Some first pass results!

At long last, here are some maps of cumulative impacts on species. Here's how I've counted things up:

for each species, identify the stressors to which it is sensitive, and then create a map of the intersection between species range and stressor range. This creates a map of how many stressors are impacting the species at any particular location/cell. These are divided into climate and non-climate stressor groups.
the species/stressor maps are flattened (so 1 = impact due to one or more stressors, 0 = no impacts from any stressors) - basically a map of impacted range.
The species impacted range maps are combined to create a map of how many species are impacted (by one or more stressors) in any particular cell.
Conversely, species ranges that are unimpacted are noted as refugia, and refugia maps are combined to create a map of how many species are NOT impacted (by any stressors) in any particular cell.

Note that the impact and refugia are still divided into climate and non-climate stressor groups. We can't just add the two maps - if we want a total impact map I'd need to calculate all stressors together farther upstream.

Note also stressor intensities or spp vulnerability weights are not accounted for here - "impact" is simply whether a stressor is present at the same location as a species is present. However, the stressor ranges are adjusted using a 95% contour volume approach - so the range includes the highest-value cells that cumulatively contain 95% of the total stressor intensity (this approach is sometimes used to define species ranges, drawing a range boundary around 95% of the total abundance). This process drops cells with very low stressor intensities.

OK, copy and paste doesn't work, and linking to the file (even in a public repo) doesn't seem to work. But drag-and-drop does, so here are some animated maps!

impact_no_cc

impact_cc_only

refugia_no_cc

refugia_cc_only

Alt Text

OK looking at these maps a few ideas and interesting things pop out to me.

First: It's not a surprise that there are more impacted species in the Coral Triange/Indo-Pacific, since there are just more species there. It might be good to normalize these numbers by a species richness map so we get a proportion of species that are impacted by one or more stressors.

Second, it generally seems like the number of impacted species in any location generally increases over time, and also that the extent of overall impacted area (grey means zero impacted species, so where purples intrude into grey) is increasing - particularly with climate stressors.

Third: I'm surprised at the patterns of climate impacts particularly in the open oceans. But makes sense I guess, keeping in mind: a) SLR is limited to the coasts (I clipped it to cells having a min depth of 10 m or less), so all the oceanic stuff is OA and SST. b) OA is generally a direct pressure on inverts, and most of what's in the open oceans is vertebrates. c) these are not necessarily where the stressors are highest, but where stressors and species threatened by those stressors (according to the IUCN assessment) coexist - so perhaps fewer tropical pelagic vertebrates have SST (or OA) listed among their threats compared to temperate and polar species (in part probably because those stressors near the poles are probably higher intensity as well).

The refugia patterns make sense. I believe if I normalize by species richness, then refugia proportions are just 1 - impacted proportions, but my math brain is broken right now.

Thoughts on any of these? Other interesting patterns?

I had the exact same reaction and idea as your first point. Definitely good to do that. I’ll keep pondering other ideas too.

Benjamin S Halpern Director, NCEAS Professor, Bren School UC Santa Barbara

On Mar 31, 2020, at 7:53 PM, oharac notifications@github.com wrote:

OK looking at these maps a few ideas and interesting things pop out to me.

First: It's not a surprise that there are more impacted species in the Coral Triange/Indo-Pacific, since there are just more species there. It might be good to normalize these numbers by a species richness map so we get a proportion of species that are impacted by one or more stressors.

Second, it generally seems like the number of impacted species in any location generally increases over time, and also that the extent of overall impacted area (grey means zero impacted species, so where purples intrude into grey) is increasing - particularly with climate stressors.

Third: I'm surprised at the patterns of climate impacts particularly in the open oceans. But makes sense I guess, keeping in mind: a) SLR is limited to the coasts (I clipped it to cells having a min depth of 10 m or less), so all the oceanic stuff is OA and SST. b) OA is generally a direct pressure on inverts, and most of what's in the open oceans is vertebrates. c) these are not necessarily where the stressors are highest, but where stressors and species threatened by those stressors (according to the IUCN assessment) coexist - so perhaps fewer tropical pelagic vertebrates have SST (or OA) listed among their threats compared to temperate and polar species (in part probably because those stressors near the poles are probably higher intensity as well).

The refugia patterns make sense. I believe if I normalize by species richness, then refugia proportions are just 1 - impacted proportions, but my math brain is broken right now.

Thoughts on any of these? Other interesting patterns?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

New maps! the differences:

The previous maps included least concern species; these contain only near-threatened and threatened. The rules for inclusion here are (prior maps neglected (b)): a) sensitive to one or more stressors b) risk status is not Least Concern, Extinct, or Data Deficient c) species is in a comprehensively-assessed group (fudged to include turtles, clupeids, and syngnathiformes for now) d) species has a range map
The percent maps are normalized by species richness. Here, richness accounts for all species that fit criteria b), c), and d) above (there are a small number of near-threatened/threatened spp that are not sensitive to any of our stressors).
I dropped the refugia maps - they're not interesting to look at. As I figured, if I'd normalized by spp richness based on criteria a) - d) above, refugia + impacted = 100%. Since I am normalizing by a slightly different spp richness, it is not exactly 100% but close enough to be boring. Still good to report in a table or SI perhaps but basically redundant as a figure.
Fixed a few bugs that I found along the way! and improved the fishing gear extraction code so can port that over to the ARF project.

impact_no_cc impact_pct_no_cc impact_pct_cc_only

Non-climate impacts, percent impacted:

Major sudden shifts in non-climate impacts (esp between Africa and Asia) need some probing. I thought it could be a thresholding issue, but not totally convinced - maybe Jamie can take a look to see if she sees something similar for the Aquaculture project (not on this repo but I'll chat with her).

As expected, the percent impacted map is way more interesting, and shows high proportional impacts in areas where we'd expect: North Sea, Baltic Sea, S China Sea, Coral Triangle. Also really highlights impacts from fishing activity in ABNJ just outside of EEZs...

Climate impacts, percent impacted:

I think those apparent hard boundaries that leave the tropics largely untouched, especially N Atlantic, S Atlantic, and the sudden edge east of Australia that curves up through Pacific Island nations, are not to do with climate but to do with climate-sensitive species whose ranges end there. Climate stressors are present everywhere, but if no sensitive/threatened spp exist there, it won't show up. I'll investigate further to verify.

Next steps

I think this is pretty good for now, and can start working on some non-map figures to see what kind of story we can tell with a paper.

Thoughts? Since fishing effort clearly impacts so many species, might be worth doing a separate analysis of just fishing-related impacts across a larger time series, since 2013 was a little while ago.

this is all great stuff. a few quick thoughts/reactions:

you flagged some odd patterns (sudden changes, etc.). Definitely will want to explore those to understand.
if no one has looked at the impact of fishing on threatened species in a spatial way before, then it would be very interesting and valuable to do this. Hard to believe it hasn't been done though . .. plus we will want to pay attention to what's published/going on with global fishing watch as they must be doing some of this too.
a cool next thing to do will be to separate out different taxa to see what patterns look like for them.

So cool!

Are the species that are included in the open, non-coastal, parts of the ocean mainly swimming fish and species with fairly large geographic/depth ranges? If so, I suspect these species aren't particularly temperature sensitive.

I think that the pattern of climate change occurring mainly in coastal regions probably makes sense . But I think this result could reflect a data limitation of deep water species in regard to sampling and physiological information. It seems like there must be deep water benthic organisms that will be sensitive to ocean acidification and warming? Maybe we don't have good knowledge of the existence/distribution of these more obscure-to-human species? Also, I suspect we don't know much about the thermal sensitivity of species in the open ocean because they are difficult to study and thermal sensitivity just hasn't been an issue for them....until maybe now. If anything, it seems benthic organisms in deep waters are going to be more sensitive to climate change than coastal organisms.

Re: weird patterns, here are a couple of maps to explain where they're coming from. Below that is response to Mel's post.

The disappearing pattern between India and Africa

... appears to be from pelagic low-bycatch fishing. It looks like an artifact from the stressor layers rather than from my 95% contour volume approach - in the stressor layers that pattern disappears in 2007. All the fishing layers seem to have odd jumps most likely due to changes in the scale and manner in which it's reported, rather than actual changes in fishing pressure.

The unclipped stressor layer

str_movie_pel_lb

thresholded stressor layer (red = dropped, below the threshold)

str_95cv_movie_pel_lb

The curved edge in the western Pacific

This is due to one (or more?) bird species sensitive to sea surface temperature, whose range does not extend into the Indo-Pacific - birds are also responsible for the pattern in the north Atlantic. Mammals, turtles, and sharks do not contribute to these patterns.

impact_birds_cc_only

stressor layer: SST covers most of the oceans, no weird boundaries

str_movie_sst

thresholded stressor layer: still no weird boundaries

str_95cv_movie_sst

Response to Mel's post:

Are the species that are included in the open, non-coastal, parts of the ocean mainly swimming fish and species with fairly large geographic/depth ranges? If so, I suspect these species aren't particularly temperature sensitive.

Exactly - the water-dwelling spp in the non-coastal areas are generally big pelagics incl tuna, billfishes, mammals, turtles, and sharks. As noted above, some birds show up as being sensitive to SST. I can dig further but may be indirect effects (e.g. SST effects on food sources) rather than direct effects which is what I'd prefer to include.

I think this result could reflect a data limitation of deep water species in regard to sampling and physiological information. It seems like there must be deep water benthic organisms that will be sensitive to ocean acidification and warming? Maybe we don't have good knowledge of the existence/distribution of these more obscure-to-human species?

While this is probably an important issue/concern, it's not an issue here. IUCN just hasn't assessed much in the way of deep benthic species, especially if we want to stick with comprehensively assessed taxa - few invertebrates at all, and the vertebrates are either the big ones or the coral reef ones.

Also, I suspect we don't know much about the thermal sensitivity of species in the open ocean because they are difficult to study and thermal sensitivity just hasn't been an issue for them....until maybe now. If anything, it seems benthic organisms in deep waters are going to be more sensitive to climate change than coastal organisms.

If only we had some sort of systematic approach to estimating a species' vulnerability to stressors based on readily-identifiable physiological traits... but that way lies madness! Though: seems like deep benthic spp would have a narrow thermal tolerance because they've evolved in a pretty stable place - so likely quite sensitive, but depending on how deep they are, how much exposure are they likely to see?

oharac / bd_chi