The review, published in the journal Discover Geoscience, followed the PRISMA 2020 guidelines and sifted through 262 initial records drawn from Scopus, Web of Science, ScienceDirect, Google Scholar, ResearchGate and citation chasing. After removing duplicates and screening for eligibility, 73 peer-reviewed studies published between 1999 and 2024 made the final cut. Three reviewers independently extracted data from each article, with a fourth reviewer reconciling discrepancies. The questions driving the analysis were deceptively simple: where and when has MEDALUS been applied, how often have researchers altered it, which indicators dominate, how often are outputs validated, and how consistently are sensitivity classes defined?

The answer to the first question reveals a field in rapid expansion. The model’s first documented application appeared in 1999 on the Greek island of Lesvos, and uptake grew slowly at first, with 14 articles, or 19.1 percent of the total, published between 2005 and 2011. But 69.9 percent of all reviewed studies appeared between 2013 and 2024, with a peak of 11 publications in 2018 and another strong showing of 10 articles in 2021. The authors attribute this surge to a convergence of enabling conditions: the opening of the Landsat archive in 2008, expanding GIS capacity in developing countries, and mounting political commitment to desertification monitoring under the United Nations Convention to Combat Desertification and Sustainable Development Goal target 15.3.

Geographically, the picture is strikingly lopsided. The model has now been applied in 23 countries across five continents, but Africa leads with 42.6 percent of applications, driven overwhelmingly by North African nations such as Egypt, Algeria and Morocco that border the Sahara. The Middle East follows with 26 percent, with Iran alone contributing 10 studies, more than any other non-Mediterranean country. Southern Europe accounts for 17.8 percent, a legacy of the model’s origins and the sustained institutional networks it created. Meanwhile, vast dryland regions with severe degradation problems, including Central Asia, East Africa, the Horn of Africa, southern Africa and Australia, have never appeared in the MEDALUS literature at all. South America and East Asia are barely represented, with Brazil’s drought-prone Nordeste and India’s Thar Desert each drawing only a couple of studies despite their scale and vulnerability.

The review’s most fundamental methodological finding concerns time. A full 97 percent of studies, 71 of 73, applied the model to a single year, producing a static snapshot of what is inherently a dynamic process. Only two studies conducted bi-temporal analyses: one examined Iran between 1984 and 1997, the other Egypt between 2003 and 2008. Because the Environmentally Sensitive Area Index aggregates multiple drivers into a single composite number, a one-off assessment cannot distinguish a critically sensitive area that is actually recovering under improved management from a moderately sensitive one that is degrading fast. Without repeated measurements, researchers cannot detect change, attribute it to specific drivers, or evaluate whether policies are working. The authors call this dominance of single-year snapshots the most serious limitation of the entire literature.

Modification, meanwhile, is nearly universal. Only one study in the entire review, the original 1999 handbook itself, used the exact 16 default indicators specified in the MEDALUS manual. Seven more came close, each omitting just one indicator, soil erosion. In total, 62 of the 73 studies altered the original indicator structure, and when the near-default studies are included, 94.5 percent of applications deviated from the original design. At the factor level, 38 studies retained all four original components while 35 introduced changes, adding groundwater, waterlogging, irrigation, erosion or socioeconomic factors, or dropping management altogether. Additions of water-related factors clustered in irrigated hotspots such as Egypt’s Nile Delta and Iran’s Khuzestan Province, where salinization and high water tables drive degradation. Notably, only 12 of the 35 factor-level modifications were explicitly justified with reference to local conditions or supporting literature; the remaining 23 offered no stated rationale.

What drives these choices? The review’s answer is uncomfortable: data availability, not environmental relevance, is the primary force shaping indicator selection. Soil erosion, a core desertification process specified in the original manual, was omitted by roughly 90 percent of studies, apparently because quantifying it is data-intensive. Static geological soil properties such as parent material, texture and depth, easily pulled from legacy global databases, dominated over dynamic indicators like organic carbon or aggregate stability that better reflect current degradation. Landsat remains the dominant satellite source for vegetation indicators thanks to its free archive and long record, even though Sentinel-2, which emerged in the later phase of the review, offers finer resolution. And rainfall, the most used climate indicator at 69.8 percent of studies, went entirely undocumented as to its data source in eight cases.

The management factor emerged as the review’s most damning case study. Policy enforcement, its most common indicator, appeared in 31 articles, yet only two, or 6.5 percent, reported how it was actually measured. Land use intensity, population density and grazing intensity fared little better. In the original framework, management indicators are scored on a subjective one-to-two scale based on expert judgment, and when that subjectivity is compounded by near-total absence of documented data sources, the management factor effectively functions as unvalidated expert opinion carrying the same computational weight as three empirically derived biophysical factors. The authors describe this opacity as the single most critical reporting deficiency they identified.

Validation, the process of checking whether model outputs correspond to reality on the ground, is almost entirely absent. Just 6 of 73 studies, 8.2 percent, performed any independent assessment of their Environmentally Sensitive Area Index outputs, and the rate barely improved over time, rising from 4.5 percent in the 1999 to 2012 period to 9.8 percent in the 2013 to 2024 period. The remaining 67 studies presented sensitivity maps as if the index were free of uncertainty. The six validation efforts fell into four categories: statistical correlation with soil quality parameters such as soil organic matter and cation exchange capacity, multivariate regression including geographic weighted regression, non-parametric testing with Spearman rank correlation and Kruskal-Wallis ANOVA, and spatial accuracy assessment against high-resolution Google Earth imagery. No study used field-based plot measurements, which the authors consider the most robust form of independent validation.

Classification practices add a final layer of confusion. The original MEDALUS scheme defines eight classes, from three grades of critical through three grades of fragile to potential and not threatened. Only 34 studies used it; the other 53.4 percent adopted simplified three-to-five-class schemes with varying names and, crucially, rarely reported the index threshold values separating the classes. A region labeled fragile under one scheme could be moderate or sensitive under another, with entirely different policy implications, and the inconsistency makes cross-study synthesis nearly impossible. The review’s prescription is clear: multi-temporal analyses to track change over decades, statistically calibrated weights with sensitivity testing, transparent documentation of data sources, ground-truth validation, and community-endorsed reporting guidelines that distinguish legitimate local adaptation from modifications that break comparability. The authors also acknowledge their own limitation, the exclusion of non-English publications, which likely understates applications in Chinese- and Arabic-speaking regions. Until these gaps close, the maps guiding land management decisions in some of the world’s most vulnerable drylands rest on foundations that have rarely been tested against the ground they claim to describe.

Subject of Research: A systematic review of global applications of the MEDALUS model for assessing desertification sensitivity from 1999 to 2024

Article Title: A systematic review of the applications of the mediterranean desertification and land use model for assessing desertification sensitivity

Article References: Aliyu, A. O., Youngu, T. T., Azua, S., Garba, S. S., Bawa, S., Akomolafe, E. A., & Bala, A. (2026). A systematic review of the applications of the mediterranean desertification and land use model for assessing desertification sensitivity. Discover Geoscience, 4(1), Article 416. https://doi.org/10.1007/s44288-026-00788-9

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00788-9

Keywords: MEDALUS model, desertification, land degradation, Environmentally Sensitive Area Index, systematic review, drylands, remote sensing, GIS, soil erosion, validation, North Africa, Middle East