Appendix C. Description of the data sets on body size and data analysis
Datasets reporting oxygen clines and patterns in body size
For our re-examination of the pattern in amphipod maximum body size we took the upper 95th percentiles of body size spectra in aquatic amphipod assemblages previously reported (Peck & Chapelle 2003; Chapelle & Peck 2004). In these publications updated information on size spectra is reported for a number of sites (Chapelle & Peck 2004) as well as new data for the high altitude Lake Titicaca (Peck and Chapelle 2003). However, it should be noted that the size data reported in the original publication (Chapelle & Peck 1999), provided at least as good, if not better, a fit with our index as those subsequently reported.
Predicting maximum amphipod body size
Oxygen availability is expected to act as a physico-chemical ceiling, limiting the maximum size that can be attained by amphipods (Chapelle and Peck 2004). In addition to environmental oxygen supply, organismal oxygen demand has been argued, on theoretical grounds, to shape the maximum body length attainable by aquatic ectotherms. Accordingly, the size difference between sites can be calculated with the following equation (Makarieva et al. 2005):
Length ratio = Lmax at site 1 / Lmax at site 2 = (fO2 at site 1 / fO2 at site 2) / (Q10(Tsite1 - Tsite2) / 10ēK)) (C.1)
where Lmax is the maximum body length (mm) and fO2 expresses the flux of supplied oxygen. Different temperatures and salinities between sites thus lead to differences in environmental oxygen supply whilst differences in the oxygen demand of the species present will be caused by temperature differences across sites. Predicted length ratios were converted to percentages of the mean to facilitate visualization. Note that amphipod body sizes are reported in millimeters, not grams, and thus the dimensions of the size data and Lmax in equation 8 correspond (see Makarieva et al. 2005 for a further discussion). We analyzed the relationship between predicted and observed amphipod maximum body size by either taking oxygen concentration as a measure of fO2, or OSI and again using an overall Q10 value of 2.0.
Effect of salinity on rate of oxygen diffusivity
So far we have assumed that the diffusivity of oxygen does not change with salinity (Stroe and Janssen 1993); increasing salinity will reduce the oxygen diffusivity (Akita 1981), but only marginally, and therefore the predicted differences are small for the range of salinities encompassed by this study. An 8% lower rate of diffusion is reported (Graham 1988) and a 4% lower rate is calculated (Akita 1981) for sea water compared to freshwater. The relationship between observed and predicted maximum amphipod body size was slightly, but consistently, improved when an effect of salinity on oxygen diffusivity was taken into account (y = 0.88·x; Pearson R = 0.906, P < 0.001; assuming the diffusivity is 8% lower in sea water). Incorporating an effect of salinity on the rate of oxygen diffusion therefore did not weaken our results; rather they were further improved.
Akita K. (1981) Diffusivities of gases in aqueous-electrolyte solutions. Indust. Eng. Chem. Fund. 20:89–94.
Chapelle G. and Peck L. (1999) Polar gigantism dictated by oxygen availability. Nature 399:114–115.
Chapelle G. and Peck L. S. (2004) Amphipod crustacean size spectra: new insights in the relationship between size and oxygen. Oikos 106:167–175.
Graham J. B. (1988) Ecological and evolutionary aspects of integumentary respiration - body size, diffusion, and the invertebrata. Amer. Zool. 28:1031–1045.
Peck L. S. and Chapelle G. (2003) Reduced oxygen at high altitude limits maximum size. Proc. R. Soc. Lond. B 270:S166–167.
Peck L. S. and Chapelle G. (1999) Amphipod gigantism dictated by oxygen availability? Reply. Ecol. Lett. 2:401–403.
Makarieva A. M., Gorshkov V. G., and Li B.-L. (2005) Temperature-associated upper limits to body size in terrestrial poikilotherms. Oikos 111:425–436.
Stroe A. Jv. and Janssen L. J. J. (1993) Determination of the diffusion-coefficient of oxygen in sodium-chloride solutions with a transient pulse technique. Anal. Chim. Acta 279:213–219.