Novel ecosystems are often defined as no-analog communities consisting of new combinations of species that assemble under new abiotic conditions. In the Anthropocene, novel systems differ from the historical state as a result of human influences where self-organizational processes prevail to make these ecosystems unlikely to revert to their historical structure/function. Degraded shrub-dominated ecosystems that develop following livestock overgrazing and drought in arid and semiarid grasslands are a potential example of novel ecosystems. Although the self-organization and stability of desertified ecosystems have been well-studied, these ecosystems have not been tested as novel ecosystems. Our objective was to test novel ecosystem concepts and hypotheses across complex landscapes at a 78,000 ha research site established in 1912 to examine the causes and consequences of desertification. We integrated long-term vegetation maps dating to 1850 with USDA data on grazing, drought, and species composition (1915 onward), and a suite of LTER data on biotic and abiotic patterns and processes (1982-present) for the Jornada Basin LTER-USDA site in southern New Mexico, USA. We hypothesized that degraded ecosystems dominated by different grass-shrub assemblages represent novel ecosystems if they meet each of four criteria: 1. different from past states, 2. human-driven, 3. self-organized, and 4. irreversible. 

All four criteria were met for some locations for each degraded shrubland type (mesquite, creosotebush, tarbush), yet not all locations were identified as novel. Because these shrub species are native to the region, an expansion of a species beyond its native habitat is needed for novelty to be recognized. Our synthesis shows that this expansion was initiated by human drivers, but then for some locations a change in soil properties occurred through time that led to a shift in species dominance and composition and further changes in soil properties. These shifts were maintained by plant-soil feedbacks and self-organization to result in novel ecosystems. On other locations, a localized spatial redistribution of soil and nutrients among grasses and shrubs within the ecosystem occurred that led to a state change, but not a novel ecosystem. Thus, arid landscapes are complex mosaics of historic grass remnants, degraded yet native shrublands, and degraded and novel shrublands. The dynamic soil-geomorphic template interacting with species characteristics is another layer of complexity that needs to be accounted for when predicting future dynamics of arid landscapes that contain novel ecosystems.