Skip to main content
NCBI home page
GeoHealth logoLink to GeoHealth
. 2026 Apr 10;10(4):e2024GH001146. doi: 10.1029/2024GH001146

Spatial and Temporal Patterns of Campylobacter Infection and Projected Habitat Suitability of Dominant Campylobacter Species in Eastern Ethiopia

Xiaolong Li 1,2, Amanda E Ojeda 3, Loic Deblais 4, Bahar Mummed Hassen 5, Mussie Bhrane 5, Gireesh Rajashekara 4, Song Liang 1,2, Jemal Yousuf Hassen 5, Sarah L McKune 1,6,7, Arie H Havelaar 2,3,7, Jason K Blackburn 2,8,✉; CAGED Research Team
PMCID: PMC13066913  PMID: 41969794

Abstract

Campylobacter is the most common bacterial cause of foodborne illness globally. Both symptomatic and asymptomatic infections with Campylobacter species have been associated with growth faltering of children in low‐resource settings, while previous prevalence studies primarily focused on diarrheal disease in children. Here, we leverage the data collected from the Campylobacter Genomics and Environmental Enteric Dysfunction (CAGED) project to characterize the spatial patterns of Campylobacter infections among infants with or without diarrhea in rural Eastern Ethiopia. Randomly enrolled infants (n = 106) were followed from birth to around 13 months, with fecal samples collected monthly. Livestock feces, drinking water, and soil samples were collected biannually. Campylobacter was detected and quantified using genus‐specific PCR and species‐specific PCR for four species. We employed a spatial filtering approach using genus‐specific data to generate smoothed prevalence surfaces by month and age group. Temporally, an upward trend of prevalence was observed as the children grew older. Spatially, high‐prevalence areas were distributed across the whole study area. To relate disease risk to environmental conditions, we used ecological niche modeling with MaxEnt to estimate habitat suitability of the genus Campylobacter and two dominant species identified by PCR results. Elevation, vegetation index, and slope were the most important contributors, and all distribution models suggested areas in the north were more likely to support the pathogen. These results inform Campylobacter infection patterns and identify target areas with higher risk of Campylobacter in low‐resource settings. This further contributes to developing effective intervention strategies in the future.

Plain Language Summary

Campylobacters are species of bacteria that cause food‐borne illnesses, leading to clinical signs including diarrhea, fever, and abdominal pain. These infections are particularly concerning in low‐resource settings where they can significantly affect the health and long‐term development of young children. Understanding how these infections spread and identifying high‐risk areas can help in developing effective prevention strategies. Here, we investigated the prevalence of Campylobacter infections in infants in rural Eastern Ethiopia, both with and without diarrhea, to understand their spread and impact on child growth. We followed 106 infants from birth to about 13 months, collecting fecal samples monthly and environmental samples from livestock, water, and soil twice a year. Campylobacter was detected using polymerase chain reaction (PCR) genetic tests, and the infection rates were mapped over time and across different locations. We found that as children aged, the prevalence of Campylobacter increased, and certain areas consistently showed higher infection rates. Using environmental data, we also predicted which areas were most suitable for the bacteria, identifying elevation, vegetation, and slope as key factors. These findings highlight regions with higher infection risks, offering valuable insights for future public health interventions to prevent and control Campylobacter infections in vulnerable populations.

Key Points

  • Prevalence of Campylobacter infections among infants in rural eastern Ethiopia showed an upward trend in their first year of life

  • Campylobacter infections were distributed across the study area, with some regions having persistently higher prevalence over time

  • MaxEnt‐based distribution models predicted northern areas were more likely to support the pathogen at both the genus and species level

1. Introduction

Species in the genus Campylobacter are known as the most common bacterial cause of foodborne illness worldwide and contribute to approximately 166 million cases each year in both industrialized and developing nations (A. H. Havelaar et al., 2015). Furthermore, a body of evidence shows that Campylobacter infections have been associated with stunting and environmental enteric dysfunction (EED), a subclinical disorder characterized by inflammation and altered function in the small intestine (Keusch et al., 2014), among children under 5 (CU5) in low‐resource settings (Amour et al., 2016; Haque et al., 2019; Lee et al., 2013; Rogawski et al., 2018). Both symptomatic and asymptomatic infections with Campylobacter spp. are linked to these health effects.

Diarrhea that can be attributed to Campylobacter colonization is often observed in Africa, and previous studies reported the prevalence of Campylobacter infection among CU5 ranged from 1.7% to 15.5% in several African countries (Gahamanyi et al., 2020). As most of the samples collected in these studies came from children with diarrhea, the actual prevalence of Campylobacter colonization among CU5 is likely to be underestimated in Africa, given that asymptomatic Campylobacter infections are often very common in these settings (Amour et al., 2016; Platts‐Mills & Kosek, 2014).

Over 25 species of Campylobacter have been identified, of which Campylobacter jejuni and Campylobacter coli are the two most commonly reported species associated with human illnesses (Kaakoush et al., 2015). Due to the clinical importance of these two species, traditional culture methods were designed to selectively isolate these two species, resulting in a large body of knowledge on disease burden and clinical manifestations specific to C. jejuni and C. coli (François et al., 2018). Owing to the development of molecular techniques and innovative isolation approaches, a number of non‐C. jejuni/coli species (i.e., Campylobacter species other than C. jejuni and C. coli) have drawn increasing attention and have been recognized as “emerging Campylobacter species” (Man, 2011). They include C. concisus, C. lari, C. upsaliensis, and C. ureolyticus. The Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development (MAL‐ED) project used both enzyme immunoassay (EIA) and polymerase chain reaction (PCR) to detect Campylobacter in stool samples and found that EIA identified a broader range of Campylobacter spp. (Platts‐Mills et al., 2014). In a subset of Campylobacter‐positive samples from MAL‐ED, non‐C. jejuni/coli species accounted for around 30% of detections by EIA. Moreover, association analysis of Campylobacter infection on child growth revealed that infection with Campylobacter species detected by EIA showed a stronger association with child stunting compared to C. jejuni/coli infection detected by PCR specific for the two species (Haque et al., 2019), implying the underlying role of non‐C. jejuni/coli species in affecting poor child growth. Our previous cross‐sectional study on Campylobacter colonization and child health outcomes (i.e., EED and stunting) among children in eastern Ethiopia suggested that non‐C. jejuni/coli species including C. hyointestinalis, C. fetus, and C. concisus detected by meta‐total RNA sequencing (MeTRS) were more prevalent in child stool samples than C. jejuni/coli (Chen, McKune, et al., 2021; Terefe et al., 2020).

Recent studies also showed the emergence of a new species, “Candidatus Campylobacter infans” (C. infans) isolated from stool samples of children under 2 years in low‐resource settings along with other non‐C. jejuni/coli species (Bian et al., 2020; Garcia Bardales et al., 2022; Parker et al., 2022). All this evidence draws our focus in this study to the broader species within the genus Campylobacter rather than just C. jejuni/coli.

Given the high disease burden, a great effort has been made to identify the risk factors and transmission pathways of Campylobacter bacteria (specifically C. jejuni and C. coli), as well as their associations with child health outcomes in Africa (Amour et al., 2016; Budge et al., 2020; El‐Tras et al., 2015; Lee et al., 2013; Lengerh et al., 2013; Luby et al., 2018; McQuade et al., 2020; Null et al., 2018; Schiaffino et al., 2021; Stewart et al., 2018). However, these patterns have yet to be fully addressed spatially or temporally. For instance, did areas with a high prevalence of Campylobacter infection among children persist in space and time? Did the geographic focus of Campylobacter infections shift over time within a specific pediatric population and how? The dynamics of such infection patterns over space and time in African countries are seldom discussed in the literature. Addressing these questions will greatly facilitate the development of effective and sustainable intervention strategies to combat this enteric pathogen among young children in low‐ and middle‐income countries (LMICs).

The spatial/temporal pattern analysis of Campylobacter infections will help capture the heterogeneity of infections on the landscape. Still, we also need to understand which environmental conditions are suitable for the pathogen to infect young children. One approach to modeling such suitability is ecological niche modeling. MaxEnt, which employs a maximum entropy approach, is one of the most commonly used methods for ecological niche modeling (Elith et al., 2006; Elith & Leathwick, 2009). Generally, ecological niche models estimate the ecological niches of species in variable space and project the potential geographic distribution across the landscape (Alexander et al., 2012; Blackburn, 2010). MaxEnt uses presence‐only data and applies a machine learning algorithm to estimate a target probability distribution that has a maximum similarity to the prior distribution (represented by random background samples) while taking account of the constraints imposed by the incomplete information derived from the presence locations (Elith et al., 2011; Merow et al., 2013; S. J. Phillips et al., 2006). MaxEnt has been widely applied in diverse ecology, evolution, and conservation biology scenarios to study species' distributions, richness and biodiversity (Elith et al., 2006). Similar to other types of species, a pathogen also has its corresponding ecological niche characterized by ecological parameters (Mullins et al., 2011). Within such a conceptualized N‐dimensional space, the pathogen can be maintained. Previous studies regarding relationships of environmental variables and Campylobacter species primarily focused on the potential impact of the natural environment on human campylobacteriosis cases (Arsenault et al., 2012; Kuhn et al., 2020; Louis et al., 2005; Soneja et al., 2016; Weisent et al., 2014), but did not investigate the species niches for Campylobacter in certain settings. Here, we used MaxEnt to estimate the habitat suitability of Campylobacter species in a low‐resource setting of Ethiopia.

The Campylobacter Genomics and Environmental Enteric Dysfunction (CAGED) project is a longitudinal study conducted in rural Eastern Ethiopia to examine the reservoirs of Campylobacter spp. infections among infants and their interactions with child health outcomes (A. H. Havelaar et al., 2022). In the present work, by using the data collected for the CAGED project, we aim to:

  1. Characterize the spatial and temporal patterns of Campylobacter infection among children under 1 year in Eastern Ethiopia, and

  2. Estimate the habitat suitability of dominant Campylobacter species in the study area using ecological niche models.

2. Materials and Methods

2.1. Study Design and Protocols

The detailed CAGED study design and protocols have been described elsewhere (A. H. Havelaar et al., 2022). Briefly, this longitudinal study was conducted in the rural Haramaya woreda (district equivalent) of Eastern Ethiopia where 10 kebeles (the smallest sub‐district administrative unit in Ethiopia) were selected as the source population (Figure 1). A total of 106 infants were randomly enrolled from the source population at birth and followed up until approximately 13 months of age. Enrollment started in December 2020 and was completed in May 2021. Up to 20 newborns were enrolled per month until the desired sample size was achieved. Starting the second month of the longitudinal study, both enrollment and follow‐ups took place simultaneously, resulting in temporally disjoint follow‐up periods for children enrolled at different months. Fecal samples were collected from enrolled infants monthly and from their mothers and siblings as well as, livestock (i.e., chicken, cattle, goat, and sheep) present in or around the homestead, and the environment (soil and drinking water) biannually. Sample collection was conducted from December 2020 to June 2022, and a total of 1073 child fecal samples and 1361 environmental samples were collected.

Figure 1.

Figure 1

Open in a new tab

Study area and enrolled households in the Campylobacter Genomics and Environmental Enteric Dysfunction (CAGED) project. Gray dots represent the locations of 106 enrolled households. Panel (a) black triangles represent the presence locations used to build MaxEnt models at the genus level (see Section 2). Panels (b, c) blue and red dots represent the presence locations for two dominant Campylobacter species, C. jejuni and C. infans, respectively. Abbreviations: AW—Adele Walta; AM—Amuma; BK—Bachake; BG—Biftu Geda; DA—Damota; GC—Gobe Challa; IO—Ifa Oromia; KR—Kuro; NG—Nageya; QD—Qerensa Dereba.

In the laboratory, DNA extraction and quantification were performed, followed by genus‐specific Taqman real‐time PCR targeting 16S rRNA to detect and quantify Campylobacter (n = 2,717 samples). A positive Campylobacter infection was defined as the cycle threshold (Ct) value of a child fecal sample being less than 35 (Deblais et al., 2023). To assess the diversity of Campylobacter infections detected in our study population, species‐specific SYBR Green real‐time PCR was performed. Species‐specific primers targeting hipO and cpn60 were used. A total of four Campylobacter species including thermophilic and nonthermophilic species were tested, including: Candidatus C. infans, C. jejuni. C. upsaliensis and C. lari. Other species were rarely, if ever, detected in infant stools. A cutoff Ct value of 35 was used to detect Campylobacter species in the human stool, animal feces, and soil (n = 2,045 samples). Nuclease‐free water was utilized as a negative control for all qPCR testing. Further details on laboratory methods are provided in Deblais et al. (2023) and in a separate manuscript describing the species composition of Campylobacter (Ojeda et al., 2025).

2.2. K‐Means Clustering for Dividing Age Groups

As previous evidence showed that the prevalence of Campylobacter infections differed between age groups of children under 2 (Amour et al., 2016), we divided our study population into four age groups to account for the confounding role of child age in Campylobacter infections and to investigate the infection patterns among different age groups.

K‐means clustering is a partitioning method that divides a data set into k clusters through an iterative process of searching for the centroid (mean) in each cluster that minimizes the within‐cluster sum of squares (Hartigan & Wong, 1979). Here, we first calculated the age (days) of the children when their fecal samples were collected in each home visit and then partitioned the child age data set into four groups. We opted for four clusters to align with a previous study on the same population, which divided children into quartiles based on age (Chen et al., 2024). While the previous study used quartiles as a straightforward statistical method, we sought to adopt a more data‐driven approach that accounts for the natural structure of our data set. The use of four groups allows for direct comparability with the prior study while providing a clustering solution reflective of the data's inherent patterns. K‐means clustering was performed in R software using the kmeans function.

2.3. Spatial Filtering Approach

To estimate the infection pattern of Campylobacter among infants over the study area and during the study period, a spatial filtering approach in the Disease Mapping and Analysis Program (DMAP, available for download at http://www.uiowa.edu/~gishlth/DMAP/) was employed to generate smoothed prevalence surfaces of Campylobacter infections based on the monthly genus‐specific qPCR data defining infection status.

DMAP is an exploratory spatial analysis tool first applied to estimate local birth defect rates with georeferenced birth data in Iowa (Rushton & Lolonis, 1996). A detailed description of the spatial filtering in DMAP can be found elsewhere (Lentz et al., 2011). Briefly, DMAP applies a circular spatial filter with a radius defined by the user, traversing grid points situated over the landscape. The prevalence at each grid intersection point is then calculated based on the number of presence/absence points falling within the filter circle which can cover multiple grid cells. In this analysis, we generated a gridded surface covering the study area with 250‐m grid cells. To determine the radius of the spatial filter, we first calculated the average k‐nearest neighbor distances at different order (k = 1, 2, …, 10) using the geographic coordinates of the 106 enrolled households in R. While exploring larger orders (k > 2), the DMAP software encountered limitations in handling these distances effectively on our relatively small landscape, resulting in computational issues and unreliable outputs. To address this, we selected k = 2 (780.53 m) as it represented a practical trade‐off, providing sufficient spatial resolution while ensuring compatibility with the software and reliability of the analysis. This approach allowed us to capture meaningful spatial patterns within the constraints of our data set and analysis tools.

The presence/absence of Campylobacter infection data were organized by age group, and DMAP estimation was performed for each month in each age group. To ensure that enough points could be used in each DMAP run, we removed those months with less than 15 data points (presence and absence records). Grid point prevalence estimates for each month and age group were then interpolated using inverse distance weighting (IDW) with a power of 2 in ArcMap 10.8.2 (ESRI; Redlands, CA) to generate continuous surfaces (raster files) of prevalence rates. To identify areas of increased prevalence, IDW‐interpolated rates were mapped as standard deviations (SD) from the mean.

Owing to the limited number of data points available for the two dominant species in comparison to the genus level, similar analyses focusing on these two specific species were not conducted.

2.4. Space Time Analysis of Moving Polygons (STAMP)

Areas of prevalence greater than 1.5 SD were used to identify persistent hot spots of Campylobacter infections over time for each age group through the STAMP approach. STAMP was designed to delineate the space‐time change of polygon movement events between two consecutive time points T1 and T2 (Robertson et al., 2007). The types of polygon movement events can be categorized as stable, extraction, expansion, generation, and disappearance based on the intersection of polygons at T1 and T2. Stable type refers to polygons that are present at both time points. In this study, a stable polygon is an area where the prevalence of Campylobacter infection has been persistently high for two consecutive months. If polygons only exist at either time point, they are categorized as generation (only present at T2) or disappearance (only present at T1). That is, the area of high prevalence only exists in one of the two consecutive months. Extraction and expansion are two special cases of generation and disappearance; extraction is the disappearance events that are adjacent to stable events, while expansion is the generation events adjacent to stable events (Figure 2).

Figure 2.

Figure 2

Open in a new tab

Types of polygon movement events defined in the Space‐Time Analysis of Moving Polygons (STAMP). Blue circles and green circles represent areas at Timepoint 1 and Timepoint 2, respectively. The overlapping region of blue and green circles is defined as stable, which exists at both time points, while the rest parts adjacent to stable are contraction type (only present at Timepoint 1) and expansion type (only present at Timepoint 2). Disappearance and generation are events independent from other areas at either timepoint.

Smoothed prevalence surfaces from DMAP were first reclassified by SD and converted to polygons in ArcMap, and the STAMP analysis was then performed for each age group separately using R package stampr (Long et al., 2018). Stable areas of high prevalence of Campylobacter infection were the focus of this study and were defined as any polygons that persist for more than two consecutive months. Since STAMP only deals with two consecutive months for each run, stable polygon layers obtained from multiple runs were overlaid by age group, and intersections of the stable layers were calculated using the “Intersect” tool in the toolbox of ArcMap.

2.5. Environmental Variables for Ecological Niche Modeling

The DMAP analysis was used to map the heterogeneity of infection on the landscape but cannot be used to relate disease risk to environmental conditions. For this purpose, we used ecological niche modeling with MaxEnt, including 13 covariates covering topography, climate, vegetation, soil, and human population density. These were selected based on relevant literature modeling the prevalence of diarrhea‐related infections locally and globally (Colston et al., 2022; Reiner et al., 2020; Sanderson et al., 2018) (Table 1). Raster layers of elevation, slope, and population counts (2020) for Ethiopia were downloaded from WorldPop (https://www.worldpop.org/) with a spatial resolution of 100 m. Daily land surface temperature (LST) and 16‐day normalized difference vegetation index (NDVI) grids were obtained from the Terra Moderate Resolution Imaging Spectroradiometer (MODIS) Land Surface Temperature/Emissivity Daily (MOD11A1) Version 6.1 product (https://lpdaac.usgs.gov/products/mod11a1v061/) and MODIS Vegetation Indices (MOD13Q1) Version 6.1 product (https://lpdaac.usgs.gov/products/mod13q1v061/), respectively. The minimum, maximum, and mean of these two products over the study area in 2021 were calculated using the Code Editor in Earth Engine (https://code.earthengine.google.com/), and 6 covariates for daily LST and 16‐day NDVI were generated correspondingly. Grids representing soil features, namely the proportion of clay particles in the fine earth fraction, organic carbon content in the fine earth fraction, and soil pH, were downloaded from SoilGrids version 2.0 (https://soilgrids.org/) at a 250‐m spatial resolution. Wind speed at 100 m above ground data were from the Global Wind Atlas (https://globalwindatlas.info/en/). As these environmental layers had varying spatial resolutions, all were resampled to 250 × 250 m grids and cropped to the boundary of the study area using the “raster” package in R.

Table 1.

Environmental Covariates Used in the MaxEnt Models

Variable Unit Spatial resolution Data source
Elevation m 100 m WorldPop (www.worldpop.org/)
Maximum daily land surface temperature, 2021 °C 1 km USGS MODIS product (MOD11A1)
Mean daily land surface temperature, 2021 °C 1 km USGS MODIS product (MOD11A1)
Minimum daily land surface temperature, 2021 °C 1 km USGS MODIS product (MOD11A1)
Maximum 16‐day NDVI, 2021 N/A 250 m USGS MODIS product (MOD13Q1)
Mean 16‐day NDVI, 2021 N/A 250 m USGS MODIS product (MOD13Q1)
Minimum 16‐day NDVI, 2021 N/A 250 m USGS MODIS product (MOD13Q1)
Population count, 2020 person 100 m WorldPop (www.worldpop.org/)
Slope degree 100 m WorldPop (www.worldpop.org/)
Proportion of clay particles in the fine earth fraction g/kg 250 m SoilGrids 2.0 (https://soilgrids.org/)
Soil organic carbon content in the fine earth fraction dg/kg 250 m SoilGrids 2.0 (https://soilgrids.org/)
Soil pH pH × 10 250 m SoilGrids 2.0 (https://soilgrids.org/)
Wind speed at 100 m above ground m/s 250 m Global Wind Atlas (https://globalwindatlas.info/en/)
Open in a new tab

2.6. Ecological Niche Modeling

MaxEnt (Version 3.4.4) software (Philips et al., n.d.) was used to model the potential geographic distribution of the genus Campylobacter as well as two dominant species identified by PCR results, C. jejuni and C. infans, over the study area (Deblais et al., 2023). To better understand the pathogen's ecological context and to enhance the predictive accuracy of ecological niche models, we used the environment samples collected from drinking water, soil, and livestock feces in enrolled households for modeling. The purpose of these samples in this study was not to characterize the diversity or distribution of Campylobacter species, but to determine the presence or absence of Campylobacter at each household. If Campylobacter was detected in any one of the environmental samples collected from a given household, that household was assigned a positive status for the purposes of ecological niche modeling. Details on the composition and distribution of Campylobacter species in these environmental samples were presented in a separate study from our research team (Ojeda et al., 2025).

As Campylobacter was identified by genus‐specific qPCR at least once from the environmental samples collected at each household (Deblais et al., 2023) during the study period, we initially had 106 presence sites available for the ecological niche modeling at the genus level. In a presence/absence framework, data points need to be spatially filtered to one point per grid cell as the resolution of the climatic conditions (here 250 m) (Joyner et al., 2010). This filtering reduced our data set to 93 presence locations (Figure 1a). Similarly, a total of 78 and 69 presence locations were obtained for C. jejuni and C. infans, respectively (Figures 1b and 1c). For all MaxEnt models, the maximum number of background points was set as 1000 due to the relatively small area of the landscape and the number of grid cells. The input data set was randomly split into 80% for training and 20% for testing, and we ran 10 replicates with a bootstrapping procedure to account for the variations between models using different sets of training/test points.

We retained all environmental covariates in the final model for two reasons. First, their inclusion ensured consistency with previous studies, facilitating comparability with existing research. Second, this approach allowed for the exploration of potential differences in covariate contributions between genus‐level and species‐level models, where some covariates might have varying impacts. By including all covariates, we maintained consistency across comparative analyses. Model performance was evaluated by the area under the receiver operating curve (AUC). An AUC of 0.5 indicates the model performance is no better than random, and, empirically, values greater than 0.8 are used as an indicator of good prediction (Kumar et al., 2020).

The average of these 10 bootstrapped models was used to estimate the habitat suitability of Campylobacter genus and each of the two species on the landscape. To make the interpretation of the suitability map easier, we described the output as “low” suitability at 0.25, “medium” suitability at 0.5, and “high” suitability at 0.75 following similar studies (Bartlow et al., 2022; Gorris et al., 2021). The uncertainty of our models was presented by the standard deviation of habitat suitability at each pixel for the 10 replicates.

We used the mean values of the built‐in permutation importance and percent variable contribution output from the 10 bootstrapped models as the metric to identify the contribution/importance of environmental covariates. In each iteration of the model training process, MaxEnt monitors the change of gain, a metric of goodness‐of‐fit comparing the average log probability of the presence samples against a uniform distribution where the gain is defined as 0, after modifying the coefficient of a specific covariate. The greater the gain is, the higher the average likelihood of the presence samples in the model. The percent contribution is obtained by converting the total gain assigned to each environmental covariate into a percentage at the end of the training process (S. J. Phillips, 2021). Unlike percent contribution, permutation importance of each covariate only depends on the final MaxEnt model. Values of a specific environmental variable are randomly permuted among both presence and background training points, and the resulting decrease in training AUC is used to measure the extent to which the model depends on that variable. A large decrease means that the corresponding variable is important to the final model. Values of decrease for all environmental variables are converted to percentages after normalization as their permutation importance.

3. Results

3.1. Prevalence of Campylobacter Infections by Age Group

The minimum and maximum child age when fecal samples were collected were 7 and 376 days, respectively. The k‐means clustering gave us four age groups: 0–95 days, 96–177 days, 178–285 days, and >285 days. The monthly prevalence of Campylobacter infections in the first age group (0–95 days) ranged from 0.08 to 0.70, covering the months of December 2020 to August 2021 (Figure 3). The monthly prevalence of Campylobacter infections in age Group 2 (96–177 days) ranged from 0.35 to 0.62 from March 2021 to November 2021. For older age groups, the monthly prevalence increased further: 0.42–0.94 for age Group 3 (178–285 days) and 0.79–1.0 for age Group 4 (>285 days). No apparent seasonality was observed in each age group, but the prevalence peaks across all age groups occurred between June and October.

Figure 3.

Figure 3

Open in a new tab

Prevalence of Campylobacter infections among infants in 10 kebeles of Ethiopia from December 2020 to June 2022 by age group. Note, age group does not directly correspond to month due to the enrollment schedule; no more than 20 children were enrolled per month.

3.2. Prevalence Surfaces From the DMAP Analysis

DMAP generated a series of smoothed prevalence surfaces for each age group by month. Spatially, areas with a higher prevalence of Campylobacter infections were distributed across the whole study area (Figure 4 and Figure S1 in Supporting Information S1). Four kebeles, Qerensa Dereba (QD) and Kuro (KR) in the north, Adele Walta (AW) in the southwest, and Nageya (NG) in the south, covered regions that had a relatively higher prevalence over time. Results from the STAMP analysis further demonstrated that these stable hot spots were distributed in central Qerensa Dereba (QD), the southeast of Kuro (KR), the south of Adele Walta (AW), as well as the southwest and southeast of Nageya (NG) (Figure 5). This spatial pattern tended to be similar among different age groups, with the exception that stable hot spots were only found in three kebeles in the south for age Group 1.

Figure 4.

Figure 4

Open in a new tab

Smoothed prevalence surfaces of Campylobacter infections in different age groups by month in 10 kebeles of Ethiopia. Age Group 1: 0–95 days; age Group 2: 96–177 days; age Group 3: 178–285 days; age Group 4: >285 days. Those eight maps from 4 months were selected as representatives to demonstrate the spatial patterns of high/low‐prevalence areas. The full set of maps can be found in the Figure S1 of Supporting Information S1. Abbreviations: AW—Adele Walta; AM—Amuma; BK—Bachake; BG—Biftu Geda; DA—Damota; GC—Gobe Challa; IO—Ifa Oromia; KR—Kuro; NG—Nageya; QD—Qerensa Dereba. (a) Age group 1 in May 2021; (b) Age group 2 in May 2021; (c) Age group 1 in June 2021; (d) Age group 2 in June 2021; (e) Age group 3 in December 2021; (f) Age group 4 in December 2021; (g) Age group 3 in January 2022; (h) Age group 4 in January 2022.

Figure 5.

Figure 5

Open in a new tab

Distribution of areas with persistently high Campylobacter infection prevalence by age group in 10 kebeles of Ethiopia. Shaded areas with colors represent regions of persistently high Campylobacter prevalence. A—D: age groups 1–4; E: overlaps among four age groups. Abbreviations: AW—Adele Walta; AM—Amuma; BK—Bachake; BG—Biftu Geda; DA—Damota; GC—Gobe Challa; IO—Ifa Oromia; KR—Kuro; NG—Nageya; QD—Qerensa Dereba.

Meanwhile, several areas on the landscape had a constant lower prevalence than the average (less than 1.5 standard deviations). The prevalence in the southern part of Amuma (AM) has been persistently lower across the four age groups. The southeast corner of Ifa Oromia (IO) and the northwest of Gobe Challa (GC) had lower prevalence only in age Groups 1 and 2; as well as northern Ifa Oromia (IO), the west part of Bachake (BK), and the south part of Damota (DA) only in age Groups 3 and 4.

3.3. Ecological Niche Models

For the genus‐level MaxEnt model, the average training AUC of 10 replicates was 0.841, with a standard deviation of 0.011 (Figure S2 in Supporting Information S1). The habitat suitability map indicated that Amuma (AM) and central Kuro (KR) in the north, the southern part of Ifa Oromia (IO) in the central west, and the southwest of Damota (DA) in the central east were highly suitable for the pathogen persistence compared to other regions (Figure 6a). There were also several areas with low habitat suitability. These were distributed in the southeastern Kuro (KR), adjacent areas of southeastern Bachake (BK) and east Damota (DA), and west parts of Gobe Challa (GC) and Nageya (NG).

Figure 6.

Figure 6

Open in a new tab

Habitat suitability maps for Campylobacter in 10 kebeles of Ethiopia. (a) genus Campylobacter (b) C. jejuni (c) C. infans. The values at each pixel were derived from the average of 10 bootstrapped replicates. Abbreviations: AW—Adele Walta; AM—Amuma; BK—Bachake; BG—Biftu Geda; DA—Damota; GC—Gobe Challa; IO—Ifa Oromia; KR—Kuro; NG—Nageya; QD—Qerensa Dereba.

C. jejuni and C. infans were recovered from environmental samples collected at 87 and 74 households, respectively. Both species were found to coexist in samples from 35 households. The suitability maps for the two dominant species (Figures 6b and 6c) showed similar patterns compared to the genus level. There was no apparent difference in the distribution of high/low suitable areas between these two species. Overall, kebeles in the north were predicted to be more suitable for the pathogen persistence than in the south. By overlaying the suitability maps to the smoothed prevalence surfaces generated from DMAP, we found that Qerensa Dereba (QD), part of Kuro (KR), and Adele Walta (AW) had both high habitat suitability and high prevalence of Campylobacter infection. Meanwhile, some areas, for example, southern parts of both Amuma (AM) and Damota (DA), had higher habitat suitability but lower prevalence.

The overall level of variation on the values at each pixel between 10 replicates of these three MaxEnt models was relatively low, with only a few regions showing a high standard deviation (Figure S3 in Supporting Information S1). That is, the habitat suitability in most regions of our study area was quite similar between the 10 bootstrap runs, suggesting relatively robust models obtained for the species distribution of Campylobacter in the study area.

Percent contribution and permutation importance are two measures of variable contributions in MaxEnt. Among the three MaxEnt models, elevation, mean 16‐day NDVI, and slope had the top 3 highest percent contribution (Table 2). However, the order of these three environmental covariates' permutation importance demonstrated a nuanced difference between models. In the genus model, the mean 16‐day NDVI had the highest permutation importance, followed by slope and elevation. While elevation had the highest value in the models for the two species. For the suitability of C. jejuni, slope became less important, and the minimum daily temperature was ranked the second highest after elevation. Similarly, the minimum daily temperature replaced the mean 16‐day NDVI and became more important in the C. infans model. For each environmental covariate in the MaxEnt model, a response curve was generated by plotting the predicted occurrence probability against the range of values of the corresponding variable. The range of the most suitable environmental conditions for the species can be identified through the response curves. The peaks in the response curves of the three most important contributors indicated that a moderate elevation of around 2100 m, a moderate mean NDVI value of around 0.40, and a gentle slope (∼5°) are highly suitable for the presence of Campylobacter at both the genus and species level (Figure S4 in Supporting Information S1).

Table 2.

Percent Contribution and Permutation Importance of Environmental Variables in Three MaxEnt Models

Variable Percent contribution Permutation importance
Genus C. jejuni C. infans Genus C. jejuni C. infans
Elevation 19.9 a 24.1 15.4 12.8 19.3 18.9
Mean 16‐day NDVI, 2021 14.5 16 15.5 20.6 14.7 8.2
Slope 12.4 17 17.4 16.7 9.3 14.5
Population count, 2020 9.6 6.2 10.5 3.2 4.4 7
Minimum daily land surface temperature, 2021 9.5 10.3 11.8 6.8 15.7 16.1
Mean daily land surface temperature, 2021 8.8 4.8 2.3 9 4.4 3.4
Maximum 16‐day NDVI, 2021 7.6 6.2 8.6 4.5 3.5 6.9
Maximum daily land surface temperature, 2021 5.6 4.5 2.9 6.5 6.2 1.1
Proportion of clay particles in the fine earth fraction 3.9 4 5.4 3.1 6.3 6.8
Soil organic carbon content in the fine earth fraction 2.6 1.1 1.5 3.7 4.2 2.5
Wind speed at 100 m above ground 2.6 1.7 2 8.2 7.2 5.7
Minimum 16‐day NDVI, 2021 2.4 3.6 5.8 3.2 3.7 6.7
Soil pH 0.6 0.6 1.1 1.6 1 2.2
Open in a new tab
a

Bold numbers represents the highest value in each column.

4. Discussion

Here, we estimated the prevalence of Campylobacter infection in a cohort of children from birth to 13 months of age in a rural part of eastern Ethiopia. We also mapped prevalence to evaluate spatial patterns of prevalence and used ecological niche modeling to predict environmental suitability for pathogen persistence using environmental covariates describing soil conditions, elevation, vegetation indices, and human population. Previously, the prevalence of Campylobacter infections (specifically C. jejuni/coli) in CU5 in Ethiopia was estimated at 14.5%, primarily based on studies of diarrheic children in hospitals or clinic centers (Gahamanyi et al., 2020). Given the high frequency of asymptomatic colonization of Campylobacter among young children in LMICs, the actual prevalence could be much higher than the estimate. The results from the formative research of the CAGED project in Ethiopia supported this assumption, showing a much higher prevalence of Campylobacter colonization than previous estimates—50% among children 11–13 months of age when those with or without symptoms are included (Chen, McKune, et al., 2021). When examining each individual age group identified by K‐means clustering, we observed a clear upward trend of the prevalence among infants as they grew older, which is consistent with the findings from the MAL‐ED study (Amour et al., 2016). As documented in detail elsewhere (Deblais et al., 2023), we observed that prevalence reached 80%–100% for children older than 285 days. Peaks of prevalence across all age groups occurred between June and October during the study period.

The generation of smoothed prevalence surfaces by age group and month suggests that Campylobacter is ubiquitous in this relatively small region, both spatially and temporally. However, that doesn't necessarily imply a uniform spatial distribution of infections with this pathogen. Local areas with persistently higher/lower prevalence over time were identified in all age groups. High‐prevalence areas were found in the north, southwest, and south of the study region, and these areas overlapped more or less through all four age groups. Several regions with constantly low prevalence were distributed adjacent to these high‐prevalence areas in the north and central areas. What environmental/non‐environmental factors contribute to this pattern is beyond the scope of this study and needs further investigation.

In this study, we first examined the habitat suitability for Campylobacter at the genus level and then for two dominant species over the study area using MaxEnt models with environmental covariates. The MaxEnt models in this study provided a robust prediction of Campylobacter distribution, and the suitability maps predicted by the MaxEnt models confirmed the environmental signal suggested by the DMAP analysis ‐ Campylobacter is ubiquitous but not uniformly distributed. Due to this ubiquitousness of Campylobacter species across the landscape, the suitability maps from both genus and species‐specific models did not show any apparent differences in the distribution of high/low suitable areas. Environmental covariates including elevation, slope, and NDVI contributed to predicting the distribution of Campylobacter genus and species. On the other hand, we also noted that some areas, for example, southern parts of both Amuma (AM) and Damota (DA), had higher habitat suitability for Campylobacter but a lower prevalence of Campylobacter infection after comparing the suitability map with the smoothed prevalence surfaces. Since ecological niche models only tackle the ecological covariates or abiotic signals (Brookes et al., 2020), this suggests that potentially biotic factors, for example, interactions between infants and other human or livestock reservoirs of the pathogen, may affect the distribution of these bacteria in the study area.

It is important to note that environmental samples, including drinking water, soil, and livestock feces, were used in this study solely to inform ecological niche models predicting the environmental suitability for Campylobacter presence. These models do not establish direct epidemiological links or confirm transmission routes from environmental sources to children. As such, while the ecological patterns observed suggest that environmental conditions may contribute to shaping the distribution of Campylobacter in this setting, further studies are needed to establish causal relationships between environmental conditions and Campylobacter carriage in children. Specifically, integrating molecular typing or whole‐genome sequencing of isolates from both human and environmental samples would provide more definitive evidence of transmission pathways. Notably, another study from our team, which applied multilocus sequence typing (MLST), has been published and explores the transmission dynamics of Campylobacter between infants and environmental sources in this population (Singh et al., 2025). That work provides complementary strain‐level evidence to support the hypotheses generated from the current ecological modeling approach.

In addition, we acknowledge that PCR‐based detection in environmental samples does not distinguish between live and dead bacteria, and therefore, the presence of Campylobacter DNA may not always reflect viable organisms capable of transmission. While our ecological niche modeling approach relies on presence/absence data to identify areas of potential environmental suitability, we recognize that this method does not confirm exposure risk or viability.

Unlike in HICs where the sources of Campylobacter infections are frequently linked to the consumption of contaminated food, the risk factors for Campylobacteriosis in LMICs are more associated with the environmental pathways, for example, contaminated drinking water and soil, and direct contact with domestic animals (Budge et al., 2020; Coker et al., 2002; Schiaffino et al., 2021). Multilevel factors are involved in these pathways. At the regional level, climatic covariates could be drivers in spreading the pathogen in the environment. For example, runoff formed during the rainy season, along with rainfall itself, might lead to the contamination of drinking water and potentially facilitate the dispersion of Campylobacter. Meanwhile, household‐level variables may play a more direct role in influencing the interaction between children and Campylobacter (Chen, Mechlowitz, et al., 2021). For example, the practice of livestock husbandry on or in the homestead could increase children's exposure to the pathogen through contacting animal feces or soil contaminated by animal feces. Thus, factors at both levels should be considered when further explaining the spatial pattern of Campylobacter infections.

Due to the relative sparseness of data points for the two dominant species compared to the genus level, we did not perform DMAP analysis on these two species to examine the monthly patterns of infections among children. This was a limitation in our study. Though suitability maps suggested no difference between both levels, it is still worthwhile to characterize infections with these two dominant species among young children in the study area and determine their temporal patterns in future studies.

Hot spots of Campylobacter infections identified through the DMAP and STAMP analyses could be targeting areas for future interventions, especially those areas with a persistently high prevalence. MaxEnt‐based distribution map has suggested that environmental covariates, including elevation, slope, and the mean NDVI, were major factors influencing the distribution of Campylobacter at both genus and species levels. It also demonstrated areas that are more likely to support the distribution of Campylobacter and should be emphasized in future Campylobacter surveillance programs in the study area. Though not addressed in this study, identifying potential factors that contribute to low prevalence in highly suitable areas for Campylobacter will be worthwhile in future studies, and relevant results will guide the development of effective intervention strategies for protecting children from Campylobacter infections and associated long‐term health effects.

Conflict of Interest

The authors declare no conflicts of interest relevant to this study.

Supporting information

Supporting Information S1

GH2-10-e2024GH001146-s003.pdf (472.2KB, pdf)

Movie S1

Download video file (398.1KB, mp4)

Movie S2

Download video file (428.4KB, mp4)

Movie S3

Download video file (375.3KB, mp4)

Movie S4

Download video file (463.7KB, mp4)

Acknowledgments

This work is a result of the CAGED Research Team whose members include: Amanda Evelyn Ojeda, Arie H. Havelaar, Abadir Jemal Seran, Abdulmuen Mohammed Ibrahim, Bahar Mummed Hassen, Belisa Usmael Ahmedo, Cyrus Saleem, Dehao Chen, Efrah Ali Yusuf, Gireesh Rajashekara, Getnet Yimer, Ibsa A. Ahmed, Ibsa Aliyi Usmane, Jafer Kedir Amin, Jason K. Blackburn, Jemal Y. Hassen, Kedir A. Hassen, Kunuza Adem Umer, Karah Mechlowitz, Kedir Teji Roba, Loic Deblais, Mussie Bhrane, Mark J. Manary, Mawardi M. Dawid, Mahammad Mahammad Usmail, Nigel P. French, Nur Shaikh, Nitya Singh, Sarah L. McKune, Wondwossen A. Gebreyes, Xiaolong Li, Yenenesh Demisie Weldesenbet, Yang Yang, Zelalem Hailu Mekuria. This project is funded by the United States Agency for International Development Bureau for Food Security under Agreement #AID‐OAA‐L‐15‐00003 as part of Feed the Future Innovation Lab for Livestock Systems and by the Bill & Melinda Gates Foundation OPP#1175487. Under the grant conditions of the Foundation, a Creative Commons Attribution 4.0 Generic License has already been assigned to the Author Accepted Manuscript version that might arise from this submission. Any opinions, findings, conclusions, or recommendations expressed here are those of the authors alone. Research reported in this publication was supported by the University of Florida Clinical and Translational Science Institute, partly supported by the NIH National Center for Advancing Translational Sciences under award number UL1TR001427.

Availability Statement

The data on which this article is based are available in Havelaar (2025).

References

  1. Alexander, K. A. , Lewis, B. L. , Marathe, M. , Eubank, S. , & Blackburn, J. K. (2012). Modeling of wildlife‐associated zoonoses: Applications and caveats. Vector Borne and Zoonotic Diseases, 12(12), 1005–1018. 10.1089/vbz.2012.0987 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Amour, C. , Gratz, J. , Mduma, E. R. , Svensen, E. , Rogawski, E. T. , McGrath, M. , et al. (2016). Epidemiology and impact of campylobacter infection in children in 8 low‐resource settings: Results from the MAL‐ED study. Clinical Infectious Diseases: An Official Publication of the Infectious Diseases Society of America, 63(9), 1171–1179. 10.1093/CID/CIW542 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Arsenault, J. , Berke, O. , Michel, P. , Ravel, A. , & Gosselin, P. (2012). Environmental and demographic risk factors for campylobacteriosis: Do various geographical scales tell the same story? BMC Infectious Diseases, 12(1), 1–12. 10.1186/1471-2334-12-318/FIGURES/2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bartlow, A. W. , Giermakowski, J. T. , Painter, C. W. , Neville, P. , Schultz‐Fellenz, E. S. , Crawford, B. M. , et al. (2022). Modeling the distribution of the endangered Jemez Mountains salamander (Plethodon neomexicanus) in relation to geology, topography, and climate. Ecology and Evolution, 12(8), e9161. 10.1002/ECE3.9161 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bian, X. , Garber, J. M. , Cooper, K. K. , Huynh, S. , Jones, J. , Mills, M. K. , et al. (2020). Campylobacter abundance in breastfed infants and identification of a new species in the global enterics multicenter study. mSphere, 5(1), e00735‐19. 10.1128/msphere.00735-19 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Blackburn, J. K. (2010). Integrating geographic information systems and ecological niche modeling into disease ecology: A case study of bacillus anthracis in the United States and Mexico. In O'Connell K., Skowronski E., Sulakvelidze A., & Bakanidze L. (Eds.), Emerging and endemic pathogens. NATO science for peace and security series A: Chemistry and biology. Springer Netherlands. 10.1007/978-90-481-9637-1_7 [DOI] [Google Scholar]
  7. Brookes, V. J. , Belkhiria, J. A. , Machado, G. , & Escobar, L. E. (2020). Ecological niche modeling: An introduction for veterinarians and epidemiologists. Review, 7, 519059. 10.3389/fvets.2020.519059 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Budge, S. , Barnett, M. , Hutchings, P. , Parker, A. , Tyrrel, S. , Hassard, F. , et al. (2020). Risk factors and transmission pathways associated with infant Campylobacter spp. prevalence and malnutrition: A formative study in rural Ethiopia. PLoS One, 15(5), 1–17. 10.1371/journal.pone.0232541 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chen, D. , McKune, S. , Yang, Y. , Usmane, I. A. , Ahmed, I. A. , Amin, J. K. , et al. (2024). Campylobacter colonization and undernutrition in infants in rural Eastern Ethiopia: A longitudinal community‐based birth cohort study. Frontiers in Public Health, 12, 1467462. 10.3389/FPUBH.2024.1467462 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chen, D. , McKune, S. L. , Singh, N. , Yousuf Hassen, J. , Gebreyes, W. , Manary, M. J. , et al. (2021). Campylobacter colonization, environmental enteric dysfunction, stunting, and associated risk factors among young children in rural Ethiopia: A cross‐sectional study from the campylobacter genomics and environmental enteric dysfunction (CAGED) project. Frontiers in Public Health, 8, 1043. 10.3389/fpubh.2020.615793 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chen, D. , Mechlowitz, K. , Li, X. , Schaefer, N. , Havelaar, A. H. , & McKune, S. L. (2021). Benefits and risks of smallholder livestock production on child nutrition in low‐ and middle‐income countries. Frontiers in Nutrition, 8, 751686. 10.3389/FNUT.2021.751686 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Coker, A. O. , Isokpehi, R. D. , Thomas, B. N. , Amisu, K. O. , & Larry Obi, C. (2002). Human campylobacteriosis in developing countries. Emerging Infectious Diseases, 8(3), 237–243. 10.3201/eid0803.010233 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Colston, J. M. , Zaitchik, B. F. , Badr, H. S. , Burnett, E. , Ali, S. A. , Rayamajhi, A. , et al. (2022). Associations between eight Earth observation‐derived climate variables and enteropathogen infection: An independent participant data meta‐analysis of surveillance studies with broad spectrum nucleic acid diagnostics. GeoHealth, 6(1), e2021GH000452. 10.1029/2021GH000452 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Deblais, L. , Ojeda, A. , Brhane, M. , Mummed, B. , Hassen, K. A. , Ahmedo, B. U. , et al. (2023). Prevalence and load of the campylobacter genus in infants and associated household contacts in rural eastern Ethiopia: A longitudinal study from the campylobacter genomics and environmental enteric dysfunction (CAGED) project. Applied and Environmental Microbiology, 89(7), e00424‐23. 10.1128/aem.00424-23 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Elith, J. , Graham, C. H. , P. Anderson, R. , Dudík, M. , Ferrier, S. , Guisan, A. , et al. (2006). Novel methods improve prediction of species' distributions from occurrence data. Ecography. 10.1111/j.2006.0906-7590.04596.x [DOI] [Google Scholar]
  16. Elith, J. , & Leathwick, J. R. (2009). Species distribution models: Ecological explanation and prediction across space and time. Annual Review of Ecology Evolution and Systematics, 40(1), 677–697. 10.1146/annurev.ecolsys.110308.120159 [DOI] [Google Scholar]
  17. Elith, J. , Phillips, S. J. , Hastie, T. , Dudík, M. , Chee, Y. E. , & Yates, C. J. (2011). A statistical explanation of MaxEnt for ecologists. Diversity and Distributions, 17(1), 43–57. 10.1111/J.1472-4642.2010.00725.X [DOI] [Google Scholar]
  18. El‐Tras, W. F. , Holt, H. R. , Tayel, A. A. , & El‐Kady, N. N. (2015). Campylobacter infections in children exposed to infected backyard poultry in Egypt. Epidemiology and Infection, 143(2), 308–315. 10.1017/S095026881400096X [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. François, R. , Yori, P. P. , Rouhani, S. , Siguas Salas, M. , Paredes Olortegui, M. , Rengifo Trigoso, D. , et al. (2018). The other campylobacters: Not innocent bystanders in endemic diarrhea and dysentery in children in low‐income settings. PLoS Neglected Tropical Diseases, 12(2), e0006200. 10.1371/journal.pntd.0006200 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Gahamanyi, N. , Mboera, L. E. G. , Matee, M. I. , Mutangana, D. , & Komba, E. V. G. (2020). Prevalence, risk factors, and antimicrobial resistance profiles of thermophilic campylobacter species in humans and animals in Sub‐Saharan Africa: A systematic review. International Journal of Microbiology, 2020, 1–12. 10.1155/2020/2092478 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Garcia Bardales, P. F. , Schiaffino, F. , Huynh, S. , Olortegui, M. P. , Yori, P. P. , Vasquez, T. P. , et al. (2022). “Candidatus Campylobacter infans” detection is not associated with diarrhea in children under the age of 2 in Peru. PLoS Neglected Tropical Diseases, 16(10), 1–12. 10.1371/JOURNAL.PNTD.0010869 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Gorris, M. E. , Bartlow, A. W. , Temple, S. D. , Romero‐Alvarez, D. , Shutt, D. P. , Fair, J. M. , et al. (2021). Updated distribution maps of predominant Culex mosquitoes across the Americas. Parasites & Vectors, 14(1), 547. 10.1186/s13071-021-05051-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Haque, M. A. , Platts‐Mills, J. A. , Mduma, E. , Bodhidatta, L. , Bessong, P. , Shakoor, S. , et al. (2019). Determinants of Campylobacter infection and association with growth and enteric inflammation in children under 2 years of age in low‐resource settings. Scientific Reports, 9(1), 17124. 10.1038/S41598-019-53533-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hartigan, J. A. , & Wong, M. A. (1979). Algorithm AS 136: A K‐Means clustering algorithm. Journal of the Royal Statistical Society. Series c (Applied Statistics), 28(1), 100–108. 10.2307/2346830 [DOI] [Google Scholar]
  25. Havelaar, A. (2025). Longitudinal data on campylobacter in infants from the CAGED study [Dataset]. Harvard Dataverse. 10.7910/DVN/1JC6RS [DOI]
  26. Havelaar, A. H. , Brhane, M. , Ahmed, I. A. , Kedir, J. , Chen, D. , Deblais, L. , et al. (2022). Unravelling the reservoirs for colonisation of infants with Campylobacter spp. in rural Ethiopia: Protocol for a longitudinal study during a global pandemic and political tensions. BMJ Open, 12(10), 61311. 10.1136/bmjopen-2022-061311 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Havelaar, A. H. , Kirk, M. D. , Torgerson, P. R. , Gibb, H. J. , Hald, T. , Lake, R. J. , et al. (2015). World health organization global estimates and regional comparisons of the burden of foodborne disease in 2010. PLoS Medicine, 12(12), e1001923. 10.1371/journal.pmed.1001923 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Joyner, T. A. , Lukhnova, L. , Pazilov, Y. , Temiralyeva, G. , Hugh‐Jones, M. E. , Aikimbayev, A. , & Blackburn, J. K. (2010). Modeling the potential distribution of Bacillus anthracis under multiple climate change scenarios for Kazakhstan. PLoS One, 5(3), e9596. 10.1371/journal.pone.0009596 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kaakoush, N. O. , Castaño‐Rodríguez, N. , Mitchell, H. M. , & Man, S. M. (2015). Global epidemiology of campylobacter infection. Clinical Microbiology Reviews, 28(3), 687–720. 10.1128/CMR.00006-15 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Keusch, G. T. , Denno, D. M. , Black, R. E. , Duggan, C. , Guerrant, R. L. , Lavery, J. V. , et al. (2014). Environmental enteric dysfunction: Pathogenesis, diagnosis, and clinical consequences. Clinical Infectious Diseases, 59(suppl_4), S207–S212. 10.1093/cid/ciu485 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kuhn, K. G. , Nygård, K. M. , Guzman‐Herrador, B. , Sunde, L. S. , Rimhanen‐Finne, R. , Trönnberg, L. , et al. (2020). Campylobacter infections expected to increase due to climate change in Northern Europe. Scientific Reports, 10(1), 1–9. 10.1038/s41598-020-70593-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kumar, A. , Kumar, A. , Adhikari, D. , Gudasalamani, R. , Saikia, P. , & Khan, M. L. (2020). Ecological niche modeling for assessing potential distribution of Pterocarpus marsupium Roxb. In Ranchi, eastern India. Ecological Research. 10.1111/1440-1703.12176 [DOI] [Google Scholar]
  33. Lee, G. , Pan, W. , Peñataro Yori, P. , Paredes Olortegui, M. , Tilley, D. , Gregory, M. , et al. (2013). Symptomatic and asymptomatic campylobacter infections associated with reduced growth in Peruvian children. PLoS Neglected Tropical Diseases, 7(1), e2036. 10.1371/journal.pntd.0002036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lengerh, A. , Moges, F. , Unakal, C. , & Anagaw, B. (2013). Prevalence, associated risk factors and antimicrobial susceptibility pattern of Campylobacter species among under five diarrheic children at Gondar University Hospital, Northwest Ethiopia. BMC Pediatrics, 13(1), 82. 10.1186/1471-2431-13-82 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Lentz, J. A. , Blackburn, J. K. , & Curtis, A. J. (2011). Evaluating patterns of a white‐band disease (WBD) outbreak in acropora palmata using spatial analysis: A comparison of transect and colony clustering. PLoS One, 6(7), e21830. 10.1371/journal.pone.0021830 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Long, J. , Robertson, C. , & Nelson, T. (2018). Stampr: Spatial‐temporal analysis of moving polygons in R. Journal of Statistical Software, 84(April). 10.18637/jss.v084.c01 [DOI] [Google Scholar]
  37. Louis, V. R. , Gillespie, I. A. , O’brien, S. J. , Russek‐Cohen, E. , Pearson, A. D. , & Colwell, R. R. (2005). Temperature‐driven campylobacter seasonality in England and Wales. Applied and Environmental Microbiology, 71(1), 85–92. 10.1128/AEM.71.1.85-92.2005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Luby, S. P. , Rahman, M. , Arnold, B. F. , Unicomb, L. , Ashraf, S. , Winch, P. J. , et al. (2018). Effects of water quality, sanitation, handwashing, and nutritional interventions on diarrhoea and child growth in rural Bangladesh: A cluster randomised controlled trial. Lancet Global Health, 6(3), e302–e315. 10.1016/S2214-109X(17)30490-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Man, S. M. (2011). The clinical importance of emerging Campylobacter species. Nature Reviews Gastroenterology & Hepatology, 8(12), 669–685. 10.1038/nrgastro.2011.191 [DOI] [PubMed] [Google Scholar]
  40. McQuade, E. T. R. , Platts‐Mills, J. A. , Gratz, J. , Zhang, J. , Moulton, L. H. , Mutasa, K. , et al. (2020). Impact of water quality, sanitation, handwashing, and nutritional interventions on enteric infections in rural Zimbabwe: The Sanitation Hygiene Infant Nutrition Efficacy (SHINE) trial. The Journal of Infectious Diseases, 221(8), 1379–1386. 10.1093/INFDIS/JIZ179 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Merow, C. , Smith, M. J. , & Silander, J. A. (2013). A practical guide to MaxEnt for modeling species’ distributions: What it does, and why inputs and settings matter. Ecography, 36(10), 1058–1069. 10.1111/J.1600-0587.2013.07872.X [DOI] [Google Scholar]
  42. Mullins, J. , Lukhnova, L. , Aikimbayev, A. , Pazilov, Y. , Ert, M. V. , & Blackburn, J. K. (2011). Ecological Niche modelling of the Bacillus anthracis A1.a sub‐lineage in Kazakhstan. BMC Ecology, 11(1), 32. 10.1186/1472-6785-11-32 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Null, C. , Stewart, C. P. , Pickering, A. J. , Dentz, H. N. , Arnold, B. F. , Arnold, C. D. , et al. (2018). Effects of water quality, sanitation, handwashing, and nutritional interventions on diarrhoea and child growth in rural Kenya: A cluster‐randomised controlled trial. Lancet Global Health, 6(3), e316–e329. 10.1016/S2214-109X(18)30005-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Ojeda, A. , Deblais, L. , Mummed, B. , Brhane, M. , Hassen, K. A. , Ahmedo, B. U. , et al. (2025). Determinants of Campylobacter species diversity in infants and association with family members, livestock, and household environments in rural Eastern Ethiopia. Gut Pathogens, 17(1), 1–15. 10.1186/S13099-025-00725-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Parker, C. T. , Schiaffino, F. , Huynh, S. , Paredes Olortegui, M. , Peñataro Yori, P. , Garcia Bardales, P. F. , et al. (2022). Shotgun metagenomics of fecal samples from children in Peru reveals frequent complex co‐infections with multiple Campylobacter species. PLoS Neglected Tropical Diseases, 16(10), 1–17. 10.1371/journal.pntd.0010815 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Philips, S. J. , Dudik, M. , & Schapire, R. E. (n.d.). Maxent software for modeling species niches and distributions (Version 3.4.4). Retrieved from http://biodiversityinformatics.amnh.org/open_source/maxent/
  47. Phillips, S. J. (2021). A brief tutorial on Maxent. Retrieved from https://biodiversityinformatics.amnh.org/open_source/maxent/Maxent_tutorial_2021.pdf
  48. Phillips, S. J. , Anderson, R. P. , & Schapire, R. E. (2006). Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190(3–4), 231–259. 10.1016/j.ecolmodel.2005.03.026 [DOI] [Google Scholar]
  49. Platts‐Mills, J. A. , & Kosek, M. (2014). Update on the burden of Campylobacter in developing countries. Current Opinion in Infectious Diseases, 27(5), 444–450. 10.1097/QCO.0000000000000091 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Platts‐Mills, J. A. , Liu, J. , Gratz, J. , Mduma, E. , Amour, C. , Swai, N. , et al. (2014). Detection of Campylobacter in stool and determination of significance by culture, enzyme immunoassay, and PCR in developing countries. Journal of Clinical Microbiology, 52(4), 1074–1080. 10.1128/JCM.02935-13 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Reiner, R. C. , Wiens, K. E. , Deshpande, A. , Baumann, M. M. , Lindstedt, P. A. , Blacker, B. F. , et al. (2020). Mapping geographical inequalities in childhood diarrhoeal morbidity and mortality in low‐income and middle‐income countries, 2000‐17: Analysis for the Global Burden of Disease Study 2017. The Lancet, 395(10239), 1779–1801. 10.1016/S0140-6736(20)30114-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Robertson, C. , Nelson, T. A. , Boots, B. , & Wulder, M. A. (2007). STAMP: Spatial‐temporal analysis of moving polygons. Journal of Geographical Systems, 9(3), 207–227. 10.1007/s10109-007-0044-2 [DOI] [Google Scholar]
  53. Rogawski, E. T. , Liu, J. , Platts‐Mills, J. A. , Kabir, F. , Lertsethtakarn, P. , Siguas, M. , et al. (2018). Use of quantitative molecular diagnostic methods to investigate the effect of enteropathogen infections on linear growth in children in low‐resource settings: Longitudinal analysis of results from the MAL‐ED cohort study. Lancet Global Health, 6(12), e1319–e1328. 10.1016/S2214-109X(18)30351-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Rushton, G. , & Lolonis, P. (1996). Exploratory spatial analysis of birth defect rates in an urban population. In Statistics in medicine. 10.1002/(SICI)1097-0258(19960415)15:7/9<717::AID-SIM243>3.0.CO;2-0 [DOI] [PubMed] [Google Scholar]
  55. Sanderson, R. A. , Maas, J. A. , Blain, A. P. , Gorton, R. , Ward, J. , O’Brien, S. J. , et al. (2018). Spatio‐temporal models to determine association between Campylobacter cases and environment. International Journal of Epidemiology, 47(1), 202–216. 10.1093/ije/dyx217 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Schiaffino, F. , Trigoso, D. R. , Colston, J. M. , Olortegui, M. P. , Shapiama Lopez, W. V. , Garcia Bardales, P. F. , et al. (2021). Associations among household animal ownership, infrastructure, and hygiene characteristics with source attribution of household fecal contamination in Peri‐urban communities of Iquitos, Peru. The American Journal of Tropical Medicine and Hygiene, 104(1), 372–381. 10.4269/AJTMH.20-0810 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Singh, N. , Thystrup, C. A. N. , Hassen, B. M. , Bhandari, M. , Rajashekara, G. , Hald, T. M. , et al. (2025). Transmission pathways of Campylobacter Jejuni between humans and livestock in rural Ethiopia are highly complex and interdependent. Gut Pathogens, 17(1), 1–16. 10.1186/s13099-025-00691-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Soneja, S. , Jiang, C. , Romeo Upperman, C. , Murtugudde, R. , Mitchell, C. S. , Blythe, D. , et al. (2016). Extreme precipitation events and increased risk of campylobacteriosis in Maryland, U.S.A. Environmental Research, 149, 216–221. 10.1016/J.ENVRES.2016.05.021 [DOI] [PubMed] [Google Scholar]
  59. Stewart, C. P. , Kariger, P. , Fernald, L. , Pickering, A. J. , Arnold, C. D. , Arnold, B. F. , et al. (2018). Effects of water quality, sanitation, handwashing, and nutritional interventions on child development in rural Kenya (WASH benefits Kenya): A cluster‐randomised controlled trial. The Lancet Child & Adolescent Health, 2(4), 269–280. 10.1016/S2352-4642(18)30025-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Terefe, Y. , Deblais, L. , Ghanem, M. , Helmy, Y. A. , Mummed, B. , Chen, D. , et al. (2020). Co‐occurrence of campylobacter species in children from Eastern Ethiopia, and their Association with environmental enteric dysfunction, diarrhea, and host microbiome. Frontiers in Public Health, 8, 99. 10.3389/fpubh.2020.00099 [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Weisent, J. , Seaver, W. , Odoi, A. , & Rohrbach, B. (2014). The importance of climatic factors and outliers in predicting regional monthly campylobacteriosis risk in Georgia, USA. International Journal of Biometeorology, 58(9), 1865–1878. 10.1007/S00484-014-0788-6 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Citations

  1. Havelaar, A. (2025). Longitudinal data on campylobacter in infants from the CAGED study [Dataset]. Harvard Dataverse. 10.7910/DVN/1JC6RS [DOI]

Supplementary Materials

Supporting Information S1

GH2-10-e2024GH001146-s003.pdf (472.2KB, pdf)

Movie S1

Download video file (398.1KB, mp4)

Movie S2

Download video file (428.4KB, mp4)

Movie S3

Download video file (375.3KB, mp4)

Movie S4

Download video file (463.7KB, mp4)

Data Availability Statement

The data on which this article is based are available in Havelaar (2025).

Articles from GeoHealth are provided here courtesy of Wiley

RESOURCES