Here are posts from Francois and Dano on Tilman et al.
Francois posted up the following lengthy comment on Tilman et al on another thread.
Background: In a recent discussion, one poster (Dano) pointed to a paper published in Science in 2001, that presumably shows how human activities other than GHG emissions have potentially dangerous effects on climate, and the environment in general. Upon a first reading of that paper, I noticed how scientifically weak it was, and in a further exchange with Dano, I offered to post a detailed critique. Here it is.
I am not a specialist in ecological sciences. However the paper’s weaknesses are immediately obvious to anyone who has written or reviewed scientific papers. I am a scientist by training (Ph.D. in physics), and have pursued an academic career for 10 years after my Ph.D. before moving to industry. I have published 56 peer-reviewed papers, and 65 talks at international conferences, as well as editing 4 conference proceedings. More importantly, I have acted as a reviewer for probably a few hundred papers, and I have sat on conference program committees, and grant committees, where I had to review proposals that were well outside my main line of research. Thus I have learned to read quickly and with a critical eye. My comments below are the ones I would have made, had I been a reviewer for that paper.
The avowed aim of the paper is to “forecast the potential global environmental impacts of agriculture” for the coming 20 to 50 years. The authors warn that “forecasts are not predictions, but rather estimates of environmental impacts should agriculture continue on the trajectories of the past 35 or more years”. To perform their forecasts, the authors have used data on 6 variables over the past 40 years (1960 to 2000), namely: use of Nitrogen, Phosphorus, global irrigation, global pasture and crop land, and global pesticide use. They have used univariate and multivariate trend estimations as a function of time, population, and GDP to extrapolate the values of those six variables in 2020 and 2050. They use those values as a basis for discussing the future environmental impacts.
While the use of univariate and multivariate trend estimations gives a seemingly rigorous and quantified basis for their forecasting, it nevertheless amounts to a rather simplistic procedure. The authors appear surprised that, despite an expected exponential growth, their variables were well fitted by a linear trend. They seem to ignore that an exponential function is well fitted by a linear function over a short enough interval. A close examination of population data by this reviewer showed that they are indeed exponential. However, and importantly for the paper’s argument, the forecasted population growth is not a simple exponential, as growth is expected to slow over the next 50 years. Developed areas, like Europe, will indeed see a population decline. Other observations point to the possible oversimplistic use of linear trends. For example, while crop and pasture lands appear to follow a linear trend from 1960 to 1988, the trend has very clearly slowed down from 1988 to 2000, and indeed there is very little growth during that period. One wonders if this is indicative of a new trend, or just a temporary phenomenon. At the very least, the authors should have asked themselves that question before blindly extrapolating the trend. On the other hand, in the Nitrogen data, that show an 8-fold increase, the exponential shape should have been more evident. That it doesn’t may be indicative of a sub-linear dependence on other variables, such as crop land. They also show data on pesticide imports expressed in dollars, without saying if the relative price per quantity has changed over that period. Finally, one may question the validity of a multivariate analysis when dealing with non-independent variables, such as time, population, and GDP. Furthermore, there are complex interactions between crop land, fertilizer, irrigation and pesticide use, which may hide more complex trends than the ones assumed by the authors. For example, what is the trend for use of pesticide vs crop land area? What does it mean that irrigated land exceeds crop land area? How is the use of fertilizer related to GDP? Is the damage due to pesticide equally bad, or are some less damaging than others, and what are their respective trends? I believe that the author’s oversimplistic analysis is the main reason why their estimates vary by as much as a factor of two, leaving the reader perplex as to the utility (or futility) of the whole exercise.
But that might be forgiven if not for the rest of the paper. Having forecasted increases in all six variables that anyone could have expected simply because the global population is expected to increase from 6 to 9.5 billions in 50 years, the authors then set out to forecasts the impacts on the environment. However, the entire discussion, which indeed constitutes the bulk of the paper, entirely fails to make the point expressed in the conclusion that: “agriculture has the potential to have massive, irreversible environmental impacts”. That is because, after having made a quantified forecast of use of fertilizers, irrigation, pesticides, crop and pasture land, the authors merely assume that because these activities have been shown to cause potential damage, they necessarily will in the future. But to prove their point, the authors needed three more elements. First, they needed to quantify past damage due to those activities. For this, they would need to use objective and quantifiable indicators of environmental damage. Secondly, they should have shown that the amount of damage has indeed followed the same trends as the activities themselves. In other words have the use of mitigation techniques increased over the given period or not? Thirdly, they need to make a convincing case that this same trend will continue in the future. But the authors have done nothing like that. Indeed, they recognize that “methods to forecast quantitatively the impact on ecosystem functioning of loss of habitat, loss of biodiversity, changes in species composition, and increased nutrient inputs need development”. But one can only wonder (and here I show my ignorance of recent progress in the field), what have ecologists done over the past 50 years?! Certainly, developing tools to quantify environmental damage would be much more useful than making hand-waving arguments about “massive impacts”. If there are such tools and the authors have not used them in their analysis, then they are even more guilty of ignoring relevant work, that would have made their analysis much more substantial and credible.
The rest of the paper is really just a general discussion of well known methods to mitigate the environmental impacts of human agricultural activities. For anyone even remotely following those issues, there is nothing fundamentally new there. In fact, an assertion such as “land use and habitat conversion are, in essence, a zero-sum game” makes this reviewer wonder if the authors have ever been to an oasis! One also wonders why the authors fail to address how social and economic factors are likely to affect the trends they discuss.
In the end, the authors arrive at the revealing conclusion that “these solutions will not be achieved unless far more resources are dedicated to their discovery and implementation”. The entire exercise turns out to be a plea for more funding!
A final word about the abstract. After failing to prove their point about massive damage, the authors nevertheless assert in the abstract that: “This eutrophication and habitat destruction would cause unprecedented ecosystem simplification, loss of ecosystem services, and species extinctions.” True or not, there is no way that this alarmist assertion is demonstrated by the bulk of the paper.
In conclusion, this paper, under the veneer of a quantitative and scientific analysis, completely fails to demonstrate the conclusion it asserts. While it is obvious to everyone that the growth of the world population will pose a great challenge to agriculture, and that we must find ways to minimize its environmental impacts, an estimation of those impacts would require a much finer analysis of the complex interactions between agricultural practices, population, and GDP, as well as an actual quantification of past damages and examination of their trend.
At best, this paper would find its place in a promotional brochure for environmental groups. It clearly does not have enough original and significant scientific contents to warrant publication in a serious and prestigious journal like Science.
Here’s Dano’s re-statement of his position:
This entire show started when I agreed with a RP Sr argument that global change has multiple causes and cited Tilman et al to illustrate future effects (hmmm…perhaps I raised hackles by stating that some sector will have to phase out oil first, I dunno).
FO jumped in and called Tilman et al. “Åalarmist’ and I asked him to back his claim, whereby he wrote a review absent any scholarship, citations or basic knowlege of the discipline showing how Tilman et al were incorrect. That is: despite the 750 words, he didn’t show how they were “Åalarmist’ – he merely didn’t like their methods (but didn’t show how his were better)…
Among the nonalarmist conclusions of Tilman et al. were that:
o 1B ha of land will be needed to feed 3B people.
o (236*10^6 MT) N, (83.7*10^6 MT) P, (530.89*10^9 ha) irrigated/pasture land, (10.1*10^6 MT) pesticide will be required in 2050.
o Should trends [projections for 2050 range from 1.9- to 4.8-fold increases] continue, by 2050, humans and other organisms in natural and managed ecosystems would be exposed to markedly elevated levels of pesticides.
o The combined total represents an average global agricultural land base in 2050 that would be 18% larger than at present.
o Just as demand for energy is the major cause of increasing atmospheric greenhouse gases, demand for agricultural products may be the major driver of future nonclimatic global change.
o The projected 50% increase in global population and demand for diets richer in meat by a wealthier world are projected to double global food demand by 2050 (N. Alexandratos, Proc. Natl. Acad. Sci. U.S.A. 96, 5908 (1999)).
o [and then concludes that various adaptive governance plans may be needed for mitigation]
Kevin justified a separate thread as follows:
I think the latter half (as well as the first half) of this thread has actually been quite interesting and informative. Although strictly speaking it is not a GW subject, the discussion of the Tilman (and the refuting of its conclusions by Willis) has been quite illuminating with some good links tht are well worth following. Perhaps as I’ve seen you do with other threads it would be a good idea to pull most of it out onto a separate thread…..
OK, here’s a separate thread continuing the discussion from here. I do not vouch for either position.