Estuaries and Coasts

, Volume 37, Supplement 1, pp 20–30

Impacts of Varying Estuarine Temperature and Light Conditions on Zostera marina (Eelgrass) and its Interactions With Ruppia maritima (Widgeongrass)

Authors

    • Virginia Institute of Marine ScienceCollege of William and Mary
  • Erin C. Shields
    • Virginia Institute of Marine ScienceCollege of William and Mary
  • David B. Parrish
    • Virginia Institute of Marine ScienceCollege of William and Mary
Article

DOI: 10.1007/s12237-013-9667-3

Cite this article as:
Moore, K.A., Shields, E.C. & Parrish, D.B. Estuaries and Coasts (2014) 37: 20. doi:10.1007/s12237-013-9667-3

Abstract

Seagrass populations have been declining globally, with changes attributed to anthropogenic stresses and, more recently, negative effects of global climate change. We examined the distribution of Zostera marina (eelgrass) dominated beds in the York River, Chesapeake Bay, VA over an 8-year time period. Using a temperature-dependent light model, declines in upriver areas were associated with higher light attenuation, resulting in lower light availability relative to compensating light requirements of Z. marina compared with downriver areas. An inverse relationship was observed between SAV growth and temperature with a change between net bed cover increases and decreases for the period of 2004–2011 observed at approximately 23 °C. Z. marina-dominated beds in the lower river have been recovering from a die-off event in 2005 and experienced another near complete decline in 2010, losing an average of 97 % of coverage of Z. marina from June to October. These 2010 declines were attributed to an early summer heat event in which daily mean water temperatures increased from 25 to 30 °C over a 2-week time period, considerably higher than previous years when complete die-offs were not observed. Z. marina recovery from this event was minimal, while Ruppia maritima (widgeongrass) expanded its abundance. Water temperatures are projected to continue to increase in the Chesapeake Bay and elsewhere. These results suggest that short-term exposures to rapidly increasing temperatures by 4–5 °C above normal during summer months can result in widespread diebacks that may lead to Z. marina extirpation from historically vegetated areas, with the potential replacement by other species.

Keywords

SeagrassTurbidityTemperatureEstuaryLightChesapeake BayClimate change

Introduction

Zostera marina (eelgrass) is an ecologically important seagrass species with worldwide temperate distribution ranging from North America and Europe to the coastal waters of Japan and Korea (Green and Short 2003). It is the dominant seagrass found in coastal and estuarine areas of the western North Atlantic including the Chesapeake Bay, extending from Quebec (Canada) at approximately 60° N to North Carolina (United States) at 35° N (Moore and Short 2006). The patterns of Z. marina seagrass abundance in the Chesapeake Bay over the past 80 years can be characterized by periods of sharp decline followed by periods of partial recovery (Orth and Moore 1983). These declines have contributed to the global pattern of decline for seagrasses which have been attributed to a combination of climatic and anthropogenic stresses including increasing turbidity and unusually high summer water temperatures (Orth and Moore 1983; Orth et al. 2006, 2010; Moore and Jarvis 2008; Waycott et al. 2009).

The effects of climate change, in particular the resulting increase in water temperature, on Z. marina populations might be especially evident in areas such as the Chesapeake Bay that are near the southern limits of its distribution along the western North Atlantic (Koch and Orth 2003). Records indicate that the Chesapeake Bay has been warming significantly since the mid-twentieth century (Preston 2004) and model predictions of increased rainfall patterns with concomitant increased sediment and nutrient inputs (Najjar et al. 2000; Polsky et al. 2000; Neff et al. 2000) will compound the stresses on this seagrass species.

High temperatures can be particularly problematic for Z. marina growth and survival. Nejrup and Pedersen (2008) found mortality to increase 12-fold in this temperate species with water temperatures between 25 and 30 °C compared with 10 to 20 °C. High temperatures can also interact with turbidity resulting in a compounding negative effect (Moore et al. 2012). Light and temperature effects interact due to an increase in Z. marina light requirements with increasing temperature which is attributed to increased respiration rates, affecting the plant’s carbon balance (Marsh et al. 1986; Moore et al. 1997; Staehr and Borum 2011). In addition, factors such as elevated sediment sulfide and low water column oxygen (Homer and Bondgaard 2001; Greve et al. 2003; Plus et al. 2003) which are related to high temperatures can also affect plant performance.

Short-term periods of stress (days to weeks) due to extreme climatic events such as storm-induced turbidity or above-average water temperatures may result in significant long-term changes in Z. marina populations. Negative impacts on Z. marina due to short-term events have been documented worldwide. Elevated turbidity for a period of 20 days was found to be the cause for mortality of transplanted Z. marina in the Chesapeake Bay (Moore et al. 1997). Reusch et al. (2005) found immediate negative effects of a summer heat wave in Germany on survival of Z. marina in their field experiment. A simulated 3-week heat wave on Z. marina collected from Italy and Denmark caused declines in both maximum electron transfer rate and effective quantum yield (Winters et al. 2011). Due to the acute nature of these extreme events, Z. marina populations may not be allowed enough time for adaptation (Zimmerman et al. 1989) resulting in significant population declines (Plus et al. 2003; Greve et al. 2003).

Z. marina in the Chesapeake Bay typically co-occurs with Ruppia maritima (widgeongrass) in a zonation pattern with R. maritima dominating in the nearshore shallows, both species growing at mid-depths, and Z. marina occupying the deeper depths where R. maritima is not found (Orth and Moore 1988). Globally, R. maritima is one of the most widely distributed seagrass species, growing in both tropical and temperate zones in a wide variety of high and low salinity habitats (Short et al. 2007). It is generally known to have greater temperature tolerances than Z. marina (Evans et al. 1986), and therefore potentially may replace Z. marina under some conditions, yet evidence for this potential change has not been well documented.

The objectives of this study were to investigate the interactive effects of turbidity and temperature on inter-annual Z. marina dominated bed persistence along a gradient of estuarine sites with increasing turbidity (Moore et al. 1997; Moore and Jarvis 2008). We hypothesized that the greatest seasonal Z. marina die-offs could be related to exposure above usual temperatures compounded by high turbidities, and that R. maritima abundance increased following these die-offs. Specifically, we quantify the patterns of short-term (monthly) and inter-annual variability of Z. marina survival (vegetative cover), and relate these trends to water temperature and light availability over 8 years of varying climatic conditions. We also compare temperature-dependent Z. marina light requirements to direct measurements of light availability and relate this to plant persistence. Finally, we quantify all changes in Z. marina and R. maritima distribution along depth gradients since 2010 and assess the potential future impacts of these changes to highlight the timing and pattern of replacement of one seagrass species by another.

Methods

Sampling Sites

Three sampling sites were chosen for this study: Goodwin Island (GI), Gloucester Point (GP), and Claybank (CB) (Fig. 1). All are located in the York River Estuary, a tributary of the Chesapeake Bay. GI (37°13′1″ N, 76°23′19″ W), the most downriver site, has supported generally persistent seagrass beds for at least the last 75 years (Orth and Moore 1983; Moore et al. 2000; Orth et al. 2010). GP (37°14′50″ N, 76°29′58″ W) is located approximately 10 km upriver from GI and represents the recent upriver limits of seagrass growth in this system (Moore et al. 1997; Orth et al. 2010). CB (37°20′55″ N, 76°36′40″ W) is currently unvegetated, but has historically (pre-1970s) represented the furthest upriver persistent seagrass area (Moore et al. 1997; Orth et al. 2010).
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig1_HTML.gif
Fig. 1

York River, Chesapeake Bay, VA, showing the location of the three sites: Claybank (CB), Gloucester Point (GP), and Goodwin Island (GI). The three transects at GP and GI run perpendicular to shore and are marked by black lines. Dark shaded areas indicate 2010 seagrass coverage

Biological Sampling

Sampling was conducted by divers along fixed transects located perpendicular to the shoreline and extending just past the last observable shoot. Three permanent transects were set up in 2004 at GI, and three at GP, as part of the National Estuarine Research Reserve System-Wide Biological Monitoring Program (National Estuarine Research Reserve System 2012). No transects were established at CB because it was unvegetated throughout the study period. Annual aerial photography (e.g., Orth et al. 2011) was checked yearly along with periodic field checks to verify the continued absence of seagrass vegetation at CB throughout the study period.

From 2004 to 2011, vegetative cover estimates and shoot lengths were assessed monthly from April to October along all transects, at intervals determined by transect length (Moore 2009). The longest transect (GI3, 700 m) was sampled for depth at every 10 m and for vegetative cover at every 20 m. All other transects (GI1, 130 m; GI2, 300 m; GP1–3, 90 m) were sampled for depth at 5 m and vegetative cover at every 10 m. A 0.5-m × 0.5-m quadrat was haphazardly tossed three times along every sampling location, and Z. marina cover was estimated visually. Starting in 2010, R. maritima coverage was also monitored. Depth data was normalized to mean lower low water (MLLW) using tide data from the U.S. NOAA, National Ocean Survey tide gauge at the U.S. Coast Guard Training Center in Yorktown, VA (37°13′36″ N, 76°28′42″ W). This location is approximately at a midpoint distance between GI and GP.

Water Quality Sampling

Water quality measurements were made at all three of the study sites from 2004 to 2010 using YSI 6600 EDS V2 multi-parameter sondes which were fixed to pilings and located between 0.25 and 0.51 m above the sediment at each site. Water quality was measured at 15-min intervals, including turbidity (in NTU), chlorophyll fluorescence, temperature, salinity, pH, dissolved oxygen, and depth. These data were managed through the Virginia Estuarine and Coastal Observing System (VECOS) database and website (Virginia Estuarine and Coastal Observing System 2012). Diffuse downwelling attenuation of photosynthetically available radiation (PAR) was determined bi-weekly adjacent to the datasondes using triplicate water column measurements of downwelling photosynthetic photon flux density measured with a LI-192SA underwater quantum sensor (LI-COR) taken every 25 cm from 10 cm below the surface to 25 cm above the bottom. YSI turbidity measurements were modeled to light attenuation by least squares regression relating turbidity to downwelling light attenuation coefficients (Kd) using the simultaneously measured light profiles and turbidities taken throughout the course of the studies. Downwelling insolation measurements of PAR were continuously recorded using a LI-190SA terrestrial quantum sensor (LI-COR) at 15-min intervals. The measurements were made at Taskinas Creek (37°24′55″ N, 76°42′53″ W), which is a Chesapeake Bay National Estuarine Research Reserve in Virginia meteorological monitoring site located along the York River approximately 12 km northwest of CB.

Compensating Light (Ic) and Light to the Bottom (Iz) Calculations

Community compensating light estimates for Z. marina at all study sites were determined using a least squares regression model that was developed for this mid-Atlantic area (Wetzel and Penhale 1983; Moore et al. 1997, 2012) that relates in situ measures of Z. marina light compensation to water temperature [log(Ic) = 0.057 × temp + 1.01; r2 = 0.99]. Water temperatures were recorded every 15 min using the YSI sondes at each site.

The percent light through water to the bottom (PLW) was calculated using Eq. (1), where Kd is modeled from the 15-min turbidity values and z is the 15-min depth values taken at each site and normalized to a common sea level datum:
$$ \mathrm{PLW}=100\times {e}^{\left({K}_{\mathrm{d}}\times z\right)} $$
(1)
PLW was then used to calculate the light to the bottom (Iz) using Eq. (2), where PAR was the 15-min reading taken from the Taskinas Creek meteorological station:
$$ {I}_z=\left(\mathrm{PAR}\times \mathrm{PLW}\right)/100 $$
(2)

Statistical Analyses

Comparisons of mean frequency distributions of water temperatures, light attenuation coefficients, and proportion of light requirements met by light availability were accomplished using Kolmogorov–Smirnov test for significance (StatView, Inc.).

A direct relationship between Z. marina abundance and water temperature was explored by calculating the mean water temperature between each sampling period, and comparing that with the relative percent change in vegetative cover during that period. A linear regression model (R Core Team 2012) was used to explain this relationship. Comparisons among the Z. marina die-off years of 2005 and 2010 and non-die-off years for June–August at GI were done using Kruskal–Wallis rank sum tests. Paired post hoc comparisons followed procedures originally published in Siegel and Castellan (1988).

Results

Z. marina Abundance

Z. marina showed seasonal patterns in integrated vegetative cover during all years, typically increasing in the spring before peaking in the early summer and subsequently declining into the fall (Fig. 2). For the two vegetated lower sites, cover was greatest during May–July 2004, when monitoring began, with coverage peaking around 45 % for GI (Fig. 2a) and 60 % for GP (Fig. 2b). Both sites also experienced large-scale declines, where virtually no above-ground biomass was present by October 2005. Z. marina at GI were able to recover from this event, increasing their coverage to 25 % by July of 2006, peaking in June of 2009 at 30 %. Z. marina beds at GP reached 12 % by June of 2006, but again showed a similar die-off to 2005 with no above-ground biomass present by the end of the year. After these 2 years of complete loss of vegetation at GP, the beds did not recover and remained unvegetated through 2011. Another large-scale decline was observed in 2010 at GI, where a complete loss of vegetation occurred from June to July. Partial recovery occurred in 2011, but by September and October of that year, coverage was at 2 %. These die-offs were not only reflected in the vegetative cover data but in the mean length measurements as well. Lowest mean shoot lengths were recorded in 2006 (22 cm) and 2011 (17 cm), the years that Z. marina was recovering from the die-offs.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig2_HTML.gif
Fig. 2

Integrative mean (±SE) Z. marina vegetative cover across all three transects at each site for all sampling dates. a GI and b GP

Cover of Z. marina beds at both GI and GP was concentrated at depths between 40 and 75 cm MLLW (Fig. 3). Throughout the summer, it was typical for plants to disappear towards the deep edge of the bed; however, during the 2005 and 2010 die-offs, Z. marina declined at all depths, with no plants remaining by the end of the growing season, even within the historically optimum depth range. During 2010, greatest declines occurred between June and July at depths between 40 and 55 cm MLLW (Fig. 3a, b). During more typical non-die-off years, declines at these mid-depths did not appear until August (Fig. 3c). Between June and July of 2010, maximum Z. marina losses occurred between 40 and 45 cm MLLW, losing 84 % of its coverage. In contrast, during non-die-off years, Z. marina at this depth range actually increased their coverage by 2 %. By August of 2010, the maximum Z. marina coverage was 5 % at the 50–55-cm depth range, while the maximum for non-die-off years was 24 % at the 40–50-cm depth range.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig3_HTML.gif
Fig. 3

Mean Z. marina vegetative cover (±SE) across all three transects at GI. Data are separated into 5-cm depth bins and reported as depth at mean lower low water (MLLW). Black circles represent non die-off years (2004, 2006–2009), and gray circles represent the 2010 die-off year. a June, b July, c August

R. maritima/Z. marina Interactions

R. maritima had not been reported present at GP or CB since at least the early 1970s (Marsh 1973, 1976; Orth et al. 2010), but co-occurs with Z. marina at all sites at GI. In 2010, R. maritima was very sparse across all three GI transects. Results from the longest (700 m) transect are reported here, as it contains the most persistent Z. marina and R. maritima beds, and is therefore best suited to examining species interactions. In the early growing season, R. maritima only dominated at the most nearshore sampling site along the transect, reaching 77 % cover in June in inshore areas (Fig. 4). After June when the Z. marina die-off occurred, R. maritima declined as well, with no plants left by September. When recovery from this event began in April of 2011, both Z. marina and R. maritima were fairly sparse with neither species above 20 % cover. In May, Z. marina began to increase in coverage, with R. maritima present but sparse, mostly growing within 300 m from shore. In June, a shift in dominant species started to occur, with Z. marina never occurring at >25 % cover, while R. maritima peaked at 70 % cover approximately 180 m from shore. R. maritima dominance continued throughout the growing season, with greatest coverage in August 2011. During this month, coverage was never below 40 % in the first 240 m from shore, while Z. marina was never above 1 % coverage in that same area. Similar to 2010, Z. marina was essentially absent from this site by September and October; however, unlike the previous year, R. maritima was still present with maximum October coverage at 33 % 300 m from shore. From 2010 to 2011, R. maritima both increased its coverage and expanded further out from shore (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig4_HTML.gif
Fig. 4

Vegetative cover of Z. marina (black bars) and R. maritima (gray bars) at every distance sampled along the GI-3 transect; 2010 (left panel) and 2011 (right panel) are represented. The months included are April (a, b), June (c, d), August (e, f), and October (g, h)

Water Quality and Physical Conditions

All sites exhibited significantly different frequency distributions (p < 0.05) for temperature, Kd, and Ic/Iz analyzed for June–August 2004–2010. The cumulative distribution of sites began to separate as temperatures reached 25 °C, with CB spending a greater percentage of time at temperatures above this level, followed by GI and GP (Fig. 5a). This trend continued until temperatures reached above 29 °C, at which points the sites did not show differences. Kd differences in cumulative frequency distributions among sites were much larger, following a trend of decreasing Kd downriver (Fig. 5b). GI only spent 22 % of the time with Kd values above 1.5 m−1, while GP and CB spent 53 and 78 %, respectively, above this level. A Kd of 1.5 m−1 in this region represents approximately 22 % of solar insolation reaching the bottom at a water depth of 1.0 m, which is a seagrass-based water clarity management goal for the Chesapeake Bay (Dennison et al. 1993). When Kd and temperature were combined to determine the frequency distribution of percentage light available relative to compensating light requirements (Iz/Ic), a significant gradient was found across the three sites (Fig. 5c). The most upriver site, CB, spent a greater percentage of time with lower available light compared to light requirements, while the most downriver site, GI, spent the least amount of time.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig5_HTML.gif
Fig. 5

Cumulative frequency distributions for June through August 2004–2010 showing the spatial gradient among the study sites along the York River for water temperature (a), light attenuation (Kd) (b), and percent light available (Iz) of compensating light requirements (Ic) (c). Graphs present mean binned frequencies of all data taken from 15-min measurements for each parameter and site

Daily mean water temperature across all sites showed seasonal differences among years when Z. marina die-offs did not occur (2004; 2006–2009) compared with 2005 and 2010. In 2005, the warming event occurred in late July through mid-August. During a week-long period in August, daily mean temperatures were approximately 4 °C warmer in 2005 (30 °C) compared with non-die-off years (26 °C) (Fig. 6). In contrast, the significant warming event for 2010 occurred from mid- to late June, where 2010 had consistently higher temperatures than non-die-off years (Fig. 6). Non-die-off years showed a very gradual increase in water temperature over this time period, increasing from 25 to 26 °C over the 2-week period. Water temperatures in 2010, on the other hand, increased rapidly from 25 °C in mid-June to 30 °C in late June. Moreover, 2010 temperatures remained higher during July before decreasing in early August.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig6_HTML.gif
Fig. 6

June to September daily mean water temperatures. Non-die off years (2004, 2006–2009) are represented by black circles, 2005 is represented by gray circles with a solid line, and 2010 is represented by gray circles with a dashed line. Mean daily SE for non-die off years is 0.03 °C, 2005 is 0.05 °C, and 2010 is 0.05 °C

Median water quality conditions for June–August at Goodwin Island showed higher water temperatures during 2005 and 2010 compared with non-die-off years (Fig. 7) (p < 0.05). Although significantly different, median light attenuation was very similar between the die-off years (2005, 1.7 m−1; 2010, 1.6 m−1) and the non-die-off years (1.7 m−1). These differences, particularly in temperature, combined so the percentage light available relative to light requirements was lower in both 2005 and 2010 than the other years (Fig. 7) (p < 0.05). During non-die-off years, the median light available was 35 % of calculated compensating light requirements, while this was only 28 % and 21 % in 2005 and 2010, respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig7_HTML.gif
Fig. 7

Water temperature, light attenuation (Kd), and percent light available (Iz) of compensating light requirements (Ic) for June through August at GI. Data are separated into non-die-off years (NDY; 2004, 2006–2009), 2005, and 2010. Represented here are the median, 25th, and 75th percentiles, and the minimum and maximum of the lower 99 % of the data taken from 15-min measurements. Different letters indicate significant differences (p < 0.05) among the years

Figure 8 shows a significant relationship (p < 0.0001) relating changes in Z. marina vegetative cover between each April–October sampling date for the 2004–2011 period, and the average water temperature during each of those sampling intervals. The overall relationship highlights the negative effect of higher water temperatures on growth with the inflection point between net increases and decreases in vegetative cover during the April–October sampling periods occurring at approximately 23.0 °C.
https://static-content.springer.com/image/art%3A10.1007%2Fs12237-013-9667-3/MediaObjects/12237_2013_9667_Fig8_HTML.gif
Fig. 8

Regression analysis showing the significant (p < 0.0001) relationship linking the mean water temperatures during each SAV sampling interval and the relative change in SAV cover observed during each interval. Data presented are for the April–October sampling periods for years 2004 through 2011

Discussion

This study was able to track monthly changes in Z. marina abundance over 8 years, relate geographical differences to water clarity gradients, and yearly differences to episodic events of above typical water temperatures. The results suggest that short-term exposures to rapidly increasing temperatures by 4–5 °C above normal during summer months can result in widespread diebacks that may lead to species extirpation from historically vegetated regions, with the potential replacement by other species. Specific rates of change in Z. marina cover were found to be negatively related to temperature at temperatures above 18 °C, with net increases observed up to approximately 23 °C and net decreases at temperatures above that inflection point. This along with our other measurements suggests time spent under conditions of higher temperatures, even for short periods, can have a direct effect on its abundance during the spring and summer.

Z. marina in this area of the Chesapeake Bay is also following a trajectory of spatial decline from upriver to downriver, due in part to decreased water clarity in upriver regions (Moore et al. 1997). At the most downriver sites in the York River system where this species still remains, periodic large-scale declines have resulted in chronically low Z. marina coverage, while R. maritima has increased its range and coverage, dominating in formerly Z. marina-dominated areas (Orth et al. 2011 and this study). The most upriver site studied here, CB, has been unvegetated with no recovery for approximately the past 40 years. This complete loss of vegetation in the early 1970s was attributed to diminished water quality and reduced light availability (Orth and Moore 1983). This pattern of loss has now spread downriver to GP. No recovery is occurring in this area, even when light availability is at levels similar to those when Z. marina was growing. Studies have suggested that recovery of Z. marina beds does not occur unless water quality actually improves to levels greater than those required for survival of existing beds (Moore 2004). Therefore, chronically reduced water clarity combined with diebacks induced by periods of thermal stress can result in seagrass loss with limited capacity for regrowth.

Goodwin Island Z. marina dominated beds were showing signs of recovery after the complete loss of above ground Z. marina vegetation in 2005 (Jarvis and Moore 2010). However, another near complete decline occurred in 2010, after a period of above-average water temperatures from mid- to late June of that year. Daily mean water temperatures during this time period showed a difference of 5 °C when compared with typical temperatures during years where Z. marina did not experience large-scale declines. This is a significant difference in temperature for this species, particularly at the range measured (25–30 °C), as Z. marina has been found to have an upper limit temperature range for growth in this Chesapeake Bay region of 28 to 30 °C (Evans et al. 1986; Orth and Moore 1986; Abe et al. 2008). High temperatures can also have indirect effects on plant growth and survival through decreased water column and sediment pO2 and their effects on growth and shoot mortality (Raum and Borum 2013) as well as reductions in inorganic carbon availability (Zimmerman et al. 1997). Our results provide field evidence for short-term stressful events causing long-term negative impacts, as both the 2005 and 2010 declines occurred after only a 1- to 2-week time period of daily mean temperatures greater than 28 °C. Further, the direct relationship between increasing water temperatures and net Z. marina cover observed here (Fig. 8) suggests that a 5 °C increase in mean water temperatures extending over the period of a few weeks to a month can reduce bed growth by nearly 50 %, even at temperatures well below the upper temperature limits.

While temperatures are stressful for Z. marina growing in this Chesapeake Bay region where there is a net decline in cover beginning at 23 °C, it may be that there are slight photophysiological adaptations among Z. marina populations growing in other regions where the upper water temperatures are different; this upper threshold may be higher or lower than observed here. Short-term thermal increases of 4–5 °C during the summer may similarly affect these different populations even though the actual temperature regimes may be different. Winters et al. (2011)) followed the responses of Z. marina from three different Adriatic populations to short term “summer heat waves” where temperatures were increased from 19 to 25 °C. They observed photosynthetic impairments that were most pronounced with poorest recovery in the higher latitude plants. Similarly, Bergmann et al. (2010) found reduced gene expression and reductions in plant growth and fitness after similar thermal stresses, with again poorest recovery in the more northern sourced plants. While similar studies have not been done with Chesapeake Bay acclimated plants, water temperatures of 19–25 °C are typically observed here during periods of maximum Z. marina growth and biomass in the late spring (Moore et al. 1996), and significant impairments would not normally be expected at temperatures of 25 °C. These preliminary comparisons suggest that it may be that rapidly increasing short-term summertime thermal stresses above the normal or average for a particular region, rather than the absolute temperature condition, can have similar negative effects on Z. marina populations worldwide.

The timing of the period of above average water temperatures in June 2010 that led to the die-off is of particular interest because of the seasonal pattern of storage of carbohydrate reserves of Z. marina growing in this region. Total non-structural carbohydrate concentrations peak in the spring/early summer in both leaves and rhizomes. This helps the plants survive the stressful summer by relying on these stored reserves (Burke et al. 1996). In 2010, Z. marina was exposed to stressful conditions during this critical time where it needed to store reserves for survival throughout the rest of the summer. This is a possible explanation for the complete loss of aboveground biomass by the end of the growing season and why a spring/summer stressful event may have a similar impact on Z. marina in this region compared to a later summer event. Both types of events may cause the plants to deplete stored carbon reserves needed to make it through the entire stressful period. The 2005 die-off event occurred after a particularly warm period in August (Moore and Jarvis 2008; Jarvis and Moore 2010), and while there was a complete die-off by October of that year, the plants gradually recovered over the next few years. Recovery from the 2010 die-off is still being monitored, but 2011 coverage was less than 2006 coverage, and by the end of 2011, another near-complete die-off was observed.

Two consecutive years of decline is particularly problematic for Z. marina growing in this region because seedlings do not typically flower until their second year (Orth and Moore 1986). After the 2005 decline, the Z. marina beds recovered initially via sexual reproduction, depleting the seed bank (Jarvis and Moore 2010). The upriver GP site experienced a second year of declines in 2006. A depleted seed bank from the initial recovery in 2006 combined with a lack of flowering shoots in the spring resulted in poor seedling establishment to initiate the recovery in 2007 (Jarvis and Moore 2010). With the downriver GI site experiencing two consecutive declines in 2010 and 2011, it is possible that these beds will follow a similar path as those upriver.

It has been well documented that Z. marina beds contain greater diversity and density of invertebrates and fish than adjacent unvegetated habitats (see review in Orth and Moore 1984). Lesser known, however, is the habitat value of a declining Z. marina bed and how it functions in this capacity under chronic low-density as opposed to a high-density meadow. Reed and Hovel (2006) found that epifaunal diversity and density in San Diego, CA was not altered until a threshold of Z. marina loss greater than 50 % was reached. Boström et al. (2002) found a threefold increase in Macoma balthica (Baltic clam) in dense Z. marina beds (>150 shoots m−2) in the Baltic Sea compared with sparse beds (<20 shoots m−2). A study in several New England estuaries found that, in general, there was greater diversity and abundance of fish in high-complexity Z. marina habitats (>100 shoots m−2) compared with low-complexity habitats (<100 shoots m−2) (Hughes et al. 2002). In 2011, integrated mean density measurements for this system ranged from a minimum of 36 shoots m−2 in September to a maximum of 318 shoots m−2 in May (Moore unpublished data). More detailed studies are needed on specific thresholds in which Z. marina loss in the Chesapeake Bay results in decreased habitat functionality; however, it is possible that the declines that have already been seen have altered the faunal communities, resulting in decreased diversity and abundance.

Habitat functionality and other ecosystem services may not only be altered by decreased Z. marina abundance but also by potential replacement of this species by R. maritima. With water temperatures predicted to continue to increase in the Chesapeake Bay, R. maritima may continue on a trajectory shown here over the last several years. Because of its greater tolerance to high temperatures, it is possible that this species will continue to increase in abundance and range, occupying areas that were previously monospecific Z. marina beds. Based on their studies of photosynthetic capacities over a range of temperatures, Evans et al. (1986) suggested that at high temperatures, R. maritima has a competitive advantage over Z. marina. Similarly, Lazar and Dawes (1991) found broad tolerances of R. maritima when photosynthetic responses were measured over three different temperatures (10, 20, and 30 °C).

Previous studies have provided evidence for the replacement of Z. marina by co-occurring marginal species in response to warmer water temperatures. Johnson et al. (2003) found that Z. marina was replaced by R. maritima in San Diego, CA during an ENSO event when water temperatures were above average. Micheli et al. (2008) reported long-term declines of Z. marina in North Carolina in conjunction with warming temperatures while Halodule wrightii (shoalgrass), a co-occurring species similar in structure to R. maritima, showed no significant changes. They experimentally found evidence for the potential future replacement of Z. marina by H. wrightii if warming continues.

Results of the example studied here suggest that replacement of one seagrass species by another is not straightforward and many factors may affect the capacity for replacement. Expansion of R. maritima as observed here at GP appeared to follow decreased abundance of Z. marina in protected, inshore, shallow areas. In highly wave-exposed shorelines and in areas of greater depths, this expansion of R. maritima may not be possible (Orth and Moore 1988). Replacement by R. maritima has not occurred in the upriver CB region studied here. This suggests that other conditions such as high turbidity may be limiting its success, recruitment may be limited, or other conditions such as physical exposure (Koch 2001) at very shallow depths may be limiting its growth there.

Aerial mapping of seagrass abundances throughout the Chesapeake Bay combined by ground survey observations by Orth et al. (2010) suggest that this pattern of R. maritima expansion after Z. marina decline in nearshore areas has been relatively widespread. The rapid replacement here of Z. marina by R. maritima by mid-summer (August) after only 1 year was dramatic and suggests that, given appropriate conditions, expansion of one species after the decline of another can occur quickly. However, this replacement theory is based on only two complete years of R. maritima data, so continued monitoring is necessary to see if this trend continues.

While this study was conducted in a relatively small system in the Chesapeake Bay, the results provide important information for the response of Z. marina in general to a changing habitat and, more specifically, to stressful episodic events that will likely become more prevalent in many areas of the world. Declining water clarity has limited Z. marina to a narrow downriver section in this system, where light availability is slightly higher than upriver and upbay regions. However, the higher clarity is not enough to offset the negative impacts of episodic summer temperature heat events. Continued climate-driven heat stress without an improvement in water clarity or inorganic carbon availability (Zimmerman et al. 1997; Palacios and Zommerman 2007) will continue to stress these types of populations that are most susceptible to impacts from climate change, possibly eliminating them in some areas. This study provides evidence that a more stress tolerant co-occurring secondary species may be released from competition and allowed to expand, though a complete replacement of ecosystem function is unlikely, and is an important area of future research.

Acknowledgments

We gratefully acknowledge field and laboratory assistance by B. Neikirk, J. Goins, S. Snyder, V. Hogge, A. Miller, R. Wright, B. Haywood, D. Tulipani, A. Knowles, and J. Austin. Sincere thanks to J. Jarvis and R. Orth for in-depth discussions and review of the manuscript. Funds for this project were provided by the Estuarine Research Reserve Division of the National Oceanic and Atmospheric Administration and the Commonwealth of Virginia. This is contribution no. 3300 from the Virginia Institute of Marine Science, School of Marine Science, College of William and Mary.

Copyright information

© Coastal and Estuarine Research Federation 2013