Stacked suitability times density: where composite risk models earn their maps

Question: How have spatially explicit niche and risk models for mosquito-borne zoonoses incorporated host and vector habitat suitability alongside human and livestock density to map transmission risk?

Answer

Across the retrieved papers, the dominant architecture is a parallel overlay: independent species distribution models for each vector and each reservoir or amplifying host, produced by different algorithms, then combined — usually multiplicatively — with gridded human and livestock density to yield a "risk" surface [11] [12] [19]. Taken together, these papers support a narrower and less comfortable position than the proliferation of such maps implies: the overlay is a defensible way to map hazard and exposure separately, but it is an assumption-laden and largely unvalidated way to map transmission. Three findings in this corpus push in that direction — that reservoir-host layers often add little where hosts are ubiquitous, so the risk geography is effectively vector-driven [12]; that human and livestock density are not monotone amplifiers of risk because contact rates and infrastructure intervene [25] [28]; and that the demonstrable gains from adding biotic or anthropogenic layers appear mainly where the composite is scored against independent occurrence data [17] [21]. The productive frontier in this literature is therefore not more layers but better-specified combination operators and honest validation.

Vector and host suitability surfaces are stacked and multiplied by density by default

The template is explicit in the West Nile virus work: six correlative models were fitted per Culex vector, the best selected on performance criteria and checked against mechanistic models and known distributions, then combined with correlative suitability for representative avian hosts and with equine and human population data to predict risk of occurrence [11]. The continental extension of that framework made the overlay components legible — Culex suitability, Passerine host suitability, human density, horse density — and projected the product to 2100 under best- and worst-case pathways [12]. The Japanese encephalitis assessment in Australia formalised the same logic at national scale, collating human, mosquito, domestic pig, feral pig and sentinel chicken detections and combining them with previously published host and vector habitat suitability predictions and estimated human, domestic pig and feral pig densities in what the authors call a multipart spatial risk assessment, distinguishing the realised from the potential niche [19]. Comparable layer-stacking with demographic denominators produces "population at risk" counts for inter-epidemic Rift Valley fever after validation against human serology [16] and for Aedes-borne viruses in Europe [4] [27].

Where competent hosts are ubiquitous, the host layer degenerates into a near-uniform multiplier and vector suitability carries the map

The European WNV overlay is unusually candid about its own variance decomposition: risk maps show areas of higher and lower risk that are "predominantly driven by vector suitability as avian hosts are widespread" [12]. This is a structural problem for host-suitability layers generally — a reservoir model whose output is near-saturated across the domain contributes scale but not geography. The inverse strategy, mapping the host side first and treating human cases as the validation target, produces genuinely heterogeneous structure instead: a traits-based estimate of WNV seroprevalence across European bird species yields a Bird Risk Index with strong spatial heterogeneity that is positively associated with the number of years with notified human cases at NUTS level [9]. Whether the host layer is informative therefore depends on host endemism and range restriction, not on modelling effort.

Nested, product and mechanistic structures generate different risk geographies than parallel overlays

Where virus occurrence models incorporated vector suitability as a nested predictor rather than as a co-equal layer, historical attribution separated land-use effects on absolute vector suitability from climatic effects on the timing of transmission risk [15]. Dengue risk in California was specified as the product of three necessary components — vector presence, temperature suitability, and introduction via travel-associated cases — with the risk threshold calibrated to locations and months where local transmission actually occurred, so that low risk arises from any single failing component rather than from an average [24]. A mechanistic R0-based metric combining temperature dependence with vector species distributions, then overlaid on human and horse densities, showed that which host stratum is used determines the answer: horses concentrate risk in south-east England around Newmarket, humans in London, and age-adjusting the human layer relocates elevated risk to coastal East Anglia [22]. Multi-host vectorial-capacity formulations go further, showing that interventions on dead-end hosts can redirect bites to reservoirs and raise transmission [23].

Human and livestock density proxies mis-specify exposure because contact, infrastructure and behaviour decouple density from biting

Florida's mismatch between high climate-based R0 and minimal dengue transmission was reconciled by estimating human–mosquito contact rates well below endemic-setting or laboratory values, with population density and the built environment moderating contact [25]. In West Africa, a socio-economic filter proxied by nighttime lights produced an "infrastructure-driven socio-economic shield" that decouples high reservoir hazard in peri-urban fringes from human incidence in dense urban cores [28]. Density can also amplify through concentration rather than contact: drought reduced mosquito abundance yet increased WNV infection rates, attributed to mosquitoes and amplifying birds concentrating around limited water [14]. Where density is modelled as a within-suitability predictor rather than a separate exposure layer, human population is simply one more covariate alongside climate, land use and surveillance data [13], and imperviousness emerges as the strongest single predictor of Culex pipiens distribution [3].

Composition and assemblage descriptors outperform density or richness layers for the biotic terms

Tick-borne Crimean-Congo haemorrhagic fever provides the clearest ablation: climate-only models over-predicted range, tick-distribution-only models underperformed in Africa, and only integrating human-biting tick distributions, livestock density and host chorotypes captured over 90% of known occurrences [21]. For WNV, non-passerine richness was negatively and passerine richness positively associated with mosquito circulation after adjustment [5], while in Italy the dilution effect appeared in ecologically intact landscapes and reversed into amplification in agricultural ones [2]. Where different risk components were decomposed, Culex abundance tracked impervious surface, infection prevalence tracked turf grass, and turf itself correlated with avian reservoir competence — three different land-cover drivers for three components of one transmission system [6]. Avian demographic timing, not just abundance, changes infected-mosquito numbers in mechanistic models [10], and scenario work that varied host and vector distributions jointly found outbreak risk diverging between locations precisely because of spatial variation in host and vector abundance [7].

Measurable gains from anthropogenic layers appear only where the composite is scored against independent occurrence data

The one direct quantification of added value in this corpus comes from Ebola spillover: combining orthoebolavirus habitat suitability with built-environment and anthropogenic drivers raised the Boyce index from 0.78 to 0.97 against historical spillover locations [17]. Ensemble approaches claim higher accuracy and transferability than individual algorithms under independent block validation [3] and are being used to flag suitable habitat in administrative regions absent from the source atlas [8].

Conflicts and gaps

Whether host layers earn their place is unresolved and system-dependent: one European WNV overlay concludes avian hosts are too widespread to structure risk [12], while Crimean-Congo and Ebola work finds biotic and anthropogenic layers indispensable and climate-only models misleading [21] [17]. The sign of the biodiversity effect is contested within one pathogen [5] [2], so a host-suitability layer built on richness has no agreed direction. Human density is simultaneously reported as risk-elevating at intra-urban scale [20] [26] and as protective through infrastructure [28]. No retrieved study compares combination operators — product, weighted sum, minimum-constraint, bivariate classification, nested predictor — on the same inputs, even though bivariate overlap is used as a reporting device elsewhere [1]. Most composite risk maps are never scored against held-out human or veterinary case data; the exceptions validate a single index rather than the full overlay [9]. Spatial-scale mismatch between 1 km suitability surfaces, NUTS- or county-level density and case data goes largely unaddressed [18].

What would change the answer

A head-to-head benchmark applying every candidate combination operator to identical vector and host suitability inputs and scoring the resulting surfaces against held-out case data with proper scoring rules would settle whether the overlay is defensible [24]. Replacing density-as-contact with empirically estimated human–mosquito contact rates would test the exposure term directly [25]. Case-based spatiotemporal models that explicitly model reporting heterogeneity and autocorrelation provide a more honest reference truth against which overlays should be scored [18], and trait- or chorotype-based host layers rather than density or richness should be the biotic default [9] [21]. Prospective surveillance in the regions these models flag as suitable but unsampled would test transferability rather than fit [8] [28].

References

[1] Poongavanan, J., Dunaiski, M., Dor, G., Kraemer, M. U. G., Giovanetti, M., Lim, A., Brady, O. J., Gutierrez, B., Baxter, C., Carrera, J. P., Fonseca, V., Alcantara, L. C. J., De Oliveira, T., Tegally, H.. "Spatiotemporal disease suitability prediction for Oropouche virus and the role of vectors across the Americas". medRxiv. 2025-06-16T00:00:00.000Z. URL: https://doi.org/10.1101/2025.02.28.25323068

[2] Marcolin, L., Ceci, N., Gobbo, F., Montarsi, F., Chiarello, G., Dorigatti, I., Di Marco, M.. "West Nile virus response to mosquito and avian biodiversity in rural environments". bioRxiv. 2026-06-27T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.26.734801

[3] Marcolin, L., Zardini, A., Longo, E., Caputo, B., Poletti, P., Di Marco, M.. "Mapping the habitat suitability of Culex pipiens in Europe using ensemble bioclimatic modelling". bioRxiv. 2025-03-18T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.17.643724

[4] Serres, K., Gambaro, F., Pietroiusti, R., Da Re, D., Vincenti-Gonzalez, M. F., da Silva Candido, D., Arsevska, E., Jacques de Dixmude, A., Quesada-Chacon, D., Mengel, M., Paprotny, D., Klitting, R., Marsboom, C., Thiery, W., Ghisbain, G., Erazo, D., Dellicour, S.. "Escalating human exposure to tropical mosquito-borne viruses in Europe". bioRxiv. 2025-12-03T00:00:00.000Z. URL: https://doi.org/10.64898/2025.12.02.691392

[5] Wang, Y., Calzolari, M., Calvi, G., Cox, V. M., Angelini, P., Dottori, M., Wint, W., Jahn, S., Marini, G., Dorigatti, I.. "Association of avian biodiversity and West Nile virus circulation in Culex mosquitoes in Emilia-Romagna, Italy". bioRxiv. 2025-08-22T00:00:00.000Z. URL: https://doi.org/10.1101/2025.08.18.670801

[6] Mackay, A. J., Yan, J., Kim, C.-H., Cara, C., Chesser, L., Alfaro, E., Avara, M., Ward, M., Magle, S., Murray, M. H., Stone, C.. "Infection and host-feeding patterns of West Nile virus vectors varies by urban greenspace composition". bioRxiv. 2025-03-28T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.24.645034

[7] de Wit, M. M., Dellar, M., Geerling, G., Boelee, E., De Jong, M. C. M., ten Bosch, Q.. "Implications of future change scenarios for mosquito-borne disease transmission in the Netherlands". bioRxiv. 2025-07-14T00:00:00.000Z. URL: https://doi.org/10.1101/2025.07.09.663836

[8] Barreaux, A. M. G., Wangu, H., Coulibaly, B., Berte, D., Coulibaly, D. K., Abdel-Rahman, E. M., Mongare, R., Adingra, P., Kalo, V., Gachoki, S., Boulange, A., Gimonneau, G., Thevenon, S., Cecchi, G., Solano, P., Kaba, D.. "Spatial distribution and habitat suitability of tsetse (Glossina spp.) in Cote dIvoire: An ensemble modeling approach to support targeted disease control". bioRxiv. 2026-08-26T00:00:00.000Z. URL: https://doi.org/10.64898/2026.08.21.746192

[9] Bastard, J., Metras, R., Durand, B.. "Mapping the Bird Risk Index for West Nile virus in Europe and its relationship with disease occurrence in humans". medRxiv. 2026-01-06T00:00:00.000Z. URL: https://doi.org/10.1101/2025.02.24.25322781

[10] Fesce, E., Cattaneo, E., Marini, G., Rosa, R., Lelli, D., Cerioli, M. P., Ilahiane, L., Rubolini, D., Chiari, M., Ferrari, N.. "Beyond competence: a mechanistic model of avian demographic drivers in West Nile virus dynamics". bioRxiv. 2026-06-18T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.15.732287

[11] Withers, A. J., Croft, S., Budgey, R., Warren, D., Johnson, N.. "Using correlative and mechanistic species distribution models to predict vector-borne disease risk for the current and future environmental and climatic change: a case study of West Nile Virus in the UK.". bioRxiv. 2024-10-22T00:00:00.000Z. URL: https://doi.org/10.1101/2024.09.12.612656

[12] Withers, A. J., Croft, S., Budgey, R., Warren, D. A., Johnson, N.. "Predicting West Nile Virus risk across Europe for the current and future conditions". bioRxiv. 2025-03-10T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.04.640552

[13] Siddiqui, T., Malysheva, N., Hartner, A.-M., Butyrin, S., Parreira, D., Genger, J.-W., Irrgang, C.. "Suitable seasons: Global monthly habitat suitability for the arbovirus vectors Aedes aegypti and Aedes albopictus in 1975-2024". bioRxiv. 2026-04-18T00:00:00.000Z. URL: https://doi.org/10.64898/2026.04.17.719149

[14] Sambado, S., Sipin, T. J., Rennie, Z., Larsen, A., Cunningham, J., Quandt, A., Sousa, D., MacDonald, A. J.. "The paradoxical impact of drought on West Nile virus risk: insights from long-term ecological data". bioRxiv. 2025-01-24T00:00:00.000Z. URL: https://doi.org/10.1101/2025.01.21.634155

[15] Holle, V., Klitting, R., Kabisch, N., Zurell, D.. "Past and future phenology changes of zoonotic vector-borne diseases under climate and land-use change". bioRxiv. 2026-06-10T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.07.730142

[16] Eskew, E. A., Clancey, E., Singh, D., Situma, S., Nyakarahuka, L., Njenga, M. K., Nuismer, S. L.. "Projecting climate change impacts on inter-epidemic risk of Rift Valley fever across East Africa". medRxiv. 2025-02-10T00:00:00.000Z. URL: https://doi.org/10.1101/2025.02.08.25321747

[17] Moir, M., Tegally, H., Moges, D. M., Dor, G., Poongavanan, J., Baxter, C., Lessells, R. J., Kraemer, M. U. G., Judge, C., Gutierrez, B., Ebengo, D. M., Kinganda Lusamaki, E., Placide, M., Muyembe-Tamfum, J.-J., Bisimwa Nachega, J., de Oliveira, T., Mavian, C. N.. "Integrating ecological and anthropogenic risk identifies emerging Ebola spillover hotspots". medRxiv. 2026-07-29T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.25.26358924

[18] Bastard, J., Assaad, C., Marti, R., Tran, A., Metras, R., DURAND, B.. "Short- and long-term causes of West Nile virus risk in Europe: a spatiotemporal model accounting for under-reporting". medRxiv. 2026-08-23T00:00:00.000Z. URL: https://doi.org/10.64898/2026.08.20.26360902

[19] Harrison, L. E., Duncan, D. H., Flegg, J. A., Price, D. J., Kurucz, N., Tolosa, X., Neville, P. J., Jansen, C. C., Potter, A., Brockway, C., Miller, K. L., Hobby, M., Burns, V., Glasgow, K. M., Draper, A. D., Schmidt, J., Weerasinghe, G., van den Hurk, A. F., Williams, D. T., Doggett, S. L., Golding, N., Shearer, F. M.. "Geospatial assessment of Japanese encephalitis virus infection risk in Australia". medRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.15.26365607

[20] Santos Vega, M., Diuk-Wasser, M., Kache, P.. "Socio-environmental typologies of dengue risk at multiple spatial scales of the urban landscape". medRxiv. 2026-03-26T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.24.26349225

[21] Estrada-Pena, A., Carnell, O., Wijburg, S. R., Holding, M., Sprong, H.. "Crimean-Congo Haemorrhagic Fever Virus Circulates within Broad Ecological Networks of Ticks and Vertebrates". bioRxiv. 2025-11-21T00:00:00.000Z. URL: https://doi.org/10.1101/2025.11.20.689440

[22] de Klerk, J. N., Pilgrim, J., Baylis, M., Blagrove, M., Medlock, J. M., Vaux, A. G. C., Bruce, R. C., Folly, A. J., Sedda, L.. "Quantifying the current and future risk of West Nile virus and Usutu virus in humans and horses across England and Wales using an R0-based suitability metric". medRxiv. 2026-10-06T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.04.26364722

[23] Fairbanks, E. L., Baylis, M., Daly, J., Tildesley, M.. "Quantifying vector diversion effects in zoonotic systems: A modelling framework for arbovirus transmission between reservoir and dead-end hosts". bioRxiv. 2025-07-31T00:00:00.000Z. URL: https://doi.org/10.1101/2025.07.24.666561

[24] Couper, L., Sipin, T. J., Sambado, S., Rennie, Z., Shanebeck, K. M., Lyberger, K. P., Collender, P. P. A., Ngo, V., Remais, J. V., MacDonald, A. J.. "Climate warming and urbanization may expand dengue transmission risk in California". bioRxiv. 2025-11-16T00:00:00.000Z. URL: https://doi.org/10.1101/2025.11.14.688544

[25] Rysava, K., Morrison, A., Paz-Bailey, G., Stanek, D., Thayer, M., Zimler, R., Johansson, M.. "Lower than expected effective human-mosquito contact may regulate dengue virus transmission in high-suitability low-incidence environments: A plausible explanation for limited dengue outbreaks in Florida, U.S.". medRxiv. 2025-12-11T00:00:00.000Z. URL: https://doi.org/10.64898/2025.12.09.25341886

[26] Marcílio Ferreira dos Santos, Andreza dos Santos Rodrigues de Melo. "Hierarchical Bayesian Modeling of Dengue in Recife, Brazil (2015-2024): The Role of Spatial Granularity and Data Quality for Epidemiological Risk Mapping". arxiv. 2025-10-16T00:00:00.000Z. URL: https://arxiv.org/abs/2510.13672

[27] Ryan, S. J., Lippi, C. J., Johnson, L. R., Meredith, J.. "Global shifts in thermal suitability and population at risk for dengue transmission by Aedes spp. mosquitoes under CMIP6 scenarios". medRxiv. 2026-07-11T00:00:00.000Z. URL: https://doi.org/10.64898/2026.07.02.26357126

[28] Simons, D.. "The Socio-economic Shield Limits Lassa Virus Spillover in Urban West Africa". medRxiv. 2026-04-17T00:00:00.000Z. URL: https://doi.org/10.64898/2025.12.17.25342147

How this survey was made

This survey answers one question: “How have spatially explicit niche and risk models for mosquito-borne zoonoses incorporated host and vector habitat suitability alongside human and livestock density to map transmission risk?”

We keep our own collection of recent preprints and refresh it every day from bioRxiv, medRxiv, arXiv, PsyArXiv, SocArXiv and ChemRxiv. For each survey, we search that collection for recent work and keep the strongest matches: papers that use the same terms as the question, and papers on the same topic in different words.

The text was written from the titles and abstracts of those papers; the full texts were not read. Statements that draw on a particular paper are marked with a number in brackets. The reference list just above is the 28 papers whose work this text draws on, chosen from the 134 papers those searches returned; the rest of the shortlist is not printed. It is not a complete review of everything published on the question.

Treat this as a starting point and a reading list, not a peer-reviewed review. The searches behind this survey ran on October 22, 2026 and covered work from the previous two years.