Northeastern Naturalist 164 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 22001144 NORTHEASTERN NATURALIST 2V1o(l2.) 2116,4 N–1o7. 32 Characteristics of Piping Plover Nesting Habitat in the Canadian Maritime Provinces Andrew W. Boyne1,*, Diane L. Amirault-Langlais2,3, and Anthony J. McCue4 Abstract - To better understand and compare regional habitat characteristics for the endangered Charadrius melodus (Piping Plover) in the Canadian Maritime provinces, we surveyed transects on nesting beaches along the Gulf of St. Lawrence and Atlantic coasts of the Canadian Maritime provinces. The beaches along the Gulf of St. Lawrence were flatter, were wider and had a higher proportion of mixed substrate than those on the Atlantic coast. While habitat use by breeding Piping Plovers was largely consistent with other studies, significant differences were found between the two regions. Piping Plovers along the Gulf of St. Lawrence were found nesting in flatter areas with a high proportion of mixed substrate and less wrack. On the Atlantic coast, Piping Plovers preferred wider and flatter sections of beach. Future conservation efforts aimed to maintain or increase populations should recognize the importance of fine-scale habitat characteristics in nes t-site selection. Introduction Charadrius melodus Ord (Piping Plover) is listed as endangered in Canada (Boyne 2001), as is the Great Lakes population in the United States, while the species is listed as threatened throughout the rest of its US range (USFWS 1985). Threats to nesting Piping Plovers in the Canadian Maritime provinces (New Brunswick [NB], Nova Scotia [NS], and Prince Edward Island [PE]) include predation (Flemming et al. 1992), human disturbance (Flemming et al. 1988), and habitat loss due to coastal development and natural changes (Plissner and Haig 2000, Wentzell 1997). Considerable efforts have been made to reduce the amount of predation and human disturbance on Piping Plover beaches in eastern Canada (Gratto-Trevor and Abbott 2011). Specifically, sections of national parks used by nesting plovers are closed during the breeding season and Piping Plover guardian programs have been established to locate and monitor nesting areas and increase public awareness (Goossen et al. 2002, Gratto-Trevor and Abbott 2011). Predator exclosures have been used on many beaches to reduce nest predation, and in some cases, predators have been removed (Gratto-Trevor and Abbott 2011, Wentzell 1997). Despite increased conservation efforts, the number of Piping Plovers nesting in the Maritimes decreased between the 1991 and the 1996 International Piping Plover Censuses 1Canadian Wildlife Service, Atlantic Region, Environmental Stewardship Branch, Environment Canada, 45 Alderney Drive, 16th floor, Dartmouth, NS B2Y 2N6, Canada. 2Canadian Wildlife Service, Atlantic Region, Environmental Stewardship Branch, Environment Canada, 17 Waterfowl Lane, Sackville, NB E4L 1G6, Canada. 3Current address - Kouchibouguac National Park, 186 Route 117, Kouchibouguac, NB E4X 2P1, Canada. 4Heffley Creek, BC V0E 1Z1, Canada. *Corresponding author - andrew.boyne@ec.gc.ca. Manuscript Editor: Greg Robertson Northeastern Naturalist Vol. 21, No. 2 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 165 (Plissner and Haig 2000). Subsequent censuses conducted in 2001, 2006, and 2011 indicate that numbers rebounded somewhat from the low of 1996 (Canadian Wildlife Service [CWS], Sackville, NB, Canada, unpubl. data; Elliott-Smith et al. 2009). However, during this same period the number of Piping Plovers increased substantially along the US Atlantic coast (Elliott-Smith et al. 2009). Unlike predation and human disturbance, the effect of habitat and habitat change on the abundance and distribution of Piping Plovers is not well understood. Natural and anthropogenic habitat change may affect Piping Plover nesting distribution, but not necessarily in a predictable manner (Cohen et al. 2009). Some natural processes, such as winter storm surges or ice scour may create habitat for nesting plovers by returning beaches to early successional stages. Conversely, in the absence of natural storm events, beach succession may progress to a point where the habitat may no longer be suitable for Piping Plover nesting. Our objective was to quantify characteristics and use of nesting habitat by Piping Plovers in the Maritime provinces. To assess the potential role of habitat on the abundance and distribution of Piping Plovers, we examined geophysical characteristics of Piping Plover nesting beaches. Our goals were to determine if there were differences in habitat (1) at the beach scale between the Gulf of St. Lawrence and the Atlantic coast, regions that likely represent two discrete subpopulations (Amirault et al., in press); and (2) at the nest-site scale between areas on beaches where Piping Plovers did, and did not, nest. Methods Site description In eastern Canada, Piping Plovers nest on open sandy beaches with little vegetative cover. Complete censuses of potential Piping Plover habitat in eastern Canada were conducted in 1991 and 1996 as part of the International Piping Plover Census, and the characteristics of breeding habitat were quantified along the Gulf of St. Lawrence and Atlantic Ocean coasts of the three Canadian Maritime provinces: NB, NS and PE (Plissner and Haig 2000). We measured habitat characteristics on a subset of beaches that were surveyed for Piping Plovers during the breeding season at least twice from 1991 to 1996 and at least three times prior to 1991 (Amirault et al. 1997, Boates et al. 1994, Island Nature Trust 1997). Seventy beaches met these criteria, all of which supported breeding Piping Plovers in at least one year. We quantified habitat characteristics on a subset of beaches (n = 29) that we selected as high priority based on highest historical use by nesting Piping Plovers (minimum of 5 individuals observed in at least one survey during 1966–1996) (Fig. 1). Transect selection and habitat sampling We measured habitat characteristics during the breeding season, 3 June–22 July 1997, along transects set perpendicular to the shoreline every 250 m. We selected the distance between transects to maximize the amount of coastline that could be practically studied, and located the first transect 50 m, parallel to the shore, from the beginning of a beach or from an access point in the middle of the beach. Each tranNortheastern Naturalist 166 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 Vol. 21, No. 2 sect ran from the “swash line” (line of wet sand from the most recent high tide) to the point where dense vegetation began in the foredune, or where high dunes or escarpments began. If we could not locate a transect 250 m from the previous transect because the beach ended, no transect was characterized. If a "blowout" (channel or hollow through a section of dune created by wind) or barrier (e.g., wharf or rock outcrop) prevented the characterization of a representative transect, we extended the distance to the next transect in 25 m increments until it was possible to measure the transect. To minimize disturbance, this protocol was also followed if a transect fell within 25 m of an active Piping Plover nest. We measured the length of transects and the distance between them with a measuring wheel (accurate to 0.1 m) and recorded the width of the beach at each transect. We used a clinometer to measure slope from the swash line to the highest point of each transect. Figure 1. Piping Plover nesting beaches where habitat was characterized along the Atlantic coast (grey circles) and Gulf of St. Lawrence (black circles) in the Canadian Maritime provinces. Northeastern Naturalist Vol. 21, No. 2 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 167 Along the linear length of each transect, we characterized visually and recorded substrate type as either cobble, sand, or mixed. Cobble substrate consisted of any combination of pebbles (1–5 cm diameter), stones (5–10 cm diameter), and/or rocks (10–20 cm diameter; see Flemming et al. 1992) with no more than 10% sand; sand substrate consisted of sand with no more than 10% cobble; and mixed substrate consisted of more than 10% of both sand and cobble substrates. The proportion of the transect that was vegetated and the proportion that contained washed up wrack, which we estimated by measuring the length of a transect with any vegetation or wrack that occurred within 1 m of the transect, were included as separate variables. We did not include lone plants or shoots and did not differentiate species of plants, making this a rough estimate of vegetation. Transect classification Gulf of St. Lawrence transects were located along the coastline of NB, PE, and the northern shore of NS. Atlantic coast transects were located along the eastern and southern shores of NS (Fig. 1). We classified transects on both the Gulf of St. Lawrence and Atlantic coast as occupied (those on both sides of a Piping Plover nest) or unoccupied (all others) by nesting Piping Plovers. In some cases, an occupied transect was not representative of the nesting habitat (e.g., a narrow section before a wide segment created by a blowout or “washover” [hollow in a dune created when the ocean breaches the dune]). In cases where there was a distinct change in habitat characteristics between a transect and the nest location, and the transect was not representative of the features of the nest site, we classified the transect as unoccupied. Statistical analyses At the beach scale, we investigated differences in habitat characteristics between Gulf of St. Lawrence (n = 18) and Atlantic coast (n = 11) beaches using an analysis of variance. In particular, we compared slope, width, proportion of substrate classified as mixed, and proportion with vegetation between regions with t-tests. All statistical analyses were conducted using R (R Development Core Team 2011). At the nest-site scale, we constructed a set of candidate models for each region (i.e., Atlantic coast and Gulf of St. Lawrence) to identify the most important habitat characteristics for nest-site occupancy of 29 high priority Piping Plover beaches (n = 75 Atlantic coast transects, 249 Gulf of St. Lawrence transects). We constructed these models within the generalized linear mixed-effects model framework to control for autocorrelation of sites at the beach level (lme4 package; Bates et al. 2011). To avoid multicollinearity among variables in the models, no pair of variables with a Pearson product-moment correlation >0.5 was included in any one model. Using second-order Akaike’s information criteria (AICc), we compared models for each region to determine which model was the best fit to the observed patterns of use (AICcmodavg package; Mazerolle 2012). For the best models, we computed parameter estimates for fixed effects with unconditional standard errors and model averaging when appropriate (Burnham and Anderson 2002). Northeastern Naturalist 168 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 Vol. 21, No. 2 Results Differences between coastal environments We observed significant differences in the physical characteristics of Atlantic coast and Gulf of St. Lawrence beaches used by nesting Piping Plovers (Table 1). Gulf of St. Lawrence beaches had a lower slope angle (t27 = 2.86, P = 0.008), were wider (t 27 = 2.44, P = 0.022), and had a higher proportion of mixed substrate (t 27 = 3.47, P = 0.002) than Atlantic coast beaches. However, the proportion of the beach with vegetation was not significantly different between regions (t27 = 0.777, P = 0.44). Differences between occupied and unoccupied transects We developed models for each region, with beach as the random effect. Candidate models for both regions included one of the substrate classes, slope or width, and proportion with vegetation and proportion with wrack in various combinations. For the Atlantic coast, slope was correlated with both sand and cobble substrates; thus, we did not combine these variables in a candidate model. For the Gulf of St. Lawrence, slope was correlated with cobble substrate only. Slope and width were correlated in both regions, as were the various substrate types. We constructed 13 candidate mixed-effect models to explain Piping Plover occupancy on the Atlantic coast. The model with the highest Akaike weight (w = 0.544) contained only beach width as a fixed effect (Table 2). The second-best model contained only slope as a fixed effect (w = 0.291). These two models independently predicted a positive association with width (i.e., preference for wider sections of beach; βwidth = 0.143 ± 0.0623 [SE]) and a negative association with slope (i.e., preference for flatter sites; βslope = -0.568 ± 0.287) when controlled for beach as a random effect. All other models for the Atlantic coast had low weight of evidence (wi ≤ 0.036) for consideration and therefore provided little additional information on habitat preferences. We also constructed 13 candidate mixed-effect models for explaining Piping Plover occupancy on the Gulf of St. Lawrence. The model that best explained occupancy contained slope, proportion of mixed substrate, proportion with vegetation, and proportion with wrack (w = 0.534; Table 3). The second-best model differed slightly in including sand rather than mixed substrate (w = 0.275). Finally, a third acceptable Table 1. Piping Plover nesting-beach characteristics along the southern Atlantic coast (n = 11) and the Gulf of St. Lawrence (n = 18) in the Maritime provinces of Canada during the 1997 breeding season. Values presented indicate mean ± standard deviation. Characteristic Atlantic Coast Gulf of St. Lawrence Slope (%) 5.4 ± 3.6 2.6 ± 1.5 Width (m) 17.3 ± 7.9 33.1 ± 20.5 Cobble (%) 25.2 ± 31.6 5.6 ± 9.5 Mixed substrate (%) 8.4 ± 8.5 33.5 ± 23.0 Sand (%) 65.2 ± 33.8 60.4 ± 23.3 Emergent vegetation (%) 4.6 ± 5.4 6.2 ± 5.5 Northeastern Naturalist Vol. 21, No. 2 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 169 model contained slope, proportion with vegetation, and proportion with wrack as fixed effects, but no substrate parameter (w = 0.124). The model-averaged parameter estimates indicate preference for flatter sites (βslope = -0.743 ± 0.202) with less wrack (βwrack = -3.00 ± 1.28) and tendency for less vegetation (βveg = -0.641 ± 2.21) when Table 3. Candidate models ranked by second-order Akaike’s information criteria with strength of evidence measures for distinguishing between used and unused beach transects for Piping Plovers along the Gulf of St. Lawrence coast in the Maritime provinces of Canada. Beach was controlled for as a random effect in all models. Rank Fixed k Δi L(gi|x) wi Model structure1 1 4 0.0 1.00000 0.534 SLOPE + MIXED + P_VEG + P_WRACK 2 4 1.3 0.51554 0.275 SLOPE + SAND + P_VEG + P_WRACK 3 3 2.9 0.23290 0.124 SLOPE + P_VEG + P_WRACK 4 1 4.8 0.09170 0.049 SLOPE 5 4 7.8 0.02035 0.011 WIDTH + COBBLE + P_VEG + P_WRACK 6 4 9.1 0.01075 0.006 WIDTH + MIXED + P_VEG + P_WRACK 7 4 12.8 0.00165 8.8E-04 WIDTH + SAND + P_VEG + P_WRACK 8 1 15.9 0.00035 1.9E-04 WIDTH 9 1 16.3 0.00029 1.5E-04 COBBLE 10 3 16.7 0.00023 1.2E-04 COBBLE + P_VEG + P_WRACK 11 1 16.9 0.00022 1.2E-04 MIXED 12 1 22.2 1.48E-05 7.9E-06 SAND 13 2 28.4 6.66E-07 3.6E-07 P_VEG + P_WRACK 1See Table 2 for parameter definitions. Table 2. Candidate models ranked by second-order Akaike’s information criteria with strength of evidence measures for distinguishing between used and unused beach transects for Piping Plovers along the Atlantic coast of Nova Scotia. Beach was controlled for as a random effect in all models. Rank Fixed k Δi L(gi|x) wi Model structure1 1 1 0.0 1.00000 0.544 WIDTH 2 1 1.3 0.53465 0.291 SLOPE 3 3 5.4 0.06644 0.036 SLOPE + P_VEG + P_WRACK 4 4 6.4 0.04123 0.022 WIDTH + COBBLE + P_VEG + P_WRACK 5 4 6.4 0.04106 0.022 WIDTH + MIXED + P_VEG + P_WRACK 6 4 6.4 0.04099 0.022 WIDTH + SAND + P_VEG + P_WRACK 7 4 6.9 0.03168 0.017 SLOPE + MIXED + P_VEG + P_WRACK 8 1 7.5 0.02302 0.013 COBBLE 9 1 7.6 0.02251 0.012 SAND 10 1 7.8 0.02037 0.011 MIXED 11 2 9.6 0.00819 0.005 P_VEG + P_WRACK 12 3 10.9 0.00436 0.002 COBBLE + P_VEG + P_WRACK 13 3 11.6 0.00310 0.002 SAND + P_VEG + P_WRACK 1SLOPE = percent slope; WIDTH = distance from swash line to dense vegetation at foredune; COBBLE = proportion of transect with substrate consisting of pebbles, stones, and/or rocks with ≤10% sand; SAND = proportion of transect with substrate consisting of sand with ≤10% pebbles, stones, and/or rocks; MIXED = proportion of transect with substrate consisting of >10% sand and >10% pebbles, stones, and/or rocks; P_VEG = proportion of transect with emergent vegetation within 1 m, perpendicular; P_WRACK = proportion of transect with wrack within 1 m, perpendicular. Northeastern Naturalist 170 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 Vol. 21, No. 2 controlled for beach as a random effect. Mixed (βmixed = 1.79 ± 0.789) and sand (βsand = -1.52 ± 0.779) substrate habitat characteristics were inversely related in respective models in accordance with their high correlation (r = 0.92). Discussion Previous studies of Piping Plover habitat use have shown that Piping Plovers prefer to nest on the widest sections of beach in flat areas with mixed substrates such as gravel, shells, sticks, and pebbles (Burger 1987, Cairns 1982, Dundas 1995, Elias-Gerkin 1994, Espie et al. 1996, Flemming et al. 1992, Prindiville Gaines and Ryan 1988). Maslo et al. (2011) suggested that low-slope, mixed-substrate areas with limited vegetation are primary nesting site characteristics common to all Piping Plover populations. Beaches on the Gulf of St. Lawrence had a lower slope, were wider and had more mixed substrate than those on the Atlantic coast. This finding, in light of the previous studies, suggests that Gulf of St. Lawrence beaches may provide betterquality nesting habitat than Atlantic coast beaches. Nesting on wide sections of beach allows birds to nest far from the vegetation line and the water, possibly reducing the risk of both predation and flooding (Espie et al. 1996, Prindiville Gaines and Ryan 1988). Flat sites are also thought to reduce risk of predation (Burger 1987); however, Piping Plovers have nested in flat micro-sites within steeper habitat (Anteau et al. 2012). Mixed substrates may provide camouflage from predators and protection from wind and sand (Flemming et al. 1992). The presence of mixed substrates may suggest that a section of beach is less susceptible to flood or storm tides as tides have not previously removed debris from these areas (Burger 1987). In a study of Piping Plover nest-site selection, Flemming et al. (1992) found that nest sites in southern NS did not differ from random sites with respect to the number of pebbles, stones, or rocks. This finding is supported by the results of our models, whereby different substrate types were not present in the top-ranking models differentiating occupied and unoccupied transects on the Atlantic coast of NS. Flemming et al. (1992) also found that nest sites had more vegetation than random sites. Conversely, we found very little evidence to include vegetation as an explanatory variable; however, the different findings may be explained by differences in methodologies between studies, particularly since our estimate of vegetation was coarse. Cairns (1977) suggested that Piping Plovers nested in vegetation in southern NS because beaches are narrow. Our results provide evidence that Piping Plovers select wider sections of beach as the primary site-level selection criterion. In addition to width, slope is likely a key factor in site-level selection along the Atlantic coast. These two coarse parameters provide the best explanation for habitat selection for Piping Plovers in southern NS. Moreover, given that beaches on the Atlantic coast are narrower and steeper than on the Gulf of St. Lawrence, these findings may suggest that plovers in this region are adapting to limited availability of suitable sites in otherwise sub-optimal habitat, similar to plovers at Lake Sakakawea in North Dakota (Anteau et al. 2012). This finding may also indicate that Northeastern Naturalist Vol. 21, No. 2 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 171 habitat selection is operating at a finer scale in this region than could be detected through this study. Flemming et al. (1992) noted that plovers on beaches along the Gulf of St. Lawrence (northern NS and eastern NB) selected nest sites with more pebbles, stones, and rocks than at random sites. Similarly, we found that the proportion of mixed substrate had the strongest positive effect to explain site use by Piping Plovers. Whereas our best models include vegetation as an explanatory variable, Flemming et al. (1992) found that nest sites along the Gulf of St. Lawrence were not associated with vegetation. Our result may be an artifact of model development in which no models were tested that included wrack without also including vegetation. The high standard error in relation to the parameter estimate for vegetation suggests that the presence of wrack is probably the stronger effect influencing model selection. Although we identified the best-fit models for these data, other models (AIC difference = 4–7; Burnham and Anderson 2002) may be reasonable under different scenarios (e.g., higher nesting density). Selection among beach slope, width, and substrate type seems to have occurred at a finer scale rather than at the beach level, suggesting that other localized factors such as presence of foraging habitat, predation, or human disturbance may be important (Cohen et al. 2009). Our findings support the results of previous studies in terms of divergent habitat preferences between the two regions and provide justification for continued efforts to manage and protect Piping Plover habitat at both broad and fine scales. Particular priority must be assigned to protecting remaining tracts of suitable habitat along the Atlantic coast to allow for population persistence and future recovery. Acknowledgments We thank K. Davidson, G.J. Robertson, T.S. Jung, J. McKnight and T. Imlay for reviewing earlier drafts of this manuscript. S.M. Boyne, C.T. Boyne, J.O. Boyne, P.R. Macdonald, G. Martin, and R.L. Gautreau helped with fieldwork. M. Elliott prepared the map. A.W. Boyne was supported by Environment Canada’s Science Horizons Youth Internship Program. Literature Cited Amirault, D.L., R.D. Chiasson, and S. Dietz. 1997. New Brunswick atlas of Piping Plover beaches. Canadian Wildlife Service, Environment Canada, Sackville, NB, Canada. Amirault-Langlais, D.L., T.L. Imlay and A.W. Boyne. 2014. Dispersal patterns suggest two breeding populations of Piping Plovers in eastern Canada. The Wilson Journal of Ornithology. 126:352–359. Anteau, M.J., M.H. Sherfy, and M.T. Wiltermuth. 2012. Selection indicates preference in diverse habitats: A ground-nesting bird (Charadrius melodus) using reservoir shoreline. PLoS ONE 7:e30347. Bates, D., M. Maechler, and B. Bolker. 2011. lme4: Linear mixed-effects models using S4 classes. Version 0.999375-42. Available online at http://CRAN.R-project.org/ package=lme4. Accessed 5 February 2012. Boates, J.S., P. Austin-Smith, G. Dickie, R. Williams, and D. Sam. 1994. Nova Scotia Piping Plover atlas. Nova Scotia Department of Natural Resources, Kentville, NS, Canada. Northeastern Naturalist 172 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 Vol. 21, No. 2 Boyne, A.W. 2001. Updated COSEWIC status report on the Piping Plover, Charadrius melodus, in Canada. Committee on the Status of Endangered Wildlife in Canada, Sackville, NB, Canada. Burger, J. 1987. Physical and social determinants of nest-site selection in Piping Plover in New Jersey. Condor 89:811–818. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach. 2nd Edition. Springer, New York, NY. Cairns, W.E. 1977. Breeding biology and behaviour of the Piping Plover in southern Nova Scotia. M.Sc. Thesis. Dalhousie University, Halifax, NS, Canada. Cairns, W.E. 1982. Biology and behavior of breeding Piping Plovers. Wilson Bulletin 94:531–545. Cohen, J.B., L.M. Houghton, and J.D. Fraser. 2009. Nesting density and reproductive success of Piping Plovers in response to storm- and human-created habitat changes. Wildlife Monographs 173:1–24. Dundas, H.A. 1995. Breeding habitat selection by Piping Plovers in southern Saskatchewan. M.Sc. Thesis. University of Saskatchewan, Saskatoon, SK, Canada. Elias-Gerkin, S.P. 1994. Piping Plover habitat suitability on Central Long Island, New York barrier islands. M.Sc. Thesis. Virginia Polytechnic Institute and State University, Blacksburg VA. Elliott-Smith, E., S.M. Haig, and B.M. Powers. 2009. Data from the 2006 International Piping Plover Census. Data Series No. 426. US Geological Survey, Reston, VA. Espie, R.H.M., R.M. Brigham, and P.C. James. 1996. Habitat selection and clutch fate of Piping Plovers (Charadrius melodus) breeding at Lake Diefenbaker, Saskatchewan. Canadian Journal of Zoology 74:1069–1075. Flemming, S.P., R.D. Chiasson, P.C. Smith, P.J. Austin-Smith, and R.P. Bancroft. 1988. Piping Plover status in Nova Scotia related to its reproductive and behavioral responses to human disturbance. Journal of Field Ornithology 59:321–330. Flemming, S.P., R.D. Chiasson, and P.J. Austin-Smith. 1992. Piping Plover nest-site selection in New Brunswick and Nova Scotia. Journal of Wildlife Management 56:578–583. Goossen, J.P., D.L. Amirault, J. Arndt, R. Bjorge, S. Boates, J. Brazil, S. Brechtel, R. Chiasson, G.N. Corbett, R. Curley, M. Elderkin, S.P. Flemming, W. Harris, L. Heyens, D. Hjertaas, M. Huot, B. Johnson, R. Jones, W. Koonz, P. Laporte, D. McAskill, R.L.G. Morrison, S. Richard, F. Shaffer, C. Stewart, L. Swanson, and E. Wiltse. 2002. National Recovery Plan for the Piping Plover (Charadrius melodus). National Recovery Plan No. 22. Canadian Wildlife Service, Environment Canda, Ottawa, ON, Canada. Gratto-Trevor, C.L., and S. Abbott. 2011. Conservation of Piping Plover (Charadrius melodus) in North America: Science, successes, and challenges. Canadian Journal of Zoology 89:401–418. Island Nature Trust. 1997. Prince Edward Island plover atlas 1997. Island Nature Trust, Charlottetown, PE, Canada. Maslo, B., S.N. Handel, and T. Pover. 2011. Restoring beaches for Atlantic coast Piping Plovers (Charadrius melodus): A classification and regression-tree analysis of nest-site selection. Restoration Ecology 19:194–203. Mazerolle, M.J. 2012. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). Version 1.24. Available online at http://CRAN.R-project.org/ package=AICcmodavg. Accessed 11 February 2012. Plissner, J.H., and S.M. Haig. 2000. Status of a broadly distributed endangered species: Results and implications of the second International Piping Plover Census. Canadian Journal of Zoology 78:128–139. Northeastern Naturalist Vol. 21, No. 2 A.W. Boyne, D.L. Amirault-Langlais and A.J. McCue 2014 173 Prindiville Gaines, E., and M.R. Ryan. 1988. Piping Plover habitat use and reproductive success in North Dakota. Journal of Wildlife Management 52:266–273. R Development Core Team. 2011. R: A Language and Environment for Statistical Computing. Version 2.14.1. Available from http://www.R-project.org/. Accessed 5 February 2012. US Fish and Wildlife Service (USFWS). 1985. Determination of endangered and threatened status for the Piping Plover. Federal Register 50:50726–50734. Wentzell, N. 1997. Piping Plover habitat-manipulation proposal. Kejimkujik National Park, Seaside Adjunct, Parks Canada, Liverpool, NS, Canada.