Abstract Hybrid landscape models integrate the functional variation contained in high-dimensional remote-sensing data with the spatial organisation of ecological patterns, thereby overcoming the traditional dichotomy between discrete patch mosaics and continuous gradients. The Pixel–Zone–Object (PZO) model provides a coherent framework in which functional pixel traits are aggregated into spatially coherent zones and the emergent objects derived from them. This review shows how four methodological lineages – variance-based partitioning, object-based segmentation, clustering-based functional-space analysis and graph-based regionalisation – converge within the hybrid paradigm. With the ESIS/Imalys Model, we introduce a predictor-free, sensor-robust and globally reproducible model that fully operationalises the PZO concept by algorithmically linking functional homogeneity with spatial contiguity. Validating hybrid models requires functional, spatial and ecological evidence as well as reproducibility analyses. Applications in biodiversity, disturbance ecology, ecosystem condition assessment, and global monitoring highlight the added value of hybrid approaches over traditional classification systems. Hybrid models extend existing paradigms by introducing a data-driven, semantics-free perspective. Advances in AI, multisensor fusion and dynamic graph-based methods will further increase their relevance for landscape ecology and global environmental observation. • Hybrid landscape models are theoretically required because classical patch–mosaic and gradient approaches cannot jointly represent functional heterogeneity and spatial coherence. • The Pixel–Zone–Object (PZO) model is introduced as an integrative framework linking functional pixel signatures, spatially coherent zones, and emergent landscape objects. • Four methodological lineages – variance-based partitioning, object-based segmentation, clustering-based functional-space analysis, and graph-based regionalisation - are synthesised into a unified hybrid paradigm. • The ESIS/Imalys model operationalises the PZO concept in a predictor-free, sensor-independent, and globally reproducible manner without training data or thematic classes. • A multidimensional validation framework is proposed that integrates functional consistency, spatial coherence, ecological plausibility, and reproducibility beyond classical accuracy metrics. • Hybrid models are shown to add value across key applications, including biodiversity analysis, disturbance and degradation monitoring, and global landscape harmonisation.
Lausch et al. (Sun,) studied this question.