Magnitude and composition of sinking particulate phosphorus fluxes in Santa Barbara Basin, California

Authors

  • Emily Sekula-Wood,

    1. Department of Earth and Ocean Sciences, University of South Carolina, Columbia, South Carolina, USA
    Search for more papers by this author
  • Claudia R. Benitez-Nelson,

    Corresponding author
    1. Department of Earth and Ocean Sciences, University of South Carolina, Columbia, South Carolina, USA
    2. Marine Science Program, University of South Carolina, Columbia, South Carolina, USA
    • Corresponding author: C. R. Benitez-Nelson, Department of Earth and Ocean Sciences, University of South Carolina, 701 Sumter St., EWS 408, Columbia, SC 29208, USA. (cbnelson@geol.sc.edu)

    Search for more papers by this author
  • Melissa A. Bennett,

    1. Department of Earth and Ocean Sciences, University of South Carolina, Columbia, South Carolina, USA
    2. Marine Science Program, University of South Carolina, Columbia, South Carolina, USA
    Search for more papers by this author
  • Robert Thunell

    1. Department of Earth and Ocean Sciences, University of South Carolina, Columbia, South Carolina, USA
    2. Marine Science Program, University of South Carolina, Columbia, South Carolina, USA
    Search for more papers by this author

Abstract

[1] The composition and bioavailability of particulate P influence marine biological community production on both modern and geologic time-scales, and continental margins play a critical role in the supply, modification, and storage of particulate P. This study examined particulate P cycling in the Santa Barbara Basin (SBB) off the coast of southern California using a ∼520 m deep-moored sediment trap deployed from 1993–2006 and a sediment core collected in 2005 directly beneath the sediment trap at 590 m. Total particulate P (TPP), particulate inorganic P (PIP), and particulate organic P (POP) were quantified using a 5-step sequential extraction method (SEDEX) that chemically separates PIP into loosely bound, oxide-bound, authigenic, and detrital P phases. POP fluxes, while similar in magnitude to other coastal regions (22 ± 10 μmol m−2 d−1) were a small component of the TPP pool (15%). Seasonal trends revealed significant increases in POP fluxes during upwelling due to increased biological production in surface waters by organisms that increased mineral ballast. High particulate organic carbon (POC) to POP ratios (337 ± 18) further indicated rapid and efficient remineralization of POP relative to POC as particles sank through the oxic water column; however, further reduction of POP ceased in the deeper anoxic waters. Loosely bound, oxide-bound, and authigenic P, dominated the TPP pool, with PIP fluxes substantially higher than those measured in other coastal settings. Strong correlations between oxide-associated, authigenic, and detrital P fluxes with lithogenic material indicated a terrestrial source associated with riverine discharge. Furthermore, more than 30% of the loosely bound and oxide-bound P was remineralized prior to burial, with the magnitude of dissolution far exceeding that of POP. These results highlight the dynamic nature of the particulate P pool in coastal ecosystems and how changes in P source can alter the composition and lability of P that enters coastal waters.

1. Introduction

[2] Continental shelves play a crucial role in the biogeochemical cycling of a number of elements [Liu et al., 2010]. These regions act as the interface between continental interiors and the open ocean via nutrient-rich river and groundwater discharge. High rates of biological production coupled with shallow water column depths further make coastal shelves disproportionately large depocenters of organic matter that can be remobilized as a result of climate induced changes in sea level. It has been estimated that ∼40% of the carbon burial in the global oceans occurs on continental margins [Müller-Karger et al., 2005]. The biological transformations that occur within these systems, however, remain difficult to constrain due to a combination of dynamic physical and biological processes coupled with anthropogenic activities. In the marine phosphorus (P) cycle, continental shelves account for more than 70% of the total P removed from the ocean each year [Wallmann, 2010]. Yet much of the P that is produced in coastal surface waters is not buried deep in the sediments, but rather recycled in the water column and surface sediments. Understanding the mechanisms that influence this P remineralization has a number of global implications, including the ocean's role in climate change.

[3] P is generally considered to limit marine production over geologic time-scales (>1000 years) [Benitez-Nelson, 2000; Paytan and McLaughlin, 2007], but the exact mechanism of how this limitation occurs remains controversial. One scenario is the shelf-nutrient hypothesis, where reduction in continental shelf area due to lower sea level during glacial times increases the net transfer of P containing material directly into the deep ocean [Broecker, 1982; Filippelli, 2008]. By bypassing the continental shelf, sinking particles are remineralized in the water column rather than buried. As a result, P recycling is enhanced and the inventory of dissolved P increases. P-enriched deep waters are subsequently brought back into the surface ocean via mixing, which enhances surface water primary production and the continued drawdown of atmospheric CO2, creating a positive feedback loop.

[4] Another possible hypothesis linking P to global climate change involves the role of oxygen. It has been hypothesized that during glacial periods increased primary production in surface waters results in a greater quantity of organic material reaching the deep ocean; thereby intensifying oxygen demand [e.g., Wallmann, 2003]. Although inconsistent, a number of studies suggest that particulate P is preferentially lost to the dissolved phase faster than particulate organic carbon (POC) is under low oxygen conditions, mainly as a result of the reduction of metal oxides and more rapid degradation of particulate organic P (POP) relative to POC [Van Cappellen and Ingall, 1994; Wallmann, 2003; Ingall et al., 2005; Wallmann, 2010]. This particulate P loss again results in an increased dissolved P inventory, creating a positive feedback loop. While the outcome is the same, the two hypotheses rely on different mechanisms that ultimately depend on the magnitude of the relative loss rates of various particulate P compounds to the dissolved phase as they sink through the water column and enter the underlying sediments.

[5] In the modern ocean, recent studies suggest that P may also limit primary production [Benitez-Nelson, 2000; Sylvan et al., 2006; Paytan and McLaughlin, 2007]. Riverine derived P is by far the largest source of P to the coastal ocean. Much of that P, however, is in the particulate phase, and is generally thought to be removed rapidly in the nearshore via sedimentation, an assumption that remains poorly tested [Benitez-Nelson, 2000]. Over the last century, coastal inputs of P have been significantly modified as a result of human activities that include agriculture, deforestation, and population growth. The ultimate impact of these alterations remain enigmatic; while P inputs may increase over the short-term, depletion of known terrestrial P reservoirs and reduction in particulate P due to river dams counteract these effects over longer timescales [Filippelli, 2008]. Furthermore, nitrogen (N) loading is also occurring, and the relative balance between the bioavailability of these two major nutrients is a key component in biological community composition [Benitez-Nelson, 2000; Sylvan et al., 2006; Paytan and McLaughlin, 2007]. Nutrient loading also leads to coastal eutrophication and a subsequent increase in both nuisance and toxin-producing algal blooms that often lead to low oxygen environs [Nixon, 1993; Anderson et al., 2002; Sylvan et al., 2006]. Therefore, understanding the composition and bioavailability of P inputs into the coastal ocean is critical for understanding current and future coastal biogeochemistry.

[6] Sinking particles play an integral role in the transport of P from surface waters through the mesopelagic and into bottom sediments. While most studies assume that the majority of sinking P is composed of marine biologically produced organic matter, others have shown that a significant fraction of the sinking total particulate P (TPP) pool in coastal settings is composed of a variety of inorganic phases as well [Paytan et al., 2003; Faul et al., 2005; Benitez-Nelson et al., 2007; Lyons et al., 2011]. These various inorganic and organic P components will be influenced in very different ways in response to changes in water column biogeochemistry.

[7] Here, we report on a 13-year time series (August 1993–December 2006) of sinking particulate P from a sediment trap deployed at ∼520 m depth in the Santa Barbara Basin, along the continental margin of southern California. Using a sequential extraction method (SEDEX), we quantify concentrations of particulate organic P and four chemically defined fractions of particulate inorganic P [Ruttenberg, 1992]. Specifically, we examine (1) if the relative magnitude and composition of P fluxes vary with season (e.g., upwelling versus non-upwelling), (2) if relationships exist between mineral (biogenic silica, carbonate and lithogenic) fluxes and particulate P, (3) if the composition and elemental ratios of sinking particles and bottom sediments suggest preferential remineralization or transformation of specific constituents, and (4) the potential implications of these results for the global P cycle.

2. Study Site

[8] Santa Barbara Basin (SBB) is located off the coast of southern California and is the northernmost extent of the Southern California Bight. Bordered by the California coastline on the north and the Channel Islands on the south, the SBB trends east-west and is approximately 100 km long and 50 km across (Figure 1). SBB has a maximum depth of ∼590 m and is flanked on the east and west by ∼200 and 475 m depth sills that restrict subsurface flow into the basin [Thunell, 1998]. Limited ventilation of bottom waters coupled with high oxygen demands associated with decomposition of organic matter result in a depletion of oxygen in bottom waters [Sholkovitz and Gieskes, 1971]. Suboxic conditions (≤2 mL O2 L−1) begin at 200 m and waters approach anoxia by 500 m depth.

Figure 1.

Map of Santa Barbara Basin. The sediment trap mooring and sediment core (indicated by an inverted triangle) are deployed near the center of SBB in a total water depth of ∼590 m.

[9] Nutrient delivery into surface waters is dominated by regional wind-driven upwelling and riverine discharge [Warrick et al., 2005; Anderson et al., 2006; McPhee-Shaw et al., 2007]. Strong northwesterly winds cause intense Ekman transport and upwelling along the north-south oriented west coast of the United States during spring and early summer. This strengthens the southward flowing California Current (CC), which injects cool, nutrient rich water into the western portion of the SBB. This seasonal upwelling (defined here as a shoaling of the 12°C isotherm at 20 m depth) is further influenced by large scale climate patterns such as the El Niño-Southern Oscillation (ENSO) [Bernal, 1981] (Figure 2a).

Figure 2.

Seasonal variations in temperature, riverine discharge, and elemental fluxes in SBB Basin from August 1993–December 2006. (a) A composite depth-temperature profile from monthly CTD casts taken at a station (34°13′58.8″N, 120°2′9″W) near the sediment trap site. The solid black line highlights the 12°C isotherm used to define upwelling versus no-upwelling periods; note the influence of the strong 1997–1998 El Niño. (b) The daily river discharge of the Ventura (filled circles) and Santa Clara (open circles) River. Note that data from the Santa Clara River is only available through September 2004. (c, d, e, g) Weekly to biweekly integrated elemental and mineral fluxes. Gaps in data reflect sediment trap clogging events. (f) The percentage of the TPP flux that is POP.

[10] Two major rivers, the Santa Clara and Ventura, empty into the eastern portion of the SBB (Figure 1) and account for approximately 90% of the terrigenous material entering the basin [Thornton, 1984]. River discharge is highly episodic with the majority occurring after brief, but intense rainstorms that usually occur during the winter (Figure 2b) [Thunell, 1998]. Although the annual supply of silica (Si), N, and P from rivers is 2–4 orders of magnitude lower than that associated with upwelling, riverine-derived nutrients may enhance primary production during wet El Niño years when runoff is high and upwelling is reduced (Figure 2c) [Warrick et al., 2005]. Warrick et al. [2005] further found that the molar ratios of dissolved river (Si:N:P = 16:5:1) and upwelling (Si:N:P = 13:10:1) macronutrient inputs are significantly different. Thus, variations in nutrient concentrations and ratios as a function of source may induce different responses in SBB primary productivity and thus particle export and composition to depth [Anderson et al., 2006].

3. Methods

3.1. Sample Collection

[11] A time series sediment trapping program was initiated near the center of SBB (34°14′N, 120°02′; Figure 1) in August 1993 [Thunell, 1998]. Over the course of the time series, the trap was deployed between 500 m and 540 m in a total water depth of approximately 590 m. Sinking particles were continuously collected by an automated Mark VI sediment trap (0.5 m2 trap opening) [Honjo and Doherty, 1988] equipped with 13 sampling cups on a rotating carousel. The mooring was deployed at 6-month intervals; therefore, each trap sample represents approximately two-weeks of collection time. Occasional disruptions in the time series data set are typically due to trap clogging associated with periods of high mass flux.

[12] Prior to sediment trap deployment, fill solutions for trap cups were prepared through the addition of sodium azide and sodium borate to filtered seawater (final concentrations of 10 g L−1 and 1%, respectively) collected from surface waters of the San Pedro Harbor at the Southern California Marine Institute. Addition of the poison minimizes aerobic bacterial degradation associated with sinking particles in the sample cups [Lee et al., 1992]. After recovery, sediment trap samples were sent to the University of South Carolina and stored at 4°C in the dark until processing. The overlying trap solutions (supernatants) of the recovered trap cups were irregularly sampled (∼50 mL), archived, and stored frozen prior to sample processing. During processing, wet trap samples were split into quarters using a rotary splitter. Three wet splits were archived and stored at 4°C. The remaining split was rinsed multiple times with deionized water (potential losses due to rinsing are discussed below) and centrifuged to remove salts, dried, and used to determine total mass flux measurements by weight. The material was subsequently ground and used for further geochemical analyses.

[13] A sediment core was collected in April 2005 at the SBB trap site to investigate changes in particulate P composition between sinking particles and bottom sediments. The core was frozen and samples from the top 5 cm obtained using a wire splitter at 0.25 cm intervals. Samples were subsequently freeze-dried and ground prior to geochemical analyses. A sedimentation rate of 0.32 cm y−1 was determined based on 210Pb-dating and using a constant sedimentation rate model according to the methods described in Krishnaswami et al. [1973] and Sekula Wood [2010]. While the sediment surface appears undisturbed, only relative ages can be assigned as no specific age horizons or shorter-lived radionuclides (e.g., 137Cs) were present in the core.

3.2. Geochemical Analyses

[14] POC, particulate N (PN), biogenic silica (opal), and calcium carbonate were analyzed in all sediment trap and core materials according to the methods described in Thunell [1998] and Thunell et al. [2007]. Lithogenic or terrestrial content was determined indirectly by subtracting the weight of biogenic material (2.5 * organic matter, biogenic silica, and carbonate) from the total sample weight [Thunell, 1998]. Sediment trap and core P concentrations were determined using a five-step sequential extraction procedure (SEDEX) described by Ruttenberg [1992]. Through a series of chemical washes coupled with physical agitation, SEDEX extracts five separate fractions of P: (1) loosely bound P, (2) oxide-bound P, (3) authigenic P, (4) detrital P, and (5) organic P.

[15] Approximately 0.02 g of sediment trap (n = 211) and 0.1 g sediment core (n = 20) material were analyzed for P, and samples were run in duplicate. In several cases (n = 29), trap samples contained insufficient material for analysis due to periods of low mass flux. Standard reference materials of estuarine sediment (National Institute of Standards and Technology (NIST) #1646a) and tomato leaves (NIST #1573a) were included in each SEDEX run to monitor extraction efficiency of inorganic and organic-rich sediments, respectively. A SBB sediment trap sample (SBB 21–9) was also measured in five separate SEDEX runs as an additional assessment of internal consistency as sediment trap material may be very heterogeneous. All supernatants and extracts, except the oxide-bound P fraction, were measured on a Beckman Coulter DU 640 spectrophotometer. Oxide-bound P fractions were analyzed on a Lachat Quikchem 8000 Automated Ion Analyzer using QuikChem Method 31-115-01-4-A. Analytical reproducibility of all reference and sediment trap materials included in multiple SEDEX runs is presented in Table 1. For all sediment trap samples analyzed, 80–86% of samples in the loosely bound, oxide-bound, authigenic, and detrital P phases, and 93% of organic P samples, exhibited <15% error between duplicate analyses.

Table 1. Analytical Reproducibility of Particulate Phosphorus Concentrations in Reference Materials During Multiple Sedex Runsa
Reference MaterialLoosely Bound POxide-Bound PAuthigenic PDetrital PPOPTPPPIP
  • a

    All values are reported as mean ± standard deviation (stdv) (μmol P g−1 sed).

Estuarine Sediment (n = 22)2.0 ± 0.92.1 ± 0.51.3 ± 0.91.1 ± 0.31.6 ± 0.38.1 ± 0.76.5 ± 0.6
Tomato Leaves (n = 16)31.9 ± 3.11.2 ± 2.21.4 ± 1.41.1 ± 0.65.2 ± 0.940.8 ± 4.035.7 ± 4.1
SBB 21–9 (Trap) (n = 10)23.8 ± 2.011.6 ± 2.814.5 ± 1.51.8 ± 0.610.2 ± 0.661.8 ± 3.951.7 ± 3.4

4. Results

4.1. Sediment Trap Particulate Phosphorus Concentrations and Fluxes

[16] Sediment trap particulate P concentrations and fluxes (P concentration x total mass flux) from August 1993–December 2006 are shown in Table 2 and in Figures 2 and 3. TPP fluxes were dominated by the particulate inorganic P (PIP) pool, which accounted for 85 ± 5% of the total. Of the PIP pool, loosely bound, oxide-bound, and authigenic P each accounted for 24–28% of the TPP flux. In contrast, detrital P was by far the smallest component, accounting for only 6%. POP averaged 15% of the TPP flux.

Table 2. Measurements of Total Mass, Mineral, and Elemental Concentrations and Fluxes in Sediment Trap and Core Samples Collected in SBBa
 Concentration (μmol g−1 sed)Flux/Accumulation Rates (μmol m−2 d−1)Upwelling Flux (μmol m−2 d−1) Mean ± stdvNon-upwelling Flux (μmol m−2 d−1) Mean ± stdvp-Value
Mean ± stdvMin−MaxMean ± stdvMin−Max
  • a

    Fluxes for sediment trap material were calculated as the product of element concentration and total mass flux (g m−2 d−1) for each trap sample. Sediment trap samples were split into upwelling and non-upwelling periods as defined by the shoaling of water column isotherms (≤12°C) above 20 m depth. Significance was tested using a t-test with p-values < 0.05 highlighted in bold. Sediment accumulation rates for the sediment core were estimated using mean core particulate concentrations and a sedimentation rate of 0.32 cm y−1 according to the method described in Krishnaswami et al. [1973].

  • b

    Values are in g m−2 d−1.

  • c

    Weight %.

  • d

    Values are based on the top 5 cm of the core.

Sediment Trap
TPP65.2 ± 16.234.9–152.7148 ± 727–547145 ± 67139 ± 840.49
PIP55.9 ± 15.824.5–143126 ± 655–493122 ± 58120 ± 760.87
   Loosely bound15.8 ± 8.83.0–59.136 ± 241–14738 ± 2733 ± 210.12
   Oxide-bound18.8 ± 12.53.1–98.840 ± 271–15339 ± 2841 ± 270.63
   Authigenic17.7 ± 5.90.3–38.441 ± 300.4–26536 ± 2145 ± 350.03
   Detrital3.6 ± 1.50.6–8.39 ± 70.1–428.0 ± 6.710 ± 70.11
POP9.8 ± 2.25.1–19.722 ± 102–6024 ± 1220 ± 8<0.001
POC3787 ± 8791676–68117557 ± 414873–25,3478924 ± 45006639 ± 3726<0.001
PN452 ± 182187–21581019 ± 5669–24691259 ± 602865 ± 493<0.001
Total Mass2.1 ± 1.2b<0.1–9.1b2.2 ± 1.1b2.1 ± 1.2b0.08
Lithogenic64.3 ± 9.6c34.7–86.2c1.3 ± 0.9b<0.1–7.7b1.3 ± 0.7b1.3 ± 0.9b0.79
Opal16.8 ± 7.9c1.8–42.7c0.3 ± 0.2b<0.1–1.1b0.5 ± 0.3b0.3 ± 0.2b<0.001
Carbonate8.0 ± 2.3c3.4–16.1c0.2 ± 0.1b<0.1–0.6b0.18 ± 0.12b0.15 ± 0.09b0.02
 
Cored
TPP45.3 ± 2.540.7–51.095.3 ± 5.2    
PIP38.2 ± 2.334.3–43.480.5 ± 4.8    
   Loosely bound1.6 ± 1.60.5–8.13.3 ± 3.4    
   Oxide-bound6.4 ± 0.74.7–8.013.6 ± 1.5    
   Authigenic24.3 ± 2.4117.1–27.851.1 ± 5.1    
   Detrital5.9 ± 2.83.0–10.312.3 ± 5.9    
POP7.1 ± 0.36.3–7.714.9 ± 0.8    
POC2742 ± 1392502–29535769 ± 292    
PN278 ± 23227–310585 ± 48    
Lithogenic71.6 ± 3.4c66.3–82.7c1.5 ± 0.1b    
Opal12.6 ± 2.3c9.7–18.0c0.25 ± 0.08b    
Carbonate8.2 ± 0.3c7.6–8.7c0.17 ± 0.07b    
Figure 3.

Seasonal changes in SBB inorganic phosphorus fractions August 1993–December 2006. The top of each panel is the percentage of the TPP flux that is composed of the individual inorganic P fraction, while the bottom of each panel depicts the sinking particle flux of the specific inorganic P pool.

[17] It is important to note that all of these fluxes should be viewed as minimum estimates. One potential difficulty in using sediment traps to determine P fluxes is that fluxes may be underestimated due to solubilization of particles in the trap solution (supernatant) during trap deployment as well as post trap recovery handling/processing [Antia, 2005; O'Neill et al., 2005]. In SBB, supernatant total dissolved P concentrations ranged from <1 to 49% and averaged 12 ± 10% of the total amount of P within the trap cup (dissolved and particulate pools combined). Most of this P loss, 75 ± 20%, further occurred as soluble reactive P (a term often used interchangeably with dissolved inorganic P). In other words, loss of P to the dissolved phase does not reflect the composition of the particulate P phases, with organic P losses disproportionately higher than inorganic losses. Corrections to account for solubilization are limited by the availability of trap cup supernatants (∼65% of all sediment trap samples recovered). Thus, particulate P measurements presented in this study are not corrected for loss to the dissolved phase.

[18] Absolute P concentrations varied widely throughout the time series, with P fluxes varying by at least a factor of 20 for all P phases. To elucidate potential drivers of this P flux variability, we examined the role of both seasonal upwelling and riverine discharge. Seasonal upwelling was defined by the shoaling of isotherms (≤12°C) above 20 m depth, following previous studies in the region (Figure 2a) [Thunell, 1998]. Daily riverine discharge rates (ft3 s−1) for the Santa Clara and Ventura Rivers were acquired from USGS (http://ca.water.usgs.gov/) monitoring gauges 11118500 and 11114000, respectively (Figure 2b). Please note that while the Santa Clara River is a known source of nutrients to the Santa Barbara Channel [Warrick et al., 2005], river discharge data is only available through September 2004.

[19] TPP, and hence PIP, closely followed total mass flux (Figures 2c and 2d, R2 = 0.74, regression not shown), a combination of lithogenic, biomineral, and organic phases. The influence of river discharge, a major source of lithogenic material, on mass, TPP, and PIP fluxes is clearly evident during the large discharge events that occurred in both the Santa Clara and Ventura Rivers in late January and early February of 1998. In contrast, while POP fluxes also mirrored mass flux, the influence of riverine discharge was not as pronounced. Rather, POP fluxes more closely reflected variations in opal flux (Figures 2e2g).

[20] The SBB undergoes strong seasonal upwelling that increases diatom abundance in surface waters, which increases the export of both POC and opal to depth [Thunell, 1998]. In order to examine the influence of this hydrographic process on P fluxes, we divided the data set into upwelling and non-upwelling periods using the criteria defined earlier. POP, %POP, POC, PN, and opal fluxes were all significantly higher during upwelling versus non-upwelling periods (p < 0.001, Table 2). Carbonate fluxes were also higher during upwelling versus non-upwelling with smaller differences in flux reflecting both biotic and abiotic sources [Thunell, 1998]. Closer inspection of POP fluxes showed a strong correlation with POC irrespective of season (Figure 4a). In contrast, POP fluxes were more strongly correlated with opal during upwelling (R2 = 0.52 versus 0.34) and with carbonate during non-upwelling (R2 = 0.62 versus 0.39), although carbonate:POP ratios between the two periods were indistinguishable (Figures 4b and 4c).

Figure 4.

POP fluxes versus (a) POC, (b) opal, and (c) carbonate. In each panel, upwelling and non-upwelling periods are depicted by filled and open circles, respectively. Best fit lines for each period are shown only if the slopes are significantly different. Otherwise, all data are fit with a single regression.

[21] Lithogenic and total PIP fluxes showed no significant seasonal differences. In fact, authigenic P fluxes were significantly higher during non-upwelling periods (Table 2). Closer examination of specific PIP phases further shows strong correlations between lithogenic and oxide-bound, authigenic, and detrital P phases irrespective of season (Figure 5). Loosely bound P fluxes showed weaker correlations with lithogenic material (R2 < 0.4, data not shown).

Figure 5.

(a) Oxide-bound, (b) authigenic and (c) detrital P fluxes versus lithogenic flux. In each panel, upwelling and non-upwelling periods are depicted by filled and open circles, respectively. All data are fit with a single regression line as there is no significant difference in slopes between upwelling and non-upwelling periods.

[22] The molar ratios of POC and PN to TPP, PIP, and POP were determined for the sediment trap time series using linear fits to all available data. Comparisons of POC to TPP and PIP, revealed molar ratios of 35 ± 3 (R2 = 0.44) and 35 ± 3 (R2 = 0.37), respectively. POC was best correlated to POP with a molar ratio of 337 ± 18 (R2 = 0.78, Figure 4a), which is significantly higher than the typical Redfield ratio of 106:1 and suggests preferential remineralization of POP within the water column [Redfield, 1958]. Similar to POC, TPP and PIP were poorly correlated with PN. The molar POC: PN ratio averaged 6.7 ± 0.2 (R2 = 0.84) and the PN: POP ratio was 43 ± 1.9 (R2 = 0.69). No clear seasonal or interannual trends were observed in these ratios.

4.2. Phosphorus Composition of Core Sediments

[23] TPP concentrations ranged from 40.7–51.0 μmol P g−1 sed, and averaged 45.3 ± 2.5 mol P g−1 sed (Table 2 and Figure 6), ∼31% less than that observed in the overlying sediment trap with the % loss evenly distributed between PIP (31%) and POP (28%). Consistent with the trap results, PIP accounted for the majority (>85%) of TPP in bottom sediments. However, unlike the composition of the overlying sediment trap, mean sediment core PIP concentrations were dominated by authigenic P followed by oxide-bound, detrital, and loosely bound P.

Figure 6.

Down core concentrations of inorganic and organic phosphorus fractions.

[24] TPP concentrations increased significantly between the upper 0–2.5 cm and lower 2.5–5.0 cm of the sediment core, mainly due to a significant increase in detrital P (p < 0.005). Loosely bound P concentrations decreased rapidly over the upper 1 cm and then reached a relatively constant concentration with increasing depth in the core. Conversely, authigenic P concentrations increased by a similar amount before reaching relatively constant concentrations with depth as well. Oxide-bound P and POP concentrations showed no significant change with depth. POC, PN, and opal concentrations also differed between the sediment trap and underlying sediments by 28, 39, and 26%, respectively, but varied little down core (Table 2). Sediment carbonate concentrations, however, were indistinguishable from those in the sediment trap samples and exhibited only small changes down core.

[25] Comparison between sediment trap P fluxes and P sediment accumulation rates (estimated from a sedimentation rate of 0.32 cm y−1 [Sekula Wood, 2010]), mean core P concentrations, porosity, and assuming a density of 2.4 g cm−3 [Krishnaswami et al., 1973], revealed a 35% difference (∼53 μmol P m−2 d−1) between the mean sediment trap TPP flux and that accumulating 50–90 m below in surficial bottom sediments (Table 2). Losses in both POP and PIP between the sediment trap and sediment core are consistent with those of POC and PN (Table 2). This suggests that in the ∼100 m depth interval above the bottom sediments and at the sediment water interface, all are being remineralized at relatively consistent rates. Please note that while the sediment surface appears undisturbed, the lack of stratigraphic markers does not allow for a direct comparison between the timing of material captured during specific sediment trap periods and the sediment core. Only relative sediment core ages could be assigned.

5. Discussion

5.1. Particulate Organic Phosphorus Trends in SBB

[26] POP is composed of a wide variety of compounds that may be more or less bioavailable than inorganic P phases depending on the biological community structure and surrounding water geochemistry [Benitez-Nelson, 2000; Paytan et al., 2003]. In the SBB, sinking POP fluxes were a surprisingly small percentage of the TPP pool throughout the time series, only 15 ± 5% (Figure 2). While this is significantly lower than the >30% of TPP typically observed in other coastal regions, the absolute magnitude of the measured SBB POP fluxes is toward the high end of the range of fluxes observed in other coastal regions (Table 3). This is likely due to exceptionally high phytoplankton biomass and productivity rates within SBB, which are among the largest measured along the southern California coast [Mantayla et al., 1995].

Table 3. SEDEX P Fluxes Measured in Sediment Traps Along the Western U.S., Gulf of California, Mexico, and Cariaco Basin, Venezuela Continental Marginsa
Sediment TrapP Concentration (μmol P g−1 sed)P Flux (μmol P m−2 d−1)Total Mass Flux (g m−2 d−1)
TPPPOPPIPLoosely Bound POxide-Bound PAuthigenic PDetrital PTPPPOPPIPLoosely Bound POxide-Bound PAuthigenic PDetrital P
  • a

    All values are reported as mean ± standard deviation (stdv.) for the duration of each respective collection period.

  • b

    Data reported in Faul et al. [2005]. Note that analyses were performed using a 4-step SEDEX procedure which combines loosely bound and iron-bound P. Values in parentheses were calculated as TPP - POP.

  • c

    Data is from Lyons et al. [2011].

  • d

    Data is from Benitez-Nelson et al. [2007].

California               
   Monterrey Bayb               
      S2HJ-FA98 1200 m36 ± 212 ± 1(24 ± 2)5 ± 113 ± < 16 ± < 13713(24)51361.1 ± 0.4
   Point Conceptionb               
      Stn. M 4050 m24 ± 19 ± 1(15 ± 1)7 ± < 16 ± < 12 ± < 152(3)11<10.2 ± 0.1
   Santa Barbara Basin               
      This Study - SBB 500 m66 ± 1610 ± 256 ± 1616 ± 919 ± 1318 ± 64 ± 2148 ± 7122 ± 10126 ± 6437 ± 2440 ± 2741 ± 308 ± 62.1 ± 1.2
                
Mexico               
   Guaymas Basinc               
      GUA 500 m39 ± 1811 ± 328 ± 189 ± 1210 ± 149 ± 52 ± 216 ± 105 ± 412 ± 83 ± 54 ± 54 ± 41 ± 10.4 ± 0.2
                
Venezuela               
   Cariaco Basind               
      CAR 130 m98 ± 5047 ± 2651 ± 39    53 ± 3130 ± 2223 ± 12    0.8 ± 0.6
      CAR 275 m82 ± 8430 ± 2353 ± 65    64 ± 4624 ± 1740 ± 36    1.0 ± 0.8
      CAR 455 m70 ± 4429 ± 1640 ± 41    41 ± 2918 ± 1422 ± 21    0.7 ± 0.6
      CAR 930 m64 ± 4728 ± 2036 ± 43    27 ± 2312 ± 1115 ± 17    0.5 ± 0.6
      CAR 1250 m55 ± 3128 ± 1427 ± 26    20 ± 1811 ± 109 ± 10    0.4 ± 0.5

[27] Within SBB, upwelling is primarily caused by strong northwesterly winds that strengthen the intensity of the nutrient-rich, southward flowing California Current. During El Niño the California Current weakens, reducing upwelling and decreasing primary production [Chavez et al., 2002]. During the upwelling season, primary production within SBB is dominated by diatoms [Anderson et al., 2008], which contributes to the significant increase in opal fluxes measured in the sediment traps at depth (p < 0.001). Haptophyte abundance also increases during upwelling, but to a lesser extent [Anderson et al., 2008], leading to a weaker increase in carbonate flux (p = 0.02). Both POP fluxes and %POP increased significantly during the upwelling season (Table 2 and Figure 2), and there is a very strong correlation between POC and POP (r2 = 0.78, p < 0.001, Figure 4), confirming the biological source of sinking POP.

[28] In previous studies, the flux of opal, along with carbonate and lithogenic material, has been shown to be an important vehicle for the vertical export of organic carbon from surface waters, i.e., the ballasting effect, where loading of organic aggregates with ballasting minerals increases the sinking rate and export efficiency of POC through the mesopelagic zone [Armstrong et al., 2002; Klaas and Archer, 2002]. Thunell et al. [2007] investigated the impact of ballasting minerals on POC fluxes in sinking particles collected in SBB and found that POC was well-correlated to carbonate (R2 = 0.70) and opal (R2 = 0.61) and to a lesser extent, lithogenic material (R2 = 0.48). Given the strong relationships between POP and POC, carbonate, and opal, we suggest that mineral fluxes have a similar ballasting effect on POP (Figure 4). The carrying capacity of carbonate for POP exceeds that of opal by more than 60% (Figure 4), but this difference is not enough to overcome the significantly higher contributions of opal to the total mass flux during the upwelling season. Thus, while opal dominates the POP flux signal, the high carrying capacity of carbonate for POP and the strong relationship between carbonate and POP fluxes throughout the year, indicate that carbonate also plays a major role in the annual transport of POP to depth.

[29] High molar ratios of POC:POP (337 ± 18, Figure 4) in SBB suggest rapid and preferential remineralization of POP within the water column relative to traditional Redfield values of 106:1 for upper ocean particulate matter [Redfield, 1958]. Similarly high POC:POP ratios have also been observed in other coastal sediment trapping studies along the North Pacific coast, the Gulf of California, and Cariaco Basin, Venezuela [Faul et al., 2005; Benitez-Nelson et al., 2007; Lyons et al., 2011]. The breakdown of P containing organic matter in oxic waters occurs as a result of enzymatic hydrolysis of specific compounds, such as phosphoesters [Paytan et al., 2003]. Given the rapid transport of opal and associated organic matter to depth, on the order of a week at most in SBB [Sekula-Wood et al., 2009], preferential remineralization of P must be occurring on shorter timescales. It is important to note that although there is little evidence for higher than expected initial POC:POP ratios of biologically produced sinking organic matter in SBB, it cannot be dismissed [Sterner and Elser, 2002].

[30] No significant seasonal trends in POC:POP (or PN:POP) were observed during upwelling, non-upwelling, or periods of high riverine discharge. While seasonally invariant elemental ratios have been observed in the Cariaco Basin, it is in direct contrast to those measured in the Guaymas Basin in the Gulf of California. In Guaymas Basin, POC:POP ratios >300 were associated with low opal fluxes (<0.2 g m−2 d−1) and hypothesized to result from slower sinking speeds of the particles [Lyons et al., 2011]. In SBB and Cariaco Basin, opal and carbonate fluxes are a factor of two to four higher than in Guaymas Basin [Thunell et al., 2007]. Thus, it appears that the abundance of mineral phases in these two systems results in more efficient transport of material from the surface to depth throughout the year.

[31] Preferential remineralization of POP relative to POC ceases below 500 m and into the underlying sediments, as indicated by the lack of observable change in sediment core POC:POP ratios. Previous studies have suggested that low oxygen bottom waters and sediments enhance the release of POP into the dissolved phase relative to POC and PN [Van Cappellen and Ingall, 1994; Wallmann, 2003; Ingall et al., 2005]. However, this remains controversial as other studies have shown no such bottom water effects [Ruttenberg and Berner, 1993; Filippelli, 1997]. High sedimentation rates and an already high POC:POP ratio may have obfuscated impacts associated with anoxic sediments on POC:POP ratios in this region.

5.2. Particulate Inorganic Phosphorus Trends in SBB

[32] PIP concentrations and fluxes composed the bulk of the TPP pool. In order to examine possible sources and potential responses of the PIP pool to changes in water column biogeochemistry, a sequential extraction technique was used where increasing water column lability is roughly defined by chemical reactivity. Loosely bound P represents that P component which is easily released into the dissolved phase. While considered a fraction of the PIP pool, it also likely includes very reactive forms of organic P. For example, diatom blooms form large aggregates that may enhance retention of loosely bound P within sticky flocculates [Alldredge and Gotschalk, 1989]. Oxide-bound P is P associated with metal (i.e., iron and manganese) oxide or oxyhydroxide coatings of particle surfaces. More efficient washing techniques in recent years reveal that these coatings are responsible for the majority of P that was initially attributed to foraminifera shells [Benitez-Nelson, 2000]. While this pool may also be bioavailable, it undergoes distinct redox chemistry, such that under low oxygen conditions, P may be released into the dissolved phase during oxide reduction. Authigenic P includes authigenic carbonate fluorapatite (CFAP), biogenic apatite (i.e., bones, teeth, etc.), and carbonate-associated P. Recently, Diaz et al. [2008] reported that intracellular polyphosphate granules released from diatoms during cell lysis or opal dissolution can serve as a nucleation point for CFAP. Therefore, the magnitude of CFAP detected in sinking particles may be related to the availability of polyphosphates during high opal fluxes. Detrital P is defined as the P intrinsic to igneous or metamorphic rocks [Ruttenberg, 1992] and is likely derived from a combination of riverine and aeolian sources. It is thus considered to be the least reactive of all of the P pools.

[33] Throughout the time series, loosely bound, oxide-bound and authigenic P dominated the PIP pool, each accounting for on average, 24–28% of TPP. Detrital P accounted for less than 6%. Furthermore, there was a strong relationship (p < 0.005) between oxide-bound, authigenic, and detrital P fluxes and the lithogenic flux (Figure 5). Weaker relationships between loosely bound P and mineral phases, POC, and lithogenic material support the dynamic nature of loosely bound P sources (p < 0.05).

[34] Lithogenic fluxes are generally assumed to be primarily composed of terrestrial material, including mineral phases in fine-grained sediments (e.g., clay particles) as well as humics and freshwater organics. Therefore, lithogenic fluxes contain oxide-bound P or particles that react rapidly in coastal seawater to form these associations. Authigenic P is likely derived from sedimentary rocks that surround the SBB, which include Miocene deposits of calcareous mudstones and dolomites [Minor et al., 2009]. Stream erosion and runoff (increased by high topography) through these units facilitates transport to the SBB and also helps to explain the weak relationship observed between carbonate and lithogenic material as well. The lack of relationship between authigenic P and opal flux suggests that transformation of polyphosphate into CFAP is either too slow a process to occur as particles sink within the SBB water column or that high authigenic P fluxes derived from lithogenic sources swamp the contribution from polyphosphate nucleation. The source of terrestrially derived detrital P is puzzling. As previously mentioned, the lithology of the SBB region is dominated by sedimentary rocks (primarily sandstones with some carbonate deposits). One possibility is that the sedimentary rocks are composed of igneous/metamorphic materials containing detrital P material. An alternative explanation is that older more cemented sedimentary rocks may have a lower chemical reactivity and therefore require a more intense chemical extraction than used in this study. For example, Shemesh [1990] determined that CFAP within the Monterey Formation, which partially supplies terrestrially derived material to the SBB, has been diagenetically altered over time via decarbonation reactions to form more recalcitrant fluorapatite.

[35] There was little to no seasonality observed in most of the PIP fluxes, except for authigenic P, which was significantly higher during non-upwelling versus upwelling (45.1 versus 36.3 μmol m−2 d−1, p < 0.05). However, there was high interannual variability associated with increases in regional riverine discharge, particularly during January–February 1998 (Figure 2). Precipitation in the SBB watershed is influenced to some degree by ENSO, such that during strong El Niño conditions (i.e., 1997–1998), precipitation and riverine discharge tend to increase, although this is not always the case [Warrick et al., 2005]. Thus, the magnitude of PIP fluxes is not only episodically and seasonally distinct from POP, but may also diverge on annual to decadal timescales in response to basin scale climate patterns.

[36] In recent years, several sediment trapping studies have investigated PIP fluxes in coastal regimes including the California coastline [Faul et al., 2005], Gulf of California [Lyons et al., 2011], and off the coast of Venezuela [Benitez-Nelson et al., 2004; Benitez-Nelson et al., 2007] (Table 3). SBB PIP fluxes are the largest, exceeding the next highest PIP fluxes, measured in Cariaco Basin at a similar depth (455 m), by a factor of four. Although the data is limited, much of the PIP flux difference appears to be due to the loosely bound, oxide-bound and authigenic P fractions. As all of these sites experience seasonal upwelling, the key to the dissimilarities in PIP fluxes may be related to differences in the magnitude of terrigenous (riverine) derived inputs, with the California Coast (SBB) > Cariaco Basin > Guaymas Basin. All three regions have strong correlations between PIP and lithogenic fluxes, suggesting terrigenous sources of PIP [Benitez-Nelson et al., 2007; Lyons et al., 2011].

[37] As mentioned earlier, much of the particulate P derived from riverine sources is thought to be removed rapidly in the nearshore via sedimentation [Benitez-Nelson, 2000]. Comparison of the sinking particulate pool with underlying core sediments therefore likely reflects the relative lability of the various PIP components. In SBB, mean sediment core PIP concentrations and accumulation rates were 30–40% lower than those observed in the overlying sediment trap, with much of this loss due to lower loosely bound and oxide-bound P concentrations and fluxes (Table 2). Decreased burial of loosely bound and oxide-bound P phases is consistent with reductive dissolution of P bearing minerals and preferential remineralization of highly labile organic P compounds under low oxygen environments [Van Cappellen and Ingall, 1994; Wallmann, 2003; Ingall et al., 2005; Wallmann, 2010]. The magnitude of the observed dissolution far exceeds that of the POP pool alone and highlights the potential importance of PIP as a major source of P to deep water P inventories that is decoupled from organic carbon cycling (Table 2).

[38] In contrast, authigenic and detrital P concentrations and accumulation rates in core sediments were higher than those observed in the sediment trap. For detrital P, surface sediment increases suggest an additional source of material, e.g., via lateral advection from basin slope sediments [Smoak et al., 2000]. Increases in authigenic P accumulation rates, however, are also likely due to transformation of other P containing phases, e.g., sink switching [e.g., Faul et al., 2005], as once in the sediments, a rapid decrease in loosely bound P concentrations over the upper cm is coincident with a similar increase in authigenic P (Figure 6). In other words, loosely bound P is rapidly degraded and released into dissolved P forms that are subsequently precipitated in situ as authigenic bearing P minerals. Sink switching of loosely bound P, rather than the organic P defined fraction, likely reflects a lower bioavailability of the already degraded sedimentary organic P pool. Otherwise, concentrations remained relatively uniform with depth, with the exception of a marked increase in detrital P concentrations between 2.5 and 4.4 cm of 31%, perhaps due to an increase in riverine discharge during that period.

6. Conclusions

[39] Results from the SBB demonstrate the dynamic nature of the particulate P pool in coastal ecosystems. TPP fluxes are dominated by inorganic phases, loosely bound, oxide-associated, and authigenic P, that all have significant terrestrial sources. POP fluxes, in contrast, while also significant in magnitude, account for just 15% of the SBB TPP flux. These various P components respond very differently to changes in regional and basin scale climatology as well as water column geochemistry. As such, each of these phases may be linked to the biogeochemistry of other elements, e.g., carbon, in unique ways. In SBB, POP fluxes are chemically linked to those of POC, and variations in particle flux are a function of the factors that influence biological productivity in surface waters (e.g., spring upwelling). In contrast, oxide-bound, authigenic, and detrital P phases are predominantly derived from terrestrial sources that enter SBB via riverine discharge. As such, large scale climate modes, such as El Niño tend to dampen SBB upwelling, but increase precipitation, thus leading to considerable changes in the magnitude and composition of the particulate P that is transported from coastal surface waters to the seafloor.

[40] Once in the water column, POP fluxes are closely tied to both opal and carbonate, which suggests that these mineral phases act as ballast for the rapid transport of POP to the seafloor. While POP fluxes do decrease between the depth of the sediment trap at ∼500 m and the seafloor, 90 m below, there is no evidence of preferential remineralization of POP relative to POC under anoxic conditions. Higher than canonical Redfield POC:POP ratios within sinking particles, however, does suggest that a fraction of the POP pool is more labile, at least in oxic waters, and is remineralized in the overlying water column.

[41] The large and rapid decrease in the loosely bound and oxide-bound P fluxes between the sediment trap and the seafloor suggests that a significant fraction of terrestrial derived P is not buried, but is rather rapidly remineralized into the dissolved phase. This is of particular interest as terrestrially derived oxide-bound P, and at least some of the loosely bound P, are indirectly tied to POC remineralization. Increased oxygen utilization at depth, due to organic matter remineralization, results in the reduction of metal oxides that release their associated P back into the dissolved phase. Thus, under low oxygen conditions, inorganic P phases significantly influence P inventories at depth, perhaps even more so than organic P phases depending on the magnitude of the PIP sinking flux.

[42] SBB is only one semi-enclosed region in the California coast. It is therefore difficult to expand the results of this study globally given the limited information that exists on P cycling in these systems. Nonetheless, similar semi-enclosed basins exist worldwide and results presented herein demonstrate the complexity of the sinking particulate P pool and the unique responses that individual P phases have to the surrounding environment. Not all coastal ecosystems have significant terrestrial inputs, but large drainage river systems input tremendous amounts of material into the nearshore that clearly influence the cycling of P to an extent that is not fully realized. More studies need to be completed in other regions in order to fully understand mechanisms of lability and the potential impacts of particulate P inputs globally on geologic and modern day timescales.

Acknowledgments

[43] We thank Eric Tappa for his continued contribution to the SBB sediment trapping program and the crew of the R/V Yellowfin for their role in deployments and recoveries of the sediment trap. The manuscript was improved by the thoughtful comments of one anonymous reviewer, the Associate Editor, and E. Ingall and G. Filippelli. This work was supported in part by the National Science Foundation Chemical Oceanography Program, grant OCE0850425.

Ancillary