1 Introduction

The dissolved organic matter (DOM) of the hydrosphere is characterized up to ca. 4–11% as discrete compounds, e.g., carbohydrates, amino acids, hydrocarbons, phenolic acids, etc. The remaining major fraction is uncharacterised at the molecular level. This complex organic matter comprises humic substances, fulvic acids and other polymeric organic matter (Benner 2002). This major fraction of DOM has been a subject of characterization by optical methods, e.g., absorption and fluorescence. The fraction exhibiting strong absorption of short wavelength (ultraviolet to blue) radiation, is termed as chromophoric dissolved organic matter (CDOM) (Coble 2007). A small fraction of CDOM, ca. 5% that emits fluorescence when irradiated by ultraviolet light (Kalle 1949) is the fluorescent DOM (FDOM). Major sources for FDOM of the coastal water include both external sources (rivers, groundwater, and anthropogenic inputs) and in situ sources (biologically produced and transformed) (Hansell and Carlson 2002; Coble 2007). FDOM has been traditionally used as tracer of different source waters. Recent developments in the identification and quantification of different fluorophores have increased the potential of FDOM to more applications, e.g., phytoplankton bloom monitoring, pollution studies, etc. (Hudson et al. 2007).

The measurement of FDOM requires small sample volumes and does not involve cumbersome isolation and purification steps. Prior studies on FDOM confined to single excitation (ex), emission (em) wavelength combination, i.e., 350 nm (ex)/450 nm (em), e.g., in the Bay of Bengal (Sarma et al. 1994). Since the early nineties, three dimensional excitation emission matrix (EEM) spectra are used increasingly for the characterization of FDOM (Coble et al. 1990, 1998; Coble 1996; Del Castillo et al. 1999; Mayer et al. 1999; Marhaba et al. 2000; Parlanti et al. 2000; Kowalczuk et al. 2003, 2005, 2009, 2010; Yamashita and Tanoue 2003, 2004; Guéguen et al. 2005; Hong et al. 2005). This technique, supplemented with parallel factor (PARAFAC) analysis provides information on various fluorophores that constitute FDOM. Five fluorophores (table 1) attributed to unique sources are commonly found in natural waters (Stedmon et al. 2003; Stedmon and Markager 2005; Murphy et al. 2008; Kowalczuk et al. 2009, 2010). The distribution and concentration of individual fluorophores (A, C, M, T and B) and fluorescence-based indices, e.g., HIX are used diagnostically to draw inferences on the source of DOM, e.g., terrestrial/marine inputs, bacterial/phytoplankton sources, and processes, e.g., humification/degradation (Coble 2007).

Table 1 Fluorophore groups’ identification.
Full size table

The Bay of Bengal is a tropical semi-enclosed basin to the north east of the Indian Ocean and receives a large freshwater influx ca. 1600 km3 yr − 1, i.e., ~50,000 m3 sec − 1 (Subramanian 1993) from the rivers draining India, Bangladesh, Nepal and Myanmar. For an oceanic surface area of 1.13%, this influx constitutes 6.6% of the global river flux. The terrestrial discharges are a vector of organic matter into the Bay. The export flux of total organic carbon (DOC+POC) to the Bay of Bengal by the Ganga–Brahmaputra (G–B) river system at the head Bay is about 650∗109 molC yr − 1 (Aucour et al. 2006). Compared to this, the flux from river Godavari, the largest of the Indian peninsular rivers is estimated as only 63 × 109 molC yr − 1 (Sarin et al. 2002). Fluxes also take place from other peninsular rivers namely Mahanadi, Krishna, Cauvery and Penna, at peak level, during the monsoon season.

Due to the complex hydrographic conditions, the western Bay of Bengal is associated with highly variable bio-optical properties and there exists a large gap of studies in this area. As a part of the Satellite Coastal and Oceanographic Research (SATCORE) programme of Indian National Centre for Ocean Information Services (INCOIS), we initiated to analyse the variability of the optical characteristics of the western coastal Bay of Bengal waters. In this paper we present a detailed study of fluorescence spectroscopic properties of CDOM in the continental shelf waters of Bay of Bengal during southwest monsoon. The objective of the paper was to identify the fluorescence signature and explore different fluorophores of FDOM for identifying the oceanographic conditions.

2 Material and methods

2.1 Study area

Stations are located as shown in figure 1 along 8 transects (T1–T8) off the East Indian coast between Bhimunipatnam and Nagapattinam. The transects are nearly opposite to rivers like Godavari (T2), Krishna (T3), Palar (T7) and Cauvery (T8); else, they are off coastal cities like Visakhapatnam (T1) and Chennai (T6). The T4 and T5 transects are on either side of Penna (Penner in figure 1) estuary. The waters of the four northern transects (T1, T2, T3, T4) are under a greater influence of the southwest monsoon than the southern transects (T5, T6, T7, T8). During this season India receives heavy rainfall and this water reaches mostly the Bay of Bengal after flowing through the Himalayan G–B river system, and peninsular rivers namely Mahanadi, Godavari, Krishna and Cauvery.

Figure 1
figure 1

Study area comprising the eight transects (T1, T2, ..., T8) and station locations (●) along 50, 100 and 200 meters isobaths (the 30 m contour is not shown due to low resolution on map) (R = river).

Full size image

2.2 Sampling and analytical methods

Seawater samples were collected from standard depths using a Seabird Rosette system fitted with 12 Niskin bottles each 1.8 litre capacity and CTD onboard FORV Sagar Sampada from 1–8 September 2009 (Cruise no. 271). The CTD salinities were calibrated with on-deck analysis of discrete samples using guideline 8400 AUTOSAL (Ocean Scientific International, Petersfield, UK). We limited our analysis to surface (i.e., 1 m depth) samples. Samples were filtered immediately after their collection through GF/F filter papers and analysed onboard for nitrite, nitrate, phosphate and silicate by following the standard photometric procedures (Grasshoff et al. 1999) with Unicam α double beam spectrophotometer. Dissolved oxygen was determined using modified Winkler titration method (Carritt and Carpenter 1966). For chlorophyll a, GF/F filters containing the particulate matter were preserved at –20°C until they were extracted with 90% acetone for 24 hr at 4°C and the absorbance measured by UV1800 Shimadzu double beam spectrophotometer in the Andhra University laboratory. The calculation for chlorophyll a (equivalent) concentration was by Jeffrey equations (Jeffrey and Humphrey 1975), and was uncorrected for phaeopigments.

2.3 Fluorescence measurements

For FDOM measurements, a portion of seawater sample was filtered immediately after collection through 0.22 μ Millipore membrane filters, prewashed by soaking for 5 minutes in MilliQ water, under dim light in amber coloured bottles. The fluorescence measurements were done for 4°C stored samples after equilibration to constant room temperature (24 ± 1°C) on Horiba Jobin Yvon Fluoromax-4 spectrofluorometer in the Andhra University laboratory. The spectrofluorometer was equipped with 150 W ozone-free xenon arc-lamp and R928P detector. The Excitation Emission Matrix (EEM) spectra were collected using excitation wavelengths from 250 to 400 nm at 5 nm intervals and the emission wavelengths from 290 to 550 nm, also at 5 nm intervals. The instrument was set-up to collect the signal in ratio mode (S/R) using 5 nm bandwidth on both excitation and emission monochromators with integration time 0.2 sec. For most samples, the absorbance at 350 nm was <0.2 (i.e., 0.02 with 1 cm cuvette). A few samples (10 nos) for which the absorbance was >0.2 were diluted by 5× with Milli-Q (organic free) water to remove inner filter effects (Kowalczuk et al. 2005) for recording the EEM spectrum. The EEM spectra were corrected for instrument configuration by using factory-supplied correction factors. The resultant spectra were Raman calibrated and the intensity expressed as Raman Units (RU). The RU is quantitatively independent of instrument specificities and therefore comparable to measurements from other instruments and on the same instrument at different times. Blank Milli-Q water EEM spectrum was subtracted from all samples to remove water Raman scattering effects. The contour plots of 3D EEM spectra were drawn using Matlab R2007a software.

2.4 PARAFAC analysis

The parallel factor (PARAFAC) analysis extracts components that represent individual fluorophores and their concentrations from the large dataset of 3D EEM spectra. The components are mathematically independent, and each can be considered as a chemical analyte or a group of strongly covarying analytes (Bro 1997). The PARAFAC was done in MATLAB R2007a using the DOMFluor, which contains N way toolbox ver 3.1 (Stedmon and Bro 2008). Before the PARAFAC analysis, high intensity EEM data attributable to the second order Rayleigh scattering were replaced with nil values (Anderson and Bro 2003). Determination of the optimum number of components was done by split half analysis and random initialisation (Stedmon and Bro 2008), as a part of the programme. The PARAFAC analysis was applied to our dataset (26 samples). The PARAFAC analysis gives the relative intensities (scores) of the components, from which the absolute intensities (I) were calculated by multiplying with the corresponding excitation and emission loadings at their λ max (Stedmon et al. 2003; Kowalczuk et al. 2009). For the nth component:

$$ \begin{array}{rll} I_n \!\!\!&=&\!\!\!\mbox{Score}_n \ast \mbox{Loading}_{n({\rm ex})} \left( {\lambda_{{\rm max}}} \right)\\ &&\!\!\!\ast \,\mbox{Loading}_{n({\rm em})} \left( {\lambda_{{\rm max}} } \right), \end{array} $$

where Score n is the relative intensity of the nth component, Loading n(ex) (λ max) is loading of the nth component at the excitation maximum, and Loading n(em) (λ max) is the loading of the nth component at the emission maximum derived from the model.

3 Results

3.1 Hydrography and nutrients

The weather was mostly dry with intermittent small spells of rain during the cruise. All the sampling stations were occupied during daytime (8 AM to 6 PM). At surface, salinity ranged from 28.86 to 34.24. The lowest salinity was at the station closest to the Godavari river mouth (T2, 30 m, figure 1). The highest salinity was also at a coastal station (T8, 30 m). In most transects, the nearshore (30 m) water did not show the typical dilution that must be seen when significant terrestrial fresh water input takes place during the (peak) monsoon season. Indeed, salinity decreased gently in the offshore direction. A general southward increasing salinity contours also characterize the study area (figure 2a). Temperature ranged from 28.48 to 30.47°C and decreased southwards (figure 2b). Dissolved oxygen (DO) concentration ranged from 119.2 to 255.8 μM (table 2; figure 2c). Nutrient concentrations were in the ranges of 0.021–0.23, 0.06–11.17, 0.05–0.36 and 0.06–6.67 μM of nitrite, nitrate, phosphate and silicate, respectively. The chl a equivalent values ranged from 0.15 to 7.24 mgm − 3 averaging at 1.97 mgm − 3, with highest values observed in T3 (figure 2d). A preliminary examination indicated that the study area consists of two distinct regions – the northern region (T1–T4) and the southern region (T6–T8). In T5, the nearshore stations (30–50 m) were similar to the southern region and the offshore stations (100–200 m) were similar to the northern region. The mean concentrations of various parameters analyzed for the two regions (table 2) show significant differences. The northern half (T1–T4 plus T5 offshore) of the study area is associated with lower mean salinity (difference between the two regional means, Δ = 0.87) and silicate (0.96 μM). While an associated decrease might be expected for nitrate, this nutrient indeed showed enrichment (0.59) than the southern half (T6–T8 plus T5 nearshore) (table 2). The DO is supersaturated (+9.8% of saturation concentration) in the northern and undersaturated (−12.8%) in the southern regions.

Figure 2
figure 2

Contours of (a) salinity, (b) temperature (°C), (c) dissolved oxygen (μM), and (d) Chl a (mgm − 3).

Full size image
Table 2 Data summary for the two regions.
Full size table

3.2 Fluorescence properties

Examples of EEM spectra are shown in figure 3. The PARAFAC analysis of EEM spectra extracted five fluorophoric components from the dataset (figure 4). The components 1–5 contain one fluorophore each namely A, M, T, C and B respectively (table 1). The mean abundance was highest for A and for both A and B fluorophores in the northern and southern regions, respectively. Other three fluorophores (M, T and C) were minor in both regions. The A, M, and C flurophores were more enriched in the northern and T and B fluorophores were more enriched in the southern region (figure 5). The A fluorophore is strongly correlated positively to the other two humic fluorophores M (R 2 = 0.97, n = 29) and C (0.99, 29) (figure 6) indicating that the three have related origins. With T, the correlation of A is less significant (0.63), and with B, there is no correlation.

Figure 3
figure 3

Examples of EEM spectra: (a) T5 transect 50 m isobath showing A, B and M peaks, and (b) T7 transect 30 m isobath showing A, T and C peaks.

Full size image
Figure 4
figure 4

PARAFAC components 1 to 5 and their excitation/emission characteristics.

Full size image
Figure 5
figure 5

Mean fraction of fluorophores in the northern transects (T1–T4) + T5 (100, 200 m isobaths) and southern transects (T5–T8) + T5 (30, 50 m isobaths).

Full size image
Figure 6
figure 6

Relationships among A, M and C fluorophores in the two regions.

Full size image

4 Discussion

4.1 North–south variability: hydrography

The lower salinity of the northern waters is because during the southwest monsoon season (June–September), heavy rains in catchment areas over the Indian subcontinent cause floods in the rivers reaching the Bay. The effect of flooding of the Ganga river is far greater – the long term mean Ganga discharge during August–September is around 40,000 m3s − 1 (annual mean: 11,400) (Jian et al. 2009). This water from the head Bay constituting the northern component of the East India coastal current (EICC) modifies the north western Bay of Bengal surface water, the effect decreasing southwards. At the same time, a second component of the EICC moves in the opposite direction from the southwestern Bay, its influence decreasing northwards (Shankar et al. 1996). Flood discharges from peninsular rivers is much smaller ca. 10,000 (annual mean: 3,300) m3s − 1 for Godavari and 5000 (2150) m3s − 1 for Krishna rivers and does not seem to dilute this component. The influence of these rivers may have been confined to near coastal region (<30 m water column) where sampling was not done in this study. In the absence of significant terrestrial influx that can otherwise cause vertical stratification, the southern region is a region of weak upwelling, as shown in CTD profiles. Surface temperature inversion was noticed at many of the southern region coastal stations, as shown for one station (figure 7). Coastal upwelling has been reported to take place away from river mouths off the southwest coast of India (Murty and Varadachari 1968; Shetye et al. 1991). The nitrate to phosphate mean ratio approaches the Redfield ratio of 16 (Redfield 1958) in the southern region (16.9) from a higher ratio (27.4) in the northern region, indicating better autochthonous conditions in the former region.

Figure 7
figure 7

Temperature profile at 30 m isobath station of T5 transect.

Full size image

4.2 Fluorophores: ex-situ and in-situ inputs and better north–south discriminators

The advent of Parafac analysis of EEM spectra (Stedmon et al. 2003) marks a significant new development that has facilitated identification and quantification of new fluorophores previously unrecognized due to the problem of overlap of closely occurring fluorophores, and thereby characterize coastal processes (Stedmon et al. 2003; Stedmon and Markager 2005; Luciani et al. 2008; Murphy et al. 2008; Yamashita et al. 2008; Kowalczuk et al. 2009; Dubnick et al. 2010; Yamashita et al. 2011; Yao et al. 2011). The predominance of A fluorophore in the study area indicates that terrestrial influence is a major factor affecting surface waters of the western Bay of Bengal. We computed the A:C and A:M ratios due to their value as source indicators (Coble 1996). As A fluorophore is distributed between two Parafac components, their total was considered to get the ratios. They were 12.8 and 5.3 for the northern region and 14 and 5.4 for the southern region and compare well with the ratios we recently reported (Chari et al. 2012) for a few stations of the western Bay of Bengal during monsoon season.

The decrease of A fluorophore in the southern region (ΔA = 0.0125 RU; 50%), conforms to the hydrographic situation of increasing marine influence (salinity). The HIX, an indicator of the extent of humification of organic matter (Zsolnay et al. 1999) was computed. It is the ratio of two areas (H/L) of emission spectrum at λ ex of 255 nm (254 nm in literature). These two areas were calculated between 300 and 345 nm for L and between 435 and 480 nm for H. Increased humification of organic material is expected to make the organic material more stable to degradation. The HIX (table 2) is higher in the northern region (ΔHIX = 0.36, 32%), and its DOM is inferred as more of riverine character than the southern region.

Among the protein fluorophores, the T fluorophore which is attributed to phytoplankton production in natural aquatic environment (Coble et al. 1990, 1998) is also higher in the northern region (table 2). But the increase is not as much (ΔT = 0.001 RU, 8%) as the increase in chl a equivalent concentration (Δchl a equivalent = 0.99, 72%) would predict. The fact that B fluorophore, which is also of phytoplanktonic origin is significantly higher in the southern region, and not lower adds to the uniqueness of the study area. The freshly formed proteins have a coiled tertiary structure, and in them, tryptophan fluorescence alone is seen (northern region) as the fluorescence energy of tyrosine, is efficiently transferred to tryptophan by the fluorescence resonance energy transfer (FRET) mechanism (Lakowicz 1999). In degraded proteins (southern region), due to the linear structures, tyrosine residues are exposed directly to the medium and the tyrosine fluorescence is also seen (Mayer et al. 1999). Thus whenever the B fluorophore is noticed, protein degradation via bacterial metabolism is inferred (Determann et al. 1996; Coble et al. 1998; Fellman et al. 2010). The B fluorophore is particularly enriched at a few (one or two) stations of the southern transects, in which dissolved oxygen is lower. It is well known that upwelling does occur in the western Bay of Bengal during southwest monsoon locally (Shetye et al. 1991) that might have brought the protein degradation products surviving bacterial metabolism to the surface.

The new nutrients introduced by upwelling should promote new biological production. As an indicator of recent phytoplankton production, BIX (Huguet et al. 2009) has been computed. The BIX was higher in the southern than the northern region (table 2; ΔBIX = 0.062, 7.3%) despite its lower chl a equivalent than the latter. This infers that phytoplankton production is indeed greater in the southern region and that its (chl a equivalent’s) lower abundance may be due to effective grazing. A southward increase of zooplankton abundance in the western Bay of Bengal has been documented (Nair et al. 1981), and held likely responsible. Lower chl a equivalent of the southern region could also have been due to lower chl a containing phytoplankton, e.g., picoplankton or dinoflagellates. But in time series observations during 2009–2012 (our unpublished results), we did not find dinoflagellates in significant numbers in the monsoon season in the western Bay of Bengal. Thus, fluorescence properties are more informative about phytoplankton production than chlorophyll a.