West Falmouth Harbor is a small (0.7 km2), groundwater-fed estuary on the western shore of Cape Cod, Massachusetts (Fig. 1b). The harbor has a mean depth of approximately 1 m, and is connected to Buzzard's Bay by a 3 m deep, 150 m wide channel. Residence time in the harbor is approximately 1 day (Hayn et al., 2014). Tidal range is 1.9 m during spring tides and 0.7 m during neap tides, with tidal currents at the mouth approaching 0.5 m s-1. The dominant source of freshwater and nutrients is groundwater. Land use surrounding the harbor is largely residential, with influence from a legacy wastewater plume within the aquifer (Ganju et al., 2012). Plant coverage in surrounding wetlands is variable, but Spartina alterniflora and Spartina patens tend to dominate, with some lesser coverage by Juncus gerardii and forbs such as Salicornia spp., Limonium carolinianum, and Solidago sempervirens (Buchsbaum and Valiela, 1987). Zostera spp. eelgrass is also present in the harbor (Del Barrio et al., 2014).

The Barnegat Bay–Little Egg Harbor estuary is a back-barrier system along the New Jersey Atlantic coast (Fig. 1c). The estuary is approximately 70 km long, 2–6 km wide, and 1.5 m deep. Bay and ocean water exchange occurs at three inlets: the Point Pleasant Canal at the northern limit, Barnegat Inlet in the middle of the barrier island, and Little Egg Inlet at the southern limit. Limited exchange through these inlets leads to a spatially variable residence time exceeding 30 days in some locations (Defne and Ganju, 2014). For the purpose of this study, sites north of Barnegat Inlet are referred to as “North Barnegat Bay”, while sites parallel to and south of Barnegat Inlet are referred to as “South Barnegat Bay”. Tides are semidiurnal and range from < 0.1 to 1.5 m, and current velocities range from < 0.5 to 1.5 m s-1 (Kennish et al., 2013; Ganju et al., 2014); there is also a pronounced south-to-north gradient in tidal range and flushing (Defne and Ganju, 2014). While the land surrounding the northern portion of the bay is developed with mixed urban–residential land use, the area south of Barnegat Inlet is less developed and retains much of the original marsh (Wieben and Baker, 2009). The salt marshes south of Barnegat Inlet are dominated by Spartina alterniflora (Olsen and Mahoney, 2001). Freshwater inputs are largest at the northern end of the bay due to the Toms River, Metedeconk River, and Cedar Creek (US EPA, 2007).

Chincoteague Bay is along the Atlantic coast of the Delmarva Peninsula (Fig. 1d). This estuary has an area of 355 km2 and an average depth of 2 m. The watershed surrounding Chincoteague Bay is 487 km2, and consists of 36 % forest, 31 % agricultural development, 25 % wetlands, and 8 % urban development (Bricker et al., 1999). Vegetation in the wetland portion is dominated by Spartina alterniflora, much like South Barnegat Bay (Keefe and Boynton, 1973). Tide range averages 0.5 m, and residence time has been estimated at 8 days (Bricker et al., 1999). The bay is connected to the ocean via two inlets: Ocean City Inlet in the north and Chincoteague Inlet in the south (Allen et al., 2007). Historically, Chincoteague Bay has been marked by extensive seagrass coverage and higher water quality, especially compared to other more developed and less well-flushed bays on the Atlantic coast (Wazniak et al., 2004).

Sampling sites were approached by both land (WF01-WF13, BB01-BB07) and sea (BB08-BB16, CB01-CB10). Sampling occurred from 25 June 2014 to 17 July 2014 (Table 1). Either a bucket (sites approached on foot) or 1 L Nalgene sampling bottle (sites approached by boat) was rinsed with native water and then used to collect a surface water sample. A pre-calibrated YSI EXO 2 multisonde, measuring fDOM, temperature, salinity, pH, turbidity, chlorophyll a fluorescence, blue-green algae fluorescence, and dissolved oxygen concentration, was placed in each sample. Excitation and emission wavelengths for the fluorescing dissolved organic matter sensor were 365 nm (±5 nm) and 460 nm (±40 nm), respectively. Measurements of each parameter were collected at 1 s intervals for approximately 60 s and averaged. For sites approached on foot, the YSI EXO was deployed immediately; for sites approached by boat, the YSI EXO was deployed later on land (in concurrence with absorbance measurements, as described below).

Temperature, turbidity, and inner filter effects (IFEs) have been shown to alter fluorescence measurements (Baker, 2005; Downing et al., 2012). For this reason, we corrected fluorescence measurements to account for temperature, turbidity, and IFEs, according to Downing et al. (2012).

A 60 mL syringe was used to draw a water sample from these buckets for absorbance measurements. Fifteen milliliters of this sample was filtered through a 0.2 µm inorganic membrane filter into a 5 cm path length cuvette. Absorbance measurements were recorded in 20 nm increments over the range of 340–440 nm (West Falmouth Harbor) or 340–720 nm (Barnegat Bay and Chincoteague Bay). Spectral slope was calculated over both the entire 340–720 nm range and the 340–440 nm range for Barnegat Bay and Chincoteague Bay to allow for direct comparison to West Falmouth Harbor and other studies (e.g., Huang and Chen, 2009; Del Castillo et al., 1999). The estimated photometric accuracy of the spectrophotometer was 0.003 absorbance units. Offsets from zero were determined for the West Falmouth Harbor CDOM spectra by running a blank sample (Milli-Q water) at 440 nm (the high end of the recorded spectrum). For Barnegat Bay and Chincoteague Bay, offsets from zero were determined by running a blank sample before measurement at each wavelength (340–720 nm). Absorbance measurements were converted to Naperian absorption coefficients as follows: a(λ)=2.303A(λ)/l, where A(λ) is the absorbance at 340 nm, l is the cell length in meters (0.05 m for this study), and a(λ) is the absorption coefficient (Green and Blough, 1994). Absorbance values at 340 nm were the highest across the range scanned, so 340 nm was chosen as the absorbance wavelength for calculating the absorption coefficient. Spectral slopes were calculated by plotting the natural log of absorption coefficient against wavelength. Due to use of the natural log, non-positive absorption coefficients were discarded to calculate spectral slope, as described in Eq. (2) (Bricaud et al., 1981): S=ln⁡(a(λ)/a(r))(r-λ), where λ is wavelength, r is a reference wavelength, a(λ) is absorption coefficient at a given wavelength, a(r) is absorption coefficient at the reference wavelength, and S is the spectral slope. The value of S shows the rate at which absorption decreases with increasing wavelength (Green and Blough, 1994). This parameter can be used to predict absorption coefficients across the spectrum based on absorption at one reference wavelength (Bricaud et al., 1981).

At each site in West Falmouth Harbor and Barnegat Bay, water samples were collected for stable carbon isotope analysis of DOC (dissolved organic carbon). Chincoteague Bay was excluded due to logistical limitations. Thirty milliliters of the collected sample was filtered through a 0.2 µm inorganic membrane filter, collected in a 40 mL glass autosampler vial that had been baked at 450 ∘C for 4 h, and sealed with caps and Teflon-faced silicon septa that had been soaked and rinsed with 10 % (by volume) HCl. Additionally, trace metal grade 12N HCl (Sigma-Aldrich) was added to each isotope water sample to achieve pH < 2. The vials were then stored at 4 ∘C. Samples were analyzed by high-temperature combustion–isotope ratio mass spectrometry (HTC-IRMS) at the USGS-WHOI Dissolved Carbon Isotope Lab (DCIL), as described by Lalonde et al. (2014). The DCIL HTC-IRMS system consists of an OI 1030C total carbon analyzer and a Graden molecular sieve trap interfaced to a Thermo-Finnigan DELTAplus XP IRMS via a modified Conflo IV. The stable carbon isotope ratios are reported in the standard δ notation relative to Vienna Pee Dee Belemnite (VPDB) and are corrected by mass balance to account for the analytical blank, which was less than the equivalent of 15 µM DOC in the sample. By comparison, the sample DOC concentrations ranged from 60.7 to 581 µM. Thus the blank correction was always less than 25 % of the sample concentration. The analytical precision of the δ13C analysis was less than 0.3 ‰. DOC concentration was calculated using a standard curve consisting of four potassium hydrogen phthalate (KHP) calibration standards quantified as the integrated volt-seconds (Vs) of the mass-44 peak on the IRMS (Lalonde et al., 2014). Peak areas were corrected for analytical blanks determined from ultrapure lab water injections.

Salinity and δ13C values for freshwater and marine end-members from West Falmouth Harbor and Barnegat Bay were used to construct isotope mixing models for the estuaries (Kaldy et al., 2005). Marine and freshwater end-members are defined as the most and least saline samples collected at each estuary. Because of the number of samples clustered near the highest salinity for each estuary, marine end-members were checked with geographic location. For West Falmouth Harbor, the site chosen as marine end-member (WF01) was taken from the mouth of the harbor where the estuary connects to Buzzard's Bay. For Barnegat Bay, the site of highest salinity (BB13) was taken from the middle of Little Egg Harbor in South Barnegat Bay. However, a more geographically intuitive marine end-member would be site BB16, near Little Egg Inlet. The only slightly lower salinity at this site (29.69 psu) as compared to BB13 (30.08 psu), along with the geographic location of BB16 at an oceanic inlet, makes BB16 a more appropriate marine end-member. Therefore, end-members used in the conservative mixing models were as follows: WF06 (freshwater), WF01 (marine), BB01 (freshwater), and BB16 (marine). The conservative mixing models (Kaldy et al., 2005) were constructed as Cmix=fCR+(1-f)CO, where Cmix is the calculated concentration for use in the mixing model, CR and CO are freshwater and marine end-member DOC concentrations, respectively, and f is the fraction of freshwater calculated from salinity: f=(SO-SM)/(SO-SR), where SM is measured salinity at a specific site, and SR and SO are freshwater and marine end-member salinities, respectively. These calculations lead to the modeled isotope ratio of each sample as δmix=[fCRδR+(1-f)COδO]/Cmix, where all subscripts and variables are the same as described for Eqs. (3) and (4).

In addition to the stable carbon isotope analysis, a “carbon-normalized CDOM” (C-normalized CDOM340) was calculated for each sample as C-normalized CDOM340=A(λ)/DOC, where DOC is dissolved organic carbon concentration (mg L-1) and A(λ) is decadic light absorbance at 340 nm (m-1). This C-normalized CDOM340 is comparable to specific ultraviolet absorbance (SUVA), a measure proven to correlate strongly with DOC aromaticity (Weishaar et al., 2003). While SUVA is typically calculated at 254 nm, the C-normalized CDOM340 calculated here provides a similar measure while accommodating this study's minimum absorbance measurement wavelength of 340 nm.

The estuary-wide average spectral slope (over the range 340–440 nm) for West Falmouth was steeper than for Barnegat and Chincoteague, with Savg equal to 0.021, 0.016, and 0.018, respectively (Table S1). At West Falmouth Harbor, spectral slope ranged from 0.013 to 0.044, with a standard deviation of 0.010. At Barnegat Bay, S ranged from 0.011 to 0.019, with a standard deviation of 0.002. At Chincoteague Bay, S ranged from 0.014 to 0.023, with a standard deviation of 0.003. Spectral slope values for Barnegat and Chincoteague were slightly steeper over the range 340–440 nm as compared to S calculated over the range 340–720 nm (Table S1 in the Supplement).

At West Falmouth, fDOM ranged from 0.63 to 10.21 QSU, with a standard deviation of 2.57 QSU. At Barnegat Bay, fDOM ranged from 12.06 to 84.40 QSU, with a standard deviation of 20.82 QSU. At Chincoteague Bay, fDOM ranged from 11.15 to 49.49 QSU, with a standard deviation of 10.95 QSU. Values observed for fDOM were within ranges reported for similar estuaries and coastal waters (Callahan et al., 2004; Clark et al., 2002; Green and Blough, 1994). Sites at West Falmouth and Barnegat Bay represented a freshwater to seawater gradient, with salinity ranging from 0.13 to 31.28 psu at West Falmouth and 3.41–30.08 psu at Barnegat. At Chincoteague Bay, salinity ranged from 25.88 to 31.85 psu. A complete salinity gradient was not sampled at Chincoteague due to the relatively high salinity found throughout the main basin of the bay, and low freshwater input. fDOM correlated inversely with salinity (Fig. 2), as expected because riverine input is typically the main external source of DOM. However, the slope and strength of the fDOM–salinity relationship differed both between and within estuaries. The steepest relationship (most rapidly decreasing fDOM signal with increasing salinity) was observed at Chincoteague Bay and in South Barnegat Bay. These two areas displayed a similar fDOM–salinity relationship; fDOM and salinity showed a slightly less negative relationship at North Barnegat Bay, and even less negative at West Falmouth Harbor.

At West Falmouth, a(340) ranged from 0.92 to 5.07 m-1, with a standard deviation of 1.02 m-1. At Barnegat Bay, a(340) ranged from 0.97 to 14.97 m-1, with a standard deviation of 3.99 m-1. At Chincoteague Bay, a(340) ranged from 1.84 to 8.38 m-1, with a standard deviation of 1.86 m-1 (Table 2). The ratio between a(340) and fDOM differed both between and within estuaries, as expected (Table S1; Fig. 3). The mean ratio of a(340) to fDOM was relatively higher in West Falmouth Harbor (1.22) than in Barnegat Bay (0.22) and Chincoteague Bay (0.17). There were two significant outliers at Barnegat Bay: BB01, which had a lower absorption coefficient (0.97 m-1) than expected based on its higher fDOM value (69.92 QSU), and BB15, which showed a much higher absorption coefficient (14.97 m-1) than expected based on its lower fDOM value (16.50 QSU). West Falmouth also demonstrated substantial variability in a(340) : fDOM ratio between sites. Chincoteague Bay, however, showed a highly consistent ratio.

The observed isotope–salinity relationship at West Falmouth Harbor and Barnegat Bay had numerous δ13C values well outside the range predicted by concentration and isotopic conservative mixing models (Table S2; Figs. 4a and 5a), which suggests an additional DOM source from within the estuaries (discussed further in Sect. 5.3). For West Falmouth Harbor, end-members of the conservative mixing model had δ13C values of -23.0 and -26.1 ‰. The observed δ13C data, however, ranged from -19.7 to -26.1 ‰, six of which were more 13C-enriched samples than the modeled range. For Barnegat Bay, end-members of the conservative mixing model had δ13C values of -22.1 and -26.7 ‰. The observed δ13C data ranged from -20.8 to -26.7 ‰, four of which were more 13C-enriched than the modeled range. The two points from North Barnegat Bay falling well above the model (Fig. 5a) correspond to sites BB04 and BB09. The two points from South Barnegat Bay falling well above the model correspond to sites BB12 and BB14. These 13C-enriched samples from Barnegat were all taken from areas near significant stretches of marsh along the western edge of Barnegat Bay. Furthermore, these samples all fall above the concentration-based mixing model for Barnegat Bay (Fig. 5b). Spatial representation of δ13C values at Barnegat Bay (Fig. 5c) shows significantly less negative δ13C values in South Barnegat Bay compared to North Barnegat Bay.

Comparison of the isotopic and optical analyses suggests a correlation between δ13C signature and fDOM-CDOM absorption ratio (Fig. 6). For both West Falmouth Harbor and Barnegat Bay, the more 13C-enriched samples also had a higher absorption coefficient per unit fluorescence. This trend is highlighted by the extremes of the data set, with the most 13C-enriched sample (WF02) displaying the highest CDOM : fDOM absorption ratio, and the least 13C-enriched sample (BB01) displaying the lowest CDOM : fDOM absorption ratio. Furthermore, West Falmouth Harbor samples had both higher CDOM : fDOM absorption ratios (-0.032, natural log scale, average) and 13C enrichment (δ13C average of -22.4 ‰) as compared to Barnegat Bay (-1.75 and -23.4 ‰, respectively).

Absorption coefficients for West Falmouth and Chincoteague were comparable to those reported for other estuaries and coastal waters (Chen et al., 2003; Green and Blough, 1994). Absorption coefficients for Barnegat Bay were somewhat higher, but within the range reported by Green and Blough (1994). Likewise, all values observed for spectral slope were within ranges reported for similar estuaries and coastal waters (Keith et al., 2002; Green and Blough, 1994), despite differences in the range over which spectral slope was calculated (400–550 nm for Keith et al., 2002; 290 nm to wavelength of absorption detection limit for Green and Blough, 1994). At Barnegat Bay and Chincoteague Bay, the range of calculated spectral slopes was quite small (Table S1). At West Falmouth Harbor, however, there was significantly more variability in spectral slope. West Falmouth Harbor is a relatively dynamic system with multiple freshwater point sources and unique mixing characteristics (Ganju et al., 2012). Considering the dramatic influence that variable sources (aquatic vs. terrestrial) and alterations (e.g., microbial and photodegradation) have on the optical properties of DOM (Spencer et al., 2009; Helms et al., 2008; De Souza Sierra et al., 1994), the variability in spectral slopes observed at West Falmouth Harbor may be attributable to the physical complexity and short residence time of this estuary. More specifically with respect to source, previous studies have shown that DOM comprised of primarily fulvic acids has steeper spectral slopes than DOM comprised of primarily humic acids (Carder et al., 1989). Considering the physical complexity and variety of point sources at West Falmouth Harbor, variable organic matter composition and spectral slope is not surprising.

The inverse relationship between fDOM and salinity observed for these three estuaries is consistent with other estuarine studies (Clark et al., 2002; Green and Blough, 1994). Differing slopes of the inverse relationships suggests the freshwater DOM sources vary between and within estuaries. This is due to differences in organic matter composition and fluorescence between the freshwater sources (Stedmon et al., 2003; Parlanti et al., 2000). South Barnegat Bay and Chincoteague Bay display a similar fDOM–salinity relationship, while South Barnegat Bay and North Barnegat Bay show a divergent relationship. South Barnegat Bay and Chincoteague Bay also have geographic and land use similarities with less development and extensive Spartina alterniflora-dominated marshes (Wieben and Baker, 2009; Olsen and Mahoney, 2001; Keefe and Boynton, 1973), whereas North Barnegat Bay is much more developed (Wieben and Baker, 2009). Furthermore, North and South Barnegat Bay appear to have different organic matter sources (determined via isotope analysis; see Sect. 5.3). This information considered together supports the idea of differing organic matter sources due to various inputs affecting fluorescence properties. As for the variability seen within West Falmouth Harbor, this is again likely attributable to the relatively low fluorescence signals observed throughout the estuary, along with the variety of freshwater inputs to this complex system.

The disparity between observed δ13C values and those predicted by conservative mixing models (Figs. 4a and 5a) suggests an additional DOM source within the estuaries. Previous studies of DOC in eastern US estuaries have suggested a marine end-member δ13C value of -24 to -22 ‰, and a freshwater end-member δ13C of -28 to -26 ‰ (Peterson et al., 1994). Observed values falling above the mixing model and approaching much more 13C-enriched values than the defined marine end-member are likely due to the influence of DOC from Spartina spp. cordgrass in nearby salt marshes. Analysis of DOC Spartina spp. by past studies has indicated a δ13C signature of about -16.4 to -11.7 ‰ (Komada et al., 2012; Chmura and Aharon, 1995). The tendency of values from this study towards this 13C-enriched signature, in combination with knowledge of Spartina coverage around the sites differing from conservative mixing models, suggests a DOM source derived from Spartina cordgrass. The influence of this end-member is particularly notable in South Barnegat Bay (specifically sites BB12 and BB14), where Spartina coverage is extensive (Olsen and Mahoney, 2001), and the δ13C of the DOC is -21.6 and -20.9 ‰ for BB12 and BB14, respectively. Although Spartina coverage in North Barnegat Bay is not as extensive as in South Barnegat Bay, the sites with DOC δ13C values that are more enriched than the conservative mixing model for North Barnegat Bay (BB04 and BB09) were taken from inland sampling locations, specifically the north bank of the lower Toms River and Reedy Creek, where stands of Spartina are present.

However, the observed 13C enrichment could also be attributed to Zostera eelgrass, which has been shown to exhibit a 13C-enriched signature (Hemminga and Mateo, 1996). For this reason, the aforementioned samples falling well above the conservative mixing models cannot necessarily be considered a result of Spartina influence. However, a comparison of site locations to known seagrass and Spartina wetland coverage can yield some indication of the most likely source of 13C-enriched DOC. Seagrass coverage maps (Lathrop and Haag, 2011) and maps of estuarine intertidal wetland coverage (United States Fish and Wildlife Service, 2015) for Barnegat Bay show intertidal wetland coverage and no seagrass coverage for sites BB09, BB12, and BB14. Site BB04 is characterized by neither type of coverage, but its inland location places it much closer to known intertidal wetland coverage (US Fish & Wildlife Service, 2015). This geographic comparison indicates Spartina as the more likely additional end-member at Barnegat Bay, though Zostera influence is still possible. Considering the movement of water and potential for mixing during residence in the estuary, this geographic analysis is by no means definitive, but does provide some insights.

For West Falmouth Harbor, sites falling well above the conservative mixing model (WF02, WF03, WF04, WF05, WF07, WF11) were compared to known seagrass (Del Barrio et al., 2014) and intertidal wetland (US Fish & Wildlife Service, 2015) coverage for West Falmouth Harbor. For sites WF03, WF05, WF07, and WF11, there is known intertidal wetland coverage and no known Zostera coverage. For site WF02, there is both intertidal wetland coverage and Zostera coverage, whereas WF04 corresponds to neither Spartina nor Zostera. This comparison yields a less clear picture of DOC sources, but this is to be expected considering the aforementioned complexity of surrounding land uses, potential DOC inputs, and limited mixing at West Falmouth Harbor. Furthermore, spatial representation of δ13C values at West Falmouth Harbor (Fig. 4c) show 13C-depleted samples in the northeastern corner of the harbor, the location of a freshwater culvert discharging groundwater (Ganju, 2011). On the whole, the conservative mixing models used in this study may not be appropriate for a system as complex as West Falmouth Harbor. Unlike the clear indication of a third end-member from the mixing model for Barnegat Bay, one could envision a more complex system with multiple additional end-members for West Falmouth Harbor (Fig. 4a and b).

We also considered the potential influence of photodegradation on the samples with DOC that was 13C-enriched in comparison to the conservative mixing model. Irradiation experiments have shown that riverine DOC becomes 13C-enriched by ∼ 3.5 ‰ and concentrations decrease by as much as 45 % over 57 days as a result of photodegradation (Spencer et al., 2009), suggesting the possibility that the aforementioned 13C-enriched samples are photodegraded terrestrial DOM. This is unlikely for samples from West Falmouth Harbor, given the very short residence time of this estuary (∼ 1 day; Hayn et al., 2014). For Barnegat Bay, however, the influence of photodegradation is possible. Sites BB12 and BB14 are in areas with residence time of ∼ 10 days, while sites BB04 and BB09 are in areas with residence time of ∼ 15–20 days (Defne and Ganju, 2014). These residence times are within the time frame over which photodegradation effects on δ13C have previously been observed (Spencer et al., 2009), which could also influence the 13C-enriched signatures observed for these samples. However, the relative lack of 13C enrichment observed at other Barnegat Bay sites with even longer residence times (e.g., BB03 and BB07; Defne and Ganju, 2014) implies that photodegradation alone likely does not explain the 13C-enriched signatures found for certain Barnegat Bay samples. Furthermore, and most convincing, the concentration-based mixing model for Barnegat Bay (Fig. 5b) demonstrates a net input of DOC into the estuary. DOC concentrations that exceed the conservative concentration-based mixing model indicate a source of DOC within the estuary. If the samples were affected by photodegradation, one would expect a net loss of measured DOC within the estuary (e.g., Spencer et al., 2009).

Further insight into the possibility of photodegradation can be derived from the C-normalized CDOM340 (Table S2). Carbon-normalized CDOM correlates strongly with sample aromaticity (Weishaar et al., 2003), which one would expect to decrease as a result of photodegradation (Hood et al., 2005). However, C-normalized CDOM340 (and thus aromaticity) is not significantly lower for the potentially photodegraded terrestrial DOM samples as compared to other terrestrial DOM samples such as BB01 and BB03 (Table S2). This lack of a drop in aromaticity does not support the possibility that the 13C-enriched samples from Barnegat Bay are photodegraded terrestrial DOM.

The variability between fDOM and CDOM absorption in these estuaries was expected based on the results of previous studies (Clark et al., 2004; Del Castillo et al., 1999; Hoge et al., 1993). West Falmouth Harbor in particular showed a different absorption coefficient to fDOM ratio as compared to the general trend for Barnegat and Chincoteague Bays (Fig. 3). We ascribe this difference to groundwater inputs, which have been shown to have lower CDOM (Shen et al., 2015; Chen et al., 2010; Huang and Chen, 2009) and are substantial in West Falmouth Harbor (Ganju, 2011). Additionally, the extremes of CDOM variability in this study can be explained by differing DOC sources within the estuaries. While the relatively uniform CDOM–fDOM relationship for Barnegat Bay results in clustering of Barnegat Bay samples (Fig. 6), this relationship is highlighted by both the Barnegat Bay outliers and the higher CDOMabs : fDOM observed for the more 13C-enriched samples at West Falmouth Harbor. Points such as the outliers at Barnegat Bay are indicative of how the CDOM–fDOM relationship can be altered in an estuary with such diverse sources and transport mechanisms. This assertion of variable CDOM–fDOM relationship depending on source is supported by the findings of Tzortziou et al. (2008), which suggested that marsh-exported DOC has a lower fluorescence per unit absorbance as compared to humic DOC originating from a freshwater source. For the two extreme outliers, 13C-enriched DOC (likely Spartina source) was associated with a lower fluorescence per unit absorbance. 13C-depleted DOC (terrestrial source) was associated with a higher fluorescence per unit absorbance. While other studies have focused on differences in the fluorescence–absorbance relationship as a function of molecular weight (Belzile and Guo, 2006; Stewart and Wetzel, 1980), the combination of CDOM optical and isotopic analyses presented here provide a connection between CDOM source and optical characteristics, as suggested by Tzortziou et al. (2008).

The effects of in situ processing on absorption properties of DOM must also be considered here. In particular, photodegradation is known to reduce the absorbance of light by DOM (Spencer et al., 2009; Kouassi and Zika, 1992). Therefore, observations of higher fluorescence per unit absorbance could be a result of photochemical effects. However, the 13C-enriched DOC samples discussed here exhibit lower fluorescence per unit absorbance than expected. This trend provides additional evidence refuting the aforementioned possibility that the 13C-enriched samples from Barnegat Bay are photodegraded terrestrial DOM (Sect. 5.4).

The variability in fDOM optical properties between and within estuaries has important consequences for light attenuation models. Continuous estimates of light attenuation are possible with continuous proxy measurements of turbidity (for sediment), chlorophyll a fluorescence, and fDOM (Gallegos et al., 2011), but Ganju et al. (2014) found that light models can be highly sensitive to the CDOM–fDOM relationship, specifically in Barnegat Bay. We applied the light model of Gallegos et al. (2011) to the individual measurements of turbidity, chlorophyll a fluorescence, and fDOM collected in this study. We explored two cases to calculate light attenuation: (1) use of the individual point CDOM : fDOM ratio and spectral slope from measurements and (2) use of an estuary-wide average CDOM : fDOM ratio and spectral slope (model parameters related to sediment particles and chlorophyll were held constant to values reported in Ganju et al., 2014). Variation in the DOM properties led to average light attenuation errors ranging from 11 to 33 % (Table 2), with individual site errors over 200 % at sites with the highest deviation from the estuary mean (site BB01, at the landward end of Barnegat Bay). This suggests that constraining optical properties of the DOM pool is critical for light modeling, and that high variability within an estuary may confound use of spatially constant parameters.