence and avian abundance (log) in the foothills of western Himalaya. The regression includes species with sample sizes of more than four individuals phylogenetically closely related species (see Ishtiaq, 2017). In contrast, Medeiros et al. (2015) found abundant hosts were more often bitten by mosquito vectors and were also more likely to be infected by Plasmodium parasites compared with less abundant hosts. In the Himalayan bird community, the vector-host encounter rates contribute to the positive relationships between host abundance and parasite prevalence; however, the temperature and mosquito abundance showed a negative effect on the Plasmodium spp. prevalence.
The seasonal dynamics in parasite prevalence are a function of vec-

| Effect of altitude and seasonal migration on parasite prevalence
In contrast to studies in wet tropics and temperate regions (

| Cross-species infections
Many lineages observed in the Haemoproteus tree were host specific at a family level, which appears to be a trait in Haemoproteus species where the host ranges are often restricted to a limited number of closely related host species (Atkinson & van Riper, 1991;Bennett & Peirce, 1988;Perez-Tris et al., 2007;Savage & Greiner, 2004). Among resident species sampled, Zosterops palpebrosus (Oriental white-eye) was the most infected species with an assemblage of Haemoproteus lineages which appears to have diversified within the Zosteropidae family. Among migrants, Acrocephalus dumetorum was the most infected species with a diverse assemblage of Haemoproteus belopolskyi, which have a relatively low genetic distance from prevalent and widespread parasites of migrant Acrocephalids and Sylvids in Europe and African (Ishtiaq, 2017). This appears to be polymorphism within H. belopolyskyi clade. The Haemoproteus lineage ACDUM1 was shared with a pink-browed rosefinch Carpodacus rodochroa, a high-altitude species which points toward potential cross-species transmission; however, as our study utilized only PCR, and not microscopy technique, there are several caveats associated with this methodology that could bias our lineage sharing results. First, the development of haemosporidians can be abortive in resistant or partly resistant hosts, resulting in no development of gametocytes (Cannell et al., 2013;Valkinũas, Ashford, Bensch, Killick-Kendrick, & Perkins, 2011). Second, the low-frequency infection could be spillover infections in resistant hosts leading to a dead end for parasite transmission, and indeed, this was confirmed with microscopy by the absence of gametocytes. Nonetheless, there was no sharing of parasite lineages between foothills and high-altitude sites, which suggests demarcation of parasite transmission zones with a varying vector and thermal gradient. In Plasmodium tree, many lineages were generalists that infected multiple host species across multiple families.
Overall, this is the first molecular study of a tropical Himalayan bird community that reveals a high diversity of Haemoproteus lineages.
While there was no seasonal variation in the prevalence of either Haemoproteus or Plasmodium, thus the lack of thermal constraints appears to facilitate year-round transmission of these parasites. Many Haemoproteus lineages were host specific and have not been reported previously. These new data contribute to our understanding of seasonal dynamics of temperature, mosquito abundance, and thermal constraints in parasite transmission in the tropics. They also provide a basis for furthering our understanding on the ecology and epidemiology of avian malaria and the spread of disease across Himalayan bird communities, which may not generally be exposed to vector and F I G U R E 8 Maximum clade credibility tree of Haemoproteus cyt-b (477 bp) lineages recovered from western Himalayan birds in India. Posterior clade probability support values above 0.5 are shown. 27 Haemoproteus lineages found in western Himalayan birds are in red text, and 46 reference Haemoproteus lineages from MalAvi database (Bensch et al., 2009) are in black text parasites throughout the year. This can have important potential implications for the risk and susceptibility to infection for naïve resident birds at higher altitudes.

ACKNOWLEDGMENTS
This study was partially funded by the Marie Curie Fellowship, European Commission (return phase: MIF1-CT-2006-040845), and Wellcome Trust/DBT India Alliance Fellowship (IA/I(S)/12/2/500629) to FI. We would like to thank Uttarakhand Forest Department for ethical approval and permission for collection of avian blood samples. We thank Dr. A. Rahmani of the Bombay Natural History Society for supplying bird rings and N. Rawat for valuable assistance in the fieldwork.

ETHICAL APPROVAL
The field experiments comply with the current laws of the India where study was performed.
We thank Uttarakhand Forest Department for ethical approval and permission for collection of avian blood samples.

CONFLICT OF INTEREST
The authors declare that they have no competing interests.

AUTHOR CONTRIBUTION
FI conceived the idea. FI, CB, and YVJ designed the experiment; FI conducted the field and laboratory experiments; CB helped with fieldwork; FI wrote the manuscript, and FI, CB, and YVJ reviewed the manuscript. All authors approved the final version of the manuscript.