Inorganic and organic sinking particulate phosphorus fluxes acrossthe oxic/anoxic water column of Cariaco Basin, Venezuela
Claudia R. Benitez-Nelson a,b,⁎, Lauren P. O'Neill Madden b, Renée M. Styles a,Robert C. Thunell a,b, Yrene Astor c
a Department of Geological Sciences, University of South Carolina, Columbia, SC 29208, USAb Marine Science Program, University of South Carolina, Columbia, SC 29208, USA
c Fundacion La Salle de Ciencias Naturales, Estacion de Investigaciones Marinas de Margarita, Isla de Margarita, Venezuela
Received 28 March 2006; received in revised form 20 December 2006; accepted 5 January 2007Available online 26 January 2007
dissolved P compounds from sinking particles and theupwelling of these products to the euphotic zone is acritical step regulating P availability and biologicalproduction in surface waters (Thomson-Bulldis andKarl, 1998; Benitez-Nelson, 2000; Karl and Bjorkman,2002; Paytan et al., 2003). Yet, little is known about theparticulate P pool with regard to its composition andspatial and temporal variability. A major topic of debate
91C.R. Benitez-Nelson et al. / Marine Chemistry 105 (2007) 90–100
is whether or not there is preferential remineralization ofsinking particulate P relative to particulate organiccarbon and nitrogen and whether this changes underoxic versus anoxic conditions (Knauer et al., 1979;Martin et al., 1987; Minster and Boulahdid, 1987;Anderson and Sarmiento, 1994; Benitez-Nelson, 2000;Karl and Bjorkman, 2002; Paytan et al., 2003). Thisdebate is further confounded by increasing evidence thatnot all of the particulate P measured in sinking particlesis organic in nature (Loh and Bauer, 2000; Benitez-Nelson et al., 2004; Faul et al., 2005). Rather, thereappears to be a significant fraction that is inorganic,associated with non-biological sources, and potentiallyless bioavailable (Paytan et al., 2003; Faul et al., 2005).
The Cariaco Basin, located along the northern marginof Venezuela, is anoxic below ∼275 m. Primaryproduction occurs mainly within the upper 20–40 mof the mixed layer and most particle production occursin the oxic zone (Thunell et al., 2000; Scranton et al.,2006). Therefore, it is an excellent place to examineparticle regeneration in both oxic and anoxic waters. Inthis study, P concentrations and organic versus inorganicP speciation in sinking particles collected from fivedifferent depths in the water column between January1996 and December 2004 are examined as part of the
Fig. 1. Bathymetry of Cariaco Basin showing the location of the sediment tramajor rivers which drain directly into Cariaco Basin are denoted by the gray
Cariaco Time Series Program (Müller-Karger et al.,2005).
2. Methods
The Cariaco Basin is a 1400-m-deep depressionapproximately 160 km long by 70 km wide located offthe central Venezuelan coast (Fig. 1). It is connected tothe Atlantic Ocean by a sill ∼100-m-deep, and twoslightly deeper channels that breech it; Canal Centinela(146-m-deep) and Canal de la Tortuge (135-m-deep).High surface production rates and restricted circulationresult in anoxic waters below ∼275 m. The depth of theoxycline varies between 250 and 320 m and isindependent of density. Rather, fluctuations in oxyclinedepth appear to be due to lateral intrusions of CaribbeanSea water that are linked to eddies along the continentalshelf (Astor et al., 2003).
Primary production in the Cariaco Basin variesseasonally and is driven by wind-induced coastalupwelling. As the Intertropical Conversion Zone(ITCZ) moves to its southern-most position, strongeasterly/northeasterly trade winds develop betweenDecember and April (Thunell et al., 1999; Müller-Karger et al., 2000; Thunell et al., 2000; Müller-Karger
p array. Dark lines within the Basin depict the 100-m isobath. The fourlines on the continent.
et al., 2001; Taylor et al., 2001; Goñi et al., 2003). Thisleads to shoaling of the nutricline, bringing nutrients tothe surface and increasing biological production in theupper 50 m (Scranton et al., 2006). During the summerand fall, the winds weaken causing cessation ofupwelling and decreased primary production (Thunellet al., 1999; Müller-Karger et al., 2000; Thunell et al.,2000; Müller-Karger et al., 2001; Taylor et al., 2001;Goñi et al., 2003). Most of the biological particle fluxappears to be driven by diatoms, with additional exportfrom coccolithophore species such as Emiliani huxleyiand fecal pellets (Thunell et al., 2000; Goñi et al., 2003).A secondary source of biologically derived particulatematter may be due to chemoautotrophic microbialproduction at the oxic/anoxic boundary at ∼275 m(Taylor et al., 2001).
Non-upwelling time periods are characterized bymaxima in rainfall (June–August) as the Atlantic ITCZmoves to its most northern position over the Venezeulancoast (Peterson and Haug, 2006). This causes anincrease in discharge rates from rivers draining SouthAmerica and it is likely that this increase also occurs inthe four local rivers that drain directly into the CariacoBasin, the Tuy, Unare, Neveri, and Manzanares, thoughdirect evidence is limited (Milliman and Syvitski, 1992).The input of terrigenous material to Cariaco Basin isgenerally enhanced during and following the regionalrainy season (Peterson and Haug, 2006).
A mooring with five sediment traps (Z, A–D) islocated in the eastern Cariaco Basin at 10°30′N and64°40′W (Fig. 1). Traps A–D have been in place sinceNovember 1995. Trap A is located in oxic waters at226±6 m. Trap B is located at 407±3 m and Trap D islocated at 1205±3 m. Trap C was located at a depth of880±2 m from Jan. 1996 to Nov. 2000, and was movedto 807±2 m in Nov. 2000. A fifth trap, Z, was added inNovember 2003 at 110 m for the first 6 months, and at150 m thereafter. All five sediment traps are cone-shaped with a 0.5 m2 opening that is covered with abaffle top to reduce turbulence. The mooring isdeployed for six-month intervals and each samplecollection cup is filled with a buffered 3.2% formalinsolution as a preservative for the accumulating organicmatter. The cups are numbered 1–13, with cup 1collecting for the two-week interval immediatelyfollowing deployment, and cup 13 collecting for the2 weeks immediately before recovery (Thunell et al.,1999; Müller-Karger et al., 2000; Thunell et al., 2000;Müller-Karger et al., 2001; Taylor et al., 2001; Goñi etal., 2003).
Sediment trap samples were sealed and refrigeratedbefore processing began, usually 1–3 weeks after
recovery. Most of the supernatant from each cup wasdiscarded, along with all obvious swimming organismsnot considered part of the particle flux. Samples werethen split into quarters using a precision rotary splitter.The quarter sample used for analysis was rinsed withdeionized water a total of three times, frozen, dried, andground (Thunell et al., 2000; Goñi et al., 2003).
Total particulate P (TPP) and particulate inorganic P(PIP) were measured using an adaptation of the Aspilamethod (Aspila et al., 1976). Particulate organic P(POP) is estimated by difference (TPP−PIP). As such,each fraction is analytically defined and the PIP fractionmay contain some acid-labile organic P-containingmolecules, such as simple sugars, whereas the POPmay contain inorganic compounds, such as pyropho-sphates (Benitez-Nelson, 2000). To check analyticalaccuracy and to monitor potential variability from run torun, a standard reference material (SRM), NIST # 1573a(Tomato leaves) was analyzed with each run. This SRMwas chosen for its similarity in P content relative to thesamples (0.216%) and because it is comprised of freshorganic material.
Particulate organic carbon (POC) and particulatenitrogen (PN) analyses were conducted according to themethods described in Thunell et al. (2000). Briefly,∼25 mg of dried, ground sample was treated with a 10%solution of phosphoric acid to remove all carbonate.Samples were rinsed with deionized water and thendried in a tin capsule prior to analysis using a PerkinElmer 2400 elemental analyzer.
3. Results
All significant (p) values were determined using atwo-tailed t-test unless otherwise noted. TPP, PIP (andtherefore POP) were measured in all of the sedimenttrap samples available from January 1996–December2004 (Table 1, n=623). Approximately 5% of thesamples were run in duplicate and SRM recovery was101±3%. TPP and PIP measurements have a standarderror of 6%. POP, found by difference, has a standarderror of 8.5%. Recently, O'Neill et al. (2005) examinedpossible diagenetic artifacts associated with sedimenttrap collections of particulate P in samples retrievedfrom Cariaco Basin. They determined that on average,∼30% of the total P within trap cups is within thesupernatants with an additional 10% of total P lostduring sediment trap processing. Most of this lostmaterial (N80%) was in the form of inorganic P. Notemporal or depth trends (e.g. oxic versus anoxicwaters) were found in the release of P to thesupernatants, implying that seasonal and depth trends
Table 1Summary of average particle fluxes and elemental ratios from 1996 to 2004
All periods (116 mg C m−2 h−1)Z a — 130 m 14 5.09 2.81 0.91 0.53 0.053 0.031 0.023 0.012 0.030 0.022 48 19 147 0.86 106 0.88A a — 226 m 171 6.35 3.91 0.79 0.49 0.064 0.046 0.040 0.036 0.024 0.017 59 15 191 0.66 56 0.44B a — 407 m 160 5.02 3.37 0.61 0.41 0.041 0.029 0.022 0.021 0.018 0.014 52 19 192 0.66 76 0.42C a — 844 m 145 3.68 2.74 0.44 0.32 0.027 0.023 0.015 0.017 0.012 0.011 51 20 205 0.66 87 0.54D a — 1205 m 131 3.27 2.38 0.39 0.28 0.020 0.018 0.009 0.010 0.011 0.010 46 18 183 0.62 107 0.63All depths 623 4.72 3.43 0.58 0.43 0.040 0.036 0.023 0.026 0.017 0.015 52 19 193 0.69 69 0.53
Average primary production rates are shown in parentheses for the time period specified.a See Methods for details on trap depths.
93C.R. Benitez-Nelson et al. / Marine Chemistry 105 (2007) 90–100
in particle phases should be maintained. Given the lackof supernatant information for many of the samplesanalyzed, and the inability to distinguish the source ofthe inorganic P to the dissolved phase (e.g. hydrolyzedfrom PIP or POP), supernatant concentrations were notadded to the measured particle concentrations or fluxesdiscussed below.
PIP comprises a significant part of the TPP pool, 52±19%. PIP concentrations decrease from an average of0.066 mmol g−1 in Z trap samples to 0.026 mmol g−1 insamples from the D trap. In contrast, POP concentra-tions decrease only within the oxic waters (0.045 mmolg−1 in Trap Z to 0.030 mmol g−1 in Trap A, pb0.01)and remain constant, ∼0.029 mmol g−1 in the samplescollected from the anoxic portion of the basin (pN0.1).These changes in trap concentration are also reflected inthe percentage of PIP within the trap samples. Forexample, with the exception of the shallow Z trap(n=14), the percentage of PIP decreases significantlybetween oxic and anoxic waters (pb0.001) and more
rapidly during upwelling. The percentage of PIP is alsolower during upwelling (48±18%) versus non-upwell-ing conditions (56±19%).
TPP, PIP, and POP fluxes vary substantially (Fig. 2,Table 1). Annual fluxes of TPP, PIP, and POP fluctuateby as much as 40% depending on the depth. Annualfluxes were divided into two periods, upwelling andnon-upwelling, where upwelling is defined by anaverage temperature over the upper 25 m of less than24.5 °C (Goñi et al., 2003). Additional upwelling timeperiods were identified based on continuity of primaryproduction. While average primary production rates (asmeasured by 14C uptake over the upper 100 m, (Müller-Karger et al., 2001) are a factor of two higher duringupwelling (163 mg C m-2 h−1) versus non-upwelling(77 mg C m−2 h−1) conditions, average POP fluxesbetween seasons are similar (not including the relativelyfew samples available from trap Z, n=14). In contrast,PIP fluxes are on average 30% higher during non-upwelling (again not including the Z trap), which
Fig. 2. TPP and POP fluxes from 1996 to 2004 in A–D traps (Z trapnot shown due to limited data). Primary production rates are integratedover the upper 100 m and gray highlighting depicts periods ofupwelling (see text for details).
corresponds to time periods of increased rainfall andriver discharge (Peterson and Haug, 2006). PIP fluxesdecrease on average by ∼78% from Trap A to D, whilePOP fluxes decrease by only ∼58%.
The molar ratio of POC and PN to TPP, PIP, and POPwas determined using a linear fit to all available data(Table 1, Fig. 3). The molar ratio of POC:POP and PN:POP across all depths and years is 193 (r2 =0.69,pb0.0001) and 24.3 (r2 =0.71, pb0.0001), respectively.In contrast, the molar ratio of POC:PIP is 68 and POCand PIP are not well-correlated (r2 =0.27). Further
analysis with depth shows a decrease in the POC:POPratio from 147 in the surface Z trap, to a constant valueof ∼190 at 226 m and below. POC:POP ratios at thesedepths also changes between upwelling and non-upwelling periods (Fig. 4), with POC:POP ratios of223 and 173, respectively (ANOVA, pb0.001). Similarrelationships between POC:PN ratios are also observed,with a POC:PN ratio of 6.4 (n=14, r2 =0.89) in thesurface Z trap, increasing to a near constant value of 8.1at 226 m and below (r2 =0.99, pb0.001). POC:PNratios, however, are only slightly elevated duringupwelling (all data=8.1, A–D traps only=8.2) versusnon-upwelling periods (all data = 7.8, A–D trapsonly=7.9; pA–Db0.01). PN:POP had no observabletrends with depth, but did have significant differences inPN:POP between upwelling (27.6) and non-upwelling(21.8, pb0.01) seasons.
In addition to sinking particulate organic matter, trapsamples were measured for CaCO3, opal, and terrestrialmaterial, where the terrestrial fraction is defined as thatportion remaining after subtraction of particulate organicmatter, CaCO3, and opal (Thunell et al. 2000). Similar toPOC, POP fluxes are significantly correlated to terres-trial fluxes (r2 =0.48, pb0.01), and the other biologicallyderived minerals of CaCO3 (r
2 =0.65, pb0.01) and opal(r2 =0.44. pb0.01), with slightly elevated terrigenous,opal, and CaCO3 to POP ratios during upwelling versusnon-upwelling periods (pb0.01).
The relationship between PIP and terrigenous,CaCO3 and opal fluxes is more complicated. Duringupwelling, overall relationships between PIP and opaland CaCO3 are insignificant (r2 b0.21), but thesignificance does increase with depth, with an r2 ofb0.1 for both CaCO3 and opal at the surface, increasingto 0.34 and 0.44 at 1250 m for CaCO3 and opal,respectively. PIP is well-correlated to terrestrial materialat almost all depths during upwelling, with an overallr2 =0.43 (pb0.001). During non-upwelling, PIP is onlywell-correlated to opal in the upper 150 m (r2 =0.79),but is significantly correlated to CaCO3 above 226 m(r2 =0.69 in Trap Z and 0.47 in Trap A) and at 1205 m(r2 =0.53). PIP is significantly correlated to terrestrialfluxes only in the two deeper traps located at ∼900 m(r2 =0.40) and 1205 m (r2 =0.48). Closer examinationof the associations between PIP and terrestrial materialduring non-upwelling periods reveals two distinctrelationships. During high terrestrial flux periods(N1.3 g m−2 d−1, also associated with high mass fluxevents of N2 g m−2 d−1), there is a significant anddifferent correlation between PIP and terrestrial fluxes(r2 =0.74) not evident throughout the rest of the year(Fig. 5).
95C.R. Benitez-Nelson et al. / Marine Chemistry 105 (2007) 90–100
4. Discussion
Cariaco Basin is a dynamic upwelling system thatresults in strong phytoplankton blooms that typicallyoccur from December to April (Thunell et al., 1999;Müller-Karger et al., 2000; Thunell et al., 2000; Müller-Karger et al., 2001; Taylor et al., 2001; Goñi et al.,2003). Although Cariaco Basin phytoplankton produc-tion currently appears to be nitrate limited (Scranton etal., 2006), a number of studies in other areas havesuggested that phytoplankton production may becomeincreasingly P stressed due to anthropogenic inputs ofexcess nitrogen and naturally induced climate fluctua-tions (Fanning, 1989; Wu et al., 2000; Karl et al., 2001).In many systems, dissolved P availability is regulated bythe upwelling of P-rich deep waters (e.g. see reviews byBenitez-Nelson 2000; Delaney 1998). Thus, the remi-neralization of sinking particles is a key parameter inunderstanding the source of nutrients to the euphoticzone. Unfortunately, there is a distinct lack of know-
Fig. 3. POC versus TPP, PIP, and POP fluxes with depth in oxic (top pane
ledge regarding the composition and magnitude ofsinking TPP.
Within the 9-year period studied, TPP, PIP, and POPfluxes in Cariaco Basin fluctuate by as much as 40% anddecrease with increasing depth. We consider thesefluxes to be minimum estimates as none of the P lostto supernatants are included in these or the followingcalculations (see above). There is no clear relationshipbetween primary production and particle flux, and POPfluxes are indistinguishable between upwelling and non-upwelling seasons (Table 1, Fig. 2). In fact, PIP fluxesactually increase by 50% during non-upwelling periods.POC fluxes are only slightly elevated (∼25%) duringupwelling versus non-upwelling seasons, even thoughprimary production rates differ by a factor of two.Rather, POC fluxes tend to covary with total mass fluxes(Thunell et al., 2000; Thunell et al., in press). Althoughterrestrially derived POC is difficult to pinpoint,particularly if from erosion of marine sedimentaryrocks (Petsch et al., 2006), there is surprisingly no
l, A–C) and anoxic (bottom panel, D–E) waters from 1996 to 2004.
Fig. 4. Elemental relationships of organic constituents in upwelling(filled squares) versus non-upwelling periods (open squares). All datafrom 1996 to 2004 is shown. A) POC versus POP, B) PN versus POP,C) POC versus PN. Linear regressions are shown for upwelling (solidlines) and non-upwelling (dashed lines) periods. Upwelling and non-upwelling slopes are significantly different at pb0.01 except for POCversus PN when Z trap data are included (e.g. when Z trap data areremoved, pb0.01).
Fig. 5. Terrigenous versus PIP fluxes from 1996 to 2004. Data withinthe dashed box is shown in the inset and contains only those pointswith terrigenous fluxes N1.3 g m−2 d−1.
evidence of terrestrial POC within the sediment traps.Almost all of the POC within the traps has classicmarine indices (Hedges et al., 1997); δ13C values rangefrom −17.6 to −22.6‰ (Woodworth et al., 2004), POC:PN molar ratios are ∼8.0 (Thunell et al., submitted forpublication), and there is a strong odd C preference ofshort chained n-alkanes (Thunell et al., 2000). Otherevidence includes the presence of only a very smallamount of lignin phenols, a tracer of terrestrial organicmatter input (Woodworth et al., 2004). Thunell et al. (in
press) conclude that Cariaco Basin organic matter fluxesare driven by association with higher density material,or “mineral ballast” that may be derived from bothmarine and terrestrial sources. This interaction enablesorganic matter, which is typically neutrally buoyant inseawater, to sink rapidly out of the upper oceanregardless of overlying primary production rates.Mineral ballast may further hamper degradation,causing preferential loss of labile material not associatedwith minerals (e.g. fecal pellets), and thereby explainlarge scale differences in the composition of sinkingorganic matter as evidenced by changes in POC:POPand POC:PN ratios within oxic waters (Fig. 3) (Hedgeset al., 2001; Armstrong et al., 2002).
There are significant decreases in the flux of materialwith depth between the A and D traps. It is interesting tonote that fluxes actually increase slightly between the Zand A traps for all constituents (Fig. 3). One possibilityis that the upper two sediment traps may have beenaffected by either partial clogging during high fluxevents (Goñi et al., 2003) or enhanced horizontal shearassociated with lateral subsurface intrusions from theCaribbean Sea (Astor et al., 2003). It may also be due tozooplankton feeding in surface waters and fecal pelletproduction at depths below the Z trap.
PIP is consistently a major component of the TPPpool (52±19%) at all depths and seasons. It is likely thatthis percentage is a minimal estimate as most of the Ploss to supernatants is inorganic in nature (O'Neill et al.,2005). This high percentage is similar to that found inseveral previous, albeit limited studies (Loh and Bauer,2000; Paytan et al., 2003; Faul et al., 2005). PIP fluxesalso decrease rapidly with depth in anoxic waterssuggesting that PIP is an important source of dissolved P
97C.R. Benitez-Nelson et al. / Marine Chemistry 105 (2007) 90–100
to marine systems. The source of this PIP, however,remains enigmatic.
Most studies that examine the sinking P poolmeasure TPP with the assumption that most of the PIPcaptured within the traps is either derived from in situcellular release or remineralization of biologicallyderived organic material (Knauer et al., 1979; Minsterand Boulahdid, 1987). Thus, it is often considered abioavailable component of the TPP pool. If PIP isderived from biogenic matter, there should be excellentrelationships between TPP, PIP and POC, as POC isderived from marine biological production (Werne et al.,2000; Goñi et al., 2003; Woodworth et al., 2004). This isnot the case (Table 1, Fig. 3). In fact the molar ratio ofPOC:TPP ranges from a low of 56 in the A trap to 107 atdepth (average=69, r2 =0.53). This is significantlylower than that expected for an upwelling systembased on canonical Redfield ratios (Redfield et al.,1963). In contrast, the ratio of POC:POP ranges from147 in the Z trap to a maximum of 205 in the C Trap(average=193, r2 =0.69). Hence, organic matter ratiosin surface waters are closer to Redfield ratios.Furthermore, the rapid increase in the POC:POP ratiowith depth (relative to POC:TPP) suggests rapid andpreferential remineralization of POP relative to PIP inoxic waters (% TPP that is IP increases by ∼11% fromthe Z to A traps). In anoxic waters, POP, POC, and PNare degraded at the same rate (constant ratios withdepth), whereas PIP continues to decrease significantly.
Additional evidence that the sinking POP pool ispreferentially remineralized in oxic waters stems fromthe contemporaneously collected monthly nutrient data.Scranton et al. (2006) compiled dissolved nutrientconcentrations and determined that the dissolved N:Pratio within the upper 100 m rarely falls below 14:1, butdecreases immediately below this depth. This decreaseoccurs simultaneously with an increase in POC:POP andPN:POC ratios, whereas the POC:PIP and PN:PIP ratiosactually decrease.
Although organic matter fluxes do not vary withseason, the ratio of POC:POP in the A–D traps does(Fig. 4), with POC:POP ratios significantly higherduring upwelling (∼223) versus non-upwelling (∼173,pb0.001). Similar patterns between upwelling and non-upwelling are observed with PN:POP ratios (27.6 versus21.8, pb0.001) and to a much lesser extent, POC:PN(8.1 versus 7.8). Given that the ratios of POC:POP in theupper 100-m trap are within error, the same betweenseasons, this implies more rapid remineralization ofPOP relative to POC in the underlying oxic watersduring upwelling periods. The cause of this more rapidremineralization during upwelling is unclear, but may be
due to: 1) higher bacterial concentrations whichfacilitate this remineralization, or 2) differences in thebioavailability of the sinking organic matter (e.g. ahigher fraction of more labile material during upwell-ing). Relationships between POP, POC, and PN arediscussed in more detail in Thunell et al. (submitted forpublication).
PIP fluxes are higher during non-upwelling periods(when primary production is relatively low and rainfallhigh, Table 1). This higher flux and the weakrelationship between PIP and POC suggest that a largefraction of the PIP pool is not derived from marineproduced sinking organic matter (Fig. 3). Instead, thetiming suggests that the most likely source of PIP isfrom terrestrially derived material that likely entersCariaco Basin from rivers (Fig. 5). During the non-upwelling season, increased rainfall from June throughAugust drives increased riverine discharge from Junethrough November from four local rivers: Tuy, Unare,Neveri, and Manzanares (Peterson and Haug, 2006).The Tuy River has by far the highest discharge rate,contributing 12×106 tons of sediment y−1 into thewestern end of Cariaco Basin (Milliman and Syvitski,1992). The Unare, Neveri, and Manzanares Rivers,while closer in proximity to the sediment trap site,contribute only ∼0.5×106 tons of sediment y−1 (Milli-man and Syvitski, 1992). Only the Manzanares Riverdischarges directly into the Cariaco Basin, while theTuy, Unare, and Neveri discharge onto the UnarePlatform. The rivers drain the Araya Peninsula, to theeast of Cariaco Basin, and the Coastal Range of theCabo Codera to the west, which are predominantlycomprised of Mesozoic metamorphic and igneous rocks(Peterson and Haug, 2006). Mountains (Cordillera de laCosta and the Serrania del Interior) and the Maturin andGuarico Sub-basins to the south are comprised ofCretaceous and Tertiary sedimentary rocks character-ized by organic carbon (and presumably P) richlimestones, cherts, and shales (Macsotay et al., 2003;Peterson and Haug, 2006). Although information islimited, phosphate deposits have been found innorthwestern Venezuela adjacent to the western CariacoBasin in the Falcón region (Rodriguez, 1981) andfurther west in the Barinas/Apure Basins (La Luna andNavay Formations) (Macsotay et al., 2003).
Closer examination of the relationship between PIPand terrestrial material during non-upwelling suggeststhat there are two distinct relationships between PIP andterrestrial organic matter that is directly related to totalmass and terrigenous fluxes (Fig. 5). When mass fluxesexceed 2 g m−2 d−1, and hence terrigenous fluxesexceed 1.3 g m−2 d−1, PIP and terrestrial fluxes are
highly correlated (r2 =0.74). Most of these high massflux events are associated with high rainfall events andin one case, an earthquake (Thunell et al., 1999).
The composition of the terrestrially derived PIPfractions remains ambiguous, but it is likely associatedwith detrital, authigenic and oxide-associated materialscommonly found in rivers and continental marginsediments (Froelich et al., 1982; Ruttenberg and Berner,1993; Berner and Rao, 1994; Vink et al., 1997). In theCariaco Basin, much of the PIP released to the dissolvedphase occurs within anoxic waters between the A and Ctraps (226–844 m, Table 1). This coincides with a peakin dissolved iron and an increase in dissolved phosphatethat are consistent with the reduction of PIP associatedmetal oxides and possible consumption by chemoauto-trophic bacteria at the oxic/anoxic interface (Ho et al.,2004). The weak relationships that exist between PIPand carbonate and opal, are likely related to PIPadsorption via metal oxides to shell surfaces. Studiesof other South American watersheds support thehypothesis of terrestrially derived oxide-associated P.For example, Berner and Rao (1994) determined that∼30% of the P within suspended particles of theAmazon River was associated with oxides and trans-ported onto the shelf. Tropical soils are further typicallycharacterized by low P concentrations that are mobilizedwith changes in iron and aluminum biogeochemistry.For example, Chacon et al. (2005) recently providedevidence that soils located in the flood plain of theMapire River in southeastern Venezuela, seasonallyrelease P under anoxic conditions due to oxidedissolution.
5. Conclusions and implications
A major topic of debate regarding the sinkingparticulate fraction of the marine P cycle is whether ornot there is preferential remineralization of organic Prelative to organic C and N (Knauer et al., 1990). Here,POP is shown to be preferentially remineralized in oxicversus anoxic waters. Although pore water processes aremuch more complicated, our results are in direct contrastto studies which have found that in anoxic sediments, Pis preferentially released relative to C (Van Capellen andIngall, 1994; Wallman, 2003). These results alsodemonstrate that PIP is a substantial fraction of theTPP pool and appears to be largely terrestrially derived.In contrast to POP, PIP fluxes appear to decreasesignificantly in anoxic waters, likely due to dissolutionof metal oxides.
Combined, this study suggests that PIP and POPcontribute significantly to the dissolved P pool depend-
ing on the available oxygen concentrations. This findinghas important implications for understanding feedbackmechanisms of nutrient availability and carbon seques-tration on continental margins under oxic versus anoxicconditions. Oxygen is important for not only under-standing nutrient regimes in the geologic past, but alsoin the present day. For example, recent evidencesuggests that there is a decreasing trend in dissolvedoxygen concentrations in the world's oceans (Emersonet al., 2002; Joos et al., 2003). Here, evidence isprovided that it is the PIP pool (as opposed to bulk TPP)which needs to be considered under anoxic conditions.In other words, under anoxic conditions, preferentialremineralization of PIP relative to POC creates a directdecoupling of the P and C cycles, where P that isreleased into the water column may enhance primaryproduction and facilitate continued C export. This isparticularly important in highly productive continentalmargin systems like the Cariaco Basin, which often havesuboxic waters at depth and large terrestrial inputs. Suchcontinental margins are hypothesized to be responsiblefor more than 40% of global C sequestration in moderntimes (Müller-Karger et al., 2005). We must note thatthis hypothesis is dependent on the magnitude of the PIPflux to other continental margins. The lack of informa-tion regarding riverine P geochemistry and PIPcomposition and fluxes in marine systems in generalmakes it difficult to assess how Cariaco Basin differsfrom other continental margins around the world.
This study has focused on the bulk characteristics ofthe sinking organic and inorganic P pools. Recentevidence has shown that preferential remineralization ofspecific organic compounds, such as phosphonates, mayoccur under anoxic conditions (Benitez-Nelson et al.,2004) and it is likely that other organic compounds arealso affected. The composition of the analyticallydefined PIP pool is probably equally complex and alsoneeds to be studied in much more detail. Thus,additional analysis and characterization of specificcompounds within the P pool is necessary to betterunderstand dissolution and degradation processes withdepth. This information will help develop a morecomplete understanding of P cycling in anoxic systems,and the global P cycle in general.
Acknowledgements
We thank E. Tappa and M. Luc for the help withsample analyses. Estacion de InvestigacionesMarinas deFundacion La Salle provided logistical support and wethank the crew of the R/V Hermano Gines for theirassistance at sea. The manuscript was greatly improved
99C.R. Benitez-Nelson et al. / Marine Chemistry 105 (2007) 90–100
by the insightful comments of two anonymous review-ers. This work was partially supported by a grant fromthe University of South Carolina Research and Produc-tive Scholarship Fund and NSF Grants OCE-0118349,OCE-0326313, and OCE-0432616.
References
Anderson, L., Sarmiento, J., 1994. Redfield ratios of remineralizationdetermined by nutrient data-analysis. Global BiogeochemicalCycles 8 (1), 65–80.
Armstrong, R., Lee, C., Hedges, J., Honjo, S., Wakeham, S., 2002. Anew, mechanistic model for organic carbon fluxes in the oceanbased on the quantitative association of POC with ballast minerals.Deep Sea Research 49, 219–236.
Aspila, K.I., Agemian, H., Chau, A.S., 1976. A semi-automatedmethod for the determination of inorganic, organic and totalphosphate in sediments. Analyst 101, 187–197.
Astor, Y., Müller-Karger, F., Scranton, M.I., 2003. Seasonal andinterannual variation in the hydrography of the Cariaco Basin:implications for basin ventilation. Continental Shelf Research 23,125–144.
Benitez-Nelson, C.R., 2000. The biogeochemical cycling of phospho-rus in marine systems. Earth-Science Reviews 51 (1–4), 109–135.
Benitez-Nelson, C.R., O'Neill, L., Kolowith, L.C., Pellechia, P.J.,Thunell, R.C., 2004. Phosphonates and particulate organicphosphorus cycling in an anoxic marine basin. Limnology andOceanography 49 (5), 1593–1604.
Berner, R.A., Rao, J.L., 1994. Phosphorus in sediments of the AmazonRiver and estuary— implications for the global flux of phosphorusto the sea. Geochimica et Cosmochimica Acta 58 (10),2333–2339.
Chacon, N., Dezzeo, N., Munoz, B., Rodriguez, J., 2005. Implicationsof soil organic carbon and the biogeochemistry of iron andaluminum on soil phosphorus distribution in flooded forests of thelower Orinoco River, Venezuela. Biogeochemistry 73 (3),555–566.
Delaney, M.L., 1998. Phosphorus accumulation in marine sedimentsand the oceanic phosphorus cycle. Global Biogeochemical Cycles12 (4), 563–572.
Emerson, S., Stump, C., Johnson, B., Karl, D.M., 2002. In situdetermination of oxygen and nitrogen dynamics in the upperocean. Deep-Sea Research. Part 1, Oceanographic Research Papers49 (5), 941–952.
Fanning, K.A., 1989. Influence of atmospheric pollution on nutrientlimitation in the ocean. Nature 339, 460–463.
Froelich, P.N., Bender, M.L., Luedtke, N., Heath, G.R., DeVries, T.,1982. The marine phosphorus cycle. American Journal of Science282, 474–511.
Goñi, M.A., Aceves, H.L., Thunell, R.C., Tappa, E., Black, D., Astor,Y., Varela, R., Müller-Karger, F., 2003. Biogenic fluxes in theCariaco Basin: a combined study of sinking particulates andunderlying sediments. Deep-Sea Research. Part 1. OceanographicResearch Papers 50 (6), 781–807.
Hedges, J.I., Keil, R., Benner, R., 1997. What happens to terrestrialorganic matter in the ocean? Organic Geochemistry 27 (5/6),195–212.
Ho, T.-Y., Taylor, G.T., Astor, Y., Varela, R., Muller-Karger, F.,Scranton, M.I., 2004. Vertical and temporal variability of redoxzonation in the water column of the Cariaco Basin: implications fororganic carbon oxidation pathways. Marine Chemistry 86 (1–2),89–104.
Joos, F., Plattner, G., Stocker, T., Kortzinger, A., Wallace, D., 2003.Trends in marine dissolved oxygen: implications for oceancirculation changes and the carbon budget. Eos, Transactions,American Geophysical Union 84 (21), 197–204.
Karl, D.M., Bjorkman, K., 2002. Dynamics of DOP. In: Carlson, C.A.(Ed.), Biogeochemistry of Marine Dissolved Organic Matter.Academic Press, San Diego, California, pp. 249–366.
Karl, D.M., Björkman, K.M., Dore, J.E., Fujieki, L., Hebel, D.V.,Houlihan, T., Letelier, R.M., Tupas, L.M., 2001. Ecologicalnitrogen-to-phosphorus stoichiometry at station ALOHA. Deep-Sea Research. Part 2. Topical Studies in Oceanography 48 (8–9),1529–1566.
Knauer, G.A., Martin, J.H., Bruland, K.W., 1979. Fluxes of particulatecarbon, nitrogen, and phosphorus in the upper water column of thenortheast Pacific. Deep-Sea Research. Part A, OceanographicResearch Papers 26, 97–108.
Knauer, G.A., Redalje, D.G., Harrison, W.G., Karl, D.M., 1990. Newproduction at the Vertex time-series site. Deep-Sea Research. PartA, Oceanographic Research Papers 37 (7), 1121–1134.
Loh, A.N., Bauer, J.E., 2000. Distribution, partitioning, and fluxes ofdissolved and particulate organic C, N, and P in the eastern NorthPacific and Southern Oceans. Deep-Sea Research II 47,2287–2316.
Macsotay, O., Ehrlich, R.N., Peraza, T., 2003. Sedimentary structuresof the La Luna, Navay and Querecual Formations, UpperCretaceous of Venezuela. Palois 18, 334–348.
Martin, J.H., Knauer, G.A., Karl, D.M., Broenkow, W.W., 1987.Vertex — carbon cycling in the Northeast Pacific. Deep-SeaResearch. Part A, Oceanographic Research Papers 34 (2),267–285.
Milliman, J.D., Syvitski, J.P.M., 1992. Geomorphic/tectonic control ofsediment discharge to the ocean: the importance of smallmountainous rivers. Journal of Geology 100, 525–544.
Minster, J.F., Boulahdid, M., 1987. Redfield ratios along isopycnalsurfaces — a complementary study. Deep-Sea Research. Part A,Oceanographic Research Papers 34 (12), 1981–2003.
Müller-Karger, F., Varela, R., Thunell, R., Scranton, M., Bohrer, R.,Taylor, G., Capelo, J., Astor, Y., Tappa, E., Ho, T.Y., Iabichella, M.,Walsh, J.J. and Diaz, J.R., 2000. Sediment record linked to surfaceprocesses in the Cariaco Basin. Eos, Transactions, AmericanGeophysical Union, 81(45), 529,534–535.
Müller-Karger, F., Varela, R., Thunell, R., Scranton, M., Bohrer, R.,Taylor, G., Capelo, J., Astor, Y., Tappa, E., Ho, T.Y., Walsh, J.J.,2001. Annual cycle of primary production in the Cariaco Basin:response to upwelling and implications for vertical export. Journalof Geophysical Research-Oceans 106 (C3), 4527–4542.
Müller-Karger, F., Varela, R., Thunell, R.C., Luerssen, R., Hu, C.,Walsh, J., 2005. The importance of continental margins in theglobal carbon cycle. Geophysical Research Letters 32 (1), L01602,doi:10.1029/2004GL021346.
O'Neill, L.P., Benitez-Nelson, C.R., Styles, R.M., Tappa, E., Thunell,R.C., 2005. Diagenetic effects of particulate phosphorus samplescollected using formalin poisoned sediment traps. Limnology andOceanography Methods 3, 308–317.
Paytan, A., Cade-Menum, B.J., McLaughlin, K., Faul, K.L., 2003.Selective phosphorus regeneration of sinking marine particles:evidence from 31P NMR. Marine Chemistry 82 (1–2), 55–70.
Peterson, L.C., Haug, G.H., 2006. Variability in the mean latitude ofthe Atlantic Intertropical Convergence Zone as recorded byriverine input of sediments to the Cariaco Basin. Palaeoceano-graphy, Palaeoclimatology, Palaeoecology 234, 97–113.
Petsch, S., Gordon, E., Longworth, B., Schillawski, S., Raymond, P.,2006. TI: sources, age and composition of dissolved andparticulate organic matter delivered to passive-margin riversystems. Eos Transactions AGU 87 (36) Ocean Sci. Meet.Suppl., Abstract OS25G-08.
Redfield, A.C., Ketchum, B.H., Richards, F.A., 1963. The influence oforganisms on the composition of seawater. In: Hill, M.N. (Ed.),The Sea. Interscience, pp. 26–77.
Rodriguez, S., 1981. The use of soil geochemistry in outliningphosphate deposits, northern and southern Venezuela. Journal ofGeochemical Exploration 15, 481–488.
Ruttenberg, K.C., Berner, R.A., 1993. Authigenic apatite formation andburial in sediments from non-upwelling, continental-margin envi-ronments. Geochimica et Cosmochimica Acta 57 (5), 991–1007.
Scranton, M., Taylor, G., Astor, Y., Müller-Karger, F., 2006. Temporalvariability in the nutrient chemistry of the Cariaco Basin. In:Neretin, L.N. (Ed.), Past and Present Marine Water ColumnAnoxia. NATO Science Series: IV: Earth and EnvironmentalSciences, vol. 64. Springer, Netherlands, pp. 139–160.
Taylor, G.T., Iabichella, M., Ho, T.Y., Scranton, M.I., Thunell, R.C.,Müller-Karger, F., Varela, R., 2001. Chemoautotrophy in the redoxtransition zone of the Cariaco Basin: a significant midwater sourceof organic carbon production. Limnology and Oceanography 46(1), 148–163.
Thomson-Bulldis, A., Karl, D., 1998. Application of a novel methodfor phosphorus determinations in the oligotrophic North PacificOcean. Limnology and Oceanography 43 (7), 1565–1577.
Thunell, R.C., Tappa, E., Varela, R., Llano, M., Astor, Y., Müller-Karger, F., Bohrer, R., 1999. Increased marine sediment suspen-
sion and fluxes following an earthquake. Nature 398 (6724),233–236.
Thunell, R.C., Varela, R., Llano, M., Collister, J., Müller-Karger, F.,Bohrer, R., 2000. Organic carbon fluxes, degradation, andaccumulation in an anoxic basin: sediment trap results from theCariaco Basin. Limnology and Oceanography 45 (2), 300–308.
Thunell, R.C., Benitez-Nelson, C.R., Müller-Karger, F., Lorenzoni, L.,Varela, R. and Astor, Y., submitted for publication. Production,standing stock, flux, and elemental composition of particulatematter in the Cariaco Basin. Global Biogeochemical Cycles.
Thunell, R.C., Benitez-Nelson, C.R., Varela, R., Astor, Y. and Müller-Karger, F., in press. Particulate organic carbon fluxes alongupwelling-dominated continental margins: rates and mechanisms.Global Biogeochemical Cycles.
Van Capellen, P., Ingall, E.D., 1994. Benthic phosphorus regeneration,net primary production and oceanic anoxia: a model of coupledmarine biogeochemical cycles of carbon and phosphorus.Paleoceanography 9, 677–692.
Vink, S., Chambers, R., Smith, S., 1997. Distribution of phosphorusin sediments from Tomales Bay, California. Marine Geology 139(1–4), 157–179.
Wallman, K., 2003. Feedbacks between oceanic redox states andmarine produtivity: a model perspective focused on benthic Pcycling. Global Biogeochemical Cycles 17 (3).
Werne, J., Hollander, D., Lyons, T., Peterson, L., 2000. Climate-induced variations in productivity and planktonic ecosystemstructure from the Younger Dryas to Holocene in the CariacoBasin, Venezuela. Paleoceanography 15 (1), 19–29.
Woodworth, M., Goñi, M., Tappa, E., Tedesco, K., Thunell, R., Astor,Y., Varela, R., Diaz-Ramos, J., Müller-Karger, F., 2004. Ocean-ographic controls on the carbon isotopic compositions of sinkingparticles from the Cariaco Basin. Deep Sea Research. Part 1.Oceanographic Research Papers 51 (2), 1955–1974.
Wu, J.F., Sunda, W., Boyle, E.A., Karl, D.M., 2000. Phosphatedepletion in the western North Atlantic Ocean. Science 289 (5480),759–762.
