This study aimed to develop a learning progression (LP) for botanical literacy and assess the impact of a garden-based education course on Thai pre-service science teachers (n = 49). The inquiry- and community-based education course was taught over a 15-week period. To measure botanical literacy, a 50-item multiple-choice test was employed. A hypothetical model of the construct was proposed and validated using a dichotomous Rasch model. Results indicated that the model fits the data well. Thus, the hypothetical model was confirmed. The validated botanical literacy comprised conceptual and procedural domains. Each domain had several core ideas and subordinate concepts located on a linear construct and arranged by increasing difficulty. For example, in the conceptual domain, the core ideas arranged from the easiest to most difficult were plant diversity, plant morphology, and plant eco-physiology. The botanical literacy of the pre-service teachers was measured before and after the instruction. To measure the learning gain, we fixed the pre- and post-item difficulty and estimated the post-instruction person ability. The means of pre- and post-instruction person ability were compared using the Welch t-test. We found that learning gain was statistically significant (p < 0.01). This study discusses implications of LP for botanical literacy in biology education.