DISCUSSION

It has been shown in several shelf environments that sediments in these systems tend to preserve a higher proportion of total organic carbon during glacial stages than during interglacial conditions (Gulf of Mexico, Jasper and Gagosian, 1990, 1993; Californian shelf, Lyle et al., 1992; Indonesian shelf, Visser et al., 2004). These elevated C_org concentrations in sediments deposited during glacial intervals may be the result of a combination of increased fluxes of terrigenous OM from continents, higher water column productivity, and increased preservation potential (i.e., reduced oxygen availability at the sediment/water interface). In addition, the higher proportion of terrigenous OM reported during glacial intervals (i.e., Jasper and Gagosian, 1990; Schubert and Stein, 1996; Visser et al., 2004) may be influenced both by higher erosional inputs from continents (Jasper and Gagosian, 1990; Visser et al., 2004) and an increased preservation potential due to the intrinsic recalcitrant nature of this material.

The use of C/N ratio and stable isotopic signatures of organic carbon (¹³C_org) are two of many possible approaches to qualitatively assess the proportional inputs of terrigenous vs. marine OM to shelf sediments (Jasper and Gagosian, 1990, 1993; Schubert and Stein, 1996; Visser et al., 2004). Terrigenous OM is relatively depleted in nitrogen, and large inputs of this material to marine sediments often result in C/N ratios >15–20 (Meyers, 1997). In addition, carbon isotope data seem to be appropriate to discriminate between input sources in environments receiving material from terrestrial plants using the C₃ pathway (–27 to –30) vs. phytoplankton (–18 to –20) or terrestrial plants using the C₄ pathway (–12 to –16) (Meyers, 1997). Hence, a combination of elemental (atomic C/N ratios) and stable isotopic signatures of sedimentary OM has been used to infer inputs of marine vs. terrigenous sources of organic matter during glacial–interglacial intermissions (Jasper and Gagosian, 1990, 1993; Goñi, 1997; Visser et al., 2004). Indeed, average C/N values in glacial shelf sediments range from 11 to close to 20, whereas interglacial sediments tend to be characterized by lower C/N values (8–11) (Jasper and Gagosian, 1990; Visser et al., 2004). In previous studies, these shifts in C/N ratios were found to be accompanied by parallel variations in carbon isotope signatures with depleted values during glacial stances (–25 to –27) vs. heavier signatures during interglacial periods (–22 to –24) (Jasper and Gagosian, 1990, 1993; Goñi, 1997). Both proxies were thus used to characterize oscillations in the terrigenous vs. marine sources of organic matter to shelf sediments with increasing C_org concentrations observed in glacial sediments probably resulting from enhanced erosion of exposed continental shelves and direct transport of recalcitrant terrigenous organic matter to bottom sediments (cf. Jasper and Gagosian, 1990, 1993; Goñi, 1997; Visser et al., 2004).

In comparison, the Peruvian margin seems to show specific differences with respect to these studies in terms of sources and inputs of OM. First, the lower OM concentrations in glacial sediments contradict the trend reported above which states that OM concentrations tend to be higher in shelf sediments during glacial periods. In the present study, the 2- to 3-fold increase in C_org and TN concentrations from the late Pleistocene to early Holocene (Fig. F2) confirms the reported decrease in OM concentrations during glacial intervals in other shallow Peruvian margin areas (Wefer et al., 1990). Wefer et al. (1990) relate this decrease in OM to a shift of upwelling cells and the oxygen minimum zone (OMZ) to deeper sites along the margin. The resulting lower water column productivity and increased oxygen availability in bottom waters of the shallow sites would have caused higher bioturbation activity and decreased the preservation potential of settling OM. In the uppermost 35 m of ODP Site 686, for example, a relationship between periods of deepening of the water column, resulting in cold, nutrient-rich water intrusions, and high OM preservation has been reported (Farrimond et al., 1990a). Hence, the observed decrease in OM in late Pleistocene section of Hole 1229E is consistent with prior work performed in the area and suggests that changes in sedimentary OM content over the recent glacial–interglacial cycles may be related to migrations of the upwelling cells and the OMZ (Wefer et al., 1990; Farrimond et al., 1990a).

In addition, the quantity of terrigenous OM in recent sediments from the Peruvian margin is exceptionally low for a continental shelf environment (Bergamaschi et al., 1997; Farrimond et al., 1990a, 1990b; Whelan et al., 1990) and suggests that very little terrigenous C_org reaches these sediments. Molecular analyses of ODP cores from the Peruvian margin have shown that OM in these sediments is predominately derived from marine primary producers (Farrimond et al., 1990a; Whelan et al., 1990). Whelan et al. (1990) suggest that the low terrigenous influx to this margin system is consistent with the arid nature of the Peru coast, which is only sparsely vegetated. In light of this information, one would thus expect to observe elemental and isotopic signatures of preserved OM to reflect an overwhelming predominance of marine sources. Paradoxically, the high C/N ratios (>12) in most sediment intervals of Hole 1229E, and some relatively light ¹³C signatures during the middle part of the Holocene (Fig. F3), suggest instead that a substantial fraction of sedimentary OM might be derived from terrigenous sources.

High C/N ratios and light carbon isotopic signatures in sedimentary OM can still be consistent, however, with a large marine source contribution. Both signatures are sensitive to diagenetic and changing environmental conditions, which may complicate source reconstructions to ancient sediments (cf. Macko and Engel, 1993). For example, in systems that are heavily dominated by marine productivity, environmental parameters such as temperature, growth rate, species, and pCO₂ variability may alter carbon isotopic signatures during synthesis (Rau et al., 1989; Fogel and Cifuentes, 1993; Macko and Engel, 1993; Johnston and Kennedy, 1998). Additionally, postdepositional alteration of stable carbon isotopic signatures has been reported in environments where selective losses of specific biomolecules (i.e., carbohydrates, lipids, and amino acids) lead to a diagenetic ¹³C enrichment or depletion in the residual organic matter (Benner et al., 1987; Macko and Engel, 1993). Both directions of fractionation have been reported in the literature, and this effect tends to be more important in sediments receiving large quantities of fresh organic matter, which undergoes substantial degradation (cf. Macko and Engel, 1993). The molar ratio of organic carbon to nitrogen has also been shown to increase due to selective diagenetic losses of nitrogen (cf. Macko and Engel, 1993). Hence, these processes (i.e., photosynthesis and diagenesis) may have an impact on the signature of organic matter ultimately preserved in Peruvian margin sediments, particularly in a system that is known to induce large remineralization rates and a strong oxygen-minimum zone within the water column (Reimers, 1982; Suess and von Huene, 1988). Therefore, one must be careful during the assessment of different inputs in relation to specific elemental and isotopic fractionation processes in this system.

Several studies have shown that source-specific biomarkers can add useful information to further constrain reconstructions of terrestrial vs. marine inputs from isotopic and elemental data (Jasper and Gagosian, 1993; Prahl et al., 1994; Goñi, 1997; Louchouarn et al., 1999; Benner et al., 2005). In Amazon Fan sediments, for example, vascular plant biomarkers (lignin- and cutin-derived molecules) have helped confirm the importance and origin of terrigenous inputs to these sediments (Goñi, 1997; Kastner and Goñi, 2003). We have thus used a similar approach to test for potential fluctuations in terrigenous organic matter inputs to sediments of Hole 1229E over the Pleistocene–Holocene transition. The lignin-derived concentrations and carbon-normalized yields in this core (Fig. F4) are extremely low in comparison to other coastal sediments (Louchouarn et al., 1999) but are consistent with prior work performed by Bergamaschi et al. (1997) on a surface sediment sample from the Peruvian margin (420 µg/gdw and 500 µg/100 mg C_org, respectively). These authors conclude that such low lignin values in this system are due to a substantial dilution of terrestrial organic matter by autochthonous material and coincide with low terrigenous inputs from an arid coastal zone (Whelan et al., 1990). The range of values observed in Hole 1229E (2–30 µg/gdw and 5–50 µg/100 mg C_org, respectively) confirms prior observations in the region (Farrimond et al., 1990a; Whelan et al., 1990) and suggests that the combination of very low riverine discharge to the Peruvian shelf, sparse vegetation on the Peru coast, and large fluxes of autochthonous materials are responsible for a minimal contribution of terrigenous organic matter to the Peruvian shelf. In addition, intrinsic ratios of lignin-derived materials demonstrate that chemically intact vascular plant material is only a minor constituent of terrigenous organic matter preserved in Peruvian margin sediments during most of the Holocene. A high acid/aldehyde ratio of lignin-derived vanillyl phenols ([Ad/Al]v) in bulk sediments (0.6) is indicative of strong oxidative degradation of parent plant materials (Goñi et al., 1993, 1998). Except for a few low values (0.3–0.4) particularly in late Pleistocene–early Holocene sediments, the (Ad/Al)v ratios in Hole 1229E range from 0.4 to 1.7 with an average of 0.6 ± 0.3 (Fig. F5), suggesting that organic matter in this core is moderately to highly degraded, particularly in the Holocene section of the core. These values do not vary in a consistent manner with respect to lignin concentrations and suggest that the oxidative degradation of lignin polymers is likely to have occurred on land rather than during postdepositional anaerobic diagenesis in this system. Similar values and conclusions were reported for a 150,000-yr profile from a core collected between Sulawesi and Borneo (Visser et al., 2004).

In addition, our core profile shows lignin compositional parameters such as the ratios of syringyl and cinnamyl phenols to vanillyl phenols (S/V and C/V ratios, respectively) that vary with no particular trend during the Holocene but may suggest a shift, particularly in C/V ratios, from the late Pleistocene to early Holocene (Fig. F5). On a first approach, these ratios are used to discriminate between taxonomic plant groups (gymnosperms vs. angiospersm) and tissues types (soft tissue vs. woody tissues) in environmental mixtures (Goñi and Hedges, 1992; Opsahl and Benner, 1995; Klap et al., 1999). A ratio of syringyl to vanillyl phenols appreciably greater than zero in such mixtures is usually indicative of the presence of at least some angiosperm tissue, whereas a ratio of cinnamyl to vanillyl phenols greater than zero suggests that nonwoody materials are present in the sample. Such signatures, however, also vary appreciably with respect to size fractions with fine particles in soils and sediments characterized by increased S/V and C/V signatures, as well as acid/aldehyde ratios due to sorption of highly degraded nonwoody constituents on clay particles (Hedges and Oades, 1997; Louchouarn et al., 1999; Hedges et al., 2000; Farella et al., 2001; Houel et al., 2006). The S/V and C/V ratios observed during the Holocene in Hole 1229E (Fig. F6) are relatively high and are consistent with mixed inputs of degraded angiosperm plant tissue appearing as either fine debris or colloidal material sorbed on mineral surfaces (cf. Houel et al., 2006). These ratios give no indication that substantial compositional changes in the terrestrial material reaching the Peruvian margin sediments have occurred during the Holocene. However, the shift in C/V and (Ad/Al)v ratios across the Pleistocene–Holocene transition (Figs. F5, F6) suggest that during the LGM, the sources of terrigenous OM were composed of less degraded woody materials, an observation that is in contrast to previous reports of no compositional changes in terrigenous OM inputs to shelf environments over glacial–interglacial transitions (Southeast Asia, Amazon Fan) (Kastner and Goñi, 2003; Visser et al., 2004). This latter finding is further significant in that it also contradicts the proposed hypothesis that grassland cover increased during the LGM in tropical systems (van der Hammen and Asby, 1994; Piperno, 1997; van der Kaars et al., 2000; Hope, 2001). It is true that the Peruvian margin does not drain a large tropical forest ecosystem as is the case for the Amazon Fan. However, the present molecular results point to increased erosional inputs of woody materials from the continents during the LGM rather than soft tissues, as would be expected under large-scale aridification of forest ecosystems. Finally, the high C/N values, low ¹³C_org signatures, and increased lignin concentrations during the late Pleistocene all point to a higher proportion of preserved OM of terrigenous origin in sediments of the LGM. This finding is in accord with prior work in other continental shelf systems. In contrast to the Indonesian shelf, however, the decreasing total C_org concentrations suggest that marine productivity may have decreased at this site during the LGM.

The increased proportion in lignin concentrations of some sediment intervals of the Holocene (0–50 cmbsf and 200–270 cmbsf) (Fig. F4) seem to suggest as well that an increased proportion of terrigenous organic matter is responsible for the lighter isotopic signatures observed during these periods. In particular, the peak in lignin concentrations occurs during a minimum in C_org and TN concentrations and ¹³C_org signatures suggesting that higher inputs of terrigenous OM during that time coincided with lower inputs of total OM and/or a decrease in the preservation of marine OM. In addition, the strong increase in C_org and TN concentrations and sudden drop in lignin concentrations at ~200 cmbsf (Figs. F2, F4) might suggest that the influx and/or preservation of marine OM increased substantially at this site. Despite this large shift in OM quantity and quality, however, ¹³C and C/N signatures change slowly to reflect, only at 100 cmbsf, signatures more typical of marine material (Fig. F3). Hence, because the fluctuations in terrigenous OM content cannot explain the low ¹³C_org values in the 150- to 200-cmbsf interval nor the downcore increase in C/N ratios during the Holocene (Fig. F3), we believe that other processes, in addition to source variations, need to be invoked to explain the apparent terrigenous elemental and isotopic signatures observed in the Holocene.

The slow "recovery" of isotopic values toward marine signatures could be due to a strong diagenetic fractionation in the sediments through regeneration of isotopically enriched components (peptides) leaving depleted residual organic matter in the sediments (i.e., lipids). The potential for selective regeneration of protein-rich material in an upwelling system such as this is important and is consistent with high C/N ratios in residual matter. Another potential source of light carbon to sediments is the active uptake, by photosynthetic organisms, of depleted CO₂ derived from organic matter regeneration at or close to the sediment/water interface (Fogel and Cifuentes, 1993). Such a process has been observed in coastal zones heavily influenced by remineralization of organic matter in the water column (Fogel et al., 1988) and may be a source of isotopically light CO₂ used by primary producers leading to depleted autochthonous organic matter (Fogel and Cifuentes, 1993). Although these two processes are not mutually exclusive, they suggest that the shifts in particular organic matter conditions may be the product of high remineralization activity within the water column and/or at the sediment/water interface. This hypothesis is consistent with the interstitial water anomalies in alkalinity, DIC, ammonium, and sulfide observed between 100 and 300 cmbsf at Sites 1228 and 1229 (Shipboard Scientific Party, 2003). Three possible explanations were provided in the report for this anomaly: (1) it may result from ongoing activity in a microbial hotspot at this shallow sediment depth, (2) it may be a chemical relic of past microbial activity and is now relaxing back to a diffusional steady state, or (3) it may be due to the recent establishment of an oxygen minimum at this water depth, causing the extinction of a bioirrigating benthos and a stimulation of sulfate reduction. The available data do not allow us to provide anything other than speculations with regard to which process may be responsible for the link between the observed particulate OM signatures and the observed anomaly in interstitial water chemicals. It seems, however, that these shifts in sedimentary parameters may be the product of both a change in OM sources and increased microbial activity in these systems.