Forest annual carbon cost: comment.

Understanding the relative influence of abiotic and biotic forces on ecosystem-level processes across broad scale remains a central question in global ecology (Schimel et al. 1996, Chapin et al. 1997, Kerkhoff et al. 2005). In their report, Piao et al. (2010) addressed how global-scale variation in annual terrestrial autotrophic respiration, Ra, varies across broad-scale gradients including temperature, biomass, and successionary age. In addition, they purported to test several predictions and observations from metabolic scaling theory, MST (West et al. 1997) on how temperature and autotrophic biomass influence rates of ecosystem metabolism and production (Enquist et al. 2007b). While I agree with Piao et al.’s emphasis on the need to assess variation in ecosystem processes across broad gradients, I question their methodology for comparing and standardizing rates of ecosystem production and disagree with their reading of MST. Issues of how to standardize flux measures in order to compare annual and instantaneous rates across sites are not just specific to Piao et al.’s study. These issues also apply to other studies assessing spatial variation in ecosystemmetabolism across broad spatial gradients (for example, see Beer et al. 2010) and are central to how we understand and quantify the relative influence of abiotic and biotic forces on the ecosystem-level processes across broad-scale gradients as well as how to use cross-site analyses to inform predictions for a warming world. There are four specific issues that can influence Piao et al.’s central conclusions: 1. Piao et al.’s methodology for comparing fluxes between sites is likely biased, because they did not correct for differences in growing season length or properly control for the scaling effects of biomass on ecosystem metabolism.—Piao et al. observed that the annual respiration of forests increased with temperature. Piao et al. also claimed that this finding is in contrast to the findings of Enquist et al. (2007b). However, one cannot compare these two studies because Piao et al.’s methodology does not follow that of Enquist et al. Specifically, Kerkhoff et al. (2005) and Enquist et al. (2007a) argued that in order to mechanistically assess the role of temperature on more instantaneous rates of ecosystem metabolism it is important to also correct for tree biomass and growing season length. Enquist et al. showed that these corrected rates of tree growth showed little to no signal with growing season temperature. There are several reasons why studies that use annual measures to compare differing sites in order to infer how climate influences ecosystem performance will be biased. By using annual temperature, the actual temperature under which most of the flux occurs will increasingly be underestimated for colder sites because biological activity is greatly reduced so that the flux is effectively ‘‘off’’ during the dormant season (see also Savage 2004). Next, any relationship between a rate, like autotrophic respiration, and some environmental attribute such as temperature depends on the timescale of measurement (see also discussion in Mahecha et al. 2010). Annual ecosystem fluxes, such as net productivity or respiration, are measured by summing more instantaneous measures throughout the year. Rates of ecosystem respiration are tightly coupled to rates of gross primary production (see Vargas et al. 2010). Similarly, the annual carbon balance of a site is constrained by the length of time autotrophs have to assimilate carbon during the year (Cannell and Thornley 2000, Vargas et al. 2010, Berdanier and Klein, in press). As a result, annual respiration measures will be limited by the gross production that occurs during the length of the growing season. The use of annual fluxes can underestimate instantaneous rate measures, especially when one does not account for the (sometimes appreciable) dormant season length (Chapin 2003, Allen et al. 2005, Kikuzawa and Lechowicz 2006). From these points, both the fluxes and temperatures of the colder sites used by Piao et al. will be underestimated by using annual values. Thus, the use of annual measures to assess physiologically based models, even one as simple as MST, will not be applicable. As a result, the methodology used by Piao et al. does not provide a strong assessment of physiologically based models such as MST, which focus on the controls on more instantaneous rates, and its application to the carbon balance of forests. To highlight the above issues I assessed growing season length and compiled ecosystem data across a similar temperature gradient presented by Piao et al. Across a broad latitudinal and temperature gradient, the length of the growing season changes (Fig. 1). Growing seasons vary from 12 months to as little as 4 months or Manuscript received 1 October 2010; accepted 14 January 2011; final version received 4 April 2011. Corresponding Editor: J. B. Yavitt. 1 Department of Ecology and Evolutionary Biology, University of Arizona, BioSciences West, Tucson, Arizona 85721 USA. 2 The Santa Fe Institute, 1399 Hyde Park Road, Santa Fe, New Mexico 87501 USA. 3 E-mail: [email protected]