Towns in ecologically sensitive areas (ESAs) serve as crucial strategic fulcrums for China's county-level development and rural revitalization, yet they also encounter conflicts within the human-environment system, primarily manifested as a trade-off between development and conservation. In the spatial dimension, towns in ESAs present remarkable spatial heterogeneity. The macro-control paradigm has become incompatible with the differentiated trends of the micro-development pattern and the demand for refined governance in towns. In the development dimension, towns demonstrate diverse spatio-temporal evolution trajectories in their sequential progress. Nevertheless, existing studies or practices still adhere to the static control measures. Excessive external constraints have constricted the towns' sustainable development in ESAs. To address the limitations of the current macro-static governance paradigm, this research develops a micro-development pattern recognition framework for towns in ESAs and suggests adaptive management and control strategies. An empirical analysis was conducted in the Baiyangdian region. The main conclusions are as follows: (1) The geographical variations in the ecosystem service structure of towns in the Baiyangdian region are pronounced. Furthermore, The high-value areas of ecosystem services are basically consistent with the pattern of Baiyangdian Wetland, and extend in a zonal pattern along the river system towards the west; (2) Transportation factors exert a crucial influence on the development dynamics of towns in the Baiyangdian region. This includes the distribution of the construction extent, which shows a decreasing trend from the transportation trunk lines to both sides, and the high-probability areas of construction are in proximity to the transportation trunk lines; (3) The micro-development pattern of towns was identified from the aspects of construction probability and construction degree. The proportion of advanced construction units is the highest, clustering around towns and expanding into the hinterland, while the proportion of potential construction units is the lowest, embedded within towns and wetlands; (4) Rainwater storage and regulation services, water conservation services, and water quality purification services will face relatively severe threats in the next 30 years, with Xiongzhou Town having the largest ecological risk patch area, while Santai Town and Xiaoli Town having the fastest expansion rate of ecological risk patches. Adaptive zoning management and control strategies for towns in the Baiyangdian region are proposed, based on eight types of micro-development pattern units (further divided by the comprehensive ecosystem service level). The study can provide a reference for optimizing the human-land system and related planning decisions in towns in ESAs.