Data check/exploratory plots

[gclawson1]

Starting this issue for data checking

I've been exploring my estimates for crop land area and production used for salmon aquaculture feeds. I really have no idea if these numbers seem "right", but they are all based on the demand of salmon feeds (and the final production values match the demand values), so theoretically they should be right. To get the crop area, I just divided each crop production raster by yield amounts per country.

I found that there is 10874 km2 of cropland globally and 2925089 of tonnage from that cropland that goes to salmon aquaculture feeds.

• First off, does this seem reasonable for the km2 estimate? I really have no idea...
• Comparing to this paper on the EU chicken meat feed industry, they estimate that the feed for EU chickens account for ~67000 km2 of cropland globally
• They also have an estimate 5.59 m2 / kg of chicken while I estimate ~3.71 m2 / kg of salmon

Table of km2 of cropland for each ingredient:

I wonder how my numbers match to yours, @cottrellr ? We can discuss in our meeting later.

[gclawson1]

For reference, the global amount of soybean area harvest (according to mapspam) is 979068 km2. So my estimate of 3436 km2 of soybeans for SPC for salmon aquaculture feed would be ~0.3% of global soybean harvest.

[cottrellr]

OK that seems like the right scale - given EU chicken production must absolutely blow global salmon production out the water. THese are my aggregated tonnage estimates of raw materials across both scenarios for my paper

[cottrellr]

But just thinking about it this was before I took out SBM as an SPC coproduct, so the conversion factors were much lower when I summed these. So I think you are in the right ball park..

[gclawson1]

Checking against pig feed use in the EU: https://www.sciencedirect.com/science/article/pii/S0048969720378372?ref=pdf_download&fr=RR-2&rr=7d6c97367b801502

They suggest "The total 2017 EU pork meat production, 23 Mtonnes (Eurostat, 2020b), lead to a total resource use of 14.5 Mha of land, 51.9 Gm3 of green water, 3.9 Gm3 of blue water, 1.23 Mtonnes of nitrogen, 0.35 Mtonnes of phosphorous, and 0.34 Mtonnes of potassium."

• So 145000 km2 of land use for pig feed used in the EU

[gclawson1]

Checking against Kuempel et al. chicken salmon paper:

Looked into that data and she reports 7561.073 km2 of feedcrop disturbance. Given that they used a different diet, different allocation approaches, and different data year than I did, seems pretty reasonable (reminder my estimate for mass allocation plant-dominant diet is ~11000 km2 of feedcrop disturbance).

Similarly the global food system paper reports 7525.14 km2 of feedcrop disturbance for salmon aquaculture

[cottrellr]

I agree and this will change with allocation strategies too - economic likely being the most dramatic change

---

Richard S. Cottrell Research Fellow in Aquaculture Sustainability Institute for Marine and Antarctic Studies College of Sciences and Engineering University of Tasmania

Theme Co-Lead, Sustainable Futures and Planetary Health Centre for Marine Socioecology University of Tasmania

Size Ecology Labhttps://www.sizeecology.org/ | Centre for Marine Socioecologyhttps://marinesocioecology.org/themes/sustainable-futures-and-planetary-health/ Google Scholarhttps://scholar.google.com/citations?user=1pLCMKIAAAAJ&hl=en | ORCIDhttps://orcid.org/my-orcid?orcid=0000-0002-6499-7503 | @RichCottrell22https://twitter.com/RichCottrell22

On 7 Jul 2023, at 7:45 am, Gage Clawson @.**@.>> wrote:

Checking against Kuempel et al. chicken salmon paper:

Looked into that data and she reports 7561.073 km2 of feedcrop disturbance. Given that they used a different diet, different allocation approaches, and different data year than I did, seems pretty reasonable (reminder my estimate for mass allocation plant-dominant diet is ~11000 km2 of feedcrop disturbance).

— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/10#issuecomment-1624348133, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJK3YJC6RXTBOCQKJJK7FJTXO4WZJANCNFSM6AAAAAAZ7TFTWQ. You are receiving this because you were mentioned.Message ID: @.***>

This email is confidential, and is for the intended recipient only. Access, disclosure, copying, distribution, or reliance on any of it by anyone outside the intended recipient organisation is prohibited and may be a criminal offence. Please delete if obtained in error and email confirmation to the sender. The views expressed in this email are not necessarily the views of the University of Tasmania, unless clearly intended otherwise.

[gclawson1]

Updated total global plots with finalized methods:

Impact maps to come next!

[gclawson1]

Species weighted mean impact (value*1000000 logged), standard deviation (using pooled varianace), and nspp maps per diet

• Still figuring out how to best visualize this. It is a bit hard to see differences in this. Obviously, for the most part, impacts on land are a bit more intense in the plant-dominant, and impacts in the ocean are a bit more intense in the fish-dominant diet. It will be important to really 1) find where the differences are, 2) show the hotspots
• Looks like ocean impacts increase in Chinese coastal areas and japan coastal areas in the fish dominant.
• Weirdly looks like impacts on Germany, Russia, northeast China land increases in the fish dominant?

[gclawson1]

Ok, here are better plots for economic allocation impacts. 1) number of species weighted mean (value*1e6 logged) in each cell under each diet 2) nspp in each cell (these are the same under each diet), and 3) standard deviation

[gclawson1]

Curious how the mean maps change across allocation types?

Looks like ocean impacts increase a bit under the different allocation approaches (see east pacific ocean, red sea). This is due to trimmings getting a much lower allocation value in the economic

[gclawson1]

I now have all of these plots across all ingredients and taxa, across ingredients by taxa, and across taxa by ingredients. I'm going to work on creating a shiny app for exploration

[cottrellr]

These plots are beautiful. What does nspp mean? number of species impacted?

[cottrellr]

And also what is the grey?

[cottrellr]

Updated total global plots with finalized methods:

Impact maps to come next!

But this is total km2, right? Not sure we should be using that as a result. It only tells us that some habitats are bigger than others so they overlap with more.

[gclawson1]

These plots are beautiful. What does nspp mean? number of species impacted? Also what is the grey?

Nspp is the the raw number of species in each cell, but you made me realize I should make those the number of species in each cell that are impacted, so I will update that.

The gray on land is no impact and the white in the ocean is no impact (or no species in the case of nspp). I need to update so that they are the same. Will probably do gray as no impact for both land and ocean.

But this is total km2, right? Not sure we should be using that as a result. It only tells us that some habitats are bigger than others so they overlap with more.

Agreed, I just wanted to see these for exploratory purposes.

Additionally, I have a shiny app coded up to show all of these maps (+more disaggregated versions), I'm just working to make it faster!

[cottrellr]

That is awesome, Gage. Stellar work. Great to hear – chat later.

---

Richard S. Cottrell Research Fellow in Aquaculture Sustainability Institute for Marine and Antarctic Studies College of Sciences and Engineering University of Tasmania

Theme Co-Lead, Sustainable Futures and Planetary Health Centre for Marine Socioecology University of Tasmania

Size Ecology Labhttps://www.sizeecology.org/ | Centre for Marine Socioecologyhttps://marinesocioecology.org/themes/sustainable-futures-and-planetary-health/ Google Scholarhttps://scholar.google.com/citations?user=X1a9t90AAAAJ&hl=en&authuser=1 | ORCIDhttps://orcid.org/my-orcid?orcid=0000-0002-6499-7503 | @RichCottrell22https://twitter.com/RichCottrell22

From: Gage Clawson @.> Date: Wednesday, 18 October 2023 at 2:46 am To: Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping @.> Cc: Richard Cottrell @.>, Mention @.> Subject: Re: [Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping] Data check/exploratory plots (Issue #10)

These plots are beautiful. What does nspp mean? number of species impacted? Also what is the grey?

Nspp is the the raw number of species in each cell, but you made me realize I should make those the number of species in each cell that are impacted, so I will update that.

The gray on land is no impact and the white in the ocean is no impact. I need to update so that they are the same. Will probably do gray as no impact for both land and ocean.

But this is total km2, right? Not sure we should be using that as a result. It only tells us that some habitats are bigger than others so they overlap with more.

Agreed, I just wanted to see these for exploratory purposes.

Additionally, I have a shiny app coded up to show all of these maps (+more disaggregated versions), I'm just working to make it faster!

— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/10#issuecomment-1766799523, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJK3YJBDFVYLQKYCNWACUULX72Y7BAVCNFSM6AAAAAAZ7TFTWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONRWG44TSNJSGM. You are receiving this because you were mentioned.Message ID: @.***>

This email is confidential, and is for the intended recipient only. Access, disclosure, copying, distribution, or reliance on any of it by anyone outside the intended recipient organisation is prohibited and may be a criminal offence. Please delete if obtained in error and email confirmation to the sender. The views expressed in this email are not necessarily the views of the University of Tasmania, unless clearly intended otherwise.

[cottrellr]

• I would go with white as no impact. Empty is intuitive for no impact

[gclawson1]

How about one of these? In the second one I included grey for the country borders, but could remove it if that is confusing

[cottrellr]

I would have the country boundaries grey. But the fill transparent and then have them on top of the raster layer. I'm sure you'll find a good approach, I leave it to you :)

[gclawson1]

Updated plot with new color scheme:

[gclawson1]

I've been doing some data checks and exploring better ways to visualize, and I came across a possible oddity regarding my trimmings pressures and impacts. Nothing is technically wrong with the way I've mapped the pressures from trimmings (I checked the code, and I am doing it the same way as I do for forage fisheries), however, as you'll see below, I'm estimating pressures (and subsequently impacts) in almost all ocean cells. This is because the fish catch associated trimmings is so widespread. However, the catch I allocate and pressures are really really small.

Here is a plot where I've just categorized cells impacts as higher in fish dominant (pink) vs higher in plant dominant (green). You'll notice that basically all of the ocean cells are pink, because there is some trimmings pressure (and impact) in nearly all ocean cells, due to pelagic fisheries included in trimmings, like some tuna species. Additionally, there are no true 0's (no difference) in the delta.

Digging deeper, here is a look at the pressures from trimmings. All of the gray areas are numbers very close to 0.

class : SpatRaster dimensions : 360, 720, 1 (nrow, ncol, nlyr) resolution : 0.5, 0.5 (x, y) extent : -180, 180, -90, 90 (xmin, xmax, ymin, ymax) coord. ref. : lon/lat WGS 84 (EPSG:4326) source(s) : memory name : sum min value : 0.00000000000001360643 max value : 17.47350792586803436279

I think there are a couple paths forward to either keep this as is, or tweak a bit:

• Keep as is, and just present it the way I currently have, where in my impact maps, the really really small numbers (basically 0) are blown out by the larger values and appear as 0 (see plot below):

• Make some assumptions about where trimmings are allowed to be sourced from in the ocean (as I think you @cottrellr have done). Basically I think this would mean that we would exclude certain parts of the ocean, no matter where the trimmings end up being traded to.

[gclawson1]

Some more plots!

We had discussed the idea of looking at the proportional impacts, rather than the raw mean impact numbers. I was imagining something similar to these plots:

My exploratory plots:

Number 1. Proportional mean impacts.

• The rows represent each raw material, the x axis is the proportion of total impact within each raw material and "All" is proportional mean impacts across all raw materials. Each panel represents a different scenario (dietary and efficiency).

Number 2. The same thing as above, except the raw materials and taxonomic types are switched in the plot. (sorry the colors are atrocious here, just the default R ones)

• Note that this is economic allocation, which is why the trimmings impacts are really really small.

Number 3. Regular mean impacts by raw material and taxon (not proportional)

Number 4. Same thing but raw material and taxon switched

[cottrellr]

This is all fascinating Gage!

Re the trimmings, yeah I think you could probably limit to it certain FAO fishing areas? I limited by just saying forage fish and trimmings come from this FAO fishing area or this FAO fishing area (i.e. everything comes from area 27, or 61, or 87). But with mine much more conceptual and not trade based that allows a little more freedom.

You could limit to the regions Biomar state their wild by products come from?

[cottrellr]

With the impact plots, I agree, from the examples (I assume Ben's work?), you could do have something very similar.

In fact, I would do mean impact aggregated for the different taxa with some variance indicated for among species differences i.e., as in the first of those three plots, where "Full" and "Coastal" could be plant versus fish-dominant.

That is essentially your first summary plot - so what do you think of a following order of plots:

• Maps of total impact for each diet (under the regular scenario).
• Summary of mean impact aggregated by taxa with variance
• Then the breakdown of the mean impact - I really like the species weighted impact as I think that paints a clearer picture (because basically all marine impacts are forage fish, I think it gives a skewed impression that forage fish is worse than say soybean, but the species weighted one adjusts).
• Then we could do comparisons of efficiency later on - or do you think in parallel?

[gclawson1]

I think we should do efficiency comparisons later on. There is a lot of info in all of these plots and keeping it to only the "regular" efficiency makes it so much easier to understand

[cottrellr]

Cool, I agree. Overload of information. I think that could be a great latter section in the paper. Can discuss key raw materials which are hotspot for prioritising for feed companies. Then look at how efficiency changes things. In the end, for species that simply cannot live around agriculture, better agriculture will still likely be problematic land use for them – so a lower raw material demand may be the most powerful thing for reducing embedded impacts.

---

Richard S. Cottrell Research Fellow in Aquaculture Sustainability Institute for Marine and Antarctic Studies College of Sciences and Engineering University of Tasmania

Theme Co-Lead, Sustainable Futures and Planetary Health Centre for Marine Socioecology University of Tasmania

Size Ecology Labhttps://www.sizeecology.org/ | Centre for Marine Socioecologyhttps://marinesocioecology.org/themes/sustainable-futures-and-planetary-health/ Google Scholarhttps://scholar.google.com/citations?user=X1a9t90AAAAJ&hl=en&authuser=1 | ORCIDhttps://orcid.org/my-orcid?orcid=0000-0002-6499-7503 | @RichCottrell22https://twitter.com/RichCottrell22

From: Gage Clawson @.> Date: Wednesday, 10 January 2024 at 9:40 am To: Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping @.> Cc: Richard Cottrell @.>, Mention @.> Subject: Re: [Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping] Data check/exploratory plots (Issue #10)

I think we should do efficiency comparisons later on. There is a lot of info in all of these plots and keeping it to only the "regular" efficiency makes it so much easier to understand

— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/10#issuecomment-1883964701, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJK3YJHQS6QGHJEBDRNCPBDYNXIKNAVCNFSM6AAAAAAZ7TFTWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOBTHE3DINZQGE. You are receiving this because you were mentioned.Message ID: @.***>

This email is confidential, and is for the intended recipient only. Access, disclosure, copying, distribution, or reliance on any of it by anyone outside the intended recipient organisation is prohibited and may be a criminal offence. Please delete if obtained in error and email confirmation to the sender. The views expressed in this email are not necessarily the views of the University of Tasmania, unless clearly intended otherwise.

[bshalpern]

I think the main issue (besides how to present the results) is how to explain/interpret the really (really!) small values. How do we anticipate and mitigate reviewer comments that the values are so small that the results are 'not meaningful....'? Let's chat through this.

[bshalpern]

overall I really like these figures and where they are headed...

[cottrellr]

I wonder whether it’s worth presenting as the mean proportion of habitat rather than extinction risk? So just avoiding that last calculation step. Probably more intuitive that 0.00005% means a pretty low impact compared to a 0.000001 extinction risk? (Pulling numbers out of the air)

Sent from my iPhone

On 11 Jan 2024, at 2:45 am, bshalpern @.***> wrote:



overall I really like these figures and where they are headed...

— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/10#issuecomment-1885215957, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJK3YJCVQUSZ4KSJMFDXF5LYN3ARFAVCNFSM6AAAAAAZ7TFTWSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOBVGIYTKOJVG4. You are receiving this because you were mentioned.Message ID: @.***>

This email is confidential, and is for the intended recipient only. Access, disclosure, copying, distribution, or reliance on any of it by anyone outside the intended recipient organisation is prohibited and may be a criminal offence. Please delete if obtained in error and email confirmation to the sender. The views expressed in this email are not necessarily the views of the University of Tasmania, unless clearly intended otherwise.

[gclawson1]

[gclawson1]

Reran all of the impacts with the new trimmings information, and the maps make a lot more sense now. I was able to exclude trimmings areas on a species basis based on the BioMar sustainability report.