Climate and Land-Use Change Reshape Invasion Risk in Tanzania’s Lake Zone, Study Finds

In the lakeside regions of Mara and Simiyu in northern Tanzania, three of the world’s most notorious invasive plants are quietly redrawing the ecological map. A new modelling study published in One Ecosystem suggests that predicting where these invaders will strike next requires far more than temperature and rainfall data alone. By combining climate projections with land-use and land-cover change for the first time at a subnational scale in this biodiversity-rich corner of East Africa, researchers have revealed that conventional climate-only models may substantially overestimate the true extent of invasion risk.

The study focused on three species with devastating ecological and economic footprints. Chromolaena odorata, a fast-growing perennial shrub from tropical America, smothers forests, grasslands and farmland while suppressing native vegetation and reducing crop yields. Lantana camara, introduced as an ornamental, has become one of the planet’s most widespread invaders, forming dense thickets in disturbed habitats. The aquatic Pontederia crassipes, commonly known as water hyacinth, blankets freshwater systems in thick floating mats that choke fisheries, block navigation and disrupt water supplies. All three rank among the most damaging invasive plants worldwide, and all three are already established in the Lake Victoria Basin.

To map where these species could thrive now and in the decades ahead, the research team employed Maximum Entropy modelling, or MaxEnt, a widely used species distribution technique that estimates suitable habitat from presence-only occurrence records. The scientists assembled hundreds of filtered occurrence points from field observations, GBIF and iNaturalist, applying a one-kilometre grid thinning procedure to eliminate spatial clustering and bias-weighted background sampling to correct for uneven survey effort across the landscape. Environmental predictors included seven biologically meaningful bioclimatic variables from the WorldClim database alongside fractional land-cover data derived from high-resolution European Space Agency satellite imagery.

The modelling framework was deliberately rigorous. Future projections drew on three distinct CMIP6 global climate models, ACCESS-CM2, MIROC6 and MRI-ESM2-0, run under four Shared Socioeconomic Pathways spanning low to very high emission futures, from SSP1-2.6 to SSP5-8.5, for the 2030s, 2050s, 2070s and 2090s. By averaging ensemble predictions across climate models and quantifying the coefficient of variation in suitable area, the researchers could explicitly measure how uncertainty grows under intensifying warming. Model performance was evaluated with multiple complementary metrics, including the area under the receiver operating characteristic curve, the Continuous Boyce Index and omission rates, rather than relying on any single statistic.

The results delivered a striking divergence between terrestrial and aquatic invaders. Under climate-only projections, Chromolaena odorata and Lantana camara retained moderate suitability under the low-emission scenario, but their highly suitable areas progressively fragmented and contracted under higher-emission pathways, particularly after 2050. By late century under SSP5-8.5, some climate models projected a near-complete loss of suitable conditions for these shrubs, while others retained only scattered residual patches. Temperature variability emerged as the dominant driver: annual temperature range alone explained nearly half of the modelled distributions of both terrestrial species, with temperature seasonality and precipitation seasonality playing substantial supporting roles.

Water hyacinth told a fundamentally different story. Pontederia crassipes maintained broad climatic suitability across virtually all scenarios and time periods, with late-century projections frequently showing relatively uniform suitability and almost no unsuitable zones. This reflects the plant’s dependence on warm, moist conditions, driven in the models by temperature seasonality and precipitation of the warmest quarter. The researchers caution, however, that climatic suitability does not equal realised occupancy. The species remains ecologically confined to permanent freshwater habitats, and its actual spread depends on nutrient enrichment from agricultural runoff, hydrological connectivity and water chemistry, factors that climate models alone cannot capture.

The study’s most consequential finding emerged when land-use projections were layered onto the climate scenarios. Suitable habitat estimates for the two terrestrial species were substantially reduced and accompanied by escalating uncertainty. For Chromolaena odorata, suitable habitat covered roughly 55 to 65 percent of the region under the low-emission pathway, but inter-model uncertainty climbed to a coefficient of variation of approximately 65 to 77 percent by 2090 under high emissions. Lantana camara fared worse still, contracting to small patches in the north-central landscape, with uncertainty exceeding 85 to 90 percent by century’s end. The discrepancy demonstrates that climate-only models can overestimate potential niche space by ignoring land-use constraints, and that invasion risk is fundamentally shaped by the interaction between broad climatic limits and local habitat availability.

Land-cover analysis revealed that croplands contribute the largest share of effectively suitable habitat for both terrestrial invaders, confirming their affinity for disturbed, human-modified landscapes. Selection ratios indicated non-random associations, with both species favouring forest and shrubland while avoiding open water. Roads, grazing lands, agricultural expansion and fallow fields all create the bare soil, increased light and repeated disturbance that these plants exploit, while livestock and transport networks accelerate seed dispersal. Meanwhile, niche overlap analysis showed moderate to high climatic similarity between the two terrestrial invaders, with Schoener’s D ranging from 0.45 to 0.81, suggesting the species could increasingly converge geographically and potentially co-invade disturbed areas, compounding pressure on native ecosystems under some future scenarios.

The implications for land managers and policymakers in Tanzania are concrete. The authors recommend concentrating surveillance and early-response efforts on agricultural zones, transport corridors and riparian systems, where models consistently agree on high suitability, since eradicating invasions early is far cheaper than fighting established populations. For water hyacinth, continued monitoring of Lake Victoria’s catchment, rivers and wetlands should be paired with nutrient-reduction strategies targeting farm runoff and wastewater. The team also advises treating late-century projections as risk estimates rather than precise predictions, given the mounting model disagreement under high emissions, and prioritising regions where multiple climate models converge.

The study is not without limitations. The water hyacinth model rests on only 21 spatially filtered occurrence records, forcing the researchers to constrain model complexity and interpret those projections cautiously. The analysis relied on a single modelling algorithm, and dispersal limitation, propagule pressure and human-mediated transport were not explicitly incorporated, meaning projected distributions represent potential suitability rather than certain future invasion. Even so, by jointly modelling three species, integrating land-use dynamics with climate futures and quantifying uncertainty at a regional scale, the research offers one of the most complete invasion risk assessments yet produced for East Africa, and a compelling argument that realistic biosecurity planning must look beyond the thermometer to the transforming landscape itself.

Subject of Research: Modelling habitat suitability of invasive alien plant species under climate and land-use change in Tanzania

Article Title: Modelling potential habitat suitability of three invasive alien plant species under projected climate scenarios and land-use/land-cover change in the Lake Zone of Tanzania

Article References: Mtenga, N., Kasongi, N., Lehmann, J., & Buttschardt, T. (2026). Modelling potential habitat suitability of three invasive alien plant species under projected climate scenarios and land-use/land-cover change in the Lake Zone of Tanzania. One Ecosystem, 11, Article e191980. https://doi.org/10.3897/oneeco.11.e191980

Image Credits: AI Generated

DOI: 10.3897/oneeco.11.e191980

Keywords: invasive alien species, MaxEnt, species distribution models, climate change, land-use change, Tanzania, Lake Victoria Basin, water hyacinth, Lantana camara, Chromolaena odorata, habitat suitability, CMIP6