The development and validation of remote-sensing-based approaches for the
retrieval of chromophoric dissolved organic matter (CDOM) concentrations requires a comprehensive understanding of
the sources and magnitude of variability in the optical properties of
dissolved material within lakes. In this study, spatial and seasonal
variability in concentration and composition of CDOM and the origin of its
variation was studied in Lake Balaton (Hungary), a large temperate shallow
lake in central Europe. In addition, we investigated the effect of
photobleaching on the optical properties of CDOM through in-lake incubation
experiments. There was marked variability throughout the year in CDOM
absorption in Lake Balaton (
There are approximately 117 million lakes on Earth greater than 0.002 km
The optical properties of lakes provide particularly useful metrics for measuring ecosystem change (Vincent et al., 1998) as they not only convey information on the quantity of particulate and dissolved material but also its quality (Williamson et al., 2014). Furthermore, understanding how the optical properties of particulate and dissolved material in lakes influence the underwater light field and water-leaving radiative signal is important for the development and application of remote sensing techniques for lake monitoring and assessment, but also for their application to lake carbon studies.
Much of the dissolved organic matter (DOM) found in lakes typically represents between 90 and 100 % of the total carbon pool (Wilkinson et al., 2013) and is derived from terrestrial inputs, transported through streams, rivers and wetlands. This allochthonous component of the DOM originates from soils, sediments and plants and is primarily composed of humic substances. The autochthonous fraction of DOM is produced mostly by phytoplankton, zooplankton and bacterioplankton and is largely composed of fulvic acids, carbohydrates, amino acids, proteins, lipids and organic acids.
Chromophoric dissolved organic matter (CDOM) is the coloured fraction of DOM. It is one of the dominant colour-forming constituents in lakes: it not only exerts a strong influence over the underwater light field and water-leaving radiance but it also has a number of important ecosystem functions. First of all, it absorbs light strongly in the ultraviolet (UV) spectrum, limiting the penetration of biologically damaging UV-B radiation that provides protection for phytoplankton and other primary producers (Hoge et al., 1995; Laurion et al., 2000; Zhang et al., 2007a; Williamson et al., 2001). In addition, CDOM can be remineralised by bacteria acting as a source of inorganic nutrients (Buchan et al., 2014), which is important for phytoplankton nutrition, thus fulfilling an important role in the development of phytoplankton blooms and lake metabolism more widely. On the other hand, studies have also shown that light absorption by CDOM can reduce the amount and quality of photosynthetically active radiation (PAR) available to phytoplankton, thereby decreasing primary production and constraining lake metabolism (Kirk, 1994; Laurion et al., 1997, 2000; Vähätalo et al., 2005). Moreover, its conservative properties with dissolved organic carbon (DOC), means CDOM is often used as a proxy for DOC. Thus, there is substantial interest in the use of CDOM as an optical tracer of DOC due to the importance of the latter in regulating physical, chemical and biological properties of lakes. It is therefore important that we develop a better understanding of the optical properties of CDOM and how these relate to the chemical composition and concentration of DOM whether driven by changes to source relationships or through the in-lake processes and transformation of the carbon pool.
Understanding how the optical properties of CDOM vary both temporally and spatially within lakes and how the observed variability influences the underwater light field is of particular importance for the development and validation of remote-sensing-based approaches for retrieving CDOM concentrations. The recent launch of new satellite missions (e.g. Sentinel-2 and -3), allied with the prospect of new hyperspectral sensors (e.g. EnMAP), has provided a new impetus for the development and application of remote sensing techniques for the assessment and monitoring of inland water quality. However, CDOM is arguably the most challenging water quality parameter for the reliable estimation of remotely sensed observations (Palmer et al., 2015) and, in spite of its importance to the physical, chemical and biological function of lakes, it remains one of the least studied parameters. Indeed, few studies have explored the application of remote sensing for the estimation of CDOM in lakes. To progress with such research, an improved understanding of the spatial and temporal variation in the optical properties of CDOM in lakes is needed.
CDOM concentration is commonly measured by its absorption coefficient
(
In addition, Weishaar et al. (2003)
introduced the specific UV absorbance parameter (SUVA
The compositional properties of CDOM vary over time in response to processes
such as microbial decomposition and exposure to UV irradiation. Previous
studies have shown the latter process, first described by Wipple
(1914) as “photobleaching”, plays a major role in the transformation of DOM
in natural waters. Exposure to solar irradiance has been shown to reduce its
capacity to absorb light; the loss of absorptivity is linked to a reduction
in molecular weight (MW), alteration of its chemical composition and an
increase in the bioavailability of DOM
(Geller,
1986; Keiber et al., 1990; Wetzel et al., 1995; Lindell et al., 1995; Corin et al., 1996; Reche et al., 1998) with implications for lake metabolism.
Most previous studies on the origin, distribution and degradation of DOM and how it influences the optical properties of CDOM have been undertaken in oceans (Andrew et al., 2013; Matsuoka et al., 2014; Hancke et al., 2014; D'Sa et al., 2014), coastal waters (Stedmon et al., 2000; Vantrepotte et al., 2007; Kutser et al., 2009; Para et al., 2013) or in high-latitude lakes (Ficek et al., 2011; Ylöstalo et al., 2014). Understandably, the bias towards high-latitude systems partly reflects the fact that this region contains a high density of humic-rich lakes. There is a relatively rich literature on DOM in temperate lakes (e.g. Zhang et al., 2011; Read and Rose, 2013; Müller et al., 2014) but few studies have focused on large shallow lakes like Lake Balaton with a continental climate; hence our understanding of the variability in CDOM optical properties in these systems is comparatively poorer. In systems such as Lake Balaton, although the DOC pool is typically smaller than at higher latitudes it still plays a significant role in regulating light availability and therefore lake metabolism, while the influence of processes such as photobleaching is also likely to be more pronounced.
In this study we explore spatial and seasonal variability in optical
properties of CDOM in Lake Balaton, a large temperate lake with a highly
continental climate. We investigate how changes in spectral absorption,
spectral slope coefficients, SUVA
With a surface area of 596 km
Lake Balaton is usually divided into four basins (south-west to north-east): Keszthely, Szigliget, Szemes and Siófok (Fig. 1b). The lake has 20 permanent and 31 temporary inflows, many of which are small streams or springs in the lake bed. The largest inflow to the lake is the Zala River, which flows through Kis-Balaton reservoir – a large semi-natural wetland system – and enters the lake in the westernmost part of Keszthely basin (Fig. 1b). The only outflow is the Sió channel in the north-east that connects the Siófok basin with the Danube River.
Lake Balaton has experienced eutrophication since the middle of the
18th century due to agricultural intensification and urbanisation
within the catchment. Since the early 1980s, significant effort has been
invested in improving its water quality
(Tátrai et al., 2000). The
construction of Kis-Balaton reservoir and wetland system was one of the main
engineering controls built to reduce nutrient inflow from the Zala River and
the overall loading within the lake. Kis-Balaton removes approximately 60 % of the annual nutrient loading to Lake Balaton
(Szilágyi et al., 1990). However, nutrient inputs from the
Zala River still result in high summer primary production in the eutrophic
(> 20 mg m
Suspended particulate matter in Lake Balaton is highly variable (spatially and temporally) due to its very shallow depth, constant mixing and susceptibility to wind-driven resuspension events (Istvánovics et al., 2004). Phytoplankton composition in the lake shows strong seasonal trends, with two annual blooms (Padisak and Reynolds, 1998; Présing et al., 2008; Hajnal and Padisák, 2008). In late summer and early autumn, cyanobacterial blooms often occur in the Keszthely basin (I), extending westwards to the Szigliget (II) and Szemes (III) basins and very occasionally to the Siófok (IV) basin (Padisak and Reynolds, 1998; Présing et al., 2008; Hajnal and Padisák, 2008). The lowest phytoplankton biomass generally occurs in February when the lake can be ice covered; a small dinoflagellate bloom may also occur in April (Mózes et al., 2006).
Spatial variability in CDOM quantity and quality was assessed over a 1-week period in July 2013 (6 stations) and a 3-week period in July 2014 (25 stations) at 31 stations over a biogeochemical gradient from the south-west in the water masses influenced by Zala River to the north-east near the outflow (Fig. 1c). Five stations were also sampled in the Kis-Balaton reservoir during the same period (2 in 2013 and 3 2014). These intensive sampling campaigns were timed to coincide with the annual summer peak in DOC to capture the maximum spatial variability likely to occur in the system.
In order to capture seasonal variability in CDOM quantity and quality, water samples were collected fortnightly at 6 long-term monitoring stations on Lake Balaton over the course of 7 months (March to September 2014). These comprised stations 01 and 03 from Keszthely basin (I), station 12 from Szigliget basin (II), station 20 from Szemes basin (III) and stations 25 and 30 from Siófok basin (IV) (for location of stations see Fig. 1c).
Water samples for DOC analysis were collected in triplicate using
acid-rinsed polypropylene bottles at 0.3 m depth below the surface. The
samples were immediately stored on ice and in the dark until they were
transferred to the laboratory for filtration. The samples were filtered
through 0.7
The spectral absorbance (
The absorption coefficient at 440 nm was used to express variation in CDOM
quantity. This wavelength was preferred over UV wavelengths because it is
more relevant to (and consistent with previous) studies on the optical
properties and remote sensing of CDOM in natural waters
(e.g. Carder et al., 1989; Nelson et al., 1998; Schwarz et al., 2002). The spectral slope for the
interval of 350–500 nm (
The
Samples for dissolved organic carbon (DOC) were measured by thermal
catalysis at 950
In order to examine the effects of solar radiation on autochthonous and
allochthonous CDOM in Lake Balaton, a 7-day in-lake incubation experiment
was undertaken during mid-July 2014. CDOM samples from Lake Balaton were
incubated in 65 mL capacity quartz tubes over 7 days under natural solar
radiation. The mean daytime lake temperature of the lake over the
experimental period was 24.6
The autochthonous CDOM was extracted from a strain of
For the CDOM
The CDOM
One CDOM
Spectral fluorescence signatures (SFS) were measured using an Instant
Screener (ISC) analyser (Laser Diagnostic Instruments Ltd., Tallinn,
Estonia). Measurements were made using a 1 cm quartz cuvette at excitation
wavelengths from 240 to 360 nm and at emission wavelengths from 260 to 575 nm with a 5 nm slit-widths for excitation and emission wavelengths.
Ultrapure water with 0.5 % NaN
There was some (pronounced in basin 1, noticeable in basin 2 and low in
basin 4) seasonal variability in the CDOM concentration measured in Lake
Balaton over a year (Fig. 2, Table 1). It should be stressed that
seasonal changes were only measured for 1 year and may not represent the
typical seasonal cycle observed over longer time periods. High
Seasonal variability of
The lowest and highest
Seasonal
Plot of
Seasonal variation in DOC was measured at six permanent sampling stations
(stations 01 and 03 from Basin Keszthely (I), station 12 from Basin
Szigliget (II), station 20 from Basin Szemes (III), stations 25 and 30 from
Basin Siófok (IV). DOC concentrations ranged from 7.63 at ST25 in April
to 19.70 mg L
Values of CDOM absorption coefficient at 440 nm, CDOM slope
coefficient between 350 and 500 nm, DOC concentration,
Values of CDOM absorption coefficient at 440 nm, CDOM slope
coefficient between 350 and 500 nm, DOC concentration,
CDOM absorption spectra for all stations (per basin) and Kis-Balaton.
Note the different
Scatter plots against distance to the main inflow [km] with loess
curve fitted to data.
Plot of
Scatter plot of
We observed an
The
Concentrations ranged from a minimum of 8.03 at ST17 (Basin III) to 10.9 mg L
Ultraviolet irradiance during the photobleaching experiment ranged from 7.79
to 42.9 MJ m
The autochthonous control samples (CDOM
Photodegradation also modified the spectral slope coefficient of the samples
(Fig. 9c). The values of
Humic-like fluorescence as indicated by
There were more than 10 orders of magnitude difference in fluorescence
intensity between CDOM
CDOM is the coloured fraction of DOC and is often the dominant light-absorbing component in lakes, particularly at blue and green wavelengths.
Previous research has shown that CDOM can be responsible for up to 80 %
of light absorption in Lake Balaton (Riddick et al., 2015) in spite of the fact that the lake also has high concentrations of
phytoplankton and non-algal particles (NAP). The high input of DOC from the
Zala River results in elevated concentrations in the western basin relative
to the remainder of the lake (from 0.169 m
The seasonal pattern in CDOM absorption and DOC concentration varied
considerably in the western basin, but was relatively constant in other
basins. The annual peak(s) in
Conversely, at the station nearest to the inflow of the Zala River the main
peak in
It is also notable that the effect of this humic-rich water from the Zala River on the biogeochemistry and light climate in Lake Balaton diminishes
rapidly through the system in summer. This can be partly attributed to the
dilution of the inflow with less humic water, but also the rapid degradation
of the highly biologically and photochemically reactive DOC originating from
Kis-Balaton during a period when microbial activity is high due to warm
water temperatures and UV irradiance is at its maximum. The collective
residence time of the Keszthely and Szigliget basins (0.25 and 0.72 years
respectively; Somlyódy and van Straten, 2012) explains
why the highly humic and labile components of the DOC entering the lake from
the Zala River are largely degraded before reaching the Szemes basin with
only the most recalcitrant DOC fractions persisting beyond the westernmost
basins. The resulting differences in CDOM composition are clearly reflected
in the variability in the CDOM absorption characteristics
(
The mean (and range) in
Very few studies have investigated seasonal variation in
The structural modifications in DOM and its coloured fractions that are in
part conveyed through variation in
Noticeably, while
Changes in humic-like fluorescence (
Similar trends were also observed in SUVA
The spatio-seasonal variability in CDOM absorption in Lake Balaton strongly suggests that photobleaching plays a key role in the processing and degradation of dissolved organic carbon as it flows through the system. Rapid degradation of allochthonous CDOM was observed (Fig. 9), which was especially pronounced at the time of year with the highest solar radiation but probably also enhanced by mineralisation by bacterial activity as a response to high water temperatures during the summer period. Dilution processes alone cannot explain the loss of DOC; therefore, it must also be due to in-lake transformation. The processing and transformation of DOC by photobleaching not only influences carbon cycling, but it also is accompanied by an increase in the transparency of the water column (Osburn et al., 2009) and changes in the optical properties (Yamashita et al., 2013) that have wider implications for the underwater light climate and primary production.
The in-lake incubations conducted in Lake Balaton provide further
substantiation for the critical role of photochemistry in the turnover of
CDOM. We observed rapid changes in the absorption properties of CDOM in
response to exposure to natural UV irradiation. In the allochthonous CDOM
treatments, the rate of degradation was higher than that obtained for
Lake Taihu by Zhang et al. (2013), who reported a 22 % decrease over 9
days. Bacterial degradation was not noticeable in the allochthonous samples
as there was almost no difference in
Photodegradation also modified the spectral slope (Fig. 9b) of the CDOM
absorption spectra. Both the spectral slope and absorption coefficient for
autochthonous CDOM were significantly lower than for allochthonous samples
(ANOVA,
The fluorescence spectra also indicate a marked difference in composition between the allochthonous and autochthonous material. The decomposition of CDOM into lower molecular weight compounds under UV-B radiation (Lepane et al., 2003) results in a significant loss of both absorption and fluorescence. The negligible fluorescence signal observed for the autochthonous CDOM samples in this study is likely due to its low concentration. In contrast, the humic-like fluorescence signal measured from allochthonous samples was initially high but decreased over the experimental period from 41.07 to 17.48 QSU (57.5 % decrease). Similarly, we observed a reduction in protein-like fluorescence from 2.06 to 1.93 QSU (6.31 % decrease; Fig. 10b) over the 7 days of the experiment. This agrees strongly with the results of previous studies showing that the fluorescence signal from humic compounds is rapidly lost through photobleaching, whereas aromatic-like fluorescence is generally not as susceptible to photodegradation. Helms et al. (2013), for example, reported an 84 % decrease in humic-like fluorescence in response to photobleaching compared to only a 47 % decrease in aromatic-like fluorescence after 68 days of continuous irradiation in a UV solar simulator.
The absorption of light by CDOM is a major determinant of water transparency in lakes and the availability of light for primary production (Kirk, 1994; Laurion et al., 1997, 2000; Vähätalo et al., 2005). Absence of measurements of the underwater light field makes it difficult to attribute its effect to this particular case, but there clearly exists evidence that the dynamic nature of dissolved organic carbon in lakes results in marked spatio-seasonal variation in both the magnitude and wavelength dependency of light absorption by chromophoric substances. This variability undoubtedly has implications, not only for the quantity of light available to photosynthetic organisms but also its quality. High concentrations of CDOM result in intense absorption of light at blue and green wavelengths but the intensity of absorption decreases exponentially with wavelength. This not only has implications for the productivity of the system (Cory et al., 2015), but also for the photo-physiology and species composition of the phytoplankton community. The intense absorption of UV light by CDOM protects phytoplankton from physiological damage and reduces the need for phytoplankton cells to manufacture UV-protective pigments. This can result in chromatic acclimation with phytoplankton in high CDOM waters investing less in UV-protective pigments (Riddick et al., 2015).
The magnitude of variability in the spectral dependency of CDOM absorption
also has implications for bio-optical models of the underwater light field
that are used to underpin remote sensing algorithms for estimation of CDOM
in lakes and other inland waters. Existing bio-optical models
(Lee et al., 2002) commonly extrapolate absorption by
CDOM in the blue to longer wavelengths using a fixed slope coefficient. We
demonstrate here that even within a single lake system significant
variability can occur in
This study revealed the high spatial and seasonal variability in the
quantity and quality of CDOM that can exist within a large, temperate
shallow lake. The variation was strongly driven by the allochthonous input
of dissolved carbon from the Zala River and its rapid transformation as it
moves through the system. The variability in the quantity and quality of
CDOM was strongly reflected in a number of readily measured optical
parameters including
Photobleaching was found to be a major factor controlling the in-lake transformation and degradation of CDOM, and a key process influencing the spatial structure CDOM throughout the system. The photobleaching rate coefficient for allochthonous CDOM was found to be higher than for autochthonous CDOM due to the greater photoreactivity of terrestrially derived compounds. CDOM in Lake Balaton is mainly terrestrial in origin and is thus rapidly degraded by exposure to UV irradiation. The implied importance of photobleaching to carbon dynamics is consistent with previous studies conducted in other inland water bodies (Zhang et al., 2013) as well as other studies carried out in shelf seas (Babin et al., 2003) and the open ocean (Helms et al., 2013).
More widely, these results provide an insight on the potential contribution
of wetlands to DOM and CDOM in lakes, not only in terms of the concentration
of CDOM but also its seasonality. The seasonal trend in CDOM observed close
to the main inflow was significantly different from that observed elsewhere
in the system. Notwithstanding the fact that most of the CDOM in Lake
Balaton would seem to be terrestrial in origin, we did observe an increase
in
The observed spatial and temporal variability in the optical properties of CDOM in this study has important implications for biogeochemical cycling in Lake Balaton but also for bio-optical models of the underwater light climate in lakes and their application in the parameterisation of algorithms for optical remote sensing of CDOM and other optically active constituents.
The data analysed in this study are accessible through the University of
Stirling DataSTORRE open access repository at
M. E. Aulló-Maestro designed and conducted the experiments with input from P. Hunter, E. Spyrakos and A. Tyler. Hajnalka Horváth and Mátyás Présing carried out data collection for seasonal measurements at Balaton Limnological Institute. Spatial measurements were taken by GloboLakes and INFORM teams. Jesús M. Torres Palenzuela contributed to fluorescence measurements at University of Vigo and Tom Preston contributed to mass spectrometry analysis at the Stable Isotope Biochemistry Laboratory at the Scottish Universities Environmental Research Centre (SUERC). M. E. Aulló-Maestro processed the data and prepared the figures with input from Pierre Mercatoris. M. E. Aulló-Maestro prepared the manuscript with the assistance of all co-authors.
The authors declare that they have no conflict of interest.
The authors gratefully acknowledge funding from the UK NERC funded GloboLakes project (REF NE/J024279/1), the EU FP7 INFORM project (Grant Agreement Number 606865) and the Hungarian Academy of Science TÁMOP-4.2.2 A-11/1/KONV-2012-0038 project. Fluorescence measurements at University of Vigo were funded by a Santander doctoral travel award. Field and lab assistance on Lake Balaton was provided by Laura Ulsig, Viktória Horváth, Anett Kelemen, Eszter Zsigmond, Piroska Rádóczy, Anna Kánicz and Ádám Szigethy. Lab assistance at University of Vigo was provided by Marta Iglesias Trigo. María E. Aullo-Maestro was supported by GloboLakes project and a University of Stirling impact studentship. The authors would like to particularly emphasise their acknowledgement to the late Matyas Presing, a valued mentor and friend who greatly contributed to this research. Edited by: B. A. Pellerin Reviewed by: two anonymous referees