The following is a comment made at SkS on the topic of how much new agricultural land will be made available in northern latitudes by global warming. It relies heavily on just one, twenty year old reference in an area in which I lack significant knowledge. It is posted primarily to provide a reference to two key sources of relevant data. I think my discussion is reasonable, but it should only used to provoke thinking on the topic rather than to take home conclusions. 

The comment follows in the sequence:

Bob Loblaw

Tom Curtis

Tom Curtis

Bob Loblaw

Tom Curtis
Bob Loblaw

"Bob Loblaw @47, your first link is indeed the map I was indicating.  I have since found a more suitable map (the third map from this site to which you have previously linked), as it shows the actual soil qualities.  Of particular interest in your dispute with Donny is the large swathes of soil in the east of Canada which have "No Capability for arable culture or permanent pasture" (pink) despite being at the same latitude as the Canadian prairies, and hence are not restricted by climate factors.  

Equally interesting are the areas of arable land (some of the highest quality) along the Peace river (north west Alberta) and the single mapped section in north east British Columbia showing a complex interlacing of arable soils (brown) adjacent to a large section of peats (black) which are unsuitable for agriculture.  The high latitude of these sections relative to the northern limit of the Canadian prairies shows that northern limit to have been set by soil type rather than climate.

Those examples, however, also show where, and how we can expect to find suitable soils in Northern Canada with increased global warming.  Both represent deposits of river silt, a process that produces suitable soil for agriculture almost regardless of climate (ie, soil in which agriculture is only restricted by climate).  There presence indicates that such suitable soils are also likely to be found in the river valleys of  the MacKenzie and other smaller northern rivers descending from the Rockies.  Comparison with the soil map of a Canada (first map, previously linked site) does show the relevant reaches of the McKenzie have luvisolic soils, and thus they are likely to provide small regions of class three or four soils (as with BC reaches of the Peace, which has a similar soil classification).  That is, the land will be potentially arable, but will require extensive conditioning to control soil pH (if I understand the description correctly).

As an aside, a more detailed soil map of Canada, and indeed of Siberia as well, can be found in the "Soil Atlas of the Northern Polar Region" (280 MB PDF; website).  Unfortunately the basis of classification in that atlas does not lead directly to an ability to classify as to whether or not the soil is arable, at least for inexpert commentors like me.

With regard to Mills (1994), closer reading shows that he finds 16 million Hectares of arable land that would be opened up from a doubling of CO2, but that the majority (15.9 million Hectares) of that will be in Alaska, while only 0.2 million hectares will be within the study area within Canada itself (page 122).   The study area within Canada is quite restricted (fig 1), so that some more might be expected in Canada, but arguably not a lot as the most suitable sites on geographical grounds are also the most likely to have been surveyed.  This difference between Alaska and Canada is consistent with the hypothesis above that river valleys will produce arable land even in climates currently to cold to sustain cropping, along with the size of the Yukon river valley.

I am not sure if this additional information reconciles you to the Mills (1994) results, but regardless I will accept the result of a peer reviewed study over the anecdotal evidence of a non-expert 100% of the time.  If you think Mills is wrong, find a more recent peer reviewed study showing that he is wrong."