Thanks to Billy Kingery for calling my attention to a recent paper by Chris Smaje, titled "The Strong Perennial Vision: A Critical Review", published in Agroecology and Sustainable Food Systems. The same journal published responses by Steve Gliessman and by Timothy Crews et al., to which Chris has responded on his blog, Small Farm Future.
These authors have probably thought about these issues more deeply than I have. And I don't want to try the patience of a couple of coauthors who are waiting for revisions on papers. So, for now, I'll just make a few comments on Chris's paper.
Among his points are 1) "there is no reason to suppose that lower-yielding perennial cereal crops cannot play a significant role in the sustainable food systems of the future" but 2) it's unrealistic to expect that they won't be lower-yielding. I tend to agree with both points. In my book, arguing for a diverse agricultural-research portfolio, I wrote:
"As explained in Chapter 7, I have my doubts about perennial grain crops. But do we really want to focus 100% of our grain-breeding efforts on annuals?"
One argument against claims that perennial grains will probably have lower yield is that apples sometimes allocate as much of their photosynthate to fruit as annual grains do to seed, as noted in a paper by Van Tassel et al. Chris points out that the apple trees in question were "physically supported and on strongly dwarfing rootstocks-- [and] are rarely grown commercially for [more than 10 years]." Chris's point is that this apple system isn't reliable evidence that perennial grasses could have high allocation to grain. I agree, but how much could we push the protein and calorie yield of tree crops? For example, could almonds be bred and managed to match the yields of annual grains? Even dwarf trees are taller than most weeds, so weed control could rely more on mowing and/or cover crops than erosion-promoting tillage or herbicides.
Chris notes that:
"some perennials (often C-strategists) can undoubtedly outperform some annuals in yearly resource capture, and this has been shown empirically in some cases, for example, in comparisons of Miscanthus with maize, alfalfa with soybeans, and perennial with annual sorghum (Piper and Kulakow 1994; DeHaan et al. 2005; Dohleman and Long 2009)"
but questions how much "breeders can manipulate plant allocation" to put extra resources acquired by perennials into seeds. He discusses this issue in some detail, but this still seems like an open question. Key to his arguments is Grimes' classification of plant strategies as competitive, stress-tolerating, or disturbance-adapted. If these strategies are really discrete and reflected throughout a plant's genome, I guess it would be hard to change. But given the apparent ease of transitions between annual and perennial (Silvertown and Dodd, 1966, cited by Smaje) or unicellularity to simple multicellularity, I suspect that it may not be too hard to change these strategies through plant breeding.
This wouldn't eliminate tradeoffs between individual-plant competitiveness and whole-crop yield, however. As Chris's paper states:
"It may indeed prove possible to select for high seed allocation in short-lived, C-strategy perennials whose growth habit enables them to accumulate more photosynthate than their annual counterparts, thereby simultaneously funding seed production and perennation. But, as Van Tassel (2012) concedes, these plants will be less competitive than co-extensive wild or weedy species--suggesting that the margin for cultivating them without resort to irrigation, tillage, fertilization, pesticide application and other such practices that perennial cereal cultivation aims to circumvent may be thin at best."
There's much more, including comments on sociological issues, so I expect this paper (and responses to it) will stimulate a lot of discussion.