Mapping the fly: mechanistic and statistical models of tsetse spatial distribution and dispersal — a scoping review

Abstract

Tsetse flies, the biological vectors of trypanosomes, represent a key constraint for food security in 38 sub-Saharan African countries. They cause sleeping sickness in humans and nagana in domestic animals. The distribution of tsetse is largely determined by climate, habitat, altitude and the presence of suitable host animals. Knowledge of the tsetse fly’s environment and movement pattern has implications for control of the fly. This work aims to describe dispersal and the spatial distribution of tsetse flies and the climatic and environmental factors that affect it. This is done through a scoping review to synthesise the types of mathematical and statistical models used in tsetse literature. Online databases were used as the main source to find articles published between 1950 and 2025. Articles that specifically looked at spatial dynamics of tsetse populations were included in the final synthesis. Ninety-one articles were included in the final synthesis. Seventy-one articles were statistical and twenty were mechanistic; most of the work focussed on Glossina pallidipes, with air temperature, rainfall and NDVI as the main climatic/environmental factors used. In an effort to ensure long-term sustainability of future programmes in a more difficult funding climate, there is currently an on-going reappraisal of tactics. Robust data-driven approaches are required to support decision making and for an optimised approach to control strategies.

1 Introduction

Vector-borne diseases, such as malaria, dengue and schistosomiasis, cause more than 700,000 deaths, and disproportionately affect populations that are poor [WHO, 2024 (1)]. Tsetse flies are the biological vectors of the protozoan parasites, trypanosomes, and are only present in Sub-Saharan Africa. They cause sleeping sickness in humans and nagana in domestic stock [Krafsur, 2009 (2)}, and represent a key constraint for food security in the 38 sub-Saharan African countries where they are present, causing billions of dollars of losses in the agriculture sector per year [Cecchi et al, 2024 (3)]. They are a major cause of rural poverty across 8.7 million km² where they occur, affecting 100 million cattle and, as such, tsetse are known as the poverty fly [Holmes, 2013 (4)]. Thirty-three species and subspecies have been identified, divided into three subgenera based on morphological differences. These subgenera are fusca, palpalis and morsitans, which are also known as forest, riverine and savannah flies based on their habitat preferences. The distribution of tsetse is largely determined by climate, vegetation, altitude, and the presence of suitable host animals. The habitat plays a major role in vector biology, affecting aspects such as tsetse distribution, movement, and host location. Kenneth Morris, an entomologist, wrote in 1939 that, “A full knowledge of the fly’s habits and of the complexities of his environmental requirements may make it possible to determine how the alteration of some small factor will lead to the disappearance of the pest” (5).

Tsetse flies are known to occur only in Africa, although there have been accounts of tsetse found in Saudi Arabia and fossil remains in Colorado [Carter, 1906 (6); Elsen et al., 1990 (7); Lambrecht, 1964 (8)]. Their distribution extends to over 10 million km² [Buxton, 1955 (9); Glasgow, 1963 (10)]. Although they occur between broad geographical limits, they are known not to occur everywhere within those limits, due to climatic or environmental constraints. Tsetse flies are absent in areas at altitudes of more than 2000 meters and the northern limits, such as in Nigeria, are set by high temperatures in hot seasons, whereas the southern limits, south of the iMfolozi River in north-east KwaZulu-Natal, South Africa may be due to complex and sometimes undefined factors [de Beer et al., 2016 (11); Buxton, 1955 (9)]. Although the tsetse fly and its puparium are well waterproofed [Bursell, 1958 (12); 1959 (13)] regions situated below the 500mm isohyet do not support tsetse flies unless there are watercourses with some vegetation along them [Glasgow, 1963 (10)]. Temperature is of great importance in the ecology of the tsetse fly. Not only does it affect interlarval and puparial durations, but it also influences the fly’s activity. Below 16 °C tsetse flies are inactive and above 32 °C they seek out refuge sites, such as rot-holes in trees, animal burrows and deep fissures in bark, in which they are inactive [Vale, 1971 (14)]. Thus, the temperatures must stay within the 16 to 32 °C range for long enough during the day for tsetse flies to remain active and be able to seek food. The puparial duration at mean temperatures <16 °C is so long that the insects’ fat reserves become depleted prior to maturation; this is one of the ways in which tsetse fly distribution is limited [Bursell, 1960 (15)]. Conversely, mean temperatures >32 °C are sufficient to ensure that all pupae die during development. Moreover, temperatures over 40 °C are rapidly lethal to both large and small species of tsetse fly and their puparia [Phelps & Burrows, 1969 (16); Hargrove and Vale, 2020 (17)]. Increasing altitude moderates temperature, and this is reflected in the distribution of tsetse flies. Female Glossina pallidipes were trapped up to an altitude of 1992 m in the southern Rift Valley of Ethiopia [Vreysen et al., 1999 (18)], but in Zimbabwe, which is farther from the Equator, they do not occur at altitudes above 1200 m [Nash, 1933 (19)]. These limits are linked to the mean annual temperature and can of course vary with climatic changes. The vegetation type at a given locality is greatly influenced by temperature and humidity. Apart from grasslands, which do not support tsetse flies, all forms of woodland, from savannah to rainforest, can provide a suitable habitat for some species of these flies, but no one vegetation type is suitable for all species [Glasgow, 1963 (10)].

A species can generally only occupy a particular site if the following key conditions are satisfied: it must have successfully dispersed there over its evolutionary and biogeographic history; the local environment must offer conditions capable of supporting a viable population; and the biotic context must favour the species’ persistence, meaning adequate resources are available and competitive pressures aren’t too severe (20). Ecological niche modelling, also known as species distribution modelling or habitat suitability modelling, has traditionally centred on environmental variables and is correlative, although more recent approaches are incorporating dispersal dynamics and biotic interactions as well, by using statistical estimation frameworks to parametrise mechanistic models and check model predictions against empirical data. To estimate dispersal rates, mark-release-recapture studies are used; these are direct methods and are limited to short distances and short timescales and difficult to use where populations are discontinuous. Genetic methods, such as phylogeographic and population genetics analyses, offer a powerful complement by providing indirect but robust estimates of gene flow, dispersal rates, effective population sizes, and landscape features that facilitate or impede fly movement. These methods are highly useful in determining whether eliminated areas are at risk of re-invasion and whether isolated pockets of tsetse are better targets for sustained elimination [Bouyer 2015 (21); Brito 2025 (22); Krafsur 2003 (23)]. Mathematical models on vector ecology have developed along two largely parallel tracks: mechanistic approaches, that describe how vectors move and persist through space; and statistical approaches that use field data to estimate ecological parameters (24). Mechanistic models are often expressed either in the form of differential equations resulting in dynamical systems or algebraic equations, reflecting a steady-state version. Statistical approaches, on the other hand, use data to construct models through statistical inference. Both approaches offer different advantages. Ecology aims to explain and understand, describe and predict. This work aims to describe dispersal and the distribution of tsetse flies. More so, the distribution of tsetse flies needs to be characterised in the context of climate change, because as with other vectors and vector-borne diseases, changing climate has and will change the current geographical distribution in ways that are complex [Paz, 2024 (25)]. In addition, efforts to control sleeping sickness have made major strides over the last two decades, resulting in a 97% reduction in new cases [WHO, 2023 (26)]. Efforts have not been as successful for controlling nagana, leading to the “Progressive Control Pathway (PCP) for Animal African Trypanosomosis” initiative by the FAO [Diall, 2017 (27)]. No vaccines exist for nagana, drug resistance is increasing, wildlife reservoirs make elimination very challenging. Although there have been successful tsetse population removal such as in Principe Island, Unguja Island, the Umfurudzi Game area, as well as the whole of the south-east lowveld in Zimbabwe, Botswana, large tracts of northern Nigeria; in the case of G. pallidipes, from the whole of KwaZulu-Natal, it still remains challenging (28). Trypanosomosis is also a neglected disease as it affects poorer people in rural areas. Stage 1 of the PCP includes developing novel tools to optimise prioritisation of intervention areas. A scoping review was carried out, therefore, to synthesise the types of mechanistic and statistical models used in the tsetse literature to study tsetse movement and dispersal, and spatial distribution in relation to climate, and ultimately how that affects control.

2 Methods

Jackson [1941 (29); 1943 (30); 1947 (31)] undertook significant field work and looked at dispersal using mark-recapture data to estimate dispersal and mortality rates for savanna tsetse species. In his works, dispersal rates were high in the first week after tsetse were released and then decreased subsequently. He hypothesised that flies moved more within ‘ambits’ of favoured vegetation types and largely restricted their movements within these ambits, moving less frequently out of these ambits or from one ambit to another. The ambits consisted of open ground in which tsetse fed, and ambits of homogenous woodland in which they resided. This meant that, although tsetse might travel long distances daily between their ‘home’ and where they feed, the overall dispersal rates are quite low. Bursell [1970 (32)] showed, however, that Jackson’s data were consistent with tsetse actually moving in a series of steps in random directions. We consider Jackson’s work as the point of departure and examine the models that came thereafter.

2.1 Scoping review

The scoping review is informed by the framework proposed by Arksey and O’Malley [2007 (33)] which has been further developed by Levac et al. [2010 (34)] and the Joanna Briggs Institute [Peters et al., 2015 (35)]. The inclusion criteria for this review are mathematical models i.e. statistical and mechanistic models (including simulation studies), which specifically looked at spatial dynamics of tsetse populations. Articles that considered only HAT or trypanosome infections were excluded. Evidence sources considered were peer-reviewed papers. Articles in English were eligible for inclusion. This scoping review does not have a protocol that has been published beforehand and is not registered.

2.2 Search strategy

A search of a database of tsetse literature on PubMed and Web of Science was used as the main source of this review. The key search terms are given in Table 1. Articles published between 1 January 1950 and 1 October 2025 were retained. Three reviewers, Abdelatif (NA), Govinder (KG) and Hargrove (JWH) evaluated the eligibility of the studies obtained from the literature searches. All articles yielded by the database searches were initially screened by their titles and abstracts, and then the full-text articles to obtain the final list of articles to be included in the synthesis. In cases of uncertainties, agreement was reached by discussion and consensus. Rayyan (36) was used to manage the results of the search, and Elicit for data-extraction. As Elicit is an AI research assistant tool, all data extracted from the articles were verified by NA.

2.3 Data charting process and data items

The following items were extracted from the articles: modelling approach used, description of method, software used, species on which the study was done, country of study and the climate and/or vegetation variables that were explicitly used in the model. The studies were grouped by whether they were mechanistic or statistical methods, as well as what specific method was used. These were presented in a table, whereas the country of study, Glossina species studied and biotic and abiotic factors used were presented graphically. All articles included in the final synthesis were summarised with each of the data items extracted and included in the Supplementary Materials (Table 1).

3 Results

3.1 Search results

A total of 746 unique articles were identified after excluding 380 duplicates. Out of the remaining articles, 641 were excluded because their abstracts, titles or full-text review did not meet the eligibility requirements (Figure 1). The total number of articles included in the final synthesis was 91 articles. Studies were excluded if they were related to modelling of HAT and/or trypanosome modelling rather than tsetse – or did not look at spatial modelling related to the distribution or dispersal of tsetse.

Figure 1

3.2 Study characteristics

The majority of the studies used statistical methods to model tsetse distribution or dispersal, with almost a third using population and landscape genetic statistics to estimate dispersal rates of tsetse (Table 2). Most of the studies took place in Kenya, followed by Uganda, Zimbabwe and Tanzania (Figure 2). Glossina pallidipes and Glossina morsitans were the most studied species (Figure 3). Articles that explicitly used biotic and abiotic factors in their models had used temperature, normalised vegetation index (NDVI) and topography variables, such as elevation and slope, to determine the effects of climate on distribution or dispersal (Figure 4).

Figure 2

Figure 3

Figure 4

3.3 Summary of models used

3.3.1 Mechanistic models

3.3.1.1 Dispersal and simulation

Rogers [1977 (38)] used a random walk model in analysing the movement of a population of G. fuscipes fuscipes. He carried out a mark-release-recapture experiment to estimate the feeding cycle, losses from the marking area, distribution of the population within the habitat and the daily movement of flies. To investigate the outcome of two-dimensional random movement, he used two methods, the first of which was a formula to predict the mean distance moved away from a starting point, assuming a constant step length. The second method was based on modelled predictions of a series of random movements away from a release point. These predictions involved probability distributions for directions moved and distances covered per step to estimate tsetse fly mortality rates. Hargrove [1981 (39)] considered a more complex model of random movement in tsetse that allowed for age-varying mortality rates, capture probability and step length. He ran computer simulations emulating Jackson’s [1941 (29); 1946 (31); 1948 (40)] field experiments and improved on Jackson’s and Rogers’ [1977 (38)] fits to the data. Later, Hargrove & Lange [1989]¹ modelled tsetse dispersal using the equivalent formulation of a diffusion process in which the rate of dispersal was given by a diffusion coefficient rather than a discrete step length. The advantage of using this was that regression techniques could be employed rather than computer simulations. The model was run using the same field experiments [Jackson, 1941 (29); 1946 (31)] and found to provide a reasonable fit to one of the experiments but not the other. This might have been a result of an overly simplified diffusion model. Williams, Dransfield and Brightwell [1992 (41)] further developed this work and used an analytic and simulation model of tsetse movement to determine efficiency of traps and how trapping will affect populations. They applied random walk processes by assuming that fly movement takes place in a series of steps and the direction of each step is randomly and uniformly distributed around a circle. The diffusion equation, which requires a single parameter, the diffusion coefficient, was considered as an alternative means of analysing the same model. The diffusion coefficient was assumed not to vary in time or space, so that the number of flies that enter a cell is proportional to the numbers in adjacent cells and the rate at which they leave the cell is proportional to the number in that cell. The model for diffusive movement was used in conjunction with a population dynamics model involving logistic growth. The model thus describes both growth in the population as well as dispersal. This model was applied by Hargrove [2003 (42)] to field data from the Rifa triangle in Zimbabwe [Vale, 1988 (43)] in a campaign to control G. pallidipes and G. morstians morsitans. In their study, Williams [1995 (44)] considered three scenarios; one with no diffusion, one with infinite diffusion where flies are infinitely mobile, and the third where diffusion rates are limited and traps kill all flies that enter the range of attraction. Fly movement can differ and be influenced by other factors such as the environment and climate and host suitability. Brightwell et al. [1992 (45)] studied the seasonal changes in the distribution of G. pallidipes and G. longipennis, by looking at the mean spread and root mean square spread, which are standardised measures. They determined the effects of environmental and climatic variables on the mean spread of the flies, and a random diffusion model where there was an increase in spread, to determine whether the spread was due to diffusion alone or whether it could be accounted for. The mean spread was determined to be a random diffusion process and not as a result of directed movement. The analysis provided a detailed account of the factors that influence seasonal change in tsetse numbers. Vale [1984 (46)] conducted a mark-release-recapture study in Zimbabwe and modelled the dispersal rates of G. morsitans and G. pallidipes for males and females as a diffusion process. Females seemed to disperse as much as males and daily displacements were found to be large. The authors noted that the work still required further development but had implications that the method was a more feasible method to determine dispersal of tsetse in small study areas. Estimates of varying dispersal rates on re-invasion of tsetse was modelled in Hargrove [2000 (28)] using a diffusion process. It was concluded that rates of re-invasion were minimised if control methods were applied at the edge of tsetse belts, and more severe when areas with cleared areas are surrounded by areas of tsetse infested land. This is in the case where control methods are not continuously maintained in the control area. Childs [2010 (47)] followed on from (41) and assumed tsetse disperse by Brownian motion down a diffusion gradient. Diffusion estimates based on other species were used to model the dispersal of G. brevipalpis and G. austeni in South Africa. Vreysen [2013 (48)] used mark-release-recapture methods to determine the mean dispersal distance of G. palpalis gambiensis by simulating a two-dimensional random walk process. They found a difference between the estimates using MRR compared to when using population genetics, due to the way the flies were released – artificially versus natural dispersal. Hargrove et al. [2021 (49) (50)] re-examined data generated through genetic analysis of material from five tsetse species to determine whether negative density-dependent dispersal, as shown by de Meeûs et al. [2019 (51)], was in fact true. By reassessing the equations and variables used, it was found that there was high measurement error in the parameters used, namely, the effective population density which is calculated by effective population size divided by the surface area occupied by the effective population. It was established that negative density-dependent dispersal does not occur in tsetse populations.

3.3.1.2 Differential equations and compartmental models

Barclay & Vreysen [2011 (52)] developed a dynamic model that considered both the life cycle of tsetse and their dispersal. The model also considered how to integrate vector methods such as sequential aerial spraying technique, traps and targets, sterile insect technique (SIT) and insecticide-treated livestock in control operations. Sensitivity analyses were carried out on the mode of dispersal, where the proportion of flies in a given cell disperses to neighbouring cells as determined by the attractiveness of that cell. Attractiveness was based on the relative carrying capacity of the cell. The second mechanism was similar to that in Williams et al. [1992 (41)] in which a constant fraction of flies in a given cell disperses to neighbouring cells each time period and the fraction remains the same for all cells. The model assumed that the net immigration was zero and that control was imposed outside of the study area as well. They determined that control by traps depends on the movement pattern of the flies, whereas effectiveness of the SIT control methods rely on whether movement is random or directed, since it is a highly density-dependent method. Parameter selections and model structures were however problematic, due to the long timestep used, small rates of dispersal and small size of the mapped cells [Hargrove, Torr & Vale, 2011 (53)]. This model was later revisited in Barclay & Vreysen [2013 (54)] and considered three aspects of dispersal; the maximum daily dispersal distance, the percentage of flies that dispersed from each cell every day and the mode of dispersal (towards more or less densely occupied cells). The two mechanisms considered previously still applied, and various analyses were done in relation to traps and targets, insecticide-treated livestock and SIT. Cecilia et al. [2021 (55)] examined spatial dispersal of G. palpalis gambiensis in heterogeneous environments using a mechanistic spatio-temporal model. A deterministic compartmental model of age-structured population dynamics was used accounting for environmental heterogeneity and density-dependence. Release-recapture data were used to assess dispersal. They determined that environmental heterogeneity affects tsetse fly population dynamics which affects control efforts by creating refuges at the interface of controlled and uncontrolled areas. They concluded that control strategies need to consider heterogeneous environments. The model evaluated complex strategies, accounting for all stages of the tsetse life cycle, spatio-temporal environmental heterogeneity and density-dependence. Longbottom et al. [2020 (56)] expanded on the model used by Lord et al. [2018 (57)], which showed that tsetse populations are highly responsive to abiotic factors, to predict where tsetse populations are most viably able to persist and which environments are most suited to the needs of the fly. Tsetse population dynamics were modelled using ordinary differential equations, incorporating temperature-dependent parameters, and then implemented separately in grid cells. This resulted in predictions of spatial monthly abundances of tsetse across Northern Zimbabwe. The analysis only considers temperature, and many other factors could have contributed to the tsetse numbers. It was also assumed that populations were closed, and therefore no movement between grid cells was allowed. Gouteux et al. [2001 (58)] considered a model that looks at density-dependent reinvasion of tsetse in an area that was subjected to removal trapping experiments. The model postulates that the population changes through a combined trapping and density-dependent reinvasion effect. They use the ‘Mitscherlist’ equation and describe two situations: if the reinvasion rate is large compared to the trapping capture rate then the captured flies are replaced by immigrant flies that come in from surrounding areas that have much larger populations; or if the reinvasion rate is small compared to the capture rate, then the population decreases substantially, so that depletion is greater from trapping than through reinvasion. The model considers climatic effects only through the estimated parameters and not implicitly, because the data used are divided into the dry season and rainy season and show that the reinvasion rate is larger during the rainy period. The model also showed that populations outside of the localised regions react by invading those regions with with low population; however, it is limited by describing when these reinvasions occur. The model is also limited when capture rates are small and high reinvasion rates apply because it will not show a noticeable decrease in population.

Peck [2012 (59)] used a spatial metapopulation model to describe the ecological relationships in tsetse sub structured populations. He used the extinction probabilities derived by Hargrove [2005 (60)] in different patches and added a colonisation rate and a network connectivity measure. Logistic regression was used to determine the variables that influenced tsetse extinction throughout the region. By using metapopulations, spatial-ecological contexts are provided and may change extinction probabilities. Yu et al. [1995 (61), 1996 (62)] developed a stochastic model to study tsetse spatial dispersal, allowing distance travelled and direction to vary randomly. Direction of the fly was assumed to be random, and distance travelled occurs in a straight line. Although it was one of the first stochastic models applied to tsetse modelling, spatial autocorrelation was not considered. Additionally, no data were used in the study and the goodness of the model fit is unknown.

3.3.1.3 Agent-based models

Lin [2015 (63)] used a dynamic agent-based model that was spatially explicit incorporating habitat data that can be utilised across East Africa, and test the impact of tsetse control techniques. Components of the model included population dynamics of flies, the ability to sense whether a wild animal is close by, and whether a patch is suitable. The tsetse spatial distribution was compared with a previous tsetse map which were in agreement. The strength of the modelling approach is the addition of complexity that more realistically resembles the tsetse ecology, finer spatial resolution, and produces a realised population map rather than realised niche.

3.3.1.4 Dynamic models

DeVisser et al. [2010 (64)] constructed a Tsetse Ecological Distribution (TED) model to explore spatio-temporal fluctuations in tsetse distribution in Kenya. It predicts tsetse presence based on habitat suitability using remote sense data, either modelled through daily maximum fly movement of an individual fly, or the advancement of tsetse populations. The model was validated using data from sites outside of the traditional belt areas to determine whether flies were present in previously unidentified areas. The model was shown to be quite sensitive to the climate variables used. Accordingly, the predicted distributions depended on which variables were used. These models present large uncertainty due to the resolution of the remote sensed data, as they make the, sometimes, unrealistic assumption that - at a certain scale - conditions are uniform. If uncertainty were reduced in these data, as well as other inputs, the model would accurately predict the presence or absence of tsetse by location. The TED model was used by Messina [2012 (65)] and McCord [2012 (66)]. McCord (2012 (66) used it to identify the locations of “control reservoirs” in an effort to optimise resource allocation in the identified areas. These reservoirs were defined as tsetse populations that are spatially limited by seasonal fluctuations to suitable habitat. Identification of these control reservoirs was given as a more cost-effective alternative to putting up barriers across fly limits, especially when considering tsetse reinvasion into cleared areas. Messina [2012 (65)] used the model to predict the effects of climate change on the tsetse distribution.

3.3.2 Statistical models

3.3.2.1 Species habitat suitability models

De Deken [2005 (67)] used an unsupervised classification method to identify clusters of similar landscape types and then predicted habitat suitability of G. fuscipes quanzensis. Accuracy of the predictions were assessed by sampling in areas where data was scarce and habitats determined suitable and less suitable. This analysis approach has weaknesses in that landscape identification was based on clusters of similar pixels of satellite images and did not depend on other climatic factors such as temperature, humidity or vegetation and tsetse distribution did not consider human activity and host availability. Accuracy of the predictions produced would need to be used with caution. Logistic regression was used to model Glossina spp. in Kenya using 1973 climate data and validating the model using data from 2000 [Moore 2010 (68)]. Projected distributions of tsetse were produced under elevated greenhouse gases. The model was compared to the FAO/PAATIS model and found to contain significant differences arising from the different research goals, variation in the environmental factors included and methodologies used. Dicko [2014 (69)] used machine learning methods, specifically regularized logistic regression and MaxEnt, to determine which approach improved prediction accuracy. Using the presence-absence validation data set, it was found that MaxEnt predicted suitability better and determined that tsetse presence data was more informative than absence data as it allows for data to be combined from different sources to inform control operations without the limitations of false absence data. Matawa et al. [2016 (70)] likewise examined habitat changes by using a species distribution model and hypothesised that changes in landcover decrease the amount of suitable habitat. They considered finer resolution remotely sensed data to improve the accuracy of tsetse distribution models and used tsetse occurrence data from 1994 to 2012 to build a MaxEnt model to predict the distribution of tsetse. The Z-score test was used to compare the proportion of suitable riparian habitat to non-riparian habitat, as well as whether the proportion of suitable habitat is significantly different from the proportion of unsuitable habitat in areas where there is dense human activity. To describe the temporal changes, the net losses and gains in tsetse were calculated taking into account the changes in habitat suitability between current year and previous base year. Logistic regression was used to investigate the relationship between drainage network and distribution of suitable and unsuitable tsetse habitat. The results from this study assess temporal changes in landcover and tsetse habitat and include stable and dynamic variables at finer spatial resolutions. Albert [2015 (71)] used autologistic regression to predict the spatial distribution of G. fuscipes fuscipes, incorporating spatial autocorrelation that was quantified using Global Moran’s I. This type of modelling approach extends the standard logistic regression by including an autocovariate term that allows the probability of an outcome to be influenced by the state of neighbouring data points. Matawa et al. [2020 (72)] built a tsetse distribution model using ensemble weighted mean modelling techniques with temporally stable and dynamic environmental variables. The models chosen to build the ensemble model were maximum entropy, generalised linear model, random forest and flexible discriminant analysis. To determine whether there was a downward trend in habitat suitability over time, the Mann-Kendall trend test was used. The use of ensemble models provided more reliability in the results. Tirados [2020 (73)] used boosted regression trees and logistic regression to produce dry season habitat suitability maps. LASSO regression was used to select the environmental factors that influenced tsetse presence/absence. This study was done to determine the impact of Tiny Targets on tsetse distribution. These models make the assumption that the traps that did not catch flies are not effective at catching those flies. However, absence of flies could be due to factors other than habitat, especially since the model was based on the dry season alone. Nnko et al. [2021 (74)] and Scoones [2017 (75)] similarly used MaxEnt species distribution modelling to predict the distribution of tsetse in Tanzania and Zimbabwe, respectively. Accuracy of the predictions was not provided, and so it is difficult to determine how well the model worked. Several studies have applied species distribution modelling techniques to assess tsetse habitat suitability and distribution. For example, Bishop et al. [2020 (76)] used random forests for G. pallidipes; Longbottom et al. [2024 (77)] employed boosted regression trees and spatiotemporal geostatistical models for G. fuscipes fuscipes; and Zhou et al. [2021 (78)] used MaxEnt to predict G. morsitans suitability under historical and future conditions. Similarly, Chikowore [2017 (79)], Gachoki et al. [2021 (80)], Gachoki et al. [2024 (81)], and Mongare et al. [2025 (82)] applied techniques including MaxEnt, generalized linear models, boosted regression trees, random forests, and partial least squares regression to examine the influence of environmental variables on the presence of teneral and non-teneral tsetse flies in Kenya. Although these types of predictive models are able to provide accurate predictions because of the underlying spatial structures that are taken into account, they require strong theoretical understanding of species’ niches [Kerr, Kulkarni & Algar, 2011 (83)]. De Beer et al. [2021 (84)] developed a MaxEnt species distribution model for G. brevipalpis and G. austeni to determine habitat suitability using both environmental and climatic variables as well as cattle density. Most of the studies considered climatic and environmental factors but few take into account biotic factors, specifically host distributions that tsetse rely on to survive.

3.3.2.2 Multivariate methods

Different statistical regression techniques were used to assess the distribution and abundance of tsetse and environmental factors. Gashuru et al. [2021 (85)] conducted a longitudinal stratified survey of G. pallidipes and G. morsitans centralis to determine the variations in occurrence between the two species. T-tests, ANOVA and logistic regression were used to determine whether apparent density was associated with area, time of year, distance and the season. The results showed that tsetse were less likely to occur outside of protected areas. Generalised additive models were used in Kargbo et al. [2025 (86)] to analyse the relationship between the tsetse species occurrence and abundance, G. morsitans submorsitans and G. palpalis gambiensis, and climate conditions in 2012–2021 and future periods (2021-2040, 2041–2060 and 2061-2080). Temperature was found to be the main variable affecting the distribution of the tsetse species assessed. Mugenyi et al. [2021 (87)] used the Zero-inflated Poisson model to explore the association between tsetse abundance and environmental and climatic variables, and tsetse density predictions mapped to show spatial distribution, whereas Wangoola et al. [2025 (88)] used a linear regression model. These types of models assess the association between the environmental and climatic variables and tsetse occurrence but do not explicitly incorporate spatial dependence or dispersal mechanisms.

3.3.2.3 Geostatistical and spatial analysis methods

Rogers et al. [1986 (89)] superimposed the distribution of tsetse on climograms of temperature and saturation deficit to determine the limits of distribution. This is a very simplified approach that does not link specific locations of climate to tsetse data. Kitron et al. [1996 (90)] used geospatial methods to determine factors affecting variations of tsetse density at a local level, using tsetse catch data to analyse the spatiotemporal distribution of flies, and remotely sensed and ground measured data to determine the association with fly density. To determine the spatial pattern of tsetse distribution and its association with environmental factors, a standardised value Moran’s I was used. Getis spatial filtering technique was used to separate the spatial component from the non-spatial component i.e. the association between trap locations from the association of catches that are not dependent on the spatial position of the traps. These measure spatial autocorrelation, identifying whether patterns are clustered, dispersed or random. Both simple and multiple regression were used to determine the correlation between tsetse density and remotely sensed/ground measured factors. These methods emphasise local variations in tsetse distribution. Rogers et al. [1994 (91)] analysed tsetse distribution using step-wise linear discriminant analysis and using the discriminating power, between tsetse and non-tsetse areas, determined by Mahalanobis distance. After selecting the variables in order of their importance, maps were created to predict the probability of tsetse occurrence in each grid cell. Although this takes into account the different climatic and habitat predictors that are deemed to influence specific tsetse distributions, the relative importance of the order of the predictors should not be heavily emphasised since most are strongly correlated. Rogers, Hay & Packer [1996 (92)] used discriminant analysis applied to temporal Fourier-processed remotely sensed vegetation and climate data to predict the presence or absence of tsetse. The model produced accuracies with an average of 82% (range of 67%-100%). Multiple discriminant analysis was used in Togo to predict the distribution of G. tachinoides after which maps were produced [Hendrickx 1999 (93)]. Validation was performed using a holdout sample. Hendrickx [2001 (94)] also used multiple discriminant analysis to optimise the use of satellite imagery in predicting the distribution of multiple tsetse species in Togo. Different analysis options were used to determine the best predictor variables, and validation was done as in the previous paper. The usefulness of the results depend on the level of accuracy which is determined by the resources allocated, and it was suggested that field surveys should be conducted at the expected tsetse limits and then tsetse abundance within the limits. Similar approaches were previously used by Rogers [1993 (95)] in Zimbabwe, and Robinson [1997 (96)] who compared discriminant analysis and maximum likelihood methods, and found that maximum likelihood produced better predictions. However, these methods lack a spatial component in that they treat each pixel of the satellite imagery as independent – spatial autocorrelation needs to be accounted in the model when mapping distribution of tsetse using environmental data (96). Robinson [1997b (97)] produced probabilities of presence and absence of three tsetse species in a univariate analysis. Landscape fragmentation analysis was applied to map tsetse densities in Burkina Faso using remote sensing data and GIS [Guerrini et al., 2008 (98)]. Fragmentation clusters were identified using a Partitioning around Medoids approach, which is a centroid-based clustering algorithm which divides a dataset in a given number of clusters by selecting medoids as centres. Apparent tsetse densities were compared in different clusters using the Wilcoxon rank sum test. This method allows for the ability to discriminate the community type from remote sensing data and uses publicly available data. Hendrickx [1999 (99)] developed digital tsetse distribution and abundance maps by incorporating spatial tsetse abundance data and remotely sensed data. Seasonal adjustments were made because surveys were carried out at different time periods. The distribution maps, in relation to the climatic and agro-ecological settings, reflect why certain distributions of tsetse exist as they do, and also allow one to create such maps based on limited field data. Biases in the tsetse density measurements and other tsetse dependent biases such as sex and age, " may, however, result in abundance maps that are not very accurate. Sedda et al. [2010 (100)] presented two space-time geostatistical kriging methods: the first of which is the standard space-time ordinary kriging that removes seasonality from tsetse catches, and the second which allows mean tsetse abundance to vary through space. This is done by estimating the trend from the sample data and then removing this trend from the apparent densities to produce the residuals that are used to model the space-time covariance function, which is then applied to predict residual abundance through space and time. This study quantified combined spatial-temporal variations in tsetse distributions, something that was not previously done. It has provided both spatial and temporal ranges to determine the optimal distances between sampling and control points and seasonal and spatial comparisons can be made within and between species. Sciarretta [2005 (101)] used linear kriging to determine the spatial distribution of combined catches of G. morsitans submorsitans, G. fuscipes and G. pallidipes in Ethiopia. Hotspots were found to be stable throughout the year and it was concluded that they could be important targets for interventions. Sciarretta [2010 (102)] used cluster analysis to similarly detect hotspots of G. morsitans submorsitans and G. pallidipes and found the two species to show an aggregated pattern along rivers and savannah habitat with heterogeneous vegetation. The spatial and temporal variability of G. palpalis palpalis using kriging was studied in Equatorial Guinea by Cano [2007 (103)] and Cano [2007b (104)] and validated using cross validation. Geostatistical models should be used appropriately and the spatial structure of data examined before-hand.

3.3.2.4 Population and landscape genetics

Adam [2014 (105)] used nine microsatellite loci to determine whether G. tachinoides could disperse between river basins in Ghana. Wright’s F-statistics were studied and estimated using Weir and Cockerham’s unbiased estimators. Hierarchical analysis was performed to determine the smallest relevant population units. Isolation by distance using Rousset’s method was performed. Habitat fragmentation was revealed as a possible reason for low dispersal and high aggregation patterns – reduced populations through control efforts will however increase the chances of re-invasion, therefore requiring more strategic programmes. Bayesian methods were used by Bateta [2020 (106)], to test for individual migrants from neighbouring sampling sites. Three genetic clusters were found with infrequent migration between them, indicating suppression rather than eradication may be the most achievable target, and efforts should be designed for each separately. Infrequent long-range migration was found to occur between distinct populations, which requires active monitoring if local eradication was successful. Wright’s F-statistics estimated through Weir and Cockerham’s unbiased estimators and isolation by distance with Rousset’s approach was used to assess the effect of vector control on population genetic structure of G. palpalis palpalis in Ivory Coast, Cameroon and Democratic Republic of Congo [Melachio 2011 (107); Berte 2019 (108)], to assess the impact of landscape fragmentation on G. tachinoides in Burkina Faso [Kone 2010 (109)], to determine whether disappearance of gene flow was due to population growth and climate effects [Ravel 2013 (110)], to assess the impacts of vector control [Ravel 2024 (111)], and dispersal among adjacent river basins [Kone 2011 (112)]. Bouyer (2009 (113)) used genetic isolation-by-distance and mark-release-recapture approaches, the study found consistent estimates of tsetse dispersal and demonstrated that landscape fragmentation strongly restricts movement. Ravel [2023 (114)] determined dispersal rates using tsetse population genetics and measuring the distance between adults and their parents using Rousset’s equation. They found an interconnected metapopulation that underwent a division leading to isolated pockets and concluded that different control strategies are needed in different regions that account for exchanges with neighbouring countries. Melachio [2015 (115)] used Wright’s F-statistics and Bayesian clustering methods and found that there was substantial gene flow between different subpopulations of G. palpalis palpalis in Cameroon. This reflects the species high dispersion ability in unfragmented suitable habitats. Mayoke (2020 (116)) found two clusters separating G. fuscipes and G. palpalis palpalis with several subclusters within, indicating that there is unrestricted dispersal of the palpalis group within Congo Brazzaville. This means that the population should be considered for eradication by using area-wide approaches, and not subpopulations. To detect first generation migrants in Kenya, the Bayesian method of Rannala and Mountain was used [Okeyo 2018 (117)]. Two genetically distinct clusters of G. pallidipes exist, separated by the biogeographic barrier Great Rift Valley; however, genetic exchanges show it is not an impermeable one. Dispersal between populations in Western Kenya is low, whereas populations in Eastern Kenya are more connected meaning more continuous and suitable habitat, leading to freer movement over generations of tsetse. Opiro [2017 (118)] used Discriminant Analysis of Principle Components, a multivariate model-free method, to describe patterns of clustering of groups of G. fuscipes fuscipes and Wright’s F-statistics, along with isolation by distance using Rousset’s procedure. The Bayesian method of Rannala and Mountain was used to detect first generation migrants. Three genetic clusters were found with high levels of gene flow between them, meaning control efforts could be undercut by re-invasion from neighbouring clusters. Ecological and geographic features allow for populations to be connected, which can be an advantage in designing control strategies. Brito [2025 (22)] used Wright’s F-statistics, isolation by distance using Rousset’s approach and Bayesian methods to determine dispersal patterns of G. brevipalpis between two parks in Mozambique. They found evidence of exchange of rare individuals even though the populations within the two parks were supposedly isolated and distant. This was attributed to passive dispersal through human transport means. De Meeus [2019 (51)] analysed data from different tsetse species using isolation by distance to investigate density-dependent dispersal i.e. when a species migrates based on the local population size. They determined that tsetse display negative density-dependent dispersal, which has major implications, since reinvasion risk is more significant in populations depleted by vector control campaigns. Microsatellite and mitochondrial DNA markers have been used in similar ways by Ouma [2005 (119)], Ouma [2011 (120)], Solano [2009 (121); 2010 (122)], De Meeus [2012 (123)], Echodu [2013 (124)], Hyseni [2012 (125)], Krafsur [2001 (126)] and Mayoke [2020 (127)] to assess genetic diversity and dispersal to aid control efforts. A habitat suitability model for G. palpalis gambiensis was corrected using a friction map that explains genetic distances between populations, or tsetse movement which revealed unconnected habitat patches that could be targeted by eradication programmes [Bouyer 2018 (128)]. This approach does not rely on expert opinion and is therefore reproducible and a way to improve species distribution models. Bouyer (2015 (21)) produced an evidence-based map of landscape connectivity and identified the most likely dispersal pathways of G. palpalis gambiensis by integrating genetic analyses, landscape friction modelling, and tsetse distribution models. Tsetse dispersal paths were identified using geostatistical methods and provided a friction map showing ecological barriers to dispersal. Saarman [2018 (129)] also used population genetics by developing a novel pipeline integrating population genetics data, environmental data and field-survey data to identify isolated geographic regions considered for tsetse control. This was validated using field surveys for tsetse in areas predicted to have suitable habitat and isolated from other suitable habitats. They built on the work by (21) by identifying environmental factors important in genetic differentiation over multiple generations and then building a habitat isolation model based on current conditions.

4 Discussion

4.1 Strengths and limitations of types of models used

Two thirds of the modelling approaches assessed in this scoping review are statistical in nature. Diall et al. (2017) (130), highlighted a few novel tools that can be used as part of PCP for nagana: (i) atlases for tsetse and nagana; (ii) distribution models, (iii) tsetse population genetics, friction and isolation; and (iv) spatially explicit benefit-cost analyses. The latter tool is outside the scope of this work, and therefore, the former three tools will be used as the basis for initial discussion. A general discussion about what considerations should be taken when choosing an appropriate model follows thereafter.

Atlases of tsetse have been developed in 14 countries, with 7 published. Atlases are geospatial databases that contain collated field data that has been centralised, harmonised and mapped. Three studies in this review have used or developed an atlas and further applied modelling to determine the effects of environmental factors in relation to distribution of tsetse. Hendrickx et al. (99) introduced area-wide recording of tsetse presence and abundance using survey data and remotely sensed environmental data to describe how tsetse distribution is affected by the environment, in Togo. Seasonal clusters were derived using hierarchical cluster analysis for the satellite data and regression to look at the effects of the environment on tsetse abundance. Spatial smoothing was applied to adjust for seasonal fluctuations resulting in maps that reflect presence risks of species. Although the techniques used may not have been the most sophisticated, it seems to be the foundational work that has led to using GIS and tsetse survey data to develop the distribution maps that followed. Mugenyi, Muhanguzi (87) obtained tsetse data from different sources using a structured protocol adapted from (131) in Uganda. They then used Zero-inflated Poisson regression models to develop a spatial predictive model on tsetse abundance with the environmental factors as predictors. The result was a tsetse prediction map with low, medium and high predicted tsetse densities. They illustrated the use of following a protocol to collate data and produce maps in a country that previously did not have any. Other than game parks, however, they did not seem to incorporate data on host availability. Mongare, Gachoki (82) used the national atlas developed in Kenya together with environmental, topographic, demographic and animal factors to develop four species distribution models, which were combined to produce an ensemble model to predict tsetse habitat suitability. The limitations of species distribution models, however, are described in more detail below. These studies illustrate the need for a centralised system to manage tsetse data (130) and to highlight the gaps in which national surveillance and intervention efforts can be directed. They also show that although these atlases have been developed in several countries already, in-depth analysis and modelling has not been conducted using the volume of data that has been collated (132).

While species distribution models are useful in describing species niches and predicting future changes in distributions based on the habitat suitability, evidence shows that there is crucial need to incorporate dispersal, spatial structures of the environment and population dynamics into these types of models (133). Mechanistic species distribution models combine the correlative nature of species distribution models and dispersal models. Joint species distribution models are also an extension that enables multiple species to be modelled simultaneously, accounting for biotic interactions that may not be explained by environmental factors alone (134). In addition, imperfect data and detection of species lead to complex ecological processes. It is not enough simply to describe the spatial variability; ecological models of distribution should explicitly model the measurement error process and incorporate uncertainty, to achieve unbiased inferences (135).

Studies using population genetics have been prominent in determining dispersal rates of tsetse and to identify isolated tsetse populations. Microsatellite markers have been used to determine dispersal distances by detecting local gene flow and have been applied across Glossina species to assess differentiation among neighbouring populations (23). These approaches are useful in informing what the most appropriate control strategies could be, where isolated populations could be targeted for elimination, and non-isolated populations could be targeted for an integrated management approach (130). These techniques make simplified statistical assumptions which may not reflect the precise theoretical population genetics processes (23, 136). A way to determine whether there are problems with the genetic inferences using these models could be to make comparisons with demographic estimates or using mark-release-recapture methods (48).

4.1.1 Considerations for model choice

The famous quote, “all models are wrong, but some are useful”, aptly summarises the assumptions and limitations inherent in ecological modelling. Model choice should be guided by the research question, the intended purpose of the model, and the availability and quality of data (Table 3). Broadly, ecological models can be classified as mechanistic or empirical. Mechanistic models aim to explain the underlying processes governing ecological systems and should provide reasonable approximations of reality while remaining “as simple as possible, but no simpler” (attributed to Einstein) (137). Empirical models, in contrast, use observations and data to construct statistical relationships that are calibrated and often validated using independent datasets. Their primary purpose is explanation, prediction or forecasting rather than explicit representation of ecological processes.

In ecology, models are often used both to explain patterns and to predict future outcomes. For example, a model may explain the relationship between species occurrence and habitat characteristics, while also predicting how species abundance may change under future environmental conditions. Predictive applications require adequate predictive performance and a clear understanding of model uncertainty. Reducing uncertainty depends on selecting appropriate explanatory variables, obtaining sufficient information, and conducting rigorous calibration and validation using independent datasets.

A key consideration in modelling is how ecological processes are represented in space and time. Models of insect dispersal are typically either spatially implicit, assuming a closed system with no movement into or out of the study region, or spatially explicit, in which movement is represented directly through random dispersal or diffusion processes (138). Spatially explicit models have become increasingly common with advances in computational power and allow researchers to investigate population dynamics at specific locations by accounting for spatial heterogeneity, organism behaviour and landscape structure (139). Additionally, ecological observations are often spatially and temporally autocorrelated, meaning that measurements that are closer in space or time tend to be more similar than those further apart. Accounting for such dependence structures is important for obtaining reliable parameter estimates and realistic uncertainty estimates. Temporal processes may be represented in either continuous or discrete time, depending on the ecological questions being addressed and the availability of data.

Many ecological systems are inherently dynamic, requiring models that account for both spatial and temporal variation. Static spatial models may be useful for describing patterns at a single time point, but understanding ecological responses to environmental change often requires spatio-temporal models (135). Such models are particularly important when evaluating the effects of global change on species distributions and population dynamics. An additional challenge arises when spatial or temporal datasets are collected at different scales, creating problems of misalignment (140). Misalignment can distort relationships between predictors and outcomes, reduce predictive performance and contribute to ecological fallacies. These issues can be addressed through data harmonisation procedures, scale-adjustment methods or statistical frameworks that explicitly accommodate differences in spatial and temporal resolution.

Models also differ in how they represent uncertainty. Deterministic models assume that a given set of inputs will always produce the same output and are often used to represent ecological mechanisms directly. For example, dispersal models may incorporate factors such as climate, habitat suitability, predation and competition to simulate population dynamics through time and space. Such models can be valuable when the objective is to understand ecological mechanisms, evaluate interventions or test alternative assumptions in systems where some data are unavailable. However, ecological systems are influenced by unpredictable environmental and biological processes (141). Consequently, deterministic models may be less suitable when accurate prediction is the primary objective. In these situations, stochastic models are often preferred because they explicitly incorporate randomness and generate probability distributions of possible outcomes. Stochastic approaches therefore provide a natural framework for representing uncertainty and forecasting ecological dynamics under varying conditions.

Forecasting models are widely used to predict future species distributions and population responses, such as under climate change. Such applications require high-quality historical data from which ecological relationships and temporal trends can be estimated. Forecasting also relies on temporal models that use time-series information in combination with exogenous environmental drivers. As forecast horizons increase, uncertainty inevitably grows. Multiple sources of uncertainty, including parameter uncertainty, environmental variability, process error and model structural uncertainty, contribute to this challenge (142).

Traditionally, many forecasting studies have relied on correlative approaches. Increasingly, however, causal and hierarchical modelling frameworks are being developed that integrate ecological mechanisms with statistical inference. These approaches retain many advantages of mechanistic models while explicitly incorporating multiple layers of uncertainty and complex ecological processes (135). Using Bayesian inference in these frameworks has been favoured in recent decades due to its ability to handle complex hierarchical systems, missing data and distinguishing between actual ecological processes such as true absences, and observation errors, such as imperfect detection in field surveys (143) (144). Bayesian inference is an alternative to frequentist inference in statistical methodology, combining prior knowledge with observed field data to generate a posterior or updated belief about an outcome.

Ecological systems are characterised by complex interactions occurring across multiple spatial and temporal scales (145). Capturing these complexities often requires balancing competing modelling objectives. Levins (145) famously described trade-offs between realism, generality and precision. Models that aim to represent all aspects of ecological systems may become difficult to parameterise, validate and interpret. In the context of tsetse fly ecology, models that attempt to represent population dynamics in detail may require numerous parameters and time-lagged processes, making calibration and validation increasingly challenging (146). When analytical solutions become intractable, simulation models offer a practical alternative by allowing complex ecological systems to be investigated computationally (147). Simulation approaches are particularly valuable in adaptive management frameworks, where models are continuously updated as new information becomes available. They allow managers to evaluate multiple scenarios, assess uncertainty and explore alternative intervention strategies. Similarly, optimisation models can be used to identify management actions that best achieve a specified objective, such as maximising disease control while minimising implementation costs (148). These approaches are particularly useful for stakeholder engagement and for guiding field surveillance and intervention programmes.

Another important modelling decision concerns the level at which ecological processes are represented. Most ecological models operate at the population level and assume that variation among individuals can be adequately represented by average rates. In contrast, agent-based or individual-based models explicitly account for individual heterogeneity and can capture emergent behaviours and spatial interactions that population-level models may miss. Although these approaches often provide greater ecological realism, as well as integrating, simulating and making use of existing information, they typically require more detailed data and can be computationally demanding (149).

Regardless of the modelling framework used, models should not replace empirical research. Rather, they should complement field studies by identifying important knowledge gaps, generating hypotheses and helping to prioritise data collection efforts (145). Models that have undergone rigorous verification and validation can provide valuable support for ecological management and decision-making. For example, they can be used to compare alternative vector control strategies, assess intervention outcomes and evaluate future risks. Increasingly, hierarchical models are being used to represent complex ecological processes; however, as model complexity increases, careful evaluation becomes even more important. Simulation studies that assess assumption violations, robustness and predictive performance are therefore essential, particularly when model outputs are intended to inform policy or management decisions (150).

4.2 Implications of changing climate and environment on tsetse distribution

With temperatures projected to increase by 1.5-3˚C in Africa in the 21st century, compared to the late 20th century, the geographical distributions of vectors and their associated diseases is expected to change [Niang et al., 2014 (151)]. The African continent is experiencing warming at a faster pace; Central Africa is expected to see higher levels of rainfall and higher temperatures, Southern Africa is expected to become drier and see increasing temperatures of 2 ˚C and West Africa an increase in the number of hot days (Africa Renewal, 2019 (152) (153).

Although many studies assess the relationships between climatic and environmental factors with tsetse distributions and predict where tsetse will occur, the exact change is unclear. To be able to establish causal relationships between climate change and vector-borne diseases, there is a need to recognise biological sensitivities to climate, meteorological evidence of climate change, and evidence of entomological and/or epidemiological change in association with climate change [Kovats et al., 2001 (154)], although these effects are difficult to separate [Lord et al., 2018 (57)]. Temperature is seen as the main driver of changes in tsetse density and distribution, since it directly influences the fly’s population dynamics via parameters such as birth and death rates. Other variables were also shown to be important, such as altitude, relative humidity, rainfall and NDVI. Habitat also greatly affects tsetse distribution and is a limiting factor, since different species are suited to different habitat types. Tsetse are very likely to emerge under climate change scenarios as significant disease challenges. Data from Zimbabwe have been used to come up with possible climate change scenarios since there is a considerable volume of data available both on climate variables and on vector abundance. It is likely that there will be shifts in the ranges suitable for tsetse, rather than large-scale expansions of fly populations. These will put vulnerable populations at risk of outbreaks and, in Africa, where socioeconomic instabilities are increasing, this can threaten the livelihoods of many people. Many studies have considered the effects of abiotic factors on tsetse, however, few considered host availability, which is a crucial factor, and may also be influenced by habitat and environmental changes.

4.3 Implications for tsetse control programmes

For many decades in the 20th century, African countries carried out their own control of tsetse and trypanosomiasis, the disease caused by trypanosomes, of which tsetse flies are the vectors in Africa [FAO, 2000 (155)]. Later, these efforts have been augmented by several internationally funded and operated initiatives. These have aimed at controlling or even eradicating tsetse and trypanosomiasis [FAO, 2000 (155)]. These include the Programme against African Trypanosomosis (PAAT), the Food and Agriculture Organization of the United Nations (FAO), the World Health Organization (WHO), the International Atomic Energy Agency (IAEA) and the African Union, Inter-African Bureau for Animal Resources (AU–IBAR) and Pan-African Tsetse and Trypanosomosis Eradication Campaign (PATTEC), which was launched by Heads of State of the African Union (AU). These initiatives all require accurate information on tsetse ecology and the geographic distribution of tsetse flies [Cecchi et al., 2024 (3)]. These are required at both a large scale and finer resolution, so that intervention strategies may be formed, priority areas determined, and field activities properly planned and implemented [Cecchi et al., 2015 (131)]. Vector control has been possible in some countries where government funding has been available [Cheneau, 1993 (156)]. These efforts were based in part on the availability of financial resources and the government’s heed to the importance of this problem, based on social as well as economic values placed on livestock industry. Certain countries such as Zimbabwe, Zambia, Botswana, Nigeria, Cameroon and Uganda were able to adopt aggressive vector control programmes. Nigeria eliminated flies from over 200,000 km² from the north of the country and Zimbabwe made substantial and sustainable progress for many years. Botswana, too, has eliminated tsetse using deltamethrin [Kgori, 2006 (157)]. However, funding has slowly eroded over time, leading to very limited control in countries that had made progress previously. Continued control efforts by international organizations and national control programmes led to fewer than 1000 sleeping sickness cases as of 2024². Control of nagana has proven more difficult (130), and is essential in improving livestock output and public health (132). PCP, which was developed for other diseases such as foot-and-mouth disease, is a data-driven approach and is grounded on achieving control using priority areas for intervention by understanding different epidemiological settings, including vectors. That requires robust data, and sound and innovative modelling approaches to determine space-time trends, mapping of risk and predicting changes in relation to climate change (132).

4.4 Limitations

This review aimed to synthesise the literature around models of tsetse dispersal and distribution in relation to climate change and control. Although attempting to be as complete as possible, it does not consider pre-prints, grey literature or articles that were not in English. It also does not include reports and books/book chapters. To this extent, is not as comprehensive, especially in covering non-Anglophone countries.

Tsetse survival is contingent on hosts, and therefore host dynamics are an essential component in tsetse ecology and should be considered where possible. This has not been covered in this review. However, as an acknowledgment of the importance of this factor on tsetse dispersal, we provide one of the few, if not only, studies to model the effects of host population density on tsetse population dynamics (158).

5 Conclusions

As funding for tsetse control becomes more constrained and climate conditions grow more challenging, control programmes are undergoing a critical reappraisal to ensure their long-term sustainability. This has pushed the field towards more rigorous, evidence-based approaches, while also encouraging a broader perspective - one that situates trypanosomosis control within the wider goals of rural development.

References

• 1

  World Health Organization. Vector-Borne Diseases (2024). Available online at: https://www.who.int/news-room/fact-sheets/detail/vector-borne-diseases (Accessed April 16, 2026).

  • Google Scholar

• 2

  KrafsurES. Tsetse flies: genetics, evolution, and role as vectors. Infect Genet Evol. (2009) 9:124–41. doi: 10.1016/j.meegid.2008.09.010

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  CecchiGPaoneMDe GierJZhaoW. The Continental Atlas of the Distribution of Tsetse Flies in Africa. Rome, Italy: Food & Agriculture Org (2024).

  • Google Scholar

• 4

  HolmesP. Tsetse-transmitted trypanosomes – their biology, disease impact and control. J Invertebr Pathol. (2013) 112:S11–4. doi: 10.1016/j.jip.2012.07.014

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 5

  TilleyH. Ecologies of complexity: tropical environments, African trypanosomiasis, and the science of disease control in British colonial Africa, 1900-1940. Osiris. (2004) 19:21–38. doi: 10.1086/649392

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 6

  CarterRM. Tsetse fly in Arabia. Br Med J. (1906) 2:1393–4.

  • Google Scholar

• 7

  ElsenPAmoudiMALeclercqM. First record of Glossina fuscipes fuscipes Newstead, 1910 and Glossina morsitans submorsitans Newstead, 1910 in southwestern Saudi Arabia. Annales de la Société belge de médecine tropicale (1990) 70:281–7.

  • Google Scholar

• 8

  LambrechtFL. Aspects of evolution and ecology of tsetse flies and trypanosomiasis in prehistoric African environment. J Afr History. (1964) 5:1–24. doi: 10.1017/s0021853700004473

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  BuxtonPA. The natural history of tsetse flies. London, H.K.Lewis. xviii + 816 pp.

  • Google Scholar

• 10

  GlasgowJP. The Distribution and Abundance of Tsetse. 241 pp. Oxford, Pergamon. (1963).

  • Google Scholar

• 11

  De BeerCJVenterGJKappmeier GreenKEsterhuizenJDe KlerkDGNtshangaseJet al. An update of the tsetse fly (Diptera: Glossinidae) distribution and African animal trypanosomosis prevalence in north-eastern KwaZulu-Natal, South Africa. Onderstepoort J Veterinary Res. (2016) 83:1–10. doi: 10.4102/ojvr.v83i1.1172

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 12

  BursellE. The water balance of tsetse pupae. Philos Trans R Soc London Ser B Biol Sci. (1958) 241:179–210. doi: 10.1098/rstb.1958.0002

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 13

  BursellE. The Water Balance of Tsetse Flies. London: Transactions of the Royal Entomological Society (1959) 111:205–235.

  • Google Scholar

• 14

  ValeG. Artificial refuges for tsetse flies (Glossina spp.). Bull Entomol Res. (1971) 61:331–50. doi: 10.1017/s0007485300057874

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 15

  BursellE. The effect of temperature on the consumption of fat during pupal development in Glossina. Bull Entomol Res. (1960) 51:583–98. doi: 10.1017/s0007485300055188

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 16

  PhelpsRBurrowsP. Lethal temperatures for puparia of Glossina morsitans orientalis. Entomol Exp Appl. (1969) 12:23–32. doi: 10.1111/j.1570-7458.1969.tb02493.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  HargroveJWValeGA. Models for the rates of pupal development, fat consumption and mortality in tsetse (Glossina spp). Bull Entomol Res. (2020) 110:44–56. doi: 10.1017/s0007485319000233

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 18

  VreysenMJB. (1999). “ The distribution and relative abundance of tsetse flies in the Southern Rift Valley of Ethiopia: preliminary survey results”, in: 25th meeting of the International Scientific Council for Trypanosomiasis Research and Control, Mombasa, Kenya, 27 September-1 October, 1999 ( OAU/IBAR), 202–213.

  • Google Scholar

• 19

  NashT. The Ecology of Glossina morsitans, Westw. and two possible Methods for its Destruction.—Part I. Bull Entomol Res. (1933) 24:107–57. doi: 10.1017/S0007485300031357

  • CrossRef
  • Google Scholar

• 20

  ThuillerW. Ecological niche modelling. Curr Biol. (2024) 34:R225–9. doi: 10.1016/j.cub.2024.02.018

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  BouyerJDickoAHCecchiGRavelSGuerriniLSolanoPet al. Mapping landscape friction to locate isolated tsetse populations that are candidates for elimination. Proc Natl Acad Sci USA. (2015) 112:14575–80. doi: 10.1073/pnas.1516778112

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 22

  BritoDRASégardAMulandaneFCCossaNVMucacheHNRavelSet al. Unravelling genetic differentiation between Glossina brevipalpis populations from two distant National Parks in Mozambique. PloS NeglTrop Dis. (2025) 19:e0012953. doi: 10.1371/journal.pntd.0012953

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 23

  KrafsurES. Tsetse fly population genetics: an indirect approach to dispersal. Trends Parasitol. (2003) 19:162–6. doi: 10.1016/s1471-4922(03)00034-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 24

  JorgensenSEBendoricchioG. Fundamentals of Ecological Modelling. Elsevier (2001).

  • Google Scholar

• 25

  PazS. Climate change: a driver of increasing vector-borne disease transmission in non-endemic areas. PloS Med. (2024) 21:e1004382. doi: 10.1371/journal.pmed.1004382

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 26

  World Health Organization. Trypanosomiasis, Human African (Sleeping Sickness) (2023). Available online at: https://www.who.int/news-room/fact-sheets/detail/trypanosomiasis-human-african-(sleeping-sickness) (Accessed April 16, 2026).

  • Google Scholar

• 27

  DiallOCecchiGWandaGArgilés-HerreroRVreysenMJCattoliGet al. Developing a progressive control pathway for African animal trypanosomosis. Trends in parasitology. (2017) 33:499–509.

  • Google Scholar

• 28

  HargroveJW. A theoretical study of the invasion of cleared areas by tsetse flies (Diptera: Glossinidae). Bull Entomol Res. (2000) 90:201–9. doi: 10.1017/s0007485300000328

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 29

  JacksonC. The Analysis of a Tsetse-Fly Population. J Anim Ecol. (1939) 8:238–246.

  • Google Scholar

• 30

  JacksonCHN. The analysis of a tsetse-fly population. II. Ann Eugenics. (1943) 12:176–205. doi: 10.1111/j.1469-1809.1943.tb02322.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 31

  JacksonC. An artificially isolated generation of tsetse flies (Diptera). Bull Entomol Res. (1946) 37:291–9. doi: 10.1017/s0007485300022203

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  BursellEM.AH.W.Dispersal and Concentration of Glossina. London: George Allen and Unwin (1970).

  • Google Scholar

• 33

  ArkseyHO'MalleyL. Scoping studies: towards a methodological framework. Int J Soc Res Method. (2005) 8:19–32. doi: 10.1080/1364557032000119616

  • CrossRef
  • Google Scholar

• 34

  LevacDColquhounHO'brienKK. Scoping studies: advancing the methodology. Implementation Sci. (2010) 5:1–9. doi: 10.1186/1748-5908-5-69

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 35

  PetersMDGodfreyCMKhalilHMcInerneyPParkerDSoaresCB. Guidance for conducting systematic scoping reviews. JBI Evidence Implementation. (2015) 13:141–56. doi: 10.1097/xeb.0000000000000050

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 36

  OuzzaniMHammadyHFedorowiczZElmagarmidA. Rayyan—a web and mobile app for systematic reviews. Systematic Rev. (2016) 5:1–10. doi: 10.1186/s13643-016-0384-4

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 37

  HaddawayNRPageMJPritchardCCMcGuinnessLA. PRISMA2020: An R package and Shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and Open Synthesis. Campbell Syst Rev. (2022) 18:e1230. doi: 10.1101/2021.07.14.21260492

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 38

  RogersD. Study of a natural population of Glossina fuscipes fuscipes Newstead and a model of fly movement. J Anim Ecol. (1977) 46:309–30. doi: 10.2307/3962

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 39

  HargroveJ. Tsetse dispersal reconsidered. J Anim Ecol. (1981) 50:351–73. doi: 10.2307/4060

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 40

  JacksonCHN. The analysis of a tsetse-fly population. III. Ann Eugenics. (1947) 14:91–108. doi: 10.1111/j.1469-1809.1947.tb02381.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  WilliamsBDransfieldRBrightwellR. The control of tsetse flies in relation to fly movement and trapping efficiency. J Appl Ecol. (1992) 29:163–79. doi: 10.2307/2404359

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  HargroveJ. Optimized simulation of the control of tsetse flies Glossina pallidipes and G. m. morsitans (Diptera: Glossinidae) using odour-baited targets in Zimbabwe. Bull Entomol Res. (2003) 93:19–29. doi: 10.1079/ber2002205

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 43

  ValeGALovemoreDFlintSCockbillG. Odour-baited targets to control tsetse flies, Glossina spp.(Diptera: Glossinidae), in Zimbabwe. Bull Entomol Res. (1988) 78:31–49. doi: 10.1017/S0007485300016059

  • CrossRef
  • Google Scholar

• 44

  WilliamsB. Modelling movement and mortality: killing tsetse flies in the field. Comput Electron Agric. (1995) 13:155–75. doi: 10.1016/0168-1699(95)00022-v

  • CrossRef
  • Google Scholar

• 45

  BrightwellRDransfieldRWilliamsB. Factors affecting seasonal dispersal of the tsetse flies Glossina pallidipes and G. longipennis (Diptera: Glossinidae) at Nguruman, south-west Kenya. Bull Entomol Res. (1992) 82:167–82. doi: 10.1017/s0007485300051695

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 46

  ValeGAHurseyBSHargroveJWTorrSJAllsoppR. The use of small plots to study populations of tsetse (Diptera: glossinidae). Int J Trop Insect Sci. (1984) 5:403–10. doi: 10.1017/s1742758400008730

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 47

  ChildsSJ. The finite element implementation of a K.P.P. equation for the simulation of tsetse control measures in the vicinity of a game reserve. Math Biosci. (2010) 227:29–43. doi: 10.1016/j.mbs.2010.05.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 48

  VreysenMJBBalenghienTSalehKMMaigaSKoudougouZCecchiGet al. Release-recapture studies confirm dispersal of Glossina palpalis gambiensis between river basins in Mali. PloS NeglTrop Dis. (2013) 7:e2022. doi: 10.1371/journal.pntd.0002022

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  HargroveJWVan SickleJValeGALucasER. Negative density-dependent dispersal in tsetse (Glossina spp): An artefact of inappropriate analysis. PloS NeglTrop Dis. (2021) 15:e0009026. doi: 10.1371/journal.pntd.0009026

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 50

  HargroveJValeG. Negative density‐dependent dispersal in tsetse (Glossina spp): red flag or red herring? Med Vet Entomol. (2021) 35:30–41. doi: 10.1111/mve.12466

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  De MeeûsTRavelSSolanoPBouyerJ. Negative density-dependent dispersal in tsetse flies: a risk for control campaigns? Trends Parasitol. (2019) 35:615–21. doi: 10.1016/j.pt.2019.05.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 52

  BarclayHJVreysenMJ. A dynamic population model for tsetse (Diptera: Glossinidae) area-wide integrated pest management. Popul Ecol. (2011) 53:89–110. doi: 10.1007/s10144-010-0224-7

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 53

  HargroveJWTorrSJValeGA. Comment on Barclay and Vreysen: published dynamic population model for tsetse cannot fit field data. Popul Ecol. (2011) 53:413–5. doi: 10.1007/s10144-010-0259-9

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 54

  BarclayHJVreysenMJ. The interaction of dispersal and control methods for the riverine tsetse fly Glossina palpalis gambiensis (Diptera: Glossinidae): a modelling study. Popul Ecol. (2013) 55:53–68. doi: 10.1007/s10144-012-0339-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 55

  CeciliaHArnouxSPicaultSDickoASeckMTSallBet al. Dispersal in heterogeneous environments drives population dynamics and control of tsetse flies. Proc Biol Sci. (2021) 288:20202810. doi: 10.1098/rspb.2020.2810

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 56

  LongbottomJCaminadeCGibsonHSWeissDJTorrSLordJS. Modelling the impact of climate change on the distribution and abundance of tsetse in Northern Zimbabwe. Parasites Vectors. (2020) 13:1–11. doi: 10.1186/s13071-020-04398-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 57

  LordJSTorrSJAutyHKBrockPMByamunguMHargroveJWet al. Geostatistical models using remotely‐sensed data predict savanna tsetse decline across the interface between protected and unprotected areas in Serengeti, Tanzania. J Appl Ecol. (2018) 55:1997–2007. doi: 10.1111/1365-2664.13091

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 58

  GouteuxJ-PArtzrouniMJarryM. A density-dependent model with reinvasion for estimating tsetse fly populations (Diptera: Glossinidae) through trapping. Bull Entomol Res. (2001) 91:177–83. doi: 10.1079/ber200185

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 59

  PeckSL. Networks of habitat patches in tsetse fly control: implications of metapopulation structure on assessing local extinction probabilities. Ecol Modell. (2012) 246:99–102. doi: 10.1016/j.ecolmodel.2012.07.016

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 60

  HargroveJ. Extinction probabilities and times to extinction for populations of tsetse flies Glossina spp.(Diptera: Glossinidae) subjected to various control measures. Bull Entomol Res. (2005) 95:13–21. doi: 10.1079/ber2004335

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 61

  YuPHabtemariamTOryangDObasaMNganwaDRobnettV. Integration of temporal and spatial models for examining the epidemiology of African trypansomiasis. Prev Veterinary Med. (1995) 24:83–95. doi: 10.1016/0167-5877(95)00465-9

  • CrossRef
  • Google Scholar

• 62

  YuPHabtemariamTOryangDObasaMNganwaDRobnettV. Stochastic model of spatial spread and control of tsetse flies (Diptera: Muscidae). Environ Entomol. (1996) 25:78–84. doi: 10.1093/ee/25.1.78

  • CrossRef
  • Google Scholar

• 63

  LinSDeVisserMHMessinaJP. An agent-based model to simulate tsetse fly distribution and control techniques: a case study in Nguruman, Kenya. Ecol Modell. (2015) 314:80–9. doi: 10.1016/j.ecolmodel.2015.07.015

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 64

  DeVisserMHMessinaJPMooreNJLuschDPMaitimaJ. A dynamic species distribution model of Glossina subgenus Morsitans: the identification of tsetse reservoirs and refugia. Ecosphere. (2010) 1:1–21. doi: 10.1890/es10-00006.1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 65

  MessinaJPMooreNJDeVisserMHMcCordPFWalkerED. Climate change and risk projection: dynamic spatial models of tsetse and African trypanosomiasis in Kenya. Ann Assoc Am Geogr. (2012) 102:1038–48. doi: 10.1080/00045608.2012.671134

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 66

  McCordPFMessinaJPCampbellDJGradySC. Tsetse fly control in Kenya's spatially and temporally dynamic control reservoirs: a cost analysis. Appl Geogr. (2012) 34:189–204. doi: 10.1016/j.apgeog.2011.11.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 67

  De DekenRSumbuJMpianaSMansinsaPWat'sengaFLutumbaPet al. Trypanosomiasis in Kinshasa: distribution of the vector, Glossina fuscipes quanzensis, and risk of transmission in the peri-urban area. Med Vet Entomol. (2005) 19:353–9. doi: 10.1111/j.1365-2915.2005.00580.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 68

  MooreNMessinaJ. A landscape and climate data logistic model of tsetse distribution in Kenya. PloS One. (2010) 5:e11809. doi: 10.1371/journal.pone.0011809

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 69

  DickoAHLancelotRSeckMTGuerriniLSallBLoMet al. Using species distribution models to optimize vector control in the framework of the tsetse eradication campaign in Senegal. Proc Natl Acad Sci USA. (2014) 111:10149–54. doi: 10.1073/pnas.1407773111

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 70

  MatawaFMurwiraAZengeyaFMAtkinsonPM. Modelling the spatial-temporal distribution of tsetse (Glossina pallidipes) as a function of topography and vegetation greenness in the Zambezi Valley of Zimbabwe. Appl Geogr. (2016) 76:198–206. doi: 10.1016/j.apgeog.2016.09.008

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 71

  AlbertMWardropNAAtkinsonPMTorrSJWelburnSC. Tsetse fly (G. f. fuscipes) distribution in the Lake Victoria basin of Uganda. PloS NeglTrop Dis. (2015) 9:e0003705. doi: 10.1371/journal.pntd.0003705

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 72

  MatawaFMurwiraAAtkinsonPM. Evaluating the impact of declining tsetse fly (Glossina pallidipes) habitat in the Zambezi valley of Zimbabwe. Geocarto Int. (2020) 35:1373–84. doi: 10.1080/10106049.2019.1576780

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 73

  TiradosIHopeASelbyRMpembeleFMiakaEMBoelaertMet al. Impact of tiny targets on Glossina fuscipes quanzensis, the primary vector of human African trypanosomiasis in the Democratic Republic of the Congo. PloS NeglTrop Dis. (2020) 14:e0008270. doi: 10.1371/journal.pntd.0008270

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 74

  NnkoHJGwakisaPSNgonyokaASindatoCEstesAB. Potential impacts of climate change on geographical distribution of three primary vectors of African Trypanosomiasis in Tanzania’s Maasai Steppe: G. m. morsitans, G. pallidipes and G. swynnertoni. PloS NeglTrop Dis. (2021) 15:e0009081. doi: 10.1371/journal.pntd.0009081

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 75

  ScoonesIDzingiraiVAndersonNMacLeodEMangwanyaLMatawaFet al. People, patches, and parasites: the case of trypanosomiasis in Zimbabwe. Hum Ecol Interdiscip J. (2017) 45:643–54. doi: 10.1007/s10745-017-9929-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 76

  BishopAPAmatulliGHyseniCPlessEBatetaROkeyoWAet al. A machine learning approach to integrating genetic and ecological data in tsetse flies (Glossina pallidipes) for spatially explicit vector control planning. Evol Appl. (2021) 14:1762–77. doi: 10.1111/eva.13237

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 77

  LongbottomJEsterhuizenJHopeALehaneMJMangwiroTCMugenyiAet al. Impact of a national tsetse control programme to eliminate Gambian sleeping sickness in Uganda: a spatiotemporal modelling study. BMJ Glob Health. (2024) 9(10):e015374. doi: 10.1136/bmjgh-2024-015374

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 78

  ZhouRGaoYChangNGaoTMaDLiCet al. Projecting the potential distribution of Glossinamorsitans (Diptera: Glossinidae) under climate change using the MaxEnt model. Biol (Basel). (2021) 10(11):1150. doi: 10.3390/biology10111150

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 79

  ChikoworeGDickoAHChinwadaPZimbaMShereniWRogerFet al. A pilot study to delimit tsetse target populations in Zimbabwe. PloS NeglTrop Dis. (2017) 11:e0005566. doi: 10.1371/journal.pntd.0005566

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 80

  GachokiSGroenTVrielingAOkalMSkidmoreAMasigaD. Satellite-based modelling of potential tsetse (Glossina pallidipes) breeding and foraging sites using teneral and non-teneral fly occurrence data. Parasit Vectors. (2021) 14:506. doi: 10.1186/s13071-021-05017-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 81

  GachokiSGroenTAVrielingASkidmoreAMasigaD. Towards accurate spatial prediction of Glossina pallidipes relative densities at country-scale in Kenya. Ecol Inf. (2024) 81:102610. doi: 10.1016/j.ecoinf.2024.102610

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 82

  MongareRGachokiSAdamEKimathiEBarreauxAMGCecchiGet al. Species distribution modeling to predict tsetse fly (Glossina spp.) habitat suitability in Kenya. Parasit Vectors. (2025) 18:378. doi: 10.1186/s13071-025-06938-1

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 83

  KerrJTKulkarniMAlgarA. Integrating theory and predictive modeling for conservation research. In: Predictive Species and Habitat Modeling in Landscape Ecology: Concepts and Applications. New York: Springer (2010). p. 9–28.

  • Google Scholar

• 84

  de BeerCJDickoAHNtshangaseJMoyabaPTaioeMOMulandaneFCet al. A distribution model for Glossina brevipalpis and Glossina austeni in Southern Mozambique, Eswatini and South Africa for enhanced area-wide integrated pest management approaches. PloS NeglTrop Dis. (2021) 15:e0009989. doi: 10.1371/journal.pntd.0009989

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 85

  GashururuRSGithigiaSMGasanaMNHabimanaRMaingiNCecchiGet al. An update on the distribution of Glossina (tsetse flies) at the wildlife-human-livestock interface of Akagera National Park, Rwanda. Parasit Vectors. (2021) 14:294. doi: 10.1186/s13071-021-04786-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 86

  KargboADafkaSOsmanAMKouaHKVieiraRFCRocklovJ. Impact of climate change and variability on the occurrence and distribution of Trypanosoma vectors in The Gambia. Parasitol Res. (2025) 124:29. doi: 10.1007/s00436-025-08475-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 87

  MugenyiAMuhanguziDHendrickxGNicolasGWaiswaCTorrSet al. Spatial analysis of G.f.fuscipes abundance in Uganda using Poisson and Zero-Inflated Poisson regression models. PloS NeglTrop Dis. (2021) 15:e0009820. doi: 10.1371/journal.pntd.0009820

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 88

  WangoolaMROgwalGMagamboHBingiPKasaweMNampalaAU. Tsetse and African Animal Trypanosomiasis status; prevalence, spatial distribution, and implications to control trypanosomiasis in Uganda. Trop Anim Health Prod. (2025) 57:191. doi: 10.1007/s11250-025-04440-5

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 89

  RogersDJRandolphSE. Distribution and abundance of tsetse flies (Glossina spp.). J Anim Ecol. (1986) 55:1007–25. doi: 10.2307/4430

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 90

  KitronUOtienoLHungerfordLOdulajaABrighamWOkelloOet al. Spatial analysis of the distribution of tsetse flies in the Lambwe Valley, Kenya, using Landsat TM satellite imagery and GIS. J Anim Ecol. (1996) 65:371–80. doi: 10.2307/5883

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 91

  RogersDHendrickxGSlingenberghJ. Tsetse flies and their control. Rev Scientifique Technique-Office Int Des Epizooties. (1994) 13:1075–124. doi: 10.20506/rst.13.4.811

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 92

  RogersDHaySPackerM. Predicting the distribution of tsetse flies in West Africa using temporal Fourier processed meteorological satellite data. Ann Trop Med Parasitol. (1996) 90:225–41. doi: 10.1080/00034983.1996.11813049

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 93

  HendrickxGNapalaARogersDBastiaensenPSlingenberghJ. Can remotely sensed meteorological data significantly contribute to reduce costs of tsetse surveys? Mem Inst Oswaldo Cruz. (1999) 94:273–6. doi: 10.1590/s0074-02761999000200028

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 94

  HendrickxGNapalaASlingenberghJHDe DekenRRogersDJ. A contribution towards simplifying area-wide tsetse surveys using medium resolution meteorological satellite data. Bull Entomol Res. (2001) 91:333–46. doi: 10.1079/ber2001103

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 95

  RogersDJWilliamsBG. Monitoring trypanosomiasis in space and time. Parasitology. (1993) 106 Suppl:S77–92. doi: 10.1017/s0031182000086133

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 96

  RobinsonTRogersDWilliamsB. Mapping tsetse habitat suitability in the common fly belt of southern Africa using multivariate analysis of climate and remotely sensed vegetation data. Med Vet Entomol. (1997) 11:235–45. doi: 10.1111/j.1365-2915.1997.tb00401.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 97

  RobinsonTRogersDWilliamsB. Univariate analysis of tsetse habitat in the common fly belt of southern Africa using climate and remotely sensed vegetation data. Med Vet Entomol. (1997) 11:223–34. doi: 10.1111/j.1365-2915.1997.tb00400.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 98

  GuerriniLBordJDucheyneEBouyerJ. Fragmentation analysis for prediction of suitable habitat for vectors: example of riverine tsetse flies in Burkina Faso. J Med Entomol. (2008) 45:1180–6. doi: 10.1603/0022-2585(2008)45[1180:fafpos]2.0.co;2

  • CrossRef
  • Google Scholar

• 99

  HendrickxGNapalaADaoBBatawuiDDe DekenRVermeilenAet al. A systematic approach to area-wide tsetse distribution and abundance maps. Bull Entomol Res. (1999) 89:231–44. doi: 10.1017/s0007485399000358

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 100

  SeddaLGuerriniLBouyerJKonéNRogersDJ. Spatio‐temporal modelling of Glossina palpalis gambiensis and Glossina tachinoides apparent densities in fragmented ecosystems of Burkina Faso. Ecography. (2010) 33:772–83. doi: 10.1111/j.1600-0587.2009.06135.x

  • CrossRef
  • Google Scholar

• 101

  SciarrettaAGirmaMTikubetGBelayehunLBalloSBaumgärtnerJ. Development of an adaptive tsetse population management scheme for the Luke community, Ethiopia. J Med Entomol. (2005) 42:1006–19. doi: 10.1093/jmedent/42.6.1006

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 102

  SciarrettaATikubetGBaumgärtnerJGirmaMTrematerraP. Spatial clustering and associations of two savannah tsetse species, Glossina morsitans submorsitans and Glossina pallidipes (Diptera: Glossinidae), for guiding interventions in an adaptive cattle health management framework. Bull Entomol Res. (2010) 100:661–70. doi: 10.1017/s0007485309990733

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 103

  CanoJDescalzoMANdong-MabaleNNdongo-AsumuPBobuakasiLBuatichéJNet al. Spatial and temporal variability of the Glossina palpalis palpalis population in the Mbini focus (Equatorial Guinea). Int J Health Geogr. (2007) 6:36. doi: 10.1186/1476-072x-6-36

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 104

  CanoJDescalzoMANdong-MabaleNNdongo-AsumuPBobuakasiLNzambo-OndoSet al. Predicted distribution and movement of Glossina palpalis palpalis (Diptera: Glossinidae) in the wet and dry seasons in the Kogo trypanosomiasis focus (Equatorial Guinea). J Vector Ecol. (2007) 32:218–25. doi: 10.3376/1081-1710(2007)32[218:pdamog]2.0.co;2

  • CrossRef
  • Google Scholar

• 105

  AdamYBouyerJDayoGKMahamaCIVreysenMJCecchiGet al. Genetic comparison of Glossina tachinoides populations in three river basins of the Upper West Region of Ghana and implications for tsetse control. Infect Genet Evol. (2014) 28:588–95. doi: 10.1016/j.meegid.2014.03.023

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 106

  BatetaRSaarmanNPOkeyoWADionKJohnsonTMirejiPOet al. Phylogeography and population structure of the tsetse fly Glossina pallidipes in Kenya and the Serengeti ecosystem. PloS NeglTrop Dis. (2020) 14:e0007855. doi: 10.1371/journal.pntd.0007855

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 107

  MelachioTTSimoGRavelSDe MeeûsTCausseSSolanoPet al. Population genetics of Glossina palpalis palpalis from central African sleeping sickness foci. Parasit Vectors. (2011) 4:140. doi: 10.1186/1756-3305-4-140

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 108

  BertéDDe MeeûsTKabaDSéréMDjohanVCourtinFet al. Population genetics of Glossina palpalis palpalis in sleeping sickness foci of Côte d'Ivoire before and after vector control. Infect Genet Evol. (2019) 75:103963. doi: 10.1016/j.meegid.2019.103963

  • CrossRef
  • Google Scholar

• 109

  KoneNDe MeeûsTBouyerJRavelSGuerriniLN'GoranEKet al. Population structuring of the tsetse Glossina tachinoides resulting from landscape fragmentation in the Mouhoun River basin, Burkina Faso. Med Vet Entomol. (2010) 24:162–8. doi: 10.1111/j.1365-2915.2010.00857.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 110

  RavelSRayaisseJ-BCourtinFSolanoPde MeeusT. Genetic signature of a recent southern range shift in Glossina tachinoides in East Burkina Faso. Infection Genet Evol. (2013) 18:309–14. doi: 10.1016/j.meegid.2013.05.024

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 111

  RavelSSégardAMolloBGMahamatMHArgiles-HerreroRBouyerJet al. Limited impact of vector control on the population genetic structure of Glossina fuscipes fuscipes from the sleeping sickness focus of Maro, Chad. Parasite. (2024) 31:13. doi: 10.1051/parasite/2024013

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 112

  KonéNBouyerJRavelSVreysenMJDomagniKTCausseSet al. Contrasting population structures of two vectors of African trypanosomoses in Burkina Faso: consequences for control. PloS NeglTrop Dis. (2011) 5:e1217. doi: 10.1371/journal.pntd.0001217

  • CrossRef
  • Google Scholar

• 113

  BouyerJBalenghienTRavelSVialLSidibéIThévenonSet al. Population sizes and dispersal pattern of tsetse flies: rolling on the river? Mol Ecol. (2009) 18:2787–97. doi: 10.1111/j.1365-294x.2009.04233.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 114

  RavelSMahamatMHSégardAArgilés-HerreroRBouyerJRayaisseJ-Bet al. Population genetics of Glossina fuscipes fuscipes from southern Chad. Peer Community Zoology. (2023) 3. doi: 10.24072/pcjournal.257

  • CrossRef
  • Google Scholar

• 115

  MélachioTTNjiokouFRavelSSimoGSolanoPDe MeeûsT. Effect of sampling methods, effective population size and migration rate estimation in Glossina palpalis palpalis from Cameroon. Infect Genet Evol. (2015) 33:150–7. doi: 10.1016/j.meegid.2015.04.023

  • CrossRef
  • Google Scholar

• 116

  MayokeAMuyaSMBatetaRMirejiPOOkothSOOnyoyoSGet al. Genetic diversity and phylogenetic relationships of tsetse flies of the palpalis group in Congo Brazzaville based on mitochondrial cox1 gene sequences. Parasit Vectors. (2020) 13:253. doi: 10.1186/s13071-020-04120-3

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 117

  OkeyoWASaarmanNPBatetaRDionKMengualMMirejiPOet al. Genetic differentiation of glossina pallidipes tsetse flies in southern Kenya. Am J Trop Med Hyg. (2018) 99:945–53. doi: 10.4269/ajtmh.18-0154

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 118

  OpiroRSaarmanNPEchoduROpiyoEADionKHalyardAet al. Genetic diversity and population structure of the tsetse fly Glossina fuscipes fuscipes (Diptera: Glossinidae) in Northern Uganda: Implications for vector control. PloS NeglTrop Dis. (2017) 11:e0005485. doi: 10.1371/journal.pntd.0005485

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 119

  OumaJOMarquezJGKrafsurES. Macrogeographic population structure of the tsetse fly, Glossina pallidipes (Diptera: Glossinidae). Bull Entomol Res. (2005) 95:437–47. doi: 10.1079/ber2005376

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 120

  OumaJOBeadellJSHyseniCOkediLMKrafsurESAksoySet al. Genetic diversity and population structure of Glossina pallidipes in Uganda and western Kenya. Parasit Vectors. (2011) 4:122. doi: 10.1186/1756-3305-4-122

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 121

  SolanoPRavelSBouyerJCamaraMKagbadounoMSDyerNet al. The population structure of Glossina palpalis gambiensis from island and continental locations in Coastal Guinea. PloS NeglTrop Dis. (2009) 3:e392. doi: 10.1371/journal.pntd.0000392

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 122

  SolanoPKabaDRavelSDyerNASallBVreysenMJet al. Population genetics as a tool to select tsetse control strategies: suppression or eradication of Glossina palpalis gambiensis in the Niayes of Senegal. PloS NeglTrop Dis. (2010) 4:e692. doi: 10.1371/journal.pntd.0000692

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 123

  De MeeûsTRavelSRayaisseJBCourtinFSolanoP. Understanding local population genetics of tsetse: the case of an isolated population of Glossina palpalis gambiensis in Burkina Faso. Infect Genet Evol. (2012) 12:1229–34. doi: 10.1016/j.meegid.2012.04.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 124

  EchoduRSistromMHyseniCEnyaruJOkediLAksoySet al. Genetically distinct Glossina fuscipes fuscipes populations in the Lake Kyoga region of Uganda and its relevance for human African trypanosomiasis. BioMed Res Int. (2013) 2013:614721. doi: 10.1155/2013/614721

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 125

  HyseniCKatoABOkediLMMasembeCOumaJOAksoySet al. The population structure of Glossina fuscipes fuscipes in the Lake Victoria basin in Uganda: implications for vector control. Parasit Vectors. (2012) 5:222. doi: 10.1186/1756-3305-5-222

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 126

  KrafsurESEndsleyMAWohlfordDLGriffithsNTAllsoppR. Genetic differentiation of Glossina morsitans centralis populations. Insect Mol Biol. (2001) 10:387–95. doi: 10.1007/s10709-006-9001-0

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 127

  MayokeAOumaJOMirejiPOOmondiSFMuyaSMItouaAet al. Population structure and migration patterns of the tsetse fly glossina fuscipes in congo-brazzaville. Am J Trop Med Hyg. (2020) 104:917–27. doi: 10.4269/ajtmh.20-0774

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 128

  BouyerJLancelotR. Using genetic data to improve species distribution models. Infect Genet Evol. (2018) 63:292–4. doi: 10.1016/j.meegid.2017.03.025

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 129

  SaarmanNBurakMOpiroRHyseniCEchoduRDionKet al. A spatial genetics approach to inform vector control of tsetse flies (Glossina fuscipes fuscipes) in Northern Uganda. Ecol Evol. (2018) 8:5336–54. doi: 10.1002/ece3.4050

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 130

  DiallOCecchiGWandaGArgilés-HerreroRVreysenMJBCattoliGet al. Developing a progressive control pathway for african animal trypanosomosis. Trends Parasitol. (2017) 33:499–509. doi: 10.1016/j.pt.2017.02.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 131

  CecchiGPaoneMArgilés HerreroRVreysenMJMattioliRC. Developing a continental atlas of the distribution and trypanosomal infection of tsetse flies (Glossina species). Parasites Vectors. (2015) 8:1–10. doi: 10.1186/s13071-015-0898-y

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 132

  CecchiGNgariNPaoneMde GierJOnyangoSBarreauxAet al. Disease intelligence and modelling for progressive control of animal trypanosomosis in Africa. Trends Parasitol. (2025) 41:734–45. doi: 10.1016/j.pt.2025.07.003

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 133

  RougierTLassalleGDrouineauHDumoulinNFaureTDeffuantGet al. The combined use of correlative and mechanistic species distribution models benefits low conservation status species. PloS One. (2015) 10:e0139194. doi: 10.1371/journal.pone.0139194

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 134

  WilkinsonDPGoldingNGuillera-ArroitaGTingleyRMcCarthyMA. Defining and evaluating predictions of joint species distribution models. Methods Ecol Evol. (2021) 12:394–405. doi: 10.1111/2041-210x.13518

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 135

  RoyleJKeryM. Applied Hierarchical Modeling in Ecology. Amsterdam: Elsevier (2016).

  • Google Scholar

• 136

  RoussetF. Inferences from spatial population genetics. In: Handbook of Statistical GeneticsChichester, United Kingdom: Wiley & Sons, Ltd. (2003). doi: 10.1002/0470022620.bbc24

  • CrossRef
  • Google Scholar

• 137

  RipleyBD. SELECTING AMONGST LARGE CLASSES OF MODELS. In: Methods and Models in Statistics: Published by Imperial College Press and Distributed by World Scientific Publishing CoLondon, UK: Imperial College Press (2004). p. 155–70. doi: 10.1142/9781860945410_0007

  • CrossRef
  • Google Scholar

• 138

  RuddWGGandourRW. Diffusion model for insect dispersal. J Econ Entomol. (1985) 78:295–301. doi: 10.1093/jee/78.2.295

  • CrossRef
  • Google Scholar

• 139

  DeAngelisDLYurekS. Spatially explicit modeling in ecology: A review. Ecosystems. (2017) 20:284–300. doi: 10.1007/s10021-016-0066-z

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 140

  UgarteMD. Banerjee, S. Carlin, B. P. and gelfand, A. E. Hierarchical modeling and analysis for spatial data. Second edition. CRC press/chapman & Hall. Monographs on statistics and applied probability 135, boca raton, florida, 2015. 562 pp. $ 99.95 . ISBN-13: 978-1-4398-1917-3 (Hardcover). Biometrics. (2015) 71:274–7. doi: 10.1111/biom.12290

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 141

  GillmanM. An Introduction to Mathematical Models in Ecology and Evolution: Time and Space. Chichester, UK: John Wiley & Sons (2009).

  • Google Scholar

• 142

  ClareJDJde ValpinePMoangaDATingleyMWBeissingerSR. A cloudy forecast for species distribution models: Predictive uncertainties abound for California birds after a century of climate and land-use change. Glob Chang Biol. (2024) 30:e17019. doi: 10.1111/gcb.17019

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 143

  EllisonAM. Bayesian inference in ecology. Ecol Lett. (2004) 7:509–20. doi: 10.1111/j.1461-0248.2004.00603.x

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 144

  BannerKMIrvineKMRodhouseTJ. The use of Bayesian priors in Ecology: The good, the bad and the not great. Methods Ecol Evol. (2020) 11:882–9. doi: 10.1111/2041-210x.13407

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 145

  JeffersJNR. Why do I need a model? In: JeffersJNR, editor. Modelling. Springer Netherlands, Dordrecht (1982). p. 15–23.

  • Google Scholar

• 146

  LevinsR. The strategy of model building in population biology. American Scientist, (1966) 54(4):421–31. Available online at: http://www.jstor.org/stable/27836590.

  • Google Scholar

• 147

  PeckSLBouyerJ. Mathematical modeling, spatial complexity, and critical decisions in tsetse control. J Econ Entomol. (2012) 105:1477–86. doi: 10.1603/ec12067

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 148

  RogersDJRandolphSE. From a case study to a theoretical basis for tsetse control. Int J Trop Insect Sci. (1984) 5:419–23. doi: 10.1017/s1742758400008766

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 149

  MurphyKJ. Agent-based models in applied ecology: Designing data-informed simulations for wildlife conservation and management. Ecosphere. (2025) 16:e70342. doi: 10.1002/ecs2.70342

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 150

  DiRenzoGVHanksEMillerDAW. A practical guide to understanding and validating complex models using data simulations. Methods Ecol Evol. (2023) 14:203–17. doi: 10.1111/2041-210x.14030

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 151

  NiangIRuppelOCAbdraboMAEsselALennardCPadghamJet al. Africa. In: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part B: Regional Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Barros, V.R., C.B. Field, D.J. Dokken, M.D. Mastrandrea, K.J. Mach, T.E. Bilir, M. Chatterjee, K.L. Ebi, Y.O. Estrada, R.C. Genova, B. Girma, E.S. Kissel, A.N. Levy, S. MacCracken, P.R. Mastrandrea, and L.L. White (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, (2014) 1199–265.

  • Google Scholar

• 152

  Available online at: https://www.un.org/africarenewal/magazine/december-2018-march-2019/global-warming-severe-consequences-africa (Accesed March 25, 2025).

• 153

  WMO. State of the Climate in Africa 2024. Geneva: WMO (2025).

  • Google Scholar

• 154

  KovatsRCampbell-LendrumDMcMichelAWoodwardACoxJSH. Early effects of climate change: do they include changes in vector-borne disease? Philos Trans R Soc London Ser B Biol Sci. (2001) 356:1057–68. doi: 10.1098/rstb2001.0894

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 155

  OuagadougouBF. Workshop on strategic planning of area-wide tsetse and trypanosomiasis control in West Africa. FAO, Rome, Italy (2001).

  • Google Scholar

• 156

  CheneauY. A systematic approach to tsetse and trypanosomiasis control. Proc FAO Panel Experts. (1993), 1–3.

  • Google Scholar

• 157

  KgoriPMModoSTorrSJ. The use of aerial spraying to eliminate tsetse from the Okavango Delta of Botswana. Acta Trop. (2006) 99:184–99. doi: 10.1016/j.actatropica.2006.07.007

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 158

  LordJSMthombothiZLagatVKAtuhaireFHargroveJW. Host-seeking efficiency can explain population dynamics of the tsetse fly Glossina morsitans morsitans in response to host density decline. PloS NeglTrop Dis. (2017) 11:e0005730. doi: 10.1371/journal.pntd.0005730

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 159

  KellyRAJakemanAJBarreteauOBorsukMEElSawahSHamiltonSHet al. Selecting among five common modelling approaches for integrated environmental assessment and management. Environ Modell Software. (2013) 47:159–81. doi: 10.1016/j.envsoft.2013.05.005

  • Pubmed Abstract
  • CrossRef
  • Google Scholar