The Amblyomma cajennense complex is represented by six species of hard ticks widely distributed from southern Texas in the United States to northern Argentina. Species within the complex feed on a variety of vertebrate hosts, primarily mammals, including humans, are the main vector of the bacterium Rickettsia rickettsii, and have been associated with the transmission of several arboviruses. In Colombia, one of the most diverse countries of the Neotropics, two species of the complex have been recorded with sympatric distribution in the northwestern part of the country: Amblyomma mixtum and Amblyomma patinoi. The latter is of medical importance since it has been confirmed as a competent vector of R. rickettsii in inter-Andean valleys of Colombia. Here, we assessed the current distribution of A. patinoi and predicted changes in its distribution over the century under different climate change scenarios. Additionally, we incorporated new records, delved deeper into the distribution of A. patinoi, and interpreted model results using the climate classification and ecoregions for the Neotropical region. The results showed the presence of A. patinoi in 79 localities distributed across 34 municipalities and seven departments of the Caribbean and Andean regions of Colombia. In addition, new records were identified in two additional departments (Valle del Cauca and Caquetá) of the Andean and Amazon regions, as well as additional records in the Department of Cundinamarca, especially in the inter-Andean Magdalena River basin. The species has been recorded in five ecoregions (Cauca Valley Dry Forests, Guajira-Barranquilla Moist Forests, Magdalena Valley Dry Forests, and Magdalena-Urabá Moist Forests) and is associated with horses, cattle, dogs, and humans. Its elevational range extends from 8 to 645 m in the Caribbean region and from 497 m to 1712 m in the Andean region and inter-Andean valleys. The potential distribution models suggest that A. patinoi presents high climatic suitability in the Caribbean areas and inter-Andean valleys. Under climate change scenarios, a possible expansion of the species distribution is observed in areas currently not recorded in the Neotropics, in countries such as Brazil, Peru, and Venezuela. However, geographical conditions, such as elevation, could limit its distribution at higher elevations.

Occurrence data

The search for occurrence records of A. patinoi was conducted on the Web of Science database, without time restrictions. Scientific and review articles were considered if they presented records of ticks, with information on their respective host (Nava et al., 2014; Faccini-Martínez et al., 2015; Acevedo-Gutiérrez et al., 2021; Quintero et al., 2021; Segura et al., 2022; Martínez-Díaz et al., 2023; Polo et al., 2024). Additionally, we searched for records using the Global Biodiversity Information Facility (GBIF; https://www.gbif.org/). To complement the information obtained from the searches, we reviewed the hard ticks deposited in the ectoparasite collection of the Veterinary Parasitology Laboratory of the National University of Colombia (CVP-UN) in Bogotá. To determine the identity of the reviewed specimens, we morphologically identified hard ticks to the species level following the extended descriptions of the species of the Amblyomma cajennense complex by Nava et al. (2014, 2017), using optical microscopy (Nikon SMZ1270 stereoscope). We did not perform molecular identification because we failed to amplify DNA from five specimens deposited in the CVP-UN collections (CVP-UN 170, 158_1, 158_2, 173_1, and 173_2). In cases where precise coordinate data were not available, the centroid at the locality or municipality level was used.

Current and potential distribution

To evaluate the current distribution of A. patinoi, we filtered the records from all information sources and assigned them to the political areas (departments) and ecoregions. For the ecoregions, we followed the classification proposed by Dinerstein et al. (2017) and specified for Colombia by García et al. (2023), as well as the climate classification according to the Köppen-Geiger scheme (Beck et al., 2018).

To estimate the potential distribution and the environmental niche models (ENMs), we used the software Wallace v2.1.1 (Kass et al., 2018) and R (R Development Core Team, 2023), in which the MaxEnt 0.1.4 algorithm was implemented through the maxnet package (Phillips et al., 2017). This implementation produces results similar to those obtained using the original MaxEnt software (Phillips et al., 2017), as supported by recent findings (Sillero et al., 2023). We used 15 bioclimatic variables (1970–2000) to construct ENMs from the WorldClim database version 2.0 (Fick & Hijmans, 2017; available: http://www.worldclim.org) with a resolution of 2.5 Arcmin (~5 km²), excluding those that combine temperature (BIO8 and BIO9) and precipitation (BIO18 and BIO19) due to their odd spatial anomalies and discontinuities between neighboring pixels (Escobar et al., 2014; Moo-Llanes et al., 2021). In addition to removing duplicates, we applied a 10 km spatial layer to eliminate records from very close areas and reduce spatial bias, using the spThin R package (Aiello‐Lammens et al., 2015; Kass et al., 2018). We defined the accessible area (M) by creating buffers with a 100 km radius around the occurrence points. This selection was based on tick dispersal, which typically occurs within relatively restricted geographic boundaries (<100 km), determined by the movement patterns of their hosts (Springer et al., 2015; Hekimoglu et al., 2023). The selected distance allows for realistically modeling potentially accessible and suitable areas, avoiding underestimating regions of high climatic suitability that could be colonized. For the projections, this region was expanded to include the countries bordering Colombia: Brazil, Ecuador, Panama, Peru, and Venezuela (the projection area was extended 28 degrees to the east, 7 degrees to the west, and 10 degrees to the south). We also included other South American countries such as Guyana, French Guyana, and Suriname (Figure S1). The occurrence records were randomly divided into two subsets: 70% for model calibration and the remaining 30% for final evaluation. Using the “random k-fold” method, the occurrence locations were randomly divided into four containers (k=4).

We built the models using feature classes that span a wide range of complexities: L (linear), LQ (linear-quadratic), H (hinge), LQH (linear – quadratic - hinge), and LQHP (linear, quadratic, hinge and product) with regularization multipliers from 0.5 to 4, and multiplier step set to 0.5. Model accuracy was evaluated using three of the most commonly used metrics: 1) Omission rates, calculated at the 10th percentile training presence threshold (Peterson et al., 2008), 2) The area under the curve (AUC) of a receiver operating characteristic plot for test occurrences (i.e., the mean of the k test AUCs). [avg.test.AUC; Peterson et al., 2011]); and 3) the Akaike Information Criterion corrected for finite sample sizes (AICc), used to evaluate the complexity of the model (Warren & Seifert, 2011). AICc was calculated only in the full model, so k partitions were not considered (Kass et al., 2018). We selected the best models based on the following criteria: 1) models with the highest AUC, and 2) if more than one best model was available, those with the lowest omission rate or AICc. Thus, we generated the final models using the best parameterizations and the complete set of occurrences with logistic results.

Climate change scenarios

We projected the ENMs for two time periods: A) Current conditions, and B) Future conditions, specifically using ECHAM6 (Stevens et al., 2013) for the years 2050 and 2070, modeled using two emissions scenarios (Representative Concentration Pathways, RCP) from the Fifth Assessment Report (AR5) (4.5 and 8.5; IPCC, 2013). The ECHAM6 model (MPI-ESM-LR) features an improved representation of radiative transfer in the shortwave (solar) spectrum, a new description of aerosol effects, enhanced surface albedo representation, including the treatment of melt zones in sea ice, and an improved representation of the middle atmosphere as part of the standard model (Stevens et al., 2013; Aguillar-Dominguez et al., 2021). Although the CCSM4 model is commonly used in Colombia due to its consideration of ENSO simulations (IPCC, 2013), we selected ECHAM6 for its specific advantages in radiative transfer and aerosol effects.

RESULTS

Occurrence data and current distribution

The results of the bibliographic review indicated the presence of A. patinoi in 75 localities across 30 municipalities and seven departments within the Caribbean and Andean regions of Colombia, including Antioquia, Bolívar, Cauca, Córdoba, Cundinamarca, Magdalena, and Sucre (Figures 1 and 2A). We found 232 occurrences in the GBIF, of which 199 provided locality or coordinate information, adding four localities in the municipalities of Nimaima, Nocaima, Caparrapí, and Útica in the Department of Cundinamarca, all situated in the Andean region. In addition, we identified new specimens in the CVP-UN collection, covering two more departments: Valle del Cauca, on the inter-Andean Cauca River basin, and Caquetá, on the Andean slopes of the Amazon basin (Figure 2A). A previously undocumented locality was also detected in the Magdalena River basin of the Department of Cundinamarca, specifically in the municipality of Mesitas (Figure 2A). Females identified as A. patinoi exhibit diagnostic traits for the species, including: the pseudoscutum has elongated cervical spots that do not merge posteriorly with limiting spots; apex of festoons with small chitinous tubercles at the inner angle; and a U-shaped genital opening with short, domed fins (Figure 1).

In total, A. patinoi has validated records in seven ecoregions of the Andean and Caribbean zones (Figure 2C, Table 2). The most represented ecoregion was the Magdalena-Urabá Moist Forests, with 17 localities across four departments (Bolívar, Córdoba, Santander, and Sucre; Figures 2B and 2C). In addition, there are records for four other ecoregions: i) the Guajira-Barranquilla Xeric Scrubland in two departments (Sucre: two localities, Bolívar: one locality; Figures 2B and 2C), ii) the Northwest Andean Montane Forests in the departments of Antioquia and Cauca (three localities, Figures 2B and 2C), iii) the Montane Forests of the Valley of the Magdalena in the department of Cundinamarca (six localities, Figures 2B and 2C), and iv) the north part of the Dry Forests of the Cauca Valley in Antioquia (Figures 2B and 2C, Table 2). Of the new records obtained through the review of biological collections, one is located in the southern part of the Dry Forests of the Cauca Valley ecoregion in the Department of Valle del Cauca (Figures 2B and 2C). The two new additional localities documented in this study are in two ecoregions in which A. patinoi had not previously been recorded: the Montane Forests of the Eastern Cordillera, in the Department of Cundinamarca, and in the Napo Moist Forests in the Department of Caquetá (Figures 2B and 2C). Amblyomma patinoi has been recorded free-ranging in vegetation and is primarily associated with horses (Equus caballus), cattle (Bos taurus), and, in some cases, humans and dogs. Its altitudinal range extends from 8 m to 645 m in the Caribbean region and from 497 m to 1712 m in the Andean region (Table 2).

Potential distribution

A total of 232 unique occurrence records were included in the dataset (Table S1). Of these, 156 data points were retained after the cleaning of occurrences. We evaluated 40 candidate models, of which we selected the best five models due to the low omission rate (OR), low complexity (AICc), and good performance (AUC) (Table S2). The best model included two linear-quadratic (LQ) features with a regularization parameter of 0.5 (OR = 0.156, AUCc = 0.92, and AICc = 8.6). Other models showed slightly lower performance values, with OR and AUC within acceptable ranges (Table S3). The variables that contributed most to habitat suitability of A. patinoi were isothermality, temperature seasonality, minimum temperature of the coldest month, annual temperature range, annual precipitation, and precipitation of the driest month. The greatest relative contribution of the environmental variables to the model was isothermality (BIO3: 11.28%), the minimum temperature of the coldest month (BIO6: 3.054%), and the precipitation of the driest month (BIO14: -3.1%). In this sense, the climatic space occupied by A. patinoi seems to be defined by areas with minimal sudden changes in temperature (high isothermality), suggesting that thermal conditions during the coldest month may influence the distribution of A. patinoi in the region. Furthermore, we highlight the relative contribution of precipitation from the driest month, which was negative, indicating that the lack of precipitation during this month is a significant limiting factor for the habitat suitability of A. patinoi in this specific region.

In Colombia, the current distribution model reveals a high suitability for A. patinoi in the Caribbean and Andean regions (Figure 2B). Specifically, its high suitability is evident in much of the Caribbean region, specifically throughout the Magdalena-Urabá Wet Forests ecoregion, which includes the north of Antioquia, as well as the departments of Córdoba, Sucre, and Bolívar. Additionally, parts of the Guajira-Barranquilla Xeric Scrub also show high suitability (Figures 2B and 2C). In the Andean region, areas of high climatic suitability are predicted mainly in the inter-Andean valleys (Figure 2B). This includes the southern part of the Magdalena-Urabá Moist Forests, encompassing parts of the departments of Santander, Boyacá and Cundinamarca, as well as the Dry Forests ecoregion of the Cauca Valley, which spans the central regions of the departments of Antioquia, Caldas, Risaralda, and Quindío (Figures 2B and 2C). Other notable areas included the Dry Valley of Patía, covering part of the departments of Cauca and Nariño, and a previously undocumented region, the Dry Forests of the Magdalena Valley, which includes parts of Cundinamarca and Huila (Figure 3B).

According to the climate classification proposed by Köppen-Geiger in Colombia, the current model indicates that the climates with the greatest suitability for A. patinoi in the Caribbean region and the inter-Andean valleys include the tropical savanna, the tropical monsoon or sub-equatorial, and parts of the rainy equatorial climates (Figure 2D). Notably, high suitability is not observed in other regions with similar climates, such as the Orinoco and Amazon regions (Figure 2D). This suggests that A. patinoi prefers warm and humid climates, thriving in areas with high temperatures and abundant precipitation, which provide optimal humidity levels for its survival.

Beyond Colombian boundaries, several ecoregions in Venezuela also exhibited a highly suitable climate for A. patinoi (Figure 3). This includes the entire region of the Maracaibo Dry Forests in the State of Zulia and portions of the Xeric Shrublands of the Coast in the State of Anzoátegui (Figure 3B). In the southern part of Colombia, a shared ecoregion with Ecuador, specifically the Northwest Andean Montane Forests, also offers favorable climatic conditions for A. patinoi. This region spans the provinces of Carchi and Imbabura in the north of Ecuador, and Azuay, El Oro, and Loja in southern Ecuador (Figure 3B). In Panama, the portions of the Chocó-Darién Moist Forests and Eastern Panamanian Montane Forest present areas of high suitability, particularly in the provinces of Darién, Embera Wounaan, Guna Yala, Madugandi, and Wargandí (Figure 3B). In Peru, the Montane Forests region of the Eastern Cordillera presents areas of high suitability, particularly in the departments of Cajamarca and Amazonas (Figure 3B). In Brazil, two notable areas of high suitability have been identified: the north of the ecoregions of the Humid Forest of the Foothills of Guyana (Amazonas) and the Savannah of Guyana (Roraima), along with moderately suitable areas in part of the Japurá-Solimões-Negro Moist Forests ecoregion (Amazonas) (Figure 3B). Additionally, a small core area within the Guiana Lowland Rainforest ecoregion, located in the states of Demerara-Mahaica and Mahaica-Berbice in Guyana, is also predicted to be suitable (Figure 3B).

In general, the ecoregions that exhibit the greatest environmental suitability for A. patinoi are the Valley Dry Forests and the Moist Forests, with four ecoregions represented for each category. Following closely are the Xeric Scrub, Montane Forests, and Dry Forests, with two represented ecoregions. Additionally, Mangroves and savannas also show high climatic suitability, albeit to a lesser extent (Figure 3).

Climate change scenarios

The projections of model transfers to future conditions of climate change indicate a high stability of suitable regions for A. patinoi in Colombia, except for the Magdalena Valley Dry Forests ecoregion, which shows a notable loss of suitable characteristics (Figures 4B and 4D). Conversely, the Guajira Xeric Scrub ecoregion gains suitability across all climate change scenarios (Figure 4). In the evaluated scenarios, stable suitability persists in regions of Panama, Ecuador, and Venezuela. However, increases in suitability are noted in parts of Brazil and Peru (Figure 4). In Brazil, the states of Maranhão and Pará exhibit ideal climatic conditions under both RCP 4.5 and 8.5 scenarios for 2050 and 2070 (Figure 4). In Peru, while there is an increase in suitability for a large part of the department of Loreto under the RCP 4.5 scenario for both timeframes (Figures 4A and 4B), a loss of suitability is predicted for the same regions under the high emissions RCP 8.5 scenario.