Plasticity in nesting adaptations of a tidal marsh endemic bird

Abstract If individuals can perceive and manage risks, they may alter their behaviors based on prior experience. This expectation may apply to nest site selection of breeding birds, for which adaptive behavioral responses may enhance fitness. Birds that nest in tidal marshes have adapted to the challenges posed primarily by periodic, monthly tidal flooding and secondarily by predation. We investigated adaptive responses in nesting behavior of the saltmarsh sparrow (Ammospiza caudacutus), an obligate tidal‐marsh‐breeding bird, using 536 nests monitored across 5 years. Using linear mixed effects models, we tested whether nest characteristics differed among nests that were successful, depredated, or flooded, and we investigated whether females made changes in nest structure and placement according to outcome of their previous nesting attempt. Nest characteristics differed among females with different nest fates. Fledged and depredated nests were built higher in the vegetation and in higher elevation areas of the marsh than those that flooded. Successful nests had more canopy cover and were comprised of a lower proportion of high marsh vegetation (Spartina patens) than those that were flooded or depredated. Females with nests that failed due to flooding constructed subsequent nests higher in the vegetation and at higher elevation than those that were successful in their prior attempt, consistent with a response to previous experience. Eighty‐five percent of females renested within the average home range core area distance (77 m), indicating a high degree of nest placement fidelity. Females for which nests were depredated in their prior nesting attempt renested at a greater distance than females for which the previous nesting attempts were successful. Our findings suggest saltmarsh sparrows exhibit plasticity in nesting behavior, which may be important for balancing selective pressures in a dynamic environment. This plasticity, however, is insufficient to enable them to adapt to the increased flooding predicted with sea‐level rise.

, and it may be one method of mitigating risks of nest failure. If breeding females can both perceive and manage risks, they may be expected to alter their nesting behaviors based on prior experience. For example, in areas of high nest predation, it may be adaptive to move away from risky sites or make alterations in nest structure to reduce the risk of failure (Beckmann, Biro, & Martin, 2015;McAuley, Longcore, & Sepik, 1990). Conversely, by exhibiting fidelity to the same breeding location yearly, birds may gain advantages that are positively correlated with breeding success, such as knowledge of food availability and predator densities (Chalfoun & Schmidt, 2012). Adaptive responses to multiple environmental factors require that a female learns specific nest site attributes and their vulnerability to specific environmental factors (Marzluff, 1988). Multiple studies on nest site selection in diverse species have found that individuals will use information on their previous breeding success to choose a current breeding site (Gavin & Bollinger, 1988;McAuley et al., 1990;Beletsky & Orians, 1991;Haas, 1998;Hunter et al. 2016). This informed nest site fidelity combined with plasticity in nest structure could lead to greater reproductive success (Chalfoun & Schmidt, 2012;Switzer, 1997).
Nest site selection is directly correlated with reproductive success in tidal marsh nesting birds, which experience high levels of nest failure due to tidal flooding (Gjerdrum, Elphick, & Rubega, 2005;Storey, Montevecchi, Andrews, & Sims, 1988). In tidal marshes, water levels fluctuate predictably with the 28-day lunar cycle, producing peaks in tide height with the new and full moons and resulting in one to two consecutive days each cycle when marshes are flooded almost entirely (Redfield, 1972). Tidal marsh specialists have adapted to the challenges of living in this harsh environment, with the trade-off being limited interspecific competition and abundant resources (Greenberg et al., 2006;Reinert, 2006). Birds that nest in tidal marshes exhibit some adaptations to mitigate reproductive consequences of tidal nest flooding, such as placement of nests at a height that exceeds the tides but is low enough to the marsh surface to minimize predation, nest repair or egg retrieval behaviors, rapid postflood renesting, and timing of nesting attempts to avoid peak seasonal tides (Greenberg et al., 2006;Reinert, 2006). Given the strong selection pressure imposed upon tidal marsh birds by periodic tidal flooding, and variation in the relative risks of predation and flooding both within and across marshes (e.g., Hunter et al., 2016, Ruskin et al., 2017a, it may also be adaptive for them to assess risks and exhibit plasticity in nesting behavior (Forstmeier & Weiss, 2004).
The saltmarsh sparrow (Ammospiza caudacutus; Figure 1) is a tidal marsh specialist with reproductive behaviors shaped by the unique selective pressures of tidal marshes. The breeding system is highly promiscuous and is characterized by scramble competition among males for mating access (Hill, Gjerdrum, & Elphick, 2010;Walsh, Maxwell, & Kovach, 2018). Neither sex is territorial and males provide no parental care (Post & Greenlaw, 1982).
Females build ground nests a few centimeters above the surface of the marsh, just above the height of the mean high water level (Gjerdrum et al., 2005;Shriver, Vickery, Hodgman, & Gibbs, 2007; see Figure 2). Nests that are initiated within three days of a high spring tide are most likely to be successful by avoiding peak tidal flooding that occurs with the 28-day lunar cycle (Gjerdrum et al., 2005;Greenlaw & Rising, 1994;Shriver et al., 2007). Nests are woven into the vegetation-primarily Spartina patens, S. alterniflora, and Juncus gerardii-and located in areas of higher elevation within the marsh, which are less affected by tidal flooding (Gjerdrum et al., 2005;Shriver et al., 2007). Nest site selection is spatially random with respect to other nesting females (Bayard & Elphick, 2010;Gjerdrum et al., 2005), suggesting that structural characteristics of the nest itself may be more important to success than where the nest is located within the preferred nesting habitat (Gjerdrum et al., 2005). While prior research has found vegetation cover characteristics to be important in nest site selection, neither these vegetation characteristics, nor nest height (the height at which the nest is placed above the ground) and substrate elevation (the elevation of the marsh surface; see Figure 2) have been found to consistently influence nest success (Gjerdrum et al., F I G U R E 1 Adult saltmarsh sparrow in Spartina patens. Photo courtesy of Nancy J. Landry F I G U R E 2 Illustration of saltmarsh sparrow nest, woven into marsh vegetation, showing structural features of the nest in relation to the marsh surface. For nests to escape flooding, tidal water levels will rise above the marsh surface, but will not exceed the height of the nest. The canopy may help to retain eggs in a flooded nest 2005; Humphreys, Elphick, Gjerdrum, & Rubega, 2007;Ruskin, Hodgman, Etterson, & Olsen, 2015;Shriver et al., 2007).
As for other tidal-marsh-breeding birds, nest failure of saltmarsh sparrows may be a result of either flooding or depredation by avian or mammalian predators, suggesting there is a trade-off between flooding and predation risks that influence nesting behaviors (Greenberg et al., 2006;Ruskin et al., 2017a). By nesting higher in the vegetation, a female's nest is more visible and susceptible to predation, while nesting closer to the marsh surface will increase the likelihood of nest flooding. Further, females may construct a canopy above the nest, which can function in retaining eggs during flooding events and may reduce predation through additional vegetation cover and concealment (Humphreys et al., 2007). A lack of territoriality, pair-bonding, or paternal care (Greenlaw & Rising, 1994;Hill et al., 2010) suggests that nest locations may be labile and females may potentially renest in new locations during the breeding season in response to nesting failure. Whether females can perceive the mortality risks facing their nests, as well as the characteristics that are associated with these risks, and alter their nesting behaviors in response is unknown.
We investigated characteristics of female nest site selection and sought to determine whether females modified their nesting behaviors as a function of prior experience. We collected data on nest characteristics, including location, marsh elevation, and structural features, to address the following questions: 1. Do nest characteristics differ among nests that are successful and those that fail due to flooding or predation? We sought to investigate differences in nest site elevation, nest height, canopy presence, and vegetation composition among nests and compared them with ultimate fates of fledged, flooded, and depredated. We predicted successful nests would be in areas of higher elevation and have characteristics that simultaneously minimize the effects of predation and flooding (e.g., presence of canopy cover, nest height above tidal flooding level, and vegetation composition primarily of S. patens).
2. Do female saltmarsh sparrows exhibit nest placement fidelity across years? We aimed to determine whether female saltmarsh sparrows returned to the same locations to nest in future years based on the fate of their previous year's nesting attempt. We hypothesized females would renest within their prior home range core area across subsequent years due to the advantages of local resource knowledge, such as fine-scale vegetation and elevation patterns.
3. Do females make changes in their nest site location and structure based on previous experiences and the outcomes of their prior nesting attempts-whether it failed due to predation or flooding or was successful? We sought to explore changes in nest placement and structure relative to a female's prior nesting success.
We expected females to alter the location, elevation, and structural characteristics of their nests in a way that would increase nesting success relative to the outcome of their prior nesting attempt. We predicted that, within a breeding season, females whose nests failed due to flooding would make structural changes to subsequent nests that would mitigate flooding failure, such as an increase in nest height, canopy cover, changes in vegetation composition, or renest in a higher elevation area of the marsh.
Furthermore, we predicted that females whose nests failed due to predation would renest at a greater distance from their previous nest, rather than modifying structural characteristics.

| Nest placement and monitoring
Sites were systematically searched for nests 2-3 times per week during each of the three annual nesting cycles. Once found, nests were revisited every 3-4 days until the nesting attempt was completed.
Nests were assigned one of three ultimate nest fates: fledged, failure due to flooding, or failure due to depredation, following Ruskin et al. (2017a,b). Nests were considered fledged if one individual from the nest reached fledging age (i.e., nests could experience partial failure prior to fledging). Nests were considered to have failed due to flooding if one or more eggs or nestlings were found immediately outside of the nest cup or the nest contents were cold and wet with the female no longer attending the nest. Nests were deemed depredated when there were signs of predatory activity, such as disturbed nests or partial remains of nestlings, and none of the chicks fledged.
Immediately upon finding a nest, we recorded structural nest measurements, including nest height (lip to ground and bottom to ground), canopy presence (coded as a categorical variable with 3 levels: full, partial, or absent), canopy cover (percentage of nest cup visible from above), and cup depth (Ruskin et al., 2017a,b; see  Gallagher, Somers, Grant, & Seliskar, 1988) were considered low marsh habitat (Roberts et al., 2017). For testing associations of nest characteristics with nest fate, we used data from 536 nests monitored across the 5-year study (see below for statistical analysis methods).
In 2015, we used a Trimble TSC3 data logger with real-time kinematic (RTK) R10 Glonass-enabled antenna (Trimble Navigation Limited, Sunnyvale, CA) and CORS base station correction (Keystone Precision, Durham, NH) to determine the surface elevation (see Figure 2) at found nests. Using these methods, we collected elevation data from 120 nests (out of the total of 536 across the 5-year study), including 12 females with >1 nesting attempt, to test for elevational influences on nest fate and changes in nest elevation over repeat nesting attempts. Because the total height of a nest above the tidal water level is a function of both the marsh surface elevation at the nest site and the height of the nest in the vegetation, we also considered a new parameter-total nest height (surface elevation + bottom nest height). We calculated total nest height by adding the marsh surface elevation at the nest to the bottom nest height.
Due to a small sample of females with multiple nesting attempts for which the first nest was depredated (n = 14), we used only females with prior nest fates of fledged (n = 74) and flooded (n = 32) to investigate changes in marsh elevation and total nest height.
To track multiple nesting attempts from the same individual throughout the breeding season and across years, females attending nests were captured at the nest using two 12-m, 36-mm mesh mist nets and uniquely marked with a numbered USGS aluminum leg band. Nest locations were recorded using a GPS unit (Garmin GPSmap 76Cx). To address hypotheses related to female nest placement across years, we used only nesting attempts from consecutive F I G U R E 3 Locations of the four marshes in the northeastern United States where saltmarsh sparrows nesting data were collected during 2011-2015 years (n = 78 females). For hypotheses about nest movement within a breeding season, we used all females with multiple nesting attempts within that breeding season (n = 45 females); however, we could not always be certain the nesting attempts were immediately sequential as there may have been intervening nesting attempts that were not found. We used GENALEX 6.5 (Peakall & Smouse, 2012) to calculate the Euclidean distance between nest locations for a given female within and across breeding seasons.

| Statistical analyses
To test our hypothesis that nest site selection and characteristics differed among nests of different fates, we compared nest site elevation and structural characteristics among nests of the three ultimate fates of fledged, flooded, and depredated. To address any potential confounding effects of nest discovery bias, that is failed nests are available for discovery for fewer days than successful nests (Mayfield 1961), we first examined our data for differences in initial discovery date for nests of different fates. Most nests were found early in the nesting cycle (mean = 8.5 days) although there were slight differences in the initial discovery dates across fates (GLMM, χ 2 = 14.44, p < 0.001). On average, flooded (7.5 days) and predated (8.1 days) nests were found 1-2 days earlier than fledged nests (9.9 days). Given that the magnitude of this difference was small and in the opposite direction of potential concern-that is, failed nests are available for discovery for fewer days than successful nests, yet we found them slightly earlier than successful nests-and that we had a very intensive nest sampling effort, we think it is reasonable to assume that discovery bias is not of concern in our study and it is appropriate to make inferences about the three nest fate categories.
We tested for differences in nest structural characteristics (bottom nest height, lip nest height, cup depth, canopy presence and cover, and vegetation composition) across the three fate categories (fledged, flooded, and depredated) using GLMMs with assumed normal errors in the package "nlme" (Pinheiro et al. 2016) in R statistical software (R Development Core Team, 2014). We tested for differences in the structural characteristics among sites (the four different study marshes) and locations (inland vs. coastal sites), using ANOVA.
Nest height and canopy presence varied significantly among locations (p < 0.001), and we therefore included location as a fixed effect covariate for mixed models with those two factors. We also incorporated female identity as a random effect to account for variation among females. To test for differences in each structural characteristic among fates, we used post hoc Tukey's highly significant difference tests for pairwise comparisons in the R package "multcomp" (Hothorn, Bretz, & Westfall, 2008). Description of all models and covariates is found in Appendix 2.
We tested for site-specific differences in marsh surface elevation using an ANOVA and Tukey's highly significant difference tests for pairwise differences. We found a significant difference in site elevation for all pairwise combinations except Eldridge Marsh and Parker River; therefore, we included site as a covariate in subsequent analyses. We then tested for differences in elevation at the nest site among nests of the three fate categories, using a GLMM with nest fate as a fixed effect, female identity as a random effect, and nest elevation as the response variable. Similarly, we also tested for differences in total nest height (surface elevation (cm) + bottom nest height (cm)) across the three fates.
Next, we examined whether sparrows made changes in nest structure in response to their previous nest outcome. We used GLMMs with a random effect of female identity, fixed effect of prior nest fate, and response variable of difference in nest characteristic measurement to test for changes in nest structure between nest attempts as a function of nest fate. For these tests, we only used the numerical measurements (nest height, cup depth, vegetation composition, percent visible), and we calculated the difference between measurements for the first and second nesting attempts.
To determine whether females moved to areas of higher elevation in renesting attempts following a flooding failure versus success, we used a GLMM with change in elevation (elevation nest 2 -elevation nest 1) as the response variable and prior nest fate as a fixed effect. Similarly, we tested for changes in total nest height by comparing total height nest 2 -total height nest 1 across the three initial nest fates. We assessed the significance of our fixed effect of fate using F tests and type II sums of squares. Description of all models and covariates for this set of analyses is found in Appendix 3.
It is possible that a relationship in changes in nesting behavior and prior fate could arise from a statistical artifact resulting from regression to the mean, due to the bounded nature of the nest data (i.e., there are minimums and maximums for all measured characteristics; Bailey et al., 2017). To rule out this possibility for any significant observed relationships, we quantified the relationship between changes in nesting behavior and prior nest fate in randomized data, following Bailey et al. (2017). To do so, we first randomized the nest order with 5,000 iterations using MATLAB.
We then calculated the change in nest behavioral measurement for the randomized data and ran the same model test for statistical associations. We plotted the distribution of p-values for the randomly generated data and calculated the percentage of models with p < 0.05.
Lastly, we examined whether female nest site fidelity was influenced by prior nest fate. We used GLMMs with assumed normal errors to test for a relationship of nest fate and distance moved by females between nesting attempts. Models included distance between nesting attempts as the response variable, fate of the prior nesting attempt as a fixed effect, and female identity as a random effect for both within-year and across-year comparisons. Significance was assessed using F tests and type II sums of squares. To test whether females showed nest placement fidelity within and across years, we used a one-sided t test to determine whether the mean distance moved between nesting attempts was greater than the average diameter of the home range core area of female saltmarsh sparrows (77 ± 13 m; Shriver, Hodgman, Gibbs, & Vickery, 2010).

| RE SULTS
We located and monitored 556 nests across the four study sites from 2011 to 2015. We assigned fates to 536 nests, of those 393 (73%) also had information on the attending female. Sample sizes and female capture rates per site and year are in Appendix 1. Only 30 (5%) nests experienced partial depredation and partial flooding (1 partial depredation and 29 partial flooding). We removed the single partially depredated nest from further analyses. Nests that experienced partial flooding were similar in structure to nests that failed due to flooding, and we therefore included them within the flooded fate category (results were the same for analyses conducted with and without the partially depredated nests). We obtained renesting data (1-5 repeat nesting events) for 98 pairs of nests from 78 individual females within a single year and for 57 pairs of nests for 45 females across years; these were the sample sizes used in the nest fidelity analyses. Of these renesting attempts, we had complete data on nesting characteristics and fate from 64 females for 84 pairs of nests (some females had >1 renesting attempt) used in models assessing changes in nest structure and location as a function of prior experience (nest fate).

| Do females exhibit fidelity in their nest placement?
Within and across breeding seasons, we found high fidelity in nest-

| Do females make changes in their nest site selection and structure based on the outcome of their prior nesting attempts?
Nest characteristics differed between the first and second nesting attempts for females (n = 84 pairs of nests), according to the fate of their first nest (Table 2; Appendix 3). We found a significant difference in the change in marsh elevation for the renesting locations of females whose prior nest was successful and those that were flooded (GLMM, χ 2 = 5.19, p = 0.02; Figure 6a). Females for which the prior nest failed due to flooding chose a site with slightly higher elevation for their subsequent nest (x = 2.7 ± 1.8 cm), while those that were successful in their previous attempt did not choose a location with a different elevation for their subsequent nest (x = −0.3 ± 1.5 cm).  Results of the randomization test on the relationship of bottom nest height and prior nest fate revealed that 63 (1.26%) out of 5,000 randomized data models yielded a significant result (Figure 7), suggesting that the significant result in our observed nest data could not be explained by random chance alone.
Although we also observed a significant relationship between the changes in nest elevation by prior nest fate, our sample size was

| D ISCUSS I ON
Nest flooding and depredation risks are important selective factors that have shaped the evolution of reproductive strategies in marsh nesting birds (Greenberg et al., 2006;Picman, Milks, & Leptich, 1993). Saltmarsh sparrows have been associated with tidal marshes for several million years (Greenlaw, 1993); during this time, they have evolved strategies to mitigate flooding risks temporally, through rapid renesting following nest failure during peak inundation periods. It has been hypothesized that timing of nest initiation relative to the flood tides is the primary adaptive nesting strategy, rather than selection of nest sites that spatially minimize flooding risk (Gjerdrum et al., 2005;Shriver et al., 2007). Our findings suggest that nest characteristics, including height, canopy presence, and elevation within the marsh, may also influence nesting success and that females exhibit plasticity in nesting behavior, which may be important for balancing selective pressures in a dynamic environment.
We found that flooded, depredated, and successful saltmarsh sparrow nests differed in height, canopy presence, and marsh surface elevation. Specifically, successful nests were built higher in the vegetation, had more canopy cover, and were in higher elevation areas of the marsh than those that flooded. Several previous studies failed to find a consistent relationship between nest structures and nest success, despite strong evidence of nest site selection preferences for characteristics that confer resistance to flooding (Gjerdrum et al., 2005;Humphreys et al., 2007;Ruskin et al., 2015;Shriver et al., 2007). Our findings are consistent with these nest characteristics having a fitness effect, and the large sample size in our study (536 nests vs. 69 −160 nests in prior studies) may have allowed more power to detect small effect sizes. Furthermore, the high resolution of the RTK elevation measurements may have enhanced the power of our comparisons in elevation of nest locations.
Another possible explanation may lie in the inclusion in our study of two inland, estuarine sites (Chapman's Landing and Lubberland Creek marshes, located in the Great Bay Estuary of New Hampshire), which have dampened tidal regimes compared to the coastal sites on which prior studies have taken place. Sparrows nesting on these marshes have been found to be among the most productive with among the lowest rates of nest flooding compared to other marshes in the species range (Ruskin et al., 2017a,b). It is possible that in these inland marshes, the adaptive (fitness) benefits of the nest site preferences are realized, whereas the extremely high rates of nest flooding experienced in coastal marshes today (which are increasing due to sea-level rise; Bayard & Elphick, 2011) override any potential adaptive benefits of nest structure differences for the range of effect sizes observed.
Tidal height has strong effects on flooding probability, and successful nests withstand higher tide heights than those that fail due to flooding (Bayard & Elphick, 2011). It is intuitive that nests that are placed higher in the vegetation will be less impacted by tidal flooding than those placed lower, and damage from multiple flooding events may have cumulative impacts (Bayard & Elphick, 2011;Walsh et al., 2016). The height of the nest in relation to tidal height, however, is While successful nests were placed higher in the vegetation than those that flooded, they were also lower than those that were depredated, supporting the idea that there is a trade-off between predation and flooding (Greenberg et al., 2006;Ruskin et al., 2017a). A trade-off between predation and flooding along a gradient of nest height also occurs in the closely related seaside sparrow (Ammospiza maritimus). Hunter et al. (2016) found that nests located higher in the vegetation had a greater probability of predation and lower probability of flooding, while those located lower in the vegetation had a lower predation probability and higher flooding probability. While optimal nest height may entail a trade-off between predation and flooding, the presence of a nest canopy may confer advantages against both threats, by providing a structure that prevents egg loss while also providing concealment (Humphreys et al., 2007). Our results were consistent with nest canopy providing protection against egg loss, as successful nests had more of a canopy presence than flooded nests (although this difference was not statistically significant when modeled using percent canopy cover rather than categorical presence of canopy, likely due to inconsistencies in observers' visual estimates of percentages in the field).
Our results did not provide clear support for the role of canopy in providing protection from predation, as successful and depredated nests did not differ significantly in canopy cover (although there was an apparent nonsignificant trend for less canopy presence in depredated nests), nor did females with depredated nests increase the amount of canopy cover in their subsequent renesting attempt.
The latter may at least in part be due to relatively small sample sizes for depredated nests, as flooding is the far greater source of nest failure in our study marshes. Additionally, while nest canopies may be beneficial against avian predators that use visual cues (e.g., crows, gulls), they may be less advantageous in protecting nests against mammalian predators (e.g., mesocarnivores), which rely strongly on olfactory cues. In our study marshes, the presence of both avian and mammalian predators may confound the expected association between nesting success and canopy cover for depredated vs. successful nests.
Making repairs or changes to nest structure or placement increases the likelihood of success in tidal marsh nesting birds (Beckmann et al., 2015;Burger, 1979). We found female saltmarsh sparrows altered their nest placement and structure in subsequent nesting attempts based on the fate of their previous nest. Behavioral plasticity, via adjustments to nest structure and site selection based on immediate environmental conditions, may be important mechanisms for species persistence in the dynamic tidal marsh environment (Refsnider & Janzen, 2012). By exhibiting plasticity in nesting behavior following a cause-specific nest failure, saltmarsh sparrows may be able to respond to the selective pressure that is stronger at a given time or place, given variation in predation and flooding risks (Ruskin et al., 2017a). We found differential changes in nest height and elevation following a failure due to flooding or predation.
Specifically, females that experienced nest flooding increased the height of their nest and placed their nests in higher elevation loca- Seaside sparrows nested at lower height in years with high predation risks, but increased nest height following failure due to flooding in years with unpredictable tidal flooding caused by wind events .
Plasticity in nesting behavior can also take the form of shifts in habitat selection to areas with a different vegetation composition or different risk of threat, for example, predation (Chalfoun & Martin, 2010). Here, we found support for our hypothesis that females that experienced predation in their previous nesting attempt renested at a greater distance than those that were successful or flooded. By renesting farther from a previous nesting attempt, a female may be able to find an area with lower predator densities and different vegetation composition, such as taller vegetation or different species, which may increase concealment. In contrast, it may be more beneficial for females that experience nest flooding to renest near their previous nest and make structural changes rather than to renest in a different location, if timing of reproduction in relation to the tidal cycle and nest structure is generally more important than nest placement within the marsh (Shriver et al., 2007).
Females are faced with nest site selection trade-offs across seasons as well. With a limited nesting window, it may be more advantageous for females to spend less time scouting for new nesting locations upon arrival on the breeding grounds and quickly begin nesting using information gained from prior nesting experiences.
This informed nest site fidelity would allow them to benefit from awareness of local environmental factors such as food abundance, tidal regime, or predation pressure (Chalfoun & Schmidt, 2012;Switzer, 1997), and, on a finer scale, elevation differences and vegetation patterns. Across years, 85% of saltmarsh sparrow females in this study returned to nest within their previous home range core area, with some renesting within a few meters of a previous nest. This high degree of nest placement fidelity may be informed by prior success in relation to flooding risk, predation pressure, and accessibility to mating opportunities. Informed site fidelity may also confer reproductive advantages in this highly promiscuous mating system (Hill et al., 2010), if mate accessibility varies spatially across the marsh.
The rapid rate of global climate change likely limits adaptive genetic changes at a population level (Berteaux, Reale, McAdam, & Boutin, 2004;Refsnider & Janzen, 2012). Mechanisms occurring at the individual level, however, such as behavioral plasticity, may provide some capacity for adapting to novel environmental effects (Refsnider & Janzen, 2012). Plasticity in nesting behavior of saltmarsh sparrows may allow them to quickly adapt to modest changes in tidal regime, habitat loss, and fragmentation.
This plasticity, however, is likely insufficient in the face of sealevel rise, which reduces high marsh habitat and modifies tidal regimes that disrupt synchronous breeding of sparrows with the 28-day tidal cycle. The direct impacts of sea-level rise are predicted to reduce the reproductive success of saltmarsh sparrows (Bayard & Elphick, 2011), which have already declined at a rate of 9% annually from 1998 to 2012 (Correll et al., 2016), leaving the species vulnerable to extinction within the next 50 years (Field et al., 2016(Field et al., , 2017. The apparent adaptive capacity of saltmarsh sparrows, however, may enhance their ability to respond to management interventions targeted to mitigate nest flooding. On the other hand, the high degree of nest site fidelity that we observed in this study raises questions about how readily sparrows will respond to marsh restoration, such as large-scale thin level sediment deposition, which would make nesting habitat unavailable for one or more breeding seasons. This would likely require sparrows to move to a new nesting location, potentially several hundred meters or more away from the location to which they have shown fidelity. Accordingly, the potential benefits and limits of sparrow's nesting plasticity should be evaluated in future restoration efforts.

ACK N OWLED G M ENTS
Primary funding for this project was provided by the United States

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
BB and AIK designed the study, with help from KMO and JW; BB and JW collected field data; BB analyzed the data, with guidance from AIK and JW; KMO provided logistical support; BB wrote the manuscript with substantial input from AIK; all authors edited the manuscript.

DATA ACCE SS I B I LIT Y
Data used in this publication, including nest structural features and fates, changes in these features and renesting distances between subsequent nesting attempts, are archived in DRYAD https://doi. org/10.5061/dryad.q64b0n2.