Incubation and hatching periods in a Mediterranean Great Tit Parus major population

Abstract

Capsule The onset of incubation relative to clutch completion is highly variable in Great Tits Parus major, and has important consequences for the duration of the incubation and hatching periods.

Aim To investigate when incubation starts relative to clutch completion, its effects on the length of the incubation and hatching periods, and which proximate factors affect all of these traits.

Methods We used data from a Great Tit population in Eastern Spain collected over 15 years. Periodic visits to the nests (daily at some stages) allowed the determination of breeding parameters of interest. General linear models were used for analyses.

Results On average, incubation started the day of laying of the last egg and lasted 13.2 days. The hatching period lasted 1.7 days. Incubation started earlier relative to clutch completion as temperatures during the laying period were higher and as clutch size increased. The incubation period was shorter if incubation started later relative to clutch completion. Hatching asynchrony increased as incubation started earlier relative to clutch completion and as the incubation period decreased.

Conclusions The onset of incubation relative to clutch completion is highly variable, and has important consequences for the duration of the incubation and hatching periods. Starting incubation before or after clutch completion could, respectively, advance or delay the hatching date, although advancement would be at the cost of increasing the degree of hatching asynchrony.

The incubation period is the time needed for the development of an embryo, assuming regular and constant attendance by the parents (Drent 1975). In passerines, incubation can start from several days before, to several days after clutch completion (Gibb 1950, Monrós et al. 1998, Ardia et al. 2006, Stenning 2008). However, the proximate causes of starting incubation sooner or later relative to clutch completion have been the subject of very few studies. From these, we know that incubation tends to start earlier relative to clutch completion if temperatures during the laying period are higher (Ardia et al. 2006, Vedder 2012; but see Wang & Beissinger 2009) and in larger clutches (Potti 1998, Wang & Beissinger 2009). Increasing temperatures as the season progresses would suggest that incubation should start earlier relative to clutch completion later in the season, but we are unaware of studies directly reporting this trend – although hatching patterns (see below) suggest that this is true.

The duration of the incubation period varies between individuals within populations. This within-population variability is most probably caused by environmental factors, because the duration of the incubation period has a low heritability (Husby et al. 2012). For example, the duration of the incubation period is shorter when temperatures during the incubation period are higher (Hepp et al. 2006, Ardia et al. 2010), and in part because of it, also late in the season (Wesołowski 2000, García-Navas & Sanz 2011). This likely because females are able to allocate more time to incubation in favourable environments, and the gradient between ambient and incubation temperature is reduced. It has also been shown that the incubation period increases with egg size (Parsons 1972, Martin & Arnold 1991), clutch size (Moreno & Carlson 1989, Hepp et al. 2005, Dobbs et al. 2006), and clutch mass (Hepp et al. 1990), because a larger number of and/or larger eggs will be more costly to incubate. However, recent studies regarding various species have found that shorter incubation periods occur in larger clutches (Wesołowski 2000, Ardia et al. 2006, García-Navas & Sanz 2011). Explanations put forward include that good-quality females could both produce larger clutches and provide higher incubation attendance, and that larger clutches cool more slowly during female absences and therefore are incubated at higher average temperatures (see also Reid et al. 2002). On the other hand, the brood patch is not completely developed and incubation behaviour not fully established until sometime after clutch completion (Bailey 1952, Jones 1971, Sockman et al. 2006), so birds starting incubation before clutch completion might then be expected to have longer incubation periods than those starting later (but see Ardia et al. 2006, García-Navas & Sanz 2011).

The most important consequence of starting incubation before clutch completion is that eggs will hatch asynchronously (i.e. hatching occurs over a period of 24 hours or more; Haftorn 1981a, Ardia et al. 2006). Thus, the same factors potentially affecting the onset of incubation relative to clutch completion (temperatures during the laying period, date, clutch size, and egg size) have been found to be related with the degree of hatching asynchrony (Slagsvold 1986, Potti 1998, Ardia et al. 2006, Stenning 2008, Theofanellis et al. 2008, Podlas & Richner 2013), the most frequent trends being an increase of hatching asynchrony with clutch size and as the season progresses.

In birds, embryonic development typically starts when eggs are actively incubated, and this usually means that eggs are constantly warm (Drent 1975, Haftorn 1983, Wiebe et al. 1998). In Great Tits Parus major, only females incubate, and they usually cover the eggs with nest material during the day until incubation starts (Haftorn & Slagsvold 1995). Therefore, the onset of incubation could be assumed to occur when eggs are seen uncovered and warm during the day, or when females are seen incubating during the day (Gibb 1950, Stenning 2008, García-Navas & Sanz 2011). Great Tit females roost in the nest during the night during the laying period (Pendlebury & Bryant 2005) until the onset of incubation, suggesting that early eggs may gain more incubation than later eggs (Lord et al. 2011). Furthermore, we found that, on many occasions, eggs were completely uncovered yet cold at the end of the laying period (Barba 1991), suggesting also that eggs may receive some incubation before the start of full incubation. Both of these processes may allow Great Tits to shorten incubation but at the expense of increasing hatch asynchrony (Cresswell & McCleery 2003).

The Great Tit is a widespread species in Europe. Its use of nest boxes has encouraged many research groups to use it as a model species for a wide range of ecological and evolutionary studies, some of them carried out over several decades (Lambrechts et al. 2010). One of these studies has been carried out in an extensive orange plantation in Sagunto, Eastern Spain, where the breeding performance of Great Tits has been followed since 1986 (Encabo et al. 2001, Tomás et al. 2012). For most of these years, data on the onset of incubation and on the duration of the incubation and hatching periods of a sample of nests are available.

Our main objective was to describe within-population variation in the onset of incubation relative to clutch completion, the factors affecting this, and its effects on the duration of incubation and hatching periods in the Great Tit population of Sagunto. To our knowledge, this is the first study dealing with the incubation and hatching periods, and factors affecting them, in a southern European Great Tit population. Based on the studies cited above, we predicted that: (1) the onset of incubation will be earlier relative to clutch completion, (2) the incubation period will be shorter, and (3) the hatching period longer if ambient temperatures are higher, as the date of laying of the first egg (or onset of incubation for incubation period) is later, and as clutch size (or clutch volume for incubation period) increases. We also tested for the effects of uncovering the eggs before the onset of full incubation, because this may indicate partial incubation (sensu Wang & Beissinger 2011). If this is the case we predict that the ‘full’ incubation period will be shorter, and hatching period longer.

METHODS

Study area and data selection

The study was performed on a Great Tit population breeding in nest boxes in an extensive orange plantation in Sagunto, Eastern Spain (39°42′N, 0°15′W, 30 m a.s.l.). Basic breeding parameters have been recorded each year since 1986 (Encabo et al. 2001, Tomás et al. 2012) during periodic visits to the nests. These were the laying date of the first egg, clutch size, and egg volume. Nest boxes were visited at least once per week, and laying of the first egg was estimated assuming that one egg was laid per day. Laying gaps (days during the laying period where no egg was laid) occurred only very occasionally under unfavourable environmental conditions (Monrós et al. 1998), and we considered that the bias they could introduce was negligible. Once a nest was found with eggs, it was visited daily from the day when the fifth egg was expected to be laid until clutch completion. Laying was assumed to have finished, and the clutch to be complete, when no more eggs where laid during two consecutive days and the female started full incubation. The length (L) and width (W) of the eggs of a sample of nests each year were measured with a calliper (±0.01 mm) when the clutch was complete. The volume (V) of each egg was calculated as V = (0.4673 × L × W²) + 0.042 (Ojanen et al. 1978, Encabo et al. 2001). Clutch volume (sum of the volumes of all the eggs in a clutch) was computed for each clutch.

For the present study, we included only first clutches laid between 1988 and 2010. As some of the parents were not individually identified (none in early years), we considered first clutches as those where the first egg was laid within 30 days from the laying of the first egg of that year (Nager & van Noordwijk 1995, Møller et al. 2014). Most pairs (70%, n = 1674 pairs) only attempted one clutch per year, but some (19%) produced a second clutch if the first clutch failed, and others (11%) attempted a second clutch after a successful first clutch (own unpubl. data, period 1988–2010). From this initial sample size we excluded nests used in experiments not part of this study which could affect the parameters under study here, and nests where the laying date and clutch size could not be determined. From these, the exact determination of the parameters of interest (onset of incubation relative to clutch completion and duration of the incubation and hatching periods) could be achieved for a variable number of nests each year (see criteria below). After application of these restrictions, we completely excluded years with less than ten valid nests for all of the studied parameters (1989, 1997, 1998, 2002–5, and 2008). From the remaining 15 years, with a total of 1098 first clutches initiated in the study area, subsamples for analyses concerning the onset of incubation included 609 nests from 15 years, those concerning the incubation period included 356 nests from 12 years, and those concerning the hatching period, 429 nests from 15 years. In all cases, selected nests are thought to be a random sample of the population.

Ambient temperature data (daily maximum and minimum) of a nearby (ca. 4 km) meteorological station were obtained from the Spanish Meteorological Agency. Daily mean temperatures were computed as the average of maximum and minimum of each day. For each nest, we computed the mean of the mean daily temperatures for the laying period (from the day of laying of the first egg to the day prior to incubation start), and for the incubation period (see below).

The incubation period

We defined the incubation period as the number of days elapsing from the day when incubation was first observed and the day prior to the hatching of the first egg. We only included nests where the incubation period was determined precisely by daily visits to the nests from at least one day before incubation started and from at least one day before hatching started. For each nest the onset of incubation relative to clutch completion was calculated by subtracting the date of start of incubation from the date of clutch completion, being ‘0’ if incubation started on the day of laying of the last egg, ‘−1’ if it started with the penultimate egg, ‘+1’ if it started the day after clutch completion, and so on (Loos & Rohwer 2004, Ardia et al. 2006).

In our daily visits, we noted whether the eggs were covered or uncovered by nest material prior to the onset of incubation. We classified nests into two groups, those in which eggs were uncovered the day of onset of full incubation (i.e. eggs were warm the first day they were uncovered), and those where eggs were completely uncovered, but cold, before starting full incubation.

The hatching period

In a sample of nests each year, nests were visited daily from incubation day 12 until all the eggs had hatched, or the remaining eggs did not hatch in two consecutive days. We used these nests to estimate the hatching period, defined as the time in days from hatching of the first egg to hatching of the last successful egg (i.e. excluding those that did not hatch) inclusive (Potti 1998). Therefore, the hatching period was one day if all the eggs hatched on the same day. Great Tits are unable to remove unhatched eggs from the nest, but they do remove nestlings which die in their first days of life. Because we made daily visits, we assumed that all disappeared eggs had hatched, and that missing nestlings had died and were taken out of the nest by the parents.

Statistical analyses

We used general linear models (GLMs) to explore the relationships between the target variables (onset of incubation relative to clutch completion, and duration of the incubation and hatching periods) and the factors predicted to affect them. The year and whether the eggs were uncovered or not before the onset of incubation were included as factors in all the models. The rest of the independent variables were included as covariates in each model as appropriate. In the model of the hatching period, the onset of incubation relative to clutch completion and its square value were included to account for a potential nonlinear relationship – for example, starting incubation before clutch completion may affect hatching asynchrony but starting incubation after clutch completion may have little or no effect. As this variable had negative and positive values, the lower value (−3) was considered to be ‘1’, and the rest of the values were arranged consequently (i.e. adding four units to each one) for this analysis, although we preserved the original nomenclature in the text and figures.

Means ± sd are presented when appropriate. The sample sizes used to estimate mean onset of incubation or the duration of the incubation and hatching periods are smaller when performing GLMs because not all nests had complete sets of measurements. Statistical analyses were carried out using SPSS v. 19.0.

RESULTS

Onset of incubation

The onset of incubation relative to clutch completion varied between years (Table 1). Yearly means were positive (i.e. incubation starting after clutch completion) for 8 out of the 15 years, negative for 6 years, and 0.00 in 1988. Incubation started from 3 days before to 6 days after clutch completion (Fig. 1). Overall, females started incubation on the day of laying of the last egg (mean = 0.06 ± 0.30, n = 15 years, including 609 nests).

Figure 1. Frequency distribution of the onset of incubation relative to clutch completion. ‘0’ is the date of laying of the last egg, negative values indicate that incubation started before clutch completion, and positive values indicate that incubation started after clutch completion. Sample sizes (number of nests) are indicated above the bars.

Table 1. Results of a GLM of the influence of year, uncovering of the eggs before incubation (UNC), mean temperature during the laying period (TLAY), laying date, and clutch size on the onset of incubation relative to clutch completion. The biological significance is shown in Figs 2 & 3 and given in the text for uncovering of the eggs before incubation.

Parameter	df	F	P
Year	14,590	2.9	< 0.001
UNC	1,590	54.1	< 0.001
TLAY	1,590	15.0	< 0.001
Laying date	1,590	0.8	0.37
Clutch size	1,590	7.2	0.007

The onset of incubation was later relative to clutch completion if the eggs were uncovered before the onset of incubation (0.7 ± 1.4, n = 134 vs. −0.2 ± 1.4, n = 475; Table 1). Of the 134 nests where eggs appeared uncovered and cold before the onset of incubation, eggs appeared uncovered on the day of clutch completion in 28 nests, eggs appeared uncovered before clutch completion in 87 nests (mean = −1.7 ± 1.7 days), and eggs appeared uncovered after clutch completion in 19 nests (mean = 2.1 ± 1.8 days). Nests where eggs appeared uncovered on the date of clutch completion started incubation 1.4 ± 0.8 days after clutch completion, those where eggs were uncovered before clutch completion started incubation, on average, on the day of clutch completion (mean = −0.1 ± 1.1 days), while those where eggs appeared uncovered after clutch completion started incubation 3.1 ± 1.6 days after clutch completion. Also, incubation started earlier relative to clutch completion as temperatures during the laying period were higher, and as clutch size was larger (Table 1; Figs 2 & 3).

Figure 2. Relationship between the onset of incubation relative to clutch completion (expressed as residuals after controlling for year, uncovering of the eggs before incubation and clutch size; see Table 1) and mean temperature during the laying period of each clutch.

Figure 3. Relationship between the onset of incubation relative to clutch completion (expressed as residuals after controlling for year, uncovering of the eggs before incubation and temperature during laying; see Table 1) and clutch size. The mean ± sd residual of onset of incubation is shown for each clutch size (sample size above bars), but the regression line was fitted using the individual data points.

Incubation period

The mean incubation period was 13.2 ± 0.4 days (n = 12 years, including 356 nests), though it varied between years (Table 2), ranging from 12.6 (years 1992 and 2006) to 13.8 days (1996). Among the factors considered to potentially affect the incubation period, we only found that incubation period was shorter as incubation started later relative to clutch completion (Table 2; Fig. 4).

Figure 4. Relationship between the incubation period and the onset of incubation relative to clutch completion. The mean ± sd residual of the incubation period is shown for each day of onset of incubation relative to clutch completion (sample size above bars), but the regression line was fitted using the individual data points.

Table 2. The results of a GLM of the influence of year, uncovering of the eggs before incubation (UNC), mean temperature during the incubation period (TINC), date of the start of incubation (ID), clutch volume, and the onset of incubation relative to clutch completion (OINC) on the duration of the incubation period as illustrated by a GLM. The biological significance is shown in Fig. 4.

Parameter	df	F	P
Year	11,316	5.2	< 0.001
UNC	1,316	0.5	0.48
ID	1,316	2.4	0.12
OINC	1,316	15.0	< 0.001
TINC	1,316	0.2	0.64
Clutch volume	1,316	1.8	0.18

Hatching period

The hatching period also varied between years (Table 3), and ranged between 1.0 and 2.9 days (mean 1.7 ± 0.6 days, n = 15 years, including 429 nests). The hatching period was longer as incubation started earlier relative to clutch completion and as the incubation period was shorter (Table 3; Figs 5 & 6). Starting incubation 5–6 days after clutch completion was rare (0.82% of the clutches; see Fig. 1), and we only have one nest with a complete data set that was included in this analysis. We repeated the analysis removing this single point, and in this case both the onset of incubation relative to clutch completion and its squared term (F _1,148 = 8.9, P = 0.003) were significant (Fig. 5). Thus, starting incubation before clutch completion had an effect on hatching asynchrony while starting on, or after clutch completion, had virtually none.

Figure 5. Duration of the hatching period (expressed as residuals after controlling for year and incubation period; see Table 3) and onset of incubation relative to clutch completion. The mean ± sd residual of hatching period is shown for each day of onset of incubation relative to clutch completion (sample size above bars), but the regression line was fitted using the individual data points. The regression line has been adjusted removing the point corresponding to a nest where incubation started six days after clutch completion (see text).

Figure 6. Relationship between the duration of the hatching period (expressed as residuals after controlling for year and onset of incubation relative to clutch completion; see Table 3) and the duration of the incubation period. The mean ± sd residual of hatching period is shown for each incubation period (sample size above bars), but the regression line was fitted using the individual data points.

Table 3. The results of a GLM of the influence of year, uncovering of the eggs before incubation (UNC), mean temperature during the laying period (TLAY), clutch size, date of the start of incubation (ID), incubation period, and the onset of incubation relative to clutch completion (OINC) and its square (OINCSQ) on the duration of the hatching period. The biological significance is shown in Figs 5 & 6.

Parameter	df	F	P
Year	9,149	3.6	< 0.001
UNC	1,149	0.7	0.40
TLAY	1,149	2.0	0.15
CS	1,149	0.5	0.47
ID	1,149	1.0	0.33
Incubation period	1,149	37.3	< 0.001
OINC	1,149	12.3	0.001
OINSQ	1,149	2.5	0.12

DISCUSSION

This study emphasizes the crucial role of the time of starting incubation relative to clutch completion on the duration of the incubation period and the degree of hatching asynchrony. The fact that passerine birds can start incubation some days before or after clutch completion has been known for decades (Gibb 1950, Zink 1959, Stenning 2008), and we confirmed it here in our Great Tit population. However, most studies have concentrated on the incubation before clutch completion, because this usually leads to hatching asynchrony, a phenomenon which has attracted much attention (Clark & Wilson 1981, Magrath 1990, Stenning 1996). The fact that many species could start incubation after clutch completion has been so overlooked that it was claimed to have been ‘discovered’ quite recently (Rowe & Weatherhead 2009: ‘we are unaware of any studies of either altricial or precocial birds that have reported a delay of incubation onset of >24 h beyond clutch completion’). But several preceding and recent studies have shown that starting incubation after clutch completion is not only common, but it also contributes to the adjustment of the hatching dates to prevalent environmental conditions, thus enhancing breeding success (Monrós et al. 1998, Cresswell & McCleery 2003, Stenning 2008, García-Navas & Sanz 2011).

Concerning proximate causes, we found that Great Tits started incubation earlier in the laying sequence when temperatures during the laying period were higher. Slagsvold (1986) suggested that higher temperatures around laying allow females to be in a better condition, so that they can start incubation before clutch completion. Potti (1998) also suggested that female condition has a causal role in the onset of incubation. Nilsson (1993) showed that female Marsh Tits Poecile palustris receiving extra food during laying started incubation earlier relative to clutch completion than control pairs, showing an energetic limitation to the start of incubation. Alternatively, starting incubation early when temperatures are higher could be viewed as a strategic decision to synchronize hatching dates with food availability (García-Navas & Sanz 2011). The negative relationship between temperature and onset of incubation also gives support for the egg viability hypothesis (Arnold et al. 1987, Amundsen & Stokland 1988), which states that the viability of unincubated eggs could decline with time, especially if temperatures are high, so incubation should start during the laying period to avoid egg losses.

Another proximate factor affecting the onset of incubation relative to clutch completion was clutch size: females laying larger clutches tended to start incubation earlier relative to clutch completion. This trend has also been found in several other species, and has been interpreted as either the willingness of females to accelerate hatching if the laying period is prolonged, since recruitment probabilities of fledglings decline with time, or as a correlation between female condition and clutch size, with females which are in better condition laying larger clutches and being able to start laying earlier relative to clutch completion (Magrath 1992, Potti 1998, Wang & Beissinger 2009, García-Navas & Sanz 2011).

For Great Tits, calculation of the incubation period in other European populations has frequently included the hatching period, and its values range between 13.5 and 16.3 days (Gibb 1950, Kluijver 1950, Winkel 1970, 1975, Haftorn 1981b). The incubation and hatching periods defined here do not overlap, so their sum (13.2 + 1.7 = 14.9 days) is comparable to the incubation periods in these previous studies. Many studies have shown that incubation periods are generally longer in species breeding at lower latitudes (Ricklefs 1969, Martin et al. 2007, Robinson et al. 2008). With the caution that available estimates were calculated slightly differently, and using different criteria, the Sagunto population has a longer incubation period than all northern populations, excluding that from Norway (16.3 days, Haftorn 1981b).

Given the predictions derived from the previous studies, it was surprising that the incubation period was only related to the day of the start of incubation relative to clutch completion. The negative relationship supports our hypothesis that incubation is less effective during the laying period, though there is also the possibility that nests which start incubation later relative to clutch completion receive more partial incubation. This contrasts with previous studies, where clutches where incubation started before clutch completion had shorter incubation periods compared to clutches where incubation started on or after laying of the last egg (Viñuela 1997, Ardia et al. 2006, García-Navas & Sanz 2011). Prolactin is one of the hormones triggering the formation of the brood patch and maintaining incubation behaviour (Lea & Klandorf 2002, Angelier & Chastel 2009). However, this hormone also causes a regression of the gonads (Dawson et al. 2001), probably creating conflicts between producing new eggs and incubation behaviour (Haftorn 1981b, Mead & Morton 1985, Sockman et al. 2000, 2006). It is, therefore, possible that, if incubation starts during the laying period, it might be less effective than starting once the physical structures and ethological repertory is fully developed, which probably occurs after clutch completion (see Runde & Barrett 1981, St. Clair 1998). Most previous studies on the duration of the incubation period have not included the day of onset of incubation relative to clutch completion among the predictor variables. Here, we show that this factor might be crucially important in determining the incubation period.

The duration of the hatching period in the studied population (1.7 days) was in the lower extreme of the range found for other Great Tit populations (1.8–4 days) (Gibb 1950, Neub 1979, Haftorn 1981b, Orell 1983, Amundsen & Slagsvold 1998, Tilgar & Mänd 2006, Theofanellis et al. 2008, Matysioková & Remeš 2011, Podlas & Richner 2013). This was not unexpected, taking into account that an average female in Sagunto started incubation the day the final egg was laid.

Our data also show the expected negative relationship between the onset of incubation relative to clutch completion and the duration of the hatching period if incubation started before clutch completion, while it virtually had no effect if it started on or after laying the last egg. Therefore, starting incubation before clutch completion could control both hatching asynchrony and date of hatching, while delaying the onset of incubation beyond clutch completion would delay the hatching date, without apparent effects on hatching asynchrony. The (relatively small) variability in the degree of hatching asynchrony observed when incubation started on or after clutch completion, could have partly been caused by some eggs receiving partial incubation before clutch completion. We have shown above that temperatures during laying and clutch size affect the onset of incubation relative to clutch completion, so these factors also have an indirect effect on hatching asynchrony.

Contrary to expectations, we found that hatching asynchrony increased as the incubation period decreased. A potential explanation for this relationship is that partial incubation, occurring over several days (both prior and after clutch completion, before the onset of full incubation) initiated embryo development, resulting in both a shortening of the (full) incubation period and an increase in hatching asynchrony. Lord et al. (2011) showed that differences in the amount of incubation received by individual Great Tit eggs prior to clutch completion significantly affected the time they needed to hatch after full incubation started. For example, Potti (1998) noticed that many Pied Flycatcher Ficedula hypoleuca clutches where incubation started after the last egg was laid exhibited hatching asynchrony; he suggested that the amount of incubation received by individual eggs prior to clutch completion was the main reason of this unexpected asynchrony.

In conclusion, temperatures during laying and clutch size were related to the onset of incubation relative to clutch completion, which in turn has important consequences for the duration of the incubation and hatching periods. Therefore, starting incubation before or after clutch completion could, respectively, advance or delay the hatching date, though any advancement would be at the cost of increasing the degree of hatching asynchrony.

ACKNOWLEDGEMENTS

We would like to thank all of the people who helped in the field during the study years, and to Will Cresswell, Alex Lord, and an anonymous reviewer for useful comments on previous drafts. We also thank the Spanish Meteorological Agency for providing temperature data, and the Spanish Ministry of Agriculture, Food and Environment for providing nest boxes.

Funding

This long-term study of Great Tits was partially supported by: GV-2517/94 (Generalidad Valenciana), CGL2004-00787, CGL2007-61395 (Spanish Ministry of Education and Science), and CGL2010-21933-C02-02 (Spanish Ministry of Science and Innovation).