Closed gclawson1 closed 7 months ago
cottrellr
commented
1 year ago Hmm strange - OK. I think I would need to go through the code - both mine and yours to compare and see how we are attacking it. I'm also just wondering whether there is a conversion factor column that is getting mixed up.
gclawson1
commented
1 year ago Yeah I'm not really sure what's going on. For the trimmings part, the species with the largest disturbance values are Tuna species for me, which probably makes sense. I am very skeptical of how much disturbance I am seeing from my trimmings estimate and how little disturbance I am seeing from my forage fish estimate though...
Also, I was curious about this part of the code:
ppr_prop_fmfo = case_when(ppr_fmfo/max_npp_t_C_yr<1 ~ ppr_fmfo/max_npp_t_C_yr,
TRUE ~ ppr_fmfo/max_npp_t_C_yr)Is there a typo in there somewhere? It seems like the case and the when parts should be different (not dividing by the same thing, the max_npp).
cottrellr
commented
1 year ago Yeah I think that part of the code is just a legacy of me rushing to try and get these results sorted pre AMSA conference. Essentially I had it to start mean_npp if PPR was lower than npp and if not set it to max. My original thinking on this was because there is of course a seasonal element to fisheries and PPR might very well exceed NPP for a cell when there is a seasonal surge in fisheries e.g. anchoveta. But the more I thought about it I realised I was penalising systems that had lower PPR by setting a different denominator. So just set the whole thing to max but hadn't gone through to tidy the code.
gclawson1
commented
1 year ago Getting back to this. I just reran my code but used the same 2019 NPP files that @cottrellr did.
What we confirmed last time we chatted was:
After rerunning with the 2019 npp data, my forage fisheries area seems somewhat reasonable:
However, my trimmings is still way higher than my forage fisheries km2:
If I were to try to match my area km2 proportionally to what Rich's is, then I would expect 100000/17446.36 = 5.73 tonnes/km2 on average. This means that I would need 1277342 (total tonnage of forage + trimmings) / 5.73 = 222849.7 km2.
Based on all of this, I am inclined to think that my trimmings catch and area estimates are the problem... The area estimates are just too large to make sense, given the tonnage is so low in comparison, and that Rich's trimmings area estimates are much lower than his forage fish area estimates. I'm going to dive into how I calculated the trimmings catch to see if there is anything going wrong/I can fix.
cottrellr
commented
1 year ago Yeah I agree, think trimmings are the problem looking at the comparison.
One aspect to it could be (because you are using trade your proportional contributions from different areas is different to mine) your data is being represented by large range species such as Tuna, so your catch is lots of cells even if the coverage per cell isn’t huge.
This becomes a problem when comparing the different regions in my analysis too. So far I have just been assuming trimmings composition is proportional to total catch. But this means, wherever there is lots of tuna, trimmings are way to tuna-dependent which completely blows out the PPR because of the super high trophic level. This meant that places like Humboldt were coming out as the most intensive which is the complete opposite to what Biomar find given the species they use for trimmings.
So looking at the Biomar sustainability report for 2022, they have how much of each species comes from catch or from trimmings and we can calculate weights based on that. I have just done this for my paper so I can walk you through when we meet?
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 9 Aug 2023, at 2:56 am, Gage Clawson @.**@.>> wrote:
Getting back to this. I just reran my code but used the same 2019 NPP files that @cottrellrhttps://github.com/cottrellr did.
What we confirmed last time we chatted was:
After rerunning with the 2019 npp data, my forage fisheries area seems somewhat reasonable:
However, my trimmings is still way higher than my forage fisheries km2:
If I were to try to match my area km2 proportionally to what Rich's is, then I would expect 100000/17446.36 = 5.73 tonnes/km2 on average. This means that I would need 1277342 (total tonnage of forage + trimmings) / 5.73 = 222849.7 km2.
Based on all of this, I am inclined to think that my trimmings catch and area estimates are the problem... The area estimates are just too large to make sense, given the tonnage is so low in comparison, and that Rich's trimmings area estimates are much lower than his forage fish area estimates. I'm going to dive into how I calculated the trimmings catch to see if there is anything going wrong/I can fix.
— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/11#issuecomment-1669981941, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJK3YJACZXKVIXTXYK3IEOLXUJVSRANCNFSM6AAAAAA3B4RGB4. 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
commented
1 year ago Yeah, that is my suspicion and exactly what I am seeing too. The species like tuna is dispersed across many more cells than the forage fisheries. This is logged for better vis, but you can see Humboldt is way higher than everything else for trimmings:
That sounds good to go over it tomorrow. I've revised and gone over my code most of today and I couldn't find anything troublesome and everything seems to be in order, so I don't think it is a coding mistake as to why I am seeing this trimmings # so high.
gclawson1
commented
1 year ago Was just curious so I compared my forage FOFM disturbance value to the chicken salmon paper and... surprisingly similar!
Chicken salmon paper: 80902 km2
Mine: 77861.26 km2 (mass allocation, plant-dominant)
Obviously different methods, but makes me feel more confident about my forage fish fofm value.
gclawson1
commented
1 year ago I figured out what was wrong with my trimmings!!!
In my code, I match the fmfo trade to the watson data, so that I can see that, for example, if 55% of the trimmings catch in USA is from XX spp, then 55% of the fish meal (and 55% of the fish oil) exported from or kept in USA for trimmings will be attributed to XX spp. (This is a bit of an oversimplification, but it is basically whats happening)
When I was flagging species in the watson data as trimmings catch or not, I figured out that I was losing many of the trimmings species due to name mismatches (particular sardinella and anchovies, some of the species with the most catch in places like Peru, a major exporter of FMFO).
Because of this, when I matched to the trade data, I wasn't attributing any anchovy or sardine catch to FMFO trimmings in Peru, instead, I was attributing only TUNA catch to the FMFO trimmings in Peru. So basically I was saying that 100% of the trimmings catch in Peru was coming from Tuna species, and consequently, 100% of the fish meal and oil from trimmings that Peru exports (which is the most globally, by far), was attributed to Tuna.
And because all of that was assumed to be Tuna catch, a species with a high trophic level, that PPR was through the roof, exploding my disturbance metric for trimmings.
Now that I have fixed the species joining problem, only <1% (!) of FMFO exported from Peru is from Tuna. I suspect this problem also applies to places like Chile, Ecuador, etc.
After rerunning my disturbance metric, my values make a lot more sense!
For a plant-dominant, mass allocation, fish meal:
forage fish disturbance = 77861.26 km2 (the same as above, since nothing changed here)
Trimmings fish disturbance = 28807.79 km2 (compared to 452806.2 km2 before...)
And consequently 28807.79/77861.26 = 0.3699887, which lines up really well with the tonnage proportion of trimmings to forage fish, and I believe the biomar estimates (~30%).
Feeling a lot better about this now!!
juliablanchard
commented
1 year ago Nice detective work Gage!
———— Sent from my phone
From: Gage Clawson @.> Sent: Saturday, August 12, 2023 8:59:19 AM To: Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping @.> Cc: Subscribed @.***> Subject: Re: [Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping] FMFO disturbance (Issue #11)
I figured out what was wrong with my trimmings!!!
In my code, I match the fmfo trade to the watson data, so that I can see that, for example, if 55% of the trimmings catch in USA is from XX spp, then 55% of the fish meal (and 55% of the fish oil) exported from or kept in USA will be attributed to XX spp.
When I was flagging species in the watson data as trimmings catch or not, I figured out that I was losing many of the trimmings species due to name mismatches (particular sardinella and anchovies, some of the species with the most catch in places like Peru, a major exporter of FMFO).
Because of this, when I matched to the trade data, I wasn't attributing any anchovy or sardine catch to FMFO in Peru, instead, I was attributing only TUNA catch to the FMFO in Peru. So basically I was saying that 100% of the trimmings catch in Peru was coming from Tuna species, and consequently, 100% of the fish meal and oil that Peru exports (which is the most globally, by far), was attributed to Tuna.
And because all of that was assumed to be Tuna catch, a species with a high trophic level, that PPR was through the roof, exploding my disturbance metric for trimmings.
Now that I have fixed the species joining problem, only <1% (!) of FMFO exported from Peru is from Tuna. I suspect this problem also applies to places like Chile, Ecuador, etc.
After rerunning my disturbance metric, my values make a lot more sense!
For a plant-dominant, mass allocation, fish meal:
forage fish disturbance from trimmings = 77861.26 km2 (the same as above, since nothing changed here)
Trimmings fish disturbance = 28807.79 km2 (compared to 452806.2 km2 before...)
And consequently 28807.79/77861.26 = 0.3699887, which lines up really well with the tonnage proportion of trimmings to forage fish, and I believe the biomar estimates (~30%).
Feeling a lot better about this now!!
— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/11#issuecomment-1675495872, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABGD7OTPJTFIRDKGBNM52ILXU22MPANCNFSM6AAAAAA3B4RGB4. You are receiving this because you are subscribed to this thread.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
commented
1 year ago Great job Gage, and I’ll forward your the code I used for the weightings in Monday, when I get to work hardcore on getting this paper finished again 😊
Sent from my iPhone
On 12 Aug 2023, at 7:21 pm, Julia Blanchard @.***> wrote:
Nice detective work Gage!
———— Sent from my phone
From: Gage Clawson @.> Sent: Saturday, August 12, 2023 8:59:19 AM To: Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping @.> Cc: Subscribed @.***> Subject: Re: [Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping] FMFO disturbance (Issue #11)
I figured out what was wrong with my trimmings!!!
In my code, I match the fmfo trade to the watson data, so that I can see that, for example, if 55% of the trimmings catch in USA is from XX spp, then 55% of the fish meal (and 55% of the fish oil) exported from or kept in USA will be attributed to XX spp.
When I was flagging species in the watson data as trimmings catch or not, I figured out that I was losing many of the trimmings species due to name mismatches (particular sardinella and anchovies, some of the species with the most catch in places like Peru, a major exporter of FMFO).
Because of this, when I matched to the trade data, I wasn't attributing any anchovy or sardine catch to FMFO in Peru, instead, I was attributing only TUNA catch to the FMFO in Peru. So basically I was saying that 100% of the trimmings catch in Peru was coming from Tuna species, and consequently, 100% of the fish meal and oil that Peru exports (which is the most globally, by far), was attributed to Tuna.
And because all of that was assumed to be Tuna catch, a species with a high trophic level, that PPR was through the roof, exploding my disturbance metric for trimmings.
Now that I have fixed the species joining problem, only <1% (!) of FMFO exported from Peru is from Tuna. I suspect this problem also applies to places like Chile, Ecuador, etc.
After rerunning my disturbance metric, my values make a lot more sense!
For a plant-dominant, mass allocation, fish meal:
forage fish disturbance from trimmings = 77861.26 km2 (the same as above, since nothing changed here)
Trimmings fish disturbance = 28807.79 km2 (compared to 452806.2 km2 before...)
And consequently 28807.79/77861.26 = 0.3699887, which lines up really well with the tonnage proportion of trimmings to forage fish, and I believe the biomar estimates (~30%).
Feeling a lot better about this now!!
— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/11#issuecomment-1675495872<https://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/11#issuecomment-1675495872>, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABGD7OTPJTFIRDKGBNM52ILXU22MPANCNFSM6AAAAAA3B4RGB4<https://github.com/notifications/unsubscribe-auth/ABGD7OTPJTFIRDKGBNM52ILXU22MPANCNFSM6AAAAAA3B4RGB4>. You are receiving this because you are subscribed to this thread.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.
— Reply to this email directly, view it on GitHubhttps://github.com/Sustainable-Aquafeeds-Project/feed_biodiv_impact_mapping/issues/11#issuecomment-1675808711, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJK3YJEP77OZ6XR62LAMMBDXU5DKJANCNFSM6AAAAAA3B4RGB4. You are receiving this because you were mentioned.Message ID: @.***>
I've taken the code from Rich's project for fisheries disturbance and used it to calculate it within mine. I'm seeing a couple of odd things that I'll document here.
After calculating the km2 of disturbance for both forage fish and trimmings that are used for FMFO, we see this (comparisons are from plant-dominant diet with mass allocation):
Here is what the log of my plant dominant, mass allocation FMFO forage fish + trimmings km2 raster looks like. The colorful area off the west coast of South America is pretty much all from trimmings:
This looks fairly different from Rich's:
I know these aren't perfect comparisons, as we used slightly different diets (I think Rich's diet has a larger proportion of the diet from FMFO) and methods (e.g., I incorporated trade data), but I feel like at the very least, my total amount of disturbance should be less than yours, not more, given the tonnage discrepancy.