The Canyonlands Dataset is a large-scale, multi-seasonal, multimodal field-robotics dataset for research on autonomous ecological monitoring and restoration of degraded rangelands. Data were gathered with a teleoperated Clearpath Husky ground robot across two field campaigns: May 2022 (early germination) and November 2022 (seed stage), at the Canyonlands Research Center near Monticello, in southeastern Utah, revisiting the same eighteen 10 m x 10 m experimental plots (six each of control, drill-seeded, and connectivity-modifier treatments) in both seasons. For each plot the dataset provides time-synchronized high-resolution forward- and downward-facing RGB imagery, 3D LiDAR point clouds, RTK-corrected GPS, IMU, and LIO-SAM odometry, together with sensor calibration, per-frame GPS and compass-heading uncertainty, and the raw ROS1 recordings. Captured under realistic field conditions (no artificial landmarks, no preplanned paths, extreme visual redundancy, sparse geometric features, and dramatic seasonal change), it is intended to support research in long-term localization and mapping, cross-seasonal data association of individual plants and microsites, and autonomous environmental monitoring in unstructured natural environments where existing cultivated-agriculture datasets are not representative.