Introduction

Scleractinian corals are conspicuously rare in shallow waters outside of the tropics. This may have been true throughout their evolutionary history (Stanley 2003), and there is evidence that the distribution of shallow-water corals is deeply rooted in the obligate endosymbiotic relationships many corals form with dinoflagellates (Symbiodinium spp.) that provide photosynthetic energy and enhance calcification (Stanley and Swart 1995; Davy et al. 2012). Temperature, light, and seawater aragonite saturation state—critical factors for photosynthesis and calcification by hermatypic corals—co-vary and decline with increasing latitude, limiting coral growth and reef formation (Kleypas et al. 1999). In contrast, macroalgae become increasingly abundant in shallow waters at higher latitudes due to increased nutrient concentrations and reduced grazing activity (Johannes et al. 1983; Miller and Hay 1996; Miller 1998). Competition with macroalgae for light and space within the photic zone is thus a further impediment to coral success in temperate regions (Miller 1995, 1998; Miller and Hay 1996).

A few hardy, ahermatypic shallow-water scleractinians exist well beyond the latitudinal limits of hermatypic corals (Schuhmacher and Zibrowius 1985). These temperate species tend to engage in facultative symbioses with Symbiodinium spp., and under sufficiently high light and temperature, symbiosis can considerably enhance their growth (Jacques et al. 1983; Miller 1995; Dimond and Carrington 2007; Hoogenboom et al. 2010). However, low temperatures during winter months, and generally suboptimal light availability due to water column turbidity, macroalgal shading and reduced solar radiation at higher latitudes means that symbiosis has limited benefit in temperate regions (Jacques et al. 1983; Dimond and Carrington 2007; Ferrier-Pagès et al. 2011). Consequently, temperate corals must rely heavily on heterotrophy to support their growth and metabolic demands (Szmant-Froelich and Pilson 1980; Ferrier-Pagès et al. 2011). Temperate corals may have higher heterotrophic potential than their tropical counterparts because of the higher net productivity of temperate regions (Muller-Parker and Davy 2001) and the potentially higher heterotrophic capacity of temperate scleractinians (Tremblay et al. 2011).

In spite of their high heterotrophic potential, few shallow-water scleractinians are found beyond 40° latitude (Veron 2000). If heterotrophy can sustain these corals at high latitudes, what ultimately limits their distributions? In this study, we evaluate a simple temperature-based model of the growth of Astrangia poculata, a widespread scleractinian in shallow waters of the northwest Atlantic that forms a facultative symbiosis with Symbiodinium psygmophilum (Symbiodinium ITS2 type B2) throughout its range (Thornhill et al. 2008; LaJeunesse et al. 2012). This coral reaches an abrupt northern range limit around Cape Cod, Massachusetts (~42°N; Peters et al. 1988). Our model is based on a field study of the growth of A. poculata near this northern limit (Dimond and Carrington 2007), where it was speculated that low temperatures restrict coral growth further north through a combination of physiological and ecological effects. Here, we test this model and find that it accurately predicts the known geographic range limit of A. poculata.

Materials and methods

Range of A. poculata

Geographic occurrence data for A. poculata were acquired from the Global Biodiversity Information Facility (GBIF; data.gbif.org, accessed 2012-03-22). Data sources included the Ocean Biogeographic Information System, the National Museum of Natural History, the Yale University Peabody Museum, and the Marine Science Institute, University of California, Santa Barbara. The dataset was imported into ArcGIS 10 (ESRI, Redlands, CA, USA) and included a total of 599 observations of A. poculata occurrence at 270 distinct locations.

Coral growth data

Growth of A. poculata was measured using methods described in detail by Dimond and Carrington (2007). Briefly, symbiotic and asymbiotic A. poculata colonies were collected and reattached with underwater epoxy to cleared plots on vertical rock surfaces at two sites at 3 m depth in Narragansett Bay, RI. For 15 months between May 2004 and July 2005, digital photographs of the corals were taken on a quasi-monthly basis and image analysis was used to enumerate coral polyps as a measure of colony growth between sampling intervals. In situ temperature records from dataloggers (Stowaway TidBit loggers, Onset Computer, Bourne, MA, USA) fastened adjacent to experimental plots at 3 m depth permitted analysis of coral growth in relation to average temperatures between sampling intervals. As shown in Dimond and Carrington (2007), the temperature–growth relationship obtained by this method was very similar to laboratory-derived measurements of A. poculata CaCO3 deposition by Jacques et al. (1983). In addition to positive growth through polyp budding and skeletal accretion, however, Dimond and Carrington (2007) also documented loss of living coral polyps during the coldest times of the year. This loss was attributed to polyp dormancy and tissue thinning throughout the winter, which had the added consequence of rendering the colony unable to protect itself from sedimentation and the encroachment and settlement of organisms such as barnacles, mussels, and macroalgae. Polyp dormancy, defined here as a persistent contraction of polyps within their calices, was consistently observed in the field during the winter months at temperatures below approximately 5 °C. Growth versus temperature data were fitted with polynomial functions in SigmaPlot 11 (Systat Software Inc., San Jose, CA, USA). These functions were used to predict spatial patterns in coral growth, as described below.

Symbiosis between A. poculata and S. psygmophilum is facultative, and symbiont densities can range naturally from very low to over 107 cm−2 (Cummings 1983; Jacques et al. 1983; Dimond and Carrington 2008). While few corals appear to completely lack symbionts, field data suggest a natural break between fully symbiotic and functionally asymbiotic colonies at Symbiodinium densities of approximately 105 cm−2 (Cummings 1983; Dimond and Carrington 2008). Below this threshold density, corals appear visibly pale, and such colonies were selected and designated asymbiotic (=azooxanthellate) by Dimond and Carrington (2007). In contrast, corals designated symbiotic (=zooxanthellate) had full brown pigmentation from Symbiodinium densities greater than approximately 106 cm−2. Importantly, below this threshold density, Symbiodinium have negligible influence on coral calcification and do not provide net photosynthesis (Jacques et al. 1983). Symbiotic states in A. poculata appear to be largely stable over time (Dimond and Carrington 2007).

Sea surface temperature data

To estimate A. poculata growth potential along the U.S. Atlantic coast, we obtained sea surface temperature (SST) records from NOAA Pathfinder 4 km advanced very high resolution radiometer (AVHRR) satellite data (www.nodc.noaa.gov/SatelliteData/pathfinder4km; Casey et al. 2010). Cloud-screened monthly climatology datasets spanning the years 1981–2008 were imported into ArcGIS 10 for manipulation and analysis.

The suitability of remotely sensed SST data for studies of near-shore benthic ecology was recently evaluated by Smale and Wernberg (2009), who compared AVHRR and MODIS satellite data to in situ dataloggers at 10 m depth at several sites along the coast of Western Australia. Satellite-derived SST was typically 1–2 °C higher than temperatures logged at depth, and AVHRR data provided better correlations with in situ loggers than MODIS data, with correlations ranging from 0.85 to 0.96. Smale and Wernberg (2009) concluded that while satellite SST is not appropriate for fine-scale (e.g., sub-kilometer) studies requiring spatially and temporally explicit shallow subtidal temperatures, it is appropriate for large-scale analyses of broad geographic patterns (Fig. 1).

Fig. 1
figure 1

Geographic occurrences of Astrangia poculata from published collection records (black dots). Star indicates the location of the field study on A. poculata growth in Rhode Island, USA (Dimond and Carrington 2007)

Full size image

Analysis of coral growth potential

To estimate growth potential of A. poculata along the U.S. Atlantic coast, monthly SST climatology data were extracted along a transect 10 km offshore of the coast at 0.5° latitude intervals from 25 to 45°N (approximately Miami, FL, to Eastport, ME). For each month, symbiotic and asymbiotic coral growth (polyps mo−1) were calculated using the temperature-growth functions obtained in Fig. 2. Monthly estimates were then summed to determine estimated net annual coral growth (polyps yr−1) along the latitudinal transect. As a measure of the potential enhancement effect of symbiosis on coral growth, we also calculated the ratio of symbiotic to asymbiotic growth along the transect.

Fig. 2
figure 2

Temperature versus growth relationships a and 95 % confidence bands b for symbiotic and asymbiotic Astrangia poculata colonies derived from field growth data and in situ temperature dataloggers. Symbiotic: y = − 0.1849 + 0.0125x + 0.0005x2, r 2 = 0.97, p < 0.0001, n = 22. Asymbiotic: y = − 0.1161 + 0.0072x + 0.0003x2, r 2 = 0.96, p < 0.0001, n = 22

Full size image

For closer inspection of estimated growth of A. poculata at its northern range limit around Cape Cod, we used the raster calculator tool in ArcGIS 10 to estimate asymbiotic coral growth for each 4 km pixel and each month of the satellite SST monthly climatology data. Asymbiotic coral growth was modeled for this analysis with the rationale that symbiotic corals are relatively uncommon around A. poculata’s range limit; a transect from 0 to 15 m depth at Ft. Wetherill, Rhode Island in July 2010 found that 8 % of corals were fully symbiotic, with no clear depth pattern (R. Rotjan, pers obs). Secondly, the predicted benefit of symbiosis drops sharply beyond the range limit (see Results). As above, estimated annual growth was determined by summing estimated monthly growth. We then created a polygon coverage of areas with net growth versus no net growth by reclassifying values as either negative (no net growth) or positive (net growth) and clipping the dataset to a 100 km buffer of the coastline. This buffer distance was chosen primarily for visual clarity rather than to represent A. poculata habitat. Our SST-based analysis applies only to shallow near-shore waters and cannot predict deep-water coral growth patterns.

Results

Figure 1 shows geographic occurrences of A. poculata published by GBIF. Although lack of observations of A. poculata from a given area could reflect lack of sampling effort and should not necessarily be interpreted as an absence, the dataset suggests that A. poculata does not occur north of Cape Cod, Massachusetts. This limit is supported by Peters et al. (1988) and by Hargitt (1914), although the latter study mentioned a second-hand account of A. poculata occurrence in Casco Bay, Maine. We are unaware of any officially documented reports of A. poculata north of Cape Cod, but anecdotal reports suggest that A. poculata occasionally occurs in Massachusetts Bay and the Gulf of Maine (D. Thornhill and R. Rotjan, pers. obs.).

Symbiotic and asymbiotic coral growth illustrates the strong temperature dependence of A. poculata polyp gain and loss (Fig. 2). Little coral growth was observed below mean temperatures of approximately 10 °C. Symbiotic corals exhibited a clear growth advantage above 10 °C, but below this temperature, the data suggest that they performed poorly relative to asymbiotic colonies, actually losing polyps at a faster rate (Fig. 2a). For clarity, 95 % confidence bands for the coral growth functions are shown in Fig. 2b. These bands are extrapolated beyond the extent of the data, hence their widening at higher values.

Estimated coral growth along the U.S. Atlantic coast based on temperature alone shows a strong latitudinal effect (Fig. 3). For this analysis, coral growth was extrapolated beyond the extent of the data, so it should be noted that summertime coastal temperatures south of New Jersey (~40°N) exceed those observed in our Rhode Island dataset. Despite considerable overlap of 95 % confidence limits for predicted symbiotic and asymbiotic coral growth over much of the range, general patterns suggest interesting trends and contrasts. Most significantly, this analysis predicted net polyp loss north of 41.9°N for asymbiotic corals and 43.0°N for symbiotic corals. Symbiotic coral growth exceeds asymbiotic coral until 43.6°N, after which asymbiotic coral growth exceeds that of symbiotic corals; this pattern is also illustrated by the ratio of symbiotic to asymbiotic coral growth, which drops sharply below 1 at 43.6°N. This ratio remains relatively high and stable at southerly locales until dropping precipitously north of ~40°N.

Fig. 3
figure 3

Latitudinal trends in predicted annual symbiotic and asymbiotic coral growth (±95 % confidence limits) along the U.S. Atlantic coast (solid lines), and the predicted benefit of symbiosis expressed as the ratio of symbiotic coral growth to asymbiotic coral growth (dotted line)

Full size image

Closer examination of spatial patterns of estimated A. poculata growth around its northern range limit shows that the model accurately predicts the coral’s actual range limit in terms of areas with and without predicted net growth (Figs. 1, 4). Astrangia poculata is effectively absent north of the southern coast of Cape Cod, making Buzzards Bay and Nantucket Sound the northernmost coastal waters with persistent documented populations of A. poculata. The model suggests that cold water intrusion along the outer Cape Cod coast may act as a barrier to coral growth there and further north. Although the model does predict net asymbiotic coral growth within part of Cape Cod Bay, there are no published records of A. poculata there.

Fig. 4
figure 4

Spatial patterns of predicted net growth of Astrangia poculata around its known range limit (Cape Cod, Massachusetts), with sea surface temperature (SST) climatology as a backdrop

Full size image

Discussion

The results of our analysis suggest that the seasonally integrated net growth of A. poculata colonies is constrained by low temperature north of Cape Cod. Simply put, the effect of increasing latitude on A. poculata annual net growth is to reduce favorable conditions and increase unfavorable conditions. Although biogeographic limits can arise from the complex interplay of demographic, ecological, and environmental factors (Gaston 2009; Sexton et al. 2009), temperature is a key factor governing organism physiology and therefore plays a significant role in the biogeography of many organisms (Pörtner 2002). Our analysis did not test alternative hypotheses for the upper latitudinal limit of A. poculata and therefore may oversimplify the processes that influence A. poculata biogeography. Nevertheless, our results demonstrated that latitudinal temperature patterns alone could successfully predict this coral’s distribution. We also provide evidence that while associating with symbiotic dinoflagellates remains largely beneficial for A. poculata around its range limit, the benefit of symbiosis declines sharply at higher latitudes.

Astrangia poculata exhibits metabolic characteristics similar to tropical corals when acclimated to tropical temperatures (Jacques et al. 1983), so warmer and longer summers should be favorable for A. poculata growth. Calcium carbonate deposition rates of A. poculata in the laboratory were highly temperature-dependent (Jacques et al. 1983) and very similar to the seasonal polyp budding rates observed in the field by Dimond and Carrington (2007). The steep increase in growth rates observed in both studies at higher temperatures may be due in part to the fact that A. poculata respiration is largely constant between 11.5 and 23 °C (Jacques et al. 1983). This ability of A. poculata to regulate its respiration rates across a broad range of temperatures may allow increased energy input to growth as temperature increases (Jacques et al. 1983). Above 23 °C, however, A. poculata respiration rates increase (Jacques et al. 1983), which probably limits the amount of energy available for growth. Because of this and because energy sources may not be sufficient to keep up with energy demand, particularly at higher temperatures (e.g., Fitt et al. 2000), our extrapolated analysis of growth at southerly latitudes beyond the range of temperatures in our dataset is purely theoretical and should be interpreted with caution. Moreover, our analysis does not account for potential local adaptation to either lower or higher temperature regimes other than those in our Rhode Island dataset. Anecdotal reports of A. poculata north of Cape Cod, for example, could reflect local adaptation of a small population to colder temperatures. Local thermal adaptation has been observed in tropical corals (Smith-Keune and van Oppen 2006) and Symbiodinium (Howells et al. 2011), and broadcast spawning species such as A. poculata generally have greater potential for local adaptation (Szmant-Froelich et al. 1980; Sanford and Kelly 2011).

During the winter, a state of torpor characterized by contracted polyps, slowed respiration rate, and growth cessation allows A. poculata to conserve energy albeit at the cost of starvation, tissue thinning, and loss of living surface area to competitors (Jacques et al. 1983; Peters et al. 1988; Grace 1996; Dimond and Carrington 2007). Colder, longer winters such as those occurring north of Cape Cod therefore result in a prolonged state of torpor that exacerbates polyp loss to competitors. Polyp expansion is not only necessary for feeding, but also protects A. poculata from burial by sediment (Peters and Pilson 1985) and is probably also critical for defense of space from both settling and neighboring sessile organisms. For the temperate northeastern Pacific cup coral Balanophyllia elegans, for example, maintaining tentacular contact with neighboring sessile invertebrates is an effective defense against their encroachment (Bruno and Witman 1996). Among the more destructive sessile invertebrates frequently found in contact with A. poculata on southern New England rocky reefs are boring clionid sponges (J. Dimond pers. obs.). Rützler (2002) found that the ability of Caribbean reef corals to defend themselves against clionid sponges was compromised when corals were stressed by either high or low temperatures. In addition to sponges and other invertebrates, seaweeds are the dominant competitor for space at the shallowest depths on temperate reefs, and under algal canopies, A. poculata adopts a flat encrusting morphology to avoid excessive algal abrasion (Miller and Hay 1996; Grace 2004). The detrimental effects of algal competition are further illustrated by the temperate coral Oculina arbuscula (likely a regional moniker of O. varicosa, Eytan et al. 2009) which co-occurs with A. poculata south of Cape Hatteras (~35°N). This species grows only in a branching morphology even at shallow depths, but benefits considerably from the grazing activity of the crab Mithrax forceps, which consumes seaweeds adjacent to the coral and gains protection from predators by closely associating itself with the coral’s branches (Stachowicz and Hay 1999). Oculina arbuscula may succumb to the effects of restricted growth and interspecific competition northward of Cape Hatteras in a similar manner to what we predict for A. poculata at its range limit around Cape Cod.

Cape Cod is a well-known biogeographic boundary where many species reach their northern, and to a lesser extent, southern range limits (Engle and Summers 2000; Wares 2002; Hale 2010). Pleistocene glaciation of the New England coast reached its southern terminus at Cape Cod and may have affected contemporary distributions and gene flow of coastal species, but oceanographic conditions including both temperature and currents are also very likely to play a role in setting species’ range limits around Cape Cod (Wares 2002). Western Gulf of Maine and Mid-Atlantic Bight coastal circulation is predominantly southward (Churchill et al. 2005; Lentz 2008), which may create unfavorable conditions for northward larval dispersal around Cape Cod (Wares 2002). On the other hand, larvae originating from coastal areas south of Cape Hatteras can be transported far northward by the Gulf Stream and its warm-core rings (Hare et al. 2002), probably as far north as the Scotian Shelf based on the high frequency of expatriated subtropical fish larvae often found there (Markle et al. 1980). Independent of larval dispersal limitation, however, low temperature has been found to limit species’ ranges northward of Cape Cod by three primary factors, summarized by Sanford et al. (2006): 1) inability to survive winter temperatures; 2) inability to reproduce at cooler temperatures; and 3) inability to successfully compete for resources with northern species. Although coastal wintertime temperatures north of Cape Cod reach similar lows as those observed to the south, low temperature conditions persist for longer periods of time at increasing latitudes, and thus, exposure duration is probably at least as important a consideration as temperature per se. With this caveat, and the fact that we did not test A. poculata performance in the presence of the suite of potential competitors occurring north of Cape Cod, our results are consistent with hypotheses 1 and 3.

Reproductive limitations to A. poculata’s distribution were not examined in our analysis but deserve consideration. Coral gametogenesis, for example, could become limited at higher latitudes and lower temperatures. Although gametogenesis in A. poculata begins in early spring, the coldest time of year in southern New England coastal waters, it is not complete until late summer and may be dependent on a sufficiently warm summer to be completed before water temperatures begin to drop again in the autumn (Szmant-Froelich et al. 1980). Szmant-Froelich et al. (1980) found that A. poculata kept in the laboratory at 20–22 °C and fed regularly were able to spawn year-round, suggesting that A. poculata gametogenesis was controlled by either temperature or food availability. Spawning during late summer in natural populations may also be critical for larval development, which is likely to be temperature-dependent. The fiddler crab Uca pugnax also reaches its northern range limit around Cape Cod, and this limit appears to be determined by the inability of larvae to develop at the lower temperatures commonly encountered north of Cape Cod (Sanford et al. 2006). It is important to recognize that other life history characteristics may limit the competency of A. poculata at higher latitudes, but here we argue that temperature effects on adult persistence are sufficient to control the coral’s range regardless of reproductive success or early life history stage survival.

Does temperature alone cause the considerable seasonal changes in coral growth observed in the field, and can latitudinal changes in temperature alone predict coral growth and persistence? Two additional factors influencing coral growth are light availability and aragonite saturation state, both of which co-vary seasonally and latitudinally with temperature (Kleypas et al. 1999). If we consider only asymbiotic corals, light availability can be excluded as a factor limiting A. poculata’s range. Aragonite saturation state has been implicated as a limiting factor for the growth of both tropical reef corals (Kleypas et al. 1999) and deep-sea scleractinians (Cairns 2007). Aragonite saturation state generally declines with increasing latitude (Kleypas et al. 1999) and also changes seasonally (Holcomb et al. 2010). In Woods Hole, Massachusetts, the northern range limit of A. poculata, seawater aragonite saturation state varies from approximately 1.5 in winter to 2.5 in summer (Holcomb et al. 2010). Holcomb et al. (2010, 2012) found that reduced seawater aragonite saturation state resulting from experimental CO2 enrichment generally decreased A. poculata calcification, but nutrient supplementation could offset the effects of low aragonite saturation, and female corals were more negatively affected than males. Experiments with two other temperate corals, Cladocora caespitosa (Rodolfo-Metalpa et al. 2010) and Oculina arbuscula (Ries et al. 2010), have shown that these corals are fairly robust to changes in aragonite saturation. Ries et al. (2010) found that O. arbuscula can maintain steady calcification rates until aragonite becomes under saturated (<1). This suggests that aragonite saturation state is unlikely to affect the range limit of O. arbuscula since supersaturating conditions persist well northward of Cape Hatteras. Even in the Gulf of Maine, aragonite saturation remains well above 1 except where periodic large discharges of freshwater occur around river mouths (Salisbury et al. 2008). It seems unlikely that the abrupt northern range limit of A. poculata around Cape Cod is delineated by changes in aragonite saturation state alone, but since temperature affects aragonite saturation state, the two factors may act synergistically in limiting A. poculata growth northward.

Our analysis suggests that A. poculata’s range limit coincides with a sharp decline in the benefit of associating with Symbiodinium. This is due to limited growth enhancement by symbiosis at lower summer temperatures and, intriguingly, an augmented polyp loss effect associated with symbiosis during colder months, suggesting that symbiotic dinoflagellates may incur costs during stressful winter periods. Thornhill et al. (2008) demonstrated that Symbiodinium ITS2 type B2 (recently described as S. psygmophilum, LaJeunesse et al. 2012), the symbiont lineage associated with A. poculata, is remarkably tolerant of low temperatures in comparison with tropical symbiont types. Despite this, the maximum photochemical efficiency of photosystem II of S. psygmophilum declined to approximately 0 at 10 °C (Thornhill et al. 2008). Additionally, S. psygmophilum had no measureable effect on the growth of A. poculata below 6.5 °C (Jacques et al. 1983). This symbiont species does, however, continue to divide at temperatures as low as 3 °C and in prolonged darkness (Dimond and Carrington 2008), suggesting that it may become heterotrophic and draw resources from its host during the winter (e.g., Steen 1986). This hypothesis is supported by the observations that asymbiotic A. poculata had higher tissue mass per unit area than symbiotic corals, particularly during the winter (Cummings 1983), and that asymbiotic corals were less susceptible than symbiotic corals to tissue damage during experimental starvation and sedimentation stress (Peters and Pilson 1985). In spite of the potential negative effects of symbiosis on the host under suboptimal conditions, our analysis also suggests that, at shallow depths with adequate light penetration (Jacques et al. 1983), symbiosis can provide a valuable boost for A. poculata growing near its range limit.

Here, we have provided support for two hypotheses concerning the biogeography of A. poculata which may apply to shallow-water temperate corals in general. We show that latitudinal declines in temperature (1) not only limit favorable conditions for coral growth, but also exacerbate unfavorable conditions that cause corals to lose surface area to competitors and (2) may be associated with a shift to net costs of associating with symbiotic dinoflagellates. Gradual northward range shifts in response to climatic warming would be predicted based on the first hypothesis, but in the meantime, further investigation into latitudinal variation in the costs and benefits of associating with symbiotic dinoflagellates would be worthwhile from both an ecological and evolutionary perspective.