Summary

This document combines all exploratory analysis for a report submitted to WWF-EPO using data from Pendrill et al (2020), the physical trade data from Kastner et al (2011) and Trase (data from Q3-20 release).

Most analyses in this markdown are also linked to the graphs available on Observable: https://observablehq.com/collection/@trase/wwf-eu

Important dates:

Aside from the connection to the Trase database, the data from Pendrill et al (2020) and Kastner et al (2011) is saved locally. Note that the data from Kastner et al (2011) cannot widely be shared, while the data from Pendrill et al (2020) is now publicly available at http://doi.org/10.5281/zenodo.4250532.

List of Trase datasets used in this report:

Note that these results contain a likely error for Paraguay to be revisited.

Exploration: Check the EU28 consumption of key commodities

First, check the EU28 consumption from TROPICAL countries using Kastner et al (2011).

Provide table for 2017 to be used in the report

Table 1: List of main agricultural commodities sourced from 112 tropical countries, imported and consumed by the EU28 in 2017. Based on COMTRADE (2020) and the trade model from Kastner et al (2011).

Note: results in the report have converted the “oil palm, fruit” into palm oil using the equivalence factors (2.65). We can generally refer to “Oil palm fruit” as “Oil palm products” for simplicity.

WWF-EPO was interested in separating the effects of trade with those of deforestation to understand the decline of deforestation associated with imports in 2005-2017 (as identified below with the Pendrill et al (2020) dataset). We can generate several time series graphs showing total tonnes of products imported from tropical countries and consumed in the EU28, focusing specifically on:

  • Commodities: Soybean, oil palm fruit, maize, wood products, coffee, cocoa (beans), beef, cotton (seed)
  • Countries: Indonesia, Malaysia, Papua New Guinea, Brazil, Argentina, Paraguay, Chile, Vietnam, Honduras, Colombia, Cote d’Ivoire, Ghana, Nigeria

In the case of soybean, we also want to show imports from non-tropical countries such as the US and Canada, and other countries on the European continent.

For this analysis we remove EU28 producing countries that would have exported within the EU to clearly separate the tropical from non-tropical countries.

We note that imports and consumption of commodities into the EU28 is roughly 70% from tropical countries and 90% from non-tropical countries (including domestic consumption).

The commodities listed as “Other” from tropical countries are:

Wheat, Rice, paddy, Barley, Oats, Millet, Sorghum, Buckwheat, Quinoa, Canary seed, Potatoes, Sweet potatoes, Cassava, Roots and tubers nes, Sugar crops nes, NA, Beans, dry, Broad beans, horse beans, dry, Peas, dry, Chick peas, Lentils, Brazil nuts, with shell, Cashew nuts, with shell, Almonds, with shell, Walnuts, with shell, Pistachios, Kola nuts, Areca nuts, Nuts nes, Groundnuts, with shell, Coconuts, Olives, Karite nuts (sheanuts), Castor oil seed, Sunflower seed, Rapeseed, Safflower seed, Sesame seed, Mustard seed, Linseed, Oilseeds nes, Cabbages and other brassicas, Artichokes, Asparagus, Lettuce and chicory, Spinach, Tomatoes, Cauliflowers and broccoli, Pumpkins, squash and gourds, Cucumbers and gherkins, Eggplants (aubergines), Chillies and peppers, green, Onions, shallots, green, Onions, dry, Garlic, Leeks, other alliaceous vegetables, Beans, green, Peas, green, Carrots and turnips, Maize, green, Mushrooms and truffles, Vegetables, fresh nes, Bananas, Plantains and others, Oranges, Tangerines, mandarins, clementines, satsumas, Lemons and limes, Grapefruit (inc. pomelos), Apples, Pears, Quinces, Apricots, Cherries, Peaches and nectarines, Plums and sloes, Strawberries, Blueberries, Grapes, Watermelons, Melons, other (inc.cantaloupes), Figs, Mangoes, mangosteens, guavas, Avocados, Pineapples, Dates, Persimmons, Kiwi fruit, Papayas, Fruit, tropical fresh nes, Fruit, fresh nes, Tea, Maté, Hops, Pepper (piper spp.), Chillies and peppers, dry, Vanilla, Cinnamon (cannella), Cloves, Nutmeg, mace and cardamoms, Anise, badian, fennel, coriander, Ginger, Spices nes, Pyrethrum, dried, Jute, Manila fibre (abaca), Fibre crops nes, Tobacco, unmanufactured, Rubber, natural, Meat, pig, Milk, Total, Eggs Primary, Sheep and Goat Meat, Meat, Poultry, Rye, Chestnut, Fonio, Currants, Flax fibre and tow, Triticale, Cherries, sour, Bambara beans

Then, we check the imports of products from ALL countries in all years (tropical and non-tropical countries) that are conusmed in the EU28 using the data from Kastner et al (2011).

Countries of interest in the analysis:

  • Tropical countries: Brazil, Paraguay, Argentina (soybean), Malaysia, Indonesia, Papua New Guinea (oil palm fruit)
  • Non-Tropical countries: EU28, United States of America, Canada, Ukraine, China, mainland, Russian Federation, Republic of Moldova, Serbia, Kazakhstan

Soybean

There has clearly been a drop in EU28 soybean imports and consumption from Brazil, and Argentina and an increase of influx of soybeans from the US and other non-tropical countries, including a boost in local production (from EU28) as well as Canada and the Ukraine.

Oil palm fruit

There seems to have been a switch from Malaysia to Indonesia as the main SE Asian source of palm oil, but also there has clearly been some diversification to the other tropical countries, namely:

  • In 2005: Colombia (0.8 Mtons), Ecuador (0.15 Mtons), Brazil (0.14 Mtons), Cote d’Ivoire (0.12 Mtons), and others
  • In 2017: Colombia (1.8 Mtons), Honduras (1.6 Mtons), Guatemala (1.4 Mtons), Ecuador (0.4 Mtons), Cote d’Ivoire (0.17 Mtons) and others

The full table of results follows:

The final step is to check the forest cover loss using the Global Forest Watch data for:

  • Brazil, Paraguay and Argentina (to relate to soybeans)
  • Indonesia, Malaysia, Papua New Guinea (to relate to oil palm fruit)

Note that in all countries, except Brazil, we use the 25% threshold for tree cover loss.

Following the method from Pendrill et al (2020) we look at the tree cover loss in 2003 and 2015 to link to exports in 2005 and 2017 in the time series. Considering both of those years, there was a drop in tree cover loss per year in Brazil (from 3.8 Mha/y to 2.2 Mha/y), while tree cover loss remained roughly flat in Paraguay and Argentina (around 0.2-0.3 Mha/y).

In contrast tree cover loss in Southeast Asia increased in Indonesia when comparing 2003 (0.5 Mha/y) to 2015 (1.8 Mha/y), as well as in Papua New Guinea (from 0.04 Mha/y to 0.2 Mha/y) and Malaysia (from 0.2 Mha/y to 0.5 Mha/y).

Note of caution: the above is not meant to replace the analysis with the data from Pendrill et al (2020), rather it is an exploratory analysis to get a high level sense of the separate trends in trade and tree cover loss, recognizing that not all of the trade is linked to tree cover loss (as was the motivation behind the work done by Pendrill et al (2020)).

The rest of the document focuses on generating the data for the individual insights

Insight 1 (I1)

Significant deforestation remains in EU imported products

First we do a high level analysis of EU28 using the Pendrill et al (2020) data. This will show the position of the EU28 compared to other economies in the world.

These graphs summarize the total embedded deforestation (ha) and CO2 emissions from deforestation for imports into the EU28.

Obtain the average deforestation and CO2 emissions per year of imports into EU countries:

Deforestation per capita in the largest EU economies, in 2017:

The next graph shows differences between 2005 and 2017 for each EU28 country. This graph did not end up being used in the report, but it is still good to see:

The following table allows you to extract the actual numbers from the above graphs.

We then derive a high level global summary to see the position of EU28 and its relation to deforestation for imports over time and compared to other countries such as China (Mainland and Hong Kong combined), the US, India and Japan.

Note: The above table has the domestic consumption removed from the dataset. You can see the ranking per country (group) and per year.

Then we look into which imported products have the greatest impacts by comparing 2017 (latest year) to the 2005-2016 mean. First in bar graph, then in treemap to also link to specific source countries.

Observation: Not all commodities are imported into and consumed in the EU28 every year so you have to be careful how you calculate the mean product deforestation per tonne per year when considering a range (e.g. 2005-2017). There were discussions back-and-fourth with WWF-EPO as the values below were different than those computed by WWF-EPO. The calculation below only averages considering years of product imported, not the total number of years considered in the range.

and the accompanying table:

Plot tree maps, considering 2017 and the 2005-2016 mean:

And the tables to search this data:

Ranking are then shown in a time series graph with a focus on consumption centers.

With a focus on individual economies within the EU28 (largest economies considered).

and the table for this data:

Insight 2

The majority of the EU’s tropical deforestation footprint is associated with a handful of commodities, localities, and companies

Now we use Trase to zoom into specific regions and to highlight the link between consumption, deforestation risk (and CO2 emissions), jurisdictions and number of companies involved.

We first consider South America (Brazil, Argentina, Paraguay) as a group with its biomes (those for which we have identified deforestation risk). In all analyses:

## # A tibble: 1 x 11
## # Groups:   year, country_of_production [1]
##    year country_of_prod~ commodity   TONS DEF_HA  tCO2 `latest == "TRU~ PCT_TONS
##   <dbl> <chr>            <chr>      <dbl>  <dbl> <dbl> <lgl>               <dbl>
## 1  2015 INDONESIA        PALM OIL  3.87e6 28895.    NA TRUE                  100
## # ... with 3 more variables: PCT_DEF_HA <dbl>, PCT_tCO2 <dbl>,
## #   DEF_PER_KTON <dbl>

We can show either a summary table based on the above, or the following graph to show the benchmarks.

## [1] " soy"

A better Figure was used in Observable to show this: https://observablehq.com/d/20b442874185e459 We then do a count of jurisdictions and traders linked to deforestation for products exported to the EU28.

Determine rankings, per EU country and commodity with impacts per biomes

And now considering the combined EU28 after revised.

## Key message 3

Zero deforestation commitments and commodity certification are highly variable, are patchy in their coverage and have yet to demonstrate a clear impact

Now we look at commitments, referring to the Amsterdam Declaration on Deforestation (with reference to figures in Key Message 1) and showing specific data on Zero Deforestation Commitments (ZDCs) by the private sector.

References

Kastner T et al (2011) Tracing distant environmental impacts of agricultural products from a consumer perspective Ecological Economics 70(6): 1032–1040, doi: 10.1016/j.ecolecon.2011.01.012

Pendrill F et al (2020) Deforestation risk embodied in production and consumption of agricultural and forestry commodities 2005–2017 (Version 1.0) [Data set] Zenodo: http://doi.org/10.5281/zenodo.4250532

END