Coexisting vegetation types in tropical landscapes can respond in contrasting ways to rainfall, despite being in the same climatic envelope. Understanding such heterogeneity in vegetation-rainfall interactions is key for predicting how ecosystems might respond to future environmental changes. Here we test whether temporal coupling between vegetation greenness and rainfall is a good indicator of ecosystem state in the landscape. For this, we study a well-preserved landscape of the Brazilian Cerrado that is formed by mosaics of contrasting ecosystems, including savannas, dry forests and gallery forests. First, we correlate the time-series of rainfall and vegetation greenness to quantify their coupling for each vegetation type. We then compare vegetation-rainfall coupling with other state variables, such as local-scale vegetation structural and functional traits, as well as differences in environmental conditions in which these vegetation types exist. Coexisting vegetation types are set in contrasting local-scale environmental conditions and have distinct responsiveness to rainfall. Commonly used structural and functional state variables, such as tree cover and tree height, do not depict such marked differences between the vegetation types, particularly for gallery and dry forests. Dry forests have the strongest coupling and decrease their greenness during dry seasons, reflecting vegetation deciduousness on nutrient-richer soils. In contrast, gallery forests increase their greenness during the dry season, when direct radiation peaks, likely due to perennial access to groundwater. Savannas are less responsive to rainfall and have a more stable greenness throughout the year. Our findings suggest that heterogeneity in local abiotic conditions contribute to determining both vegetation distribution and ecosystem states in these tropical savanna landscapes. Changes in these conditions as a result of climate and land-use changes will likely alter the distribution of vegetation types in the future. Our functional metric may thus be useful for assessing future responses of tropical ecosystems to changes in precipitation.