An assessment of assumptions and uncertainty in deuterium‐based estimates of terrestrial subsidies to aquatic consumers

Abstract

The deuterium ratio ( H/ H) in tissue is often used to estimate terrestrial subsidies to aquatic consumers because of strongly differentiated values between terrestrial and aquatic primary producers. However, quantitative deuterium-based analyses of terrestrial resource assimilation are highly dependent on several poorly defined assumptions. We explored the sensitivity of these estimates to assumptions regarding environmental water contributions to consumer deuterium content (ω) and algal photosynthetic hydrogen discrimination (ε ). We also tested whether C/ C and H/ H-based estimates of terrestrial resource assimilation give similar outcomes. The average of the 12 experiments that have directly estimated proportional contributions of environmental water to consumer tissue H/ H was 0.27 ± 0.11 (mean ± SD), with similar values for invertebrates and fish. Conversely, of the 28 field studies that have used H/ H to characterize aquatic food webs, all but one assume a value that is less than our current best estimate, usually substantially less. A reanalysis of the raw data from four recent case studies indicates the calculated terrestrial contribution to aquatic consumers is extremely sensitive to this assumption. When the authors' original assumptions were used (i.e., ω = 0.16 ± 0.05), the estimated proportional contribution of terrestrial resources to aquatic consumers (θ ) averaged 29 ± 17%, and when ω = 0.27 was used the average estimated assimilation of allochthonous resources was ≈0.00. A compilation of published photosynthetic hydrogen discrimination values for microalgae averaged ε = -150 ± 27‰ (SD, n = 99), and a sensitivity analysis showed the outcomes of these calculations were also strongly influenced by uncertainty in ε . There was no statistical association between C/ C and H/ H-based estimates of terrestrial subsidies (r = -0.12, n = 274). This analysis indicates that the assumptions in deuterium-based estimates of terrestrial resource assimilation are highly influential but poorly constrained; therefore, the impact of these assumptions on calculated outputs must be carefully assessed and thoroughly reported. Due to the highly uncertain assumptions inherent in deuterium-based analyses, we urge much more caution when using this approach to estimate terrestrial subsidies to consumers in aquatic ecosystems.

Terrestrial contributions to Afrotropical aquatic food webs: The Congo River case

“…When both δ 13 C and δ 2 H data were included, the separation between primary sources in the Congo River mainstream was more efficient and the mixing model estimated a strong contribution of terrestrial C 3 plants to both aquatic invertebrates and fish (mean, 74% and 52%, respectively; Figure ). Unfortunately, we did not have δ 2 H measurements for algal material, either as phytoplankton or periphyton, or alternatively any robust primary consumer to serve as a unambiguous algae grazer or planktivore, which only directs us to assume the isotopic composition of algae to be close to −150 ± 27‰ (Brett et al, ). Although the empirical values of this source are unknown in the Congo River, we do not expect large changes on our model estimations as demonstrated by the sensitivity analysis performed using contrasting δ 2 H values for the algal source (Table ), which supports our model assumptions (Brett et al, ).…”

“…Unfortunately, we did not have δ 2 H measurements for algal material, either as phytoplankton or periphyton, or alternatively any robust primary consumer to serve as a unambiguous algae grazer or planktivore, which only directs us to assume the isotopic composition of algae to be close to −150 ± 27‰ (Brett et al, 2018). Although the empirical values of this source are unknown in the Congo River, we do not expect large changes on our model estimations as demonstrated by the sensitivity analysis performed using contrasting δ 2 H values for the algal source (Table S2), which supports our model assumptions (Brett et al, 2017).…”

“…This algal production included potential pelagic (phytoplankton) and benthic (periphyton) microalgae together because unfortunately we were not able to obtain a pure sample for any of these items in our sampling. The reported algal discrimination from water or DIC for either phytoplankton and periphyton was reported to be similar for δ 2 H in a meta-analysis study (Brett et al, 2018), but some differences (~4‰) can be found for δ 13 C (Vander Zanden & Rasmussen, 1999). To accommodate for the additional uncertainty associated with the variability of C isotope fractionation during photosynthesis, the DIC isotopic variation was then increased by 50%.…”

Fruit of the forest ‐ larval sea lamprey Petromyzon marinus are fuelled by allochthonous resources

“…All trophic enrichment factors were doubled when modelling the resource use of fishes other than ammocoetes, as these fishes were all secondary consumers. Aquatic consumers obtain approximately 28% of their hydrogen from drinking water, therefore the consumer data must be corrected to account for this prior to estimating the dietary contribution (Brett et al, ). Although water samples were not obtained when sampling, we estimated that the values would be approximately −60 ‰, equivalent to values recorded for precipitation in Fredericton, New Brunswick, (−55 ± 10 ‰, unpubl.…”

“…Consequently, algae growing within a river and leaves growing on trees on the riverbank often have δ 2 H isotope ratios that differ by over 100‰ (Tanentzap et al, ). As the fractionation in δ 2 H between consumer and prey is small by comparison, δ 2 H is a potentially useful marker to distinguish between allochthonous and autochthonous contributions to lotic food webs (Brett et al, ; Tanentzap et al, ). As consumers' stable isotope ratios reflect their diet during the preceding 1–6 months, depending on the turn‐over rate of specific tissues (Vander Zanden et al, ), seasonal sampling provides an opportunity to ascertain temporal changes in consumers resource use (Hayden et al, ).…”