Time-Specific Patterns of Nest Survival for Ducks and Passerines Breeding in North Dakota

Abstract

In many bird species, survival can vary with the age of the nest, with the date a nest was initiated, or among years within the same nesting area. A literature review showed that patterns of survival vary in relation to nest age and date and are often contradictory. Inconsistencies could be a result of temporal variation in the environment or life-history differences among species. We examined patterns of nest survival in relation to nest age, date, and year for several duck and passerine species nesting at a single location in North Dakota during 1998–2003. We predicted that if environment shaped nest survival patterns, then temporal patterns in survival might be similar among three species of upland nesting ducks, and also among three species of grassland passerines nesting at the same site. We expected that survival patterns would differ between ducks and passerines because of relatively disparate life histories and differences in predators that prey on their nests. Nest survival was rarely constant among years, seasonally, or with age of the nest for species that we studied. As predicted, the pattern of survival was similar among duck species, driven mainly by differences in nest survival associated with nest initiation date. The pattern of survival also was similar among passerine species, but nest survival was more influenced by nest age than by date. Our findings suggest that some but not all variation in temporal patterns of nest survival in grassland birds reported in the literature can be explained on the basis of temporal environmental variation. Because patterns of survival were dissimilar among ducks and passerines, it is likely that mechanisms such as predation or brood parasitism have variable influences on productivity of ducks and passerines nesting in the same area. Our results indicate that biologists and managers should not assume that temporal environmental variations, especially factors that affect nest survival, act similarly on all grassland birds.

Résumé

Chez plusieurs espèces d'oiseaux, la survie peut varier avec l'âge du nid, la date d'initiation du nid ou entre les années dans une même aire de nidification. Une revue de la littérature a montré que les patrons de survie varient selon l'âge du nid et la date et qu'ils sont souvent contradictoires. Ces incohérences pourraient être le résultat d'une variation temporelle dans l'environnement ou de différences dans le cycle vital des espèces. Nous avons examiné les patrons de survie en fonction de l'âge du nid, de la date et de l'année pour plusieurs espèces de canards et de passereaux nichant au même endroit dans le Dakota du Nord de 1998 à 2003. Nous avons prédit que si l'environnement façonne les patrons de survie des nids, alors les patrons temporels de survie peuvent être similaires entre trois espèces de canards nichant sur des terres émergées et aussi entre trois espèces de passereaux des prairies nichant sur le même site. Nous nous attendions à ce que les patrons de survie soient différents entre les canards et les passereaux en raison de leurs histoires naturelles relativement disparates et de différences au niveau des prédateurs de leurs nids. La survie des nids était rarement constante entre les années, les saisons ou l'âge des nids pour les espèces que nous avons étudiées. Comme prévu, les patrons de survie étaient similaires entre les espèces de canards et étaient principalement influencés par les différences associées à la date d'initiation des nids. Les patrons de survie étaient également similaires entre les espèces de passereaux, mais la survie des nids était davantage influencée par l'âge du nid et la date. Nos résultats suggèrent que certaines variations dans les patrons temporels de survie des nids chez les oiseaux des prairies, telles que rapportées dans la littérature, peuvent être expliquées par des variations environnementales temporelles. En raison de différences dans les patrons de survie entre les canards et les passereaux, il semblerait que les mécanismes tels que la prédation ou le parasitisme de nid ont des influences variables sur la productivité des canards et des passereaux nichant dans la même région. Nos résultats indiquent que les biologistes et les gestionnaires ne devraient pas assumer que les variations environnementales temporelles, notamment les facteurs qui affectent la survie du nid, agissent de façon similaire sur tous les oiseaux des prairies.

NUMEROUS STUDIES OF avian nest success have shown that nest survival can vary with the age of the nest, with the date a nest was initiated, or across breeding seasons within the same nesting area. However, few consistent, time-specific patterns of nest survival appear to exist among ecologically similar species. For instance, nest survival in some species has been shown to vary discretely among the laying, incubation, or brood-rearing stages (e.g., Winter 1999, Davis 2003). Linear increases or decreases in relation to nest age (Stephens et al. 2005, Churchwell et al. 2008) and either peaks or troughs in the period between nest initiation and fledging (Pieron and Rohwer 2010) also have been described. Even more complicated cubic age patterns have been identified for passerine species (Grant et al. 2005). Nest survival for some species increased or decreased linearly with date during the breeding season (Klett and Johnson 1982, Winter et al. 2004) or exhibited a curvilinear pattern between early and late nesting (Pieron and Rohwer 2010). Some studies reported no influence of age, date, or year on nest survival (e.g., Emery et al. 2005, Koper and Schmiegelow 2007). Unfortunately, interpretation of time-specific variation in nest survival and comparisons across studies are often difficult because of confounding effects related to life history and spatial sources of environmental variation.

Nest survival can be considered a product of a species' life-history traits coupled with environmental influences (e.g., landscape metrics, weather events, management activities). There are ecological reasons to expect that species nesting syntopically might respond similarly to forces that influence nest survival. For example, it is generally assumed that grassland passerines benefit from management actions intended to protect or improve prairie habitats for ducks, although scant information is available to evaluate this assumption (Ball et al. 1994, Koper and Schmiegelow 2007). Conversely, one could expect differences to exist among duck or songbird species or between these two taxonomic groups, based on body size, life-history traits, or the degree of dependence on upland prairie during the nesting cycle (i.e., passerines require grasslands for feeding, nesting, and brood rearing, whereas ducks utilize grasslands mainly for nesting).

Although many studies have evaluated effects of age and date on nest survival, relatively few were designed for this purpose, and even fewer discuss probable mechanisms behind patterns of survival in the context of a species' life history or environmental factors. Indeed, our review of the literature did not locate a single contemporaneous study of time-specific effects on nest survival in ducks and passerines residing in the same habitat. Here, we present results of a study of nest survival in six species of grassland birds nesting in North Dakota: Gadwall (Anas strepera), Mallard (A. platyrhynchos), Blue-winged Teal (A. discors), Clay-colored Sparrow (Spizella pallida), Savannah Sparrow (Passerculus sandwichensis), and Bobolink (Dolichonyx oryzivorus). Our primary objective was to compare time-specific patterns of nest survival among species within ducks and passerines, and then examine similarities and differences between ducks and passerines. Specifically, we predicted that more ecologically distant species (e.g., passerines vs. ducks) would differ in patterns of survival because these two groups have relatively disparate life histories.

Methods

Study area.—Our study was conducted on the 23,900-ha J. Clark Salyer National Wildlife Refuge (NWR) in Bottineau County, North Dakota (about 48°45′N, 100°50′W). We eliminated many spatial environmental effects (e.g., area and landscape effects) on nest survival by restricting our work to a single 400-ha tract comprising seven contiguous units that varied from 41 to 69 ha (Grant et al. 2011). All units had similar vegetation structure and composition at a bird-territory scale (e.g., <5 ha) and similar habitat-area and edge metrics (e.g., distance to cropland edge) at a local landscape scale (Grant et al. 2010, 2011). The plant community was mixed-grass prairie, consisting of a needle grass—wheat-grass (Stipa—Pascopyrum) association intermingled with two exotic grasses, Kentucky Bluegrass (Poa pratensis) and Smooth Brome (Bromus inermis). Grasslands were interspersed with short (<1.0 m) brush dominated by Western Snowberry (Symphoricarpos occidentalis). Climate is semiarid to subhumid continental, with average monthly temperatures ranging from -15°C in January to 20°C in July. Annual precipitation was variable among years 1998–2003 but was similar to the long-term average of 43 cm.

Nest monitoring.—We systematically searched for and monitored nests of grassland birds from mid-April to late July each year (1998–2003). To find nests, two observers pulled a 25-m weighted rope that had metal cans attached every 0.5 m, which flushed adult birds from their nests (Davis 2003). Searches were conducted between 0700 and 1600 hours CST. Each of seven study plots was systematically searched six to eight times during the breeding season. We also located nests fortuitously and, for passerines, by observing behaviors of adult birds (Grant et al. 2005). We marked nests with survey flags placed 3–5 m to the north and south of the nest, with the top of the flag placed just above the average height of the vegetation. To estimate age, we candled eggs from each nest during laying or incubation (Weller 1956, Lokemoen and Koford 1996) or aged nestlings (passerines only) from voucher photographs of known-age young (e.g., Jongsomjit et al. 2007). We monitored duck nests every 5-10 days until nest fate (i.e., successful or unsuccessful) could be determined. We defined a successful duck nest as having at least one hatched egg, determined by the presence of detached shell membranes (Girard 1939). We visited passerine nests every 2 to 5 days until young were near the estimated age of fledging; we then visited nests daily to minimize uncertainty in assigning final nest fates (Grant et al. 2005). A passerine nest was considered successful if at least one host young survived to minimum age of fledging (we omitted nests parasitized by Brown-headed Cowbirds [Molothrus ater]). We used behaviors of the parents (e.g., alarm calling, carrying food), presence of young near the nest, nestling age at the previous visit, evidence of nest disturbance, evidence of nestling mortality, and presence of feces or feather scales of young at the nest to aid in assigning nest fate (Davis 2003, Grant et al. 2005).

Modeling nest survival.—Daily nest survival is the probability that a nest survives a given day, conditional on it being active at the beginning of that day. We used the logistic-exposure method (Shaffer 2004a) to investigate time-specific patterns in daily nest survival. Nests survived the interval between visits if at least 1 egg or nestling was alive on the latter visit, or if at least 1 egg hatched (ducks) or young fledged (passerines) on or before the final visit. Date, nest age, and nest stage were time-varying explanatory variables. Logistic-exposure models assume a logistic relation (i.e., logit link function) between daily nest survival and explanatory variables. With one exception, assumptions are the standard ones used in other common approaches to modeling nest survival (e.g., Dinsmore et al. 2002, Rotella et al. 2004). Assumptions include the following: (1) nest fates are correctly determined and are independent among nests, (2) nest discovery and revisitation do not affect survival, (3) all nest visits are recorded, (4) nest ages are correctly determined, (5) nest checks are conducted independently of nest fate, and (6) daily survival rates are homogeneous and conditionally independent from day to day. The one exception is in how logistic-exposure models handle time-specific variables. Logistic-exposure models do not allow daily, time-specific variables, but instead use interval-specific values. This structure has negligible effect on estimates of age or date effects, especially when intervals between visits are short, as they were in the present study (Rotella et al. 2004, Shaffer 2004a). For each interval, we assigned midpoint values of age and date and the nest stage observed at the beginning of the interval. Logistic-exposure models were fitted with PROC GENMOD (SAS Institute 2004) following procedures of Shaffer (2004a). Nests that had already either failed or fledged young when found, those that showed evidence of egg depredation when found, and all nests that likely were abandoned because of investigator influence (e.g., nests abandoned within 2 days of being found) or that contained eggs broken by investigators were excluded from nest survival analyses. Nests abandoned for reasons other than investigator influence were treated as nest failures.

We used an information-theoretic approach and Akaike's information criterion adjusted for sample size (AIC_c) to identify models (i.e., those with the lowest AIC_c score) that best described the data (Burnham and Anderson 2002). We present the Akaike model weight (w_i) for each candidate model (Burnham and Anderson 2002). The Akaike weight for a particular model represents the weight of evidence for that model, given the data and the candidate models considered (Burnham and Anderson 2002:75). We used the effective sample size (n_eff; Rotella et al. 2004) to compute AIC_c (i.e., n_eff = total number of days that nests were known to survive + total number of intervals in which a failure occurred).

Recognizing the inconsistent patterns in nest survival reported in the literature, we adopted a flexible modeling approach that allowed us to describe age and date patterns for each species individually. We considered six model structures for assessing the relation between daily nest survival and nest age: (1) constant survival, (2) stage-specific constant survival, (3) survival linearly related to age, (4) stage-specific linear survival, (5) survival nonlinearly related to age via a quadratic polynomial function, and (6) survival nonlinearly related to age via a cubic polynomial function (passerines only; Fig. 1). We considered three model structures for describing the relation between daily nest survival and date: (1) constant survival, (2) survival linearly related to date, and (3) survival nonlinearly (quadratic polynomial) related to date. Combinations of the above model structures gave rise to 15 and 18 models for duck and passerine species, respectively. Quadratic age models for ducks and cubic age models for passerines were based on the premise that one inflection point may be necessary to accommodate the transition from laying to incubation, and an additional inflection point may be needed to describe the transition to brood rearing in altricial species (e.g., Grant et al. 2005, Pieron and Rohwer 2010).

We used a three-step approach to identify parsimonious models that recognized not only effects of nest age and date on daily nest survival, but also appropriately included effects of year and study plot. Plot effects were treated as fixed because there were only seven plots and they were contiguous, not a random sample from some larger population. In step 1, we evaluated the performance of 18 (15 for ducks) basic model forms after adding effects of plot and year (1998–2003) to account for unmeasured sources of variation characterizing individual study plots and years. We advanced models with ΔAIC_c < 8 to the next step, unless a particular model was within 0–2 units of a hierarchically simpler model and had essentially the same log-likelihood as the simpler and better model (Burnham and Anderson 2002:131). In step 2, we investigated whether year and plot were necessary. Specifically, we evaluated the performance of models advancing from step 1, all of which included effects for year and plot, against otherwise identical models except that year, plot, or both year and plot were excluded. Step 3 was necessary only if year effects were present in models advancing from step 2. In step 3, we added interaction terms with year to models that received support from step 2 to determine whether age and date patterns were consistent among years. From this third and final step, we identified a final set of models for each species; we report step-3 model-selection results for models with w_i ≥ 0.10.

Model selection based on AIC will support variables whose 85% confidence intervals exclude zero (Arnold 2010). Thus, to illustrate effects of nest age on survival, we show model-averaged daily survival rates with 85% confidence intervals, unless survival varied among years. In these cases, to avoid excessive clutter, daily survival rates are shown for each year without confidence intervals. To illustrate effects of nest survival in relation to date, we use “period survival” to refer to the probability that a nest survives the period from nest initiation through hatching of the first egg (ducks) or fledging of the first young (passerines). Period survival is the product of daily survival rates for each day in the period. When daily survival varies with ordinal date, period survival varies with nest initiation date (Shaffer and Thompson 2007). We used equation 2 in Shaffer and Thompson (2007) to estimate period survival in relation to nest initiation date from model-averaged daily survival rates described above (for example SAS code, see Shaffer 2004b).

From results of the above analyses, we created an omnibus model form flexible enough to preserve observed variation in each species, but general enough to apply to all species, by combining terms from the top model for each species (but see below for an exception involving the Bobolink). We did this separately for duck and passerine groups, and we added species main effects to the omnibus model form and compared the resulting model with a second model that contained both main effects and interactions between species and time-specific variables. To account for species differences in laying intervals, we rescaled the age variable so that incubation commenced at time 0 for all species. Age and date were then standardized to a mean (± SD) of 0.0 ± 1.0 to permit meaningful assessments of their relative effects.

Similar to the analysis comparing species within taxonomic groups, we identified an omnibus model form sufficiently flexible to approximate observed time-specific patterns for ducks and passerines. To create the omnibus model, we combined terms from the top model for each duck or passerine species. We compared a baseline model containing an effect of taxonomic group (passerine or duck), within-group species effects, and effects of age and date against a model in which taxonomic group was allowed to interact with temporal effects. Because passerines are altricial and because we could not account for posthatch survival in ducks, we excluded exposure days during the nestling stage, limiting comparisons to laying and incubation periods.

Results

We determined the fates of 2,927 nests of the six species between 1998 and 2003. Individual species modeling revealed uncertainty regarding the most appropriate time-specific model for each species, with two to four models each carrying ≥0.10 model weight and collectively carrying 0.72–1.00 model weight (Table 1). Despite the model-selection uncertainty, some general patterns emerged. Daily nest survival varied with either age of the nest or date within the breeding season for all species except, possibly, the Bobolink (Table 1; ΔAIC_c ≥ 37 for constant survival model for species other than the Bobolink). Year effects were present in top models for all species except the Bobolink. The Gadwall and Bobolink were the only species to show any evidence of plot effects.

Duck nest survival patterns.—The top two models for Gadwalls captured 78% of total model weight and indicated that nest survival was variable by date and year (Table 1). A difference between laying and incubation stages was suggested by the remaining model for Gadwalls, but the estimated difference in daily survival rate between stages (0.02) was slight (Fig. 2). We observed an interaction between year and date (e.g., date = -0.69 ± 0.43 in 2001 and date = 0.54 ± 0.32 in 2002; here and below, we present means ± SE). Both date and age or stage effects were present in all top models (cumulative model weight = 1.00) for the Mallard (Table 1). Mallard nest survival increased with age ( age = 0.12 ± 0.05) and declined with date ( date = -0.02 ± 0.01; Figs. 2 and 3). Date effects were apparent in both top models (cumulative model weight = 0.72) for Blue-winged Teals, and the general pattern was one of survival decreasing with nest initiation date (e.g., date = -0.05 ± <0.01 in 2000), with the exception of 1998 date = 0.01 ± 0.02; Fig. 3).

We used an omnibus model with quadratic age and quadratic date effects that varied with year to examine species-specific effects. We found much stronger support for the model without species interactions (AIC_c = 3,165.8, w_i = 1.0 vs. AIC_c = 3,185.7, w_i = 0.0) and therefore concluded that age- and date-related survival patterns were similar among duck species. Linear age effects ranged from -0.09 ± 0.13 to 0.23 ± 0.13, and 85% confidence intervals overlapped 0.0 in all years. Linear date effects varied from -0.66 ± 0.11 to 0.20 ± 0.16. Survival decreased with date in 2000 ( date = -0.66 ± 0.11) and 2001 ( date = -0.32 ± 0.10); 85% confidence intervals overlapped 0.0 in remaining years.

Passerine nest survival patterns.—The top models for Claycolored Sparrows (cumulative model weight = 1.0) and Savannah Sparrows (cumulative model weight = 0.92) provided clear evidence for age-specific survival patterns approximated by a cubic relationship (Table 1). Age-specific patterns of nest survival were nearly identical for the two species (for both: age = 0.52 ± 0.13, age^*age = -0.05 ± 0.01, and age^*age^*age = 0.001 ± <0.001) and exhibited a general pattern of lowest survival soon after initiation, increasing sharply until early—middle incubation, gradually decreasing to a low shortly after hatch, and then increasing during the nestling period until fledging (Fig. 4). Survival of Clay-colored Sparrow nests did not vary with date, but survival of Savannah Sparrow nests gradually declined from May through July ( date = -0.009 ± 0.004; Table 1 and Fig. 5). Absolute survival rates were variable among years, but the pattern of survival was similar among years for both species (Figs. 4 and 5; ΔAIC_c ≥; 12 for models with year × age or year × date interactions). In contrast to the results for Clay-colored and Savannah sparrows, the top two models for Bobolinks (cumulative model weight = 0.69) provided evidence of stage-specific linear effects (Table 1). The general pattern was one of increasing survival during laying ( age = 0.23 ± 0.40) and incubation ( age = 0.05 ± 0.06), an abrupt drop in survival immediately following hatch and then a sharp increase until fledging ( age = 0.24 ± 0.08; Fig. 4). However, the constant-survival model was within 2.3 AIC_c units of the top-supported model (Table 1).

We chose the cubic-age, linear-date model form with additive year effects as the omnibus model form for comparisons of survival patterns among passerine species. Although the top model for Bobolinks showed a stage-specific linear pattern rather than a cubic-age pattern, the sample size for Bobolinks was small (n_eff = 1,207) compared with sample sizes for either Clay-colored Sparrows (n_eff = 7,673) or Savannah Sparrows (n_eff = 5,925). The latter species both demonstrated the cubic-age pattern; thus, we chose it when forming the omnibus model. We reasoned that if the cubic-age pattern was inadequate for Bobolinks, our comparison would show a difference among species. The model without species interactions was better supported (AIC_c = 4,887.2; w_i = 0.94) than the model with interactions (AIC_c = 4,892.6; w_i = 0.06); thus, we concluded that survival patterns were similar among passerine species. Parameter estimates for linear age = -0.28 ± 0.07), quadratic ( age*age = 0.44 ± 0.07), and cubic age components ( age^*age^*age = 0.27 ± 0.04) had 85% confidence intervals that did not overlap 0.0. The estimated date effect ( date = -0.07 ± 0.04), although negative, was barely distinguishable from 0.0.

Comparing duck and passerine nest survival patterns.—We used the quadratic-age, quadratic-date model form with year interactions as our omnibus model form for comparisons of duck and passerine groups. The model with group × age and group × date interactions (AIC_c = 5,567.8; w_i = 0.99) outperformed the model without an interaction between ducks and passerines (AIC_c = 5,576.5; w_i = 0.01). Passerines generally showed stronger effects of age, whereas ducks showed stronger effects of initiation date. For example, standardized parameter estimates in 2003 were age = -0.36 ± 0.17, age^*age = 0.03 ± 0.02, date = -4.2 ± 1.9, and date^*date = 0.001 ± 0.001 for ducks; and age = 1.4 ± 0.41, age^* age = -0.03 ± 0.01, date = 0.019 ± 3.27, and date^*date = -0.00 ± <0.001 for passerines.

Discussion

Nest survival in relation to age of the nest.—Except for the Mallard, age effects on duck nest survival were not evident in our study. Survival increased with nest age for Mallards in our study and for ducks elsewhere in North Dakota and Alaska (Klett and Johnson 1982, Grand 1995, Garrettson and Rohwer 2001, Stephens et al. 2005). Two studies identified age-specific patterns of survival strikingly similar to that of Mallards in our study (Hoekman et al. 2006: fig. 6, Traylor et al. 2004: fig. 3). By contrast, survival was not related to nest age for ducks breeding in Canada (Emery et al. 2005) or in Alaska (Walker et al. 2005). Nest survival was more complicated for ducks breeding in eastern North Dakota, where a curvilinear pattern was evident from initiation to hatching (Pieron and Rohwer 2010).

In our study, nest survival varied with age of the nest for Clay-colored and Savannah sparrows, and possibly for Bobolinks. Nest survival was related to age in a curvilinear manner for Clay-colored and Savannah sparrows. The Bobolink pattern of survival was generally similar to that of the sparrows, except during incubation (Fig. 4). Nest survival varied with nest age for six species of passerines breeding in southern Saskatchewan (Davis et al. 2006). Notably, Davis et al. (2006) also observed a cubic age effect for Baird's Sparrows (Ammodramus bairdii), Savannah Sparrows, Western Meadowlarks (Sturnella neglecta), and Chestnutcollared Longspurs (Calcarius ornatus), similar to that we identified for Clay-colored and Savannah sparrows. The cubic age effect was first reported for Vesper Sparrows (Pooecetes gramineus) and Clay-colored Sparrows breeding at a location <40 km from our study site (Grant et al. 2005). Kerns et al. (2010) reported the cubic age effect for Clay-colored Sparrows and Savannah Sparrows breeding at sites <100 km west of our study, as did a concurrent study of Clay-colored Sparrows nesting in the same region (R. K. Murphy unpubl. data). Skagen and Yackel Adams (2010) recently described the cubic age effect for Lark Buntings (Calamospiza melanocorys) breeding in the shortgrass prairie region. An age-specific pattern of nest survival was identified for Savannah Sparrows (R. K. Murphy unpubl. data), similar to that observed for Bobolinks in our study. Lloyd and Martin (2005) described a linear decline in survival from initiation through fledging for Chestnut-collared Longspurs breeding in northeastern Montana, as did Churchwell et al. (2008) for Dickcissels (Spiza americana) nesting in Oklahoma.

Nest survival in relation to date of the nesting season.—In our study, survival of ducks was usually highest for nests initiated earlier in the breeding season. Likewise, nest survival was higher early in the season for Northern Pintails (Anas acuta) breeding in Alaska (Flint and Grand 1996). By contrast, several studies report higher late-season nest survival for ducks breeding in the Prairie Pothole Region (Klett and Johnson 1982, Greenwood et al. 1995, Garrettson and Rohwer 2001, Koper and Schmiegelow 2007) and elsewhere in Alaska (Grand 1995, Walker et al. 2005), similar to our data for Gadwalls in 1998 and 2002 and for Bluewinged Teals in 1998. Data were mixed for another study in prairie Canada, where nest survival was higher early in the season for cover types receiving habitat management (e.g., delayed grazing, delayed haying, rotational grazing), but higher late in the season for nests placed in unmanaged cover types (e.g., idled grassland, shrubland, woodland; Emery et al. 2005). Hoekman et al. (2006) found little evidence of seasonal variation in survival of Mallards nesting in Ontario, except for some cases in which nests were destroyed by haying late in the season. The pattern of nest survival was curvilinear with date for ducks breeding in eastern North Dakota; survival was highest for nests initiated during the middle of the breeding season (Pieron and Rohwer 2010).

We observed decreasing nest survival with date for Savannah Sparrows. Grant et al. (2005) observed a similar, but more compelling, pattern of survival in relation to initiation date for Clay-colored and Vesper sparrows, as did R. K. Murphy (unpubl. data) for Claycolored and Savannah sparrows nesting in western North Dakota. Nest survival also decreased seasonally for Clay-colored Sparrows and Bobolinks, but not for Savannah Sparrows, at tallgrass prairie sites located in North Dakota and Minnesota (Winter et al. 2004). By contrast, Davis et al. (2006) observed increased survival with date for passerines nesting in Saskatchewan. Nest survival was unrelated to date for passerines breeding in southern Alberta (Koper and Schmiegelow 2007).

Comparing survival patterns within and between ducks and passerines.—As expected, patterns of survival associated with nest age and initiation date differed between ducks and passerines. Nest survival in ducks was influenced more by initiation date than by nest age, similar to data reported for ducks elsewhere (Flint and Grand 1996, Emery et al. 2005, Walker et al. 2005). We predicted that upland nesting ducks subjected to the same environmental influences would show similar patterns in nest survival in relation to age of the nest and initiation date. Age and date effects were generally similar among Mallard, Gadwall, and Blue-winged Teal nests. The pattern of survival in response to initiation date best illustrates this prediction, with all species having higher nest survival early in the season (Fig. 3). This pattern suggests that there may have been benefits for ducks nesting early in the season at our study site. Fitness may increase for females or broods if higher-quality nests are initiated early in the season (Blums et al. 2005), when the density of nests or of nest predators is lower (Nams 1997, Grant et al. 2006), or if predators such as small mammals are associated with specific vegetation height and density parameters that change throughout the season (e.g., Dion et al. 2000). Early-hatched offspring may have survival advantages over later-hatched young during both the pre- and postfledging periods (Rohwer 1992, Amundson and Arnold 2011). Successful early nesting may confer additional advantages for females, in that they (1) expend less energy producing only one clutch, (2) have less exposure to predators because renesting is not a factor, (3) have larger clutches, and (4) have more time for molting and recovery prior to migration in the fall (Devries et al. 2003, Krapu et al. 2004, Emery et al. 2005, Arnold et al. 2010). Subtle differences in clutch size, length of incubation, nest attendance, nest defense, and nest-site selection exist among species and can influence nest survival (Duebbert et al. 1986, Klett et al. 1986, Afton and Paulus 1992, Gloutney et al. 1993, Feldheim 1997, Gunness and Weatherhead 2002). The similar effects of age and date on nest success among the duck species in our study suggest that subtle differences in life histories were not particularly important in shaping patterns of nest survival.

In our study, passerines had absolute rates and patterns of time-specific nest survival that were more similar among years than in ducks, and age was a consistently more important predictor of nest survival than initiation date. Age effects also were more important than seasonal effects for passerines nesting elsewhere in North Dakota (Grant et al. 2005) and Saskatchewan (Davis et al. 2006). Effects of nest age may reflect activities at the nest that are intrinsic to altricial species (e.g., food begging by young, delivery of food, removal of fecal sacs). Because of these constraints, altricial species may be less plastic (than ducks) in terms of their response to environmental factors that affect nest survival. A cubic age effect is more common than anticipated by Grant et al. (2005), given that few subsequent studies have looked for this effect (Davis et al. 2006, Kerns et al. 2010, Skagen and Yackel Adams 2010, R. K. Murphy unpubl. data, present study).

Patterns of survival were dissimilar among ducks and passerines, which suggests that mechanisms, especially predation, differentially influenced productivity of syntopically breeding populations of ducks and passerines (Koper and Schmiegelow 2007). Although ducks and grassland passerines share certain nest predators, the basic suite of predators differ between taxonomic groups. Midsized mammalian carnivores are important predators of duck nests (Sargeant et al. 1993, Sovada et al. 2000), whereas ground squirrels (Spermophilus spp.), small mammals (e.g., Peromyscus spp., Microtus spp.), and Brown-headed Cowbirds are more important predators of passerine nests (Pietz and Grantors 2000, Grant et al. 2006). Parasitism by Brown-headed Cowbirds is an additional factor that influences nest survival in grassland passerines (Davis and Sealy 2000, Grantors et al. 2001, Kerns et al. 2010). Differences in the relative importance of time-specific variables between ducks and passerines suggest that duck populations may be more amenable to management actions that influence survival (e.g., reducing predators or enhancing nesting habitat early in the nesting season; Emery et al. 2005) than passerines. Our results indicate that biologists and managers should not assume that temporal environmental variation, especially factors that affect nest survival, act similarly on all grassland birds (e.g., Koper and Schmiegelow 2007).

Acknowledgments

Funding was provided by the U.S. Fish and Wildlife Service (USFWS), Region 6 Divisions of Refuges and Wildlife and Migratory Bird Management; and by the U.S. Geological Survey (USGS), Northern Prairie Wildlife Research Center. We are grateful to S. K. Davis for sharing his experience and methods for locating and monitoring nests of grassland passerines. S. L. Jones provided support throughout the project. E. M. Madden and G. B. Berkey were pivotal in completing the study. C. M. Aucoin, B. Bedard, B. Cofell, K. J. Berg, K. M. BausChristopherson, M. J. Friel, J. Gault, S. L. Finkbeiner, K. M. Hansen, S. Kuzyk, R. A. Laubhan, S. A. Marshall, R. T. McManus, M. P. Nenneman, and J. J. Thury assisted with data collection. Comments by M. A. Sovoda, T. W. Arnold, and two anonymous reviewers improved the manuscript. T. K. Buhl assisted with analyses. The findings and conclusions in this article are those of the authors and the USGS and do not necessarily represent the views of the USFWS. The use of trade, product, or firm names in this article is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Literature Cited

Author notes