Data presentation in Santa Catalina Mountains plant study misleading

Here is a good example of how data presentation can give one a misleading impression of what is really happening.

A University of Arizona press release claims: “Warming Climate Pushes Plants Up the Mountain.”  That refers to a recently published paper by Richard Brusca et al. “Dramatic response to climate change in the Southwest: Robert Whittaker’s 1963 Arizona Mountain plant transect revisited.”  See complete paper here.  (The older paper referred to is Whittaker, R. W., and W. A. Niering. 1964. Vegetation of the Santa Catalina Mountains, Arizona. I. Ecological classification and distribution of species. J. Arizona–Nevada Acad. Sci. 3:9–34.)

The Mt. Lemmon/Catalina highway has elevation markers every 1,000 feet from the Tucson Valley to the top of Mount Lemmon, spanning over 6,000 feet of elevation change.  Whittaker recorded plant species which occurred in each 1,000-foot division in 1963.  Brusca repeated the survey in 2011 to see if plant species changed elevation in response to climate change.

Before I comment further, please look at Brusca’s graphical comparison of the two surveys below to see what first impression you get.  Here is Brusca’s caption to the graph:

“Figure 1. Summary of elevation range of the 27 most common upland montane plants along the Catalina Highway. White bars are 1963 elevational range data from Whittaker and Niering (1964), the two terminal (stippled) 1000-ft bands denoting Whittaker’s upper- and lowermost 1000-ft vegetation bands. Black bars represent 2011 elevation data from this study. To be as conservative as possible, a change in a species elevation limit (high or low) was noted only if that species was found outside (above or below) the upper- or lowermost 1000-ft band. Thus, if anything, we underestimate the elevational change in the species since 1963 (see Materials and Methods). Following this protocol, 15 species show an unambiguous increase in lower elevation, four show an increase in upper elevation, and eight show a decrease in upper elevation.”

Brusca plant study

My first impression was that yes, indeed, 26 of the 27 plant species had their lowest first occurrence at a higher elevation in 2011 compared to 1963, thereby confirming the hypothesis.  But then I read Brusca’s paper.

Here is the revelation: Whittaker did not record actual elevations of plant occurrence.  He recorded only their presence or absence within a particular 1,000-foot elevation range.  That means for Whittaker if, for instance, a plant occurred at an elevation of 6,001 feet, it would be recorded the same as if it first occurred at 6,999 feet.  In both cases the 6,000-7,000 foot range bar would be filled in.

Brusca recorded the actual elevations and his black bars in the graph reflect that.  His presentation  of the data is not equivalent to Whittaker’s; he therefore compares “apples and oranges.”   If Brusca had presented his data the same way Whittaker had, the graph would look much different. For instance, look at the top green section.  If Brusca had presented his results the same way Whittaker did, then the second, third, fourth, fifth, seventh, and ninth bar comparisons would be identical rather than showing a rise of lowest elevation for the species.  Overall, by my count, 11 of the 27 species have increased their lowest elevation, but looking at the graph, one gets the first impression that 26 of the 27 species increased their lowest elevation.

If the difference in presentation were corrected, then of the eight species in the top green section only two would show an increase in lowest elevation rather than seven as implied by the presentation.

There is another problem also.  Whittaker started his survey at 2,500 feet, but Brusca started at 3,500 feet.  That affects the appearance of several comparisons in the bottom pink range of the graph.  One additional possibly confounding piece of information: Brusca wrote, “we did not sample in any areas of significant forest fire history since 1963.”  Where some plants growing at certain lower elevations, but destroyed by fire?

The thesis of the Brusca paper is stated in the first sentence of the abstract: “Models analyzing how Southwestern plant communities will respond to climate change predict that increases in temperature will lead to upward elevational shifts of montane species.”  Athough Brusca presents a graph showing increasing temperatures at the Tucson airport (probably due to urban heat island effect) he presents no evidence that the temperature in the mountains has in fact increased.

My overall impression is that the results are not as dramatic as stated.  For some perspective, however, one should note that during the last glacial epoch, the Tucson Valley was forest and grassland, and the Clovis people were hunting mammoths in the valley.  As the climate warmed, temperate species retreated to the higher mountain elevations to produce our “sky islands.”

See another critique of this paper at WUWT here, which concludes” Overall, while it is an interesting study and I commend them on repeating the earlier transect, I’d say they have totally failed to demonstrate that the plants are responding in any meaningful or predictable way to the vagaries of the local climate …” In a comment to that post, Dr. Tim Ball points out that “It virtually ignores the role of precipitation, especially in determining lower level boundaries. It also ignores the aspect of the mountain slope that is a south facing slope in the Northern Hemisphere have very different evapotranspiration rates. Also there is the effect of cold air drainage that can create cooler night time conditions in lower levels. These are just a few of the factors that can explain what is happening and apparently ignored.”

Advertisements