Response to Comment: Terrestrial support of pelagic consumers in unproductive lakes—Uncertainty and potential in assessments using stable isotopes

Abstract

The use of stable carbon isotopes (d 13 C) has played a key role in estimation of the proportion of aquatic consumer biomass derived from terrestrial organic matter (OM; i.e., allochthony; Meili et al. 1996; Grey et al. 2001; Pace et al. 2004). However, the use of d 13 C for assessing allochthony has shortcomings because of the small natural separation between terrestrial and aquatic isotopic end members and the difficulty in physically separating autotrophic phyto-plankton for d 13 C analysis from other components of particulate organic carbon (POC). These shortcomings are especially problematic in unproductive lakes where the phytoplankton are dominated by small and mixotrophic species, and where the internal photosynthesis is low compared to the input of terrestrial OM (Algesten et al. 2004; Jansson et al. 2008). Several alternative analyses and approaches have been tested to overcome these methodo-logical limitations, including compound-specific analyses of phytoplankton biomarkers (Pace et al. 2007; Van Den Meersche et al. 2009; Berggren et al. 2014), manipulation of phytoplankton d 13 C by addition of 13 C-labeled dissolved inorganic carbon (Pace et al. 2004; Taipale et al. 2008), addition of 13 C-enriched OM (Karlsson et al. 2007; Bartels et al. 2012), and various mass balance and modeling approaches (Marty and Planas 2008; Mohamed and Taylor 2009; Berggren et al. 2010). Still, a generally applicable method is lacking, implying that the problems with assessing d 13 C of phytoplankton is a major limitation in the use of d 13 C for estimating allochthony with the accuracy needed for detailed understanding of food web dynamics. As a consequence, researchers have started to use alternative isotopic tracers and especially the stable hydro-gen isotope (d 2 H). The terrestrial and aquatic end-member d 2 H values have been shown to be clearly separated (Doucett et al. 2007), and the method has been used to quantify allochthony of consumers in lakes where the use of conventional d 13 C analysis has been problematic (Cole et al. 2011; Solomon et al. 2011; Wilkinson et al. 2013a). Although issues related to, e.g., dietary water and analytical procedures need to be carefully addressed to reduce uncertainties in estimates of allochthony using d 2 H (Solo-mon et al. 2009; Soto et al. 2013), the technique has a large potential for advancing our understanding of patterns and magnitude of allochthony in consumers of many aquatic systems. In Karlsson et al. (2012), data on metabolism, d 2 H, and stable nitrogen isotopes were used to show that terrestrial OM supported all parts of the food web in a humic lake (Upper Bear Lake in boreal Sweden). These results suggest preferential use of autochthonous-based resources and that there is an upper level of terrestrial support of higher trophic levels in lakes. They then went on to compare these results with those concurrently obtained using d 13 C analyses. Brett (2014) questions the d 13 C of phytoplankton used for estimates of consumer allochthony in Upper Bear Lake and, thereby, also the ''key results'' of the study. However, Karlsson et al. (2012) based the key results on d 2 H and further explained the choice of d 2 H over d 13 C by the fact that the d 13 C approach has limitations due to small separation between terrestrial and aquatic primary producers and the fact that the d 13 C of phytoplankton is variable and difficult to measure. The Karlsson et al. (2012) paper then compared the findings regarding allochthony based on d 2 H with allochthony estimated from d 13 C data using previously published end-member d 13 C values from similar lake types in boreal northern Sweden. Only zooplankton was included in this comparison because it is not possible to estimate allochthony for zoobenthos and fish by using the d 13 C data. Karlsson et al. (2012) did not attempt to estimate the d 13 C of end members in Upper Bear Lake but rather compared values with results obtained using the end-member d 13 C used in previous studies. Such a comparison we believe is of interest for the scientific community. The comparison shows a close relationship between the two methods. Thus, the point by Brett (2014), i.e., that allochthony may be overestimated by assuming too high 13 C depletion in phytoplankton, cannot be reconciled with the fact that the d 2 H data show similar results as the 13 C data. Brett's comment criticizes the method in which Karlsson et al. (2012) used an unusually depleted phytoplankton d 13 C value, which was taken from Karlsson et al. (2003). Brett (2014) points out that Karlsson et al. (2003) used a too-high 13 C fractionation factor (e p 5 223.7% for forested lakes) for phytoplankton photosynthesis com-pared to the ''contemporary literature'' (the Karlsson et al. [2012] article did not speculate on the e p). We are aware that the e p in Karlsson et al. (2003) is high when compared with many other studies, but we also note similarities with a recent study of 18 boreal clear-water and humic lakes in Canada, where e p was estimated from d 13 C of phytoplank-ton biomarker fatty acids to 221.8% 6 3.6% (Berggren et al. 2014). However, the d 13 C of phytoplankton is not only dependent on e p , but also, to a large extent, on the d 13 C-CO 2 , which varies largely between lakes. Our point here is that d 13