Change how trees are assigned into PFTs #117
Description
At the moment, individuals from the SAFE census data are assigned into PFTs as follows:
- First, a base PFT species classification is used (based on literature; Kohler and Huth, 1998) to assign a set of species into 4 PFTs (emergent, overstory, understory and pioneer trees).
- The maximum height for each of these PFTs is then calculated based on the individuals contained within the species from each PFT based on SAFE census data
- The PFT maximum height is then used as a threshold value to assign the remaining individuals from the census (i.e., the ones belonging to species that were not part of the base classification)
This approach has a few issues:
- Emergent, overstory and understory refer to positions in the canopy, but pioneer refers to life-history strategy. However, a tree can in theory be a pioneer AND an overstory tree (e.g., if its crown reaches overstory height).
- In the second step (where a PFT is assigned based on maximum height), pioneer trees are those with maximum height greater than understory PFT but smaller than overstory PFT. For some trees this may be correct, but for others it may be the case that these should be seen as overstory PFT. The reason why I went with this approach in the first place is because it allowed assigning trees into PFTs based solely on DBH / height. This approach may still be desired in certain cases (i.e., when detailed census data is not available), but I think we can use a more detailed approach for assigning PFTs.
As an alternative to assigning PFTs based solely on maximum height, I will see if the criteria can be expanded based on those measured by Dobert et al. (2017): https://doi.org/10.1111/1365-2745.12794 (e.g., maximum height, dispersal, fruit, life-history, pollination). I believe @annarallings has already been working on narrowing these criteria down to potential PFTs (see https://imperiallondon.sharepoint.com/:x:/r/sites/VirtualRainforestProject-PQ/Shared%20Documents/Data/Functional%20Groups/Functional%20Groups/[SAFE_functionalgroups_PLANT_v3_20250130.xlsx](https://imperiallondon.sharepoint.com/:x:/r/sites/VirtualRainforestProject-PQ/Shared%20Documents/Data/Functional%20Groups/Functional%20Groups/SAFE_functionalgroups_PLANT_v3_20250130.xlsx?d=w16e17cf4b1d249e9b8a2ecc4912309d9&csf=1&web=1&e=YE13E3)?d=w16e17cf4b1d249e9b8a2ecc4912309d9&csf=1&web=1&e=YE13E3)
What this alternative, more complex, approach will likely look like:
- Use first step described above to calculate PFT maximum height for emergent, overstory, understory, pioneer (i.e., still using the base classification from the literature)
- Then, link the info for each species from Dobert et al. (2017) to the SAFE census data
- Then, instead of assigning remaining species based on maximum height, split the "base PFTs" into more detailed PFTs based on 1) PFT maximum height and 2) Dobert's info. This does mean that this approach requires detailed census data down to species level as input (we have this, but future users may not).
- As part of the previous step, pioneer trees will no longer be a separate PFT. Instead, pioneer species will fall under overstory trees, but will have different functional trait values (reflecting differences in life-history; e.g., different wood density, SLA, maximum height, etc.).
- Similarly, the overstory PFT may be split into several "subPFTs" (e.g., if their fruit type is different (berry/nut) or if their dispersal method is different).
As a result, this approach should:
- Provide a pretty straightforward way to include additional PFT complexity in the model without needing a lot of new code functionality, I think. Note that this form of additional PFT complexity does not necessarily mean a better representation of differences in functional behavior (this will all depend on how detailed PFT variables can be parameterised across PFTs). However, it will definitely improve the way animals have access to specific supplies (e.g., nutrients from berries will only be consumed by animals that eat those), and potentially we can do something similar for dispersal/regeneration (i.e., presence of a certain animal may affect dispersal/regeneration through an animal-seed link).
- Improve the accuracy of how pioneer trees are implemented in the model (in other words, a more correct implementation of life-history strategy)
- Note that the original approach can still work as a more simplistic approach (e.g., when detailed census data is not available). Instead of having "subPFTs" with different values, there would just be the main PFTs (e.g., emergent, overstory and understory), but no classification of species into PFTs based on Dobert's data.