Unveiling temporal variation in dissolved organic matter (DOM) with high-frequency spectroscopic measurements in a shallow eutrophic lake

ABSTRACT

Dissolved organic matter (DOM) plays an important role in biogeochemistry of lake ecosystems. Studies measuring DOM at short intervals in lakes are rare, and thus its short-term dynamics are largely unknown. We investigated DOM temporal variation in large, shallow, eutrophic Lake Võrtsjärv (Estonia) during one growing season (May–Sep 2016) using a field-deployable in situ spectrometer to measure absorbance spectra (wavelength range 200–708 nm) at a 2 h interval coupled with monthly discrete water sampling. Collected spectra were analyzed together with some in-lake variables, lake metabolic rates, and meteorological and hydrological data using boosted regression tree (BRT) and random forest (RF) models. Different spectral parameters were used to assess total and allochthonous DOM quantity and relative share of autochthonous DOM. All parameters (i.e., DOM quantity and quality) varied on a large scale. For example, dissolved organic carbon (DOC) concentrations ranged from 12.0 to 17.3 mg L⁻¹. High levels of DOM were mainly of allochthonous origin, and a strong relationship with inflow indicated the same. The relative share of autochthonous DOM increased with rising air temperature as primary production rose in warm water; however, we found no direct relationships with gross primary production. RF and BRT models explained up to 38% and 63% of DOM temporal variability, respectively. Our results showed that monthly water samples did not capture large variation in DOM. Therefore, high-frequency measurements using in situ spectrometry improve temporal representativeness of DOM monitoring in lakes compared to traditional sampling methods.

KEYWORDS:

• boosted regression tree
• coloured dissolved organic matter (CDOM)
• dissolved organic carbon (DOC)
• in situ spectrometer
• lake monitoring
• random forest

Acknowledgements

The Estonian Ministry of Environment and Estonian Environment Agency supported data collection within the national monitoring program.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Estonian Research Council grants PRG709, PRG1167, and PRG1954, and by Estonian University of Life Sciences grant P210160PKKH.