Pair bond and breeding success in Blue Tits Parus caeruleus and Great Tits Parus major

Ibis (25), 147, 92 18 Blackwell Publishing, Ltd. Pair bond and breeding success in s Parus caeruleus and s Parus major MIRIAM PAMPUS*, KARL-HEINZ SCHMIDT & WOLFGANG WILTSCHKO Fachbereich Biologie der J.W. Goethe-Universität Frankfurt, Zoologie, D-654 Frankfurt, Germany Data from 939 nests of the Parus caeruleus and 18 nests of the P. major from nestboxes provided in superabundance in mixed forest study sites between 1976 and 21 were analysed to examine the effects of mate retention on breeding success and the relationship between mate fidelity and site fidelity. Most birds retained their former partner (76% in s and 65% in s). The probability of a pair divorcing was affected by male age in s, divorce being more likely in pairs with first-year males. pairs breeding together for a second season bred earlier, but had no higher breeding success than pairs breeding together for the first time. In s laying date and start of incubation tended to be earlier in pairs breeding together for a second season, but hatching and fledging dates were not earlier than in other pairs. pairs breeding together for two consecutive seasons bred earlier in the second season than in the first, but breeding success did not differ significantly between years. In both species, breeding performance did not differ between pairs that divorced after a season and pairs that stayed together. Thus breeding success did not determine whether a pair divorced or bred together again. Neither s nor s improved their breeding performance through divorce. females even had fewer fledglings in the year after divorce than in the year before. Mate retention affected breeding site fidelity. females had greater breeding dispersal distances between consecutive years when re-mating than when breeding again with the same mate. In s both males and females dispersed more when re-mating than when retaining the former partner, suggesting that mate retention increased the chance of retaining the breeding site. In both species, breeding dispersal distances did not differ between pairs that divorced and pairs in which one mate disappeared. Because no major advantage of mate retention was evident, we suggest that mate retention evolved under different conditions than those found in study sites with high breeding densities and a superabundance of artificial nesting sites. Long-term pair bonds are common in many large long-lived bird species but do also occur in small short-lived passerines (Dhondt & Adriaensen 1994, Choudhury 1995, Ens et al. 1996, Dubois & Cézilly 22). Thus, natural selection must have favoured birds that stay together with their mate for more than one season. Mate retention might be favourable either because changing mates is costly or because the familiarity with the old mate (and perhaps breeding site) improves lifetime reproductive success (Ens et al. 1996). Mate retention can increase lifetime reproductive success in various ways: the most direct way is to influence reproductive *Corresponding author. Email: m.pampus@gmx.de performance itself. Although a positive effect of mate retention on reproductive success is evident in several long-lived species (Coulson 1966, Fowler 1995), the advantage of long-term pair bonds is less clear in short-lived species such as s Parus major and s P. caeruleus. Mate retention (Perrins & McCleery 1985) and divorce (Dhondt 1986, Lindén 1991, Dhondt & Adriaensen 1994, Choudhury 1995, Ens et al. 1996, Dubois & Cézilly 22) have been regarded as two sides of a strategy to increase reproductive success in s and s as both occur regularly in these species (Winkel & Winkel 198, Perrins & McCleery 1985, Dhondt & Adriaensen 1994, Dhondt et al. 1996, Blondel et al. 2, Saitou 22). Although finding or fighting for a new mate (and perhaps territory) can be costly, 24 British Ornithologists Union

Pair bond and breeding success in Blue and s 93 and reproductive success might even decrease after a change of mate (Ens et al. 1996), divorce can also be advantageous under certain conditions and might therefore form an adaptive strategy. To what extent divorce is adaptive can be examined through two main hypotheses. First, the incompatibility hypothesis suggests that a change of mate might be beneficial for both partners if they do not complement each other well, for example because they are closely related (Ens et al. 1996). Secondly, the better option hypothesis suggests that a change of mate should only benefit one partner, who leaves the other in favour of a partner or territory of higher quality (Orell et al. 1994, Ens et al. 1996). Both hypotheses predict an improvement in reproductive success through divorce. By contrast, there are also nonadaptive explanations for the occurrence of divorce and mate retention. A mate can be forced away from its partner by a third individual (Choudhury 1995, Ens et al. 1996). A splitting up of pairs can also occur by chance, the mates accidentally losing each other (Dhondt & Adriaensen 1994, Saitou 22). Remating with the same partner may occur by chance as well, when both partners return independently to the same breeding site and meet there again. In cases where non-adaptive reasons explain either materetention or divorce, no difference in reproductive success between faithful and divorced pairs and others should be expected. In very short-lived species such as s and s with mortality rates of around 5% (Perrins & McCleery 1985, Dhondt & Adriaensen 1994, Dhondt et al. 1996), relatively few cases of mate retention and divorce can be studied, i.e. cases in which both members of a breeding pair are known to have survived to a subsequent breeding season. Therefore, only few long-term studies provide adequate data sets to study this matter and the results of these studies are, in part, contradictory. In Wytham Woods, UK, Perrins and McCleery (1985) found that mates breeding together for a second season had slightly larger clutches and a slightly greater number of surviving offspring than mates breeding together for the first time. Pairs that divorced afterwards had significantly smaller clutches and tended to lay later than pairs which stayed together, indicating that pairs may divorce because of poor breeding success. In Japan, Saitou (22) found no difference in breeding success between faithful and divorced s. Faithful pairs laid earlier, had larger clutches and fledged more offspring than the population mean, while breeding performance of pairs that divorced afterwards did not differ from the population mean (Dhondt & Adriaensen 1994). Female s bred more successfully after divorce than did pairs that remained together (Dhondt & Adriaensen 1994). In Corsica, Blondel et al. (2) found no difference in breeding success between faithful and divorced Blue Tit pairs. Again, female s improved their breeding success through divorce, supporting the better option hypothesis, whereas males did not do better after divorce. In nests with unhatched eggs, both mates could improve their hatching rate through divorce, supporting the incompatibility hypothesis. However, divorce status did not directly predict the number of unhatched eggs in the nest (Kempenaers et al. 1998). If mate retention or divorce affects breeding success, this might occur in various stages of the breeding cycle and to different degrees, finally increasing lifetime reproductive success. The true measure of lifetime reproductive success is the total number of descendants of an individual. Because this cannot be determined in a free-living tit population, most studies have used clutch size, number of fledged offspring and laying or fledging date as measures of reproductive success, assuming that early laying and fledging leads to a better survival of the offspring (Magrath 1991, Dhondt et al. 1996, Naef-Daenzer et al. 21). Apart from fledging date, the condition of offspring when fledging should have a large influence on their probability of survival (Magrath 1991, Dhondt et al. 1996, Naef-Daenzer et al. 21). Some studies have used the number of offspring recruited to the population as a measure of the number of surviving offspring (e.g. Perrins & McCleery 1985). This can be misleading, however, because recruitment rates are often very low (about 3% in our own long-term study) and juvenile dispersal is difficult to distinguish from juvenile mortality in many tit populations. Several factors can cause differences in the timing of breeding between pairs breeding together for a second season and newly formed pairs. Faithful pairs may start nest building earlier than newly formed pairs because they spend less time on courtship behaviour. On the other hand, males of faithful pairs might invest more time and energy in feeding the female during egg formation, possibly leading to earlier laying, and during incubation, which might shorten the incubation period. Faithful pairs might also be more efficient or invest more in feeding the hatchlings, thus shortening the hatchling period.

94 M. Pampus, K.-H. Schmidt & W. Wiltschko Apart from mate fidelity, both s and Great Tits show high breeding site fidelity (Kluyver 1951), yet little is known about how these two factors (mate and site fidelity) influence each other. Breeding site fidelity could explain the occurrence of mate retention in a non-adaptive way (see above). Breeding dispersal distances might also indicate which bird is leaving when a divorce occurs. In this paper we address the following questions. Can s or s improve their breeding success through mate retention? Can one or both partners improve their breeding success through divorce? Do faithful pairs start breeding earlier and do they have shorter breeding periods than other pairs? Can non-adaptive reasons account for the occurrence of mate retention and divorce? How do breeding site fidelity and mate fidelity influence each other? In addition to frequently used measures such as clutch size, and numbers of hatchlings and fledglings, we use hatchling and fledgling weight as a measure of offspring quality determining the chances of survival (Magrath 1991, Dhondt et al. 1996, Naef-Daenzer et al. 21) and the chances of settling early in a good territory (Wilson 1992). To determine how the timing of breeding is affected by mate retention, we examined different dates from nest building to fledging and the duration of incubation and of hatchling period. METHODS Study sites We analysed data from a long-term study of s and s (Schmidt 1983) collected in three different study sites in the area of Schlüchtern, Hessia, Germany (5 19 N, 9 28 E, 2 3 m above sealevel). Two of the three study sites consisted of mixed deciduous forest: one of 1 ha contained 18 artificial nestboxes, the other of 2 ha had 3 nestboxes. The nestboxes were distributed in rows 25 m apart; the distance between nestboxes within a row was also 25 m. In these two study sites, mean breeding density was 17.23 and 16.19 pairs per 1 ha and 42.8 and 7.34 pairs per 1 ha, respectively. The third site consists of 3 nestboxes linearly distributed along the forest edge, in shrubs and meadows along both sides of a highway over a distance of 3.5 km. One of the study sites with even distribution was visited daily; the other two study sites were visited at least once a week during the breeding season. Data-set The data-set includes data from 1976 to 21 from the two study sites with even distributions of nestboxes and data from 1992 to 1997 from the study site with a linear distribution of nestboxes. The incidence of mate retention and divorce in the populations was calculated from 939 nests and 18 nests where mating status was known. To study the effects of pair bond on reproductive success, 837 nests and 88 nests were examined. This data-set includes only nests where the identity of both parents was known and no predator influence on the brood could be detected. Second broods and replacement broods as well as nests where the male was proven to be polygynous were excluded from the analysis. Because the parents are not caught before the hatchlings are at least 5 days of age (in order to minimize the risk of brood abandonment), nests where no young reached this age are excluded from the dataset. This might bias the results, because if pairs breeding together for a second year had fewer early brood losses than newly formed pairs, this could not be detected. To examine the timing of breeding we used only data collected on a daily basis in one of the three study sites. Standardization of data In order to take out year effects, the data for every nest in each year were standardized by subtracting the population mean of the year and dividing by the standard deviation of the year (i.e. as z-scores). Thus, the breeding performance of each pair is expressed in relation to the population mean of the year in which it bred. Although most tests were carried out using z-scores, we sometimes show the unstandardized values also in order to facilitate understanding of the possible biological meaning of the results. Data analysis First, we carried out a MANOVA testing the breeding parameters against the age classes of the male and female (1st year or older). Only pairs with known age of both mates were included in this analysis. Because pairs breeding together for a second season can only occur in the group of old males with old

Pair bond and breeding success in Blue and s 95 females, these faithful pairs were excluded from the analysis of age effects. Because age effects were apparent both in timing of breeding and in breeding success, we only used data from pairs of older males and older females to analyse the effect of mate retention. We tested whether breeding performance or timing differs between pairs breeding together for a second year and pairs breeding together for the first time, using a t-test. To determine whether pairs split up because of poor breeding success or stayed together because of good breeding success, we compared the breeding performance of pairs in the year before splitting up with pairs in the first of two consecutive years of breeding together. To determine whether individuals can improve their breeding performance by either staying with the old mate or divorcing and breeding with a new mate, we used the Wilcoxon test for paired samples. In divorced birds we tested whether breeding performance after divorce differed from breeding performance before divorce. In pairs breeding together for a second season we tested whether breeding performance differed between the first and second year. We then analysed the influence of pair bond on territoriality, comparing the distances between breeding sites of consecutive years in birds retaining their former partner, birds after divorce and widowed birds. Age classes Age was determined by plumage (Svensson 197) and all birds were divided into the two categories young (Y = in the first year of life) and old (O = in the second year or older). Thus, four types of pairings are possible: young female with young male (YFYM), young female with old male (YFOM), old female with young male (OFYM) and old female with old male (OFOM). Mating status In any year a bird can either be breeding with its former mate (MR = mate retention) or with a new mate (NP = new pair). Breeding with a new mate can be due to the divorce of the former pair (SP) or to a loss of mate (LM). Splitting up was defined as breeding with a new mate, while the former mate is observed breeding in the same or a following year. Loss of mate was defined as breeding with a new mate, when the former mate was not recorded breeding again in any following year. Breeding performance As measures of reproductive success, we used clutch size (number of eggs laid in the clutch), number hatched, mean hatchling weight on the 1th day (where day 1 = hatch day), weight of the heaviest nestling on the 1th day, number fledged, mean fledgling weight (= weight on the 15th day), weight of the heaviest fledgling, hatching rate and fledging rate (number fledged divided by number hatched). As measures of the timing of breeding we recorded end of nest building period (when the first lining material is found inside the nestbox), laying date of the first egg, onset of daytime incubation, hatching date, fledging date, period of incubation (from beginning of daytime incubation to hatching) and nestling period (days from hatching to fledging). For the timing parameters only data from one study site were used, which was visited on a daily basis. Fledging date is defined as the first day the nestbox was found empty of nestlings. The exact fledging day is either this day or one day earlier. Breeding dispersal To examine breeding site fidelity in relation to mate fidelity, only data from the two study sites with evenly distributed nestboxes were used. The distance between nestboxes used by an individual bird in consecutive years was measured using a 1 : 1 topographical map. If a second brood or replacement brood was started in a different nestbox, the distance was taken between the second nestbox and the nestbox where the first brood was recorded in the following year. RESULTS Most of the pairs where both mates survived to the next breeding season bred together again. Divorce occurred in 24% of pairs and in 35% of pairs (Table 1). Table 1. Proportion of divorced and retained partners in Blue and pairs. Percentages are given in parentheses. Total number of pairs 939 18 Both survived 92 (9.8) 12 (11.9) Stayed together 6 (65.22 a ) 91 (75.83 a ) Split up 32 (34.78 a ) 29 (24.17 a ) a Of pairs where both survived.

96 M. Pampus, K.-H. Schmidt & W. Wiltschko Table 2. MANOVA on breeding success with male age, female age and interaction of male and female age as explanatory variables. Male age Female age Male age* Female age Dependent variable Age class Mean a sd a n F P F P F P Clutch size YMYF.728.9666 35 YMOF.13.9352 11 OMYF.981.9269 188 OMOF.2412.9334 172 2.12.157 8.495.4 3.125.77 Number hatched YMYF.65.9593 35 YMOF.255.987 11 OMYF.368.9137 188 OMOF.151 1.232 172 1.95.163 2.47.121 1.23.312 Number fledged YMYF.479.9314 35 YMOF.1583.7313 11 OMYF.1755.7637 188 OMOF.2873.8988 172 4.664.31 3.13.77.994.319 Clutch size YMYF.655.942 261 YMOF.311.9393 13 OMYF.651.929 181 OMOF.769 1.537 174.552.458 1.465.226.546.46 Number hatched YMYF.215.9218 261 YMOF.873.9894 13 OMYF.586.8613 181 OMOF.529 1.537 174.248.619 2.348.126..985 Number fledged YMYF.1391.8835 261 YMOF.111.8314 13 OMYF.1849.7864 181 OMOF.1682.8354 174.67.413.13.719.9.923 a Data are z-scores. In order to take out year effects, the population mean for the year is subtracted from each single value. The result is then divided by the standard deviation for the year. Negative z-scores represent values smaller than the population mean. For example, the clutch size of pairs of young male and young female (YMYF) is.728 sd units smaller than the population mean while the clutch size of old male and old female pairs (OMOF) is.2412 sd units larger than the population mean. YMYF = 1st year male and 1st year female; YMOF = 1st year male and at least 2nd year female; OMYF = at least 2nd year male and 1st year female; OMOF = at least 2nd year male and at least 2nd year female. Age and breeding performance To separate the effect of mate retention on breeding performance from a possible effect of age we analysed breeding performance in relation to male and female age. Because pairs breeding together for a second year can only be found in the class of older males and older females, we excluded these faithful pairs from the analysis of age effects. The age class of mates did not significantly affect mean clutch size, number of hatchlings or number of fledglings. In s female age affected clutch size: older females layed significantly larger clutches; male age affected the number of fledged offspring: pairs with an older male fledge significantly more offspring than pairs with a first-year male (Table 2). pairs with at least second-year females tended to have a lower hatching rate than pairs with a first-year female, although this was not statistically significant (Table 3). In s there is a significant effect of age class of mates on fledging rate, which is highest in pairs consisting of an old male and old female, and lowest in pairs of first-year mates (Table 3). In both Blue and Great Tits nestling weight is not affected by age class of mates (Table 4). In s female age affects laying date: second-year females layed significantly earlier than females in their first year. In s the combined age of the male and female affected fledging date.

Pair bond and breeding success in Blue and s 97 Table 3. Kruskal Wallis test on difference in hatching and fledging rate between age classes of pairs. Kruskal-Wallis (df = 3) Variable Age class n Mean a sd a χ 2 P Hatching rate YFYM 31.17 1.2 YFOM 19.893.23 OFYM 11.931.16 OFOM 175.9 1.1 2.653.448 Fledging rate YFYM 32.59.93 YFOM 188.225.81 OFYM 11.282.49 OFOM 173.259.72 11.195.11 Hatching rate YFYM 265.83.9 YFOM 183.141.85 OFYM 13.69 1.7 OFOM 175.147 1.4 7.717.52 Fledging rate YFYM 255.175.83 YFOM 179.16.79 OFYM 128.129.83 OFOM 171.238.58 1.825.69 a Data are z-scores (see footnote to Table 2). YMYF = 1st year male and 1st year female; YMOF = 1st year male and at least 2nd year female; OMYF = at least 2nd year male and 1st year female; OMOF = at least 2nd year male and at least 2nd year female. Offspring of older male and older female fledge significantly earlier than offspring of pairs with at least one first-year mate (Table 5). Age and pair bond pairs with a first-year male were more likely to divorce and pairs with an older male were less likely to divorce than would be expected by chance (Table 6). Pair bond and breeding performance Table 7 gives mean values for breeding success from unstandardized data and shows the results of a t-test using standardized values (z-scores), comparing the mean breeding success of newly formed pairs with that of pairs after mate retention. Neither clutch size nor number of hatched and fledged offspring differed significantly between newly formed pairs and pairs breeding together for a second season. There is no difference in hatching and fledging rate between newly formed pairs and pairs breeding together for a second season (Table 8). The offspring of mates that stayed together did not differ significantly in weight from those of newly formed pairs, although nestlings of faithful pairs were slightly heavier for both Great Tits and s (Table 9). Pair bond and timing of breeding pairs bred earlier after mate retention than when with a new partner, although, for the end of nest building and for laying date, this difference was not significant (Table 1). In s each part of the breeding cycle began about 4 days earlier in pairs breeding together for a second season than for those in newly formed pairs. In s no significant difference in timing of breeding was found between newly formed pairs and pairs that had stayed together. The duration of incubation and of the nestling period did not differ between the two groups of pairs. In s breeding retaining a mate seems to save time. However, birds that did this might have been early breeders for other reasons, e.g. because they were in better condition. We tested this by using a Wilcoxon test to compare the timing of breeding in the first and in the second year of pairs breeding together for two seasons. To eliminate age effects on the timing of breeding the analysis was based only on pairs of older males and older females. sample sizes were too small for testing.

98 M. Pampus, K.-H. Schmidt & W. Wiltschko Table 4. MANOVA on nestling and fledgling weight with male age, female age and interaction of male and female age as explanatory variables. Male age Female age Male age* Female age Dependent variable Age class Mean a sd a n F P F P F P Mean nestling weight on 1th day Weight of heaviest nestling on 1th day Mean nestling weight on 15th day Weight of heaviest nestling on 15th day Mean nestling weight on 1th day Weight of heaviest nestling on 1th day Mean nestling weight on 15th day Weight of heaviest nestling on 15th day YMYF.118 1. 8 YMOF.7.83 27 OMYF.13.88 48 OMOF.72.84 58 1.353.246.62.84.464.496 YMYF.123.96 8 YMOF.45.76 27 OMYF.168.95 48 OMOF.77.85 58 2.349.127.2.962.397.529 YMYF.17 1.7 8 YMOF.131.95 27 OMYF.195.84 48 OMOF.112.82 58 1.548.215.69.436 1.913.168 YMYF.153 1.6 8 YMOF.78.9 27 OMYF.17.89 48 OMOF.112.84 58 1.622.24.38.538 1.61.34 YMYF.175.79 31 YMOF.2 1.9 9 OMYF.7.97 26 OMOF.377.79 17 1.466.23 1.775.187.191.663 YMYF.223.86 31 YMOF.72 1.5 9 OMYF..98 26 OMOF.182.9 17.536.466 1.15 1.15.63.83 YMYF..84 31 YMOF.32 1.17 9 OMYF.8.87 26 OMOF.274.93 17.259.612.528.47.756.387 YMYF.8.91 31 YMOF.179 1.7 9 OMYF.64.7 26 OMOF.278.9 17.885.35.162.688 1.448.232 a Data are z-scores (see footnote to Table 2). YMYF = 1st year male and 1st year female; YMOF = 1st year male and at least 2nd year female; OMYF = at least 2nd year male and 1st year female; OMOF = at least 2nd year male and at least 2nd year female. In pairs breeding together for two consecutive seasons the start of incubation and the hatching and fledging dates lay significantly earlier in the second than in the first year, indicating that the earlier onset of breeding is a true effect of mate retention (Table 11). Breeding performance as a determinant of pair bond duration The breeding success in one year might influence the decision of mates as to whether to breed together again or divorce and find a better partner for the next season. We tested whether the breeding performance of pairs that were going to split up differed from that of pairs that would stay together. For this analysis we used the unstandardized data, and assumed that the total number of offspring would be the factor on which a decision was based rather than the number of offspring relative to the mean of the population (z-scores). Because in s both the frequency of divorce and the breeding performance were affected by the age of one or both of the partners, we analysed two groups of pairs separately

Pair bond and breeding success in Blue and s 99 Table 5. MANOVA on timing of breeding with male age, female age and interaction of male and female age as explanatory variables. Male age Female age Male age* Female age Variable Age class Mean a sd a n F P F P F P End of nest building YMYF.3.91 122..992 1.633.22.565.453 YMOF.69.75 42 OMYF.77.92 8 OMOF.137.85 77 Laying date of 1st egg YMYF.43.96 122.65.8 8.836.3.6.937 YMOF.321.61 42 OMYF.11.78 8 OMOF.34.62 77 Start of incubation YMYF.72.78 122.246.62 1.573.211.19.891 YMOF.175.7 42 OMYF.14.77 8 OMOF.234.66 77 Hatching date YMYF.6.77 122.441.57 2.141.144.36.849 YMOF.176.68 42 OMYF.12.75 8 OMOF.252.61 77 Fledging date YMYF.37.83 122.347.556 2.66.17.223.637 YMOF.163.76 42 OMYF.62.85 8 OMOF.18.67 77 Incubation period YMYF.35.93 122.598.44.43.837.123.727 YMOF.3 1.3 42 OMYF.97.99 8 OMOF.8 1.4 77 Nestling period YMYF.171.96 122.776.379.143.76.9.924 YMOF.134 1.17 42 OMYF.49.93 8 OMOF.15.93 77 End of nest building YMYF.141.91 49.25.875.137.712 1.332.25 YMOF.1.97 2 OMYF.68 1.5 48 OMOF.173.88 28 Laying date of 1st egg YMYF.5.82 49.994.321 1.16.315.852.358 YMOF.36 1.29 2 OMYF.38.92 48 OMOF.278.77 28 Start of incubation YMYF.216.68 49.35.582.692.47 3.26.84 YMOF.66.99 2 OMYF.42.81 48 OMOF.4.8 28 Hatching date YMYF.218.64 49.556.457.5.946 3.39.83 YMOF.74 1.5 2 OMYF.61.86 48 OMOF.34.78 28 Fledging date YMYF.12.76 49 2.327.129.4.841 6.355.13 YMOF.262 1.2 2 OMYF.58.93 48 OMOF.357.65 28 Incubation period YMYF.82.75 49.48.49 1.93.297.268.65 YMOF.162.99 2 OMYF.12.87 48 OMOF.136.88 28 Nestling period YMYF.17.8 49.25.875 2.734.1.38.58 YMOF.151.76 2 OMYF.66.78 48 OMOF.86.93 28 a Data are z-scores (see footnote to Table 2). YMYF = 1st year male and 1st year female; YMOF = 1st year male and at least 2nd year female; OMYF = at least 2nd year male and 1st year female; OMOF = at least 2nd year male and at least 2nd year female.

1 M. Pampus, K.-H. Schmidt & W. Wiltschko Table 6. 2 2 χ 2 test on the effect of age class of mates on the probability of mate retention and splitting up of pairs. YM OM Total YF OF Total YM OM Total YF OF Total MR Observed 36 45 81 45 36 81 24 34 58 35 23 58 Expected 4.9 4.1 81 46.4 34.6 81 25.6 32.4 58 32.4 25.6 58 SP Observed 16 6 22 14 8 22 14 14 28 13 15 28 Expected 11.1 1.9 22 12.6 9.4 22 12.4 15.6 28 15.6 12.4 28 Total 52 51 13 59 44 13 38 48 86 48 38 86 χ 2 5.536.462.569 1.483 P.19.497.451.223 YM = 1st year male; OM = at least 2nd year male; YF = 1st year female; OF = at least 2nd year female; MR, mate retention; SP, splitting up. Table 7. t-test on differences in breeding success between newly formed pairs and pairs breeding together for a second season. Only pairs of at least second year male and female (OMOF) are included. Variable Mating status n Mean a sd a t P Clutch size NP 174.236 (9.99).934 (1.6) MR 68.323 (1.15).956 (1.62).642.521 Number hatched NP 173.143 (8.9) 1.26 (1.97) MR 68.299 (9.32).956 (1.7) 1.84.279 Number fledged NP 174.279 (8.16).95 (2.47) MR 68.45 (8.68).696 (1.88).481.631 Clutch size NP 174.77 (1.84) 1.54 (1.87) MR 56.187 (11.5).833 (1.62).84.423 Number hatched NP 174.53 (9.21) 1.54 (2.4) MR 56.71 (9.7).96 (2.7).789.431 Number fledged NP 174.168 (8.68).835 (2.68) MR 56.251 (8.95).837 (3.4).643.521 a Data are z-scores (see footnote to Table 2). Unstandardized values are given in parentheses. NP, newly formed pair; MR, mate retention. according to age class in this species: pairs with a first-year male (YMYF and YMOF) and pairs with a second-year (or older) male (OMYF and OMOF). In s nestling weight could not be analysed because the sample sizes were too small. Breeding performance did not differ between divorcing and faithful pairs in either groups or s (Table 12). Mate retention and divorce might be two sides of a strategy to improve breeding success. To test this we investigated whether individuals had improved their breeding success from one year to the next through either splitting up or retaining the old mate. We carried out a Wilcoxon test using standardized data (z-scores). Only pairs where both mates were older than 1 year in the first year were included in the analysis. Pairs that stayed together did not improve their breeding performance between the first and second year of breeding, and neither males nor females improved their breeding performance through splitting up (Table 13). females even tended to have fewer hatched offspring and had significantly fewer fledglings in the year after splitting up than in the year before. Although the number of pairs consisting of older males and older females that divorced was too small to allow testing (n = 3), the data for females point in the same direction. In all three cases the clutch size and number hatched were smaller in the second year than in the first. Hatching rate, fledging rate and number fledged were smaller in the second year in two of the three cases. In males no trend was detectable

Pair bond and breeding success in Blue and s 11 Table 8. Mann Whitney test on differences in hatching rate and fledging rate between newly formed pairs and pairs breeding together for a second season. Only pairs of at least 2nd year male and female (OMOF) are included. Variable Mating status n Mean a sd a Z Mann Whitney P Hatching rate NP 174.96 (.89) 1.1 (.13) MR 68.78 (.92).92 (.9) 1.25.211 Fledging rate NP 174.242 (.92).75 (.19) MR 68.272 (.93).92 (.93) 1.125.261 Hatching rate NP 174.147 (.85) 1.4 (.18) MR 56.15 (.88) 1. (.13).61.951 Fledging rate NP 171.238 (.94).58 (.15) MR 55.28 (.92).69 (.23).917.359 a Data are z-scores (see footnote to Table 2). Unstandardized values are given in parentheses. NP, newly formed pair; MR, mate retention. Table 9. t-test on differences in nestling and fledgling weight between newly formed pairs and pairs breeding together for a second season. Only pairs of at least 2nd year male and female (OMOF) are included. Variable Mating status n Mean a sd a t P Mean nestling NP 61.38 (13.64).887 (1.28) weight (1th day) MR 25.184 (13.89).949 (1.38).666.57 Weight of heaviest NP 61.43 (15.).918 (1.37) nestling (1th day) MR 25.251 (15.35).885 (1.25).959.34 Mean nestling NP 67.87 (16.63).793 (1.42) weight (15th day) MR 28.78 (16.71) 1.16 (1.29).46.963 Weight of heaviest NP 67.64 (17.88).799 (1.32) nestling (15th day) MR 28.147 (18.9).929 (1.3).444.658 Mean nestling NP 18.29 (9.12).85 (.96) weight (1th day) MR 7.664 (9.6).814 (1.48).998.329 Weight of heaviest NP 18.11 (9.9).932 (1.14) nestling (1th day) MR 7.564 (1.43).782 (1.47) 1.159.258 Mean nestling NP 25.124 (11.18).946 (.78) weight (15th day) MR 8.344 (11.32) 1.32 (.65).56.58 Weight of heaviest NP 25.185 (11.97) 1.37 (.79) nestling (15th day) MR 8.739 (12.38).799 (.81) 1.379.178 a Data are z-scores (see footnote to Table 2). Unstandardized values are given in parentheses (weight in g). NP, newly formed pair; MR, mate retention. (n = 3). In males breeding success did not differ between attempts made before and those after a divorce. Pair bond and breeding dispersal females dispersed further between consecutive years when re-mating than if retaining their mate (Table 14). males kept close to their former nesting hole without any significant difference between groups. The difference in breeding dispersal distances between sexes is highly significant in s (Mann Whitney U-test, n = 241, P <.1). In s both males and females dispersed further after either losing a mate or divorcing than when retaining the former mate (Mann Whitney

12 M. Pampus, K.-H. Schmidt & W. Wiltschko Table 1. t-test on differences in timing of breeding between newly formed pairs and pairs breeding together for a second season. Only pairs of at least 2nd year male and female (OMOF) are included. Variable Mating status n Mean a sd a t P n Mean a sd a t P End of nest NP 84.133 (18.29).88 (8.13) 33.83 (17.58).852 (6.41) building period MR 36.445 (14.64).823 (8.63) 1.814.72 12.14 (14.64).678 (6.93).814.42 (year day b ) Laying date NP 87.235 (114.8).688 (7.42) 31.243 (112.45).838 (6.) (year day b ) MR 34.56 (11.79).83 (8.19) 1.856.66 11.315 (11.79).559 (6.21).266.792 Start of NP 11.19 (125.3).8 (7.7) 38.241 (124.63).923 (5.79) incubation (year day b ) MR 4.662 (12.7).868 (6.72) 3.82.2 13.45 (12.7).75 (6.34).737.465 Hatching date NP 11.27 (137.53).79 (7.64) 38.217 (138.5).893 (5.67) (year day b ) MR 4.662 (133.23).877 (6.55) 2.991.3 13.291 (133.23).81 (6.1).265.792 Fledging date NP 11.154 (156.92).824 (8.33) 38.227 (157.79).87 (6.2) (year day b ) MR 4.621 (152.5).849 (6.67) 3.12.3 13.35 (152.5).837 (6.3).446.658 Incubation NP 11.64 (12.5) 1.94 (1.63) 38.6 (13.42).916 (1.46) period (days) MR 4.21 (12.53).785 (1.9).227.821 13.281 (12.53).96 (1.17).978.333 Nestling NP 11.19 (18.1).962 (1.68) 38.1 (18.66).958 (1.21) period (days) MR 4.41 (18.23).657 (.77).361.719 13.136 (18.23) 1.111 (1.29).42.676 a Data are z-scores (see footnote to Table 2). Unstandardized values are given in parentheses. b 1 = 1 April. NP, newly formed pair; MR, mate retention. Table 11. Timing of breeding in the first and the second year of breeding together. Wilcoxon signed rank test Difference between years n Mean rank Sum of ranks Z P End of nest building period 2nd year earlier than 1st year 16 13.13 21. 2nd year later than 1st year 9 12.78 115. 2nd year same as 1st year 1.278.21 Laying date of 1st egg 2nd year earlier than 1st year 13 13. 169. 2nd year later than 1st year 9 9.33 84. 2nd year same as 1st year 1.38.168 Start of incubation period 2nd year earlier than 1st year 2 18.45 369. 2nd year later than 1st year 1 9.6 96. 2nd year same as 1st year 2.88.5 Hatching date 2nd year earlier than 1st year 23 16.13 371. 2nd year later than 1st year 7 13.43 94. 2nd year same as 1st year 2.849.4 Fledging date 2nd year earlier than 1st year 23 15.26 351. 2nd year later than 1st year 6 14. 84. 2nd year same as 1st year 2.887.4 Duration of incubation period 2nd year earlier than 1st year 9 17.22 155. 2nd year later than 1st year 21 14.76 31. 2nd year same as 1st year 1.594.111 Duration of nestling period 2nd year earlier than 1st year 18 15.67 282. 2nd year later than 1st year 12 15.25 183. 2nd year same as 1st year 1.18.39 Data are z-scores (see footnote to Table 2). Only pairs where both mates were at least in their 2nd year of life (OMOF) when first breeding together are included. NP, newly formed pair; MR, mate retention.

Pair bond and breeding success in Blue and s 13 Table 12. Mann Whitney test on differences in breeding performance before mate retention and before splitting up. Species/ age class Variable Mating status n Mean a sd a Mann Whitney Z P YMYF and YMOF OMYF and OMOF Clutch size MR 35 9.97 1.58 SP 16 9.31 1.62 1.473.141 Number hatched MR 35 8.77 2.43 SP 16 8.69 1.54.843.399 Number fledged MR 35 8.46 2.1 SP 16 7.31 2.98 1.46.16 Hatching rate MR 35.87.18 SP 16.94.6.84.41 Fledging rate MR 34.95.1 SP 16.83.28 1.232.218 Mean nestling weight (1th day) MR 7 14.6 1.69 SP 5 14.16 1..244.88 Weight of heaviest nestling (1th day) MR 7 15.61 1.42 SP 5 15.36.96.244.87 Mean nestling weight (15th day) MR 11 16.39 1.7 SP 8 16.38 1.34.33.741 Weight of heaviest nestling (15th day) MR 11 17.99.94 SP 8 17.43 1.37.83.46 Clutch size MR 44 9.91 1.41 SP 6 1.67 1.37.995.32 Number hatched MR 44 8.82 1.92 SP 6 9.83 1.17 1.352.176 Number fledged MR 44 7.98 2.55 SP 6 8.17 1.47.167.867 Hatching rate MR 44.88.13 SP 6.92.6.535.593 Fledging rate MR 44.9.21 SP 6.84.18 1.316.188 Mean nestling weight (1th day) MR 15 13.54 1.29 SP 2 13.39 1.16.298.766 Weight of heaviest nestling (1th day) MR 15 14.99 1.9 SP 2 14.75 1.77.75.94 Mean nestling weight (15th day) MR 17 16.79 1.6 SP 5 16.55.84.823.411 Weight of heaviest nestling (15th day) MR 17 18.12 1.39 SP 5 18.44.53.235.814 Clutch size MR 59 1.85 1.76 SP 31 1.71 1.51.636.525 Number hatched MR 59 9.46 2.15 SP 31 9.77 1.5.32.762 Number fledged MR 59 8.86 2.24 SP 31 9.19 2.6 1.67.286 Hatching rate MR 59.87.15 SP 31.92.1.95.366 Fledging rate MR 59.94.12 SP 31.93.22.676.499 a Unstandardized values. SP, splitting up; MR, mate retention, YMYF = 1st year male and 1st year female; YMOF = 1st year male and at least 2nd year female; OMYF = at least 2nd year male and 1st year female; OMOF = at least 2nd year male and at least 2nd year female.

14 M. Pampus, K.-H. Schmidt & W. Wiltschko Table 13. Wilcoxon signed rank test on differences in breeding performance before and after mate retention/splitting up. Species/ Sex Mating status Variable Difference between years n Mean rank Sum of ranks Wilcoxon signed rank test Z P MR Clutch size Number hatched Number fledged Hatching rate Fledging rate MR Clutch size Number hatched Number fledged Hatching rate Fledging rate SP Clutch size Female Number hatched Number fledged Hatching rate Fledging rate SP Clutch size Male Number hatched Number fledged Hatching rate Fledging rate 11 15 6 9 17 6 14 12 6 14 12 6 15 11 6 11 7 3 7 11 3 8 1 3 6 12 3 9 9 3 7 3 8 2 9 1 6 4 5 5 3 7 5 5 5 5 7 3 4 6 11. 15.33 12.78 13.88 12.43 14.75 9.57 18.8 14.93 11.55 9.86 8.93 1.64 8.77 9.56 9.45 12.25 8.13 1.39 8.61 4.57 7.67 5.63 5. 5.33 7. 6.5 4. 5.6 5.4 6. 5.29 4.6 6.4 5.8 5.2 4.71 7.33 7.5 4.17 121. 23. 115. 236. 174. 177. 134. 217. 224. 127. 18.5 62.5 74.5 96.5 76.5 94.5 73.5 97.5 93.5 77.5 32. 23. 45. 1. 48. 7. 39. 16. 28. 27. 18. 37. 23. 32. 29. 26. 33. 22. 3. 25. 1.384.166 1.537.124.38.97 1.54.292 1.232.218 1.2.316.479.632.392.695.523.61.348.727.459.646 1.784.74 2.9.37 1.172.241.51.959.968.333.459.646.153.878.561.575.255.799 Data are z-scores (see footnote of Table 2). Only pairs where both mates were at least in their 2nd year of life (OMOF) when first breeding together/before splitting up are included.

Pair bond and breeding success in Blue and s 15 Table 14. Kruskal Wallis one-way ANOVA on differences in median distances between nestboxes of consecutive years between different status groups of pair bond. Median distance between nestboxes (m) Species Sex MR n LM n SP n χ 2 P Female 42.2 39 97.6 75 117.4 26 24.73 <.1 Male 42.2 39 51.7 116 47.1 24 2.8 ns Female 28.6 3 6.3 64 6.6 8 13.23 <.1 Male 28.6 3 6.4 46 67.2 6 1.93 <.5 MR = mate retention, breeding with the same mate in consecutive years, LM = loss of mate, re-mating between consecutive years due to the death of the mate, SP = splitting up, re-mating between consecutive years while the former mate is still alive. U-test, n = 124, P >.5). In both species the median dispersal distances did not differ after losing a mate or divorcing (Kruskal Wallis one-way ANOVA, : n = 241, P <.5; : n = 124, P <.5). DISCUSSION Mate retention proved to be the regular mating strategy in s and s. In the population studied, 75.8% of pairs and 65.2% of Blue Tit pairs bred together for a second season when both mates had survived, while the other 24.2% and 34.8%, respectively, split up to breed with a different partner. These divorce rates are considered to be neither very high nor very low for these species: divorce rates in s of 49% (Dhondt et al. 1996) and in s of 8 85% (Dhondt & Adriaensen 1994) have been reported. Both mate retention and divorce might be strategies to increase reproductive success. In the populations studied, however, no direct improvement in breeding success through either mate retention or divorce could be detected. and pairs breeding together for a second season did not have greater breeding success (in terms of clutch size, number of hatchlings and fledglings, fledging rate, weight of hatchlings and fledglings) than did newly formed pairs. We found no improvement in breeding performance between the first and second years of breeding together. Contrary to the findings of Dhondt and Adriaensen (1994), divorce even seemed to decrease the breeding success of females, which had significantly fewer fledglings, relative to the population mean, in the year after divorce than they had before. The breeding success of pairs that later divorced did not differ from that of pairs that had stayed together. This is contrary to the findings of Perrins and McCleery (1985). Thus, in the populations studied, neither s nor Blue Tits divorced as a reaction to poor breeding success, nor did they stay together because of good breeding success. Mate retention did have an effect on the timing of breeding, however. In pairs breeding together for a second season, nest completion and the start of laying tended to be earlier, and the start of incubation, hatching and fledging were significantly earlier than in pairs breeding together for the first time. Incubation and nestling periods were not shorter in pairs breeding together for a second season than in newly formed pairs. Thus s did not shorten their breeding cycle through mate retention but just started, and therefore finished, breeding earlier than other pairs. Tits might improve their lifetime reproductive success through early breeding because offspring fledged early in the year have more time to gain weight and foraging experience before winter, thus improving their probability of surviving into the next season (Magrath 1991, Dhondt et al. 1996, Naef-Daenzer et al. 21). Furthermore, the adults have more time to recover from the breeding season, and to moult. However, in the population studied, the advancement in timing of breeding in faithful pairs only amounts to 4 days on average and thus would not seem to create a major advantage for birds retaining their former mate. Overall, the influence of mate retention on reproductive performance is very small. In this population neither mate retention nor divorce function as strategies to increase reproductive success. This could mean that mate retention is advantageous in a different context. If mate retention does not increase breeding success it might increase the probability of survival in the non-breeding season and thus may affect lifetime reproductive success.

16 M. Pampus, K.-H. Schmidt & W. Wiltschko Mates that stay together during the winter might increase their probability of survival by achieving a higher social rank in the winter flock. It has been shown that the social status of s (Lambrechts & Dhondt 1986) and Willow Tits P. montanus (Ekman 199, Hogstad 1992) is correlated positively with overwinter survival. A female can increase her social rank by mating with a dominant male who keeps competitors away while she is feeding (Ekman 199). Whether s are paired in winter is not clear (Gosler 1993). Yet there is evidence of a pair bond in s between two years of breeding together. We found that in winter, pairs of colour-marked individuals visited artificial food supplies more frequently with their mate of the last and the following year than with any other conspecific breeding in the same area. Feeding patterns of colour-marked individuals suggest a stable pair bond in winter (data not shown). Males in their second year of life or older usually dominate over first-year males, which are in the majority in the population (Saitou 1979). Between the older males there is a dominance hierarchy that depends on body size and other factors (Koivula et al. 1993). It might therefore be advantageous for a female to leave her former mate to pair with an older and/or higher-ranking male (Smith 1992, Dhondt & Adriaensen 1994, Otter & Ratcliffe 1996). A female might also divorce her mate in order to move to a better territory (Choudhury 1995, Ramsey et al. 2). Orell et al. (1994) found that in the Willow Tit the probability of divorce depended on the age of the female. Ramsey et al. (2) found the same in the Black-capped Chickadee P. atricapillus as did Dhondt and Adriaensen (1994) in s. In these three studies, divorce occurred more often in pairs with a first-year female than in pairs with an older female. The Willow Tit females divorced in favour of an older, usually high-ranking male. The Black-capped Chickadee females divorced in favour of a higher-ranking male than the previous mate. We found an influence of male, instead of female, age on the frequency of divorce in s. Pairs with first-year males divorced more often, and pairs with older males less often, than expected by chance. Female age had no influence on the frequency of divorce. If the observed divorces were initiated by the females, an issue which we cannot prove, female s in our population seem to divorce young males in favour of older ones, as do Willow Tits. s in our study do not benefit from divorce in terms of reproductive performance, but they might benefit in terms of lower winter mortality. In s we found no age effect on the frequency of divorce. Mate retention and divorce may also occur for non-adaptive reasons. However, it seems unlikely that a former pair should re-mate by chance considering the high breeding densities in the study sites and the short distances between nesting sites. Furthermore, we consider the pair bond to last during winter as stated above, at least in s. The greatest median dispersal distance after divorce was 117.4 m, which was found in female s; Great Tits only moved about 65 m after splitting up. These short distances imply that divorce did not occur by actively shifting to a better territory. An accidental divorce resulting from the mates occupying different winter home-ranges does not seem likely, either. Saitou (22) showed that divorce in the occurred more often in pairs where one or both mates were non-resident in winter. In our study sites, however, only very few if any of the s leave the area in winter (data not shown). Although divorce might not be beneficial for either of the former mates, they might be forced into it by a third, unpaired but socially dominant individual, which in this way achieves an opportunity to breed (Choudhury 1995). The effect of male age on the frequency of divorce that we found in the Great Tits could be related to the occurrence of forced divorce. Older males, being dominant over first-year males (Saitou 1979), might be able to chase away a first-year opponent more easily than an older one so that divorce occurs more often in pairs with a firstyear male. Why does mate retention hold no detectable advantage? In most populations of s and s studied mate retention is the main strategy, and only a few individuals are involved in a divorce, although in some populations divorce rates can be higher than mate retention rates (Dhondt et al. 1996). To evolve as the main mating strategy mate retention therefore must have been beneficial in most populations for most individuals. The fact that no major advantage of mate retention has been found in the populations so far studied leads to the conclusion that mate retention must have evolved under different conditions than those we find in our study sites today. Undoubtedly, in most parts of the ranges of the and, habitats have

Pair bond and breeding success in Blue and s 17 changed dramatically through human impact. In Europe breeding densities are often so high that some pairs do not defend a territory but breed inside the territory of another pair (Gosler 1993). This might be related to forest fragmentation, creating forest edges and light open stands of deciduous trees, which make good habitats for s and s (Harrap & Quinn 1996), but it might also have resulted through the widespread provision of artificial nesting holes. Most studies of breeding biology in the and are based on study sites with a superabundance of nestboxes, so that competition for breeding sites is much reduced or eliminated. The small breeding dispersal distances we found obviously reflect this situation of high breeding densities and high densities of artificial breeding sites. In s, and in females, the median breeding dispersal distances after divorce and after losing a mate were greater than the dispersal distances after mate retention. Thus mate retention seems to enable s to keep their former breeding sites. When breeding with a new partner, both male and female s move further away from the former breeding site, with no difference between loss of mate and divorce. males stayed close to their former nesting sites, independent of pairing status, confirming the findings of Blondel et al. (2). This does not necessarily mean that females are the active partners in a divorce in Blue Tits (Cézilly et al. 2); in cases in which a male mates with a new female this female might chase away the former one. The superabundance of nestboxes enables most birds to stay and breed again within their former feeding grounds, independent of whether or not they retained their mate. The median breeding dispersal distance of both s and s was less than 125 m, which is regarded as the approximate diameter of a territory (Lindén 1991). Under natural conditions the situation might be very different. Assuming that breeding sites are limited, a considerable number of birds might be excluded from breeding each year, and faithful pairs might have a better chance of breeding again in the same nesting hole. There is evidence that prior residency increases the ability to defend a territory or nesting site (Koivula et al. 1993). It is not clear though how much the female contributes to defending the nest-site. Hinde (1952) observed that female s sometimes take part in aggression against another pair. Perhaps the male is supported or encouraged by the female, making it easier for him to defend the former nesting site. Provided that mates keep a pair bond in winter, or re-establish it before new pairs are formed, they can stay close to the nesting site and defend it as soon as the breeding season begins. No time is lost in searching for a suitable nesting hole or in forming a new pair bond. Our data show that newly formed pairs start breeding about 4 days later than faithful pairs. This small delay through re-mating did not influence the number and quality of fledged offspring. With a limited number of nesting holes, searching for a breeding site (or a new mate, see below) might delay breeding further and thus lead to a reduced breeding success. Birds that (have to) remate may even be excluded from breeding in the following years if they are not able to keep their former nesting hole. This would have an enormous effect on lifetime reproductive success. Although long-term pair bonds seldom exceed two breeding seasons owing to the high mortality rates of s and s, birds breeding together for 2 years could double their lifetime reproductive success compared with birds losing their partners or divorcing after their first breeding season without getting the chance to breed again. On the other hand, the number of suitable breeding sites might not be a limiting factor in populations at very low densities (T. Weso4owski pers. comm.), especially in unmanaged forests like the last European primary forest in Bia4owie a National Park, Poland (Tomia4ojc et al. 1984). In such populations, mate retention might be beneficial because the cost of finding a new mate is high, especially if the pairs remain on the territory in winter and do not form flocks (Dhondt & Adriaensen 1994). Thus the original advantage of mate retention might lie in retaining a well-known breeding territory and a suitable nesting hole and in reducing the cost of finding a new partner. This advantage does not exist in populations of very high densities and under the artificial condition of a superabundance of nesting holes of equal quality. We thank Christopher Perrins and two anonymous referees for valuable comments on the manuscript, and we thank numerous students who helped collect the data. REFERENCES Blondel, J., Perret, P. & Galan, M.-J. 2. High divorce rates in Corsican s: how to choose a better option in a harsh environment. Oikos 89: 451 46.