The past couple of weeks I’ve been learning about biocultural restoration in Hawai’i, and I have some article ideas I’ve been stewing on. But in the meantime, I wanted to write about a different topic that I’ve been sitting on for a few months now.
In this article, I share something I’ve learned about: namely, sourcing images of plants from GBIF (the Global Biodiversity Information Facility). [side note: GBIF also has animal/fungi/bacteria/etc. records] While this topic isn’t directly related to restoration per se, I’ve found in my own work and in writing blog articles for this series that being able to find good plant images is useful for communicating science.
Challenges of Sourcing Plant Images
In my experience, I’ve found that sourcing images of specific plant species is not always easy. The key challenges I’ve encountered are finding images that are high-quality, legally usable, and reliable in terms of species identification. While search engines now allow image filtering by license, the image results don’t always correspond to the exact species (e.g., they may turn up other species in the same genus) And while websites from cooperative extension agencies, herbaria, and the like can have high-quality, reliable images, they are often copyrighted.
Enter GBIF, a network which–among other things–aggregates species occurrence records (and images!) from a range of sources, including crowdsourced websites (e.g., iNaturalist), herbaria, natural history museums, and other similar platforms. Every image on GBIF has some sort of Creative Commons license, which means they can all be used for noncommercial purposes. What’s more, many of these images even include suggested attribution language that one can simply copy and paste, to easily give credit to the image creator. This makes image attribution pretty straightforward.
Step-by-step Walkthrough
Let’s say I’m looking for images of Casuarina equisetifolia, a widespread tree that’s common in the Pacific Islands. I made this video to demonstrate how I would do this.
As I mention in the video, it often takes some time browsing through images to get the content and quality I want. This is not always obvious from the thumbnails displayed in the gallery.
A few caveats to consider. Especially with crowdsourced records from sources like iNaturalist, one should be careful about species identification. For unfamiliar species, trying to verify images against a published flora would be advisable if correct identification is critical. In practice, I have not actually gone to this degree of verification. Instead, I look through a large number of images to get a sense of distinguishing features; this assumes that the masses are generally correct in their species ID. Also, keep an eye out for images with a “NoDerivatives” version of a CC license, which does not allow any image manipulation (e.g., cropping).
Final Thoughts
At this point, GBIF is generally the first place I go to find plant images. The sheer volume, clear taxonomic categorization, legal clarity in terms of licenses, and ease of attribution are the key factors that make GBIF my go-to. With some practice and patience for sifting through images, I generally find the images I’m looking for here. I recommend it for sourcing plant images for published materials like field guides, presentations, and blog articles. I hope you give it a try!
AI RESPONSIBILITY RUBRIC
This rubric shows human vs AI contribution across stages of developing the article. It was generated by AI and reviewed by Taylan, making adjustments as needed.
--------------------
RESEARCH/VERIFICATION
Human 70% | AI 30%
[==============......]
AI: Claude (Sonnet 4.6)
Core knowledge came from Taylan's direct experience with the subject matter. AI performed supplementary verification, caught factual nuances, and flagged a practical implication that was incorporated into the final article.
--------------------
CONCEPT/PLANNING
Human 60% | AI 40%
[============........]
AI: Claude (Sonnet 4.6)
Taylan drove the core concept and made all decisions about scope, audience, and what to include or cut. AI contributed to structuring and iterating on the outline and raised framing questions that shaped the final angle.
--------------------
WRITING
Human 95% | AI 5%
[==================.]
AI: Claude (Sonnet 4.6)
Taylan wrote the entire draft and produced the video walkthrough independently, with no AI involvement in either. AI contribution was limited to a specific caveat that Taylan incorporated into the draft.
--------------------
EDITING/REFINEMENT
Human 60% | AI 40%
[============........]
AI: Claude (Sonnet 4.6)
Taylan did initial editing pass. Then AI identified specific errors and style issues; Taylan reviewed all suggestions and made final decisions about what to accept.
--------------------
AI Tools: Claude (Sonnet 4.6)