Experimental design : scaling up in time and space , and its statistical considerations

  • J - Ieincmeyer, Mark us Reichstein
  • Published 2011


Accur'.ltc measurement of t he soil C02 emu x is critical for the assessment of lhe carbon budget of terrestrial ecosystems, since it is the main pathway for assill1­ ~ilated carbon to return to the atmosphere, and only small changes in the soil COz effl ux rate migh t ha\'e important implications on the nct ecosystem carbon bal­ ance. Due to thi s central role ill the terrestria l carbon cycle, soil CO2 efflu x has been measured throughout all biomes, and covering all principal vegetation t) pes. Using simp lified regressions of soil CO2 efflux mea­ surements reported in the scientific literature, the total amount ofc:ubon emitted as C02 by soils world" ide has been es timated at approximately 68-80 Pg (1995; Raich el fl l. , 2002), represenri ngthe second largest carbon flux between ecosystems and the atmosphere. Th is amount is more than ten times the current rate of fossil fuel combustion and indicates tha t each year aroun d 10% of the atmosphere's CO2 cycles ttu'ough the soil (Prentice et al., 2001). Thus, even a small change in soil respi­ ration could significantly intensify, or mi tigate, current atmospheric increases Ofe02, with potential feedbacks to climate change. In f:let, soils store more than twice as much carbon globall) than the atmosphere (Bolin, 2000) an d consequently contain a large long-term potentiall for the carbon cycle cl imate feedback . . \pplying results from small-scale experi ments to larger areas is neces­ sary in order to undersrand the potential role of soils in sequestering or releasing carbon under changed cli­ maric conditions, and LO inform management and IDolicy makers about likely consequences of land-use changes on carbon fluxes and stocks in specific regions. As in the example of the global estimate mentioned above, there is inevitably a need to estimate the soil CO2 efflux over vast areas thaI it is im possible to cover appropriatel " b) actual measurements, and for time scales beyond the scope of measured data, particular! :;. where fut ure predictions are required. Scaling up from sparse and infrequent measure­ ments to the level of, fo r example, catchment, region or even continental or global scales, bears a considerable degree of uncertainty, making such extrapolations d if­ ficult. The scope of rhj s chapter is to in troduce a range of requiremen ts thal are crirjeaJ to faciljtate meaning­ ful extrapolation of results obsened on small scales to allow making estimates of soil CO2 efflux over larger areas and longer time scales. The aim is to provide a general overview in order to enable the reader to design a suitable measuring strategy towards such extrapola­ tions, mainly at the plot and landscape scale. .\Ileas lI ring techniques can be broadly di\ ided into (I) chamber-based, (2) soil profile and (3) eddy covari­ ance approaches. Chamber-based measurements pro­ vide by fa r the majority of publisht:d results, and the general considerations of heterogeneity are valid for all measuring approaches. "Ve therefore concentrate on chamber-based measurements to illusu'ate experimen ­ tal designs for dealing with natural variations in soil CO2 effl ux. Soil profile methods allow a vertical resol ution of the origin of su rface fl ux contributions, thus provi d­ ing critical insight into carbon allocation within soils by roots and contributions to the heterotrophic flu x com­ ponent for different soil depths. However, soil profiles inherently create considerable disturbance both during installation and sampling (Fang and Moncrieff, 1998) and are difficult to replicate within plots. For successful applications of this technique, please refer to Tang et al. (2003), L iang et al. (2004) and D a\'idson et al. (2006). Eddy coyariance has been applied to measure soil sur­ face CO2 flux with some success (see c.g. L aw et al., 1999; Janssens et aI., 2000; Wilson and J\ leyers, 2001; Subke and Tenhunen, 2004). This technique has the

4 Figures and Tables

Cite this paper

@inproceedings{Ieincmeyer2011ExperimentalD, title={Experimental design : scaling up in time and space , and its statistical considerations}, author={J - Ieincmeyer and Mark us Reichstein}, year={2011} }