Carry-over effects on reproduction in food-supplemented wintering great tits

Bird winter-feeding has become a popular backyard activity around the world, particularly in northern regions of Europe and America with cold winters. However, the short-and long-term ecological consequences of such artificial feeding remain inconclusive. In seasonal environments, timing of breeding is a crucial aspect that can strongly influence reproductive output and ultimately fitness. Individual condition at the start of the breeding season is especially important in determining breeding success, by influencing the onset of and investment in breeding. However, empirical evidence on the effects of winter feeding on avian breeding performance remain equivocal. We studied onset of reproduction (laying date) and breeding investment (clutch size) over seven consecutive seasons in a population of wild great tits Parus major in southern Sweden. During the first three years of study, no experimental manipulation was undertaken, while over the last four years the study area was exposed to either supplemented or unmanipulated winter feeding conditions. Breeding was positively affected by supplementary feeding during winter, as birds breeding in the supplemented area increased their clutch size compared to birds from the control area, although laying date remained unaffected by winter feeding. Since differences in clutch size were absent during the three years prior to the experimental manipulation, the results suggest that winter supplementary feeding, and not inherent differences between the two areas, was the reason for the observed effect. Both breeding parameters varied over the years of study, although the effects of the experimental manipulation on clutch size remained consistent, which suggests a carry-over effect of winter feeding on subsequent breeding performance.


Research
availability (Grüebler and Naef-Daenzer 2008).Especially at high latitudes, where time and weather constraints exert strong selective pressures (Pakanen et al. 2016), delayed onset of reproduction has a negative impact on nestling physiology with potential long-term consequences (Kalinski et al. 2019).Furthermore, birds need to decide how much resources should be devoted to a given breeding event, a decision that involves several successive stages, from territory defence and nest building, the number and quality of eggs produced, and the level of parental care that can span well beyond the nesting period (Williams 2012).Avian parents will need to invest successively in each of these events, by carefully considering their own condition, the quality of the environment and the expected outcome of their investment, which optimally will lead to a positively correlated response among stages in successful breeders (Sockman et al. 2006).However, in many cases, resource constraints may lead to trade-offs among reproductive traits, for example the quality and number of eggs laid (Williams 2012).Still, the number of eggs produced, i.e. clutch size, is among the most determinant investment decisions to be taken at the breeding onset and is considered a robust measurement of breeding investment in life-history and ecophysiological studies of oviparous animals (Sockman et al. 2006).The trade-off between individual investment in a specific breeding event and self-maintenance is instrumental for determining current versus future reproductive success to maximize individual life-time reproductive success (Stearns 2000).However, given the fluctuating nature of environmental conditions, such traits need to be plastically adjusted to prevailing conditions.Understanding the plasticity of such responses requires field experiments and long-term monitoring of wild populations (Verhulst and Nilsson 2007).
Individual condition at the start of the breeding season is a key factor in determining the onset and resource allocation to any breeding event (Williams 2012).The tight energy budget during winter when non-renewable food sources drastically decrease while general environmental conditions deteriorate may affect early-breeding decisions (Carey andDawson 1999, Williams 2012).Small passerines survive these conditions by a process of acclimatization that ultimately leads to an overall increase in energy expenditure (Broggi et al. 2019).Thus, winter survival and individual condition at the start of reproduction depend on the ability to obtain the necessary resources to cover this increased energetic challenge, a cost that can be alleviated when human-provided food is available (Brittingham and Temple 1988, Lahti et al. 1998, Nilsson et al. 2020).The way individuals cope with harsh conditions during winter is thus of major importance since it may have immediate, but also long-term, carry-over effects on future life-history stages (Harrison et al. 2011).
Several studies have attempted to reveal the effects of food supplementation during winter on breeding performance with varying results.While the use of artificial feeders advanced laying dates in blue tits Cyanistes caeruleus (Robb et al. 2008b) and increased clutch size in the gray jay Perisoreus canadensis (Derbyshire et al. 2015), other studies in blue tits showed no effects (Plummer et al. 2013a, Crates et al. 2016) or even detrimental consequences on overall breeding success (Plummer et al. 2013b).In addition, results may change between years and populations, for example in blue tits Cyanistes caeruleus in the UK (Robb et al. 2008b, Plummer et al. 2013a, b, Crates et al. 2016).Furthermore, the supplementation timing with respect to the breeding onset, the long-term direct consequences (i.e.carry-over effects), as found by Grieco et al. (2002) in blue tit females that postponed their egg laying after being food-supplemented in the previous breeding season, are acknowledged as critical aspects that remain poorly understood (Ruffino et al. 2014).
Winter feeding of wild birds has become widespread in recent decades in many parts of the world and has become one of the most popular human-wildlife interactions (Reynolds et al. 2017).However, very few studies have investigated the short-or long-term consequences during multiple years, comparing the same areas both with and without winter feeding (Robb et al. 2008a, Ruffino et al. 2014).Here we studied the breeding parameters in a wild population of great tits Parus major over a 7-year period, from which the last 4 years involved a winter-feeding experiment.We studied the effects of winter supplementary feeding on breeding density, laying date and clutch size during the subsequent breeding seasons.Female great tits exposed to experimental winter feeding exhibited a decreased use of nocturnal hypothermia, implying that food predictability may be an important factor affecting energy management during winter in this population (Nilsson et al. 2020).Therefore, we predicted that if natural food sources are a limiting factor in the wintering population, population density should increase in the fed area and food-supplemented individuals should advance their laying date and increase their clutch size as compared to non-supplemented tits.

Study area and species
The great tit is a small (~18 g) songbird inhabiting temperate forests across Eurasia.The species feeds mostly on insects and arthropods during the breeding season but gradually switches to a seed-based diet during winter (Gosler 1993).Being a hole-nesting passerine, it readily uses artificial nest-boxes for breeding and winter roosting, as well as artificial feeders during winter.The study area (~5 km 2 = 470 ha) is in a restrictedaccess forest within the Vomb municipality, southern Skåne (SE) (55.67°N, 13.55°E) (Fig. 1).The forest is composed of stands of scots pine Pinus sylvestris and diverse broad-leaved species of different age and variable understory.The forest is surrounded by agricultural fields with sparse farmhouses.Part of the study area has been provided with nest-boxes since 1988, and since 2013, the area has had 300-400 wooden nest-boxes attached to tree trunks ~50 m apart and 1.5 m above the ground.A sedentary great tit population is settled year-round in the area using these nestboxes for breeding and for winter roosting.

Experimental approach
From beginning of November 2015, 11 feeders were installed hanging from a tree branch away from the tree trunk, 1.5 m above the ground.Each feeder hung vertically and consisted of a PVC tube of 15 cm diameter and ~50 cm length, with an opening at the base, where a tray collected the seeds from the tube.We filled feeders with a mixture of unhusked sunflower seeds and husked peanuts and refilled them every few days to ensure a permanent supply of food.Feeders remained active throughout the winter from early November until the end of March when they were removed, a procedure that was repeated during four non-breeding seasons from 2015-2016 to 2018-2019.
Since first laid eggs appeared around late April, the experimental food provisioning stopped at least 5 weeks before the start of egg laying and ensured that reproductive females, of this non-hoarding species, did not directly rely on supplementary food for egg production.Each feeder was placed 500 m apart in an area covering one third of the study area (fed area from here onwards, Fig. 1).Thus, the study area was divided in two experimental areas, a northern fed area and a southern area that remained unmanipulated (hereafter the non-fed area, Fig. 1), separated by a buffer zone of ~1 km width.The only inhabited farmhouse close to the study area is in the northern part, within the experimental fed area, ensuring that no feeders were at close range from birds of the non-fed area, located at a minimum 1 km distance from any feeder (Fig. 1).Previous studies have shown northern wintering great tit populations to exhibit high site-fidelity to their artificial feeding locations (Pakanen et al. 2018).
We ringed wintering individuals during the non-breeding season by capturing them when roosting in the nestboxes within both experimental areas, and by mist netting at the feeders.All birds were measured, and colour banded to allow future identification without need of recapturing.Repeated nest box checks and mist netting, together with routine observations at the feeders of colour banded individuals allowed us to assess their site fidelity.

Monitoring of reproductive parameters
We recoded 974 breeding events during the breeding seasons of 2013-2019. Initially (2013-2016)), 293 nestboxes were spread over the study area.To extend nestbox availability in the unfed and buffer areas, we added 100 nestboxes to these areas in the winter of 2016-2017.We visited nestboxes on a weekly basis to detect the onset of egg laying and clutch size.From 2016 onwards, we captured the parents at some point after the chicks were 10 days old.Most parents were captured at the nestbox by means of mechanical traps, or with mist nets in the nest vicinity.Some colourringed individuals were identified by their colour bands without a need for capturing.We only included first breeding attempts per breeding season in this study.Supposed replacement clutches, i.e. new clutches appearing immediately after an active nestbox was abandoned in the vicinity, were removed from the data-set.Clutch size was considered complete whenever incubation by females was certified or suspected (based on the form of the nest), but clutches of less than two eggs were considered to have been interrupted and were excluded from the dataset.

Statistical analyses
We analysed variation in clutch size and laying date in relation to experimental treatment with Generalized linear mixed models.In both models, treatment was included as a fixed effect, with nestbox as random factor in proc GLIMMIX SAS 9.4.We also included year (as a fixed factor) and the interaction with treatment to account for possible temporal patterns.We initially tested for the difference between experimental areas on the years before the start of the experiment (2013)(2014)(2015) to exclude potential systematic differences between areas, and afterwards we restricted the data set to the experimental years (2016)(2017)(2018)(2019).Models were estimated by REML with type III tests, and df by the Satterthwaite method.Nestbox occupation rate was studied between fed and non-fed areas, before and after the experimental procedure.Number of occupied with respect to available nestboxes was analysed with a logistic regression, with experimental period (pre-experimental period 2013-2015 versus experimental period 2016-2019), experimental treatment (Fed versus Non-fed) and their interaction, as implemented in proc GENMOD SAS 9.4 (SAS Inst.Inc. 2009).Results from mixed models are presented with the F values, and W values for the logistic regression, together with df and p values.Parameter estimates ± SE are provided for continuous predictors.All p-values are two-tailed.All continuous variables fulfilled the requirements of normality.

Results
To study individual movement between the areas, we recorded the breeding area of great tit individuals, which had been captured or observed during winter.Out of 173 such individuals (84 males and 89 females) recorded more than once from autumn 2015 onwards, 168 (97%) remained faithful to the experimental area where they were initially detected.Three individuals moved from the buffer to the fed area (2 males and 1 female), one female from the unfed to the buffer area, and one male from the fed to the unfed area.Altogether, the results indicate that great tits in our study area are extremely faithful to their breeding and wintering grounds.Furthermore, breeding densities did not change as a result of the winter-feeding experiment.No difference in nestbox occupation rate was detected between pre-experimental years (2013)(2014)(2015) and experimental years (2016-2019), in either of the experimental areas (Fed: Wald test-W 1 = 3.31; p = 0.08; Non-fed: Wald test-W 1 = 0.57; p = 0.45; Fig. 2).
During the experimental years, birds from the fed area laid larger clutches (Fed: 9.23 ± 1.10) than those from the non-fed area (Non-fed 8.80 ± 0.11; F 1,188 = 8.21; p = 0.005; Fig. 3).Clutch size changed among years (F 3,188 = 5.82; p = 0.001; Fig. 3), a pattern that was unrelated to the feeding experiment (interaction year × treatment F 3,185 = 1.20; p = 0.31).However, during the winter-feeding experiment, birds from the fed area did not start egg laying earlier than birds from the non-fed area, albeit we found a tendency for fed birds to lay almost a day earlier (Fed: 27.98 ± 0.43) than birds from the non-fed area (Non-fed 28.85 ± 0.47; F 1,199 = 1.88; p = 0.17; Fig. 4).Laying date varied among years (F 3,199 = 14.9; p < 0.001; Fig. 4), a pattern that was unrelated to the feeding experiment (interaction year × treatment F 3,196 = 1.91; p = 0.13).During the three years prior to the winter-feeding experiment (2013)(2014)(2015), no difference between experimental areas was detected, neither in clutch size nor in laying date (all p > 0.1).However, also during this time period, we found differences between years for both clutch size (F 2,106 = 10.9;p < 0.001) and laying date (F 2,104 = 116.0;p < 0.001), without interactions between year and area (p > 0.6 for both clutch size and laying date).

Discussion
Winter food supplementation affected one of the breeding parameters studied as birds laid larger clutches when breeding within the winter-feeding area.However, birds from the winter-supplemented area did not start egg-laying earlier than birds from the non-fed area.Our data from seven years of study -three years without experimental supplementation, four years with winter food supplementation -indicates that density of breeding great tits was not affected by the experimental procedure, and breeding birds remained faithful to their respective wintering areas.Thus, our results suggest that winter feeding can have carry-over effects into the subsequent breeding season that are expressed in a larger investment in the number of eggs laid.
Reduced food availability during winter may affect population dynamics by limiting winter survival or constraining early reproductive investment, a situation that may be buffered in artificially food-supplemented populations.While food supplementation during reproduction has generally a positive effect on breeding performance (Ruffino et al. 2014), the effect of food supplementation before the onset of breeding varies remarkably, even within the same species.For example, blue tits breeding in the southwest (50°N) and in the south of the UK (51°N) showed no significant effects on laying date or clutch size when fed during winter (Plummer et al. 2013a, Crates et al. 2016), whereas in a northern population (54°N), winter feeding positively affected early breeding parameters (Robb et al. 2008b).Such differential carry-over effects might depend on the type of food artificially offered (Plummer et al. 2013a) or the predictability of food provisioning (Crates et al. 2016).However, as the two studies reporting significant carry-over effects (Robb et al. 2008b; this study) are from somewhat more northerly locations, it may suggest a dependence on environmental and natural food availability conditions that varies latitudinally.Furthermore, the pronounced between-year differences in both laying date and clutch size emphasises the importance of multi-year studies.Here, we show an effect on clutch size consistent across the four experimental years.
In the present study, we show that great tits from a winter supplemented area produced larger clutches, although they did not advance the breeding onset, which could be driven by several non-exclusive explanations.Firstly, it is likely that timing of breeding is affected by a multitude of environmental factors like temperature and precipitation, in addition to food availability (Svensson and Nilsson 1995).Secondly, although clutch size and laying date are two highly correlated breeding traits, their expression might depend on the different partners in the breeding pair.While males have a profound effect on the onset of egg laying via territorial behaviour and nest site selection, nest-building and clutch size rely to a large extent on female investment (Evans et al. 2020).Previous studies in the same population indicate that females (not males) gain more advantage from winter feeding than males as indicated by a different use of nocturnal hypothermia, in the fed compared  to the non-fed area (Nilsson et al. 2020).Thus, it could be possible that winter-feeding would promote differences in energy reserve levels between sexes, which could explain why we observe modulation of the breeding trait to be more heavily dependent on the condition of females.Finally, winter-feeding may have altered the trade-off between the size and number of eggs.If females in the fed area produced smaller eggs, females in the two areas may not differ in general investment into breeding.However, previous supplemental feeding experiments before and during egg laying among blue tits in a nearby area, did not affect egg size (Nilsson and Svensson 1993).
The observed carry-over effect of winter feeding on clutch size could be the result of a direct physiological effect resulting from a surplus of food, allowing such supplemented birds to increase investment in clutch size.However, supplemented birds could also be using winter food availability as a cue to adjust their breeding investment, independent of their body condition.Interestingly, Rytkönen and Orell (2001) found that great tits at the northern distribution range tended to produce a larger number of eggs compared to the number of young they could successfully raise.This apparent maladaptation was tentatively explained by the authors in terms of poor feeding performance in a newly colonized area.However, the realised clutch size might also have resulted from a misinterpretation of environmental quality as this population mostly relies on human-provided food during winter (Orell 1989).
Thus, winter feeding may be beneficial under certain conditions, but detrimental in others (Plummer et al. 2013a, Catto et al. 2021), potentially creating an 'ecological trap' (Barea andWatson 2013, Hale andSwearer 2016).Although our experiment did not result in an increased breeding density, winter feeding may have population consequences by attracting and improving survival of less-fit individuals, thereby increasing population density, and reducing average individual condition (Robb et al. 2008a).Additionally, winter food may attract predators and pathogens, influencing subsequent breeding performance.Further studies are needed to reveal whether supplementary feeding of wintering great tits leads to an ecological trap.
The fact that great tits in our study area exhibit a high site fidelity, suggests that our results do not originate from habitat sorting, resulting in differences in quality of the individuals between the experimental areas, but rather different physiological condition.Several potential mechanisms might explain cascading effects from winter to breeding.Food supplementation can directly affect timing of moult and subsequent spring migration phenology in several migrant species (Danner et al. 2013, 2015, Cooper et al. 2015).However, in resident species like blue and great tits, the mechanism responsible for the carry-over effect might instead be due to effects on individual physiological condition, with consequences for the trade-off between current parental effort and self-maintenance affecting future reproduction.Many physiological mechanisms involving energy metabolism, immunity, endocrine regulation and oxidative stress may interplay, with varying outcomes for example due to shared currency or metabolic paths (Cohen et al. 2012).For example, as part of such a trade-off, additional resources before breeding may improve the parental oxidative balance (Badás et al. 2015).In line with this, increased oxidative stress resulting from raised energy expenditure during breeding was reduced because of supplementary feeding in North American red squirrels Tamiasciurus hudsonicus (Fletcher et al. 2013).
Our results add to the accumulating evidence on the substantial and long-term consequences of supplementary feeding in wild animal populations (Robb et al. 2008a, Ruffino et al. 2014, Broggi et al. 2020).As supplementary feeding is becoming a more popular activity worldwide, it is important to reveal the underlying physiological mechanisms that link life-history stages, their effects on individual fitness, and the causes and consequences of the variation observed among species and populations, to increase beneficial use of food supplementation and prevent potential harmful consequences on wild fauna.

Figure 1 .
Figure 1.Study area located in Vombs fure in southern Sweden.Different experimental areas (Fed and Non-fed) are highlighted, with respect to inhabited areas in red.

Figure 2 .
Figure 2. Percentage of occupation (%) as the proportion of nestboxes occupied by great tits Parus major between 2013 and 2019 in two experimental areas.Occupation in the winter food-supplemented area (FED) is represented by filled columns, whereas the non-supplemented area (NONFED) is represented by open columns.Numbers above each column indicate the number of occupied boxes in relation to the total number of available boxes in the areas.The dashed line between 2015 and 2016 indicates the start of the winter-feeding experiment.

Figure 3 .
Figure 3. Clutch size of breeding great tits Parus major between 2013 and 2019 in the two experimental areas.Mean yearly clutch size and SE in the winter food-supplemented area (FED) is represented by filled columns, while the non-supplemented area (NONFED) is represented by open columns.The dashed line between 2015 and 2016 indicates the start of the winter-feeding experiment.

Figure 4 .
Figure 4. Laying date (1st April = 1) of breeding great tits Parus major between 2013 and 2019 in the two experimental areas.Mean yearly laying date and SE in the winter food-supplemented area (FED) is represented by filled columns, while the non-supplemented area (NONFED) is represented by open columns.The dashed line between 2015 and 2016 indicates the start of the winter-feeding experiment.