I can remember when I was growing up in a community in rural South India, we had a garden growing bananas, tomatoes, spinach, lemons, and mangoes. We took care of the garden as a group activity. Even though we were growing these for the communal use, sometimes we were very much invested with the ones that we planted. Everyday after school, we would rush to the garden and see whose banana trees or tomato plants were growing fast. We often wondered why some plants grew faster than others and were also curious to know why some plants produced more fruits than others.
For some children, a school garden, or any other garden for that matter, can generate a sense of wonder and curiosity. So many activities can be developed to further enhance our classroom activities by connecting classroom mathematics with school-based gardens. For instance, we can study about rates of change, inventing new functions, visualizing arts and patterns, productivity, marketing, etc. These sorts of activities offer an alternative to or an extension to existing curriculum.
I think that one of the issues with garden-based learning could be attributed to difficulties in assessing or obtaining measurable outcomes. For those of us interested in assessment tools in relation to school-based garden, there are resources out there to do just that. For instance, Cornell University provides an evaluation toolkit, in addition to many lesson plans and activities.