Introduction

American horseshoe crabs (Limulus polyphemus) spawn on sandy estuarine beaches throughout their range (Maine, USA to Yucatan, Mexico). The species is economically valuable as eel and whelk bait and for the production of Limulus amoebocyte lysate (LAL) used in the pharmaceutical industry for the detection of bacterial endotoxin (Atlantic States Marine Fisheries Commission 1998; Shuster et al. 2003; Novitsky 2009). Limulus is a key species in the estuarine food web, and its eggs provide essential food for a number of migratory shorebirds (Botton 2009). Horseshoe crabs are an important focal point for ecotourism and marine education in the middle Atlantic region of the USA (Riepe 2001; O’Connell et al. 2009).

Spawning behavior and mate selection in horseshoe crabs have been studied extensively (e.g., Brockmann and Smith 2009; Brockmann et al. 2015). In brief, males and females in amplexus, often accompanied by unattached “satellite” males, migrate to sandy estuarine beaches to spawn on intertidal beaches. Some horseshoe crab populations exhibit strong lunar periodicity, with the largest number of animals spawning on Spring tides (new and full moons) (Rudloe 1980; Barlow et al. 1986; Watson and Chabot 2010), but in other studies, the association between spawning activity and lunar phase is weak (Ehlinger et al. 2003; James-Pirri et al. 2005; Leschen et al. 2006; Cheng et al. 2016). Fertilization is external; the females lay clusters of ca. 1000 to 5700 eggs and deposit them at depths of 15 cm or more (Weber and Carter 2009; Botton et al. 2010), a depth where sediments on estuarine beaches are rarely reworked by wave action (Jackson et al. 2005a). Embryonic development takes place within the sand, culminating in the hatching of first instar (trilobite) larvae after about 1 month. The larvae emerge from the sand and are briefly planktonic before assuming a benthic existence (Botton et al. 2010).

As estuarine-dependent animals, horseshoe crabs will be impacted by rising sea levels associated with global warming, a phenomenon that will have significant impacts on coastal infrastructure and ecosystems (FitzGerald et al. 2008; Williams and Gutierrez 2009; Defeo et al. 2009; Nicholls and Cazenave 2010). Since 1900, global sea level has risen by some 1.7 mm year−1, with the most rapid rate of increase (3.2 mm year−1) occurring from 1993 to 2010 (Intergovernmental Panel on Climate Change 2013). Glacial melt and thermal expansion of ocean water account for the majority of sea level rise (Nicholls and Cazenave 2010). These global processes are predicted to continue throughout the twenty-first century, leading to further increases in sea level (Intergovernmental Panel on Climate Change 2013). In some areas, notably the middle Atlantic region of the USA, rates of relative sea level rise will be even greater than the global average because of land subsidence and the compaction of sediments resulting from the withdrawal of freshwater from coastal aquifers (Williams and Gutierrez 2009).

Jamaica Bay, New York (Fig. 1), is a highly urbanized and eutrophic estuary; it has limited inflow from natural rivers or streams and receives the majority of its freshwater and pollutant inputs from water treatment plants, storm sewers, and combined sewer overflows (O’Shea and Brosnan 2000; Beck et al. 2009). The construction of a jetty at Breezy Point in the 1930s to stabilize Rockaway Inlet, along with development on the Rockaway Peninsula, has limited sand overwash during storm events (Messaros et al. 2012). Much of the low-lying Jamaica Bay system was considered to be highly vulnerable to sea level rise, even prior to Hurricane Sandy in late October, 2012 (Rosenzweig et al. 2011; Solecki 2012). The bay is experiencing rapid losses of beaches and tidal marshes, attributed to global sea level rise, land subsidence, and eutrophication (Gornitz et al. 2002; Hartig et al. 2002; Wigand et al. 2014), and our study site at Plumb Beach (Fig. 1) has an insufficient supply of sand to maintain itself (Psuty et al. 2013).

Fig. 1
figure 1

Map of the study area in 2012, prior to the beach nourishment project showing the locations of the nourishment and reference sections

Full size image

Satellite imagery clearly shows that the western portion of Plumb Beach (closest to the highway), exposed to the greatest fetch, experienced rapid and severe erosion following a prior beach nourishment project in 1992 (Fig. 2). The majority of the sand moved eastward, but there has also been some deposition at the far westernmost section (Psuty et al. 2013). The most severely eroded section had been stabilized with sandbags and rubble fill to protect the adjacent Belt Parkway (Fig. 3a), a major highway with an average daily traffic volume of over 140,000 vehicles (New York State Department of Transportation 2017). In contrast, the net west to east transport of sediment from the longshore drift has resulted in substantial sand accretion at the eastern section, resulting in a wide and relatively undisturbed sandy beach (Figs. 2 and 3b).

Fig. 2
figure 2

Google Earth images of Plumb Beach, Jamaica Bay. a 1994, less than 2 years after a beach nourishment project. b 2003, note the net movement of sand from left (west) to right (east). c July 2012, prior to the beach nourishment project. Note the continued deposition of sand to the east and the intrusion of the sandy beach into the former salt marsh. d May 2015, after the completion of the nourishment project. The white bars show the approximate locations of the three transects at each section where the core and sediment samples were taken. Note the two groins and bulkhead on the western section. All images were taken from an altitude of 1.83 km

Full size image
Fig. 3
figure 3

a Western portion of Plumb Beach in summer, 2012 prior to the start of the beach nourishment project, showing the presence of sandbags and rubble fill in the intertidal zone. The guard rail of the Belt Parkway can be seen in the upper left of the figure. b Eastern portion of Plumb Beach in summer, 2012 (photographs by M. Botton)

Full size image

Despite the highly urbanized character of Jamaica Bay, Plumb Beach is one of the most important spawning locations for horseshoe crabs in New York State (Sclafani et al. 2009). Prior research at Plumb Beach was limited to a few high tide beach surveys in 1998 (Hanna 2001). In a bay-wide survey, Botton et al. (2006) concluded that horseshoe crab abundance in Jamaica Bay was limited by the availability of suitable spawning habitat rather than by water quality. Well-oxygenated sandy beaches of medium to coarse sand, away from the influence of hydrogen sulfide, are considered ideal (Botton et al. 1988; Penn and Brockmann 1994; Vasquez et al. 2015). If beaches are stabilized by groins, bulkheads, or revetments, sediment grain size characteristics may be affected and horseshoe crabs may be prevented from accessing portions of the beach foreshore (Jackson and Nordstrom 2009; Jackson et al. 2015).

The US Army Corps of Engineers (USACE), partnering with the New York City Department of Parks & Recreation, performed a beach restoration project at Plumb Beach in Fall 2012 to stabilize the eroding shoreline and protect the adjacent highway (USACE 2012) (Figs. 1, 2d, and 3a). Beach nourishment took place in Fall 2012 and involved the pumping of 97,098 m3 of sand from a nearby channel onto 610 m of the western section of Plumb Beach (http://beachnourishment.wcu.edu/, Fig. 4). Hurricane Sandy delayed the final stages of the project (Fig. 2d), including the planting of dune vegetation and construction of two terminal groins and a 61 m stone breakwater until Summer 2013, after the horseshoe crab spawning season. The primary goal of the USACE project was the protection of the highway, but it did provide us with an opportunity to assess the impacts of beach nourishment on horseshoe crabs using a BACI (before-after control impact) design. Beach nourishment is an important engineering tool to maintain and restore shoreline habitats, and should ideally have beneficial effects on the ecological community, in addition to fulfilling the desired purpose of shoreline protection (Schlacher et al. 2014). Although beach nourishment has been conducted on some estuarine shorelines, notably in Delaware (Jackson and Nordstrom 2009), there has yet to be a BACI-type study that examines the use of these restored shorelines by horseshoe crabs.

Fig. 4
figure 4

Western section of Plumb Beach in December, 2012 following completion of the beach nourishment project (photograph by M. Botton)

Full size image

The purpose of this study was to assess the effect of the beach nourishment project on spawning horseshoe crabs at Plumb Beach. Our specific objectives were to evaluate and contrast horseshoe crab spawning abundance, egg densities pre- and post-nourishment at the severely eroded western section of Plumb Beach in comparison to the eastern (reference) section. We also assessed beach geomorphology and sediment texture throughout the study, because previous studies have shown that these are important influences on nest-site selection by spawning horseshoe crabs (Botton et al. 1988; Jackson et al. 2007; Jackson and Nordstrom 2009).

Materials and Methods

Study Area

Plumb Beach (Fig. 1) is located in Jamaica Bay, New York, near the mouth of the estuary. The shoreline at Plumb Beach is under the jurisdiction of the National Park Service as a unit of Gateway National Recreation Area. Common recreational activities at Plumb Beach include windsurfing, fishing, and birding. The western section is the area where beach nourishment took place, and the eastern section serves as our reference site.

Densities of Spawning Adults

Surveys of spawning adult horseshoe crabs in New York State have included the eastern (reference) section of Plumb Beach (approximately 0.9 km) since 2005 in the New York Horseshoe Crab Monitoring Network, a citizen science survey coordinated state-wide by the Cornell University Cooperative Extension of Suffolk County, New York State Department of Environmental Conservation and locally by the New York City Audubon Society. The survey was expanded to include the western (nourishment) section of Plumb Beach (approximately 0.85 km) in 2012. Sampling dates (n = 12 per year) were standardized to bracket the evening Spring tides (full and new moons) by surveying 2 days before the Spring tide, the date of Spring tide, and 2 days after the Spring tide in May and June in 2012 (pre-nourishment) and 2013–2015 (post-nourishment) to capture the predicted peak spawning activities. In the eastern region, we followed the sampling protocol of Smith et al. (2002a) where teams counted all animals in 100 randomly selected 1 m2 quadrats positioned within the surf zone at the time of predicted high tide ± 30 min, along the 0.9 km of beach. Males and females were identified and counted separately, and only those animals that were more than halfway within the quadrat were included. Because horseshoe crab densities in the western section were much lower throughout the 4-year study, in lieu of using quadrats, the teams counted all visible horseshoe crabs at or near the water’s edge while walking the entire 0.85 m length of beach (Sclafani et al. 2013). Spawning counts are expressed as the Index of Spawning Activity (ISA), which was the average number of spawning females per square meter counted on the evening (highest) high tides (Smith et al. 2002a).

Egg Density

The density of horseshoe crab eggs is a useful index of habitat suitability (Smith et al. 2002b; Pooler et al. 2003; Botton et al. 2006). We selected six sampling transects at Plumb Beach at three roughly equidistant positions at both the eastern (reference) and western (nourishment) sections (Fig. 2d). Within each transect, eggs were sampled at random points within a 15 long × 3 m wide rectangle centered at the mean tide level, where egg density is known to be greatest (Botton et al. 1994; Smith et al. 2002b; Weber and Carter 2009). Cores were collected at approximately 2–3-week intervals during the spawning season in 2012 (pre-nourishment) and 2013–2015 (post-nourishment) using a 3.8-cm diameter PVC tubing to a depth of 20 cm. On each sampling date, we collected ten cores from each transect (= 30 cores each for the reference and nourishment sites).

We placed sand from each core into labeled plastic bags and refrigerated them until analysis (within 1 week). Each core was washed through a 1.0-mm sieve to separate the eggs from the sand, and all horseshoe crab eggs, embryos, and first instars (trilobites) were staged (Botton et al. 2010) and counted separately. Eggs that were greenish-blue were considered viable; those that were black, brown, or purple were considered dead.

For the BACI design, the null hypothesis was that the differences in densities between the nourishment and references sites would be unchanged following beach nourishment. Means for total stages (i.e., combined eggs, embryos, and larvae) were first calculated separately for each sampling date for the reference and nourishment sites on log-transformed data. The differences between the means of the two sites on each sampling date were calculated, and the differences for the pre-treatment (2012) and post-treatment (2013–2015) years were then compared. As it was noted that time of sampling had a strong effect on the numbers of total stages, the comparison was done as an analysis of covariance (ANCOVA), with pre-and post-treatment as the factor of interest, and time of sampling as a covariate. The relationship to sampling date within year was not linear, but was approximated quite well by a parabola (i.e., low abundance in early and late season, with a peak in mid-season), so the time covariate was included as a quadratic. In addition to the main analysis on total stages, separate analyses were run for live eggs, embryos, and trilobites. In the case of the embryos, including a quadratic term for time of sampling did not improve the fit of the model, and so it was omitted. In the case of the trilobites, including time as a covariate did not improve the fit of the model, and so was omitted. All of these analyses were done using Systat v12 (Systat Software Inc., San Jose, CA).

Geomorphology and Sediment Composition

We measured beach width at the same transects where eggs were sampled (3 west, 3 east) from the Spring high tide line to the beginning of the intertidal flat, and slope within the mid-tide region was determined using the technique described by Emery (1961). At each site, we collected sediment samples from three equally spaced stations along the transect from high to low water, designated as upper, middle, and lower intertidal. The compressive strength of the surface mid-beach sediment was measured using a spring-operated pocket penetrometer (Forestry Suppliers Inc., model 77114). We collected approximately 300 g of sand at each station from the surface (0–5 cm) and deep (15–20 cm) layers, and we measured the distance from the sediment surface to the anaerobic layer (black sand), if present. Sand was dried at 60 °C for 1 week., and approximately 15 g of the dried sand was set aside for the determination of organic carbon, expressed as the percent loss on ignition after 5 h at 500°C in a muffle furnace (Byers et al. 1978). The remaining dried sand was then fractionated using a sieve series with 4 mm, 2 mm, 1 mm, 500 μm, 250 μm, 125 μm, and 63 μm mesh openings. All grain-size characteristics, including percent gravel, geometric mean, sorting, and skewness, were calculated using GRADISTAT Version 8.0 for Excel versions 2007–2010 (http://www.kpal.co.uk/gradistat.html) developed by Blott and Pye (2001).

Results

Densities of Spawning Adults

In 2012, prior to beach nourishment, there were very few female horseshoe crabs spawning at the western section of Plumb Beach (ISA = 0.003; Fig. 5). Likewise, in 2013, the first Spring following beach nourishment, there were very few females spawning on the western beach (ISA = 0.001). From 2014 onward, ISA’s have increased at the western section, reaching 0.23 in 2015. Similar patterns of abundance were seen for males and total individuals (data not shown). The operational sex ratio (ratio of spawning males to spawning females) has been consistently male-biased at the nourishment site, ranging from 1.63 males/females in 2012 to 3.79 in 2014.

Fig. 5
figure 5

Index of Spawning Activity (number of female horseshoe crabs per quadrat or per m2 ± SE) in 2012 (pre-nourishment) and 2013–2015 (post-nourishment) at the western (nourishment) and eastern (reference) sections of Plumb Beach

Full size image

Spawning densities have been consistently higher at the reference site than at the nourishment site throughout the study (Fig. 5). The highest ISA (0.458) was found in 2013, and the lowest ISA (0.068) was found in 2015, with similar patterns of abundance for males and total individuals (data not shown). Operational sex ratios at the reference site have ranged between 2.26 (2013) and 2.82 (2014). Overall, ISA’s have always been consistently higher for the reference site than the nourishment site, although the disparity between sites was greatest in 2012 and 2013 and has since been reduced.

Egg Density

Beach nourishment of the western section of Plumb Beach in Fall 2012 was not followed by a detectable increase in horseshoe crab egg deposition in the first three post-nourishment seasons, 2013–2015. The western area had significantly fewer horseshoe crab eggs and later developmental stages (embryos and trilobite larvae) in comparison to the eastern section both before nourishment (2012) and in the 3 years following nourishment (2013–2015) (Fig. 6). The pattern of egg deposition at the eastern section varied annually; in 2012, spawning began relatively early and the peak count of live eggs occurred on 22 May; whereas, in 2013, 2014, and 2015, the peaks in live eggs occurred several weeks later. The percentage of cores having ≥ 1 horseshoe crab egg during May and June was low at the western section (range 0–27%) compared with the eastern section (range 30–100%) (Fig. 7).

Fig. 6
figure 6

Mean number (± SE) of horseshoe crab developmental stages (combined eggs, embryos and trilobites) per 20 cm core in 2012 (pre-nourishment) and 2013–2015 (post-nourishment) at the western (nourishment) and eastern (reference) sections of Plumb Beach

Full size image
Fig. 7
figure 7

Percentage of core samples having one or more horseshoe crab developmental stages (combined eggs, embryos, and trilobites) in 2012 (pre-nourishment) and 2013–2015 (post-nourishment) at the western (nourishment) and eastern (reference) sections of Plumb Beach

Full size image

With respect to total stages per core (Fig. 6), the differences between the means at the eastern and western sections were not significantly different before (2012) and after (2013–2015) the beach nourishment project (F 1,15 = 0.227, p = 0.641). There were significant effects in the ANCOVA model for sampling date (F 1,15 = 7.152, p = 0.017) and the quadratic term for date (F 1,15 = 8.888, p = 0.009). With respect to live eggs, the patterns were similar to total stages. Differences between sites before and after nourishment were non-significant (F 1,14 = 0.720, p = 0.410), and both sampling date (F 1,14 = 5.436, p = 0.035) and the quadratic term for date (F 1,14 = 8.052, p = 0.013) were significant. There was a small, but significant site effect for embryos (F 1,11 = 5.592, p = 0.037) and a non-significant site effect for larvae (F 1,10 = 3.447, p = 0.093).

Geomorphology and Sediment Composition

In 2012, prior to beach nourishment, the western region of Plumb Beach was severely eroded, especially on the central and eastern transects where a sand bag revetment had been constructed following a 2009 storm to curtail erosion that threatened the adjacent parking area, bicycle path, and highway (Figs. 2c and 3a). With access to the upper foreshore precluded, the existing areas of beach were quite narrow compared to the reference site (Table 1). We found that many parts of the foreshore at the western section were virtually impenetrable because of the amount of hard fill (e.g., pieces of brick, asphalt, cinderblock) just below the sediment surface. When we sampled patches of sand, we noted the presence of very coarse rubble or anaerobic sand at depths that could affect horseshoe crab egg laying (Table 1). The sediments at the western section of Plumb Beach in 2012 were significantly coarser (t 32 = 2.16, p < 0.04; Fig. 8a) and had a greater percentage of gravel-sized particles (> 4 mm) compared with the eastern section (t 32 = 2.51, p < 0.02; Fig. 8b). Sediments at the western section were more heterogeneous (poorly sorted) than the reference site (t 32 = 2.36, p < 0.03; Fig. 8c), but there was no significant difference in skewness (t 32 = 1.19, p > 0.25; Fig. 8d). There was a small, but statistically significant difference in the percent organic carbon between western (\( \overline{x} \) = 0.16%) and eastern (\( \overline{x} \) = 0.10%) sediments (t 32 = 2.27, 32 df, p = 0.03).

Table 1 Geomorphological characteristics of Plumb Beach before (2012) and after (2013–2015) the beach nourishment project
Full size table
Fig. 8
figure 8

Sediment characteristics at the western (nourishment) and eastern (reference) sections of Plumb Beach in 2012 (pre-nourishment) and 2013–2015 (post-nourishment) (means ± SE). a Mean grain diameter. b Percent gravel. c Sorting. d Skewness. e Compressive strength

Full size image

Beach nourishment at the western portion of Plumb Beach was completed in late October 2012, just prior to the landfall of Hurricane Sandy. The project created a sandy beach seaward of the prior foreshore (Figs. 2d, 4), with the majority of the sand deposited in the central and eastern region of the nourishment site. The nourished beach was wider but with similar slope to the pre-construction conditions (Table 1).

We observed several potentially important differences in sediment grain-size characteristics between the nourished and reference sites in 2013. Sediments at the nourished beach were significantly finer (t 34 = 2.77, p < 0.01; Fig. 8a) and had significantly less gravel than the reference beach (t 34 = 3.43, p < 0.005; Fig. 8b), which was the reverse of the pre-nourishment conditions. Sediments at the nourished site were significantly more homogenous than at the reference site (t 34 = 3.48, p < 0.002; Fig. 8c), and were fine-skewed rather than coarse-skewed (t 34 = 2.70, p < 0.02; Fig. 8d). As a consequence of these textural differences, sediments at the mid-tide area of the nourished beach were significantly harder than at the reference site (t 28 = 18.44, p < 0.0001; Fig. 8e). We also noted the presence of black (anaerobic) sand at depths of 5–8 cm in the middle to lower foreshore of the restored beach, whereas no anaerobic sand was encountered at the reference site in any of the 4 years (Table 1). In 2013, there was no significant difference in sediment organic carbon between western (\( \overline{x} \) = 0.55%) and eastern regions (\( \overline{x} \) = 0.53%; t 34 = 0.05, p > 0.47).

In 2014 and 2015, many of these sediment differences between sites remained, although some of the differences were not as large. Mean grain diameter remained finer (Fig. 8a) and percent gravel lower (Fig. 8a) at the nourishment site than at the reference site, but only the difference in percent gravel in 2015 was statistically significant (t 34 = 2.34, p < 0.025). Sand at the nourishment site remained more homogeneous (Fig. 8c) and fine skewed (Fig. 8d) compared to the reference site. Sediments at the nourishment site were significantly harder than the reference site in 2014, but the differences were no longer significant in 2015 (Fig. 8e) when all 45 determinations from the nourishment site were pooled. However, sediments at the nourishment site in 2015 were significantly (p < 0.0001) harder on the easternmost (\( \overline{x} \) = 0.54) and central (\( \overline{x} \) = 0.57) transects than at the westernmost transect (\( \overline{x} \)= 0.11).

We also noted some differences in sediment texture at the reference site between years. For example, in 2013, sediments at the reference site had a significantly higher percentage of gravel (t 34 = 5.26, p < 0.0001 and a larger mean grain diameter (t 34 = 7.64, p < 0.0001) than in 2012. However, it is unclear whether these changes were related to the beach nourishment project or to the effects of Hurricane Sandy. Percent gravel, sorting, skewness, and hardness did not show any clear temporal trends at the reference sites between 2012 and 2015.

Discussion

The primary rationale of the Plumb Beach nourishment project was to provide necessary protection for infrastructure, including a vital highway along the south shore of Brooklyn that was threatened by erosion (USACE 2012). Hurricane Sandy, one of the most destructive tropical storms to ever impact the middle Atlantic region, made landfall on 29 October 2012, and caused catastrophic damage to low-lying areas of Jamaica Bay. The just completed Plumb Beach nourishment project was widely credited by public officials with saving the highway (Castagna 2013). The timing of the Plumb Beach project was designed to avoid direct disturbances to spawning adult horseshoe crabs or their eggs. It began in Fall 2012 when there were few adults in the vicinity. There were virtually no horseshoe crab eggs, embryos, or larvae at the eroded western section of Plumb Beach by June and very few remaining at the eastern section (reference site) by late July (Fig. 6). Therefore, there was little chance of any adverse impacts on horseshoe crab eggs, embryos, or larvae from construction-related activities that began in Fall 2012. Similarly, the 2013 spawning season concluded before the continuation of the final stages of the project, including the construction of two terminal groins and a stone breakwater. It is unknown whether there were large numbers of horseshoe crabs entrained during the dredging process as has been seen in some USACE beach nourishment projects (Ray and Clarke 2010), but we saw no unusual accumulations of damaged or dead horseshoe crabs to suggest this was a problem.

The beach nourishment project at Plumb Beach appears to have had limited benefits to spawning horseshoe crabs during the first 3 years following the addition of sand. Before beach nourishment (2012), the highly eroded western section had very little horseshoe crab spawning activity and few eggs deposited in beach sediments in comparison to the eastern (reference) section (Figs. 5, 6). The absence of horseshoe crabs from the western section in 2012 was not surprising because of the sandbag revetment in the middle to upper foreshore. Armoring a shoreline frequently leads to disproportionately large effects on upper foreshore habitats (Dugan et al. 2011). The presence of coarse debris in the intertidal zone also limited the suitability of the habitat for egg-laying females.

Despite the fact that the nourished beach was wider and the rubble and sandbags had been buried, horseshoe crabs have continued to show a strong preference to spawn at the reference site in the 3 years following nourishment. When comparing densities of eggs and total stages (eggs, embryos, and larvae), the null hypothesis that the differences in densities between the nourishment and references sites would be unchanged following beach nourishment was accepted. The lack of long-term population data on horseshoe crabs at these sites does limit our interpretation of the results. In the present study, peak egg densities at the Plumb Beach reference site ranged between 130,000–239,000 eggs m−2, values that exceeded the peak densities at 4 of the 5 previously studied sites in Jamaica Bay (Botton et al. 2006). Hanna (2001) counted spawning horseshoe crabs during several full and new moon periods along a 1530 m stretch of Plumb Beach in 1998, and found that most of the animals were found in the region corresponding to our reference site. Her peak count was 241 females; assuming most of these females were found along the water’s edge, this would be equivalent to an ISA of 0.158, which is within the range that we estimated from 2012 to 2015. Our highest average count (combined males + females) at the reference site was 1271 animals (= 1495.km), in 2013. We caution that this would be a very conservative estimate of the spawning population because (1) the quadrats do not sample submerged animals spawning more than 1 m from the shoreline and (2) there may be multiple pulses of spawning activity during the May–June period that involve different individuals.

Whereas the ISA on the western section has increased following nourishment (Fig. 5), we have yet to see a parallel increase in egg deposition (Figs. 6 and 7). There are two possible explanations for this disconnect. First, animals counted during the spawning surveys include all visible mated pairs and satellite males at the shoreline. We speculate that some of these females in the western section may approach the beach, but refrain from laying eggs once they detect that the sediments are less than ideal. This hypothesis is consistent with the somewhat more male-biased operational sex ratios on the nourishment site in 2014 and 2015. Second, observers have noted that spawning activity on the western section is spatially patchy, with most of the mated pairs and single males concentrated to the west of the westernmost terminal groin, in a small section of beach where some limited accretion of sand is taking place (Fig. 2). Because this section of beach was not sampled for eggs, we may have missed the egg input that this area is making. But clearly, in the part of the nourishment site between the two groins, there are far fewer eggs deposited than at the reference site, and the differences between sites are similar before and after nourishment.

We hypothesize that differences in sediment quality may be associated with the low utilization of the nourished beach by spawning horseshoe crabs in the 3 years following beach nourishment. The nourished and reference beaches were more similar in superficial appearance in 2013 than before nourishment (compare Figs. 3 and 4), but the regions differed in mean grain size, percent gravel, sorting, and skewness (Fig. 8a–d). Sediments at the nourished site were finer, more uniform, and fine-skewed, but at the reference site, sediments were coarser, poorly sorted, and coarse-skewed. These qualities affect the compressive strength of the sediments, such that the nourished beach was significantly harder than the reference site (Fig. 8e). The fine, relatively homogenous sand at the nourished site may also limit diffusion of O2 into the interstices; dark gray, anaerobic sediments were found at the nourished site (especially in the lower and mid-tide areas) but never at the reference site. The presence of H2S, associated with low O2 conditions is known to deter female horseshoe crabs from depositing their eggs (Botton et al. 1988; Penn and Brockmann 1994; Vasquez et al. 2015) and low O2 significantly delays embryonic development (Funch et al. 2016). The nourishment site is somewhat more exposed to wave action than the reference site, especially during south to southwest winds, and this factor could also be important in explaining site differences.

The general experience, gained mainly from studies on high energy coastal environments, is that armored shorelines have limited ecological functionality in comparison to sandy beaches. Beaches tend to be narrower in the vicinity of seawalls and revetments, and armored beaches typically have diminished accumulation of wrack, reduced macrobenthic abundance and biomass in the upper foreshore, and support fewer birds (shorebirds and gulls) than unarmored beaches (Dugan et al. 2008, 2011). Consequently, beach nourishment is viewed as a preferred option to preserve and protect coastal environments, although comparatively, few studies have specifically evaluated the effects of shoreline armoring of estuarine and bay beaches (Nordstrom 2005). The biological effects resulting from beach nourishment projects include impacts to biota at the borrow site, physical disturbances to the fauna at the nourished beach caused by burial and/or crushing from heavy equipment, interference with larval recruitment, and mis-matches in sediment composition, beach slope, etc. between the pre- and post-nourished beaches (Peterson and Bishop 2005; Peterson et al. 2006; Wilber et al. 2009; Manning et al. 2014).

Horseshoe crabs require sandy estuarine beaches as spawning habitat (Botton et al. 1988; Smith et al. 2002b), and the loss of beach as a result of shoreline armoring is an important factor in the decline of horseshoe crab populations in southeast Asia (Botton 2001; Berkson et al. 2009; Hsieh and Chen 2009) and North America (Botton et al. 1988; Mattei et al. 2015). Jackson et al. (2015) found that the placement and arrangement of bulkheads along Delaware Bay beaches determined the degree of impact. Bulkheads located above the spring high tide line do not currently affect habitat suitability, but if bulkheads were low on the beach, there was a loss of suitable spawning habitat above the structure, and an increase in sediment activation below the bulkhead that could accelerate the exhumation of eggs. In some locations, small unarmored sections located between bulkheaded areas can be suitable spawning habitats, as well as sinks for eggs entrained by wave action (Jackson et al. 2015). However, even those bulkheads that are presently above the spring high tide will gradually come into play with rising sea levels, and beach nourishment may ultimately be needed to maintain these habitats and the upland property and infrastructure.

Beach nourishment projects for shore protection have taken place on several Delaware Bay beaches in the State of Delaware, including important horseshoe crab spawning beaches such as Pickering Beach, Kitts Hummock, Bowers Beach, and Slaughter Beach (PBS&J 2010). On nourished foreshores in Delaware Bay, sediments at 15 cm depth were finer and better sorted than on un-nourished sites (Jackson et al. 2005b), a finding that is consistent with our results from Plumb Beach. Suitable levels of dissolved oxygen are important for Limulus egg viability and development (Penn and Brockmann 1994; Funch et al. 2016) and sediments with > 10% fine sand have sharply lower levels of dissolved oxygen than coarser sediments (Brafield 1964). Coarse sand and gravel was added to experimental plots at Bowers Beach, Delaware Bay, to test the hypothesis that this augmentation would lead to an increase in spawning activity and egg developmental success (Jackson et al. 2007). It was concluded that the quantity of deposited material was insufficient to cause significant changes in mean grain size or sorting, suggesting that the initial selection of the source sediment in a nourishment project is critical. By transplanting eggs in pouches and placing them within the beach, Jackson et al. (2007) found similar percentages of egg development comparing nourished and reference sites. Avissar’s modeling study (Avissar 2006) predicted that nourished beaches in Delaware Bay would be less suitable than natural habitats for horseshoe crabs because they would tend to be finer with less dissolved oxygen, and hotter (darker sand color). In our study, anaerobic sediments were not found at the reference site, but were seen in more than half the samples from the nourished beach (Table 1) and this factor could have influenced nest-site selection (Botton et al. 1988).

Sea level rise will continue to threaten sandy beach ecosystems for the foreseeable future (FitzGerald et al. 2008; Schlacher et al. 2008; Williams and Gutierrez 2009; Defeo et al. 2009; Nicholls and Cazenave 2010). It is a particular concern in urban areas such as New York City, where retreat from the coastal zone is not a practical option, and shorelines will need to be maintained by armoring, periodic beach nourishment, or other engineered solutions (Rosenzweig et al. 2011; Solecki 2012). Viable coastal habitats, including beaches, salt marshes, and mangroves are important in protecting coastal properties against sea level rise and storm damage (Arkema et al. 2013) and they have greater ecological value compared with armored shorelines (Dugan et al. 2011). A recent review (Gittman et al. 2015) predicted that as much as 1/3 of the coastline of the USA may be armored by 2100. In an era of rising sea levels, the prioritization of habitats for beach nourishment should consider both the value of the protected property as well as the potential benefits to fauna such as horseshoe crabs and sea turtles (e.g., Katselidis et al. 2013), both of which are dependent on sandy beaches for reproduction.

Although the Plumb Beach project did not result in a measurable increase in the density of horseshoe crab eggs over the course of this study, the differences in sediment texture were greatest in the first year following nourishment (2013) and have become smaller over time (Fig. 8). Should this trend continue, the nourished beach may eventually become more suitable for horseshoe crabs. Beach-dependent animals do not necessarily respond positively to the new habitat immediately, as it may take time for the geomorphology to equilibrate to the environmental conditions. For example, loggerhead sea turtles (Caretta caretta) experienced decreased nesting success on nourished beaches in Florida, but the effects lasted for only one season before recovering (Rumbold et al. 2001; Brock et al. 2009).

The future conservation of horseshoe crab populations depends on sound fishery management (Atlantic States Marine Fisheries Commission 1998; Smith et al. 2009, 2017; Millard et al. 2015) and the continued availability of critical spawning and juvenile nursery habitats (Botton 2001; Berkson et al. 2009; Hsieh and Chen 2009). The results of our study suggest that horseshoe crabs may experience only limited short-term benefits of beach nourishment, although in our view, this is a preferable alternative to the use of bulkheads or revetments. The likelihood of successful restoration of horseshoe crab spawning and nursery grounds will be enhanced with a better understanding of site-specific hydrodynamics and geomorphology (Hsieh and Chen 2009).