Rapid urbanization has been one of the primary drivers of ecosystems change in the Guangdong-Hong Kong-Macau Bay Area (the Great Bay Area). To understand these spatio-temporal dynamics, it is important to simulate ecosystem trend and patterns during regional land use optimization, preventing and controlling the ecological risks of urbanization. Based on land use-land cover (LULC) data for the years 2000, 2005, 2010, 2015 and 2018, this paper analyzed the evolution of ecosystem types and landscape patterns in this region. A CA-Markov model was used to simulate ecosystem pattern in 2025. The results showed that: (1) the forests, farmland and urban land were the main ecosystem types in the Greater Bay Area in 2018, accounting for 53.99%, 22.67% and 14.51% of the total area, respectively. (2) From 2000 to 2018, the area of farmland, forest, and wetland decreased by 1983 km², 740 km² and 278 km² respectively, and the area of urban and grassland increased by 2896 km² and 103 km², respectively. The growth in urban areas came from the conversion of surrounding farmland, woodland and wetland. Due to poor management, portions of woodland degraded to grassland, leading to the growth of the latter. The change rate of ecosystem types in the Greater Bay Area had obvious periodical characteristics. The fastest period of urban expansion was from 2000 to 2005, leading to a rapid decline in farmland. The rate of urban expansion and farmland reduction slowed after 2005.The speed of forest reduction accelerated in 2000-2015 and slowed down in 2015-2018. The grassland area increased from 2000 to 2015, with a slow growth rate from 2000 to 2010 and a rapid growth rate from 2010 to 2015, then decreased from 2015 to 2018. From 2000 to 2018, the wetland area continued declining with the same speed. (3) Under the effects of demographic urbanization and LULC urbanization, the landscape pattern changed significantly. Fragmentation and separation increased, and the diversity and evenness of the landscape decreased. The Contagion index (CONTAG) increased and the Division index (DIVISION) increased, indicating that the fragmentation and heterogeneity of the landscape increased. The largest patches of farmland and forestland were destroyed, and the Number of Patches index (NP) increased. Urban patches showed a spreading growth, while the urban landscape's degree of fragmentation and separation fell. (4) The ecosystem pattern in 2025 was simulated based on a CA-Markov model, and the accuracy test was carried out by comparison with the actual data in 2015; the Kappa coefficient was 0.94. Compared with 2018, the urban area is projected to increase by 609 km², extending outward along the larger urban patches, mainly in Guangzhou, Dongguan and Foshan prefectures. The area of farmland and forest ecosystem is projected to decrease by 309 km² and 316 km², respectively. The areas of grassland ecosystem and wetland ecosystem are projected to remain unchanged. Based on the change and simulation of ecosystem patterns in the Greater Bay Area, it was found that the farmland protection red line, ecological protection red line and the protection of important ecological space integrity (all national policies) played important roles in reducing the ecological risk of urbanization in a rapidly urbanizing region.