My life as told through R

I was recently reflecting on some of the ways that my research has shifted in focus over the past 10 years or so that I have been an independent (or semi-independent) researcher.  It occurred to me that the story of my research can be perhaps most succinctly summarized through changes in the R packages that I’ve used most through time.

When I first started using R (~2006), almost all of my analyses revolved around the ‘VEGAN’ package.  VEGAN is the ‘community ecology package’ and provides user-friendly ‘tools for descriptive community ecology. It has most basic functions of diversity analysis, community ordination and dissimilarity analysis.’  I used vegan to look at basic patterns of diversity and composition on the islands of Lago Guri and then in the large 50-ha forest dynamics plots run by the CTFS (with whom I was doing a postdoc). These analyses all focused around simple presence absence or abundance matrices and looked at core questions in community ecology.

From VEGAN, I transitioned to ‘SPLANCS’.  SPLANCS is the package for ‘Spatial and Space-Time Point Pattern Analysis’.  I had moved into the realm of spatial statistics and was looking at detailed spatial patterns in the CTFS databases.  Eventually the CTFS developed its own package which included many functions built around VEGAN and SPLANCS.

Now, the most common package in my R codes is ‘RASTER’ which is the package for ‘Geographic data analysis and modeling’.  The use of RASTER reflects my shifting focus on larger-scale biogeographic and macroecological questions.  This shift in turn reflects the expanding nature of anthropogenic disturbances. In order to understand how we are screwing with the planet in big ways, we need to do big research.

Beyond just using different packages, there has also been a shift toward writing and developing more and more of my own personal handwritten code.  When I was an R infant working with VEGAN, my analyses were built around the available functions. The questions I asked were constrained by what VEGAN could do for me.  Now I build my own code around the questions I am interested in and use the functions available from VEGAN, SPLANCS, RASTER and any other package, just to flush out the sub routines.  Now my analyses are built around the questions – not the other way around.  So not only does my shifting use of R reflect the changing focus of my research, it also reflects changes in the very way I think and do research.

No avocados (or beer!) for you

The upwithclimate team has been thinking a lot lately about our diet can influence climate change.  But as we have shown previously with our work on bananas, the converse is also true: climate change influences our diet.

As a recent example, the popular burrito chain Chipotle just announced that “Increasing weather volatility or other long-term changes in global weather patterns, including any changes associated with global climate change, could have a significant impact on the price or availability of some of our ingredients…In the event of cost increases with respect to one or more of our raw ingredients we may choose to temporarily suspend serving menu items, such as guacamole or one or more of our salsas, rather than paying the increased cost for the ingredients.”

In another example, a study published in PNAS last year showed that with climate change, the extent of area suitable for viticulture (wine production) is predicted to decrease by “25% to 73% in major wine producing regions by 2050”.  Even under more conservative climate change projections, the extent of wine growing areas is predicted to shrink by 19% to 62% by 2050.

Climate change is also causing reductions in coffee production (more here).

And chocolate

And corn

And wheat

And BEER!!!


Bad Bovines, Part 2

The second of our formal journal responses was published today in PNAS!  It is reproduced below:

Increasing preference for beef magnifies human impact on world’s food web

Kenneth J. Feeley & Brian Machovina

Bonhommeau et al.’s paper, “Eating up the world’s food web and the human trophic level,” (1) provides a valuable perspective on the role of human food consumption within the global ecosystem. However, the ranking of human beings at a similar trophic level as other animal species downplays the effects that humans have on the Earth in comparison to other species. The sheer volume of food consumed by humans and our growing preference for inefficient food sources cause us to have increasingly disproportionate impacts on the global ecosystem in relation to other species, even of the same trophic level.

The importance of dietary preference is exemplified by differences between China and the USA.  China’s HTL increased from 2.1 in 1989 to 2.2 in 2009.  This increase in HTL was driven by a more than doubling of pork consumption in China over the past two decades.  Over the same period total beef consumption in China increased six-fold – from approximately 1 million tonnes in 1989 to >6 million tonnes in 2009 (2). Beef is an extremely inefficient food source; the land area required to produce a kilogram is beef is >2.5 times greater than required for pork and >3 times greater than for poultry (3, 4).  As such, the area of land required to meet China’s food demands is growing at a faster rate than required based on population growth and increasing HTL alone. Roughly three times as much land area was required to meet China’s demand for meat in 2009 as in 1989 (approximately 470,000km2 in 1989 vs. 1,380,000km2 in 2009; estimates based on annual consumption of beef, pork and poultry [2] and global average land requirements for production [3]).  Over this period, China’s population increased by approximately 18% and per capita meat consumption increased by 135%.  Together these two factors account for roughly 90% of the increase in required land area.  An additional 71,000km2 of increased land demand is due to a doubling in the relative consumption of beef (beef constituted 4% of meat consumed in 1989 and 9% in 2009).  In the USA, population size increased by 22% and per capita meat consumption increased by 6% (2).  These two factors alone would have resulted in a 30% increase in the area of land required for meat production. However, in contrast to China, the USA has decreased relative consumption of beef (beef accounted for 40% of meat consumed in 1989 and 33% in 2009).  Consequently, the total land area required to fulfill the USA’s food demands increased by only 21%.

Despite a relatively-low HTL (1), humans have a dominant role in the world’s food web, appropriating approximately 10% of total net primary productivity for food production purposes alone (1, 5). This amount will almost invariably increase in the future due to growing population sizes and concurrent increases in the HTL; this pressure on the Earth’s systems can be further magnified by a rapidly-growing preference for beef and other inefficient food sources.

Reply by Bonhommeau et al. 

1.         Bonhommeau S, et al. (2013) Eating up the world’s food web and the human trophic level. Proceedings of the National Academy of Sciences.
2.         Food and Agriculture Organization of the United Nations (2013) FAOSTAT database.  (
3.         Elferink EV & Nonhebel S (2007) Variations in land requirements for meat production. Journal of Cleaner Production 15(18):1778-1786.
4.         Gerbens-Leenes PW, Nonhebel S, & Ivens WPMF (2002) A method to determine land requirements relating to food consumption patterns. Agriculture, Ecosystems & Environment 90(1):47-58.
5.         Imhoff ML, et al. (2004) Global patterns in human consumption of net primary production. Nature 429(6994):870-873.

Online tools to visualize global climate

In constant evolution, climate represents the long-term mean weather conditions of an area. The ability to determine how natural and anthropogenic influences affect the evolution of climate is not trivial. The understanding of climate requires large historical records at a multiple spatial scales that for non-specialist might be difficult to use because of the large volume of data and the lack of simple and dynamic user-friendly interfaces that integrate these data.

With the intent of allowing not only the scientist, but also the general public to interact (play) with climate models at multiple spatial and temporal scales and explore how the climate is changing through time, several groups have created easy to use internet-based tools. These tools are particularly useful for teaching, but can also help us to identify global patterns and information gaps that can have a considerable impact in the understanding of regional and local climates.

The figure is an example of an illustration produced with Climate Reanalyzer. It represents the change in global temperature from 1979 to 2013. The color spectrum represents temperature change, with negative values representing cooling while positive values warming.
The figure is an example of an illustration produced with Climate Reanalyzer. It represents the change in global temperature from 1979 to 2013. The color spectrum represents temperature change, with negative values representing cooling while positive values warming.

Here are links to some of those tools. I am sure that there are more. Generally each site has detailed information about where do the data come from and how they were processed.

Some illustrations can be quite shocking!