A series of field experiments explore the characteristic fractionation of the $\sp{13}$C isotope by land plants on the North American continent, as seen in CO$\sb2$ emitting from plants and soils to the canopy layer during a diurnal cycle. CO$\sb2$ concentrations are reported for 495 discrete air samples taken within forests and over tundra at ten, rural sites ranging from 9$\sp\circ$N to 69$\sp\circ$N (Barro Colorado, Panama; Chamela, Mexico; Cuyamaca, CA; Yosemite, CA; Scotia Ridge, PA; Barnard, VT; Hamilton, MT; Rock Lake, Alberta; Bethel, AL; and Toolik, AL).

$\delta\sp{13}$C, $\delta\sp{18}$O and N$\sb2$O concentration are reported for 236 samples of CO$\sb2$ extracted cryogenically from the air samples. The results show the intercepts, $\delta\sp{13}$C$\sb{\rm I}$, of least-squares fits to the isotopic and reciprocal concentration at each site to range progressively from $-$28% near the equator to $-$23% near the Artic Circle. The latitudinal trend toward greater fractionation within the closed, tropical canopy is consistent with previous hypotheses regarding cyclic enrichment and water-use-efficiency relations, but is inconsistent with the hypothesis that $\sp{13}$C enrichment simply follows greater insolation. The mean value found for $\delta\sp{13}$C$\sb{\rm I}$, $-$25% ($\pm$1.6%), is in close agreement with nominal values used in global computer modelling of the biosphere-atmosphere CO$\sb2$ flux. Variability in samples from soil enclosure experiments and between years at some sites suggests that multiple factors may cause spatial and temporal heterogeneity in the $\sp{13}$C fractionation of as much as 2% to 3%.

Anomalous N$\sb2$O or $\delta\sp{18}$O values identify 90% of the $\delta\sp{13}$C data departing significantly ($>$2 sigma) from the least-squares fit for each site. N$\sb2$O concentrations range from 267 ppb to 3,882 ppb, while N$\sb2$O corrections to $\delta\sp{13}$C range from +0.06% to +1.95%. 20% of all samples require N$\sb2$O-based correction to the $\delta\sp{13}$C data that depart from the nominal +0.23% correction by more than 1%, suggesting that applying a constant correction for N$\sb2$O, or no correction at all may expose such assessments of characteristic isotopic composition in biospheric-atmosphere CO$\sb2$ exchange to an additional uncertainty exceeding 1%.