Which stressors? and how to quantify species-level weights/scores?

[oharac]

Stressors to include

@Melsteroni and I did a quick crosswalk of the IUCN threat classification system against the layers available in the new CHI paper. Some were very obvious analogues - light pollution in CHI <-> light pollution (9.6.1) in IUCN, similar for shipping lanes. Others will take some text mining to clarify, e.g. climate-related habitat shifting includes SLR but not necessarily; climate "other impacts" may include OA but again would need to dig further into the text. Categories so far look like they cover everything in the new paper:

• Direct human (1.1-1.3 - residential/commercial dev, 5.1 and 5.3 - human intrusions and disturbance)
• Fishing (5.4.1-5.4.4 - fishing and harvesting aquatic resources)
  • can separate artisanal from commercial
  • within commercial, keywords and habitat checks on species can help differentiate pelagic vs demersal
  • Hab destructive vs not might be similar to the gear type separation I'm doing for ARF
  • low vs high bycatch might be trickier without getting into the underlying Watson data to separate out targeted fish from non-targeted...
• Nutrient pollutants (9.3.1 - agricultural and forestry effluents - nutrient loads)
• Organic chemical pollutants (9.3.3 - agricultural and forestry effluents - herbicides/pesticides)
• Light pollution (9.6.1)
• SLR (11.1 - climate change/severe weather: habitat shifting and alteration)
  • keyword search to refine
• SST (11.3 - climate change/severe weather: temperature extremes)
• OA (11.5 - climate change/severe weather: other impacts)
  • keyword search to refine

Are there other stressors we could or would like to include? e.g. aquaculture? UV? benthic structures? marine debris?

Weighting of stressor impact on species

I think it would be super interesting to use weightings to determine the impact of a stressor on a species' range beyond just impacted/not impacted. But the weightings add a new level of complexity and will attract deeper scrutiny.

• Not all species have a numeric "score" for the impact of a particular stressor.
  • We could also look at severity (rapid declines vs slow declines) to fill in some gaps.
  • Taxa-level mean scores for gapfilling?
• Threat identification based on text-mining to tease out a single stressor could raise questions
  • e.g. "climate change - other impacts" for a given species may include OA but maybe also UV; does a score of 5 (medium impact) apply equally to both? or could OA be a 1 and UV be a 9 (making up numbers here for an example). Text mining is not likely to be able to easily differentiate these.
• Fishing impacts are presumably different for a species depending on whether it's targeted or not, and what specific gear types are present.
  • The CHI layers aggregate the impacts in a way that make it tough to identify actual impacts at the species level.
  • We'll already need keyword checks to determine pelagic/demersal, high/low bycatch, destructive/non-destructive, so we're already making the species-level scores for this threat pretty fuzzy. Add in targeted/non-targeted and we'd have to dig far deeper into the Watson data... which will dramatically slow progress on this!