DISCUSSION

The C-N variations in the Site 1040 sediment section conceivably resulted from a combination of biogeochemical and sedimentological factors worthy of brief discussion here. Carbon and nitrogen isotope composition of marine sediment can commonly be related to the source (e.g., marine or terrestrial component) of the organic matter. The marine component is characterized by high ¹³C values near –21 and ¹⁵N values of about +8 with low C/N of <20, whereas the terrestrial component is characterized by low ¹³C values of about –27 and ¹⁵N values of about +1.8 with high C/N from 20 to >200 (Peters et al., 1978; Meyers, 1992; Minoura et al., 1997). However, primary biogenic productivity (Calvert et al., 1992; Minoura et al., 1997; Freudenthal et al., 2001; Ettwein et al., 2001; Pattan et al., 2003; Higginson et al., 2003) or diagenesis (Peters et al., 1978; Sweeney et al., 1978; Rau et al., 1987; Minoura et al., 1997; Sadofsky and Bebout, 2004) are more commonly invoked to produce the variations of C and N concentrations and isotope compositions of marine sediment sections. A variety of processes involving possible isotope fractionation in the water column and sediment section, including N fixation, degradation, nitrification/denitrification, and postburial alteration, have been examined for the effects on C and/or N isotopic compositions. In general, N fixation by phytoplankton at or near the water surface produces little N isotope discrimination; thus, the resultant N compound has ¹⁵N values close to that of atmosphere (Brandes et al., 1998). However, other processes can produce significant C and N isotope fractionation. The uptake of dissolved inorganic carbon (DIC) and dissolved inorganic nitrogen (DIN) by microbacteria results in organic matter with ¹⁵N and ¹³C values lower than values for DIC and DIN because of preferential consumption of light C and N isotopes. The magnitude of isotope discrimination depends on the nutrient supply. In a nutrient-rich environment, isotope discrimination results in ¹⁴N enrichment in the photosynthetic products, but ¹⁵N enrichment in the remaining dissolved nitrate (Voss et al., 1996). If the nutrients are limited, little isotope discrimination occurs, and the ¹⁵N values of the organic matter should be similar to values for the nitrogenous substrate (Wada and Hattori, 1978; Ostrom et al., 1997). Degradation of particulate organic matter during deposition can increase the C and N isotopic ratio of the remaining organic matter (Thunell and Kepple, 2004).

Postburial diagenetic alteration can also influence the ¹⁵N values of the sedimentary organic and inorganic N fractions, largely according to redox conditions (Gong and Hollander, 1997; Freudenthal et al., 2001; Lehmann et al., 2002). In the relatively oxidized environment near the sediment/water interface, aerobic organic matter degradation can shift residue organic matter isotopically to slightly heavier values; in anoxic sediments, fixation of ammonium between clay lattices results in a decrease in the ¹⁵N values of the inorganic and total N (Altabet and Francois, 2001; Freudenthal et al., 2001). The shifts of ¹⁵N values by these processes are generally regarded as being small (less than ±1) (Freudenthal et al., 2001). However, Lehmann et al. (2002) reported much larger ¹⁵N shifts (3) resulting from diagenetic alteration in laboratory incubation experiments. For organic C, some authors have suggested that diagenesis can fractionate C isotopes (Benner et al., 1987; Lehmann et al., 2002), whereas others showed that organic C is isotopically resistant to water-column and/or postburial diagenesis (Meyers and Eadie, 1993; Schelske and Hodell, 1995) and applied ¹³C as a tracer of paleoproductivity and atmospheric pCO₂ levels (Hollander and McKenzie, 1991; Schelske and Hodell, 1991; Fontugne and Calvert, 1992; Brenner et al., 1999).

In summary, the C-N signatures of diagenesis and changes in productivity in seafloor sediments are fairly clear. Diagenesis may in some cases cause a decrease in C/N with decoupled C-N variations of concentrations and isotope compositions (Müller, 1977), whereas productivity changes tend to produce C-N covariance in concentrations and isotope compositions at relatively constant C/N ratios (see discussion by Sadofsky and Bebout, 2004). Without significant superimposed diagenetic effects, linear relationships between C- and N-isotope compositions can in some cases be interpreted as reflecting sources of organic matter (marine vs. terrestrial; e.g., Minoura et al., 1997).

A consideration of the downhole trends in sediment N concentrations and pore water ammonium concentrations indicates that diagenesis did not play a significant role in producing the C-N variations at Site 1040. The sediments show large downhole increases in both N concentrations (from 832 to ~2000 ppm) and ¹⁵N values (from +3.5 to +6.6) within the top ~130 mbsf and then gradual downhole decreases in N concentrations (from ~2200 to ~1500 ppm) and ¹⁵N values (from about +6 to about +4). Ammonium contents of the interstitial waters are high, reflecting the high sedimentary TN concentrations in the section, and show dramatic downhole increases to ~150 mbsf, then decrease downhole to greater depth. Compared with Site 1039, where sediment C-N concentrations are considerably lower (except in the uppermost part of the section drilled at Site 1039), the concentrations in interstitial water at Site 1040 are as much as 10 times higher for ammonium and 1000 times higher for methane (Kimura, Silver, Blum, et al., 1997). The downhole increases of ammonium concentrations in the pore waters in the upper 180 m indicate increasing diagenetic N release (see Fig. F2). However, the downhole increases of sedimentary TN concentrations in the same interval indicate that any downhole loss of N during diagenesis is overwhelmed by the trend of increase in the sedimentary N reservoir apparently related to other factors.

Downhole variations in the concentrations and isotopic compositions of TOC and carbonate similarly show no obvious signature of diagenesis. Carbonate in the sediments shows no downhole trend in concentration but locally varies significantly in C- and O-isotope compositions (Fig. F4). TOC concentrations remain relatively constant throughout the sediment section (Fig. F3A, F3B), and the downhole increase in N concentrations therefore results in a downhole decrease in TOC/TN ratios, from >18 to <8, within the upper 100 m of the section (Fig. F3C). This variation in TOC/TN resembles a diagenesis signal, as seen in the decrease in TOC/TN downhole at ODP Site 1149 (see Sadofsky and Bebout, 2004; cf. Morris, Villinger, Klaus, et al., 2003); however, the lack of any correlated TOC concentration or isotope shift together with the large downhole increase in TN over this interval argues against diagenesis as the primary effect producing the downhole variation in sediment TOC/TN (Fig. F5).

We suggest that the downhole decline in TOC/TN ratios at Site 1040, including the striking difference between the compositions in the upper ~130 m and those in the wedge section below (i.e., above the décollement), largely reflects variations in the sources of the organic matter deposited in this section. At shallower levels (the less deformed Subunit P1A and the uppermost part of Subunit P1B), the upsection trend of increasing TOC/TN ratios from ~8 at 100 mbsf to >18 at the surface could represent a record of increased terrestrial (continental) organic contribution, largely in sediments interpreted as being debris flow deposits and turbidites and which contain abundant volcanic clasts presumably derived in the volcanic arc. Our N concentration and isotopic data are consistent with this hypothesis, showing upsection decrease from higher TN concentrations and ¹⁵N values (more marine signature) at ~130 mbsf to lower TN concentrations and ¹⁵N values (more terrestrial signature) at the top of the section (see Fig. F2). However, the concentrations and isotopic compositions of TOC (Fig. F3A, F3B) in the wedge sediments show no obvious variations over this same interval, perhaps because of a similarity in TOC concentrations and ¹³C values of the terrestrial organic matter (e.g., C3 plant-dominant organic matter characterized by low ¹³C values of approximately –25) and marine organic matter. Interestingly, these sediments display homogeneous major and trace element compositions similar to those of the Costa Rica andesites (Kimura, Silver, Blum, et al., 1997). It is unknown whether a significant change in the terrestrial organic component in the sediment could be achieved without producing noticeable change in the overall whole-sediment composition. The increasing terrestrial inputs with time, possibly accompanied by declining marine organic productivity resulting from changes in turbidity, could have resulted in the upsection decreases in N concentrations and ¹⁵N values with increases in TOC/TN ratios (Fig. F3). This relationship is particularly apparent for the sediments in the upper 50 m containing abundant rock fragments and turbidites but fewer bacterial remnants (Kimura, Silver, Blum, et al., 1997).

In the more deformed wedge below ~130 m (see Fig. F1), deformation and the lack of age constraints (see Morris, Villinger, Klaus, et al., 2003) prevent any detailed interpretation of downhole variations in C and N concentrations and isotopic compositions; however, this part of the section likely preserves C-N signatures of sediment sources. Overall, the lower part of the wedge sediment section at Site 1040 displays C-N compositions different from those in the upper ~130 m, showing relatively constant concentrations and isotopic compositions of both TN and TOC, with some more subtle trends (see Figs. F2, F3). TOC and TN concentrations, ~1–2 wt% and 1500–2200 ppm, respectively, are similar to those in the upper part of the section drilled at Site 1039 (1.5–2.5 wt% TOC; 1500–2500 ppm TN) (see Fig. F1), implying similar overall organic depositional conditions to those at present outboard of the trench at Site 1039 and influenced by the high productivity related to proximity to the shore (see Li and Bebout, 2005). However, the TOC and TN isotopic compositions are different, with the Site 1040 wedge sediments below ~130 mbsf having ¹³C_TOC values near –24 and ¹⁵N values near +5, and the uppermost part of the Site 1039 section having ¹³C_TOC values near –22 and ¹⁵N values near +7. This difference could reflect a larger terrestrial organic component in the wedge sediments at Site 1040 (deposited at >3 Ma) relative to the upper part of the Site 1039 section (Pleistocene to the present). On a plot of ¹³C_TOC vs. ¹⁵N (Fig. F6), the data for Site 1040 are deflected toward lower ¹³C_TOC values from the mixing line between terrestrial and marine components defined by Minoura et al. (1997) (solid mixing line on Fig. F6), perhaps a hint of diagenetic shift similar to that at Site 1149 (see field for Site 1149 data on Fig. F6; data from Sadofsky and Bebout, 2004). Alternatively, the terrestrial organic material fed into the Site 1040 section had more negative ¹³C values, relative to that deposited more recently at Site 1039, perhaps nearer –28 (see the dashed mixing line on Fig. F6).

The dissimilarity of the more deformed wedge sediment at Site 1040 below ~80 mbsf and the upper sediment section at Site 1039 does not necessarily preclude formation of the deformed wedge (see Fig. F7) by older accretion. Morris et al. (2002) suggested that the uniformly low ¹⁰Be concentrations in the Site 1040 wedge sediments could reflect their greater age (older than 3–7 Ma) relative to the sediments in the upper part of the Site 1039 section also sampled at Site 1040 below the décollement. One possibility would be that the more deformed wedge sediments sampled below ~80 mbsf at Site 1040 but above the décollement represent accretion before 3 Ma and that no significant accretion has occurred at this locality since the mid-Pliocene (3 Ma). This accretion would have been approximately contemporaneous with, or somewhat older than, deposition of the lower hemipelagic section at Site 1039, from ~120 to ~140 mbsf. The lower part of the hemipelagic section at Site 1039 is also similar to the more deformed wedge sediments at Site 1040 in its lower content of microfossils (E. Silver, pers. comm., 2005). However, it is problematic to directly compare the deformed wedge section at Site 1040 with the Pliocene section at Site 1039, as Site 1039 must have been located at a greater distance from the trench at ~3 Ma (~260 km using the modern convergence rate of 8.8 cm/yr), and the wedge sediments at Site 1040 could also be somewhat older than the Site 1039 Pliocene hemipelagic section (see Morris et al., 2002; Morris, Villinger, Klaus, et al., 2003). Interestingly, organic C and N concentrations in the more deformed wedge sediments at Site 1040 are both higher than in the Pliocene hemipelagic section at Site 1039 (0.5–1.0 wt% TOC; 500–1000 ppm TN; data from Li and Bebout, 2005), perhaps consistent with an increase in organic productivity toward the Pliocene margin. Also, the lower ¹⁵N values of the more deformed wedge sediments (from +3 to +5) relative to the ¹⁵N values in the Pliocene section at Site 1039 (from +5 to +7) could reflect an increased terrestrial organic component nearer the trench and continent.

A further implication of the Site 1040 deformed wedge representing Pliocene accretion is that this accretion could conceivably have reduced the delivery of the uppermost, more ¹⁰Be rich part of the Pliocene sediment section to depths below the Costa Rica arc (because of subduction accretion), perhaps contributing to the modern low-¹⁰Be nature of this arc segment (see Valentine et al., 1997; Morris et al., 2002). If this were the case, it would not be necessary to call upon erosion to "dilute" the ¹⁰Be signal in the modern Costa Rica arc (see Vannucchi et al., 2003, and discussion in Morris et al., 2002). The absence of a part of the hemipelagic section beneath the modern Costa Rica arc could also produce the proportionally smaller contribution of hemipelagic sediments and larger contribution of carbonate-rich sediment noted for that arc relative to the Nicaragua arc (see Morris, Villinger, Klaus, et al., 2003). However, one important remaining issue related to this consideration of possible paleoaccretion involves the chemical compositions of the sediments (e.g., low Ba concentrations) in the more deformed wedge sediments relative to the presently subducting hemipelagic sediment section as sampled at Site 1039 and below the décollement at Site 1040 (cf. Kimura, Silver, Blum, et al., 1997).

Another, seemingly more likely possibility is that the more deformed wedge sediment section at Site 1040 represents mostly older slope apron sediments also sampled at Site 1041 (8 km upslope from Site 1040), variably deformed within the wedge. The uppermost section of the toe of the wedge is known to move downslope into the trench through creep, slumping, or debris flow (see discussion by Morris, Villinger, Klaus, et al., 2003). This process, if it involved older sediment, could be difficult to distinguish from a scenario of paleoaccretion involving slope sediment, of similar source to that at Site 1041, which was deposited beyond the wedge into the paleotrench then relatively quickly offscraped. The difference between these processes could be somewhat semantic, as it is likely that any sediment deposited onto the extreme toe of the wedge in the Pliocene would also have been deposited somewhat into the trench and onto the top of the sediment section on the incoming plate, where it could then be subducted or offscraped into the deformed wedge. Kimura, Silver, Blum, et al. (1997) reported that the wedge sediments at Site 1040 are more similar in composition to sediments in the slope apron sampled at Site 1041 (e.g., in SiO₂, Fe₂O₃, TiO₂, and Ba concentrations) than to the hemipelagic sediments at Site 1039 (also see data for Site 1043 in Valentine et al., 1997). The uppermost ~80 m of the Site 1040 section consists of less deformed debris flows and turbidites with somewhat different lithology and C-N compositions than the deformed wedge below (Figs. F2, F3), but also similar in major and trace element compositions (SiO₂, Fe₂O₃, TiO₂, and Ba concentrations) to the slope apron sediments recovered at Site 1041 8 km upslope from Site 1040 (see Kimura, Silver, Blum, et al., 1997; Silver, 2000). The chemical compositions and lithologies of these sediments in the upper 80–100 m certainly appear consistent with their being less deformed parts of the slope sediment apron, perhaps ranging in age to more recent than the Pliocene (see the somewhat elevated ¹⁰Be in the uppermost part of the Site 1040 wedge section in fig. 4 of Morris et al., 2002).

We did not analyze the C and N concentrations and isotopic compositions of the Site 1041 slope apron sediment section, the upper ~220 m of which is Pliocene to Pleistocene in age (see Kimura, Silver, Blum, et al., 1997). However, TOC and TN concentrations reported by Kimura, Silver, Blum, et al. (1997) (1–1.5 wt% TOC; 1000–2000 ppm TN) for Site 1041 are similar to and only slightly lower than those we report for the deformed wedge sediments at >130 mbsf at Site 1040 (1–2 wt% TOC; 1500–2200 ppm TN) (Figs. F2, F3). The uppermost, less deformed ~100 m of the Site 1040 section has lower N concentrations (decreasing to as low as 1000 ppm) and shows an upsection trend of decreasing ¹⁵N values, relative to values in deeper parts of the wedge section (Fig. F2), that we believe could reflect a less deformed and more intact record of changing organic sources/types.

The effect of fluid flow on the C and N concentrations and isotopic compositions of sediments in and near the fault zones and décollement does not appear to be as obvious as the effects on other elements, as recorded by the pore fluid compositions. In the upper fault zone at ~180–193 mbsf at Site 1040, although sedimentary TN, TOC, and carbonate all show hints of decreasing concentrations and isotopic compositions, the magnitudes of the possible shifts are very small (Figs. F2, F3, F4). In the middle fault zone at ~230–240 mbsf, there is no notable difference in the concentrations and isotopic ratios of TN, TOC, and carbonate relative to those of sediments in the adjacent walls. From top to bottom within the décollement zone, sediments show hints of slightly increased TN, TOC, and carbonate concentrations and ¹⁵N, ¹³C, and ¹⁸O values, but no changes in carbonate ¹³C values (Figs. F2, F3, F4). These changes could be related to fluid flow. Any decreases in TN, TOC, and carbonate concentrations and isotopic compositions in the upper zone could be explained by selective removal of these components during interaction between the sediments and fluids. The décollement zone is divided into a brittle upper part and a ductile lower part, which act as conduit and barrier to fluid flow, respectively (Tobin et al., 2001). However, the variations of concentrations and isotopic compositions of N, TOC, and carbonate are small compared with the effects of productivity and diagenesis unrelated to fluid-sediment interactions, making it difficult to distinguish any possible fluid-related effects from the other heterogeneities, and there are no sharp changes in all these parameters at the boundary between the upper sediment wall and the décollement itself.

The effect of fluid on the sedimentary geochemical C pool is possibly represented in some parts of the Site 1040 section away from fault zones and the décollement. One sediment sample from ~76 mbsf contains carbonate with ¹³C of –26.1 and ¹⁸O of 36.4, which are consistent with biogenic carbonate derived from methane hydrate (von Rad et al., 1996) originating from the decomposition of organic matter at depths of 5–10 km (Hensen et al., 2004; Suess et al., 2001). Authigenic carbonates are in general characterized by extremely low ¹³C and high ¹⁸O values, which strongly depend on the environment (Canet et al., 2003; Hensen et al., 2004). As another possible fluid effect, five sediment samples from ~240 to 303 mbsf contain minor carbonate with ¹³C values ranging from –5.2 to –8.6 (Fig. F4B), considerably lower than that of normal marine sedimentary carbonate (0 ± 2) (Hoefs, 1987) and that of carbonate throughout the rest of the section. These unusually ¹³C-depleted carbonates could also link to the low-¹³C material (methane and other hydrocarbon compounds) introduced by fluids from greater depths (Kopf et al., 2000).

In our consideration of sedimentary subduction input and arc volcanic output fluxes based on study of C and N in the sediment section at Site 1039, we proposed several uncertainties that complicate efforts at mass balance (Li and Bebout, 2005). One of these is the devolatilization of C and N during very early subduction beneath the wedge. In the fault zones and the décollement, dissolved ammonium in fluids could originate either locally in sediments in the zones or at greater depth, perhaps as deep as 2–10 km (Kopf et al., 2000). Contribution of deeply-sourced ammonium could increase the TN concentrations in the fault zones and produce anomalies in the concentrations and isotopic compositions of TN of sediments or ammonium of interstitial water, as was observed for Cl, Li, B, propane, and some other cations (Silver et al., 2000; Ruppel and Kinoshita, 2000; Morris, Villinger, Klaus, et al., 2003; Kimura, Silver, Blum, et al., 1997; Kopf et al., 2000; Saffer and Screaton, 2003; Müller, 1977). However, obvious concentration or isotopic composition anomalies were not observed in TN of sediment and pore fluid ammonium (Fig. F2). Possible explanations for this lack of an obvious fluid flow signature are that (1) any deeply generated components were diluted through mixing with more locally derived pore water in the fault zones or décollement or diffusive exchange with interstitial water in sediments near fault zones, (2) concentrations of deeply generated ammonium are very low in the fluid infiltrating from greater depth, or (3) concentrations and isotopic compositions of the deeply-sourced fluid did not contrast markedly with those of the more locally derived pore fluids and thus left no signature of their passage.

The lack of an obvious difference in the ammonium concentrations of the fluid in the faults and décollement, relative to those in pore fluids away from these structures, could imply relatively minimal loss of sedimentary N during compaction and deformation at somewhat greater depths experiencing higher temperatures. This hypothesis is consistent with the experimental observation by You and Gieskes (2001) that decomposition of organic matter in the sediment is not important until temperatures exceed 350°C. Sadofsky and Bebout (2003) presented TN concentrations and ¹⁵N values for metamorphosed clastic sedimentary rocks thought to have been subducted to >30 km depths and experienced peak temperatures of <350°C beneath an active accretionary prism and metamorphosed. Sadofsky and Bebout (2003) also demonstrated that the majority of the TN is retained, although largely transferred into clays and low-grade white micas, to these depths without significant change in ¹⁵N values relative to seafloor compositions. However, these authors demonstrated the loss of a relatively high ¹⁵N component of N at very shallow levels (3–5 km, the Coastal Belt of the Franciscan complex exposed in the California Coast Ranges) and suggested that this loss at <200°C corresponded to removal and mobilization of a more loosely adsorbed component, perhaps nitrate. A study of N concentrations and isotopic compositions in Schistes Lustres and Lago di Cignana metasedimentary rocks (Italian Alps) seemingly documents retention of much of the N inventory in these sediments subducted to depths of as much as 90 km and at temperatures as high as ~625°C (Busigny et al., 2003; see discussion by Sadofsky and Bebout, 2004).

Yet another uncertainty may affect considerations of the broader MAT-to-arc mass balance of N, related to the transit time for the off-trench sediment sections to be subducted beneath the volcanic arc. Turner et al. (2000) suggested that it takes up to several million years for subducted material to be transported from the trench and erupted in arcs. A simple calculation for the MAT indicates that ~2–3 m.y. is required for subducting sediment sections in the MAT to arrive beneath the arc. However, the TOC and TN concentrations of sediments at Site 1039, presently just outboard of the MAT (see Figs. F1, F7), show large increases for the period of 2 Ma to present. One model presented by Li and Bebout (2005) attributes the increase of TOC and TN concentrations (higher productivity) to the greater availability of light and nutrients when the Site 1039 sediment section moved toward the coast with time. A second model suggested by Li and Bebout (2005) attributes the increase of TOC and TN concentrations at Site 1039 to the closure of the Central America Seaway and consequent changes of seawater circulation and climate. If the latter is the case, the sediment section presently beneath the Central America volcanic arc and contributing to arc volcanic gases could contain dramatically different and probably much lower C and N concentrations, given their age of >2–3 Ma. This would cause significant errors in the estimates of sedimentary N inputs into the Central America margin as presented by Li and Bebout (2005). However, most of the more deformed wedge sediments at Site 1040, which appear to be older than 3–4 Ma (Morris et al., 2002), contain very high TOC and TN (1500–2200 ppm N; 1–2 wt% TOC) (Table T1; Figs. F2, F3), nearly as high as concentrations in the uppermost part of the sediment section at Site 1039 (1500–2500 ppm N; 1.5–2.5 wt% TOC) (Li and Bebout, 2005; also note the repetition of the upsection increase in TOC content in the subducting section at Site 1040 beneath the décollement in Lutz et al., 2000). Therefore, it is likely that sediment sections deposited at or near the trench along the older Central America margin (>3 Ma, as represented by the deformed sediment sections sampled at Site 1040) likely experienced similar effects of increased productivity near the continent and trench, and it appears reasonable to use the Site 1039 section as an approximation of the sediment section presently beneath the Central America arc (Li and Bebout, 2005). The Pliocene-aged, lower hemipelagic section at Site 1039 (~120–140 mbsf) contains lower TN and TOC (500–1000 ppm N; 0.5–1.0 wt% TOC), consistent with its having been more distant from the trench, as much as ~260 km (see discussion in Li and Bebout, 2005) at approximately that time.