1. Introduction

Population and ecosystem dynamics are key ecological processes to monitor as ecosystems undergo anthropogenic alterations due to habitat fragmentation and loss (Fahrig 2003; Mantyka-Pringle et al. 2012) and climate warming (Parmesan & Yohe 2003; Scheffers et al. 2016). Species have responded through changes in ecological processes including shifts in phenology (Parmesan & Yohe 2003; Visser & Both 2005), changes to foraging behaviour (Mahan & Yahner 1999), altered habitat use/distribution (Mantyka-Pringle et al. 2012; Kortsch et al. 2015), and reduced reproductive/survival rates with resulting declines in population abundance (Fahrig 2003; Scheffers et al. 2016). These changes in species abundances and distributions can lead to altered community structure and trophic interactions (Rall et al. 2010; Molinos et al. 2015; Scheffers et al. 2016) as well as regime shifts (Petchey et al. 1999; Kortsch et al. 2014), with implications for ecosystem function and stability (de Ruiter et al. 1995; Neutel et al. 2002; Rall et al. 2010). Changes in community structure are especially critical to ecosystems where higher trophic levels are vulnerable to anthropogenic change because altered top predator population dynamics can cause cascading effects (Shackell et al. 2010).
Examining energy dynamics over time can provide insights into ecological responses to both natural and anthropogenic change. Bioenergetics has been studied at individual/species levels using ingestion and assimilation rates (Bailey & Mukerji 1977; Cressa & Lewis 1986), prey consumption estimates (Lantry & Stewart 1993), and metabolism (Lam et al. 1991). Furthermore, broader-scale energetics studies have documented patterns in population energetic requirements (Markussen & Øritsland 1991; Ryg & Øritsland 1991; Ernest et al. 2003) and ecosystem energetic dynamics across trophic levels (Sakshaug et al. 1994). Bioenergetics research at various scales is useful for monitoring ecological patterns given that alterations in individual energetic balances may lead to changes in population dynamics (Yodzis & Innes 1992; Humphries et al. 2004). Thus, understanding temporal dynamics in energetics and relationships to environmental conditions may provide insights into the mechanisms influencing population dynamics and improve our ability to predict how populations may respond to future stressors.
The Arctic marine ecosystem has experienced rapid and extensive changes in sea ice in response to climate warming (Comiso 2002; Stirling & Parkinson 2006; Stroeve & Notz 2018; IPCC 2019). In particular, reduced sea ice extent and earlier sea ice breakup are major factors that influence the life history of many Arctic marine species (Comiso 2002; Stirling & Parkinson 2006; Meier et al. 2014), especially sea ice-dependant marine mammals (Laidre et al. 2008, 2015; Post et al. 2009; Wassman et al. 2011). For example, polar bears (Ursus maritimus ) are particularly vulnerable to sea ice decline (Stirling et al. 1999; Stirling & Derocher 2012) because they rely on sea ice for movement, reproduction, and as a platform from which to hunt their main prey, ice-associated seals (Stirling & Archibald 1977; Smith 1980). As both a top predator and a species sensitive to sea ice conditions, polar bears are particularly useful for monitoring changing Arctic marine ecosystem dynamics. The Western Hudson Bay (WH) polar bear population is an example of a long-term monitoring program where individuals have been captured and measured over three decades, which provides a unique opportunity to examine energetic dynamics relative to sea ice habitat. Declines in WH polar bear body condition (Sciullo et al. 2016), reproductive rates (Stirling et al. 1999), survival (Regehr et al. 2007), and abundance (Lunn et al. 2016) have all been associated with climate warming. Such changes to population dynamics are influenced by individual condition and energy balances (Yodzis & Innes 1992; Humphries et al. 2004), which in turn are driven by alterations in energy intake and expenditure (Pagano et al. 2018). The open water period Hudson Bay, during which polar bears fast on land, has lengthened (Stern & Laidre 2016) and an increase to a 180 day fasting period is predicted to result in increased starvation and mortality rates (Molnár et al. 2010, 2014; Pilfold et al. 2016). It is therefore important to examine energetic dynamics at various levels and long-term monitoring can provide important insights into top predator bioenergetic responses to climate warming and implications for ecosystem dynamics.
Energetics has been examined in polar bear populations using a fat condition index (Stirling et al. 2008), metabolic rates (Pagano et al. 2018), body condition metrics and fasting (Atkinson & Ramsay 1995; Robbins et al. 2012; Rode et al. 2018), and lipid content (Sciullo et al. 2016). Additionally, the use of body measurements to estimate individual energetic stores can provide insights into energetic dynamics. For example, storage energy and energy density have been used to quantify energy budgets for individual polar bears (Molnár et al. 2009, 2010; Sciullo et al. 2016). Storage energy represents the energy that is available for maintenance, reproduction, and growth, and is influenced by energy intake and expenditure (Molnár et al. 2009, 2010; Sciullo et al. 2016). However, because not all energy is available for use when individuals are fasting, energy density is another useful metric as it accounts for the energy content per unit mass (Molnár et al. 2009, 2010; Sciullo et al. 2016). These measures are both informative for understanding changes in individual energy balances, as well as predicting changes in population dynamics in response to future conditions.
We used data on population abundance, age/sex structure, and morphometrics collected from WH polar bears to estimate the population energy density and storage energy from 1985 to 2018. Our objectives were to: 1) examine temporal dynamics of energy in the WH population, 2) assess the influence of environmental conditions on population energy, and 3) explore lagged effects of environmental variables. In addition, we analyzed energy dynamics within the population to provide insights into intra-population variation and examine the vulnerability of different age/sex classes based on energy balances. This research increases our understanding of the temporal and intra-population energetic patterns of a top predator experiencing habitat loss due to climate warming, as well as potential implications for Arctic marine ecosystem dynamics.