Testing the use of bulk organic δ13C, δ15N, and Corg:Ntot ratios to estimate subsidence during the 1964 great Alaska earthquake

Quaternary Science Reviews 113 (2015) 134e146

Contents lists available at ScienceDirect

Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev

Testing the use of bulk organic d13C, d15N, and Corg:Ntot ratios to estimate subsidence during the 1964 great Alaska earthquake Adrian M. Bender a, b, *, Robert C. Witter a, Matthew Rogers c a

U.S. Geological Survey, Alaska Science Center, Anchorage, AK, USA Geological Science, Western Washington University, Bellingham, WA, USA c ENRI Stable Isotope Laboratory, University of Alaska, Anchorage, Anchorage, AK, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 2 June 2014 Received in revised form 21 September 2014 Accepted 25 September 2014 Available online 3 November 2014

During the Mw 9.2 1964 great Alaska earthquake, Turnagain Arm near Girdwood, Alaska subsided 1.7 ± 0.1 m based on pre- and postearthquake leveling. The coseismic subsidence in 1964 caused equivalent sudden relative sea-level (RSL) rise that is stratigraphically preserved as mud-over-peat contacts where intertidal silt buried peaty marsh surfaces. Changes in intertidal microfossil assemblages across these contacts have been used to estimate subsidence in 1964 by applying quantitative microfossil transfer functions to reconstruct corresponding RSL rise. Here, we review the use of organic stable C and N isotope values and Corg:Ntot ratios as alternative proxies for reconstructing coseismic RSL changes, and report independent estimates of subsidence in 1964 by using d13C values from intertidal sediment to assess RSL change caused by the earthquake. We observe that surface sediment d13C values systematically decrease by ~4‰ over the ~2.5 m increase in elevation along three 60- to 100-m-long transects extending from intertidal mud flat to upland environments. We use a straightforward linear regression to quantify the relationship between modern sediment d13C values and elevation (n ¼ 84, R2 ¼ 0.56). The linear regression provides a slopeeintercept equation used to reconstruct the paleoelevation of the site before and after the earthquake based on d13C values in sandy silt above and herbaceous peat below the 1964 contact. The regression standard error (average ¼ ±0.59‰) reflects the modern isotopic variability at sites of similar surface elevation, and is equivalent to an uncertainty of ±0.4 m elevation with respect to Mean Higher High Water. To reduce potential errors in paleoelevation and subsidence estimates, we analyzed multiple sediment d13C values in nine cores on a shoreperpendicular transect at Bird Point. Our method estimates 1.3 ± 0.4 m of coseismic RSL rise across the 1964 contact by taking the arithmetic mean of the differences (n ¼ 9) between reconstructed elevations for sediment above and below the 1964 earthquake subsidence contact. This estimate compares well with independent subsidence estimates derived from post-earthquake leveling in Turnagain Arm, and from microfossil transfer functions at Girdwood (1.50 ± 0.32 m). While our results support the use of bulk organic d13C for reconstructing RSL change in southern Alaska, the variability of stable isotope values in modern and buried intertidal sediment required the analysis of multiple samples to reduce error. Published by Elsevier Ltd.

Keywords: Quaternary Sea level Stable isotopes Paleoseismology Alaska

1. Introduction Tectonic subsidence during the Mw 9.2 great Alaska earthquake in 1964 lowered Turnagain Arm near Girdwood, Alaska by 1.2e1.8 m (Fig. 1; Plafker, 1969). The subsidence resulted from

* Corresponding author. U.S. Geological Survey, Alaska Science Center, Anchorage, AK, USA. Tel.: þ1 541 601 6185. E-mail addresses: [email protected], [email protected] (A.M. Bender). http://dx.doi.org/10.1016/j.quascirev.2014.09.031 0277-3791/Published by Elsevier Ltd.

regional vertical displacement of southern Alaska induced by coseismic slip on the Alaska-Aleutian megathrust (Plafker, 1965). Regional displacements in 1964 formed a broad trough of subsidence, extending from Kodiak Island to College Fjord in Prince William Sound (Fig. 1), which caused sudden relative sea-level (RSL) rise that submerged shorelines, drowned coastal spruce forests and shifted estuary mud flats inland over peat-forming marshes and swamps. Sharp mud-over-peat stratigraphic contacts record these environmental shifts along Turnagain Arm, including sites at Girdwood and Bird Point (Fig. 2). Although the amount of coseismic

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

135

Fig. 1. Tectonic setting of southcentral Alaska above the eastern Alaska-Aleutian subduction zone. Shaded areas depict the distribution and magnitude of regional vertical coseismic displacements (blue, subsidence; red, uplift) that occurred during the 1964 Mw 9.2 great Alaska earthquake (after Plafker, 1969). Contour lines extrapolate between post-earthquake survey measurements, and approximate vertical deformation in meters.

subsidence in 1964 was measured by differencing pre- and postearthquake leveling along Turnagain Arm (~1.7 m, Small, 1966; Wood, 1966), Hamilton and Shennan (2005) used microfossil transfer functions to estimate a 1.50 ± 0.32 m rise in RSL caused by earthquake subsidence at Girdwood in 1964. However, sparse abundance, poor preservation, or the absence of microfossils at some coastal sites in Alaska may hamper the use of this method to reconstruct RSL changes related to the earthquake deformation cycle (e.g., Wilson et al., 2005a). Reconstructing the vertical component of displacement over multiple earthquake deformation cycles provides critical data that aid assessments of regional seismic and pan-Pacific tsunami hazards.

Here, as an alternative to microfossil-based methods, we explore the potential of using stable carbon and nitrogen isotope ratios in coastal sediment as sea-level indicators. Several studies have shown that d13C values and ratios of organic carbon to total nitrogen (Corg:Ntot) in contemporary intertidal and sub-tidal surface sediment vary with respect to elevation and reflect the origin of organic matter in intertidal sediment (e.g., Chmura and Aharon, 1995; Wilson et al., 2005; Lamb et al., 2006). For example, recent studies have applied stable carbon isotopes in sediment to reconstruct RSL changes in the salt marshes of New Jersey (Kemp et al., 2011) and central Oregon (Engelhart et al., 2013). These studies compare RSL estimates derived from stable carbon isotope values in

136

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

Fig. 2. (a) Turnagain Arm and field sites, southeast of Anchorage (map from Google). (b) Bird Point field site, showing location of transects AeA0 , BeB0 , and Bird Point tide gauge. (c) Panorama showing Bird Point transect AeA0 and the digital level used for field surveys. (d) Panorama of Bird Point transect BeB0 . (e) Girdwood field site showing location of transect CeC0 and Girdwood tide gauge. (f) Panorama of Girdwood transect CeC0 .

sediment to analyses of fossil foraminifera that inhabit distinct elevation zones in intertidal environments. This study uses stable isotope analyses of bulk organic sediment to estimate the amount of sudden RSL rise caused by subsidence during the 1964 earthquake at Bird Point on Turnagain Arm (Fig. 2b). We quantify the contemporary relationship between surface sediment d13C values and ground elevation with respect to the tidal frame using a straightforward linear regression. We then apply the linear relation to reconstruct RSL rise in 1964 by using sediment d13C values to estimate the paleoelevation of mud above, and peat below the stratigraphic contact marking earthquake subsidence in 1964. The results compare well with estimates of measured subsidence following the 1964 earthquake, but highlight complications in using d13C values to estimate RSL changes due to the inherent variability in sediment d13C values from one site to the next. This inherent variability subjects estimations of paleoelevation and RSL rise at a single site to potentially large error. To reduce potential error, we reconstructed RSL rise in multiple cores along a shore-perpendicular transect at Bird Point.

1.1. Using stable isotope ratios to reconstruct relative sea level change In contemporary intertidal and sub-tidal surface sediment, d13C and Corg:Ntot ratios vary with respect to elevation and reflect the origin of organic matter in the sediment in relation to the tidal frame (Wilson et al., 2005b; Lamb et al., 2006). Therefore, the relationship between stable isotope values in sediment and the tidal frame may provide useful sea-level indicators for reconstructing RSL change. Sea-level indicators (e.g., microfossils, coastal sediment, shore platforms, strandlines) that are most useful have a specific relationship between the elevation range of the indicator and a reference water leveldthe indicative meaning (van de Plassche, 1986; Shennan et al., in press). Among the first studies to distinguish sediment source using carbon isotope ratios, Emery et al. (1967) distinguished Holocene freshwater peat sequences from saltmarsh sediment at Cape Cod in the northeastern United States. Wilson et al. (2005b) have shown that d13C values increase and Corg:Ntot ratios decrease from supra-tidal, through intertidal,

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

and sub-tidal environments, despite decomposition effects. Bulk organic d13C and Corg:Ntot values have been used successfully to reconstruct relative sea-level change (e.g., Bentley et al., 2005; Wilson et al., 2005a; Zong et al., 2006; Kemp et al., 2011; Engelhart et al., 2013), demonstrating that the method is a reliable and independent approach for assessing the evolution of coastal environments and complements microfossil-based approaches (Lamb et al., 2007; Wilson and Lamb, 2012). Stable isotope compositions, usually presented as delta (d) values, measure the ratios of heavy and light isotopes in a sample, and represent deviations in parts per thousand (‰) from a known standard. Standards for reference include carbon in the international standard (Vienna)Pee Dee Belemnite [(V)PDB], and nitrogen gas in the atmosphere. For an in-depth treatment of isotope calculations, see Deines (1980). Sources of organic carbon in intertidal sediment reflect contributions from two principal sources: (1) an autochthonous component composed of accumulated in situ plant material growing in the sediment or forming organic sediment (e.g., peat), and (2) an allochthonous component composed of tidal-derived organic carbon. The tidal-derived source may reflect a combination of riverine and marine phytoplankton in addition to particulate organic matter (POM) from terrigenous environments (Wilson et al., 2005a; Lamb et al., 2007; Yu et al., 2010). For example, marine macroalgae and phytoplankton typically have higher d13C values that differentiate them from terrestrial vegetation (Chmura and Aharon, 1995) due to the source of carbon inherited during photosynthetic respiration. Higher d13C values in marine plants has been attributed to photosynthetic fixation of bicarbonate (HC0 3 ), which is enriched in the heavier 13C isotope compared to dissolved CO2 in marine water (Stephenson et al., 1984). Because the concentration of bicarbonate is pH dependent, bicarbonate is the principal inorganic carbon source in marine water, whereas dissolved CO2 is the principal inorganic carbon source in freshwater (Keeley and Sandquist, 1992), which may influence phytoplankton d13C values along the axis of an estuary. The higher d13C values in marine plankton (19 to 24‰) relative to terrestrial vegetation (~27‰) also may reflect kinetic fractionation of dissolved inorganic carbon during planktonic photosynthesis (Peterson and Fry, 1987). Wilson et al. (2005a) concluded that ground elevation in intertidal environments was the principal control on the relative contributions of d13C from overlying vegetation and tidal-derived POM in surface sediment. Gradients in suspended POM d13C along the axes of several European estuaries show distinct riverine and marine end-member values (e.g., Middelburg and Nieuwenhuize, 1998), which have been attributed to variations in the contribution of freshwater versus marine particulate organic d13C to intertidal sediment (Wilson et al., 2005a). The presence of C4 plants (e.g., Spartina spp.), typically enriched in d13C, may also affect variations of d13C in intertidal sediment (see Smith and Epstein, 1971; Chmura and Aharon, 1995). However, because intertidal environments in Alaska typically lack C4 plants, this source of carbon likely has little influence on sediment d13C values at Alaska sites. Corg:Ntot ratios can distinguish aquatic algal from vascular land plant sources of organic material in sediment and are often used to complement d13C values (Meyers, 1994; Wilson et al., 2005a; Lamb et al., 2006). Both freshwater and marine algae have Corg:Ntot ratios that range from 4 to 10, which differ markedly from ratios typically >20 measured for terrestrial vegetation (Meyers, 1994). Studies of Corg:Ntot as an indicator of sediment source in the Pearl River estuary in southern China attribute a seaward decrease in sediment Corg:Ntot values to a seaward decrease in the contribution of

137

terrestrial organic matter and a concomitant increase in the contribution of marine algae (Zong et al., 2006; Yu et al., 2010). Marine sources of nitrogen are enriched in d15N compared to terrestrial sources of nitrogen (Owens, 1987). The largest source of nitrogen is atmospheric N2, which is well mixed and has a near constant d15N value of 0‰ (Peterson and Fry, 1987, and references therein). Variations in d15N values are principally controlled by the rate of nitrogen supply that often limits plant growth. In marine environments substantial isotopic fractionation may result from nitrification and denitrification and to a lesser degree via photosynthesis by phytoplankton (Peterson and Fry, 1987). In the Tamar estuary of southwest England, Owens and Law (1989) demonstrated that d15N values in freshwater sediment (þ2.3 ± 0.9‰) were statistically lower than d15N values in estuary sediment (2.7e5.6‰). However, the observed variations reflected complex patterns of sediment transport rather than simple mixing in the estuary of marine and terrestrial nitrogen sources. Sediment d15N values in the Shelde estuary of northwestern Europe increase from 6.9 ± 1.7‰ in the upper estuary to 9.0 ± 1.1‰ in the lower estuary, reflecting a seaward decrease in the terrestrial component of nitrogen (Middelburg and Nieuwenhuize, 1998). Thornton and McManus (1994) reported distinct differences in d15N values of lake (0.2e4.0‰) and estuary (2.6e10.6‰) sediment in the watershed of the Tay estuary in Scotland, but concluded that biogenetic alteration (diagenesis) of nitrogen made d15N an unreliable proxy for the origin of organic material. Variation in sediment d15N values also may reflect post-depositional fractionation of nitrogen due to more rapid loss of 14N than 15N during decomposition in soils (Peterson and Fry, 1987). The decomposition of organic matter changes d13C, d15N and Corg:Ntot values in sediment (Peterson and Fry, 1987; Lamb et al., 2006). While decomposition of vegetation may account for differences between d13C in surface sediment and overlying vegetation (Malamud-Roam and Ingram, 2001), relative changes in Holocene sediment d13C appear to be preserved (Lamb et al., 2006). For example, despite the effect of decomposition that shifted Holocene sediment d13C compared to modern sediment values, Wilson et al. (2005a) found that the relationships between elevation, d13C and Corg:Ntot was sufficiently robust to distinguish sediment from different intertidal environments. The loss of labile organic matter during decomposition leads to a decrease in bulk sediment d13C values from 2 to 6‰ in Holocene sediment (Benner et al., 1987). However, the effects of soil decomposition on d15N may hinder efforts to relate elevation with stable nitrogen isotope ratios in intertidal environments. Relative sea-level changes inferred from Holocene sediment d13C and Corg:Ntot values compare well with changes in paleoenvironment inferred from microfossils (Bentley et al., 2005; Wilson et al., 2005a, b; Zong et al., 2006). Kemp et al. (2011) and Engelhart et al. (2013) analyzed d13C values in bulk organic sediment alongside foraminiferal assemblages to reconstruct relative sea-level change in central Oregon and New Jersey salt marshes. Kemp et al. (2011) identified the elevational range of d13C values in samples with and without agglutinated foraminifera to reconstruct relative sea level at sites along the southern Atlantic coast of New Jersey. At the Siletz River in Oregon, Engelhart et al. (2013) used d13C, total organic content and Corg:Ntot ranges in modern surface sediment to estimate coseismic subsidence during the great Cascadia subduction earthquake of A.D. 1700. Subsidence estimated from changes in d13C values across the stratigraphic contact compared well with foraminiferal estimates, which indicated 0.71 ± 0.56 m and 0.88 ± 0.39 m of subsidence in A.D. 1700, respectively.

138

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

2. Field and laboratory methods We recorded tidal variations and investigated transects across intertidal zones at three Turnagain Arm sites accessible by road: two at Bird Point and one at Girdwood. Two pocket marshes surveyed at Bird Point were easy to access, allowed surveys along short transect between the mudflat and upland, and preserved geological evidence for tectonic subsidence in 1964, including the standing dead snags of spruce trees killed by sudden RSL rise. We also reoccupied the Girdwood marsh site studied by Hamilton and Shennan (2005)dthis site contains an extensive ghost forest drowned in 1964 that is visibly rooted in the buried pre-1964 peat, exposed in a ~1 km-long, ~1.5 m-tall coastal bluff. The sites also provided suitable locations to record tidal water level variations. To record tidal variations of water levels at each site we deployed a HOBO-brand pressure transducer and compensatory barometer at Bird Point (Fig. 2b) from 10 May to 10 June 2013 (61 days), and another HOBO pressure transducer at Girdwood (Fig. 2e) from 12 June to 10 July 2013 (29 days). Used as tide gauges in Turnagain Arm, we anchored the pressure transducers to bedrock at elevations submerged by high tide that were safely accessible at each site. We leveled the sensors to a temporary benchmark repeatedly to verify vertical stability during deployment. We used HOBOware processing software to correct tide data for atmospheric pressure and water salinity. Elevation surveys used a Topcon DL-502 electronic digital level to vertical profiles along modern transects from tidal flat to freshwater peat-forming environments at each site (Fig. 2). To characterize the relationship between intertidal surface sediment geochemistry and elevation in Turnagain Arm, we sampled 1 cm3 of surface sediment at 10-cm vertical intervals along each transect (Figs. 2 and 3). We also mapped the vertical distribution of dominant vascular plants on transects at Bird Point to better characterize the extent of intertidal zones at these sites (Fig. 4). We sampled across the 1964 contact at Bird Point and Girdwood to measure changes in sediment geochemistry across the contact, and, if present, determine whether such changes reflect coseismic subsidence documented by post-earthquake surveys (Plafker, 1969). At Bird Point we collected and logged 11 gouge cores on a ~45 m-long transect nested along line BeB0 in Fig. 2b. We cored BeB0 because it offered the shortest and topographically smoothest transition from upland to tidal flat, and hence the best chance at capturing the full range of intertidal environments and sampling the isotopic compositions of pre- and post-1964 sediment. Cores along transect BeB0 revealed the 1964 subsidence contact as the first mud-over-peat contact at depths ranging from 40 to 60 cm. We collected ~1 cm3 samples of mud (sandy silt) and peaty sediment (slightly-to non-humified rooted herbaceous peat) from above and below the contact in each core. We also logged a ~1.5 m-tall coastal bluff at Girdwood that shows evidence of subsidence in 1964. At this location, Girdwood outcrop GC01 (Fig. 2e), we took 26e1 cm3 samples from the modern surface to below the 1964 earthquake contact. We sampled every 10 cm in mud (sandy silt) above the 1964 contact, and sampled every cm across a gradational contact into the buried peat (silty rooted herbaceous peat). We used acid rinse pre-treatment methods described by Kemp et al. (2010) to prepare surface, core, and outcrop sediment samples for carbon and nitrogen stable isotope analyses. Brodie et al. (2011a, 2011b) demonstrate that, in comparison with other sample pre-treatment methods for dissolving inorganic carbon, acid rinsing depletes Corg, Ntot, and accompanying stable isotopes. However, Craven et al. (2013) show that this depleting effect is unlikely to influence RSL reconstructions based on d13C, d15N, or Corg:Ntot because the loss is well within sediment isotopic and

elemental ranges encountered across intertidal elevations. All samples were dried overnight at 40  C, crushed and mixed with mortar and pestle, bathed in 100 mL 5% hydrochloric acid for 18 h to dissolve inorganic carbon, and homogenized using a “bead beater” instrument. Sub-milligram aliquots of homogenized samples were analyzed by continuous flow isotope ratio mass spectrometry at the ENRI Stable Isotope Lab at the University of Alaska, Anchorage to measure total organic carbon and nitrogen concentrations and determine values of d13C (vs. VPDB), d15N (vs. air), Corg, and Ntot. Estimates of instrumental uncertainty related to the mass spectrometer analysis are derived from long-term records of results from internal quality control standards at the ENRI Stable Isotope Laboratory. Materials with a known isotopic value and a matrix similar to the analyte, powdered peach leaf in this case, are treated as unknowns during each analytical run. The 2s estimates for the 13C and 15 N analyses (0.12‰ and 0.19‰, respectively) are derived from variance in the long-term results of these quality control standards. 3. Results 3.1. Measuring tides in a semidiurnal hypertidal estuary Tides in Turnagain Arm have a maximum range of 10 m, ranking them the sixth highest tides in the world among Earth's hypertidal estuaries (Greb and Archer, 2007; Archer, 2013). During spring flood tides a 1.2- to 1.8-m-high bore wave forms and travels up the estuary at ~16 km/h. Mean lower low water in Turnagain Arm is not suitable as a tidal datum because the elevation of most channel thalwegs is above low tide and base flow contributed by streams entering the arm attenuates the arrival of the incoming tide. As a result, tidal time series from pressure transducers deployed in Turnagain Arm record asymmetrical tidal curves that reach a constant base level at low tide (Fig. 3). The extreme range of tides prevented measurement of base level flow during low tide at Girdwood. We used mean higherhigh water (MHHW) as the reference water level because of the complicated tides in the arm and the observation that intertidal marshes predominantly form above MHHW. Temporary tide gauges deployed at Bird Point and Girdwood (Fig. 2) provided time series data with which we calculated sitespecific MHHW datums following the approach of Hamilton and Shennan (2005). Linear regressions of high tide water levels measured at Bird Point (n ¼ 126) and Girdwood (n ¼ 54) versus corresponding high tide elevations recorded at the nearby Anchorage NOAA tide station #9455920 (http://www.co-ops.nos.noaa.gov) produced the slopeeintercept equations used to calculate MHHW at both sites (Fig. 3). With these equations we used MHHW at Anchorage (8.89 m, MLLW datum) to estimate the MHHW datums relative to the tide gauge sensors at Bird Point (8.19 m ± 0.23 m regression standard error) and Girdwood (3.32 m ± 0.13 m). The highest range of tide measured at Bird Point was more than 9 m. Vertical survey measurements at Bird Point and Girdwood, including all core sites and surface samples, are referenced to the respective site's MHHW datum. 3.2. Distribution of vascular plants in Turnagain Arm salt marshes Intertidal zonation of perennial vascular plants in Alaskan salt marshes has been attributed to the effects of salinity, waterlogging and soil type (Snow and Vince, 1984). At Susitna Flats on the north side of Cook Inlet, ~80 km north of Bird Point, Snow and Vince (1984) studied the distribution of dominant vascular saltmarsh species characteristic of distinct intertidal zones. Puccinellia nutkaensis and Triglochin maritimum colonized the uppermost mudflats exhibiting the highest salt tolerance for vegetation flooded most often by the highest tides. Carex ramenskii and C. lyngbyaei

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

139

Fig. 3. (A) Simple linear regression plots of corresponding high tide water depths at Bird Point and Girdwood, and high tide elevations at NOAA Anchorage tide gauge 9455920. We used these regressions to calculate site-specific mean higher-high water at each site. (B) Tidal variations in water depth measured by temporary tide gauges at Bird Point and Girdwood water elevation at NOAA tide gauge 9455920 at Anchorage, west of Turnagain Arm, during the indicated period. High tides in these plots are regressed in Fig. 3A. See text for detailed discussion.

characterized higher sedge marshes located inland at sites that rarely experienced tidal inundation. Poa eminens inhabited levees and drift piles flooded by only extreme high tides. With the exception of T. maritimum, which grew in all zones, the salinity of soils hosting these plants suggested that the relative salt tolerance of each species controlled its seaward distribution in the marsh (Snow and Vince, 1984).

At Bird Point, we used descriptive guides of vascular plants (e.g., Hulten, 1968; Vince and Snow, 1984; Pojar and MacKinnon, 1994) to identify dominant salt marsh species and measure their vertical distributions along two surveyed profiles (Fig. 4). Along profile AeA0 , P. nutkaensis marked the lower limit of vegetation at ~e0.2 m (all elevations are reported relative to MHHW), below which extended unvegetated mudflats. The sedge marsh zone began at

140

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

Fig. 4. Vertical elevation profiles plotted with respect to local MHHW datum along three modern surface transects: (A) Bird Point AeA0 ; (B) Bird Point BeB0 ; and (C) Girdwood CeC0 . Plots under each profile show the corresponding d13C, d15N and Corg:Ntot values for surface sediment samples along each transect.

about the elevation of MHHW marked by the lower growth limit of C. ramenskii and T. maritimum. C. lingbyaei populated higher elevations of the sedge marsh with an upper limit of ~1.0 m. Locally, C. ramenskii grew down to elevations as low as 0.2 m where freshwater channels incised the mudflat. Marshes above ~1.0 m often flooded by standing water barred by low levees were characterized by Potentilla egedii, Equisetum spp., Calamagrostis spp., Rumex spp., and Cicuta douglasii. Forested uplands above ~1.2 m rarely flooded by tides included Alnus spp., Salix spp., and Picea spp. 3.3. d13C, d15N, and Corg:Ntot in surface sediment To characterize the relationship between elevation and contemporary values of d13C, d15N, total organic carbon (Corg) and total nitrogen (Ntot), we analyzed surface sediment samples

collected on survey transects from tide flat, through intertidal marsh, and into freshwater upland environments at Bird Point and Girdwood (Fig. 4). We collected 31 and 29 surface sediment samples on Bird Point transects AeA0 and BeB0 respectively, and 30 samples on transect CeC0 at Girdwood. Variation of surface sediment geochemistry with elevation is relatively consistent between sites, but shows considerable scatter from the lowest and highest intertidal elevations (Fig. 4). Values of d13C in modern sediment range from 29.36‰ at 1.27 m to 24.71‰ at 0.09 m on transect AeA0 , 28.83‰ at 1.07 m to 23.89‰ at 0.43 m on transect BeB0 , and 29.12‰ at 0.96 m to 24.27‰ at 0.83 m on transect CeC0 (MHHW vertical datum). Independent linear regressions of surface sediment d13C values and elevation yield R2 values of 0.57, 0.63, and 0.64 respectively for AeA0 , BeB0 , and CeC0 .

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

Values of d15N in modern sediment range from 1.27‰ at 0.58 m to 2.24‰ at 0.61 m on transect AeA0 , 1.52‰ at 1.59 m to 2.72‰ at 0.43 m on transect BeB0 , and 0.73‰ at 1.34 m to 3.20‰ at 0.33 m on transect CeC0 (MHHW vertical datum). Independent linear regressions of surface sediment d15N values and elevation

141

yield R2 values of 0.33, 0.59, and 0.51 respectively for AeA0 , BeB0 , and CeC0 . Ratios of Corg:Ntot in modern sediment range from 10.35 at 0.44 m to 50.43 at 0.93 m on transect AeA0 , 11.27 at 0.35 m to 26.98 at 0.97 m on transect BeB0 , and 10.92 at 0.83 m to 46.20 at

Fig. 5. Simple linear regression plots of d13C, d15N values and Corg:Ntot ratios against elevation with respect to mean higher-high water (MHHW). These regressions combine geochemical and surveyed sample elevation values from Bird Point on transects AeA0 and BeB0 , and from Girdwood transect CeC0 . We used the slopeeintercept equation for the d13C regression to calculate paleo-elevations for d13C values of sediment above and below the 1964 contact at Bird Point (Fig. 7).

142

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

1.39 m on transect CeC0 (MHHW vertical datum). Independent linear regressions of surface sediment Corg:Ntot ratios and elevation yield R2 values of 0.28, 0.56, and 0.44 respectively for AeA0 , BeB0 , and CeC0 . To quantify the modern relationship between elevation and d13C, d15N, and Corg:Ntot at these sites, we plot d13C and d15N values and Corg:Ntot ratios from each transect, and use a simple linear regression of sample geochemical values (n ¼ 84 for each parameter) against elevation (Fig. 5). Regressions of these data predict that a one meter elevation gain between tide flat and high marsh corresponds to surface sediment values of d13C that decrease by 1.58‰, d15N values that decrease by 0.96‰, and Corg:Ntot ratios that increase by 2.67. Among the three geochemical sea-level-indicator candidates that we test by this method, the d13C regression has the best fit (R2 ¼ 0.56) and smallest regression standard error (average ¼ ±0.59‰, or ±0.4 m MHHW). 3.4. Values of d13C in pre- and post-1964 sediment Linear regression shows that contemporary values of d13C decrease as elevation increases across transects of the sites we investigated. This regression also has a higher R2 value and lower regression standard error than linear regressions of d15N values and Corg:Ntot ratios (Fig. 5). Based on these calculated relationships, we hypothesized that shifts in d13C values from peat to mud across the 1964 contact should reflect the change in contemporary d13C values that we documented between tide flat (~25‰) and freshwater upland environments (~28‰). At Girdwood outcrop GC-01, values of d13C range from 26.12 to 28.02‰ in the mud above the 1964 contact, and from 26.89 to 28.57‰ in the buried peat (Fig. 6). Across the gradational 1964

contact at 60.5 cm depth, d13C values shift a modest 0.40‰, from peat (27.64‰) into organic-rich sandy silty mud (27.24). Values of d13C vary above and below the contact, including an excursion of roughly 1‰ that occurs in mud between 55.5 and 52.5 cm depth. Given the wide range of d13C values observed in pre- and postsubsidence sedimentary units at GC01 we anticipated problems with accurately reconstructing single point elevations based on isotope values across the 1964 contact at a single point. Additionally, the variability encountered in modern intertidal sediment d13C values produces a regression standard error equivalent to ±0.4 m MHHW for individual reconstructed elevations. To reduce the influence of these sources of error on reconstructed pre- and post-earthquake elevations, we increased sample density by taking 11 cores along Bird Point transect BeB0 and analyzing values of d13C in samples of muddy sediment above, and peaty sediment immediately underlying the 1964 contact in these cores (Fig. 7A). Of the three modern transects we completed, we cored BeB0 because it is the shortest and topographically smoothest transition from upland marsh to tidal flat. We reasoned that if the modern surface roughly reflects the pre-1964 surface, then a coring transect across BeB0 would offer the best chance at capturing the full range of marsh environment isotopic compositions in the pre-1964 peat. Samples from above and below the contact in cores BPP4 and BPP6 revealed nearly indistinguishable d13C valuesdwe assume that this is a result of sampling too close to either side of the contact (samples were ~1 cm apart), and do not include these values in subsequent analyses. In the remaining 9 cores, samples of mud from above the 1964 contact have d13C values that range from 26.45 to 24.88‰ (Fig. 7B). Peaty sediment from below the contact in these cores has d13C values that range from 28.60

Fig. 6. Girdwood outcrop GC-01. Schematic stratigraphy and soil descriptions correspond with outcrop photo, dashed line indicates 1964 contact, and black boxes indicate samples. Sample d13C values (black line), Corg:Ntot ratios (red line), and corresponding paleoelevations (calculated using sample d13C values and the d13C regression in Fig. 5) are plotted by depth. See text for detailed discussion.

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

143

Fig. 7. Bird Point core transect BeB0 . (A) Depicts core locations and depths along transect BeB0 , soil types intersected, depth and lateral continuity of the inferred 1964 contact, and sample locations in each core above and below the contact. (B) Plot of d13C values for mud and peat samples in cores taken along transect BeB0 . Samples from above and below the contact in cores BPP4 and BPP6 had almost indistinguishable d13C values, which we assume was the result of sampling too close to either side of the contact (samples were ~1 cm apart). We did not include these indistinguishable values here or in subsequent analyses.

to 26.83‰, increasing toward Turnagain Arm (seaward) (Fig. 7B). Values of d13C increase 3.03 to 0.94‰ (standard deviation ¼ 0.60) from peat into mud across the 1964 contact, an average increase of 1.98‰ across the contact on transect BeB0 . 4. Estimating coseismic subsidence To test if the increase in d13C values from peat to mud across the 1964 contact is consistent with the amount of coseismic subsidence that occurred in Turnagain Arm during the 1964 earthquake, we used the slopeeintercept equation for the regression of d13C against elevation to calculate paleoelevations for each sample. Using this regression, reconstructed elevations for peat below the contact should approximate the pre-1964 earthquake surface elevation. Similarly, reconstructed elevations for mud above the contact should approximate the elevation of the surface following earthquake subsidence in 1964. Therefore, the difference between reconstructed peat and mud elevations provides an estimate of coseismic subsidence (Fig. 8). However, the modern transects sampled in this study extend only to the top of each site's tidal

range. Therefore, the upper extent of the tidal range places an upper bound on the regressions used to reconstruct paleoelevation. As a result, the reconstructed paleoelevations near the upper end of the tidal range provide a minimum constraint on the amount of subsidence. The d13C regression standard error (±0.4 m) quantifies the uncertainty of the reconstructed elevations. The reported precision is limited to the decimeter in order to reflect the precision of the 10-cm vertical intervals of surface sediment samples on transects AeA0 , BeB0 , and CeC0 . Samples across the 1964 contact in cores along Bird Point transect BeB0 have mud d13C values that correspond to paleoelevations between 0.15 and 0.65 m (MHHW). d13C values for peat samples correspond to paleoelevations between 1.7 and 0.6 m (MHHW). The arithmetic mean of the differences between reconstructed mud and peat elevations provides an estimate of 1.3 ± 0.4 m of coseismic subsidence at Bird Point during the 1964 earthquake (Fig. 8). This subsidence estimate is within 0.4 m of estimates derived from post-earthquake leveling along Turnagain Arm (~1.7 m, Small, 1966; Wood, 1966), and within 0.2e1.0 m of two independent microfossil transfer function estimates at

144

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

Fig. 8. d13C -based subsidence envelope for Bird Point transect BeB0 . Predicted mud and peat elevations bound the bottom and top of the blue envelope, and are calculated using the d13C values for mud and peat in Fig. 7B as the y variable in the slopeeintercept equation for d13C Fig. 5. Regression standard error for each predicted value is the dashed line that bounds the light gray outer envelope. The average subsidence estimate (1.3 m) is the arithmetic difference between the average predicted elevations for peat and mud. The reported error (±0.4 m) is the average regression standard error calculated for peat and mud predicted elevations. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Girdwood (1.5 ± 0.32 m, Hamilton and Shennan, 2005) and Bird Point (~1.9e2.3 m, Watcham et al., 2013). We also used the method described above to calculate paleoelevations for d13C values in samples from Girdwood exposure GC01 (Fig. 6). Reconstructed elevations range from 0.62 to 1.68 m MHHW within the pre-1964 peat at GC-01, reflecting a variability of 26.89 to 28.56‰ d13C values within this key unit, and allowing for a wide range of mostly inaccurate subsidence estimates across the contact. At GC-01 we logged the 1964 contact at 60.5 cm depthdsamples from 1-cm on either side of the contact document an increase of 0.40‰ d13C from peat (27.24‰) into mud (27.64‰), yielding 0.3 ± 0.4 m of subsidence (a >1 m underestimate). The >1‰ variability of d13C values and corresponding reconstructed elevations observed down section at GC-01 shows that a pair of samples taken at any single location across the 1964 contact may bias the elevation estimate and lead to considerable error. To reduce potential errors that might result from isotopic variability at a single point, we analyzed multiple samples above and below the 1964 contact from multiple cores on Bird Point transect BeB0 . Our multi-core approach increased sample density in order to generate multiple subsidence estimates across the marsh environment. This method also allows a test of the internal consistency among the samples. Also, we found that taking the arithmetic mean of subsidence estimates across the transect reduces error contributed by possible outlying points in modern isotopic data, and results in an estimate at Bird Point that compares well with several independent estimates of earthquake subsidence in 1964. 5. Discussion Linear regressions of data from modern surface transects at Bird Point and Girdwood demonstrate that d13C values decrease with

increasing elevation above the MHHW reference level in the Turnagain Arm tidal frame. Comparatively large regression standard error and low R2 values for d15N and Corg:Ntot along the same transects make these parameters less reliable proxies for reconstructing RSL changes along the coast of southern Alaska. However, the relationship between contemporary sediment d13C and elevation at these sites is clear, and we apply it to estimate subsidence across the 1964 contact at Bird Point (Fig. 8). We applied a linear model relating sediment d13C to elevation to estimate RSL change caused by earthquake subsidence in 1964 at Bird Point. This method estimates 1.3 ± 0.4 m of RSL rise that is within 0.4 m of measured subsidence from post-earthquake leveling along Turnagain Arm (~1.7 m, Small, 1966; Wood, 1966), and within 0.2e1.0 m of two independent microfossil transfer function subsidence estimates at Girdwood and Bird Point (1.5 ± 0.32 m, Hamilton and Shennan, 2005; ~1.9e2.3 m, Watcham et al., 2013). The close agreement between our Bird Point BeB0 results and other independent subsidence estimates lends confidence in using linear regressions of d13C values to reconstruct relative sea-level changes in Alaskan coastal environments. However, results from Girdwood outcrop GC-01 (Fig. 6) demonstrate that d13C can have considerable local variability. The results from Bird Point transect BeB0 demonstrate that averaging multiple samples may be necessary to achieve accurate results when sediment d13C values display considerable scatter. The considerable down-section variability of sediment Corg:Ntot ratios from the Girdwood outcrop suggest that complicated mixing of multiple sources of organic material may explain the low subsidence estimate at the Girdwood site (Fig. 6). The relative contributions of tidally-derived POM (possibly phytoplankton) appear to deflect Corg:Ntot values in the peat below the 1964 contact toward very low values (<15) that are typical of mud flat sediment (Fig. 5). Strangely, these anomalously low ratios in peaty

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

sediment are lower than all but one ratio in muddy sediment above the 1964 contact. The challenges in reconstructing RSL change at the Girdwood site may be the result of organic material contributed from multiple sources, including riverine and terrestrial POM coming out of Glacier Creek, in addition to daily tidal inundation. The method used here differs from other approaches that apply bulk organic d13C and Corg:Ntot values to sea level reconstructions. For example, Kemp et al. (2011) and Engelhart et al. (2013) used a multiproxy approach that combined d13C in bulk organic sediment with the presence or absence of agglutinated foraminifera. In this study, we did not integrate microfossil proxies into the analyses because our purpose was to assess the reliability of stable isotopes in bulk organic sediment as an alternative RSL indicator for use in Alaskan sites where biostratigraphic methods are challenged by sparse abundance, poor preservation, or an absence of microfossils. Our method further diverges from the method used by Engelhart et al. (2013) by using simple linear regression analyses to relate bulk sediment geochemical parameters to elevation rather than more sophisticated cluster analyses. The cluster analyses used by Engelhart et al. (2013) allow an objective way of recognizing the indicative meaning (indicative range and reference water level) for specific, statistically-defined groups of the various parameters. Because the geochemical data from sites along Turnagain Arm showed no clear groupings, or ‘clusters,’ like that frequently observed in foraminiferal assemblages, we opted to use a straight-forward linear regression to relate elevation to bulk sediment geochemistry. The relative advantage of using linear regression on d13C values in multiple cores is illustrated at Bird Point, where Watcham et al. (2013) used microfossil-based transfer functions to calculate 3.4 ± 1.2 m of subsidence across the 1964 contactdan estimate that is double the subsidence measured by post-earthquake leveling, and 2.1 m greater than our d13C-based estimate. Watcham et al. (2013) suggest that their approach may overestimate subsidence at Bird Point in part because of the small (n ¼ 42) modern microfossil training set in Turnagain Arm, and the lack of good analogues for peat diatoms in the regional microfossil training set. Similarly, Shennan et al. (2014) reported that the variable preservation of diatom species in silt stratigraphy, and sparse abundance of tidal flat diatoms hamper the quantitative analysis of coseismic relative sea-level change at sites on the Copper River Delta, Alaska. We suggest that where such problems inhibit microfossil-based approaches to reconstructing past relative sea-level changes, linear regression analysis of d13C values in multiple samples may provide a good alternative. The lower coseismic subsidence estimates derived from sediment d13C values may result from error contributed by three different processes. First, decomposition of organic matter can result in a decrease in bulk organic sediment values of 2.0‰e6.0‰ (Benner et al., 1987). While it appears that d13C values of the pre-1964 peat increase toward lower intertidal elevations (Fig. 7B), as is shown in modern sediment d13C values, our data cannot address whether or not decomposition has had a differential effect on d13C values between sedge marsh-derived sediment and mudflat sediment. A second process that may influence d13C values across the 1964 contact is differential mixing of tidal-versus riverine-derived POM along Turnagain Arm. This process could explain the lower d13Cbased subsidence estimate at the Girdwood site, directly adjacent to the mouth of Glacier Creek, compared to Bird Point. Different contributions of terrigenous and riverine POM transported by Glacier Creek may explain the lower d13C values in mud directly above the 1964 contact at the Girdwood site (Fig. 6). In addition, high variability with depth displayed by Corg:Ntot ratios, especially near the 1964 contact, implies heterogeneous inputs of POM from both tidal and riverine sources.

145

Finally, we noticed that the mud-over-peat contact marking subsidence in 1964 at the Girdwood site was gradual in nature rather than sharp like correlative contacts observed at outcrops nearby. The gradual contact suggests that sediment mixing by bioturbation or strong tidal currents following the 1964 earthquake (e.g., Greb and Archer, 2007) may have introduced sediment with lower d13C values into muddy deposits that overlie the 1964 peat. Further testing of linear regression analysis of sediment d13C values to reconstruct RSL change related to the earthquake deformation cycle will assess the fidelity of coseismic displacement estimates, especially at sites of known earthquake related deformation. 6. Conclusions We tested the use of bulk organic d13C, d15N, and Corg:Ntot ratios in intertidal sediment from sites along Turnagain Arm to estimate subsidence during the 1964 great Alaska earthquake. Linear regressions of all three sediment parameters versus elevation show systematic variations across the intertidal zone from the mud flat to high marsh. However, considerable variability in the data presented challenges that required analyses of multiple samples to reduce potential errors. Linear relationship between sediment d13C and elevation (R2 ¼ 0.56) provided the most reliable proxy for reconstructing paleoelevation of the ground surface before and after the 1964 earthquake. Regressions of sediment d15N values and Corg:Ntot ratios versus elevation (R2 ¼ 0.33 and 0.05, respectively) provide less reliable proxies for estimating RSL change at sites along Turnagain Arm. The results of this research lead to the following conclusions: (1) Values of d13C in modern surface sediment at three sites decrease with increasing elevation above MHHW in the Turnagain Arm tidal frame. (2) Values of d13C also differentiate between terrestrial peaty sediment below, and muddy estuarine sediment above the 1964 earthquake contact in 9 cores taken in a cross-marsh transect at Bird Point. (3) A linear regression of modern d13C values versus elevation was used to reconstruct paleoelevations for samples of mud above and peat below the 1964 contact in nine cores taken at Bird Point. By taking the arithmetic mean of the nine differences between peat and mud paleoelevations, we calculate 1.3 ± 0.4 m of subsidence, which agrees with independent estimates of Turnagain Arm subsidence in 1964. (4) Detailed analysis of d13C values in a single section at Girdwood reveals a >1‰ range of d13C values within the pre-1964 peat. This internal variability may reflect some combination of decomposition diagenesis, differential POM mixing near the mouth of Glacier Creek, or sediment mixing across the 1964 contact. Such complicating factors present potential errors in reconstructing paleoelevation that were overcome by multiple paired samples in multiple cores that intersected the 1964 subsidence contact. (5) Where problems of sparse abundance, poor preservation, or absence of microfossils inhibit biostratigraphic methods of reconstructing past earthquake-driven RSL changes, analyses of bulk organic d13C in intertidal sediment may provide a reliable alternative for reconstructing coastal environmental change in Alaska. Acknowledgments This work was supported in part by two undergraduate research grants to A. Bender, including (1) a University of Alaska, Fairbanks (UAF) Center for Global Change Student Research Grant with funds

146

A.M. Bender et al. / Quaternary Science Reviews 113 (2015) 134e146

from the Cooperative Institute for Alaska Research and the UAF Vice Chancellor for Research and (2) the Fran Ulmer Transformative Research Award from the University of Alaska, Anchorage. The U.S. Geological Survey's Alaska Earthquake Hazards Project supported A. Bender and R. Witter. We thank LeeAnn Munk, Casey Saenger and Peter Haeussler for discussions that motivated the research and guidance that facilitated the application of a new approach in Alaska. Abby Gahm, Lillian Pickering, Kelly Thomson and Ray Witter assisted in the field. We thank two anonymous reviewers for constructive comments that improved the manuscript. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. References Archer, A.W., 2013. World's highest tides: hypertidal coastal systems in North America, South America and Europe. Sediment. Geol. 284e285, 1e25. http:// dx.doi.org/10.1016/j.sedgeo.2012.12.007. Benner, R., Fogel, M.L., Sprague, K.E., Hodson, R.E., 1987. Depletion of 13C in lignin and its implications for stable carbon isotope studies. Nature 329, 708e710. Bentley, M.J., Hodgson, D.A., Sugden, D.E., Roberts, S.J., Smith, J.A., Leng, M.J., Bryant, C., 2005. Early Holocene retreat of the George VI Ice Shelf, Antarctic Peninsula. Geology 33, 173e176. Brodie, C.R., Leng, M.J., Casford, J.S., Kendrick, C.P., Lloyd, J.M., Yongqiang, Z., Bird, M.I., 2011a. Evidence for bias in C and N concentrations and d13C composition of terrestrial and aquatic organic materials due to pre-analysis acid preparation methods. Chem. Geol. 282 (3), 67e83. Brodie, C.R., Heaton, T.H., Leng, M.J., Kendrick, C.P., Casford, J.S., Lloyd, J.M., 2011b. Evidence for bias in measured d15N values of terrestrial and aquatic organic materials due to pre-analysis acid treatment methods. Rapid Commun. Mass Spectrom. 25 (8), 1089e1099. Chmura, G.L., Aharon, P., 1995. Stable carbon isotope signatures of sedimentary carbon in coastal wetlands as indicators of salinity regime. J. Coast. Res. 11, 124e135. Craven, K.F., Edwards, R.J., Goodhue, R., Rocha, C., 2013. Evaluating the influence of selected acid pre-treatment methods on C/N and d13C of temperate inter-tidal sediments for relative sea level reconstruction. Ir. J. Earth Sci. 31 (1), 25e42. Deines, P., 1980. The isotopic composition of reduced organic carbon. In: Fritz, P., Fontes, J.C. (Eds.), Handbook of Environmental Isotope Geochemistry. The Terrestrial Environment, vol. 1.A. Elsevier, Amsterdam, pp. 329e406. Emery, K.O., Wigley, R.L., Bartlott, A.S., Rubin, M., Barghoorn, E.S., 1967. Fresh water peat on the continental shelf. Science 158, 130e137. Engelhart, S.E., Horton, B.P., Vane, C.H., Nelson, A.R., Witter, R.C., Brody, S.R., Hawkes, A.D., 2013. Modern foraminifera, d13C, and bulk geochemistry of central Oregon tidal marshes and their application in paleoseismology. Palaeogeogr. Palaeoclimatol. Palaeoecol. 377, 13e27. http://dx.doi.org/10.1016/ j.palaeo.2013.02.032. Greb, S.F., Archer, A.W., 2007. Soft-sediment deformation produced by tides in a meizoseismic area, Turnagain Arm, Alaska. Geology 35 (5), 435e438. http:// dx.doi.org/10.1130/G23209A.1. Hamilton, S., Shennan, I., 2005. Late Holocene relative sea-level changes and the earthquake deformation cycle around upper Cook Inlet, Alaska. Quat. Sci. Rev. 24 (12e13), 1479e1498. http://dx.doi.org/10.1016/j.quascirev.2004.11.003. Hulten, E., 1968. Flora of Alaska and Neighboring Territories, a Manual of the Vascular Plants. Stanford Univ. Press, Stanford, Calif, p. 1008. Keeley, J.E., Sandquist, D.R., 1992. Carbon: freshwater plants. Plant Cell Environ. 15, 1021e1035. Kemp, A.C., Vane, C.H., Horton, B.P., Culver, S.J., 2010. Stable carbon isotopes as potential sea-level indicators in salt marshes, North Carolina, USA. Holocene 20 (4), 623e636. http://dx.doi.org/10.1177/0959683609354302. Kemp, A.C., Vane, C.H., Horton, B.P., Engelhart, S.E., Nikitina, D., 2011. Application of stable carbon isotopes for reconstructing salt-marsh floral zones and relative sea level, New Jersey, USA. J. Quat. Sci. 27 (4), 404e414. http://dx.doi.org/ 10.1002/jqs.1561. Lamb, A.L., Wilson, G.P., Leng, M.J., 2006. A review of coastal palaeoclimate and relative sea-level reconstructions using d13C and Corg:Ntot ratios in organic material. Earth-Sci. Rev. 75, 29e57.

Lamb, A.L., Vane, C.H., Wilson, G.P., Rees, J.G., Moss-Hayes, V.L., 2007. Assessing d13C and C/N ratios from organic material in archived cores as Holocene sea level and palaeoenvironmental indicators in the Humber Estuary, UK. Mar. Geol. 244 (1e4), 109e128. http://dx.doi.org/10.1016/j.margeo.2007.06.012. Malamud-Roam, F., Ingram, B.L., 2001. Carbon isotopic compositions of plants and sediments of tide marshes in the San Francisco estuary. J. Coast. Res. 17, 17e29. Meyers, P.A., 1994. Preservation of elemental and isotopic source identification of sedimentary organic matter. Chem. Geol. 114 (3), 289e302. Middelburg, J.J., Nieuwenhuize, J., 1998. Carbon and nitrogen stable isotopes in suspended matter and sediments from the Schelde Estuary. Mar. Chem. 60 (3), 217e225. Owens, N.J.P., 1987. Natural variations in nitrogen-15 in the marine environment. Adv. Mar. Biol. 24, 389e451. Owens, N., Law, C.S., 1989. Natural variations in 15N content of riverine and estuarine sediments. Estuar. Coast. Shelf Sci. 28 (4), 407e416. Peterson, B.J., Fry, B., 1987. Stable isotopes in ecosystem studies. Annu. Rev. Ecol. Syst. 293e320. Plafker, G., 1965. Tectonic deformation associated with the 1964 Alaska earthquake the earthquake of 27 March 1964 resulted in observable crustal deformation of unprecedented areal extent. Science 148 (3678), 1675e1687. Plafker, G. , 1969. Tectonics of the March 27, 1964, Alaska Earthquake. US Government Printing Office, p. 74. Pojar, J., MacKinnon, A. (Eds.), 1994. Plants of the Pacific Northwest Coast. B.C. Ministry of Forests and Lone Pine Publishing, Redmond, WA, p. 526. Shennan, I., Bruhn, R., Barlow, N., Good, K., Hocking, E., 2014a. Late Holocene great earthquakes in the eastern part of the Aleutian megathrust. Quat. Sci. Rev. 84, 86e97. http://dx.doi.org/10.1016/j.quascirev.2013.11.010. Shennan, I., Long, A.J., Horton, B.P. (Eds.), 2015. Handbook of Sea-level Research. John Wiley & Sons, Ltd, Chichester (in press). http://www.wiley.com/WileyCDA/ WileyTitle/productCd-1118452585.html#see-less-auth. Small, J.B., 1966. Vertical Bench Mark Displacement, pt. 1 of Alaskan Surveys to Determine Crustal Movement. U.S. Coast and Geod. Survey, p. 24. Smith, B.N., Epstein, S., 1971. Two categories of 13C/12C ratios for higher plants. Plant Physiol. 47, 380e384. Snow, A., Vince, S.W., 1984. Plant zonation in an Alaskan salt marsh: II. An experimental study of the role of edaphic conditions. J. Ecol. 72 (2), 669e684. Stephenson, R.L., Tan, F.C., Mann, K.H., 1984. Stable carbon isotope variability in marine macrophytes and its implications for food web studies. Mar. Biol. 81 (3), 223e230. Thornton, S.F., McManus, J., 1994. Application of organic carbon and nitrogen stable isotope and C/N ratios as source indicators of organic matter provenance in estuarine systems: evidence from the Tay Estuary, Scotland. Estuar. Coast. Shelf Sci. 38 (3), 219e233. van de Plassche, O., 1986. Sea-level Research: a Manual for the Collection and Evaluation of Data. GeoBooks, Norwich. Vince, S.W., Snow, A.A., 1984. Plant zonation in an Alaskan salt marsh, I Distribution, abundance, and environmental factors. J. Ecol. 72, 651e667. Watcham, E.P., Shennan, I., Barlow, N.L.M., 2013. Scale considerations in using diatoms as indicators of sea-level change: lessons from Alaska. J. Quat. Sci. 28 (2), 165e179. http://dx.doi.org/10.1002/jqs.2592. Wilson, G.P., Lamb, A.L., Leng, M.J., et al., 2005a. d13C and C: N as potential coastal palaeoenvironmental indicators in the Mersey Estuary, UK. Quat. Sci. Rev. 24, 2015e2029. Wilson, G.P., Lamb, A.L., Leng, M.J., et al., 2005b. Variability of organic d13C and C:N in the Mersey Estuary, U.K. and its implications for sea-level reconstruction studies. Estuar. Coast. Shelf Sci. 64, 685e698. Wilson, G.P., Lamb, A.L., 2012. An assessment of the utility of regional diatom-based tidal-level transfer functions. J. Quat. Sci. 27 (4), 360e370. Wood, F.J. (Ed.), 1966. Operational Phases of the Coast and Geodetic Survey Program in Alaska for the Period March 27 to December 31, 1964, v. 1 of the Prince William Sound, Alaska, Earthquake of 1964 and Aftershocks. U.S. Coast and Geodetic Survey, p. 263. Yu, F., Zong, Y., Lloyd, J.M., Huang, G., Leng, M.J., Kendrick, C., et al., 2010. Estuarine, coastal and shelf science. Estuar. Coast. Shelf Sci. 87 (4), 618e630. http:// dx.doi.org/10.1016/j.ecss.2010.02.018. Zong, Y., Lloyd, J.M., Leng, M.J., Yim, W.W.S., Huang, G., 2006. Reconstruction of Holocene monsoon history from the Pearl River Estuary, southern China, using diatoms and carbon isotope ratios. Holocene 16 (2), 251e263. http://dx.doi.org/ 10.1191/0959683606hl911rp.