Winter feeding inﬂuences the cost of living in boreal passerines

The plastic regulation of internal energy reserves is acknowledged as the main adaptive response to winter conditions of resident small birds in northern latitudes, a strategy that may be altered whenever human-supple-mented food is available. We investigated the effects of supplementary feeding on the energy management strategy of two wild passerine species, the Willow Tit Poecile montanus and Blue Tit Cyanistes caeruleus , wintering in boreal conditions by measuring body mass and the energy cost of living, i.e. basal metabolic rate. Individuals of both species were heavier, larger and exhibited a higher energy cost of living when captured at the feeders than were individuals captured away from feeders. Fed Willow Tits expended more energy in maintenance, although this difference disappeared once mass was accounted for. Conversely, Blue Tits at feeders had higher mass-adjusted energy cost of living, but only at low ambient temperatures. The results indicate that winter feeding has species-speci ﬁ c effects on overall energy management strategy and modi ﬁ es the response to environmental conditions of wintering passerines.

Endotherms wintering at high latitudes struggle to survive as environmental conditions deteriorate at the same time as food availability diminishes. Small birds survive this situation through seasonal acclimatization based on an increased metabolic output to meet thermogenic demands (Broggi et al. 2004), and the enlargement of internal energy reserves (Blem 2000).
The adaptive acquisition and storage of reserves is the main mechanism to maintain energy balance in wintering small birds (Pravosudov & Grubb 1997, Brodin 2007. Body mass (M b ) fluctuations resulting from daily increase and overnight depletion of internal reserves constitute the core of the 'optimal body mass' (OBM) theory of energy management in wintering small birds (Moiron et al. 2018). This daily M b cycle is predicted to widen as temperatures and food predictability decrease (i.e. higher evening M b and lower morning M b ) (Pravosudov & Grubb 1997).
In addition to internal reserves, wintering small birds save energy through regulation of body temperature and the cost of living (i.e. basal metabolism) as part of their energy management strategy (McKechnie & Lovegrove 2002, Broggi et al. 2019. M b and basal metabolic rate (BMR) are phenotypically integrated traits that are highly and positively related (McNab 1997, Hayes 2001. The interrelation between the two traits varies individually according to tissue composition and proportions, which change with physiological, ecological and environmental conditions (Klaassen & Biebach 1994, Rønning et al. 2008, Broggi et al. 2019. Providing supplementary food to wild birds is becoming an increasingly popular human activity which presumably allows individual birds to obtain their optimal energy reserve level more readily (Robb et al. 2008). However, winter-feeding may have wider physiological and ecological consequences that only recently have started to be studied (Robb et al. 2008). For example, increased food availability and predictability may come at the cost of impoverished diet, higher predation risk and pathogen exposure, with negative fitness consequences (Barea & Watson 2013).
The Blue Tit Cyanistes caeruleus and the Willow Tit Poecile montanus are two similar small-sized (~10 g) species of the family Paridae that are widespread across Europe. The Blue Tit is a temperate species inhabiting deciduous forests and has expanded northwards into boreal regions in recent decades (Valkama et al. 2011). Blue Tits at the northern margin of the distribution spend winters mostly near human habitation where, like the Great Tit Parus major, they rely on feeders for survival, a habit that presumably explains the successful expansion of both species (Orell 1989). However, in contrast to Great Tits, a fraction of the Blue Tit population persists during winter in the reedbeds and birch forests along the coastline, away from inhabited buildings and feeders (J. Broggi & K. Koivula pers. obs.). The Willow Tit is a highly sedentary forest species that is well adapted to survive winters without the need for humanprovided food. Once settled, individuals remain all year round within their territory, where they hoard scattered food-caches for winter survival (Lahti et al. 1998).
Additionally, Willow Tits prioritize energy-saving mechanisms, allowing them to buffer the effects of environmental variation in comparison with other temperate species such as the Blue Tit (Broggi et al. 2003(Broggi et al. , 2019. Thus, although the two species have a similar size and face the same thermal challenges, they maintain different seasonal energy management strategies (Broggi et al. 2019). However, recapture evidence from both species suggests that whenever individuals find a feeder, they rarely move away from such a steady food source, particularly during cold periods (Lahti et al. 1998, Pakanen et al. 2018).
Here we studied how permanent access to supplementary food influences M b and winter BMR in wild Willow and Blue Tit populations in boreal conditions. We hypothesize that feeder users will enjoy an increased food predictability that will affect their optimal reserve level. Further, the overall energy budget will possibly increase, leading to a higher energetic cost of living, measured by BMR. Following the OBM theory, birds at feeders are expected to reduce their M b and increase their BMR. We expect that proximity to feeders reduces sensitivity to environmental conditions. Further, we expect the two species to differ in their response, as boreal specialists such as the Willow Tit should be better adapted to withstand local conditions compared with temperate species such as the Blue Tit, and thus be less affected by food supplementation.

Study area and experimental approach
We analysed M b and BMR of 61 wintering Blue and Willow Tit individuals in the Oulu region, Finland (65°N, 25°30'E) between 2003 and 2010. We tested the effect of artificial food supplementation by comparing birds of both species from permanently fed and nonfed areas (Fig. 1). In the fed area, birds were captured by funnel traps baited with unhusked peanuts near the Oulu University campus (19 Blue Tits and 23 Willow Tits). Data from these fed birds were selected from a larger dataset that has been already published to match in time (year and season) with the non-fed birds (see below) (Broggi et al. 2019). Supplementary food was provided near the University campus from early September to late March, and feeders were refilled on a weekly basis to ensure a permanent supply. The Oulu study area is a forest mosaic of mixed tree species of different ages dominated by conifers Pinus sylvestris and Picea abies and diverse birch Betula sp. species. Average winter temperatures fall below À10°C with a permanent snow cover that remains for 5 months. Birds from the non-fed areas were captured with mist-nets by luring them with sound recordings, and small feeders were installed on the same capture day (six Blue Tits and 13 Willow Tits). Non-fed Blue Tits were trapped in coastal areas c.40 km south of Oulu, in an area of reedbeds and coastal birch forest away from any urban centres (Fig. 1). The thick snow layer (> 2 m) and frozen sea that characterizes the area from November until late April prevents any access to the summer cottages present and ensures that there is no supplementary feeding during winter months. Non-fed Willow Tits were captured in the Oulu study area to the north of the town (Fig. 1), where resident flocks remain faithful to their territories after settling down in early winter as first-year birds (Koivula & Orell 1988).
The different capture methods did not affect recapture probability and did not bias the samples obtained, as shown by a subsample of individuals trapped with tape lures and mist-nets at the fed area (Broggi 2006, pp 21-28;J. Broggi in prep.). All birds were trapped in the afternoon between midday and 14:00 h, measured for biometric variables, and immediately transported to the University laboratory. Birds were transported in ringing bags and then maintained in individual cages for a short period (< 2 h) of time before night-time measurements.
Cages consisted of small (0.3 m 3 ) metallic mesh cubes covered with a blind fabric to minimize stress. Ad libitum food (unhusked peanuts and mealworms) and water supply were provided before BMR measurements were taken (see below). Time spent in the cages did not affect BMR measurements (J. Broggi unpubl. data).

Metabolic measurements
BMR is defined as the average minimal oxygen consumption under post-absorptive digestive conditions during the resting phase of the daily cycle of non-growing, non-reproductive animals at thermoneutrality (McNab 1997). BMR is the most widely used energetic measure in biological and medical studies and constitutes a major part of the daily energy expenditure in wintering birds (Broggi et al. 2004), potentially affecting overwinter survival (Nilsson & Nilsson 2016). Oxygen consumption of (up to) four individuals each night was used to measure BMR in an open-circuit respirometer in a dark climate chamber at a constant temperature of 25°C, within the thermoneutral zone of the species (Reinertsen & Haftorn 1986). In brief, outdoor dry air was pushed through four independent mass-flow controllers (FMA-A2407; Omega Engineering, Inc., Norwalk, CT, USA) at 300 mL/min into plexiglass metabolic chambers (1.6 L) containing each individual bird. Constantly flowing incoming and outcoming air was scrubbed for CO 2 and H 2 O, with soda-lime and silica scrubbers, then directed through a multiplexer that sent air from each channel in bouts of 12 min to the oxygen analyser (S-3A; Ametek, Pittsburg, PA, USA). Each cycle lasted an hour and included a fifth channel with reference air to control for the analyser bias. Recorded data were visually inspected, and the lowest 5 min of steady measurements were converted to mL O 2 /min following Hill (1972). BMR measurements were invariably taken between midnight and 04:00 h, ensuring the post-absorptive state of all individuals. After the measurement night, birds were released at the point of capture.

Statistical analyses
We used general linear models to explain variation in M b , BMR and mass-adjusted BMR. Mass-adjusted BMR was studied by analysing BMR as the dependent variable, with M b fitted as a covariate. Whole-animal and mass-adjusted BMR can be considered different traits that provide information on different scales of the individual energy expenditure (Hayes 2001). Whole-animal BMR reflects the overall energy cost of maintenance of an individual, and variation can be interpreted as a general increase in the overall size with a concomitant increase in the energy cost of living, which may sometimes be paralleled by changes in organ sizes and/or the proportion of tissues with different metabolic activity (Petit et al. 2014). Mass-adjusted BMR is the energetic cost per unit mass, and variation probably involves a substantial change in the metabolic intensity resulting from differences in tissue proportions that allow the metabolic comparison of individuals of different weight (Mueller & Diamond 2001, Rønning et al. 2008. Individual variables included age (first winter or older) and sex as categorical predictors. Willow Tits during the non-breeding season can be sexed reliably through sexual dimorphism in wing length (Koivula & Orell 1988), and thus only tarsus and wing length (mm) were considered as continuous predictors of body size and sex. Environmental variables considered were day of capture (1 October = 1), winter of study (2003)(2004) = 1) and minimum temperature (T min ) on capture day (in°C). The quadratic component of day of capture was included to test for non-linear effects of date, which reflect photoperiodic cyclicity over the non-breeding season. Each variable was tested individually together with treatment (supplementary-fed vs. non-fed) as a fixed factor and their interaction in all analyses, except for the analyses of mass-adjusted BMR, where M b was included as covariate. Models with significant variables are presented (Table 1) together with the F, df and Pvalues of the last predictor incorporated. Parameter estimates AE se are provided for significant continuous predictors. All P-values are two-tailed. All continuous variables fulfilled the requirements of normality.

RESULTS
Blue Tits at feeders were heavier than birds from nonfed areas after sex and size were accounted for (Fed: 12.103 g AE 0.147 vs. Non-fed: 10.928 g AE 0.263; F 1,20 = 15.43; P < 0.001; Fig. 2). However, such differences in M b were not reflected in either BMR or massadjusted BMR (Table 1; Fig. 2). Blue Tit males were heavier than females, and heavier Blue Tits had longer wing and tarsus length (Table 1). Otherwise, M b was not related to age or any of the environmental parameters considered (Table 1). These patterns persisted between fed and non-fed areas, as none of the interactions between these variables and treatment were significant (all P > 0.05). However, BMR was negatively influenced by T min , a pattern that differed between fed and non-fed treatments for mass-adjusted BMR, while all the other variables remained non-significant (all P > 0.05; Table 1). Fed Blue Tits showed a higher massadjusted BMR with decreasing T min , whereas the effect was not significant in non-fed birds (Table 1; Fig. 3).
Willow Tits were heavier and had a higher BMR at feeders (Table 1; Fig. 2). However, mass-adjusted BMR did not differ between fed and non-fed areas (Table 1; Fig. 2). Patterns of variation in M b , BMR and mass-adjusted BMR were independent of any size or age differences, as shown by non-significant interactions between these variables and treatment (all P > 0.05). Willow Tit age and size were related to M b but not to BMR or mass-adjusted BMR. Juvenile Willow Tits were heavier than adults, and heavier Willow Tits had longer wings (Table 1). Willow Tit M b , BMR and mass-adjusted BMR were not related to T min or year of study (Table 1, Fig. 3). However, BMR and mass-adjusted BMR were non-linearly related to capture date, with a drop in midwinter (Table 1; Fig. 3). None of these seasonal patterns differed between experimental areas, as shown by nonsignificant interactions with treatment (all P > 0.05).

DISCUSSION
Small birds wintering in boreal regions respond to supplementary feeding with substantial physiological changes involving body mass and energy expenditure. Both Blue and Willow Tits had higher M b when captured at the feeders (+6.29% and +4.48%, respectively). Additionally, access to feeders promoted increased BMR in Willow Tits (+7.78%), whereas Blue Tits exhibited a conditional response to the immediate environmental conditions. Blue Tits with access to feeders had higher metabolic intensity (i.e. mass-adjusted BMR) when minimum temperatures dropped. Winter supplementary feeding affects the energy strategy and response to environmental conditions of the two studied species differently.
Body mass was higher in both species studied as a response to winter food supplementation, suggesting that the studied populations may be food-constrained. Further, access to feeders did not promote a reduction in M b , as predicted by the OBM, but vice versa (Brodin 2007, Moiron et al. 2018. Even well-adapted boreal species seem to make use of extra food whenever available, suggesting that local conditions are extreme and that energy-saving mechanisms are not cost-free (Pravosudov & Grubb 1997). BMR was higher as a result of access to feeders only in Willow Tits, a response probably linked to the higher M b . Whenever more food is ingested, a larger processing capacity is required, which may lead to increased size in digestive and cardiopulmonary organs (Mueller & Diamond 2001, Nilsson 2002, Petit et al. 2014. Blue Tits did not change their overall energy budget, as BMR remained practically unaffected by the presence of feeders. However, Blue Tits at the feeders responded to the environmental conditions by increasing their metabolic intensity (i.e. mass-adjusted BMR) when T min decreased. Changes in mass-adjusted BMR or just M b would be indicative of a substantial tissue reorganization (Hayes 2001) and increased internal reserves, respectively (Klaassen & Biebach 1994). These results suggest that Blue Tits react to an immediate environmental challenge only when food availability is guaranteed.
Boreal and temperate specialist species such as Blue and Willow Tits differ in their metabolic response to food supplementation rather than in body mass, suggesting a relevant role of energy expenditure over internal reserves in energy management in these species, in line with previous studies (Broggi et al. 2019). The changes in metabolism in both species suggest that the variation in M b not only resulted from regulation of internal reserves but is also affected by other mechanisms such as changes in organ sizes and tissue composition (Petit et al. 2014). Here we show that supplementary feeding of wild wintering birds promotes a higher M b as a result of an improved food availability, which is also parallelled by changes in the energy cost of living and its response to environmental conditions. These results highlight the multifaceted nature of winter energy management in small passerines by showing that changes in body mass are not only the result of internal reserve variation (Broggi et al. 2019). Winter supplementary feeding has species-specific physiological and ecological implications, which certainly need to be addressed to comprehend fully the impact on wild animal populations.
We are in debt to numerous field assistants and staff from the Oulu University Research Facility who provided logistic and technical support and field assistance. Jan Ake Nilsson provided valuable comments on earlier drafts of the manuscript. The manuscript was improved by the comments of Simon Butler and two anonymous referees.
J.B. has been funded by the Andaluc ıa Talent Hub Program launched by the Andalusian Knowledge Agency, co-funded by the European Union's Seventh Framework Program, Marie Skłodowska-Curie actions (COFUND -Grant Agreement no. 291780) and the Junta de Andaluc ıa. The study was supported by the Academy of Finland project nos. 102286 and 47195, and the Thule Institute of the University of Oulu (E.H.).