Hydrobiologia

, Volume 691, Issue 1, pp 157–169

Parameterization of chlorophyll a-specific absorption coefficients and effects of their variations in a highly eutrophic lake: a case study at Lake Kasumigaura, Japan

Authors

    • Faculty of Life and Environmental SciencesUniversity of Tsukuba
  • Nobuhiro Zaitsu
    • Graduate School of Life and Environmental SciencesUniversity of Tsukuba
  • Yuta Sekimura
    • Graduate School of Life and Environmental SciencesUniversity of Tsukuba
  • Bunkei Matsushita
    • Faculty of Life and Environmental SciencesUniversity of Tsukuba
  • Takehiko Fukushima
    • Faculty of Life and Environmental SciencesUniversity of Tsukuba
  • Akio Imai
    • National Institute for Environmental Studies
Primary Research Paper

DOI: 10.1007/s10750-012-1066-4

Cite this article as:
Yoshimura, K., Zaitsu, N., Sekimura, Y. et al. Hydrobiologia (2012) 691: 157. doi:10.1007/s10750-012-1066-4
  • 192 Views

Abstract

The chlorophyll a-specific absorption coefficient (\( a_{\text{ph}}^{*} \left( \lambda \right) \)) in a highly eutrophic lake can show characteristics distinct from that in the ocean due to the differences in the structure and composition of phytoplankton. In this study, investigated the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) in Lake Kasumigaura, a highly eutrophic lake in Japan, in association with the package effect and the effect of accessory pigments, and carried out the parameterization of \( a_{\text{ph}}^{*} \left( \lambda \right) \). Although \( a_{\text{ph}}^{*} \left( \lambda \right) \) did not vary spatially, it did show significant temporal variation, with a particularly high value after spring-bloom. This high \( a_{\text{ph}}^{*} \left( \lambda \right) \) in spring was attributed to a lower package effect and a higher proportion of carotenoid than the other samples. Although the value of \( a_{\text{ph}}^{*} \left( \lambda \right) \) was correlated with the concentration of chlorophyll-a (Chl-a), the correlation coefficient was lower than those reported in the ocean. Some lake-water samples showed variations of the package effect and the effect of accessory pigments that were independent of the concentration of Chl-a, and these independent variations resulted in the weak correlation between \( a_{\text{ph}}^{*} \left( \lambda \right) \) and the concentration of Chl-a. Together, these results suggest that the factors controlling \( a_{\text{ph}}^{*} \left( \lambda \right) \) in highly eutrophic lakes are distinct from that in ocean samples.

Keywords

Chlorophyll a-specific absorption coefficientTemporal/spatial variationsHighly eutrophic lake waterPackage effectEffect of accessory pigments

Introduction

The absorption coefficient of phytoplankton (aph(λ)) is one of the major parameters used to define the light absorption of natural water, and the investigation of properties of aph(λ) is important to elucidate the optical properties of water. The relation between aph(λ) and concentrations of chlorophyll a (Chl-a) has been intensively studied for the last two decades, and significant positive correlations have been widely reported (Bricaud et al., 1995, 2004; Cleveland, 1995; Babin et al., 2003; Zhang et al., 2007, 2010; Dmitriev et al., 2009). Because of the significant correlation between aph(λ) and Chl-a, their ratio, i.e., the chlorophyll-a-specific absorption, (\( a_{\text{ph}}^{*} \left( \lambda \right) \)), has been applied in remote sensing algorithms for estimating the concentration of Chl-a and further the primary production (Ishizaka, 1998; Oyama et al., 2010; Yang et al., 2011). Thus, the investigation of temporal/spatial variations and the parameterization of \( a_{\text{ph}}^{*} \left( \lambda \right) \) are important to enhance the accuracy of the algorithms (Hoogenboom et al., 1998).

The temporal/spatial variations of \( a_{\text{ph}}^{*} \left( \lambda \right) \) have been widely investigated in the ocean and coastal regions for its parameterization purposes, and significant positive correlations with the concentration of Chl-a have also been well documented (Hoepffner & Sathyendranath, 1992; Bricaud et al., 1995, 2004; Cleveland, 1995; Suzuki et al., 1998; Babin et al., 2003; Stæhr & Markager, 2004). The causes of variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) have also been studied, and the package effect and the effect of accessory pigments were suggested to be the major factors (Kirk, 1975; Morel & Bricaud, 1981; Hoepffner & Sathyendranath, 1992; Bricaud et al., 1995, 2004). The package effect is derived from the overlapping of chloroplasts resulting in a decrease in the pigments contributing to beam absorption in aph(λ) measurements. This effect theoretically depends on cell size and intracellular pigment concentrations, and \( a_{\text{ph}}^{*} \left( \lambda \right) \) is affected by the package effect at all the wavelengths absorbed by pigments (Morel & Bricaud, 1981). The effect of accessory pigments results from the absorption by accessory pigments represented by carotenoids and chlorophyll b (Chl-b)—the pigment composition relates to the effect (Hoepffner & Sathyendranath, 1992). Since carotenoids absorb light only around 450 nm, the effect of accessory pigments on \( a_{\text{ph}}^{*} \left( \lambda \right) \) is dominantly found around this wavelength.

Babin et al. (2003) studied the effects of cell size and pigment composition on the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) in different coastal regions, and found that these effects made different contributions to \( a_{\text{ph}}^{*} \left( \lambda \right) \) according to the water region. Bricaud et al. (2004) studied the effects of cell size, pigment composition, and the package effect on the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) in the ocean, and found that the package effect had the dominant influence; they used the package effect index (\( Q_{\text{a}}^{*} \)(λ)) for measurement of the package effect, as proposed by Morel & Bricaud (1981). In contrast to those in the ocean, the variations of \( a_{\text{ph}}^{*} \left( \lambda \right) \) in lake water and their effects are poorly understood, although intensive studies have been carried out for Lake Taihu, a highly eutrophic lake in China (Zhang et al., 2007, 2010; Le et al., 2009; Sun et al., 2009).

Trophic status affect not only the concentration of Chl-a but also the package effect and the effect of accessory pigments (Bricaud et al., 2004 and references therein), with the result that the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) has been well correlated with the concentration of Chl-a in the ocean and coastal regions covering oligo-, meso- and eutrophic conditions (Bricaud et al., 1995, 2004; Babin et al., 2003; Stæhr & Markager, 2004). On the other hand, an insignificant correlation between \( a_{\text{ph}}^{*} \left( \lambda \right) \) and Chl-a was also reported (Hoepffner & Sathyendranath, 1992). Since the concentration and composition of pigments of phytoplankton in highly eutrophic/turbid lakes are probably more affected by factors such as light availability (Moore et al., 1995; Fujiki & Taguchi, 2002) than by the nutrient condition, the relations among concentration of Chl-a, package effect, and effect of accessory pigments may be different from that in the ocean and coastal regions covering wide range of the trophic status. In addition, most of the models of the relation between \( a_{\text{ph}}^{*} \left( \lambda \right) \) and Chl-a have been reported for the range of Chl-a lower than 100 mg m−3 (Bricaud et al., 1995; Stæhr & Markager, 2004), but the concentration of Chl-a in eutrophic lakes sometimes lies outside of this range.

To obtain new information on the characteristics of \( a_{\text{ph}}^{*} \left( \lambda \right) \) in highly eutrophic lake waters, we comprehensively examined the temporal/spatial variations in \( a_{\text{ph}}^{*} \left( \lambda \right) \), along with the parameterization of \( a_{\text{ph}}^{*} \left( \lambda \right) \), in a highly eutrophic lake. In addition, the package effect and the effect of accessory pigments on the variations of \( a_{\text{ph}}^{*} \left( \lambda \right) \) were quantitatively evaluated.

Methods

Sampling

Lake Kasumigaura is the second-largest lake in Japan and is characterized by highly turbid and eutrophic conditions. Although the nutrient conditions are constantly highly eutrophic, the concentration of Chl-a changes dynamically depending on the season, and a remarkable increase in Chl-a is observed every spring in response to blooming. The structure of phytoplankton community shows seasonal variation (NIES database; http://db.cger.nies.go.jp/gem/inter/GEMS/database/kasumi/top.html). Diatom (especially Aulacoseira spp. and Thalassiosiraceae spp.) dominate in spring and autumn. Cyanobacteria can be found in all seasons, but large biomass of Microcystis aeruginosa are observed in summer. In winter, the community is mainly composed of the diatom, cyanobacteria (Planktothrix spp.), and green algae (especially Dictyosphaerium spp. and Scenedesmus spp.).

Sampling cruises were conducted 12 times from August 2009 to June 2010. Surface water was collected from 26 sites on September 1 and December 15, 2009 and March 17 and May 18, 2010 to cover all seasons, and the water samples at five sites representing major water bodies in Lake Kasumigaura (sites 6, 10, 13, 14, and 17), were obtained on all the sampling dates (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig1_HTML.gif
Fig. 1

Lake Kasumigaura and sampling sites

Surface water was kept in polycarbonate bottles in the dark at the in situ temperature and was readily filtered by Whatman GF/F glass–fiber filters after it was brought back to the laboratory (about an hour after the sampling). The filter samples for the measurements of chlorophylls and carotenoids were immediately frozen and stored at −40°C until the analysis.

Measurements

Absorption coefficients (m−1) of particulate matter (ap(λ)) and tripton (atrp(λ)) were measured for the filter sample using a dual-beam spectrophotometer (UV-3100PC; Shimadzu) equipped with an integrating sphere according to Mitchell (1990). In order to measure the atrp(λ), pigments in particulate matter on the filter were extracted by methanol in the dark (2°C, 12 h) as described in Kishino et al. (1985). The optical density of the filter sample (ODf(λ)) was measured between 350 and 800 nm. In order to correct the path length amplification effect of the ODf(λ), the following equation was used according to Mitchell (1990):
$$ {\text{ODs}}\left( \lambda \right) = a \times {\text{ODf}}\left( \lambda \right) + b \times {\text{ODf}}\left( \lambda \right)^{ 2} , $$
(1)
where ODs(λ) is the optical density of particulate matter as a suspension, and a and b are the constants obtained from the relation between ODs(λ) and ODf(λ). While Mitchell (1990) concentrated particulate matter by filtration and measured the ODs(λ) for the particulate matter resuspended in pure water, in the present study, the ODs(λ) for lake water was measured directly with setting a corresponding filtrate at reference side using the same spectrophotometer system and quartz cuvette (1 cm width), because the lake water was highly turbid and could provide high values of optical density with low noise and without the requirement of concentrating the particulate matter. The relation between ODf(λ) and ODs(λ) obtained at station 10 on each sampling date is shown in Fig. 2. A clear relation was confirmed among sampling dates with little variation. Although the relations among different sites on each sampling date are not shown, this variation was also small. A polynomial regression line was fitted for the relation, and the constants of a and b in Eq. 1 were obtained as 0.442 and 0.200, respectively. The relation between ODf(λ) and ODs(λ) obtained in this study was close to those of Bricaud & Stramski (1990) and Hoepffner & Sathyendranath (1992). The ap(λ) and atrp(λ) were calculated from ODs(λ) using Eq. 2:
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig2_HTML.gif
Fig. 2

Relation between ODf(λ) and ODs(λ) obtained for the data at station 10 at every sampling date (N = 12) at intervals of 10 nm between 350 and 700 nm

$$ a_{\text{p/trp}} \left( \lambda \right) = 2.303 \times {\text{ODs}}\left( \lambda \right) \times {\text{GL}}^{ - 1} , $$
(2)
where GL is the geometrical light pass length (m) calculated by dividing the volume of the filtered sample (m3) with the clearance area of the filter (m2). The aph(λ) was obtained as the difference between the ap(λ) and the atrp(λ). Then, aph(λ) was divided by the concentration of Chl-a to calculate \( a_{\text{ph}}^{*} \left( \lambda \right) \) (m2 mgChl-a−1).

Chlorophylls (i.e., Chl-a, -b, and -c) in the filter sample were extracted with methanol for 24 h in the dark at 2°C. The concentration of chlorophylls was determined according to SCOR-UNESCO (1966).

In order to extract the carotenoids, the filter sample was put into a brown vial with acetone (5 ml) and homogenized by ultrasonication for 10 min in ice-cold water, and then the sample was stored for 12 h in the dark at 2°C. After the extraction, the solution was passed through a disk filter (DISMIC-13HP; ADVANTEC), and the filtrate was applied to the measurement of carotenoids. Carotenoid concentrations were determined by high-performance liquid chromatography (LC-20A system; SHIMADZU) equipped with a photodiode array detector (SPD-M20A; SHIMADZU) with reference to Kraay et al. (1992). A reverse-phase column (Nova-pack C18, 3.9 mm × 150 mm; Waters) was used to separate carotenoids with a flow rate of 0.8 ml min−1; the gradient program is shown in Table 1. Standards of major carotenoids, including peridinin, fucoxanthin, alloxanthin, diatoxanthin, zeaxanthin, lutein, and β-carotene (Wako Pure Chemical Industries, Ltd.) were used to determine the concentrations.
Table 1

Time schedule of the gradient mode in the measurement of carotenoids

Time (min)

Solvents (%)

A

B

C

0

60

40

0

5

0

100

0

10

0

80

20

20

0

50

50

24

0

30

70

31.5

0

30

70

33

0

0

100

38

0

0

100

39

0

100

0

43

0

100

0

44

60

40

0

48

60

40

0

The solvents (Wako Pure Chemical Industries, Ltd.) were as follows: A was 0.5 M ammonium acetate in methanol/water (85:15, v/v), B was acetonitrile/water (90:10, v/v), and C was 100% ethyl acetate

Package effect index

Morel & Bricaud (1981) suggested \( Q_{\text{a}}^{*} \)(λ) to represent the package effect. The \( Q_{\text{a}}^{*} \)(λ) is defined as the ratio between the specific absorption coefficients of phytoplankton as an intact cell (i.e., \( a_{\text{ph}}^{*} \left( \lambda \right) \)) and the specific absorption coefficients of the same phytoplankton material as a solution (\( a_{\text{sol}}^{*} \left( \lambda \right) \)). The equation is represented as follows:
$$ Q_{\text{a}}^{*} \left( \lambda \right) = a_{\text{ph}}^{*} \left( \lambda \right)/a_{\text{sol}}^{*} \left( \lambda \right) $$
(3)

If there is no package effect (i.e., no underestimation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) due to the effect), then \( Q_{\text{a}}^{*} \)(λ) is 1, the theoretical maximum value. \( Q_{\text{a}}^{*} \)(λ) decreases with increases in the effect and reaches a theoretical minimum value of 0. Low \( Q_{\text{a}}^{*} \)(λ) indicates a high package effect, resulting in a large underestimation of \( a_{\text{ph}}^{*} \left( \lambda \right) \).

The \( a_{\text{ph}}^{*} \left( \lambda \right) \) at 675 nm is derived from Chl-a, Chl-b, and divinyl Chl-b (Bricaud et al., 1995). In this study, the proportions of Chl-b in chlorophyll pigments were small and relatively constant (9 ± 4%, under submission). In addition, the absorption of Chl-b at 675 nm is considerably smaller than the absorption of Chl-a (Bricaud et al., 2004; Fig. 1). Divinyl Chl-b has been found in specific kinds of prokaryotic prochlorophytes in the oligotrophic open ocean (Goencke & Repeta, 1992), but their biomass in Lake Kasumigaura is limited because of the dominance of eukaryotic phytoplankton such as diatoms (NIES database; http://db.cger.nies.go.jp/gem/inter/GEMS/database/kasumi/top.html). Thus, \( a_{\text{ph}}^{*} \)(675) in Lake Kasumigaura was mainly derived from Chl-a. Since Bricaud et al. (1995) reported that the \( a_{\text{sol}}^{*} \)(675) of Chl-a was 0.0207 m2 mgChl-a−1, the \( Q_{\text{a}}^{*} \)(675) was obtained as the ratio of \( a_{\text{ph}}^{*} \)(675) to \( a_{\text{sol}}^{*} \)(675).

Results

Variations of Chl-a concentration

The concentrations of Chl-a fell in a range between 36.6 and 214.4 mg m−3 with a mean value of 79.5 ± 32.8 mg m−3, and a large and significant temporal variation was observed (analysis of variance (ANOVA), P < 0.001) (Fig. 3). Increases in the concentration were observed in September (90.0 ± 15.1 mg m−3) and especially in March (123.3 ± 48.3 mg m−3), representing the obvious spring bloom. The concentrations were drastically decreased in May (58.6 ± 19.1 mg m−3). On the other hand, no significant variation among sampling sites was confirmed. The temporal variation of Chl-a concentration is summarized in Table 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig3_HTML.gif
Fig. 3

Seasonal change in the concentration of Chl-a. The error bar indicates standard deviations

Table 2

Concentrations of Chl-a, and values of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675)

Date

Chl-a (mg m−3)

\( a_{\text{ph}}^{*} \)(440) (m2 mgChl-a−1)

\( a_{\text{ph}}^{*} \)(675) (m2 mgChl-a−1)

Ave

SD

Max

Min

Ave

SD

Max

Min

Ave

SD

Max

Min

2009/8/12

62.5

8.1

76.5

54.3

0.022

0.003

0.025

0.016

0.014

0.003

0.017

0.009

2009/9/1

75.1

16.2

117.7

39.7

0.023

0.003

0.034

0.013

0.016

0.001

0.023

0.010

2009/9/9

90.0

15.1

112.8

75.1

0.017

0.002

0.019

0.015

0.013

0.001

0.015

0.012

2009/10/7

53.3

8.4

63.3

39.8

0.020

0.002

0.024

0.018

0.014

0.002

0.017

0.012

2009/12/15

64.1

9.5

81.4

45.1

0.019

0.006

0.031

0.012

0.010

0.003

0.013

0.006

2010/1/13

71.6

11.8

86.0

54.1

0.021

0.003

0.025

0.016

0.016

0.001

0.018

0.015

2010/2/10

77.4

13.9

95.0

58.9

0.030

0.003

0.033

0.025

0.019

0.001

0.020

0.017

2010/3/10

104.4

34.6

172.4

75.6

0.022

0.001

0.023

0.021

0.017

0.001

0.018

0.016

2010/3/17

117.1

38.6

187.7

70.7

0.025

0.003

0.032

0.019

0.021

0.002

0.024

0.018

2010/4/7

123.3

48.3

214.4

78.5

0.024

0.002

0.028

0.021

0.016

0.001

0.018

0.015

2010/5/18

58.6

19.1

91.7

36.6

0.039

0.006

0.051

0.028

0.024

0.002

0.029

0.020

2010/6/9

65.0

26.8

116.9

43.9

0.028

0.007

0.034

0.016

0.020

0.004

0.023

0.013

Ave average, SD standard deviation, Max maximum, Min minimum

Variations and parameterization of \( a_{\text{ph}}^{*} \left( \lambda \right) \)

The values of aph(440) and aph(675) ranged from 0.58 to 4.8 m−1 with a mean value of 1.97 ± 0.82 m−1 and from 0.41 to 3.39 m−1 with a mean value of 1.34 ± 0.60 m−1, respectively. In our preliminary study, we found that the concentration of Chl-a was significantly correlated with both aph(440) (R = 0.82, P < 0.001) and aph(675) (R = 0.90, P < 0.001) (under submission). The spectra of \( a_{\text{ph}}^{*} \left( \lambda \right) \) showed specific peaks around 440 and 675 nm (Fig. 4). Large variations of the values were observed among spectra: \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) varied from 0.012 to 0.051 m2 mgChl-a−1 and 0.006 to 0.029 m2 mgChl-a−1, respectively. The mean values at 440 and 675 nm obtained from all data are summarized in Table 3. The coefficients of variation at 440 and 675 nm were 31.7 and 29.0%, respectively. The ratio of \( a_{\text{ph}}^{*} \)(440)/\( a_{\text{ph}}^{*} \)(675) was 1.50 ± 0.27.
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig4_HTML.gif
Fig. 4

Spectra of \( a_{\text{ph}}^{*} \left( \lambda \right) \) for all data

Table 3

Averages and standard deviations of \( a_{\text{ph}}^{*} \left( \lambda \right) \) at 440 and 675 nm at 440 and 675 nm obtained for all data set, data set except for May 2010 and data in May 2010, respectively (m2 mgChl-a−1)

\( a_{\text{ph}}^{*} \)(λ)

All data

August 2009–April 2010 and June 2010

May 2010

\( a_{\text{ph}}^{*} \)(440)

0.026 ± 0.008

0.023 ± 0.005

0.040 ± 0.007

\( a_{\text{ph}}^{*} \)(675)

0.018 ± 0.005

0.016 ± 0.004

0.024 ± 0.002

The data from the five sites obtained on every sampling date (N = 60) were used to study a temporal variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \), and the data obtained on four sampling dates at 26 sites (N = 104) were applied to the study of spatial variation. The temporal variations of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) are shown in Fig. 5. Significantly high values of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) were observed in May (ANOVA, P < 0.001). On the other hand, significant spatial variations of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) were not observed (Fig. 6). Although the mean values appeared to increase from the upper to the lower stream (i.e., from stations 1 and 14 to stations 13 and 23, respectively), the large standard deviation at each site exceeded the variations among the sites, resulting in the insignificant spatial variations. The temporal variations of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) are shown in Table 2 along with the concentrations of Chl-a. The mean values obtained for May and the other dates are also summarized in Table 3. The values at 440 and 675 nm in May were higher than all the values obtained on the other dates. The standard deviations obtained for the divided periods (i.e., May and the other months) were lower than those of the entire period.
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig5_HTML.gif
Fig. 5

Temporal changes in \( a_{\text{ph}}^{*} \left( \lambda \right) \) at 440 and 675 nm

https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig6_HTML.gif
Fig. 6

Spatial changes in \( a_{\text{ph}}^{*} \) at 440 and 675 nm. Dashed lines divide the different water streams

The variation of \( a_{\text{ph}}^{*} \)(440) as a function of Chl-a concentration is shown in Fig. 7. The correlation between \( a_{\text{ph}}^{*} \)(440) and Chl-a was statistically significant (P < 0.01, R = 0.29), and the following power function was obtained:
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig7_HTML.gif
Fig. 7

Variation of \( a_{\text{ph}}^{*} \)(440) as a function of the concentration of Chl-a. The solid line represents the power function regression line obtained by least squares fitting

$$ a_{\text{ph}}^{*} \left( {440} \right) = 0.0718 \times \left[ {{\text{Chl-}}}a \right]^{ - 0.2458} ,$$
(4)
where [Chl-a] is the concentration of Chl-a. Although the value of \( a_{\text{ph}}^{*} \)(440) tended to decrease with the increase in Chl-a, some data, especially the data from August to December 2009, went against the trend, with both \( a_{\text{ph}}^{*} \)(440) and Chl-a showing low values.
The intercepts and slopes in the model representing the relation between \( a_{\text{ph}}^{*} \)(440) and Chl-a were obtained from 400 to 700 nm and are listed in Table 4. The values of \( a_{\text{ph}}^{*} \left( \lambda \right) \) at 440 and 675 nm were estimated from the concentration of Chl-a in the range from 10 to 200 mg m−3 using the coefficients in Table 4 and were compared with those of Bricaud et al. (1995) and Stæhr & Markager (2004) (Table 5). The estimated \( a_{\text{ph}}^{*} \left( \lambda \right) \) values in the present study were higher than the values in these previous reports.
Table 4

Slopes, intercepts, and correlation coefficients (R) of power functional regression lines representing the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) as a function of the concentration of Chl-a between 400 and 700 nm with 4-nm intervals

λ (nm)

Intercept

Slope

R

λ (nm)

Intercept

Slope

R

400

0.054

0.263

0.33

552

0.013

0.187

0.18

404

0.057

0.264

0.33

556

0.012

0.179

0.17

408

0.056

0.249

0.31

560

0.011

0.183

0.18

412

0.063

0.261

0.33

564

0.010

0.171

0.16

416

0.065

0.258

0.32

568

0.010

0.160

0.15

420

0.068

0.262

0.33

572

0.011

0.186

0.19

424

0.068

0.254

0.32

576

0.011

0.185

0.18

428

0.069

0.251

0.31

580

0.010

0.141

0.14

432

0.071

0.250

0.31

584

0.009

0.123

0.13

436

0.072

0.246

0.30

588

0.010

0.135

0.14

440

0.072

0.245

0.29

592

0.011

0.152

0.15

444

0.069

0.244

0.29

596

0.012

0.173

0.18

448

0.064

0.241

0.28

600

0.013

0.178

0.19

452

0.060

0.239

0.28

604

0.013

0.167

0.18

456

0.053

0.225

0.26

608

0.015

0.197

0.21

460

0.050

0.221

0.25

612

0.015

0.185

0.21

464

0.049

0.222

0.25

616

0.016

0.180

0.20

468

0.051

0.239

0.27

620

0.017

0.183

0.22

472

0.052

0.251

0.28

624

0.017

0.173

0.21

476

0.053

0.261

0.29

628

0.017

0.170

0.21

480

0.054

0.272

0.30

632

0.017

0.163

0.20

484

0.052

0.269

0.29

636

0.017

0.168

0.20

488

0.052

0.268

0.29

640

0.016

0.153

0.18

492

0.051

0.269

0.29

644

0.017

0.173

0.21

496

0.049

0.264

0.29

648

0.017

0.177

0.22

500

0.047

0.267

0.29

652

0.017

0.178

0.23

504

0.045

0.269

0.29

656

0.015

0.129

0.16

508

0.043

0.276

0.29

660

0.017

0.127

0.17

512

0.040

0.276

0.29

664

0.018

0.095

0.13

516

0.037

0.278

0.28

668

0.022

0.104

0.15

520

0.033

0.269

0.26

672

0.025

0.100

0.14

524

0.028

0.249

0.24

676

0.030

0.128

0.19

528

0.025

0.243

0.23

680

0.032

0.152

0.21

532

0.020

0.210

0.20

684

0.033

0.178

0.25

536

0.019

0.216

0.20

688

0.031

0.206

0.29

540

0.017

0.204

0.19

692

0.027

0.249

0.34

544

0.015

0.196

0.18

696

0.021

0.276

0.35

548

0.016

0.216

0.20

700

0.018

0.344

0.39

The values were obtained for the data excluding October and December in 2009

Table 5

Values of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) estimated for Chl-a in the range of 10–200 mg m−3 using the relations between \( a_{\text{ph}}^{*} \)(440) and Chl-a concentration (this study, Bricaud et al., 1995, and Stæhr & Markager, 2004)

Chl-a (mg m−3)

10

50

100

150

200

\( a_{\text{ph}}^{*} \)(440)

 This study

0.041

0.028

0.023

0.021

0.020

 Bricaud et al. (1995)

0.015

0.011

0.009

0.008

0.007

 Stæhr & Markager (2004)

0.023

0.018

0.015

0.013

0.012

\( a_{\text{ph}}^{*} \)(675)

 This study

0.022

0.018

0.017

0.016

0.015

 Bricaud et al. (1995)

0.012

0.011

0.010

0.009

0.009

 Stæhr & Markager (2004)

0.014

0.012

0.011

0.010

0.009

Effects on the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \)

Package effect

The variations of \( Q_{\text{a}}^{*} \)(675) as a function of Chl-a are shown in Fig. 8a. \( Q_{\text{a}}^{*} \)(675) varied between 0.31 and 1.39. Although some values exceeded the theoretical maximum of 1 because of a slight absorption of Chl-b, \( Q_{\text{a}}^{*} \)(675) can still be considered a reasonable index for assessing the trend of the packaging effect because of the constant proportion of Chl-b. There was no correlation between \( Q_{\text{a}}^{*} \)(675) and Chl-a in this study. There were data showing similar concentrations of Chl-a but obviously different \( Q_{\text{a}}^{*} \)(675) values. In particular, the data of October and December in 2009 showed scatterings distinct from those of other data; both Chl-a and \( Q_{\text{a}}^{*} \)(675) showed low values. The \( Q_{\text{a}}^{*} \)(675) obtained from May to June in 2010, the period after the spring bloom, was significantly high (ANOVA, P < 0.001), demonstrating that the underestimation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) by the effect was smaller than those in other periods.
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig8_HTML.gif
Fig. 8

Relations of the concentration of Chl-a with a\( Q_{\text{a}}^{*} \)(675) and b the Car/Chl-a ratio in each investigated period. The regression line of the relation between Chl-a and the Car/Chl-a ratio is shown as a solid line in “b

Effect of accessory pigments

The effect of accessory pigments was evaluated based on the ratio between carotenoids and Chl-a (Car/Chl-a ratio). The Car/Chl-a ratios as a function of Chl-a are shown in Fig. 8b. The ratios fall in a range from 0.2 to 1.0. Although the Car/Chl-a ratio was correlated with Chl-a (R = 0.46, P < 0.01), some data, especially the data of October and December in 2009, included a low Car/Chl-a ratio and low Chl-a concentration and were far from the regression line. The values of the Car/Chl-a ratios obtained in May and June were obviously quite high (exceeding 0.8), while those of the other periods fell in a similar range from 0.2 to 0.6, demonstrating that the \( a_{\text{ph}}^{*} \left( \lambda \right) \) in May 2010 was highly enhanced by the carotenoids in Lake Kasumigaura.

Contributions of the package effect and the effect of accessory pigments to the variation of \( a_{\text{ph}}^{*} \)(440)

The variation of \( a_{\text{ph}}^{*} \)(440) was attributed to the package effect and the effect of accessory pigments. Because we assumed that the data points with a similar Car/Chl-a ratio were also similarly affected by the accessory pigments, the distribution of the Car/Chl-a ratio was assessed to extract a group of data points with similar Car/Chl-a ratios (Fig. 9). The numerous data points distributed across the ratio range of 0.3–0.4 (N = 51) were thus selected as the data group that was similarly affected by the effect of accessory pigments. The pigment composition of the extracted data is shown in Table 6.
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig9_HTML.gif
Fig. 9

Distribution of the Car/Chl-a ratio

Table 6

Pigment composition of the samples with a Car/Chl-a ratio of 0.3–0.4 (N = 51)

Pigments

Composition (%)

Chl-a

48.0 ± 12.9

Chl-b

9.9 ± 4.9

Chl-c

25.7 ± 12.5

Peridinin

2.6 ± 1.9

Fucoxanthin

4.0 ± 2.4

Alloxanthin

0.9 ± 0.7

Diatoxanthin

0.3 ± 0.3

Lutein

1.9 ± 0.8

Zeaxanthin

2.8 ± 1.4

β-carotene

4.0 ± 1.7

The \( a_{\text{ph}}^{*} \)(440) of all data and the extracted data were compared with \( Q_{\text{a}}^{*} \)(675) (Fig. 10). Although the variations of \( a_{\text{ph}}^{*} \)(440) of all data were attributed to not only the package effect but also the effect of the accessory pigments, the \( a_{\text{ph}}^{*} \)(440) of all data showed a significant correlation with \( Q_{\text{a}}^{*} \)(675) (R = 0.88, P < 0.001) because the Car/Chl-a ratio was also correlated with \( Q_{\text{a}}^{*} \)(675) (R = 0.65, P < 0.001). On the other hand, the variation of \( a_{\text{ph}}^{*} \)(440) of the extracted data was assumed to be derived from only the package effect, because the extracted data were affected similarly by the accessory pigments. The extracted data also showed a significant correlation between \( a_{\text{ph}}^{*} \)(440) and \( Q_{\text{a}}^{*} \)(675) (R = 0.76, P < 0.001). An exponential function well represents the variations of \( a_{\text{ph}}^{*} \)(440) in accordance with the variation of \( Q_{\text{a}}^{*} \)(675) and shows the highest correlation coefficients. Therefore, the exponential curves were fitted to all the data, and the extracted data, and Eqs. 5 and 6 were obtained, respectively, as follows:
https://static-content.springer.com/image/art%3A10.1007%2Fs10750-012-1066-4/MediaObjects/10750_2012_1066_Fig10_HTML.gif
Fig. 10

Relations between \( Q_{\text{a}}^{*} \)(675) and \( a_{\text{ph}}^{*} \)(440) of all datasets (open circles, N = 144) and of the set of extracted data points showing similar Car/Chl-a ratios and pigment compositions (closed circles, N = 51). The solid line and dashed line represent the regression lines of all datasets and the extracted dataset, respectively

$$ a_{\text{ph}}^{*} \left( {440} \right) = 0.0099 \times \exp \left[ {1.0753 \times Q_{\text{a}}^{*} \left( {675} \right)} \right] $$
(5)
$$ a_{\text{ph}}^{*} \left( {440} \right) = 0.0130 \times \exp \left[ {0.6976 \times Q_{\text{a}}^{*} \left( {675} \right)} \right] $$
(6)

Equation 5 represents the variation of \( a_{\text{ph}}^{*} \)(440), which was influenced by both the package effect and the effect of accessory pigments, with \( Q_{\text{a}}^{*} \)(675). On the other hand, the variation of \( a_{\text{ph}}^{*} \)(440) shown in Eq. 6 was derived from only the package effect, because of the similar pigment compositions of the extracted data. The slope of Eq. 6 was smaller than that for all data.

Discussion

Variations and parameterization of \( a_{\text{ph}}^{*} \left( \lambda \right) \)

The variations of \( a_{\text{ph}}^{*} \left( \lambda \right) \) critically affect the accuracy of parameterization of \( a_{\text{ph}}^{*} \left( \lambda \right) \) and the following models. The values of \( a_{\text{ph}}^{*} \left( \lambda \right) \) at 440 and 675 nm in Lake Kasumigaura varied greatly. Significant temporal variations of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) were observed, while spatial variations were not observed among sampling sites. In May, \( a_{\text{ph}}^{*} \)(440) and (675) showed significantly elevated values. Zhang et al. (2007) also reported that the slopes of the relation between aph(440) and Chl-a (i.e., \( a_{\text{ph}}^{*} \)(440)) in Lake Taihu, China, were different between summer and winter. The standard deviations of \( a_{\text{ph}}^{*} \left( \lambda \right) \) were decreased by dividing the data into two periods, namely, the month after the spring bloom (i.e., May) and all the other months. The \( Q_{\text{a}}^{*} \)(675) and Car/Chl-a in May were significantly higher compared to the values in the other months, demonstrating that the \( a_{\text{ph}}^{*} \)(440) and (675) in May were affected by a low package effect and a high proportion of accessory pigments.

Le et al. (2009) reported values of \( a_{\text{ph}}^{*} \)(440) (0.056 ± 0.033 m2 mg−1) and \( a_{\text{ph}}^{*} \)(675) (0.021 ± 0.011 m2 mg−1) in Lake Taihu, a highly eutrophic lake. Their \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) values were higher than those obtained in this study (0.026 ± 0.008 m2 mg−1 at 440 nm, and 0.018 ± 0.005 m2 mg−1 at 675 nm). Although their \( Q_{\text{a}}^{*} \)(675) (0.35 ± 0.12) was lower than that in the present study (0.86 ± 0.25), indicating the decrease in \( a_{\text{ph}}^{*} \left( \lambda \right) \) by the package effect in Lake Taihu, the ratio of \( a_{\text{ph}}^{*} \)(440)/\( a_{\text{ph}}^{*} \)(675) was higher than that in the present study (1.50 ± 0.27). These results indicate that the higher values of \( a_{\text{ph}}^{*} \left( \lambda \right) \) in Lake Taihu were derived from the high proportion of carotenoids (i.e., the effect of accessory pigments). Since the \( a_{\text{ph}}^{*} \)(440)/\( a_{\text{ph}}^{*} \)(675) can vary depending on the phytoplankton community (Le et al., 2009), the difference in phytoplankton composition resulted in the different effects of accessory pigments between Lake Kasumigaura and Lake Taihu.

Many studies have reported a significant correlation between the \( a_{\text{ph}}^{*} \left( \lambda \right) \) and the concentration of Chl-a (Bricaud et al., 1995, 2004; Babin et al., 2003; Stæhr & Markager, 2004), and \( a_{\text{ph}}^{*} \left( \lambda \right) \) usually decreases with increasing in Chl-a concentration. On the other hand, other authors have observed that changes in \( a_{\text{ph}}^{*} \left( \lambda \right) \) were independent of Chl-a concentration (Hoepffner & Sathyendranath, 1992). In addition, relatively low \( a_{\text{ph}}^{*} \left( \lambda \right) \) values compared to Chl-a concentrations (as distinct from the usual trend of an decrease in \( a_{\text{ph}}^{*} \left( \lambda \right) \) values with increasing concentration of Chl-a) were observed in the highly eutrophic Lake Taihu (Le et al., 2009). In this study, the low \( a_{\text{ph}}^{*} \)(440) and Chl-a values were observed from August to December in 2009, resulting in the lower correlation coefficient compared to the values observed in other studies (Bricaud et al., 1995; Stæhr & Markager, 2004). In the ocean, it has been reported that the concentration of Chl-a, the package effect and the effect of accessory pigments are all influenced by the nutrient condition (Raven, 1986; Yentsch & Phinney, 1989; Bricaud et al., 1995; Agawin et al., 2000), indicating that the Chl-a and the \( a_{\text{ph}}^{*} \left( \lambda \right) \) are both controlled by the nutrient condition, resulting in the association of their variation. The studies in which a significant correlation was observed between the values of \( a_{\text{ph}}^{*} \left( \lambda \right) \) and the Chl-a concentrations covered a wide range of nutrient conditions from oligotrophic to eutrophic water (Bricaud et al., 1995, 2004; Babin et al., 2003; Stæhr & Markager, 2004). On the other hand, the trophic status (highly eutrophic condition) in Lake Kasumigaura was constant, indicating that there are other factors controlling the \( a_{\text{ph}}^{*} \left( \lambda \right) \) and the Chl-a concentration, such as light availability. The large impact of light condition to the \( a_{\text{ph}}^{*} \left( \lambda \right) \) has been well reported (Moore et al., 1995; Fujiki & Taguchi, 2002). Tomioka et al. (2011) reported that underwater irradiance affects to the bloom of M. aeruginosa in Lake Kasumigaura. Since M. aeruginosa shows distinct pigment composition (especially the one represented by the existing of phycocyanin) from other phytoplankton and small cell size, the light availability can be one of the important factors to determine the package effect and effect of accessory pigment. Thus, the plankton community structure was probably controlled by factors such as light availability other than the nutrient condition in Lake Kasumigaura, resulting in the independent variations of package effect and effect of accessory pigments from the concentration of Chl-a.

The relations between \( a_{\text{ph}}^{*} \left( \lambda \right) \) and the concentration of Chl-a have been used in models of primary production (Bricaud et al., 1995; Stæhr & Markager, 2004). The \( a_{\text{ph}}^{*} \left( \lambda \right) \) values at 440 and 675 nm estimated from the concentration of Chl-a using the relation obtained in this study were higher than those in the ocean (table 5). The cell size and pigment composition is depending on the phytoplankton community (Le et al., 2009). The cell size of diatoms in lake water has been reported to be smaller than that of marine diatoms (Litchman et al., 2009). The cyanobacteria, which usually dominate in highly eutrophic lakes such as Lake Kasumigaura, are classified as picoplanktons (Uitz et al., 2006). These difference of cell characteristics may have resulted in the difference of \( a_{\text{ph}}^{*} \left( \lambda \right) \) between the ocean and eutrophic lake water, although more comprehensive studies examining the difference in cell characteristics between the ocean and lake water are needed.

Contributions of the package effect and the effect of accessory pigments to the variation of \( a_{\text{ph}}^{*} \)(440)

The package effect and the effect of accessory pigments are major factors controlling the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) not only in the ocean (Bricaud et al., 1995, 2004; Babin et al., 2003) but also in lake water (Le et al., 2009). This study obtained two equations representing the relations between \( a_{\text{ph}}^{*} \)(440) and \( Q_{\text{a}}^{*} \)(675) (Eqs. 5, 6). The variation of \( a_{\text{ph}}^{*} \)(440) represented by Eq. 5 was affected by the package effect and the effect of accessory pigments. Therefore, the range of variation of \( a_{\text{ph}}^{*} \)(440) due to both effects was calculated to be 0.030 m2 mgChl-a−1 by Eq. 5 and the maximum/minimum values of \( Q_{\text{a}}^{*} \)(675). On the other hand, Eq. 6 obtained for the extracted data represents the variation of \( a_{\text{ph}}^{*} \)(440) due only to the package effect, since the data were equally affected by the effect of accessory pigments. The range of variation of \( a_{\text{ph}}^{*} \)(440) was calculated to be 0.018 m2 mgChl-a−1 by Eq. 6 and the maximum/minimum values of \( Q_{\text{a}}^{*} \)(675). This value corresponds to 60% of the variation due to both effects (i.e., 0.030 m2 mgChl-a−1), indicating that the package effect and the effect of accessory pigments contributed to 60 and 40% of the variation of \( a_{\text{ph}}^{*} \)(440), respectively. Thus, the package effect dominantly contributed to the variation of \( a_{\text{ph}}^{*} \)(440) in Lake Kasumigaura.

Conclusions

  1. 1.

    The values of \( a_{\text{ph}}^{*} \)(440) and \( a_{\text{ph}}^{*} \)(675) showed significant temporal variations, and constant values were obtained by dividing the period into the post-blooming and other periods.

     
  2. 2.

    The package effect and the effect of accessory pigments in eutrophic lake water varied independently of the concentration of Chl-a, and thus the variation of \( a_{\text{ph}}^{*} \left( \lambda \right) \) was also independent of the concentration of Chl-a in eutrophic lake water.

     
  3. 3.

    The package effect dominantly contributed to the variation of \( a_{\text{ph}}^{*} \)(440) in Lake Kasumigaura.

     

Acknowledgments

This research was supported by the “Global Environment Research Fund by the Ministry of the Environment Japan (B-0909)”. The apposite comments and suggestions of reviewers considerably helped in improving the manuscript.

Copyright information

© Springer Science+Business Media B.V. 2012