Numerical and functional responses of breeding passerine species to mass occurrence of geometrid caterpillars in a subalpine birch forest: a 30-year study




In northern Fennoscandia, the geometrid moths Epirrita autumnata and Operophtera brumata have cyclicities in density with mass occurrence at 10-year intervals. The larvae of Epirrita and Operophtera attain a size of 2–3 cm and 1.5–2 cm, respectively, and are nutritious food items for passerine birds. To examine whether these larvae have any numerical and/or functional influence on a passerine bird community (mountain birch forest in Budal, central Norway) during a 30-year period (1972–2001), I estimated their abundance (number of larvae per 100 sweeps) in the birch canopy, and the densities of breeding birds in the passerine community. In addition, from 1972 to 1998, I monitored the nesting success of five of the bird species. The foraging pattern of the most abundant bird species and their gizzard contents (adults and nestlings) were examined in 1972–78 (covering population peaks of both the geometrids). Population peaks of Epirrita occurred in 1975–76, 1985–86 and 1996, and of Operophtera in 1976–77, 1986–87 and 1997–98. The passerine community consisted of eight species that were territorial in all 30 years, one species in 26 years, three species in 14–21 years and three species in 1–4 years. Only the Brambling Fringilla montifringilla population responded numerically to the fluctuations of Epirrita and Operophtera. Brambling was also the only species whose mean clutch size varied between years, and this correlated positively with the density of Epirrita. The mean annual nesting success of Willow Warbler Phylloscopus trochilus, Bluethroat Luscinia svecica and Common Redpoll Carduelis flammea tended to be higher in years with mass outbreaks of Epirrita, but was significantly so only for Reed Bunting Emberiza schoeniclus. The abundance of Operophtera larvae showed no influence on the nesting success of any bird species. The passerines foraged more frequently in the birch canopy in the Epirrita outbreak years (1975–76) than in the years before or after. Gizzard analyses of five adult passerine species and their nestlings showed that Epirrita was the main food item in 1974–76. Even though Operophtera occurred in large numbers in birch trees in 1976 and 1977, only a few larvae were found in the gizzards of the passerines. None of the passerines showed an increase in their population density in the year following the larval outbreaks, but the densities of Willow Warbler and Bluethroat increased in the succeeding year, indicating a higher return rate for these species. The study shows the existence of a dietary response and also indicates a reproductive response to the changes in the abundance of Epirrita in mountain birch forest. The lack of numerical response in the passerines (except the Brambling) to the fluctuation in Epirrita contrasts with the pattern described for passerine communities in northern temperate deciduous forests in North America, where Lepidoptera caterpillars periodically have mass outbreaks.

The majority of passerine species breeding in temperate deciduous forests feed largely on arthropods from broad-leaved foliage, and defoliating caterpillars in particular seem to be preferred (Royama 1970, Busby & Sealy 1979, Martin 1987, Holmes 1988). Lepidoptera larvae are the major food items for many passerine birds during the breeding season and comprise most of the food given to their nestlings (e.g. Busby & Sealy 1979, Hogstad 1988, Perrins 1991). Several bird species show numerical responses to the abundance of caterpillar species (Gibb 1960, Enemar et al. 1984, Virolainen 1984, Lindström 1987, Crawford & Jennings 1989, Holmes et al. 1991, Sherry & Holmes 1991, Rodenhouse & Holmes 1992, Hogstad 2000a). An abundance of arthropods, especially caterpillars, has been found to improve the breeding success of many species and lead to an increased breeding density the following year (e.g. Virolainen 1984, Holmes et al. 1986, 1991, Lindström 1987, Holmes & Sherry 1988, Stenning et al. 1988, Sherry & Holmes 1991, Rodenhouse & Holmes 1992, Johnson & Geupel 1996). Although many studies have demonstrated that food may limit egg production, reproductive success and parental survival, the importance of food limitation as a regulating process is controversial (e.g. Martin 1987, Tremblay et al. 2003).

In northern Fennoscandia, the Autumnal Moth Epirrita autumnata and the Winter Moth Operophtera brumata (Lepidoptera: Geometridae) feeding on Mountain Birch Betula pubescens ssp. czerepanovii have a cyclicity in density of 9–10 years (Tenow 1972). Even though Epirrita also occurs at lower altitudes, mass outbreaks are restricted to the subalpine birch region at high altitudes. Operophtera is found in more peripheral parts of Fennoscandia, and also appears in central and northwestern Europe. The attacks by larvae of these geometrid species on the trees are relatively synchronized (Tenow 1972, Hogstad 1997). Because the decrease in larval abundance after a population peak coincides with chemical changes in birch leaves, a delayed induced defence mechanism in host plants, resulting in deterioration of food for the larva, has been suggested to be responsible for this cyclicity (e.g. Haukioja 1991, Tenow 1996, Kaitaniemi et al. 1998, but see Ruohomaki et al. 2000, Selås et al. 2001).

The larvae of Epirrita and Operophtera attain a size of 2–3 cm and 1.5–2 cm, respectively, and are probably more nutritious food items for passerines than are many chitinous arthropods. Being exposed on the foliage, the larvae are readily available to canopy-foraging insectivorous birds. However, little is known about the effect these larvae have on the foraging pattern of birds, their breeding success and survival, and what bird species are affected.

In subalpine birch forests, the Brambling Fringilla montifringilla and the Common Redpoll Carduelis flammea are key species in the passerine community. However, in contrast to the other species in the community, these species have no breeding site tenacity and select nesting sites within their breeding range mainly according to yearly variations in the food conditions (Mikkonen 1983, Enemar et al. 1984, Lindström 1987, Cramp & Perrins 1994, Hogstad 2000a). Even though the nestling diet of both Bramblings and Redpolls consists mostly of moth larvae in peak years for Epirrita (Hogstad 1988, 1996a), the population density of Redpolls seems to be related to the seed crop of the birch following rich flowering the preceding year, whereas that of Bramblings is, apparently, related to the abundance of Epirrita. I therefore expected that (1) the nomadic Brambling, a rather long-lived bird (Cramp & Perrins 1994), should be able to track the moth abundance without delay and thus respond numerically to the moth fluctuations, whereas the Redpoll, and the strongly philopatric passerines, should not. However, whether these moth larvae influence the nesting success of passerines is unknown. Because small passerines are observed eating considerable numbers of geometrid caterpillars in subalpine forests (Ytreberg 1972, Arvidsson & Klaesson 1986, Lindström 1987, Hogstad 1988, Thingstad & Fjeldheim 1999) and also feed their young with such larvae, it may be expected that: (2) in peak years of geometrid larvae, small passerines should forage more in the vegetation where the geometrid larvae occur, i.e. in the birch canopy; (3) birds should eat more larvae when larvae are abundant; (4) the breeding success (number of offspring produced) of insect-eating passerine species may be better in geometrid peak years than in years with a low abundance of such larvae; and (5) if so, an increase in their densities (i.e. increase in their return rate) in the year following the mass outbreak of larvae may be predicted.

To evaluate these suggestions, I estimated the abundance of E. autumnata and O. brumata larvae and the population densities of breeding birds belonging to the passerine community in a mountain birch forest in central Norway during 1972–2001. The nesting success of five of these passerine species was studied in 1972–98. Because year-to-year differences in weather and abundance of predators may be pronounced in the area, and affect the nesting success of some species (Hogstad 1988, 2000a), I have also related the nesting success to climatic factors and nest predation. In 1972–78, I studied the foraging patterns of the nine most abundant passerine species, and analysed the gizzard contents of adults and nestlings. During these seven years, the observations cover one population peak of each of the geometrids; from the population increase to its crash.


Fieldwork was carried out in a homogeneous subalpine mountain birch forest in the northern boreal zone, in Budal (62°45′N, 10°30′E), central Norway. The forest is unmanaged, and due to a short growing season (usually from the beginning of June until late August), the habitat has remained fairly constant. The forest extends from 750 to 900 m above sea-level, and the general tree height is 3–6 m. The shrub layer, rather dense in some places, comprises mainly Junipers Juniperus communis. Patches of willows, Salix spp., especially S. lapponum, S. glauca and S. phylicifolia, are found locally in moist habitats. Dwarf Birch Betula nana is common in more open locations.

Most of the study area is covered by snow until early May, and the snow cover is usually 30–50% in the last 10 days of May when the first passerines arrive in the area. The weather is very variable during the breeding season. During the study period, the mean temperature in June varied between 6.7 (1993) and 12.5 °C (1986, Fig. 1), and the daily temperature between −3 and +25 °C. In some years, up to 5 days in June had a temperature below zero and 7 days experienced snowfall. It rained for up to 27 days in June, with up to 72 mm of rainfall in a week. As the yearly number of rain-days in June is connected with low temperature, there is a negative correlation between the number of rain-days and the mean temperature in June (r = −0.62, P < 0.001, n = 30).

Figure 1.

Meteorological data representative for the study area during the breeding season (amount of precipitation and mean monthly temperature in June). The data are from Berkåk meteorological station, about 25 km west of the study area and 450 m above sea-level. Owing to lack of data from Berkåk in 1980–82, data from Kvikne meteorological station (about 20 km southwest of the study area) were used.

Fieldwork took place mostly from the last week of May to mid-July, when most birds leave the area. The density of the breeding birds was estimated by the territory mapping method in accordance with international recommendations (Anon. 1970, Bibby et al. 1992), within one permanent, small study plot of 30 ha (100 × 3000 m), about 800 m asl. Additional fieldwork, such as acquiring information on nesting events and the presence of colour-ringed birds in the area, was carried out in an area of 3.5 km2, which included the study plot. The passerine community consisted of eight species that were territorial in all the 30 years (1972–2001: Tree Pipit Anthus trivialis, Bluethroat Luscinia svecica, Fieldfare Turdus pilaris, Redwing T. iliacus, Song Thrush T. philomelos, Willow Warbler Phylloscopus trochilus, Brambling, and Reed Bunting Emberiza schoeniclus), one species (Redpoll) in 26 years, three species (Dunnock Prunella modularis, Redstart Phoenicurus phoenicurus, Willow Tit Parus montanus) in 14–21 years and three species (Lesser Whitethroat Sylvia curruca, Wood Warbler Phylloscopus sibilatrix, Bullfinch Pyrrhula pyrrhula) that occurred in 1–4 years.

Both Epirrita and Operophtera are univoltine, i.e. have one generation per year. The eggs, laid in September–October, overwinter and hatch at budburst the following spring. Newly hatched larvae are usually found in the first week of June. The Epirrita larvae live freely on the birch leaves, whereas the Operophtera larvae live inside birch leaves spun together. After five instars, the larvae of both species are fully grown at the beginning of July. The full-grown larvae descend from the birch tree canopy onto the forest floor where they spin a cocoon and pupate. The pupal period lasts about 1.5–2 months, whereupon the new generation flies and the cycle is completed (Tenow 1972).

The abundance of geometrid moths was estimated by collecting the larvae with a sweep-net from branches on the lowest 4 m of birch trees. Analyses of the vertical distribution of caterpillars (Epirrita) in 100 birch trees (lower vs. upper halves of the canopy of 5- to 6-m-high trees) on 4 July 1974 using sweep-netting showed no significant difference (t-test, t18 = 0.20, ns). Each year, 6–28 collections, each of 100 sweeps, were taken in the first days of July, i.e. when the larvae were in their last two instars: 4 and 5. The annual index is the mean number of larvae per 100 sweeps. Morphological changes in the larvae were studied during a population peak (1973–78, see Hogstad 1996b). The fresh body weight of larvae was determined using a balance to an accuracy of 0.1 mg. The size of the larvae is expressed as the width of the head capsule because the body is soft and the larvae grow relatively evenly during their development; the head capsule, by contrast, is sclerotic and does not grow within each of the five instars.

In 1972–78 arthropods that were collected were determined to family or genus level. They were divided into seven groups: three larval groups (Epirrita, Operophtera, other larvae) and four imagine groups (Coleoptera, Diptera, Hymenoptera, others). The number of arthropods caught annually in the birch canopy varied between 930 and 12 391, the mean being 4043.

Information on the foraging sites and gizzard content (oesophagus and stomach) of the passerines was collected from mid-June to about 10 July in 1972–78. Data on foraging sites were obtained by recording where the birds were first observed attacking prey. The following categories of foraging sites were recorded: birch trees (inner and outer parts of the foliage), shrub layer (< 2 m tall) and ground (including herbaceous vegetation and dead branches on the forest floor). I made only one foraging record for each bird per tree, and no more than three consecutive records of any individual were collected in a day. The yearly number of foraging records were as follows: Tree Pipit 15–61, Bluethroat 19–111, Fieldfare 58–216, Redwing 69–157, Song Thrush 8–32, Willow Warbler 490–808, Brambling 120–911, Redpoll 39–256, Reed Bunting 18–156. The foraging niche breadth (B) of these three niche dimensions is expressed using Levin's index:


where pi is the proportion of observations falling in the ith of n categories. B can vary from 1 to n (see Price 1975).

Gizzards of adult birds (mostly males) and nestlings (mostly one per nest and 7–8 days old) were conserved in 70% ethanol immediately after they were killed and later examined under a binocular microscope. The birds were killed under a licence from the Directorate for Nature Management. A total of 97 and 67 gizzards of adults and nestlings, respectively, were examined (Table 1).

Table 1.  Numbers of gizzards examined from adult birds and nestlings (in parentheses) collected from mid-June to about 10 July.
Bluethroat1 1 (1) 1 (1)1 (1)1 (1)1 (1) 6 (5)
Fieldfare1 (6) 1 (6)
Redwing 1 (3) 1 (4)1 (3)4 (1)1 (7) 8 (18)
Willow Warbler2 (1)12 (2) 7 (2)4 (2)2 (3)1 (2)1 (2)29 (14)
Brambling512 (4)10 (8)7 (1)3 (2)2 (2)140 (17)
Common Redpoll41 5
Reed Bunting2 1 (2) 1 (2)1 (1)1 (1)1 (1)1 8 (7)

Nesting data were compiled for all the species, but due to a small sample size for some, the nesting success has been determined for only five species. Nests were mostly visited every 2–4 days until nesting failed or the young fledged. The yearly nesting success was calculated by dividing the number of young fledged by the number of eggs in complete clutches for Redpoll (open nest, most low in birch), Bluethroat and Reed Bunting (open nest on or close to the ground). For Willow Warbler (nest on the ground, domed with a small entrance hole on the side, making it difficult to count the nestlings without disturbing the birds) and the Brambling (nest often high up in trees), nesting success was determined as the proportion of nests where young fledged (without counting the number of young) relative to the number of nests with complete clutches.

Nest predation varied from year to year. Nest predators present in the area were Common Ravens Corvus corax (one pair), Hooded Crows Corvus corone cornix (one or two pairs), Pine Martens Martes martes, Stoats Mustela erminea and Weasels Mustela nivalis. Judged by the number of tracks seen in the snow in winter and spring, the densities of the last two species varied from year to year, and were highest when small rodents were at peak numbers. Accurate density indices of these mustelids could not be assessed due to their cryptic way of life (cf. Marcström et al. 1990). Heavy nest predation by mustelids has been demonstrated for tree-nesting birds in years with low densities of small rodents, their primary prey (e.g. Järvinen 1985, Sørensen et al. 1990). As disturbed lining in several nest cups emptied of eggs may indicate that mammalian predators had visited the nests, I have related the nest predation of the birds to rodent density (index 0 = no tracks in the snow, no rodents observed during the breeding season; 1 = few tracks, rodents rarely observed; 2 = new tracks seen almost daily, rodents regularly observed; 3 = many tracks, rodents frequently observed).

All tests are two-tailed, and were performed using SPSS 11.0. Means are presented ± 1 sd.


Abundance of the moth larvae in 1972–2001

The population densities of the moth larvae varied considerably. Epirrita larval numbers peaked in 1975–76, 1985–86 and 1996, and Operophtera in 1976–77, 1986–87 and 1997–98 (Fig. 2). The populations of the two species fluctuated fairly synchronously (Spearman rank correlation rs = 0.62, P < 0.001, n = 30), but the peaks of Operophtera were one year later than those of Epirrita. Thus, the correlation between population densities of Epirrita (year t+ 1) and Operophtera (year t) was higher (rs = 0.73, P < 0.001, n = 29).

Figure 2.

Fluctuations in the abundance of Epirrita (solid line) and Operophtera (broken line) during 1972–2001.

Population fluctuations of the passerines in relation to larval abundance

Only the Brambling population responded numerically to the fluctuations of Epirrita (rs = 0.82, P < 0.001) and Operophtera (rs = 0.52, P < 0.01). The other species in the passerine community revealed no numerical response to caterpillar abundance (rs =−0.31 to 0.31, ns). When controlling for Operophtera fluctuations over time in a partial correlation analysis, the population density of Bramblings still correlated with that of Epirrita (r27 = 0.59, P = 0.001). In addition, when controlling for fluctuation of Epirrita in a partial correlation analysis, the density of Bramblings (r27 = 0.50, P < 0.01) correlated significantly with Operophtera numbers.

Even though all the passerine species studied largely ate caterpillars in the Epirrita mass outbreak years, and, although not well quantified, increased their nesting success in such years (see below), none of the birds increased their population density in the years following each of the three mass outbreaks (see Table 2). However, Willow Warblers and Bluethroats increased their densities 2 years after the outbreak years (Willow Warbler: 5–17 territories (terr.)/km2, from a mean of 41.5 terr./km2 in the mass outbreak years to 59.3 terr./km2 2 years later; Bluethroat: 2–7 terr./km2, from a mean of 5.7 terr./km2 to 9.7 terr./km2), indicating that these species might have had a higher return rate.

Table 2.  Increase (+), decrease (–) or similar (0) population density (territories/km2) of passerines compared with that in the mass outbreak year, one (year t + 1) or two (t + 2) years after mass outbreaks of Epirrita (year t) in the study area. The density of Epirrita in the outbreak year of 1966 is not quantified (see Hogstad 1968).
t + 1t+ 2t + 1t+ 2t + 1t+ 2t + 1t+ 2
Tree Pipit– – 0– 0+
Bluethroat+– +– ++
Redstart+– 0    
Willow Warbler+– +++
Brambling– – – – 
Redpoll00– – – 
Reed Bunting++– +
Total community+– – ++

Clutch size and nesting success

The yearly mean clutch size for Willow Warblers (mean for 1972–98: 6.1 ± 0.53, n = 64), Bluethroats (6.6 ± 0.51, n = 16), Reed Buntings (5.2 ± 0.39, n = 39) and Redpolls (5.2 ± 0.49, n = 58) did not vary over the years (Kruskal–Wallis test, H = 7.09–23.59, ns), whereas that of Bramblings (5.6 ± 0.73, n = 95) did (H = 48.1, df = 24, P < 0.01). The variation in Brambling clutch size could be due to differences in climatic factors (mean June temperature, rainfall or number of rain-days) as well as the abundance of geometrids. A multiple stepwise regression analysis revealed that the abundance of Epirrita accounted for 27% of the overall variation in the clutch size of the Brambling, whereas the remaining factors did not contribute significantly towards explaining the clutch size variation. Thus, the yearly mean clutch size of Bramblings was positively correlated with the density of Epirrita (rs = 0.50, P = 0.01, n = 25), but not of Operophtera (rs = 0.39, ns). The other bird species showed no significant relationships with caterpillar density.

The mean yearly nesting success of Willow Warblers and Bramblings was about 80% and that of Bluethroats, Redpolls and Reed Buntings about 65% (Table 3). Although the nesting success of these species was slightly higher (except for Brambling) in years with a mass outbreak of Epirrita (> 30 larvae/100 sweeps) compared with the other years, the difference was only significant for Reed Buntings. The abundance of Operophtera did not influence the nesting success of the species (t = −0.58 to 1.10, ns).

Table 3.  Mean nesting success (%, ± sd) of five passerine species during 1972–98, and their nesting success in years with low abundance of Epirrita (< 30 larvae/100 sweeps) and high abundance (> 30 larvae/100 sweeps: 1975, 1976, 1985, 1986, 1996). Figures in parentheses denote the number of years.
Species1972−98Epirrita abundancetP
  • a

    Nesting success given as the ratio number of nests where young fledged (without counting the number of young) related to the number of nests with complete clutches.

  • b

    Nesting success calculated as the number of young fledged from the number of eggs in complete clutches.

Willow Warblera79.9 ± 19.0 (20)78.1 ± 20.2 (15)85.0 ± 15.9 (5)−0.690.50
Bramblinga85.2 ± 15.1 (25)85.9 ± 16.3 (20)82.8 ± 10.3 (5)  0.400.70
Bluethroatb66.7 ± 40.2 (18)61.5 ± 38.9 (13)80.0 ± 44.7 (5)−0.870.40
Redpollb74.7 ± 26.5 (16)68.3 ± 27.1 (12)94.0 ± 12.0 (4)−1.810.09
Reed Buntingb64.6 ± 38.9 (24)57.6 ± 39.0 (20) 100 ± 0 (4)−2.140.04

Variation in the nesting success of the species (Table 3) and the relationships involving Epirrita abundance, rodent abundance (as an index of nest predation), amount of rain or number of rain-days in June are shown in Table 4. A multiple stepwise regression analysis revealed that the number of rain-days accounted for 27% of the overall variation in the nesting success of Willow Warblers, whereas the remaining factors did not contribute significantly towards explaining the nesting success. Likewise, the amount of rain accounted for 29% of the variance in nesting success of Bluethroats, and rodent abundance explained 20% of the variation in the nesting success of Bramblings. However, even though rainy days influenced the nesting success of Willow Warblers and Bluethroats, these species were affected less by periods of cold, rainy weather than were Bramblings. Thus, in the cold, wet summers of 1978, 1979 and 1981, nests with eggs or nestlings of Bramblings were abandoned, and several birds left the area during the second half of June, whereas Willow Warbler nests were apparently not affected. Furthermore, Bramblings caught in June 1981 had no subcutaneous fat, visible as fat deposition in the furculum, in contrast to birds caught in 1980 (see also Hogstad 1982). However, in 1996, when a mass outbreak of Epirrita occurred, there were several days of exceptionally rainy weather in the nestling period, but no effect on the breeding success of Bramblings was recorded.

Table 4.  Correlation values (Pearson) between the mean yearly nesting success of the species studied and some factors.
SpeciesAbundance of EpirritaAbundance of OperophteraRodent densityMean temperature in JunePrecipitation (mm) in JuneNo. of rainy days in June
  • *

    P < 0.05 in a two-tailed analysis.

Willow Warbler  0.28  0.11−0.09  0.22−0.19−0.52*
Brambling−0.03  0.13  0.45*−0.20−0.20−0.11
Bluethroat  0.29−0.09  0.19  0.25−0.54*−0.26
Redpoll  0.33  0.14−0.19  0.27  0.02−0.18
Reed Bunting  0.30−0.11  0.26−0.31−0.30  0.17

Abundance of arthropods in the birch canopy in 1972–78

The mean total number of arthropods and the mean number of each of the seven groups caught per 100 sweeps in the birch canopy varied considerably from one year to another (Fig. 3; Kruskal–Wallis test for each of the groups: H = 44.5–103.8, df = 6, P < 0.001). The large numbers in 1975–77 were mainly due to the abundance of diptera in 1975 and 1976 and Operophtera larvae in 1976 and 1977. In 1978, there were fewer phytophagous insects than in any other year when arthropods were determined and counted, namely 1971–81 (cf. Hogstad 1988).

Figure 3.

Yearly mean number of arthropods collected per 100 sweeps with a net in birch trees from mid-June to about 10 July. Figures below the year denote the number of collections, each of 100 sweeps.

The abundance of insect larvae in the birch trees in 1972–78 rose gradually from 6.6% (percentage proportion of the total number of arthropods caught by sweep-netting) in 1972 to 78% in 1977, and then declined to 6.2% in 1978 (1972 = 6.6%, 1973 = 13.1%, 1974 = 30.2%, 1975 = 24.8%, 1976 = 51.7%, 1977 = 78.0%, 1978 = 6.2%). The larvae consisted mostly of Epirrita in 1972–74 and 1978 (65–80%, mean = 73%) and Operophtera in 1975–77 (53–96%, mean = 76%), whereas larvae of Diptera, Coleoptera, Hymenoptera and Lepidoptera other than geometrids were present in small numbers.

The yearly mean size of the Epirrita larvae (expressed as the width of the head capsules) and the mean body weight of the 5th instar were relatively stable during 1973–76 (capsule width 1.71–1.75 mm, mean = 1.72 mm, n = 648; mean fresh body weight 86 mg, n = 198), but decreased in the crash year of 1977 (capsule width 1.68 mm, n = 35; mean body weight 45 mg, n = 10) and in 1978 (capsule width 1.58 mm; mean body weight 29 mg, n = 2). The capsule width of the 5th instar of Operophtera decreased evenly from 1.45 mm in 1973 to 1.29 mm in 1978, whereas the mean body weight remained fairly stable in 1974–77 (36–44 mg, mean = 41 mg) but dropped to 22 mg in 1978 (see Hogstad 1996b).

Insect imagines were of Homoptera, Heteroptera, Diptera, Coleoptera and Hymenoptera. Even though diptera (mostly small Nematocera and Cyclorrhapha) accounted for a large proportion of the arthropods, they did not constitute a major proportion of the biomass in the sweep-net catches. Aranea and Opiliones occurred regularly but in small numbers.

Foraging patterns of the birds

Spatial patterns: the use of trees, bushes or ground

As expected, the passerine species foraged most frequently in the birch canopy in the Epirrita outbreak years (Fig. 4). For instance, the birds foraged more in birch trees and less in the field layer during the mass outbreak of Epirrita in 1975–76 compared with the preceding years (1972–74: Willow Warbler: χ2 = 79.13; Brambling: χ2 = 221.03; Reed Bunting: χ2 = 33.14; Tree Pipit: χ2 = 26.25; Redpoll: χ = 15.17, all P < 0.001, df = 2) or after the outbreak (1977–78: Willow Warbler: χ2 = 74.23; Brambling: χ2 = 208.30; Reed Bunting: χ2 = 19.98; Bluethroat: χ2 = 11.23; Tree Pipit: χ2 = 21.98; Redpoll: χ2 = 10.63, all P < 0.01, df = 2). When the foraging patterns of these species in the years before and after the outbreak were compared, no differences were found (χ2 = 0.36–5.87, df = 2, ns). Accordingly, the niche breadths of each of the foliage-foraging species, Willow Warbler, Brambling and Redpoll, narrowed in 1975–76 compared with the 5 years of 1972–74 and 1977–78 (Willow Warbler: from 1.82 to 1.54; Brambling: from 2.02 to 1.41; Redpoll: from 2.08 to 1.83). By contrast, the niche breadths of the usually ground-feeding species expanded: Tree Pipit from 1.52 to 2.05, Bluethroat from 2.21 to 2.75, Reed Bunting from 1.78 to 2.26 and Redwing from 1.03 to 1.72. Fieldfares and Song Thrushes, which usually find their food on the ground, were also observed foraging in the birch canopy attacking larvae during the outbreak years.

Figure 4.

Foraging substrate (ground, bushes, birch trees) used by the foliage-foraging species, Willow Warbler, Brambling and Redpoll, and the ground-foraging species, Tree Pipit, Bluethroat, Redwing and Reed Bunting, from mid-June to about 10 July in 1972–78. Figures denote the number of foraging records.

Foraging density

To estimate the relative importance of the different height strata (trees, bushes or ground) as foraging substrates for the birds over the years 1972–78, I calculated the yearly proportion of their use of these strata on the basis of the foraging records used in Figure 4. These values were then multiplied by the density of each species (number of territories/km2; Table 5), and summed over species, yielding a measure of the ‘foraging density’. As the majority of foraging records were from birch trees, only foraging density values for trees are given. Even though the foraging indices are only an indirect measure, they probably provide a relative index of the proportional representation of the yearly foliage utilization of the birds during the 3-week sampling period, i.e. when the geometrid larvae were at their largest. When the data for the nine most abundant species were pooled, a significant positive correlation was found between the yearly foraging density of the birds in birch trees and the abundance of Epirrita (Fig. 5; rs = 0.89, P < 0.01, n = 7) but not of Operophtera (rs = 0.24, ns). Although the density of the birds (except Bramblings) fluctuated independently of the abundance of Epirrita, their utilization of the birch canopy as a foraging site was clearly dependent on the abundance of Epirrita.

Table 5.  Yearly densities (D = number of territories/km2) of the nine most abundant species in the passerine community and their proportion (P) of foraging records in birch trees (in relation to trees, bushes, ground).
Tree Pipit 50.24 20.17 30.19  50.49 70.61 20.33 70.14
Bluethroat 70.12130.14130.17  30.26 50.26 30.11120.08
Fieldfare300.00420.00470.00 500.00500.00430.00150.00
Redwing 70.00100.00100.03 100.08150.11 80.00 70.00
Song Thrush 20.00 30.00 20.06  50.13 50.11 20.00 30.00
Willow Warbler670.70530.74500.72 430.82430.75380.74600.66
Brambling330.61420.61470.67 500.93500.73430.60150.52
Common Redpoll100.54 30.49130.55 200.73170.61130.53 00.00
Reed Bunting170.04180.03230.09  30.17120.20120.05170.05
Foraging density (D × P)75.1 69.0 79.9 101.6 89.3 62.4 50.2 
Mean foraging density 8.4  7.7  8.9  11.3  9.9  6.9  5.6 
Figure 5.

The relationship between the yearly abundance of Epirrita larvae and the pooled ‘foraging density’ (see text for explanation) of nine passerine species in birch trees over the years 1972–78.

Gizzard analyses of adult birds

Epirrita dominated and comprised on average 93% (84–100%) of the larval content in the gizzards of the passerines from 1973 to 1976, i.e. the years when the abundance of Epirrita far exceeded 10 larvae/100 sweeps. In 1972 and 1978, none of the gizzards examined contained Epirrita, but some larvae were found in Willow Warblers, Bramblings and Redpolls in 1977 (Fig. 6).

Figure 6.

The percentage of larvae and imagines of arthropods found in gizzards of adult birds of the most abundant species in 1972–78.

It is noteworthy that the Brambling was the only species whose gizzard content of Epirrita (%) almost synchronously followed Epirrita abundance (Fig. 7; r = 0.98, P < 0.001, n = 7). Although the gizzard content of Epirrita (%) in Willow Warblers and Reed Buntings was also positively correlated with the abundance of Epirrita (r = 0.86 and 0.82, respectively, both P < 0.05, n = 7), they had the highest proportion of Epirrita in their gizzards in 1974, the year before the Epirrita mass outbreak. Operophtera occurred in large numbers in the birch trees in 1976 and 1977, but only a few larvae were found in the gizzards analysed.

Figure 7.

The percentages of Epirrita larvae in the sweep-net samples (stippled) and those found in gizzards of adults (broken line) and nestlings (solid line) of the most abundant bird species in 1972–78.

Of the insect imagines, Coleoptera and Hymenoptera were the main prey items of most passerine species. Of the Coleoptera, Curculionidae and Chrysomelidae dominated numerically. About 95% of the curculionids were Polydrosus undatus and P. ruficornis. Of the chrysomelids, Syneta betulae was the dominant species. Except for Tenthredinidae (Hymenoptera), which made up about 5–10% of the total prey items, the other insect orders were represented in only small numbers.

Gizzard analyses of nestlings

Although the gizzards of the nestlings generally contained the same insect groups as were found in adult birds, the mean proportion of larvae in nestling gizzards was nearly twice as high (c. 60% in each of the five passerines examined; Table 6) as in adults (c. 35%). The mean total number of larvae in the nestling gizzards did not vary in different years in the case of Willow Warblers, Bramblings, Bluethroats or Redwings (Kruskal–Wallis; H = 0.30–5.54, ns). Epirrita made up on average about 80% of the total number of all larvae (1973–77) in nestlings (67–95%, mean 81%) as well as in adults (67–100%, mean 85%) of these passerines, whereas Operophtera occurred sporadically in 1976 and 1977 when this species made up about 7% and 24%, respectively, of the total number of larvae.

Table 6.  The yearly mean proportion (%) of all larvae (1972–78) and the proportion of Epirrita among these larvae (mean for 1973–77) in gizzards of adults and nestlings. N denotes the number of gizzards.
NLarvae totalEpirritaNLarvae totalEpirrita
Bluethroat 634 67 56580
Redwing 849100186595
Willow Warbler2930 97145775
Brambling4035 87175587
Reed Bunting 829 75 75967
Mean:9135 85616081

The proportion of Epirrita was already relatively high in nestling gizzards of the five passerine species in 1973, even though the abundance of Epirrita in the birch trees was low (Fig. 7). As expected, the proportion of Epirrita was high in 1974–76.

The relationships between the abundance of Epirrita and the gizzard contents of Epirrita in adults and nestlings were significant or nearly so in both adult and nestling Willow Warblers (r = 0.86, P = 0.01, n = 7; r = 0.73, P = 0.06, n = 7, respectively), Bramblings (r = 0.98, P < 0.001, n = 7; r = 0.82, P < 0.05, n = 6, respectively), Redwings (r = 0.85, P = 0.07, n = 5; r = 0.85, P = 0.07, n = 5, respectively) and Reed Buntings (r = 0.82, P < 0.05, n = 7; r = 0.89, P < 0.05, n = 5, respectively), but this was not the case for Bluethroats (r = 0.66, ns, n = 6; r = 0.56, ns, n = 5, respectively).


This study has demonstrated a change in the foraging pattern and the existence of a dietary response, and has also indicated a reproductive response in breeding passerines, to changes in the abundance of Epirrita in a subalpine birch forest. I am not aware of other long-term studies that have documented such effects from Epirrita on northern passerines. The niche breadth narrowing or expanding of the foliage-foraging species and the ground-foraging species, respectively, agrees with optimal foraging theory, because the birds specialized on the abundant food resources, i.e. Epirrita. As predicted, the Brambling was the only species that responded numerically to the fluctuations of Epirrita. A similar numerical response to Epirrita fluctuations was found in a subalpine birch forest in Ammarnäs, Swedish Lapland (Enemar et al. 1984, Lindström 1987), although the correlation was not as strong as in Budal. The small discrepancy in the relationship between Epirrita and Brambling densities in the two areas may be connected with habitat differences: areas with rich meadow birch forest in Ammarnäs vs. oligotrophic heath birch forest in Budal. Furthermore, in Ammarnäs, the highest Brambling density was already reached during the build-up phase and then decreased even though caterpillars were still abundant.

Because the Epirrita populations fluctuate cyclically, with peak densities attained about every 10 years, both population peaks and crashes are predictable. Thus, the rather long-lived, nomadic Brambling may be able to evaluate the abundance of Epirrita larvae and choose to breed in areas where Epirrita are abundant. This suggestion seems to be strengthened by the findings that Bramblings were largest (largest wing lengths) in years with mass outbreaks of Epirrita (Hogstad 1985a, Lindström 1987). It may therefore be suggested that larger and older (cf. Hogstad & Røskaft 1986) birds move into the area in years with a good food supply. Furthermore, larger Brambling females had larger clutches and laid heavier eggs (Hogstad 2000a, 2000b) than smaller females. Together, these findings suggest increasing breeding success with age, making age a determinant for the reproductive success of Bramblings.

The dependency of Bramblings on the larvae was most obvious in the cold, rainy summers in Budal, as the breeding success of the birds was poor when the Epirrita larvae were uncommon but normal when they were abundant. As in Budal (Hogstad 1982), the summer of 1981 was cold and wet in Finland, and in Lapland Bramblings produced very few young although the Bluethroat and Redstart maintained the same success as in other years, i.e. a mean of 73–77% (Hilden et al. 1982). The poor success of Bramblings in both Lapland and Budal was due to females leaving their nests unattended for long periods, deserting their nests or having their nests robbed by Hooded Crows or mustelids during the cold period. Although weather conditions and nest predation may also play a decisive role in the reproductive success of Willow Warblers (Hogstad 1985b, Arvidsson & Klaesson 1986, present study), this species was far less affected than Bramblings.

An explanation for the behaviour differences between Bramblings and the other passerines breeding in the same area may be that the species have different stress responses to harsh conditions. Breeding Bramblings exposed to acute stress increase their levels of the stress hormone corticosterone more rapidly than Willow Warblers (Silverin 1998). Because of its nomadic tendency, even in the breeding period, the Brambling may be poorly adapted to a variable food situation (Hogstad 2000a). The greater resistance to stress among Willow Warblers compared with Bramblings (Silverin et al. 1997) allows them to continue breeding despite severe and disruptive environmental conditions. However, when food is in good supply, i.e. when Epirrita larvae are abundant, the breeding success of Bramblings was found to be normal even under severe climatic conditions. Being more vulnerable to poor weather in years when their food is scarce, but being able to withstand bad weather in years when food is abundant, has also been found in the Nuthatch Sitta europaea and in tit species (Matthysen 1989).

Epirrita is clearly favoured by all birds in preference to all other arthropods, and Operophtera larvae were apparently eaten by the birds far less than would be expected from their abundance, biomass and accessibility. Even in 1977, when Operophtera were abundant and there were very few Epirrita, Operophtera larvae comprised only a small fraction of the gizzard content of both adult birds and their nestlings. Operophtera obviously had little or no measurable effect on bird reproduction.

Although there were relatively few Epirrita in 1973, Bluethroats, Willow Warblers, Bramblings and Redwings fed their nestlings with such larvae 3–4 times more than they did in the crash year of 1977 when the abundance of Epirrita was at the same level. This may be connected with the quality of the larvae. The mean body weight of the 5th instar (the last) was fairly stable (about 80 mg) when the population density was increasing (1973–76), but was reduced by about 50% in the crash year (1977). Although Operophtera were about four times more abundant than Epirrita in 1977, and their mean body weight was similar to that of Epirrita that year, the birds scarcely ate Operophtera larvae or fed their nestlings with them. As an induced defence mechanism in the birch trees increases with the abundance of defoliating larvae (Haukioja 1991, Tenow 1996), this may result in the food for the larvae deteriorating in quality. Because mass outbreaks of Operophtera follow 1 year after those of Epirrita, it may be speculated that the quality of the Operophtera larvae is poorer, i.e. low nutrient content or digestibility, and therefore these larvae are little used as food by the birds. On the other hand, the different behaviour of the two moth species, Epirrita larvae being in nearly constant motion on birch leaves whereas Operophtera remain curled up in a birch leaf, may partly explain the different preferences of the birds.

Although all the species fed on Epirrita in peak years for the larvae, and Willow Warblers, Bramblings, Bluethroats and Reed Buntings tended to have better reproductive success in such years, none of the species increased in abundance in the year following a peak. However, Willow Warblers and Bluethroats increased their densities in the second year after peak years, and it may therefore be suggested that, because of better nesting success, these species responded to the mass abundance of Epirrita by increasing density. Although not quantified, Epirrita also had a mass outbreak in Budal in 1966 (Hogstad 1968). Thus, four population peaks of the larvae occurred during 1965–2001. If this 1966 peak is included (Table 2), it appears that the population density of the Willow Warbler tends to increase in the first year after an Epirrita peak year. Moreover, the densities of Willow Warblers and Bluethroats increased 2 years after the peak (Willow Warbler, t6 = −2.37, P = 0.056; Bluethroat: t6 = −3.54, P = 0.017), in contrast to the other species. This indicates that a high abundance of Epirrita larvae as food for Willow Warblers and Bluethroats leads to a high return rate (i.e. better survivorship) of these bird species in the subsequent years, possibly because of a decrease in mortality immediately after fledging. There is little information about the return rate of Bluethroats, but that of adult Willow Warblers is relatively high. In southern Finland, 39% of males and 17% of females were recaptured in years after ringing (Tiainen 1983), and in southern England 38% (both sexes; Lawn 1982), or 30% and 17% for males and females, respectively, returned to their former territories (Pratt & Peach 1991). In Sweden, the return rate was 27% in a subarctic birch forest in the north, and 30% in a rich deciduous forest in the south (Silverin et al. 1997).

The present study shows that in years with a mass occurrence of Epirrita, passerines expanded their foraging upwards into the canopy, feeding themselves and their nestlings with the Epirrita larvae. The occurrence of Epirrita also tends to increase the nesting success of several passerine species, which in turn may affect the community structure. However, only one species, the Brambling, responded numerically to the Epirrita fluctuations. This contrasts with the pattern described for passerine communities in northern temperate deciduous forests in North America where defoliating Lepidoptera caterpillars (e.g. Spruce Budworm Choristoneura fumiferana, Gypsy Moth Lymantria dispar and Heterocampa guttivitta) have periodic and infrequent major outbreaks (cf. Morse 1989). Some specialists among American spruce warblers may not nest within large areas for long periods unless mass outbreaks of some Lepidoptera species occur. The abundance of these species often becomes extremely high during major outbreaks, and their numbers fluctuate in parallel with the caterpillar biomass of the previous summer (Holmes 1990). Some warblers respond to an abundance of caterpillars by raising two broods instead of one (Holmes et al. 1991, Rodenhouse & Holmes 1992). As these warblers reached their peak numbers 1 or 2 years later than the caterpillars, breeding success was probably involved. Because these birds can reduce the abundance of Lepidoptera larvae on tree foliage (Holmes 1990), it was suggested that food during the breeding period is an important limiting factor in northern temperate deciduous forest. Such specialists apparently do not exist in subalpine birch forest in Scandinavia (e.g. Enemar et al. 1984, this study), and there is probably sufficient food to raise young in most years.


I thank O.A. Indset, I. Stenberg and O. Tovmo for field assistance, M. Munkönen, O. Tenow and P.G. Thingstad for valuable comments on the manuscript, and R. Binns for improving the English.