Northeastern Naturalist 380 S.M. Baillie, D.H. Brunton, and A.W. Boyne 22001144 NORTHEASTERN NATURALIST 2V1(o3l). :2318,0 N–3o9. 63 Should I Stay or Should I Go: Influences on Roseate Terns’ (Sterna dougallii) Decisions to Move the Chicks Shauna M. Baillie1,3,*, Dianne H. Brunton2, and Andrew W. Boyne3 Abstract - In this study, we investigated the relative influence of habitat variables on the decision by Sterna dougallii (Roseate Tern) parents to move from (movers) or to stay at (stayers) the nest after chick hatch. At Country Island, NS, Canada, 75% of the 21 Roseate Tern breeding pairs in this study were movers. Using a model-selection approach, we found that the chicks were more likely to be moved from nest sites in cobble beach habitat with low vegetation height and high nest densities of congener terns. However, differences in reproductive parameters among movers and stayers were not statistically significant. Though we could not establish whether moving the chicks or staying were adaptive strategies, we provide firm evidence that Roseate Tern chicks are moved to areas of lower tern densities. Chicks move further away from other terns as they age, perhaps as a mechanism to avoid kleptoparasitism as their nutritional requirements increase. Based on our findings, Roseate Terns appear more likely to rear their chicks to fledging at the original nest site when nest densities of other tern species are low (≤ 0.02 nests/m 2) in highly vegetated areas. Thus, to enhance Roseate Tern productivity in places where they are endangered, such as Atlantic Canada, we suggest that species recovery programs place artificial nest cover, e.g., next boxes and wooden logs, in areas with potential for taller veget ation growth that are suboptimal nesting habitat for S. paradisaea (Arctic Tern)and S. hirundo (Common Tern). Introduction Colonial breeding birds, e.g., Laridae, Sternidae, and Sulidae, tend to nest in close proximity, in part, to gain protection from threats such as avian and mammalian predators (Lack 1968). This group protection can manifest in aggressive forms of anti-predator behavior, such as contact strikes and group mobbing (Burger et al. 1993, Hatch 2002, Nguyen et al. 2006, Nisbet 2002, Palestis 2005, Ramos and Del Nevo 1995, Shealer and Burger 1992). A negative consequence of colonial nesting and its associated competition, however, is increased conspecific and heterospecific aggression that can be lethal to young (McNamara et al. 2000). Many studies have shown that competitive adult aggression and infanticide function to prevent interference from unrelated chicks during food transfers and to avoid kleptoparasitism by other adults (De Leon et al. 2002; Gochfeld et al. 1998; Ramos 1998, 2003; Ratcliffe et al. 1997). Besnard et al. (2006) showed that nesting Larids responded to increases in agonistic displays by moving their chicks. Studies that compare reproductive success among nests with chicks that move or stay would add to our 1Canadian Wildlife Service, 45 Alderney Drive, Dartmouth, NS B2Y 2N6, Canada. 2Ecology, Behaviour and Conservation Group, Institute of Natural Sciences, Massey University, Private Bag 102-904, North Shore Mail Centre, Auckland 0630, New Zealand. 3Department of Biology, Life Science Centre, Dalhousie University, 1355 Oxford Street, Halifax, NS B3H 4R2, Canada. *Corresponding author - Shauna.Baillie@Dal.Ca. Manuscript Editor: Greg Robertson Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 381 understanding of whether moving the chicks away from high nest densities confers a fitness advantage. Sterna dougallii Montagu (Roseate Tern), S. sandvicensis (Latham) (Sandwich Tern), and some gulls and skimmers (J. Burger, Rutgers University, Piscataway, NJ, pers. comm.) are among the few semi-precocial species that move chicks tens of meters away from nesting aggregations once the chicks are mobile (Besnard et al. 2006, Brown and Morris 1995, Gochfeld et al. 1998, Stienen and Brenninkmeijer 1999, Whittam and Leonard 1999). Distinctively, Roseate Terns in Northwest Atlantic breeding colonies always nest among S. paradisaea Pontoppidan (Arctic Tern) and/or S. hirundo L. (Common Tern) (Gochfeld et al. 1998, Hatch 2002, Nisbet 2002). Generally, Arctic and Common Terns arrive at the colony first and begin courtship activity, while Roseate Terns arrive three weeks later and choose nest sites among the established Arctic and Common tern territories (Gochfeld et al. 1998, Toms et al. 2006). To our knowledge, of these three tern species, only Roseate Terns move their chicks to alternate brooding sites away from the original nest location shortly after hatching (Gochfeld et al. 1998, Hatch 2002, Nisbet 2002). Though Arctic and Common Terns eventually do move their chicks from the nest site, they wait until the chicks are older. Another major difference between Roseate Terns and neighboring congeners is that Roseate Terns prefer more complex structural cover around their nests, often sited under vegetation or objects, whereas Arctic and Common Terns usually nest in the open (Gochfeld et al. 1998, Hatch 2002, Nisbet 2002). Several studies have shown that after chick hatch, Arctic and Common Terns will attack adult Roseate Terns, kleptoparasitize chick meals, and even kill Roseate Tern chicks (Burger and Gochfeld 1991, Gochfeld et al. 1998, Palestis 2005). Roseate Terns faced with an aggressive and potentially life-threatening opponent must decide whether to flee or to fight (Krebs and Dawkins 1984). If passive avoidance, or hiding, is an alternate strategy to aggression, then nest density of neighbors, structural cover at the nest, and the potential of improved cover away from the nest may influence whether a Roseate Tern chick is moved away from or remains in the nest (Stienen and Brenninkmeijer 1999, Varpe and Tveraa 2005). In this study, we set out to investigate whether or not moving the chicks is an adaptive strategy against aggressive neighbors. To address this problem, we asked the following questions: 1) What habitat variables influence the decision on whether a chick is moved or stays in the nest? 2) What habitat variables change as a chick moves, i.e., do nest densities decrease with distance moved? and 3) Is there a reproductive advantage to moving the chicks? For the purposes of our study, Roseate Tern chicks that were moved by their parents one meter or more from the original nest were classified as “movers”. Chicks that remained within a 1-m radius of the nest until fledging were termed “stayers”. If movers are truly fleeing from high congener nest densities, then we would expect nest density to significantly decrease in the local environment of the mover. If fleeing has fitness benefits, then movers should experience higher survival to fledging than stayers. Roseate Terns are listed as endangered by the Species at Risk Act in Canada (http://www.sararegistry. gc.ca/species/speciesDetails_e.cfm?sid=40) and by the United States Fish and Northeastern Naturalist 382 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 Wildlife Service (USFWS) in the northeastern United States (http://ecos.fws.gov/ speciesProfile/profile/speciesProfile.action?spcode=B07O). Thus, the findings of this study have conservation implications key to the promotion of optimal nesting and chick-rearing habitat for less aggressive and often outnumbered species, such as the Roseate Tern in the Northwest Atlantic. Methods Field-Site Description Country Island is a 19-ha island located 5 km south of Davidson Point near Country Harbour in Guysborough County, NS, Canada (45o06'N, 61o32'W; Fig. 1). The island is low-lying with a vast, open heath-scrub plateau favored as a nesting location by hundreds of Arctic and Common Tern pairs in the southern half of the island. The north-northeastern perimeter of the heath plateau is lined by a fern plateau vegetated mainly with Rubus occidentalis L. (Black Raspberry), ferns, Angelica sp., and tall grasses, and there are small stands of Picea alba (White Spruce) to the north of this area. The east coast of southern Country Island is lined with rocky cobble ridges and boulder beaches, and there are grassy slopes and banks on the west and south coasts that give way to cobble beaches (Table 1). The study area was within the mixed tern colony in the treeless southern half of the island. At the time of this study in 2005, approximately 362 pairs of Common Terns, 846 pairs of Arctic Terns, and 41 pairs of Roseate Terns nested on the island (Toms et al. 2006). Twenty-one Roseate Tern nests were used in this study. Twenty nests of the 41 Roseate Tern nests located on Country Island had to be excluded due to either nest failure (n = 10), or because chicks disappeared before they could be followed for a minimum of 10 days after hatching (n = 10) (A.W. Boyne, unpubl. Table 1. Description of habitat categories, discussed in the text and listed in Table 2, associated with the mixed-species tern colony at Country Island, NS, during 200 5. Habitat category Description Cobble beach Low-gradient beach comprised of at least approximately 75% cobble sediment, and often patchily covered with mats of Beach Pea Lathyrus japonicus. Cobble beach ridge Steep-gradient cobble beach berm with up to 50% cobble sediment mixed with 50% rock and boulders. Boulder beach Low-gradient beach with substrate composition of at least 50% b oulder. Grassy slope Low-gradient grass-covered land mass. Grassy bank Where grass-covered land mass ends and there is a vertical drop to the beach. Lagoon bottomland Grassy wet depressions near lagoon or pond between the beach berms and plateaus. Fern plateau Flat upland plateau vegetated with mainly raspberry, fern, and tall grasses ~1 m in height. Heath plateau Flat upland plateau vegetated with mainly heath scrub, primarily Empetrum nigrum L. (Black Crowberry) and short herbaceous plants ~10 cm or less in height. Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 383 data). We conducted this research under the approval of the Migratory Bird Act and CWS Bird Banding Office permits, as well as approval of the Canadian Council on Animal Care. Nest substrate and density measurements To describe the relative composition of nest-site substrate surrounding each of the 21 Roseate Tern nests in this study, we estimated the percentage of boulders (≥ 50 cm), rock (49–20 cm), cobble (< 20 cm), and vegetation (or bare soil) within a 1-m radius of each Roseate Tern study nest. To describe the density of tern nests, Figure 1. Map showing the location of Country Island, Guysborough County, NS, Canada (45o06'N, 61o32'W). The stippled area on the Country Island inset illustrates the rocky shoreline. Rose diagrams represent four clusters of Roseate Tern nests and the cardinal direction and maximum distance (m) that chicks travelled from those nests. Distance travelled, up to 71 m, is represented by the vertical axis inside t he rose diagrams. Northeastern Naturalist 384 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 the number of Arctic, Common, and Roseate Tern nests were counted within a 5-m radius around each Roseate Tern nest. We pooled Arctic and Common Tern nest counts because Arctic Tern nests could not be distinguished from Common Tern nests when adults were not present, chicks often took cover as we approached, and the determination of species by eggs is unreliable. The Arctic/Common Tern and Roseate Tern nest counts were converted to density estimates (nests/m2) for data analysis. As the Roseate Tern parents started to move their chicks, we recorded composition of substrate and nest density at each subsequent chick location, or brooding site. Model selection and statistical analyses Logistic regressions in Program R 3.0.1 (R Development Core Team 2013) using the GLM package were used to model the decision to stay (0) or to move (1) as a function of the main of effects of the four variables (percent boulder, percent cobble, Arctic/Common Tern nest density, and vegetation height) and their two-way interactions, as well as the intercept-only (Null) model and a final model with all four variable terms. These four variables were reduced from an original 12 variables using a principal components analysis (PCA) performed in the PRINCOMP R package. We chose variables with the longest PCA loading vectors that explained PCA scores of Roseate Tern nests (Fig. 2). For each model, second-order Akaike’s information criterion (AICc), change in AICc (ΔAICc) and Akaike weights (wi) were calculated. The models with a ΔAICc between 0 Figure 2. Principal components analyses (PCA) used to reduce the number of variables from 12 to four. The longest loading vectors (arrows radiating outward) that explained clusters of Roseate Tern nest (numbers) PCA scores for the first two principal components (Comp. 1 and 2) were selected from a) the first PCA and entered into b) a second PCA. The ovals highlight the final four variables used to build logistic regression models (percent boulder, percent cobble, Arctic/Common Tern nest density, and vegetation height). S = stayer nests and M = mover nests. Analyses and graphs were generated in Program R (R Development Corp.) using the PRINCOMP package for R. Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 385 and 2 were considered the best explanatory models for the dependent variable (Burnham and Anderson 2002). Roseate Terns typically lay two eggs; therefore, to avoid confusion among chicks within the same nest, the first chick to hatch was marked with indelible ink as A-chick and the second as B-chick. To avoid pseudoreplication within multi-chick nests, only A-chicks were used in statistical analyses. For all linear regressions, arcsine transformations were used to meet assumptions of homoscedascity for nest density (nests/m2) data. Results were presented as arithmetic means ± standard error (SE), unless otherwise stated. The program used for all statistical analyses other than the logistic regressions and PCA was SPSS 17.0 for Windows (IBM, New York, NY). Chick movement observations and investigator disturbance Roseate Tern chick-movement patterns of 16 nests, comprised of 16 A-chicks and 8 B-chicks, were measured by observing the parents arriving from sea with fish meals in their bills for the young chicks and recording the location of the chick. During these observations, investigators were concealed from the birds by a portable blind on the edge of the tern colony. In order to minimize disturbance, after the chick disappeared from sight and its parents returned to sea, a researcher marked the chick’s last seen location with flagging tape and a hand-held global positioning system. Observers did not leave the blinds for more than five minutes at a time, and observation time was limited to two hours per day in any given area, once every three days. At a later time, the direction and distance of a flagged chick location from its original nest was measured using a compass and retractable tape measure. Stayers and movers received the same amount of investigator disturbance throughout the study. Because Roseate Tern colony attendance is typically low during the day, the chances of alarming the parents whose chicks we were observing was low. The observation blind was effective. For example, we on occasion could observe both parents luring chicks leapfrog fashion on the ground with fish meals in their bills; if the parents were aware of (or threatened by) observer presence, it is unlikely the Roseate Terns would have remained on the ground engaged in this activity. Though complete avoidance of the effects of investigator disturbance is not possible in any behavioral ecology study, we kept disturbance to a minimum, and we are confident the movements measured are a result of a reproductive strategy for the movers given breeding conditions rather than a response to investigator presence. Reproductive success and mass at last r ecapture Roseate Tern nests (n = 21) and their subsequent chick-brooding sites, were visited every two to three days to determine clutch size (eggs laid per nest), hatching success (chicks hatched per egg), and fledging success (chicks fledged per chicks hatched). We calculated breeding success as the product of hatching and fledging success. Eggs from which chicks had begun to hatch were checked daily for accurate hatch-date information. On the day after hatching, a stainless steel USFWS identification band was placed on the leg of each chick. Every two to three days for a maximum of four visits, chick mass was measured (to nearest 1 g) using a Pesola Northeastern Naturalist 386 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 Figure 3. Distribution of Roseate Tern nests on Country Island, NS, during the 2005 breeding season. The map shows the location of nests from which chicks had dispersed (black dots) and had not dispersed (white stars) after chick hatch. The area with hatched lines is the heath plateau where Arctic and Common Terns densely aggregate, the diagonal lines represent the lagoon and the stippled areas represent the cobble, rock, and boulder beaches. This figure is not drawn to scale, and is compressed in an east– west direction. spring scale. Any chick was considered to have fledged if it reached the age of 15 days, a standard measure of fledging success for Roseate Terns used by seabird monitoring programs throughout eastern North America (McKnight et al. 2005). The typical age of fledging is 22–24 days for Arctic Terns (Hatch 2002) and 22–28 days Common Terns (Nisbet 2002). Due to the difficulty of reaching each chick at exactly 15 days of age, we report mass at last recapture (g) at 15 ± 2 days of age in this study. Results Description of nest sites The 21 Roseate Tern nests in this study were found in four main clusters located within the outer perimeter of the main Arctic and Common Tern mixed-species colony (Fig. 3; Tables 1, 2). No Roseate Tern nests were located within the center of the Arctic and Common Tern nesting colony in the relatively open and flat heath plateau. Roseate Tern stayer nests (n = 5) were located under man-made nest shelters and/or flotsam (e.g., large cement slab, wooden log, wooden nest boxes) (Fig. 3, Table 2). Most mover nests (n = 16) were situated on open cobble beach, though three were built under human-made nest shelters and/or flotsam (e.g., old car tire, wooden logs). Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 387 Table 2. Roseate Tern mover (M) and stayer (S) nest site locations (using NAD83), number of chicks hatched from eggs and number of chicks fledged per nest on Country Island, NS, during the 2005 breeding season. Information on chick fate could not be determined for four nests, numbers 5, 22, 25, and 33. The habitat category descriptions can be found in Table 1. “Other” represents nest shelters and other man-made or natural structures used for nest sites. Nest Chicks Chicks ID # Northing Easting M/S hatched fledged Habitat category Other 1 45.10025 -61.54105 S 2 2 Fern plateau Nest box 2 45.10022 -61.54069 S 2 1 Fern plateau Log 5 45.10035 -61.54071 M 2 NA Fern plateau 6 45.10039 -61.54012 M 2 1 Cobble beach ridge 11 45.10056 -61.53999 M 1 0 Cobble beach ridge 13 45.10027 -61.54017 M 1 1 Cobble beach ridge 15 45.10096 -61.53980 M 1 1 Cobble beach Old car tire 16 45.10100 -61.53977 M 1 1 Cobble beach 17 45.10096 -61.53973 M 1 1 Cobble beach 18 45.10106 -61.53977 M 2 2 Cobble beach 19 45.10036 -61.54014 M 2 2 Cobble beach ridge 20 45.10025 -61.54095 S 2 2 Fern plateau Log 22 45.10022 -61.54074 M 2 NA Fern plateau Log 24 45.10096 -61.53975 M 1 1 Cobble beach 25 45.10088 -61.53973 S 1 NA Cobble beach 27 45.10084 -61.53975 M 2 2 Cobble beach 30 45.09949 -61.54200 M 2 1 Cobble beach 33 45.10096 -61.53973 M 1 NA Cobble beach 34 45.10101 -61.53967 M 2 0 Cobble beach 36 45.09951 -61.54232 M 1 1 Cobble beach Log 38 45.10117 -61.54528 S 1 1 Grassy bank Cement slab Influences on the decision to move or to stay The “cobble”, “vegetation height” and “Arctic/Common Terns + cobble” models were selected as the best explanatory models for whether a chick became a mover or a stayer (Tables 3, 4). The AICc model-selection best model, “cobble”, showed that Roseate Tern chicks were more likely to become movers if their nest was situated in cobble beach habitat (Fig. 4). The percentage of vegetation around stayer nests is obviously important from inspection of Figure 3. However, percent vegetation was highly correlated with vegetation height in our PCA, and we used vegetation height in the logistic regressions because of its stronger explanatory power (Fig. 2). The “vegetation height” model was the second best model according to our AICc model selection (Table 3), which revealed that Roseate Tern chicks are more likely to become stayers in tall vegetation (Fig. 5a). Vegetation height was >50 cm at stayer nests, and 100 cm tall at 3 out of 5 stayer nests. In contrast, the majority of mover nests were located in vegetation that was less than 50 cm tall (Fig. 5a). The third best model, “Arctic/Common Terns + cobble”, was not ranked much lower than the second model, “vegetation height”, according to the ΔAICc, and their wi were nearly identical (Table 3). Roseate Tern chicks were more likely to become movers both in cobble substrate and when tern densities are high. However, our parameter estimate for Arctic and/or Common Tern nest density, within the “Arctic/Common Terns + cobble” Northeastern Naturalist 388 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 Table 4. Model effects and coefficients for the best logistic regression models, and the intercept only (Null) model, on factors that influence the decision of whether or not a Roseate Tern chick moves away from or remains in the nest after hatching, retained by change in Akaike’s information criterion for small sample sizes (ΔAICc; Burnham and Anderson 2002). Abbreviated terms are df = degrees of freedom, wi = Akaike weights, and SE = standard error. Model effects df AICc Residual df Residual deviance ΔAICc wi Null 20 25.1 20 23.1 4.3 0.03 Cobble 20 20.7 19 16.7 0.0 0.31 Vegetation height 20 22.4 19 18.4 1.7 0.13 Arctic/Common Terns 20 22.5 18 16.3 1.8 0.13 Coefficients Parameter estimate SE Null 1.16 0.51 Intercept -0.001 0.67 Cobble 6.68 3.98 Intercept 4.18 1.65 Vegetation height -0.04 0.02 Intercept -0.21 0.77 Arctic/Common Terns 8.38 14.91 Cobble 5.89 4.18 model, was the only parameter estimate to have 0 (zero) within its SE bounds (Table 4), suggesting over-dispersion of this variable and weak effects at best. This result is not a surprise given the small sample of stayers. Nonetheless, the data shows that Arctic and/or Common Tern nest densities were three times higher around the nests of movers (mean: 0.06 ± 0.01 nests/m2; median: 0.06; n = 16) than stayers (mean 0.02 ± 0.02 nests/m2; median: 0.0; n = 5) (Fig. 5b). A statistical outlier, Nest #38, was the one exception to our findings that stayer nests had lower Arctic and/or Common Terns around them (Fig. 5b). Nest #38, was located on a grassy slope under a cement slab surrounded by relatively higher densities of Arctic and/or Common Table 3. Entire model set listed in order of rank based on change in Akaike's information criterion for small sample sizes (ΔAICc) and Akaike weights (ωi) (Burnham and Anderson 2002). Model variables: cob = percent cobble, veght = vegetation height, TERN5m = Arctic/Common Tern nest density within 5 m of study nests, and bould = percent boulder. Models (glm in R, family= “binomial”) AIC K AICc Δi e (-0.5Δi) ωi Rank cob 20.666 1 20.713 0.000 1.000 0.313 1 veght 22.365 1 22.412 1.699 0.428 0.134 2 TERN5m * cob 22.330 2 22.471 1.759 0.415 0.130 3 cob * bould 22.586 2 22.727 2.015 0.365 0.114 4 bould * veght 23.139 2 23.280 2.568 0.277 0.087 5 TERN5m 23.458 1 23.505 2.792 0.248 0.078 6 TERN5m * veght 24.121 2 24.262 3.550 0.170 0.053 7 cob * veght 24.302 2 24.443 3.731 0.155 0.048 8 Intercept only 25.053 0 25.053 4.340 0.114 0.034 9 TERN5m * bould 25.258 2 25.399 4.687 0.096 0.030 10 TERN5m * cob * bould * veght 26.185 4 26.667 5.954 0.051 0.016 11 bould 27.051 1 27.098 6.385 0.041 0.013 12 Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 389 Terns (0.1 nests/m2). When the AICc model selection is repeated excluding this outlier, the best model was the “Arctic/Common Terns + cobble” model (results available upon request). Changes in habitat variables as chicks moved Arctic and/or Common Tern and Roseate Tern nest densities within 5-m radii of chick brooding sites, after Roseate Tern chicks left their original nests, significantly decreased with chick age (F1,48 = 24.7, P < 0.001, r2 = 0.3 and F1,48 = 11.2, Figure 5. Comparison of a) vegetation height and b) nest densities surrounding Roseate Tern mover (n = 16) and stayer (n = 5) study nests. The box-and-whisker plots show the quartiles, medians, and outliers of nest-density data for pooled Arctic and/or Common Terns, as well as Roseate Terns, within a 5-m radius of the study nests. Figure 4. Comparison of nest-site characteristics among movers and stayers. Bar plots show the relative proportion of four Roseate Tern nest-site substrate variables (within a 1-m radius) among movers (n = 16) and stayers (n = 5). Northeastern Naturalist 390 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 P = 0.002, r2 = 0.2, respectively). The youngest and oldest chicks we observed to be moved were 4 and 25 days of age, respectively. We found that chicks kept moving away from (not toward) the nest as they aged and, as such, age was a good indicator of travel distance from the original nest. The (accumulative) straight-line distance from their original nests increased positively and significantly with chick age (F1,47 = 24.1, P < 0.001, r2 = 0.3). The total straight-line distance that movers traveled after leaving their nests ranged from 3 to 71 m, and the direction of travel varied depending on the original nest location (see Fig. 1). Roseate Tern movers on the eastern cobble beach and cobble beach ridge moved northeastward and eastward into increasingly boulderstrewn areas (Fig. 6a, b). Roseate Tern movers at the edge of the fern plateau moved northwestward to a few meters deeper within the fern plateau (Fig. 6c). Movers on the south cobble beach, however, moved westward within the same substrate type and composition (Fig. 6d). Reproductive success of movers versus stayers The data show that estimates for stayer hatching and fledging success were 4 and 14% higher, respectively, than for than movers; however, these differences were not significant (Table 5). Of the 32 chicks that hatched, 20 were confirmed to have Figure 6. Changes in nest- and subsequent brooding-site substrate composition for chicks that moved within the a) cobble and boulder beach north-northeast of the main tern colony, b) cobble ridge east of the colony, c) fern plateau, and d) southern cobble beach areas of Country Island (x-axis: chick age category in days [d]: 1 = 0–2 d, 2 = 3–5 d, 3 = 6–8 d, 4 = 9–11 d, 5 = 12–14 d and 6 = 15–17 d). Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 391 Table 5. A comparison of the reproductive parameters of Roseate Tern nests in which parents either moved the chicks (movers) or did not (stayers). Mass at last recapture is reported for chicks at age 15 ± 2 days. P-values were evaluated at significance level α = 0.05. Movers (n = 16 nests) Stayers (n = 5 nests) Test, statistic, P-value Hatching success (%) 96 (24/25 eggs) 100 (8/8 eggs) Mann-Whitney U, U = 37.5, P = 0.6 Fledging success (%) 74 (14/19 chicks) 86 (6/7 chicks) Mann-Whitney U, U = 23.5, P = 0.7 Breeding success (%)A 71 86 - # eggs laid/nest (mean ± SE) 1.6 ± 0.1 (25 eggs) 1.6 ± 0.2 (8 eggs) t-test, t19 = -1.4, P = 0.9 # chicks hatched/nest (mean ± SE) 1.5 ± 0.1 (24 chicks) 1.6 ± 0.3 (8 chicks) t-test, t19 = -0.4, P = 0.7 # chicks fledged/nest (mean ± SE) B 1.1 ± 0.2 (14 chicks/13 nests) 1.5 ± 0.3 (6 chicks/4 nests) t-test, t15 = -1.2, P = 0.3 Mass at last recapture (g) (mean ± SE)C 82.0 ± 2.6 (9 chicks) 87.3 ± 5.0 (3 chicks) t-test, t10 = -1.0, P = 0.3 ABreeding success is the product of hatching and fledging success . BSix chicks died, 6 went missing before 15 days of age, and 20 survived of the 32 chicks that hatched. The 6 chicks that went missing came from a total of 4 nests (3 mover nests with 5 chicks, and 1 stayer nest with 1 chick) before chicks reached the fledge age of 15 days (see Table 2). Therefore, only 17 of the 21 nests in this study were used for fledging success est imates. CAdditionally, we were not able to capture 8 more chicks during age 13–17 days to obtain mass at last recapture measurements, leaving only 12 chicks for mass at last recapture measurements out of the 20 chicks for wh ich fledging success estimates were calculated. Northeastern Naturalist 392 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 survived to fledge age, 6 chicks died and another 6 went missing before 15 days of age (Table 2). Because the fate of the six missing chicks could not be determined, fledging-success estimates for four (3 movers and 1 stayer) of the 21 nests in this study could not be obtained. Overall, 1.2 ± 0.6 SD (n = 17 nests) Roseate Tern chicks fledged per nest, and this parameter did not vary statistically among movers (n = 13 nests) and stayers (n = 4 nests) (Table 4). Chick mass at last recapture of movers (n = 9) on average was 5.3 g (6%) lower than that of stayers (n = 3), but this difference was not significant (Table 5). Discussion Habitat variables that influence the decision to move We found that cobble nest-site substrate, vegetation height, and the density of Arctic and/or Common Tern nests in cobble areas were the most important factors influencing whether Roseate Terns moved their chicks, or not, shortly after chick hatch. Generally, Roseate Tern chicks were more likely to become movers if nesting was initiated within cobble beach areas with short vegetation and higher congener nest densities. Cobble areas are attractive to and support high densities of nesting Arctic and Common terns, which may then cause some tern chicks to move in response to aggression. Of course, these areas are also more exposed to predators and weather, so aggression (and density) may not be the only cause of chick movements. Our empirical measurements, however, show that the tern nest densities surrounding Roseate Tern movers were three times higher than that of stayers. Seventy-seven percent of the Roseate Tern nests in our study were movers; thus, the majority of chicks were relocated by their parents within days of hatching. It is not uncommon at other colonies for some Roseate Tern parents to move chicks away from their nests, but this behavior exhibits considerable plasticity (Gochfeld et al. 1998, Ramos 2003). We wondered, why move the chicks at all? Perhaps the movers we observed in this study were avoiding kleptoparasitism and other aggressive interactions from congeners (Burger and Gochfeld 1991, Gochfeld et al. 1998, Palestis 2005, Stienen et al. 2001). At Country Island, NS, during our study, only the Roseate Tern and not the other two species of tern, Arctic and Common Terns, moved their chicks. Roseate Tern colony attendance is known to be lower than that of Arctic and Common Terns due to differences in foraging niche that require Roseate Terns to spend more time at sea (Burger and Gochfeld 1991, Gochfeld et al. 1998). Thus, Roseate Terns may hide the chicks to protect young from predators, while alleviating the constraint of constant parental vigilance at the colony (Donehower et al. 2007). Further, since other tern species are aggressive toward unattended chicks, it may not be advantageous for Roseate Terns to leave their chicks near congeners for predator protection. A comparative study that quantifies the level of physical aggression and kleptoparasitism of congeners and predators toward Roseate Terns would help disentangle the relative influence of predators and congeners on chick movement. An exception to our general finding that stayers nest amongst lower tern densities involved one of the five Roseate Tern stayer nests that we found under a large Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 393 concrete slab in the midst of Arctic and/or Common Tern densities almost twice as high as that for most movers. Through investigation of the substrate composition of Roseate Tern nests in this study, we found that stayers were most often situated within vegetation substrate rather than the cobble beaches. Correspondingly, all but one of the 15 Roseate Tern nests within cobble beaches became movers. Additionally, all stayers nesting within the vegetation substrate (4 out of 5 stayers) were found under an object, e.g., nest box or wooden log, which provided physical cover before the vegetation grew taller as the chick-rearing season advanced. Altogether, our findings suggest that structural cover and nest substrate play an important role in influencing the decision to move or to stay . Habitat changes as Roseate Tern chicks move Arctic and/or Common Tern nest densities decreased significantly within the local environment of Roseate Tern chicks as they moved away from their nest sites, and to subsequent brooding sites. Among these movers, we show that chicks tended to continuously move away from the nest with age, and many kept moving up to 71 m away. This behavior has been documented previously in a minority of semi-precocial species, where both parents physically lead chicks away from neighbors and displace them to areas with lower nest densities (Stienen and Brenninkmeijer 1999). Movers on the east coast cobble beaches of Country Island indeed moved away from high nest densities, but at the same time, percent substrate composition of boulders generally increased. Boulders and their deep crevices provide better hiding places than the low structural relief of cobble beaches. Similarly, movers in the fern plateau moved into greater physical cover within the dense and quickly growing foliage of the fern plateau. Despite the fact that we found no change in substrate composition on the southernmost cobble beach as chicks moved, sample sizes there were too low (n = 2) to draw conclusions. Our answer to the question of whether Roseate Terns are fleeing from their congeners after chick hatch is confounded by the fact that when chicks are led away from neighbors, physical cover increases. An elegant determination of whether parents are explicitly moving their chicks away from congeners, would involve a common garden experimental design in which paired controls of known mover and stayer groups are compared under varying nest-density conditions, yet within identical substrate and structural cover (see Aguilar et al. 2008, Maxson et al. 2007). Unfortunately, such an experiment is not feasible at Country Island considering the low numbers of Rosea te Terns and their endangered species status. A study on the reproductive consequences of moving or staying could be performed at Roseate Tern colonies with larger population sizes, e.g., in the northeastern United States. Is there a reproductive advantage to moving the chicks? If moving the chicks is an adaptive strategy, then there should be a reproductive advantage to moving the chicks. However, we found no statistically significant difference in fledging success or chick body condition between movers and stayers, nor in any other reproductive parameter. The data show that stayers, in fact, had Northeastern Naturalist 394 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 higher reproductive estimates than movers. This finding was consistent for both breeding success and chick mass at last recapture parameters. It could be that low sample sizes have confounded our results. Studies elsewhere were able to provide evidence of a fitness advantage to moving the chicks. For example, a study on evasion behavior in Larids experimentally confirmed that Sandwich Terns in the Netherlands maximized chick body condition by moving chicks away from kleptoparasitic Larus ridibundus (L.) (Black-Headed Gull) (Stienen and Brenninkmeijer 1999). Thus, for Roseate Terns it would seem that if the nest location is good it pays to stay, but if the location is bad it pays to move. The problem is that we cannot know what the reproductive success of the movers would have been if they had stayed. Also, it is unknown as to whether the same pairs of Roseate Tern parents at Country Island tactically move their chicks year after year. Future studies should address parental decision patterns regarding moving the chicks through time, e.g., with increased parental experience. At the very least, multiple years of study on known breeding pairs would improve sample sizes and statistical power, and may help determine whether moving the chicks is an adaptive strateg y. Conclusions Overall, our observations lead us to believe that Roseate Tern parents at Country Island were placing their chicks in lower nest densities and better physical cover after the chicks hatched. We could not infer a fitness advantage or disadvantage to moving the chicks mainly because movers and stayers were not found in the same habitat type, which precluded application of stringent study controls. To resolve the hypothesis that Roseate Terns “flee” aggressive congener neighbors as an adaptive strategy after the chicks hatch, future longitudinal studies would have to show that moving the chicks is reproductively advantageous and that nest densities decrease independently from changes in nest substrate as the chicks move. We recommend that Roseate Tern species recovery programs in Atlantic Canada encourage nesting in the areas where we have shown chicks to be less likely to move, i.e., areas other than within the center of cobble beaches, through placement of nest boxes at multispecies tern colonies in locations with potential for taller vegetation, e.g., at the grassy edges of raspberry scrub or fern fields. The employment of wooden logs and other flotsam could be useful as an alternative to next boxes. Finally, nestenhancement initiatives should prioritize areas within low densities (≤ 0.02 nests/ m2) of other tern species whenever possible. Acknowledgments This study would not have been possible without the advisement and support of the Canadian Wildlife Service (CWS) and CWS research-contract funding to S.M. Baillie. Many great thanks to Emily MacDonald, who assisted in the field through every stage of this project. Julie McKnight (J.M.) championed the logistics of opening and closing the island field season. Laura Simms and Bradley Toms provided their time and assistance in the field. Warm thanks to the Manthorne family for transportation of supplies and personnel to and from the island. The fieldwork was performed alongside an ongoing Country Island tern research and monitoring program established in 1998 in conjunction with Nova Northeastern Naturalist Vol. 21, No. 3 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 395 Scotia Department of Natural Resources, Dalhousie University, the Canadian Roseate Tern Recovery team, and the Canadian Coast Guard. Greg Robertson, Marty Leonard, and J.M. offered valuable suggestions on earlier versions of the manuscript. Further advice and suggestions were offered by Joanna Burger, Jeremy Hatch, Ian Nisbet, Norman Ratcliffe, and our anonymous referees. We conducted this research under the approval of the Migratory Bird Act CWS research and Environment Canada Bird Banding Office permits, as well as approval of the Canadian Council on Animal Care. Literature Cited Aguilar, T.M., R.I. Dias, A.C. Oliveira, and R.H. Macedo. 2008. Nest-site selection by Blue-black Grassquits in a Neotropical savanna: Do choices influence nest success? Journal of Field Ornithology 79:24–31. Besnard, A., N. Sadoul, and J.-D. Lebreton. 2006. First quantitative comparison of aggression between crèching and non-crèching Larid species. Waterbirds 29:481–488. Brown, K.M., and R.D. Morris. 1995. Investigator disturbance, chick movement, and aggressive-behavior in Ring-billed Gulls. Wilson Bulletin 107:140–152. Burger, J., and M. Gochfeld. 1991. Reproductive vulnerability: Parental attendance around hatching in Roseate Sterna dougallii and Common S. hirundo Terns. Condor 93:125–129. Burger, J., D.A. Shealer, and M. Gochfeld. 1993. Defensive aggression in terns: Discrimination and response to individual researchers. Aggressive Behaviour 19:303–311. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. Springer, New York, NY. De Leon, A., J.A. Fargallo, V. Polo, and J. Potti. 2002. Adult aggression during the postguard phase in the Chinstrap Penguin, Pygoscelis antarctica. Polar Biology 25:355–359. Donehower, C.E., D.M. Bird, C.S. Hall, and S.W.K. Kress. 2007. Effects of gull predation and predator control on tern nesting success at Eastern Egg Rock, Maine. Waterbirds 30:29–39. Gochfeld, M., J. Burger, and I.C.T. Nisbet. 1998. Roseate Tern, Sterna dougallii. No. 370, In A. Poole and F. Gill (Eds.). The Birds of North America. The Academy of Natural Sciences, Philadelphia, PA, and the American Ornithologists’ Union, Washington, DC. Hatch, J. 2002. Arctic Tern, Sterna paradisaea. No. 707, In A. Poole and F. Gill (Eds.). The Birds of North America. The Academy of Natural Sciences, Philadelphia, PA, and the American Ornithologists’ Union, Washington, DC. Krebs, J.R., and R. Dawkins. 1984. Animal signals: Mind reading and manipulation. Pp. 380–402, In J.R. Krebs and N.B. Davies (Eds.). Behavioral Ecology: An Evolutionary Approach, 2nd Edition. Blackwell Scientific, Oxford, UK. Lack, D. 1968. Ecological Adaptations for Breeding in Birds. Methuen, London, UK. Maxson, S.J., J.R. Fieberg, and M.R. Riggs. 2007. Black Tern nest-habitat selection and factors affecting nest success in northwestern Minnesota. Waterbirds 30:1–6. McKnight, J., J. Rock, and A.W. Boyne. 2005. Country Island tern restoration project. Annual report: Year 7. Canadian Wildlife Service, Environment Canada, Dartmouth, NS, Canada. McNamara, J.M., T.S.S. Kely, J.N. Webba, and A.I. Houston. 2000. A dynamic gametheoretic model of parental care. Journal Theoretical Biology 205:605–623. Nguyen, L.P., K.F. Abraham, and E. Nol. 2006. Influence of Arctic Terns on survival of artificial and natural Semi-Palmated Plover nests. Waterbirds 29:100–104. Nisbet, I.C.T. 2002. Common Tern Sterna hirundo. No. 618, In A. Poole and F. Gill (Eds.). The Birds of North America. The Academy of Natural Sciences, Philadelphia, PA, and the American Ornithologist’ Union, Washington, DC. Northeastern Naturalist 396 S.M. Baillie, D.H. Brunton, and A.W. Boyne 2014 Vol. 21, No. 3 Palestis, B.G. 2005. Nesting stage and nest defence by Common Terns. Waterbirds 28:87–94. R Development Core Team. 2013. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at http:// www.R-project.org. Ramos, J.A. 1998. Nest-site selection by Roseate Terns breeding on Aride Island, Seychelles. Waterbirds 21:438–443. Ramos, J.A. 2003. Intraspecific aggression by Roseate Tern adults on chicks in a tropical colony. Waterbirds 26:160–165. Ramos, J.A., and A.J. Del Nevo. 1995. Nest-site selection by Roseate Terns and Common Terns in the Azores. Auk 112:580–589. Ratcliffe, N., D. Richardson, S.R. Lidstone, P.J. Bond, C. Westlake, and S. Stennet. 1997. Host selection, attack rates, and success rates for Black-headed Gull kleptoparasitism of terns. Waterbirds 20:227–234. Shealer, D.A., and J. Burger. 1992. Differential responses of tropical Roseate Terns to aerial intruders throughout the nesting cycle. Condor 94:712–719. Stienen, E.W.M., and A. Brenninkmeijer. 1999. Keep the chicks moving: How Sandwich Terns can minimize kleptoparasitism by Black-headed Gulls. Animal Behaviour 57:1135–1144. Stienen, E.W.M., A. Brenninkmeijer, and C.E. Geschiere. 2001. Living with gulls: The consequences for Sandwich Terns of breeding in association with Black-headed Gulls. Waterbirds 24:68–82. Toms, B., S.M. Baillie, A.W. Boyne, L. Simms, and J. McKnight. 2006. Country Island tern restoration project. Annual report: Year 8. Canadian Wildlife Service, Environment Canada, Dartmouth, NS, Canada. Varpe, O., and T. Tveraa. 2005. Chick survival in relation to nest site: Is the Antarctic Petrel hiding from its predator? Polar Biology 28:388–394. Whittam, R.M., and M.L. Leonard. 1999. Predation and breeding success in Roseate Terns, Sterna dougallii. Canadian Journal of Zoology 77:851–856.