First identification and seasonal dynamics of hepatozoon and microfilariae in Diploderma swinhonis (squamata: agamidae) from Taiwan

Abstract

Hemoparasites in reptiles remain underexplored in East Asia, particularly among endemic lizards. We present the first molecular and morphological study of blood parasites in Diploderma swinhonis, a widespread tree lizard endemic to Taiwan. We examined six populations from three locations using an integrated approach combining microscopy (morphology, morphometry) and molecular approaches (Sanger sequencing, phylogenetic analysis). Phylogenetic analyses revealed that Hepatozoon sequences from this host are genetically identical to H. ophisauri, suggesting a broader host and geographic range for this species. In contrast, microfilariae formed a distinct lineage within Oswaldofilariinae, with over 10% divergence from known taxa, indicating a likely undescribed species. Despite high lizard densities in some locations, overall infection prevalence was low (5.3%; n = 1166 ). Relative risk analysis showed significantly higher infection rates in adults than in juveniles, supporting an age-related effect. Seasonal peaks in infection varied between parasite taxa and locations, reflecting differences in transmission routes and life histories. Newly designed primers targeting cytochrome b (for Hepatozoon) and COI (for microfilariae) demonstrated higher detection sensitivity than traditional markers, especially in low-parasitemia individuals. These results highlight the importance of appropriate molecular tools for detecting blood parasites in under-studied reptiles and contribute new insights into the ecological and taxonomic diversity of hemoparasites on East Asian Islands.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-19878-8.

Introduction

Reptilian hemoparasites comprise a diverse array of blood-associated parasitic agents, including protozoans of the phylum Apicomplexa¹, kinetoplastids of Euglenozoa², juvenile stages of nematodes³, and certain bacteria and viruses⁴. These parasites are prevalent among reptiles and exhibit greater diversity than those infecting mammals and birds⁴. Such diversity may be attributed to reptiles’ relatively fixed habitat types, restricted home ranges and lower mobility⁴, and their long evolutionary history⁵. For instance, the well-studied genus Plasmodium, the causative agent of malaria, includes over 250 described species, the majority of which infect reptiles⁶.

Hemoparasitic infections in lizards can lead to notable physiological and ecological consequences, including anemia^7–9, damage to internal organs¹⁰, and even mortality⁶. Beyond their direct physiological impacts, these parasites can modify predator–prey relationships and influence both intra- and interspecific competition^11,12. The dynamics of host-parasite interactions are shaped by processes occurring across spatial, temporal, and ecological scales from individual hosts to entire communities^13,14. Investigating parasitism patterns in wild populations enhances our understanding of the ecological and evolutionary roles of hemoparasites, while also contributing valuable insights to epidemiological and conservation studies¹⁵. Accurate detection and quantification are essential, particularly in systems with low parasite prevalence or multiple concurrent infections^16,17.

Recent advances in molecular techniques have significantly improved the detection and identification of reptilian hemoparasites. Diagnostic primers have been developed for various lizard blood parasites^18–22, allowing for more accurate assessments of parasite diversity and prevalence. These tools have enabled researchers to conduct more comprehensive surveys, particularly in tropical regions where parasite diversity is presumed to be high. As a result, our understanding of hemoparasite communities has expanded most rapidly in biodiversity hotspots. For instance, most known species of lizard malaria occur in tropical areas, with the highest species richness documented in the Americas and Africa⁶.

In contrast, research on reptilian hemoparasites in East Asia remains limited. To date, only very few species of lizard malaria, such as Plasmodium sasai²³, has been reported from the region, infecting lacertid lizards such as Takydromus tachydromoides in Japan, T. smaragdinus in the Amami Islands, and T. sexlineatus in Thailand²⁴. Similarly, very few species of saurian microfilaria, such as Conispiculum sp., has been documented, from the Okinawa tree lizard (Diploderma polygonatum)²⁵. These sparse records reflect a pronounced research bias and likely underestimate the true diversity of lizard hemoparasites in East Asia. Our field observations in Taiwan indicate that Diploderma swinhonis, a common and widespread species, may also host multiple hemoparasites, including Hepatozoon sp. and microfilariae. Yet, no systematic studies on lizard blood parasites have been conducted in Taiwan or neighboring areas. The lack of research represents a critical gap in understanding the ecological and evolutionary dynamics of hemoparasitism in the region.

In this study, we investigated the diversity, prevalence, and ecological correlates of hemoparasites in the most common endemic Taiwanese tree lizard, Diploderma swinhonis (Fig. 1). Using an integrated approach combining molecular identification, morphological features, and field-based ecological observations, we addressed the following questions: (1) Do Taiwanese populations of D. swinhonis harbor known or undescribed species of Hepatozoon and filarioid parasites? (2) How are parasite prevalence patterns shaped by geography, seasonality, and host characteristics such as age and sex? (3) Can morphological variation among island populations reveal potential cryptic diversity not captured by genetic markers? (4) How do newly developed molecular tools compare with traditional microscopy and commonly used primers in detecting hemoparasite infections? By integrating molecular diagnostics with ecological and morphological data, this study provides the first systematic assessment of blood parasites in Taiwanese reptiles and contributes new insights into hemoparasite taxonomy, transmission ecology, and detection methodology in East Asian lizard hosts.

Fig. 1.

Sampling sites include northern (Taipei, New Taipei City), eastern (Hualien, Taitung), western Taiwan (Pingtung), and offshore islands (Green Island, Orchid Island). Yellow and white circles indicate PCR-positive and PCR-negative sites, respectively.

Results

Methodological comparison

Among the 1,166 samples evaluated, the combination of microscopy and molecular methods yielded to a total of 18 Hepatozoon-positive and 44 microfilaria-positive individuals. Molecular methods exhibited higher sensitivity than microscopy for detecting both Hepatozoon and microfilariae (Table 1). PCR identified 18 Hepatozoon-positive (PCR⁺/POS⁺ = 18/18 = 100%) and 42 microfilaria-positive (PCR⁺/POS⁺ = 42/44 = 95.5%) individuals, whereas microscopy detected only 9 (50%) and 33 (75%) positive cases, respectively. These results suggest that molecular techniques are more effective for detecting hemoparasite infections than conventional microscopy.

Table 1.

Proportion of PCR-positive cases (PCR⁺) among all positive cases (POS⁺, defined as individuals confirmed positive by either microscopy or PCR), stratified by parasitemia level (low vs. high), and corresponding false negative rates (FNR) for each primer.

Hemoparasites	Marker	Low parasitemia level		High parasitemia level		Total
		PCR+/POS+	%	PCR+/POS+	%	PCR+/POS+	%	FNR
Hepatozoon	18S rDNA	6/13	46%	4/5	80%	10/18	56%	44%
	cytochrome b	12/13	92%	5/5	100%	17/18	94%	6%
Microfilariae	COI	19/20	95%	22/24	92%	41/44	93%	7%
	12S DNA	4/20	20%	16/24	67%	20/44	45%	55%

High: > 10 parasites per 1,000 red blood cells; low: ≤ 10 parasites per 1,000 red blood cells or undetected.

To further evaluate diagnostic performance, we compared the detection sensitivity of different genetic markers across parasitemia levels. For Hepatozoon, cytochrome b primers showed high sensitivity both in low-parasitemia individuals (detecting 12 out of 13 cases; PCR⁺/POS⁺ = 92.3%, FNR = 7.7%) and all 5 high-parasitemia samples (PCR⁺/POS⁺ = 100%, FNR = 0%). In contrast, 18S rDNA primers detected only 6 of 13 low-parasitemia samples (PCR⁺/POS⁺ = 46.2%, FNR = 53.8%) and 4 of 5 high-parasitemia samples (PCR⁺/POS⁺ = 80.0%, FNR = 20.0%).

For microfilariae, COI primers amplified parasite sequences in 19 out of 20 low-parasitemia samples (PCR⁺/POS⁺ = 95.0%, FNR = 5.0%) and in 22 out of 24 high-parasitemia samples (PCR⁺/POS⁺ = 91.7%, FNR = 8.3%). The 12S rDNA primers, however, showed reduced sensitivity, detecting 4 out of 20 low-parasitemia cases (PCR⁺/POS⁺ = 20.0%, FNR = 80.0%) and 16 out of 24 high-parasitemia cases (PCR⁺/POS⁺ = 66.7%, FNR = 33.3%). These results indicate that although all four markers are capable of detecting hemoparasites, cytochrome b primers outperform 18S rDNA for Hepatozoon, and COI primers outperformed 12S rDNA for microfilariae.

Molecular identification of hepatozoon in the tree lizard

We successfully acquired 18S rDNA sequences from nine Hepatozoon-positive D. swinhonis lizards collected in Taitung (n = 2), Green Island (n = 5), and Orchid Island (n = 2) (GenBank Accession No. PV474131 – PV474155). The sequences from Taitung are precisely identical to the GenBank entries assigned to Hepatozoon ophisauri (Tartakovskii, 1913) (p-distance = 0.0000; Fig. 2A), isolated from the Kashmir rock agama (Laudakia tuberculata) in Pakistan (OR976039–41)²⁶. Sequences from Green Island and Orchid Island were more similar to KF939624 (p-distance = 0.0017, with 1-bp difference), which was originally obtained from a king ratsnake (Elaphe carinata) collected in Shanghai, China²⁷.

Fig. 2.

Maximum likelihood (ML) phylogenetic tree of hemogregarines based on (A) 18S rDNA and (B) cytochrome b sequences. Node support values represent 1,000 ultrafast bootstrap replicates and Bayesian posterior probabilities. Specimens from this study were grouped within Hepatozoon ophisauri.

In addition, we sequenced the cytochrome b gene from our three sampling regions, including Taitung (N = 3), Green Island (N = 11), and Orchid Island (N = 2). All Taiwanese sequences formed a clearly monophyletic clade in the phylogenetic tree (Fig. 2B). Phylogenetically this clade is closest to H. catesbianae (isolated from the host Lithobates clamitans)²⁸ and H. clamatae (also from L. clamitans)²⁹. However, given the relative scarcity of Hepatozoon cytochrome b sequences in GenBank, cytochrome b is suitable for detecting Taiwanese samples but remains insufficient to establish a fully resolved phylogeny.

Molecular identification of microfilariae in the tree lizard

Phylogenies inferred from mitochondrial COI (Fig. 3A) and 12S rDNA (Fig. 3B) were used to assess interspecific divergence. Phylogenetic trees revealed a bifurcating topology within the Filarioidea. One major clade comprised members of the subfamilies Dirofilariinae, Onchocercinae, and Splendidofilariinae. Specimens obtained from D. swinhonis clustered within the Oswaldofilariinae clade, forming a lineage closely related to Oswaldofilaria chabaudi and O. petersi. The genetic divergence between our specimens and their closest relatives exceeded 12.8–13.4% in 12S rDNA and 18.6–23.9% in COI sequences. These levels of divergence suggest that the microfilariae observed in D. swinhonis cannot be assigned to any currently described species.

Fig. 3.

Maximum likelihood (ML) phylogenetic trees of microfilariae based on (A) mitochondrial COI and (B) 12S rDNA sequences. Node support values represent 1,000 ultrafast bootstrap replicates and Bayesian posterior probabilities. Specimens from this study were placed within the subfamily Oswaldofilariinae and may represent a novel taxon.

Morphological comparison

Morphology of hemoparasites were evaluated by examining blood smears from D. swinhonis individuals collected in Green Island and Orchid Island using light microscopy. For Hepatozoon sp. (Fig. 4), there are three among the seven morphological variables that represent significant differences between the two islands (Table 2), including area of the parasites (PA; Wilcoxon test: Z = −2.9, p = 0.004), nucleus width (PNW; Z = −4.1, p < 0.001), and nucleus area of parasites (PNA; Z = −4.3, p < 0.001). Hepatozoon in Green Island is significantly larger than that in Orchid Island.

Fig. 4.

Micrographs of Diff-Quik–stained blood films showing Hepatozoon ophisauri in Diploderma swinhonis. (A–D) Specimens from Green Island: (A) illustration of measurement parameters; (B) single gamont in an erythrocyte; (C) two gamonts in a single erythrocyte; (D) extracellular gamont. (E–F) Specimens from Orchid Island, showing significantly smaller parasite area (PA; Wilcoxon test: Z = –2.9, p = 0.004), parasite nucleus width (PNW; Z = –4.1, p < 0.001), and parasite nucleus area (PNA; Z = –4.3, p < 0.001) compared to those from Green Island.

Table 2.

Comparison of morphological traits (mean ± SE; Wilcoxon rank-sum test) between Green Island and Orchid Island.

Hepatozoon	Green Island (n = 43)	Orchid Island (n = 21)	Oswaldofilaria	Green Island (n = 10)	Orchid Island (n = 230)
PL (μm)	14.6 ± 0.1	14.2 ± 0.2	PL (μm)**	59.8 ± 0.9	54.7 ± 0.5
PW (μm)	3.3 ± 0.1	3.3 ± 0.1	PW (μm)***	9.3 ± 0.0	6.3 ± 0.1
PL/PW	4.4 ± 0.1	4.3 ± 0.1	PL/PW***	6.5 ± 0.1	8.7 ± 0.1
PA (μm2)**	45.5 ± 1.0	39.3 ± 1.5
PNL (μm)	5.1 ± 0.1	5.1 ± 0.2
PNW (μm)**	2.8 ± 0.1	2.3 ± 0.1
PNA(μm2)**	12.6 ± 0.3	9.7 ± 0.5

**: p < 0.01; ***: p < 0.001. Abbreviations: PL = parasite length; PW = parasite width; PL/PW = length/width ratio; PA = parasite area; PNL = parasite nuclear length; PNW = parasite nuclear width; PNA = parasite nuclear area.

For microfilariae (Fig. 5), all three variables represent significant difference between the two islands, including parasite length (PL; Z = 2.7 p = 0.006), parasite width (PW; Z = 5.3, p < 0.001), and parasite shape (L/W; Z = −4.8, p < 0.001) (Table 2). The microfilariae in Green Island is also significantly larger than that in Orchid Island.

Fig. 5.

Micrographs of Diff-Quik–stained blood films showing microfilariae in Diploderma swinhonis. (A) Illustration of measurement parameters; (A–B) specimens from Green Island; (C–D) from Orchid Island. Parasites from the two islands exhibited significant morphological differences in parasite length (PL; Z = 2.7, p = 0.006), width (PW; Z = 5.3, p < 0.001), and shape ratio (L/W; Z = –4.8, p < 0.001).

Prevalence across season and region

Seasonal variation in infection prevalence showed distinct patterns across locations and parasite taxa (Fig. 6). For Hepatozoon ophisauri (Table S1), prevalence was highest in Taitung during autumn (5.9%), whereas Green Island showed a peak in spring (6.5%) and a minor increase in winter (0.97%). Prevalence on Orchid Island remained low across all seasons, peaking slightly in autumn (1.1%).

Fig. 6.

Seasonal prevalence of (A) Hepatozoon ophisauri and (B) Oswaldofilaria sp. in Diploderma swinhonis populations from three regions in Taiwan: Taitung (orange), Green Island (green), and Orchid Island (blue). Bars represent the percentage of infected individuals detected by microscopy and molecular methods in each season.

In contrast, Oswaldofilaria sp. exhibited consistently higher prevalence on Orchid Island, with the highest infection rate observed in autumn (11.3%), followed by winter (8.4%) and summer (5.8%). Prevalence on Green Island was comparatively low but detectable in summer (0.98%) and winter (0.97). No microfilarial infections were detected in Taitung throughout the year (Table S2).

Relative risk (RR) in host

On Orchid Island at Chungai Bridge, adult lizards had a significantly higher risk of infection than juveniles, with an RR of 13.4 (p < 0.001, Table S3), indicating a 13.4 times greater risk for adults compared to juveniles. No significant difference was found in other populations and other groupings (Table S3).

Discussion

Identification of hepatozoon ophisauri and new distribution records

At present, species delimitation in hemoparasites and the extent of interspecific divergence across different genes are not well established. In our phylogeny reconstructed from 18S rDNA sequences (Fig. 2A), pairwise p-distances between most sister clades fall at approximately 3–8%. We therefore used this interval as a rough reference when comparing it with our sequence data.

Molecular analyses of the Hepatozoon-positive samples from D. swinhonis revealed high sequence identity with published 18S rDNA entries in the GenBank database. Specifically, sequences from Taitung were found to be identical to a reference sequence of Hepatozoon ophisauri (Tartakovskii, 1913), originally discovered from the European snake lizard Pseudopus apodus. This species was later redescribed by Zechmeisterová et al.³⁰ in 2019 using both molecular and morphological characters. Our identification was based on direct sequence comparison, which indicated 100% identity sequences with the Pakistan sample from the Kashmir rock agama (Laudakia tuberculata) (GenBank Accession No. MW495114), strongly supporting their conspecificity. Also given that L. tuberculata and D. swinhonis both belong to the family Agamidae, this match also suggests a likely host-parasite association within the same reptilian clade.

Interestingly, sequences obtained from Green Island and Orchid Island closely matched those of Hepatozoon chinensis, initially described from the king ratsnake (Elaphe carinata) in Shanghai, China²⁷. Snake hosts are typically infected with Hepatozoon via two routes: ingestion of an invertebrate containing sporulated oocysts, or predation on an infected intermediate vertebrate host such as a frog, lizard, or rodent^4,31. As invertebrates are not common prey for E. carinata, it is likely that infection occurs through predation on infected lizards^31,32. This trophic transmission pathway can lead to substantial synonymy between Hepatozoon species infecting reptiles at different trophic levels. Given the high genetic similarity, H. chinensis may in fact represent a junior synonym of H. ophisauri.

Notably, several GenBank entries labeled as H. chinensis are genetically identical to sequences assigned to H. ophisauri, highlighting the presence of overlapping genetic identities among nominal species. This inconsistency is further compounded by the inclusion of other taxa within the same clade, including H. erhardovae Krampitz, 1964³³, H. annularis (El-Naffar, Mandour, and Mohammed, 1991)³⁴, and H. ayorgbor Sloboda et al., 2007³², all of which exhibit highly similar or identical 18S rDNA sequences despite originating from taxonomically and geographically disparate hosts. Given that H. ophisauri has nomenclatural priority and includes sequences identical to the sequences obtained in this study, we consider it the most appropriate designation for the Taiwanese lineage of Hepatozoon identified in this study. Nevertheless, the presence of genetically indistinct lineages across diverse hosts suggests unresolved species boundaries and highlights the need for integrative taxonomic approaches combining molecular, morphological, and ecological data to properly delimit species in this genus.

The known host range of H. ophisauri has expanded in recent years, with confirmed infections in agamid lizards such as L. nupta and L. agrorensis, and suspected records in Trapelus megalonyx and Phrynocephalus maculatus from Pakistan²⁶. Our study adds D. swinhonis as a new agamid host for this lineage, and the first confirmed record of H. ophisauri in East Asia. This geographic extension further supports the hypothesis that this parasite has a broader host and ecological range than previously recognized.

Identification of microfilariae and morphological divergence across islands

Similarly, using the phylogenies inferred from mitochondrial COI (Fig. 3A) and 12S rDNA (Fig. 3B), we made a rough assessment of interspecific sequence divergence of microfilariae Among closely related taxa. Inspection of these trees indicates that p-distances between closely related species for both markers are approximately 4–9%. We therefore use this range as a rough benchmark for species delimitation in this study.

The phylogenetic position of the microfilariae observed in D. swinhonis indicates that these parasites likely belong to the subfamily Oswaldofilariinae. Our sequences clustered closely with Oswaldofilaria chabaudi (host = Tropidurus torquatus; Squamata: Iguania: Tropiduridae) and O. petersi (host = Crocodilurus amazonicus; Squamata: Lacertoidea: Teiidae), and more distantly with some species of Foleyellides spp. and Neofoleyellides spp. However, the genetic divergence between the Taiwanese samples and their closest known relatives exceeds 12% in both COI and 12S rDNA sequences—a threshold well beyond typical species-level differentiation in filarial nematodes. These data strongly support the presence of a previously undescribed, cryptic lineage of Oswaldofilariinae in Taiwanese agamids. While morphological data for adult worms remain poorly studied across potential taxa, the molecular evidence suggests that D. swinhonis harbors a unique filarioid parasite not yet reported elsewhere.

Another noteworthy indication of potential diversity is the marked morphological differentiation among island populations. Microfilariae from Green Island were consistently larger than those from Orchid Island in length (59.8 ± 0.9 μm vs. 54.7 ± 0.5 μm) and width (9.3 ± 0.0 μm vs. 6.3 ± 0.1 μm), and difference in shape with lower PL/PW ratios (6.5 ± 0.1 vs. 8.7 ± 0.1). All comparisons were highly significant (Wilcoxon rank-sum test, p < 0.01; Table 2). These differences may reflect local adaptation, phenotypic plasticity, or underlying cryptic diversity not fully resolved by current molecular markers.

A similar pattern of morphological divergence was also observed in Hepatozoon. Although 18S rDNA and cytochrome b sequences showed no island-specific variation, parasites from Green Island were significantly larger than those from Orchid Island in several traits, including parasite area (45.5 ± 1.0 vs. 39.3 ± 1.5 μm²), nuclear width (2.8 ± 0.1 vs. 2.3 ± 0.1 μm), and nuclear area (12.6 ± 0.3 vs. 9.7 ± 0.5 μm²) (p < 0.01; Table 2). These findings support the hypothesis that both parasite groups may exhibit local phenotypic variation in response to host physiology or environmental conditions. In the case of microfilariae, such size variation may also reflect circulatory adaptations, as suggested by McLaren³⁵, who noted that microfilarial morphology can vary with the fluid dynamics within the host.

Parasite dynamics in the lizard

Relative risks (RR) of infection were significantly higher in adult lizards compared to juveniles at one of the sampling sites, Chungai Bridge on Orchid Island (RR = 13.4, p < 0.001), indicating a strong age-dependent effect consistent with theoretical expectations³⁶. Age is associated with longer exposure time and larger body size, both of which contribute to higher parasite loads^37,38. A longer exposure period increases the probability of contact with vectors, while a larger host may present a more conspicuous or attractive target to parasites.

This age effect is further supported by observations from recaptured individuals. Four D. swinhonis lizards were recaptured during the study, three of which were parasite-free at first capture. However, two of these individuals later tested positive for hemoparasites after 7- and 9-month intervals, respectively. These cases underscore the value of mark-recapture data in understanding temporal infection dynamics and highlight the need for long-term monitoring in future field studies.

In terms of prevalence, our data revealed low hemoparasite infection rates (overall less than 3%) in D. swinhonis, which are considerably lower than those reported for some insular lizard species, such as Podarcis spp. (95–100%)^39,40. Although high host densities were observed on Green Island and Orchid Island, they were not associated with higher parasite prevalence. This discrepancy may reflect low vector abundance or a recent introduction of hemoparasites—both factors previously proposed in studies of other lizards^41,42 and in seabird systems^43,44. On small oceanic islets, wind exposure, high salinity, and the scarcity of freshwater habitats may limit the availability of suitable invertebrate vectors⁴⁵.

We also found that Hepatozoon and microfilariae exhibited different seasonal prevalence peaks across sites (Fig. 6). For Hepatozoon, the highest prevalence occurred in spring on Green Island, but in autumn in both Taitung and Orchid Island. In contrast, microfilariae peaked in summer on Green Island and in autumn on Orchid Island. These asynchronous patterns likely reflect differences in transmission ecology. Hepatozoon utilizes invertebrate definitive hosts (e.g., mites), whereas filarioids rely on vertebrates as definitive hosts and arthropods as intermediate hosts^31,46.

Our understanding of how these infections are shaped by local climatic variation and their interactions with other potential infection sources remains limited. Although seasonal variation of Hepatozoon infections has been discussed more extensively in endothermic hosts (e.g., birds or dogs)^47–49, comparable studies in ectotherms are far fewer. To our knowledge, most work quantifying Hepatozoon infection intensity has focused on the individual level, for example by tracking long-term fluctuations in captive animals^50,51. Our study is among the few to analyze population-level infection in an ectotherm. One limitation is that the populations we examined may lie near the northern distributional limit of this Hepatozoon species in Taiwan, as several more northerly populations yielded no detections. Consequently, overall prevalence was low, which makes our assessment of seasonality more susceptible to stochastic variation.

Comparison of detection methods and primer performance

Two diagnostic methods, i.e., microscopy and PCR assays, were evaluated for their effectiveness in detecting hemoparasites. PCR assays, particularly those using newly developed primers, demonstrated higher sensitivity than microscopy for both Hepatozoon and microfilariae (Table 1).

Most previous studies have used 18S rDNA primers to detect Hepatozoon infections¹⁹. The commonly used primer set HepF300/HepR900, designed by Perkins & Keller¹⁸, targets conserved regions of Hepatozoon 18S rDNA. However, our results show that these primers were effective only in specimens with high parasitemia levels. The limited performance in low-parasitemia samples may be due to non-specific amplification of homologous host DNA, thereby reducing detection accuracy. In contrast, PCR using our newly developed cytochrome b primers exhibited high sensitivity even in low-parasitemia samples, of which many of which were not detectable by microscopy. A similar pattern was observed for microfilarial detection. PCR primers previously developed for distantly related nematodes, such as Onchocerca volvulus, Ascaris suum, and Caenorhabditis elegans⁵², showed limited performance in detecting low-level infections. By contrast, our newly designed COI primers yielded significantly higher sensitivity across both high- and low-parasitemia samples, outperforming both microscopy and existing molecular assays.

These findings indicate that our newly developed primers for both Hepatozoon and microfilariae provide superior detection sensitivity compared to traditional methods. We recommend their adoption as reliable diagnostic tools for field surveys and ecological studies. Nevertheless, both PCR and microscopy have complementary strengths. Although PCR offers superior sensitivity, morphological examination remains essential for species differentiation—especially in newly surveyed regions, where existing primers may not reliably detect locally endemic or cryptic lineages.

Methods

Survey of infection in populations across taiwan

We conducted an initial field survey in 2014 to assess the potential occurrence of hemoparasites in Diploderma swinhonis across Taiwan. The sampling regime was designed to include diverse geographic regions across Taiwan, encompassing New Taipei City (n = 55), Taipei (n = 50), Hualien (n = 22), Taitung (n = 21), Pingtung (n = 30), and two nearby islets—Green Island (n = 30) and Orchid Island (n = 73) (Fig. 1). Parasite detection methods followed the protocols described in subsequent sections. This preliminary survey revealed that D. swinhonis was infected with Hepatozoon and microfilariae only in Taitung, Green Island, and Orchid Island (Fig. 1). Consequently, our study focused on these two hemoparasite taxa in populations from these three locations, where parasite prevalence is expected to stabilize during peak infection seasons.

Sample collection for the three infected areas

Based on the preliminary findings, we conducted a more intensive and regular sampling effort from April 2015 to April 2016 across six localities within the three focal regions: Liyushan Park in Taitung; Cross Mountain Trail and Ameishan Trail in Green Island; Xiaotian Pond, Weather Station, and Chungai Bridge in Orchid Island. From 2014 to 2016, a total of 1,166 lizards were captured and sampled. Each lizard was weighed to the nearest 0.01 g using an electronic scale, and snout-vent length (SVL) was measured to the nearest 0.01 mm using a digital caliper (Mitutoyo, Kanagawa, Japan). We recorded sex, age class, ectoparasite load, and any external symptoms of each individual.

Blood samples were collected during the toe-clipping marking procedure. A single blood smear was prepared per individual, air-dried, and fixed using Diff-Quik stain. Slides were stored at room temperature. All lizards were released at their original capture sites immediately after processing.

Microscopic observation and morphological measurements

Hemoparasites were detected by examining blood smears from D. swinhonis individuals collected in Green Island and Orchid Island using light microscopy. Infection status was assessed under a 400 × Leica optical microscope equipped with a Canon 5100D digital camera. All blood smears were examined by the same observer (YZW), who thoroughly scanned each slide, spending approximately six minutes per smear.

Once hemoparasites were detected, morphological measurements were taken using ImageJ 1.48 (National Institutes of Health, USA), with image scale calibrated by a stage micrometer. For Hepatozoon, we measured parasite length (PL), parasite width (PW), parasite area (PA), nucleus length (PNL), and nucleus width (PNW), following O’Dwyer et al.⁵³ (Fig. 4). For microfilariae, we measured PL and PW following Van As⁵ (Fig. 5).

Because the data did not meet the assumptions of parametric tests, we applied Wilcoxon rank-sum tests to compare parasite morphologies between Green Island and Orchid Island using JMP 16.0 (SAS Institute Inc., USA).

Molecular detection

Genomic DNA was extracted from toe tissue samples using the EasyPure Genomic DNA Spin Kit GT-100 (Bioman, Taiwan). Polymerase chain reactions (PCRs) were performed to detect potential infections of Hepatozoon and microfilariae. Reactions were carried out in a 20 μl volume containing 100–150 ng of genomic DNA, 10 μl of 2 × GoTaq® Green Master Mix (containing 400 μM each of dATP, dGTP, dCTP, and dTTP, and 3 mM MgCl₂), and 0.5–0.75 μl (10 μM) each of forward and reverse primers, where primer sequences and annealing temperatures are listed in Table S4. PCR procedures comprised an initial heating period of 3 min with 95 °C, 35 cycles of 94 °C for 45 s, annealing temperature (Table S4) for 30 s, 72 °C for 1 min, and a final extension at 72 °C for 5 min. PCR products were visualised by electrophoresis on 1.2% agarose gels stained with FluoroStain DNA Fluorescent Staining Dye (SMOBiO, Taiwan). Samples yielding positive amplicons were sequenced in both directions using an ABI 3730 automated sequencer (Genomics BioSci & Tech Corp., Taipei, Taiwan). The resulting sequences were checked and edited using Sequencher 4.9 (Gene Codes Corp., Boston, MA), and compared against available sequences in the GenBank database. Parasite identity was confirmed by BLASTn to ensure specific amplification from the hemoparasite genome.

Phylogenetic analyses

We included sequences of closely related taxa from the GenBank database in our phylogenetic analyses based on previous studies of hemogregarines (18S rDNA and cytochrome b) and onchocercid nematodes (COI and 12S rDNA). Four datasets were assembled: (1) partial 18S rDNA of hemogregarines (1,061 bp), consisting of 140 sequences including 9 from our samples, and 131 published sequences representing Hepatozoon, Dactylosoma, Haemogregarina, Hemolivia, Karyolysus, and other members of Adeleorina, with three Adeleidae taxa used as outgroups (Table S5); (2) partial cytochrome b of hemogregarines (552 bp), consisting of 30 sequences including 16 from our samples, and 14 published sequences (Table S6); (3) partial mitochondrial COI from filarial nematodes (570 bp), consisting of 196 sequences including 27 from our samples, and 169 sequences from six subfamilies of Onchocercidae (Dirofilariinae, Icosiellinae, Onchocercinae, Oswaldofilariinae, Splendidofilariinae, and Waltonellinae) and two sequences from Thelaziidae as outgroups (Table S7); and (4) partial mitochondrial 12S rDNA from filarial nematodes (416 bp), comprising 153 sequences including 20 from our samples, and sequences from above-mentioned taxa (Table S8).

All sequences were aligned using MUSCLE⁵⁴ implemented in MEGA 6.0⁵⁵. Phylogenetic trees were reconstructed for each gene using both maximum likelihood (ML) and Bayesian inference (BI) approaches. ML analyses were conducted in IQ-TREE 2⁵⁶ using ModelFinder Plus⁵⁷ to select the best-fit nucleotide substitution models based on the Bayesian information criterion (BIC): TVM + F + I + R2 for 18S rDNA, K3Pu + F + G4 for cytochrome b, GTR + F + I + R6 for COI, and TIM3 + F + I + R5 for 12S rDNA. Branch support was assessed using 1,000 ultrafast bootstrap (UFBoot) replicates⁵⁸. For Bayesian analyses, we used MrBayes 3.2.6⁵⁹, with substitution models selected using jModelTest 2.1.10⁶⁰ based on BIC: TPM3u + I + G for 18S rDNA, TIM1 + I + G for COI, and TPM1uf + I + G for 12S rDNA. As some models were not available in MrBayes, we used the closest available alternatives: GTR + I + G for 18S rDNA, K2P + I + G for cytochrome b, GTR + G for COI, and HKY + I + G for 12S rDNA. Each MCMC analysis was run for 4,000,000 generations with sampling every 1,000 generations, and the first 25% of trees discarded as burn-in. Convergence was assessed by ensuring that the average standard deviation of split frequencies fell below 0.01. Nodes with UFBoot values ≥ 95 and Bayesian posterior probabilities (PP) ≥ 0.95 were considered strongly supported.

Prevalence and relative risk evaluation

Seasonal prevalence was calculated for each population as the percentage of infected individuals among all sampled lizards in each season. To evaluate factors influencing hemoparasite load, we applied a relative risk (RR) framework to assess the association between host characteristics (specifically age and sex) and infection prevalence. Individuals were considered infected if hemoparasites were detected by either microscopy or molecular diagnostic methods. This analysis was conducted using data from three localities with sufficient sample sizes: Cross Mountain Trail and Ameishan Trail on Green Island, and Chungai Bridge on Orchid Island. Infections by Hepatozoon and microfilariae were combined for this analysis.

Relative risk (RR) was calculated to evaluate the association between parasite infection and categorical explanatory variables. RR was computed as the ratio of infection prevalence among the exposed group to that among the non-exposed group, i.e., the infection risks between adults and juveniles, and between males and females. Only lizards with confirmed sex and age were included in this analysis using the formula:

Fisher’s exact test was applied to determine whether the observed differences in RR between groups were statistically significant. All statistical analyses were performed using JMP 16.0 (SAS Institute Inc., USA).

Methodological comparison

We compared the sensitivity of microscopy and molecular diagnostic methods in two steps. First, a contingency table was constructed to identify false-negative results associated with each method. Second, we evaluated the sensitivity of different genetic markers used in molecular detection. Based on microscopy observations, samples were categorized into high and low parasitemia levels (high: > 10 parasites per 1,000 red blood cells; low: ≤ 10 parasites per 1,000 red blood cells or undetected). To assess marker performance, we analyzed Hepatozoon using 18S rDNA and cytochrome b, and microfilariae using COI and 12S rDNA. Detection rates were then compared between high and low parasitemia groups for each marker.

Author contributions

Y.-Z.W. and S.-M.L. conceptualized the ideas and designed the research; Y.-Z.W. conducted fieldwork and data collection; Y.-Z.W., C.-C.S., and Y.-C.C. analyzed the data; Y.-Z.W. and S.-M.L. wrote the original draft; Y.-C.C. and S.-M.L. reviewed and edited the final manuscript. All authors made significant contributions to the drafting process and approved the final version for publication.

Funding

This work was supported by National Science and Technology Council, Taiwan (grant numbers MOST 111–2621-B-003–001-MY3, NSTC 112–2621-B-003–002-MY3, and NSTC 114–2621-B-003–003-MY3).

Data availability

The datasets supporting the conclusions of this article are available in the NCBI (National Center for Biotechnology Information) GenBank repository under the accession numbers PV466197–PV466205, PX227321–PX227336, PV474131–PV474157, and PV491278–PV491297. The aligned sequences and tree files are provided in supplementary files.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

All procedures conducted during this study were approved by Institutional Animal Care and Use Committee (IACUC), National Taiwan Normal University and complied with Taiwan’s Wildlife Conservation Act. Diploderma swinhonis is not listed as a protected species in Taiwan, and all individuals were released at their original capture sites immediately after sampling, with handling time per lizard under 5 min.