As discussed, we've decided to build our Plant Selection functionality as its own self-contained application with both a client (the React component) and a server (the plant-data API). Having this level of isolation allows us to focus purely on the problem of selecting plants, but there's no reason that this couldn't be integrated into a larger code base at a later time.

Our directory structure is roughly as follows:

EveryPlantSelectionApp/
├── server/
│   ├── package.json
|   ├── babel.config.js
│   ├── index.js
|   └── plantData/
│       ├── plantData.js
│       ├── plantData.test.js
|       └── data.json
└── client/
    ├── package.json
    ├── webpack.config.js
    ├── babel.config.js
    ├── app/
    |   ├── index.jsx
    |   └── components/
    |       └── PlantSelectionInput/
    └── dist/
        ├── main.js (bundling target)
        └── index.html

In addition to reducing complexity for us (the programmers) the separation of server and client means that the server-side application (that is, the Plant Selection API) can be run on its own distinct server if necessary, while the client can be served up statically from a CDN, requiring only the server-side's address in order to access its REST API.