Plenary05: Experiments on competition between Great and Blue Tit: Effects on Blue Tit reproductive success and population processes

Dhondt, A.A. & Adriaensen, F. 1999. Experiments on competition between Great and Blue Tit: Effects on Blue Tit reproductive success and population processes. In: Adams, N.J. & Slotow, R.H. (eds) Proc. 22 Int. Ornithol. Congr., Durban. Ostrich 70 (1): 39–48.

Great Tits Parus major and Blue Tits Parus caeruleus compete for food and for cavities. By providing nest boxes with entrance holes small enough to exclude Great Tits, but allowing access to Blue Tits, either by themselves or in combination with standard nest boxes we manipulated the breeding densities of Blue and Great Tits independently. We compared populations differing in experimental densities over five-year periods. Components of Blue Tit reproductive success are adversely affected when either Blue Tit or Great Tit densities are experimentally increased. At high Blue Tit density adult survival does not change. When small-holed nest boxes are provided and Blue Tit density is increased, the number of local born males that recruit into the population increases considerably, so that the proportion of local born males is much higher than at low density. Similarly the local recruitment rate (the number of local recruits per adult breeding in the previous year) increases strongly for males, but not for females. The presence of small-holed nest boxes increases habitat quality, especially in winter when Blue Tits use these boxes for roosting, so that juveniles disperse less, and breeding density increases

Keddy (1989, p. 2) defines competition as ‘the negative effects which one organism has upon another by consuming, or controlling access to, a resource that is limited in availability’. This definition is brief and precise and: (1) It reminds us of the fact that for competition to take place a resource must be limited in availability, i.e. there is not enough of it, or it takes too long to find it, or access to the resource is hindered by interactions with other individuals. (2) It tells us that the effect of competition primarily operates at the level of the individual.

The negative effects of competition operate at different levels of organisation. At the level of the individual competition can affect: (1) reproduction (delayed start of reproduction and/or reduced reproductive output in a single attempt, both resulting in a lower lifetime reproductive success); (2) survival (reduced chance to survive); (3) niche use or habitat selection (reduced access to high quality sites or habitat, resulting in a reduced reproductive success or a reduced chance to survive); (4) dispersal (increased likelihood that an individual will leave a site).

Individuals with different traits differ in ‘competitive ability’, and hence will be differently affected by competition.

The effects on individuals are translated into effects at the population level. When competition is more severe one or more population processes are adversely affected (reduced reproductive rate, reduced survival rate, decreased immigration rate, increased emigration rate). These effects on population processes further translate into lower density and/or relatively more young age classes (Hairston 1980), and result in a reduced per capita growth rate at a given density.

Populations subjected to different competition pressures should evolve to become different. Hairston’s (1980) very elegant field experiments using plethodonthid salamanders show that such may be the case. Differences between individuals can be in life-history traits, or in morphological/physiological/biochemical traits (Schoener 1983). In laboratory populations such evolutionary changes can be rapid. (Pimentel et al. 1965)

Finally, competition also has effects at the community level: Competition is assumed to influence community composition (number of species, and their relative abundance), and realised niche width (Wiens 1989).

Competition occurs over a limited resource. When the resource is less abundant or the number of competitors higher, the effect of competition should be more apparent and/or easier to detect. There are therefore two possible ways to study competition experimentally: (1) by manipulating the abundance of competitors and (2) by manipulating the limiting resource.

It tends to be relatively easy to reduce numbers, but more difficult to increase their abundance. In order to manipulate resource abundance one must assess which resource is in short supply, and be able to manipulate its abundance.

John Wiens (1989) reviewed the literature on field experiments in birds testing for the existence of interspecific competition. He concluded that manipulations of resources (6/12 providing evidence for the existence of interspecific competition) were apparently less successful to demonstrate competition than density alterations (16/23). It should be pointed out that in 5/12 experiments involving resource manipulations, food was added for some Parus species; and 11/23 experiments reporting density manipulations also involved Parids. An additional four studies used other hole-nesting species. Of the 35 studies reported by Wiens, 20 involved some cavity nesting species. In more recent studies cavity nesters have also played an important role. This suggests that parids are a good model to study competition experimentally because there is evidence for competition for food and for cavities, and because experimental manipulations are relatively easy.

A detailed review of the literature on field experiments concerning interspecific competition yielded some surprises.

(1) Very few studies exist in which population parameters have been compared in detail between populations subjected to different treatments over extended periods of time. Most studies only look at a limited number of population parameters, and most last a year or less.

(2) Most experimental manipulations involve a comparison between plots in which the numbers of an alleged competitor were reduced (experiment), or not changed (control). Very few, if any, experiments kept the numbers of one competitor constant, but increased the density of the other species by manipulating resources.

(3) Only 8/35 studies reviewed by Wiens (1989) were both replicated and had adequate controls.

(4) Nobody had asked whether or to what extent a change in interspecific competition in a field experiment could or would result in a rapid evolutionary response. In fact, Connell (1980, 1983) questioned whether any such co-evolutionary change between competitors would occur at all, because, unlike in interactions between trophic levels (i.e. predator–prey, parasite–host), the interaction between competitors is not continuous.

In west-European managed forests natural cavities are in short supply. When artificial nest boxes of appropriate dimensions are provided, the breeding densities of secondary cavity nesters, such as Great and Blue Tit can increase two- or threefold over the situation without these nest boxes (East & Perrins 1981; pers. obs.). The effects on Pied Flycatchers Ficedula hypoleuca can be even more dramatic (Lundberg & Alatalo 1992). Furthermore, adding nest boxes can also have indirect effects. Thus Red-cockaded Woodpeckers Picoides borealis were more successful when clusters of nest boxes were added, because they were less frequently excluded from their own cavities by Eastern Bluebirds Sialia sialis and Southern Flying Squirrels Glaucomis volans (Loeb & Hooper 1997). Adding nest boxes, thereby increasing the density of cavity nesters, can have negative effects on the abundance of open nesters through presumed interspecific competition (Hogstad 1975; Bock et al. 1992) or positive effects through presumed heterospecific attraction (Monkkonen et al. 1990).

When nest boxes suitable for both Great and Blue Tits were provided, Minot & Perrins (1986) found that the relative abundance of Blue Tits increased with the total density of the nest boxes until a certain threshold was reached. When nest boxes were superabundant Great Tit breeding density was limited by intra-specific competition, mainly for territories, and Blue Tit breeding density by interspecific competition, whereby there remained open space between Blue Tit territories (Dhondt et al. 1982). Blue Tits occupied the better quality areas only, better quality being defined as areas in which the birds laid a larger clutch (Dhondt et al. 1992)

Because Great Tits are larger than Blue Tits, one can regulate access to the next boxes by varying the size of the entrance hole. Large-holed boxes can be used by both species; small-holed nest boxes can be used by Blue Tits only. When given the choice in an aviary, Blue Tits preferred large-holed over small-holed nest boxes for roosting (Dhondt & Eyckerman 1980; Kempenaers & Dhondt 1991). When a Great Tit was added to the aviary, Blue Tits more frequently roosted in small-holed nest boxes (Kempenaers & Dhondt 1991). In forests, the number of Blue Tits found roosting in nest boxes during winter was very low when only large-holed boxes were present, and increased significantly when small-holed boxes were offered either by themselves, or together with large-holed nest boxes (Dhondt et al. 1991). In study plots with small-holed nest boxes (either by themselves or in combination with large-holed nest boxes) the breeding density of Blue Tits was 41–68% higher compared to plots in which only large-holed boxes were present (Dhondt et al. 1991). The fact that by manipulating nest box type during winter one can influence the subsequent breeding density of tits makes it possible to study the influences of both intra- and interspecific competition experimentally.

In all experiments we used high densities of wooden nest boxes with an entrance hole diameter of 32 mm (large-holed) and/or with an entrance hole diameter of 26 mm (small-holed).

Many studies have shown either through correlative analyses of long-term descriptive data sets, or through field experiments, that both intra- and interspecific competition during the breeding season can have an effect on reproductive traits of tits. Competition during the breeding season is mainly for food, since the diets of the two species overlap to a large extent. The intensity of competition, however, varied between studies. Many studies showed an effect of Blue Tit abundance on Great Tit reproduction, but the reverse was rarely studied (Dhondt 1989). We will here limit our discussion to intra- and interspecific effects on Blue Tits and refer to an earlier review for a summary of opposite effects (Dhondt 1989).

Very few studies have reported density-dependent effects on Blue Tit reproduction. Perrins (1990) found an inverse correlation between Blue Tit clutch-size and Blue Tit breeding density in two plots within Wytham Woods near Oxford, but not in a third. Dhondt et al. (1992) found such an effect in one of two plots. Van Balen & Potting (1990) found no density dependent effects on any trait in three study plots. Effects of Great Tit density on Blue Tit reproduction are even rarer. The only study reporting such an effect is that of Perrins (1990) who found an inverse relationship between Blue Tit clutch-size and Great Tit density in one of three plots near Oxford.

Taking advantage of the fact that it is possible to manipulate the density of Great and Blue Tits independently, we wanted to test experimentally effects of interspecific competition between Great and Blue Tits, and possible effects of intra-specific competition among Blue Tits in a comprehensive and long-term study. As concerns competition during the breeding season we replicated experiments in habitats of different quality, to determine to what extent competition would be more intense in poorer quality breeding conditions, where resources should be more limiting. Since we can ‘create’ Blue Tit populations at higher and lower densities, we also wanted to determine which of the four population processes would be mainly responsible for generating and maintaining the high densities.

The experiments were carried out in two sets of study plots in Northern Belgium about 60 km apart. More details on these plots can be found in Dhondt & Eyckerman (1980), Dhondt et al. (1984) and in Dhondt (1989b). The names in brackets are those used in the above references, if different from those used in this paper.

Ghent: all plots are isolated fragments.

(1) Plot MA (Maaltepark): 10 ha; mixed deciduous suburban park, with a few conifers. 79 large-holed nest boxes between 1963–1978; an additional 50 small-holed nest boxes 1979–1983.

(2) Plot SO (Soenen): 8 ha; mixed deciduous rural park. 89 large-holed nest boxes 1967–1977 and 1979–1983. 80 small-holed and 9 large-holed nest boxes in 1978.

(3) Plot GT (Gontrode): 18 ha; mixed deciduous, mainly oak; rural. 108 large-holed nest boxes 1972–1976; 98 small-holed and 10 small-holed 1977–1983.

(4) Plot HP (Hutsepot): 27 ha; mixed deciduous, mainly beech; rural. 184 large-holed nest boxes 1964–1983.

Antwerp: plots are within a large forested area:

(1) Plot T: 12.5 ha; mixed deciduous, mainly oak, in the ‘Peerdsbos’, a large wooded estate of ca. 150 ha; 80 (1979), then 120 (1980–1983) large-holed nest boxes; 120 small-holed nest boxes (1984–1988); 120 large-holed and 60 small-holed nest boxes 1989–1993.

(2) Plot B: 12.5 ha mixed deciduous, mainly oak, in the ‘Peerdsbos’ at 600 m from Plot T; 59 small-holed and 118 large-holed nest boxes 1979–1993 (in 1979 only 59 large-holed boxes).

(3) Plot L: 7.5 ha; mixed deciduous park, adjacent to Plot T. 50 large-holed nest boxes 1980–1988.

During the breeding season boxes were checked once a week and their contents noted, providing information on laying-date, clutch-size, number of hatched young, and number of fledglings. All fledglings were banded, and in Antwerp the young were also weighed and their tarsus measured at 15 days of age. Adults were trapped when feeding young and identified or banded, weighed and measured.

During winter the nest boxes were checked at least twice, and birds identified or banded.

All measurements of the same individual bird were compared, and if necessary, correction factors for systematic observer differences calculated and applied. Breeding season weights are used without corrections; tarsus length and wing length are used after correcting for observer.

The basic experimental approach used was as follows: We chose pairs of study plots and compared these during two periods of 5 years each. In the control plot the experimental setup did not change between the periods. In the experimental plot the nest box configuration was changed between the two periods, resulting in a different breeding density of one species. Table 1 summarises the method used to describe nest box configurations in this paper. A significant plot*period interaction term, in a PROC MIXED SAS procedure (SAS Institute 1997), with year randomised within period, was considered evidence of competition.

We compared the following reproductive traits:

(1) Julian laying date: Julian date when the first egg was laid in complete first clutches.

(2) Clutch-size: Mean number of eggs laid in complete first broods.

(3) Nesting success: Proportion of eggs producing fledglings in first broods from which at least one young fledged (arc sine transformed for analyses).

(4) Fledglings/pair: number of young fledged from first broods that produced at least one fledgling.

(5) Nestling body mass: Body mass of first brood nestlings at 15 days of age, excluding nests from polygynous males.

(6) Adult mass: Body mass of adult trapped when feeding nestlings aged 8–12 days.

If competition was important we should find that laying date was delayed and/or that clutch-size, nesting success, fledglings/pair, or mass was reduced in the situation with more intense competition.

The Blue Tit densities in all periods included in this paper are shown in Fig. 1.

In Table 2 we compare between-period density changes in the control and experimental plots. To remove any effect of a change in density caused by period effects rather than by experimental treatment effects we compared the density ratios of the experimental plot in the two 5-year periods (where the nest box configuration was changed) to that in the control plot (where it was not changed). The ratio of these ratios is the treatment effect.

For example in Plot MA the post-treatment density (small and large-holed nest boxes) was 1.86 pairs/ha; the pre-treatment density (large-holed nest boxes only) was 1.00 pairs/ha. The increase caused by adding small-holed nest boxes was not times 1.86, because in the control plot breeding density increased from 1.00 to 1.20. The increase caused by the treatment was times 1.55 (1.86:1.20).

Since we expected an increase in Blue Tit breeding density when we added small-holed nest boxes, we used a one-tailed significance level for the PERIOD*PLOT interaction from the PROC MIXED analysis with PLOT, PERIOD and PLOT*PERIOD as factors. In the Plot MA and Plot SO example P = 0.017.

In all cases in which small-holed boxes were added to a study plot the mean change over a 5-year period was higher than in a control plot where the nest box configuration remained unchanged. In all cases, except one, this difference was statistically significant (Table 2).

We will use the following terminology: Blue Tit density is ‘high’ in any plot in which small-holed nest boxes are present, and ‘low’ when this is not the case. This terminology is meant to be understood in relative terms, since in plot MA, for example, the mean ‘high’ Blue tit density is 1.86 pairs/ha, whereas in plot T the mean ‘low’ Blue Tit density is 1.76 pairs/ha. This simply reflects the fact that the habitat quality in plot T (as in all Antwerp plots) is much better than in most of the Ghent plots. Great Tit density is ‘high’ in plots in which the density of large-holed boxes was high, and ‘low’ (and mostly unknown) in plots in which only small-holed nest boxes were available.

In the experimental plot we compared a 5-year period with low Blue Tit densities (only large-holed boxes) and a 5-year period with high Blue Tit densities (both large and small-holed boxes). In the control area the situation remained unchanged. That comparison can be made at Ghent by comparing Blue Tit reproduction between plots MA (changed from large-holed to mixed) and SO (large-holed boxes in both periods), and at Antwerp by comparing plot T (large holed to mixed) and Plot B (mixed in both periods) (Table 3).

In both regions laying was relatively later in the experimental plot, when Blue Tit density was higher. Thus at Ghent in the control plot SO laying was earlier during the second 5-year Period than in the first 5-year Period, whereas it remained unchanged in the experimental plot MA. At Antwerp the period effect was highly significant, laying being much earlier during the second 5-year Period in both plots. The difference between the two periods, however, was 8 days in the control plot, compared to only 5.6 days in the experimental plot, resulting in a significant Plot*Period interaction.

At Ghent nesting success was significantly affected by intra-specific competition (Plot*Period P = 0.03), whereas at Antwerp this was not at all the case.

Body mass measures were available at Antwerp only. Fifteen-day nestling body mass was 2.6% lower when Blue Tit density was high in the experimental plot, as compared to when low Blue Tit density was low.

The adult body mass analysis yielded a significant three-way interaction term sex*plot*period (P = 0.0025), indicating that the period effect differed between plots and sexes. Hence we performed analyses separately for males and females. Male mass did not change with the intensity of competition, but female mass did decrease by about 3% in the high Blue Tit density period in the experimental plot.

In both Ghent and Antwerp we therefore find evidence for the existence of intra-specific competition. If the effect on laying date is the result of increased competition, this must be for food during the early part of the breeding season. Effects on nesting success and on body mass must be the result of increased competition for food during the nestling stage. A reduced fledgling mass could result in reduced survival after fledging (Perrins 1979).

In the experimental plot we compared a 5-year period with low Great Tit densities (only small-holed boxes) and a 5-year period with high Great Tit densities (both large and small-holed boxes). In the control area the situation needed to remain unchanged. That comparison could only be made in the Antwerp plots where plot T (the experimental plot) had small-holed nest boxes only in the breeding season 1984–1988, and where both box types were present during the 1989–1993 season. In Plot B the treatment throughout was 32+26.

The results (Table 4) are surprising because we find a hint that Blue Tit clutch-size may be adversely affected by Great Tit numbers (as Perrins & McCleery 1989 found), but especially in that we find significant effects on nestling body mass (one-tailed P = 0.006) and on female (one-tailed P-value = 0.017) but not on male body mass. The overall analysis of adult mass in which Plot, Period and Sex were included as main effects and all sex*factor interactions were also added, yielded several significant interaction terms: sex*period (P = 0.03) and an almost significant sex*plot (P = 0.06). We therefore analysed the two sexes separately.

For Blue Tits we tested both for a possible effect of intra- and interspecific competition, because, according to Reynoldson & Bellamy (1971) intra-specific competition must exist for interspecific competition to be possible.

In the Antwerp plots we found no significant effect of either on clutch-size, nesting success or number of young fledged per pair. We did find, however, that 15-day-nestling body mass and adult female, but not male body mass, were significantly reduced at higher densities of either species, and that laying was relatively later when Blue Tit density was experimentally increased.

In the Ghent comparison we could only study a possible effect of intra-specific competition. In this region, where the overall feeding conditions are less favourable, we found intra-specific effects on laying date and on nesting success. The latter effect was not detected at Antwerp.

All in all our results show that interspecific competition between Great and Blue Tit during the breeding season does occur. It seems to be stronger in areas/years in which food conditions are poorer.

Our results also show that interspecific competition does not occur in every year, implying that longer-term studies are more likely to detect them. This was suggested by Schoener (1983) in his review, although he underlined that very few long-term experimental studies on interspecific competition exist.

When large numbers of small-holed nest boxes are provided before winter the number of Blue Tits found roosting in the nest boxes in winter increases 3–6 fold and the breeding population in the following breeding season increases by 20–60% (Dhondt et al. 1991). Aviary experiments show that, given the choice, Blue Tits prefer large-holed boxes for winter roosting, but shift to small-holed boxes when Great Tits are present in the aviary. Winter competition for roosting site, therefore, plays an important role in the life of Blue Tits. By alleviating interspecific competition for cavities through the provision of small-holed nest boxes, Blue Tits acquire access to adequate winter roosting sites, and this results in a higher breeding density.

The basic equation describing changes in numbers within populations states that numbers increase by birth and immigration, and decrease by death and emigration. If we want to understand what causes a change in the number of breeding birds in our resident populations we must re-write that equation somewhat. Gains are not from births but from recruits (first time breeders). Losses are from mortality of birds that have bred previously, and only very rarely from emigration (Dhondt & Eyckerman 1980). The breeding population thus consists of first-time breeders and birds that already bred in the study plot in a previous season (survivors). Since most birds breed in nest-boxes, and since nearly all breeding birds are marked or identified, it is possible to distinguish between these two groups.

First-time breeders can be further split into two groups: immigrant recruits (born outside the study plot, mostly of unknown origin), and local-born recruits (born in the plot in which they later breed).

If we compare the breeding population (N) in two consecutive years t–1 and t, the change in numbers is defined as

In any one year the breeding population can be subdivided into the three groups defined above: adults that bred in the previous year (AD); local born recruits (LR); immigrant recruits (IR) or

The change between year (t–1) and year (t) is then

The change in numbers is the sum of the following three rates:

The reason we use the number of same-sex adults, rather than breeding pairs is that polygyny in Blue Tits is rather common (Dhondt 1989b), and hence the number of females is often larger than the number of males.

When we provided small-holed nest boxes Blue Tit breeding populations shift from a ‘low’ to a ‘high’ density. How does that increase come about, and how is it maintained?

In order to increase the density, at least one of the ‘rates’ must increase, but the new ‘high’ density can be maintained without any of the rates having to be different after the density increased. Of course, if the rates were the same before and after the density increase, the numbers of birds in each group will be larger. If at the new high density one rate is higher/lower than before, then one or both other rates must have decreased/increased. An implication of all this is that we need to evaluate what happens in the year of the change from low to high density separately from what happens in the subsequent high density years.

Our strategy to understand possible changes in population processes was to carry out two independent analyses: (1) To model adult survival rates using SURGE, a program that uses capture–recapture data of individually marked birds (Lebreton et al. 1993); (2) to compare the proportion among new recruits that was born locally or immigrated from elsewhere (provides information on changes in the ratio of local versus immigrant first time breeders); (3) and to compare the local recruitment rates between periods: This provides information on the locally realised fitness from the perspective of the breeding adult.

In order to model adult survival rates adequately one needs to capture a high proportion of the breeding adults in each year. Furthermore, in order to have sufficient power to distinguish between different models, sample sizes need to be large. Because one or both conditions were not met in some of the Ghent plots, the model selected by SURGE in the Ghent plots was one of constant survival in the periods studied (results not shown). If interspecific competition had an effect, we were unable to detect it.

In the Antwerp sites the populations tended to be larger, and 85% or more of the adults were trapped in each year. We feel, therefore, that survival analyses are meaningful for this data set.

Adult survival varied significantly between years, since the s_t, p-model was much better that the s, p model (constant survival over time). The s_t, p-model was compared to three models in which an effect of the experimental increase in Blue Tit density was assumed: (1) The survival rate differed between plot T and B only in 1984, the year of the increase in density. In all other years survival rates in Plots B and T were the same. This added one survival parameter. (2) Survival in B and T were the same for 1979–1982, but different in the other 10 years. This gave a model with 4+10+10 survival parameters. (3) Survival was different in the years 1979–1982, but the same in all later years. In this model there were 4+4+10 survival parameters.

We studied variation in Local Recruitment Rate (LRR) using PROC MIXED and the Macro GLIMMIX (SAS 1997) with three main effects (PLOT, PERIOD, SEX) and all interactions. YEAR was randomised within PERIOD. The three-way-interaction term was highly significant (F_2,33 = 6.11, P = 0.006), indicating that the change in LRR between periods differed between plots and sexes. We therefore continued the analyses for each sex separately and found a significant PLOT*PERIOD interaction for males (F_2,11 = 6.87. P = 0.01), but not for females (F_2,11 = 0.11. P = 0.90). A more detailed examination of the results (Table 8) shows that the significant interaction is caused by the PERIOD1*PLOT_T term being significantly different from zero. This implies that there are no significant differences between the LRR in the year of change, and in later years.

In the MA–SO comparison the PLOT*PERIOD interaction in a PROC MIXED analysis was not statistically significant for the LRR analysis (F_1,8 = 2.33, P = 0.16), although the LRR increased from 0.02 to 0.11 in MA, compared to no change in plot SO (Table 9). The increase in the proportion local first time breeders in Plot MA (from 0.02 to 0.18) was significantly larger that the increase in Plot SO from 0.03 to 0.07 (F_1,8 = 6.99, P = 0.03).