4. DISCUSSION

The four species of Arctic-breeding birds in our comparison exhibited strong variation, both within years across species and within species across years, in multiple metrics related to the timing of and investment in reproduction. This variation was also expressed during the period of chick growth, but the species-specific responses during this phase of the breeding cycle often contrasted with those expressed during the pre-lay and nesting phases. Our prediction that spring temperatures and snow cover would have a greater effect on the timing of nest initiation and clutch size in geese compared to semipalmated sandpipers and longspurs was generally supported and may reflect the differing role that endogenous reserves play among the taxa during reproduction (see below). Our prediction that arthropod abundance would have relatively little effect on the growth of altricial longspur chicks was also supported, but we found no evidence that resource abundance influenced the growth of semipalmated sandpiper chicks or snow goose goslings. Furthermore, the effects of subspathacea biomass on the growth of brant goslings were opposite of our predictions, emphasizing the role of factors other than resource abundance in juvenile growth.

4.1 Annual adjustments to reproductive phenology and investment

Of the four species in our comparison, only semipalmated sandpipers meaningfully adjusted their arrival date in response to environmental conditions (snow cover), likely a reflection of the species’ dependence on exogenous reserves derived from arthropod prey that only become available as snow melts and temperatures warm (Holmes & Pitelka, 1968; MacLean & Pitelka, 1971). In contrast, snow geese and longspurs consistently arrived the earliest at our study site each spring. Snow geese employ their robust bill and specialized foraging techniques to access nutritious sub-surface roots and shoots (Iacobelli & Jefferies, 1991), while longspurs can subsist on seeds (Custer & Pitelka, 1978) prior to the emergence of arthropods. Brant are relatively dependent on food resources that emerge only as temperatures warm and melting snow exposes appropriate foraging substrates (Lewis et al., 2020), but unlike semipalmated sandpipers, brant also carry significant endogenous reserves that can serve as buffers when food resources are inaccessible (Hupp et al., 2018).
Subsequent phases of the reproductive cycle of the species in our comparison further reflected the species’ positions on the endogenous-exogenous continuum. Larger bodied geese are capable of carrying comparatively larger endogenous reserves to the Arctic that they can then invest in egg production and incubation (Klaassen et al., 2006). Conversely, shorebirds and passerines derive virtually all egg nutrients and reserves for self-maintenance after their arrival to the breeding area. All species advanced nest initiation with earlier snowmelt, but the effect was strongest for geese. Likewise, warmer spring temperatures resulted in earlier nest initiation by geese, but not for semipalmated sandpipers and longspurs. In years when Arctic phenology is advanced, geese can use their reserves to begin egg development in late migration or shortly after their arrival on the nesting area so as to initiate nests early and better time the hatch of offspring with peak nutrient availability (Klaassen et al., 2006; Nolet et al., 2020). Compared to the large-bodied goose species that can rely on endogenous reserves, the pre-lay periods for semipalmated sandpipers and longspurs were relatively long, reflecting the fact that individuals of both species must first forage to acquire the necessary exogenous resources prior to producing eggs. Relative to snow geese, brant invest more exogenous nutrients into eggs, and in years when spring is advanced are more likely to initiate follicle development during migration (Hupp et al., 2018). The shorter pre-lay durations in brant during years of earlier snowmelt likely reflected the species’ tendency to initiate follicle development prior to arrival on the nesting area. Follicle development in snow geese breeding on the Colville River Delta mainly occurs after arrival in the Arctic (Hupp et al., 2018), and their pre-lay interval is less variable relative to environmental conditions.
Although endogenous investment in eggs gives geese an advantage in advancing egg development and nest initiation when Arctic phenology is advanced, female geese may use reserves for self-maintenance at the expense of reproductive investment in years when snow and cold temperatures persist (Barry, 1962; Raveling, 1978; Reed et al., 2004). We detected reduced clutch size in response to delayed snowmelt in both brant and snow geese, and to lower temperatures in brant (Fig. 3). Because brant invest relatively more endogenous reserves into eggs than the other species we studied, they are more likely to reduce clutch sizes in colder springs or when late snowmelt delays nest initiation and induces females to use reserves for self-maintenance (Barry, 1962). Late springs likely delayed new plant growth that brant also rely on for egg nutrients (Hupp et al., 2018). In contrast, phylogenetic constraints (MacLean, 1972) likely preclude semipalmated sandpipers from varying clutch sizes (Sandercock et al., 1999), although previous research indicates that shorebirds can regulate egg size within clutches in response to seasonal variation (Martin et al., 2018; Sandercock et al., 1999). Longspurs exhibited the greatest variation in clutch size during our study (1–7 eggs), but this variation was not related to environmental conditions. Such variation may instead reflect the influence of other factors (e.g., female age or breeding experience; Sæther, 1990; Stearns, 1976) that we did not measure.
In addition to adjusting investment in clutches, Arctic-breeding birds may forego breeding altogether in response to extreme environmental conditions (Ganter & Boyd, 2000; Schmidt et al., 2019), a response believed to reflect a trade-off between current reproductive investment and future survival (Linden & Møller, 1989; Roff, 2002). The nesting effort of semipalmated sandpipers was greatest in years with warm temperatures and early snow melt, but the other three species did not meaningfully moderate nesting efforts in response to these variables. This variation again suggests the likely role of exogenous reserves in modulating the reproductive output of semipalmated sandpipers. Shorebirds lay clutches that constitute a relatively high proportion of their body mass (Rahn et al., 1975; Ricklefs, 1984), and semipalmated sandpiper females at our study site were apparently unable to acquire sufficient arthropod resources to initiate nests in cold springs with extensive snow cover. In contrast, brant and snow geese could rely on endogenous reserves, and longspurs seed resources in lieu of arthropods, to ensure nesting opportunities.
Notably, the four species in our study share the overarching life-history trait of being migratory animals. Previous research has suggested that migratory species are more vulnerable to climate change due to the potential decoupling of relevant seasonal cues across a species’ range (Both et al., 2010; Møller et al., 2008; Robinson et al., 2009). We documented interannual responses among these four migratory species, however, that demonstrated a high degree of interspecific response to shared stimuli among diverse taxa. The species at our study site use two flyways (Pacific [brant and snow geese] and Central [snow goose, semipalmated sandpiper, and longspur] Americas flyways) across a mix of marine (brant and semipalmated sandpiper), seasonal wetland (snow goose and semipalmated sandpiper), and agriculture/prairie landscapes (snow goose and longspur) to migrate to Arctic breeding grounds. Despite the use of varied migratory routes and habitats, these species nonetheless adjusted their timing of breeding in ways that tracked spring environmental conditions at the breeding site. Indeed, long-term information for these and other species at this site exhibit a general advancement of arrival dates in concert with warming spring conditions (Ward et al., 2016), patterns noted more broadly in other studies (Jonzén et al., 2006; Thorup et al., 2007; Van Buskirk et al., 2009). Migratory birds exhibit life histories that are predicated on exploiting diverse, ephemeral landscapes (Greenberg & Marra, 2005; Newton, 2008), and employ flexible physiologies that permit the rapid hypertrophy and subsequent atrophy of respiratory, digestive, and circulatory functions (Piersma & van Gils, 2011). Such behavioral and physiological adjustments enable large-scale movements and undoubtedly also serve as buffers in a changing world. So although migratory species may theoretically be predisposed to a decoupling of seasonal cues across their large ranges, these four species responded predictably to prevailing environmental conditions by adjusting their reproductive timing and investment. Migratory birds, by virtue of their intrinsic life histories, may thus accommodate the effects of a warming Arctic better than previously appreciated.

4.2 Variation in chick growth

Interannual adjustments in the timing of and investment in breeding may reflect adaptive responses to prevailing environmental conditions, but nests must hatch and chicks must grow and survive in order for such adjustments to be propagated in an evolutionary context (see Charmantier & Gienapp, 2014). The rate of chick growth provides insights into fitness-related variables (e.g., survival, recruitment, lifetime reproductive output) that are otherwise extremely difficult to measure in most bird species. Curiously, resource abundance was not a meaningful predictor of chick mass of either insectivore species in our study. Obviously, food abundance directly affects chick growth, which suggests that either we did not measure arthropod abundance in a way that reflected real abundance or that the range of abundances that we measured at our site did not limit growth. For the former supposition, this same sampling protocol has been successfully employed by others (see Kwon et al., 2019 for overview), and Saalfeld et al. (2019) specifically determined that this sampling technique described variations in arthropod abundance that predicted the chick mass of two shorebird species (dunlin Calidris alpina and pectoral sandpiperCalidris melanotos ) that are closely related to semipalmated sandpipers. Thus, it is more likely that arthropod abundances were not limiting during our period of study. Despite measuring a nearly 300-fold variation in arthropod abundance (1.5–427.3 mg 3-day-1 sample) during periods of chick growth from 2015–2017, our measurements did not apparently reflect conditions that affected the growth of insectivore chicks. For semipalmated sandpipers, warmer temperatures in the 3-day period prior to recapture were associated with larger same-age chicks, and there was a positive relationship between temperature and arthropod abundance at our study site (Pearson’s r = 0.62, P <0.001). In general, arthropod abundance was low at lower temperatures and increased at temperatures >5°C, but, importantly, we also documented periods of high temperature that coincided with low arthropod abundance (see Results) during which we nonetheless observed high age-specific chick masses. Further, we documented mass loss or static growth of recaptured semipalmated sandpiper chicks during a period of unusual cold (average 3.4°C from 7–13 July 2015), and encountered dead and emaciated chicks of red phalaropes (Phalaropus fulicarius ) that presumably had starved. This suggests a thermogenic trade-off for semipalmated sandpiper chicks at higher temperatures wherein the high cost of thermogenesis (Bakken et al., 2002; Schekkerman et al., 2003) may be minimized and growth maximized (McKinnon et al., 2013), permitting rapid chick growth even during periods of relatively low food abundance.
Resource abundance was a meaningful predictor of the growth of only brant, but the negative parameter estimate of this metric (Table 4) was counterintuitive. Previous research has demonstrated an inverse relationship between forage biomass and forage quality (i.e., nitrogen content) in graminoids like subspathacea (Doiron et al., 2014; Lameris et al., 2017), a relationship which may account for this finding. There is a strong positive relationship between nitrogen content and demographic variables like gosling growth and survival (Doiron et al., 2015; Manseau & Gauthier, 1993; Person et al., 2003; Sedinger & Raveling, 1986), and in our study it may be that periods of high subspathacea biomass had correspondingly low values for nitrogen content. The NDVI-based techniques of Hogrefe et al. (2017) are not suitable for detecting the small but meaningful differences in the nitrogen content of subspathacea . Anecdotally, Hogrefe et al. (2017) determined the biomass and percent nitrogen of 221 samples ofsubspathacea from 2012–2015 at our study site, and although NDVI values could not accurately predict the nitrogen content of these samples, there was a strong negative relationship between biomass and nitrogen content (Pearson’s r = -0.52, P <0.001). Thus, higher age-specific body mass of brant goslings associated with lower forage biomass may reflect aspects of forage quality that we were unable to measure. In contrast, we did not find that higher subspathaceabiomass negatively affected the growth of snow goose goslings. The larger snow goose goslings may be able to accommodate lower qualitysubspathacea compared to smaller brant goslings due to a greater intake and processing capacity (e.g., Lesage & Gauthier, 1997; Manseau & Gauthier, 1993; Richman et al., 2015). Further, larger snow goose goslings were associated with cooler temperatures, a result contrasting with semipalmated sandpipers. This may again indirectly reflect aspects of food quality (Dickey et al., 2008) rather than thermal constraints on growth per se (but see Fortin et al., 2000), as subspathacearesponds to warm temperatures with increased vegetative growth (i.e., resource biomass) and decreased nitrogen content (i.e., resource quality; Doiron et al., 2014; Lameris et al., 2017). Although we do not fully understand the mechanisms, our study does demonstrate that warmer temperatures during brood rearing can differentially affect growth rates of avian herbivores versus insectivores.
Of note, none of the environmental covariates that we assessed meaningfully predicted the mass of longspur chicks, the only altricial species in this comparison. Previous research at a nearby site in Arctic Alaska documented seasonal declines in the growth of longspur chicks, as well as negative effects of low arthropod abundance and cold temperatures (Pérez et al., 2016). At our study site, however, other factors were apparently more important in modulating the growth of longspurs. It may be that aspects of parental quality that we did not measure (e.g., nest-site selection [Martin et al., 2000, Lloyd & Martin, 2004], chick provisioning [Davies, 1986; Limmer & Becker, 2009]) buffered deleterious effects of temperature and resource abundance that affected the chicks of precocial species at our site.
An environmental variable that received wide support across our comparisons was the timing of nest initiation with respect to snow cover. For brant, snow geese, and semipalmated sandpipers, chicks from nests that were initiated before or near the annual date of 50% snow cover were larger than chicks from nests that were initiated relatively later (Table 4). The positive effect of early initiation on chick growth has been documented in other studies of Arctic-breeding birds (Cooch et al., 1991; Ruthrauff & McCaffery, 2005; Sedinger & Flint, 1991), and indicates that early nest initiation rather than resource abundance is a more important factor in regulating chick growth at our study site. This suggests the role of potential factors (e.g., parental quality [Clutton-Brock, 1984; Forslund & Pärt, 1995], carry-over effects [Harrison et al., 2011]) that we could not measure. Other researchers, however, have specifically detected seasonal declines in the growth of snow goose goslings related to food abundance (Lepage et al., 1999; Lindholm et al., 1994). These studies were conducted at breeding sites with degraded grazing lawns and low-quality food compared to that on the Colville River (Hupp et al., 2017), emphasizing how spatial variation in ecological factors—in this case, food quality and abundance—can differentially affect the demographic response of the same populations (Sedinger et al., 2001).

4.3 Conclusion

Life-history traits that afford flexible responses to variable environmental conditions are favored in highly seasonal and unpredictable environments like the Arctic. Traits that in turn promote evolutionary changes in a population are further expected to be subject to strong selection pressure under climate-warming scenarios (Berteaux et al., 2004; Hoffmann & Sgrò, 2011; Williams et al., 2008). Temperatures across all seasons are projected to increase on Alaska’s Arctic Coastal Plain due to climate change (IPCC, 2013), and increases in warming have already led to long-term advances in snow melt and longer snow-free seasons in Arctic Alaska (Cox et al., 2017; Hinzman et al., 2005; Stone et al., 2002). Brant and snow geese generally responded more flexibly to variation in temperature and snowmelt during the pre-lay and nesting periods than did semipalmated sandpipers and longspurs. In contrast, we detected potentially deleterious effects of increased temperature on brant and snow goose goslings, while semipalmated sandpiper chicks responded favorably to warmer conditions. Thus, brant and snow geese may possess traits that are beneficial during one phase of the reproductive cycle (e.g., relative flexibility along the endogenous-exogenous spectrum) and others which may be detrimental at another phase (e.g., temperature-mediated sensitivity to food quality during juvenile growth). For the Arctic-breeding birds in our study, these contrasting responses underscore the importance of assessing the effects of climate variability across multiple phases of the reproductive cycle (Nolet et al., 2020).