An insect (Argyresthia retinella, Lep., Yponomeutidae) outbreak in northern birch forests, released by climatic changes?

Authors


Dr O. Tenow, Division of Forest Entomology, Swedish University of Agricultural Sciences, S-750 07 Uppsala, Sweden. Telefax: +46–18 67 28 90.

Abstract

1. In the early 1990s, birch Betula pubescens L. forests in north-western Norway were damaged by the bud- and shoot-mining larvae of Argyresthia retinella not previously known for outbreaks. In 1993–96, the outbreak was mapped and changes in attack intensity and foliage structure were quantified by sampling birch twigs along transects from coast to inland. Results were considered in relation to a variable climate.

2. The outbreak extended 400 km along the coast, mainly within the lowest 100–150 m a.s.l. It started in the late 1980s and the attack intensity culminated in 1993–94. Repeatedly, 30–50% of the leaf-carrying shoots were damaged or killed. Trees compensated by sprouting clusters of shoots from undamaged shoots and, hence, increased foliage clumping. Eventually, twigs and branches died.

3. The overwintering post-diapause egg, the larva and the adult stages were considered the most sensitive to changes in temperature climate. The air temperature, averaged for a combination of these stages, was calculated for each year, as well as deviations from the long-term (1868–1996) average.

4. The 5-year running average of deviations approached or exceeded + 1 °C three times: in the 1930s, around 1960 and around 1990. The third peak coincided with the outbreak of the 1990s. There are no reports of an A. retinella outbreak in either the 1930s or 1960s. However, in one photographic documentation from 1940, clumped foliage structure of birch suggests an outbreak in the 1930s.

5. Severe A. retinella attacks appear to be a fairly new phenomenon, possibly connected to recent high temperature deviations. If so, with the present air temperature climate, outbreaks may occur at intervals of ≈20–25 years. With a trend of decreasing air temperature, outbreaks may be less frequent. At higher temperatures, natural or from anthropogenic global warming, the outcome is more uncertain although more frequent outbreaks may occur initially.

6. It is recommended that foresters learn to identify damage made by A. retinella and include observations on this insect in their reporting. Monitoring A. retinella damage in the North and elsewhere, should contribute to the understanding of the outbreak ecology of this insect.

Introduction

In 1991, severe damage to the foliage of the birch Betula pubescens L. forest was observed along the north Norwegian coast in the county of Nordland. Many leaves were wilted. Between the living, leaf-carrying shoots (short and long shoots; for definitions, see Maillette 1982) many short shoots were dead, giving the foliage a sparse appearance. Apparently, the damage had accumulated for some years. In 1992, similar damage was observed at Alta Fjord, about 400 km NE of the first area (Solberg et al. 1994). In both instances, the attacking insect was larval Argyresthia retinella Zell. (Lepidoptera, Yponomeutidae), mining and killing the buds of birches (Johansen & Kobro 1996). This species is indigenous to Norway and the rest of Fennoscandia (Svensson et al. 1987) but was not previously known as an outbreak insect in Scandinavia or elsewhere.

Birch forest is the dominant natural ecosystem in northern Norway. Its significance as a source for building material, fuel and fencing, and for shelter, reindeer husbandry, game hunting, recreation and tourism is fundamental. The seriousness and seeming novelty of this outbreak and its coincidence with a sequence of years (1987–93 1995) with mild (see Vaer 1989–95) and short (cf. Myneni et al. 1997) winters make it interesting in several respects, including its possible dependence on climate variability and future climate-dependent incidence.

Here, we report on the geographical distribution of the damage caused by A. retinella, changes in attack intensity over years, compensatory reactions and structural changes of birch foliage, and on the climatic background of the outbreak. Occurrence of earlier outbreaks in the region and the dependence of outbreaks on climate variability, natural or anthropogenic, are discussed. We wish to call the attention of forestry authorities to this threat to the forest.

Materials and methods

Argyresthia retinella zeller 1839 (birch bud moth)

Descriptions of the life history of A. retinella mainly relate to continental or western European conditions (Petersen 1924; Schütze 1931; Werner 1958; Friese 1969; Agassiz 1987). Recent studies have demonstrated that these descriptions do not fully apply to populations in northern Norway (F. Elverum & A.C. Nilssen, unpublished data). Recently summarized data on bionomics (Svensson 1993) should be representative for southern Scandinavia and continental Europe.

Species determination keys for larvae and imagines of Argyresthia have been published by Werner (1958) and Friese (1969), respectively. In our material, young larvae were light green with a brown head capsule. Full-grown but still feeding larvae were 5–6 mm long and light-brown with a transparent contour (Fig. 1a). Larvae about to pupate were whitish.

Figure 1.

Argyresthia retinella and its host, the birch Betula pubescens L: (a) larva on shoot (photo supplied by Tor Johansen); (b) short shoots with wilted leaves; (c) clusters of new shoots from unattacked shoots; (d) dying tree in the Tverrelva valley, Alta, Norway, 12 July 1994.

In continental and western Europe, larvae of A. retinella mine buds and shoots of Salix caprea L., Betula and Quercus from April to June (Schütze 1931; Werner 1958; Agassiz 1987). In Troms, northern Norway, young larvae were found in developing buds of B. pubescens from ≈20 May and full-grown ones on ≈1 July 1995 (determined by rearing from larval to adult stage; F. Elverum & A.C. Nilssen, unpublished data). Attacked shoots are recognizable because their youngest leaves are wilted and brown (see Fig. 1b). When older, damaged leaves display a wilted and brown margin. If the shoots are dissected, the larva can be found tunnelling the shoot axis, the new bud at the tip of the axis, or the petiole of a wilting leaf. The tunnelled petioles and buds easily break off, exposing brown faecal powder and spun silken threads, which reveal the cause of the damage even after the larva has left. Observations indicate that each larva may mine more than one shoot.

In Troms, full-grown A. retinella larvae spin their cocoons and pupate in the litter of the forest floor at the beginning of July. The ectoparasitoid Bracon intercessor Nees (Hym., Braconidae; det. K.-J. Heqvist) was found to parasitize A. retinella larvae from Tromsø.

In continental Europe, the moths fly in June and the first half of July (Friese 1969), in Estonia and southern Scandinavia, in (June) July and the first half of August (Petersen 1924; Svensson 1993). In Troms, moths were seen flying from the beginning of August to ≈10 September (in 1995). At Alta in Finnmark, one observation was made on 21 August (O. Tenow). Female moths preferably lay their yellow-white eggs, about 1 × 1·5 mm in size, under lichen crusts (Parmelia olivacea (L.) Nyl.) on trunks and branches of birch trees. In continental Europe, the eggs hatch before winter and the young larvae overwinter (Werner 1958). In contrast, north Norwegian A. retinella overwinters in the egg stage (F. Elverum & A.C. Nilssen, unpublished data). After egg hatch in spring, the young larvae move to the shoots where they start tunnelling.

A. retinella has been reported from all provinces of Sweden except for the northern half of Swedish Lapland (Svensson et al. 1987). It is also known from Norway and Finland although its distribution has not been mapped in detail (Svensson et al. 1987).

In our study, the origin of damage was determined from larvae (according to Werner 1958) and/or faeces and silken threads in attacked shoots, or by rearing individuals to imagines (F. Elverum & A.C. Nilssen, unpublished data).

Damaged trees and forests

Short shoots of the birch normally live for several years, but when attacked by A. retinella they often die because the preformed buds of the next year's leaf rosettes are destroyed. Long shoots are also tunnelled but survive and heal if vigorous. A swollen and sometimes cracked shoot base is then the only visible sign of the attack. Long shoots on branches weakened by repeated attacks become shortened and may die. Non-attacked shoots react by sprouting clusters of new long shoots (Fig. 1c). A birch forest, heavily attacked for the first time, looks brownish at a distance because of the many wilted leaf rosettes. Later, when dead shoots have accumulated over some years and the green shoots have become sparse, the forest has a darker impression due to the denuded parts of branches and twigs.

Degree and dynamics of attack

On 3–12 July 1993, 20–22 June and 2–13 July 1994, 19–21 June and 29 June–7 July 1995, and 25–30 June 1996, the amounts of damaged shoots (long and short shoots pooled) were quantified along a W–E transect (68 22′–68 45′N, 14 45′–19 45′E) and a N–S transect (69 57′–68 55′N, 23 05′–23 55′E; not in 1996) in northern Scandinavia. The selected localities (Table 1; Fig. 2) were sampled at 1–5 stations upslope; the altitudes were determined with a pocket altimeter. The number of stations per locality was adjusted to topography. At each station, 10 trees were selected within a 50 × 50 m-plot by using the sides of the plot as a coordinating system and choosing the birch tree nearest to each of 10 points, randomized within the plot. From each tree one twig in mid-crown was clipped with a pole pruner. For each sampled twig (carrying more than 100 living shoots), the number of damaged shoots was counted and expressed as a percentage of the total number of shoots.

Table 1.  Localities and altitudes for quantitative sampling of Argyresthia retinella damage on birch in northern Scandinavia in 1993–96 (qualitative observations within parentheses)
LocalityValley floor–
forest line (m a.s.l.)
Stations (m a.s.l.)Years
  • *

    Only in 1993.

  • N and S = north and south of the lake Vuolitjavri.

  • a and b = 150 and 50 cm above ground, i.e. above and below the snow cover as indicated by the height of ‘table’ birches.

  • §

    In 1995, Joatka was substituted by Suoluvuopmi within the same transect at about the same altitude and distance from the sea.

W–E transect
Melbu0–25020, 120, 235(1990–) 1993–96
Løbergsbukta0–32030, 110, 250(1991–) 1993–96
Gratangen0–480120, 230, 310, 390(1990–) 1993–96
Engmo130–420135, 200, 260, 350(1990–) 1993–96
Frihetsli220–500260, 360, 450(1991–) 1993–96
Abisko350–610(350*) 375, 410 (480, 600*)(1984–) 1993–96
N–S transect
Alta0–(285)10(1992–) 1993–95
Joatka N384–450400, 435(1990–) 1993–95§
Joatka S384–410a, 410b(1991–) 1993–95§
Maze270–500300, 360, 460(1990–) 1993–95
Suoppatroavvi325–(416)350(1990–) 1993–95
Figure 2.

Northern Fennoscandia with sampling localities in W–E () and N–S (•) transects.

In addition to these regular quantifications of the attacks in 1993–96, qualitative observations were made on A. retinella damage in 1990, 1991 and 1992.

Regional mapping of the attack

The geographical distribution of the attack was mapped (scale 1 : 250 000) during car drives within, between and beyond the two study transects in 1991–96 and its origin was confirmed at intermittent stops. Along the coast, the survey extended southward to Skjomen Fjord (10 km south of Narvik) and north-eastward to Tana River and further to the east to Vadsø on the Varanger peninsula (Fig. 3). Coast-to-inland drives were also made to Karasjok in Norway, and to Kilpisjärvi and Utsjoki in Finland. Upper limits of the damage were read from the pocket altimeter when passing. The distribution of damage with regard to exposure to the sun was mapped by recording aspects of slopes with damaged and undamaged forest along a 150-km stretch (Road 19) in the W–E transect on 11–12 September 1991.

Figure 3.

Outbreak area (hatched) of Argyresthia retinella on birch in northern Norway 1991–96. Tromsø climate station is indicated.

Structural changes in birch

At Melbu (Fig. 2), the forest was severely damaged in the upper part of the slope and undamaged in the lower part. At Løbergsbukta the situation was reversed. At these stations, structural changes in the birch foliage caused by the attack were studied. On 19 June and 3 July 1993, twigs were sampled at Løbergsbukta, one from each of 10 birch trees at each of three altitudinal stations. On the first date, the total number of living shoots (damaged and undamaged) per twig was counted and twig length estimated by measuring the main twig and all its ramifications >10 cm in length and expressing shoot density as the number of shoots per metre. On the second date, the distance between the widest spaced, neighbouring, living shoots on the same axis of each sampled twig was measured. Together, the two estimates indicate whether there had been an enhanced production of new shoots to compensate for the mortality of shoots on damaged twigs.

On 19 and 21 June 1995, at Løbergsbukta and Melbu, respectively, one twig from each of 10 birch trees was taken at the lowest and highest of the three altitudinal stations. The leaf-carrying part of each twig with its ramifications was clipped into pieces, 20 cm in length, and the number of living (damaged and undamaged) shoots per piece was counted for a comparison of the degree of clumping of shoots in undamaged and damaged forest.

Temperature climate

Strong gradients approximately perpendicular to the coast exist for many climatic parameters in northern Scandinavia, one of the most pronounced being winter air temperatures (cf. Johannessen 1970; Wallén 1970). Often low temperatures are the limiting factor for insect occurrence in northern regions (e.g. Bale 1993) and given the limited knowledge of response of A. retinella to environmental factors only air temperature in relation to the outbreak is treated in the following analysis.

The possible role of air temperature was investigated by using data from Tromsø (69 57′N, 18 57′E, 115 m a.s.l.), which is situated in the centre of the outbreak area (Fig. 3). Average monthly air temperatures and absolute monthly maximum and minimum temperatures were available for analysis. When comparing the time sequence of the observed outbreak with the average temperature variations, the length of the monthly periods was chosen to coincide as closely as possible with the observations of the insect stages in the field (see above): egg in diapause (October–February), egg in post-diapause (March, April), larva (May, June), pupa (July) and imago (August, September). For biological reasons (see Discussion), analyses of temperature data were restricted to the period corresponding to the egg in post-diapause, larva and imago stages.

Statistical tests

Before analyses of temporal and spatial differences in A. retinella attacks were made, data (percentages) were arcsine transformed for normality. Data were tested for significance with anovas. If significant, the anovas were complemented with Tukey's HSD test for multiple comparisons for differences between years, stations and altitudes.

Results

Regional extent of attack

Larvae of A. retinella caused visible damage to the forest from Melbu, farthest to the west, to Gratangen, on the mainland, in the W–E transect. Along the coast, from Narvik via Gratangen to Alta, the northernmost station in the N–S transect, Argyresthia-damaged forests were frequent (Fig. 3). No A. retinella damage was observed far inland or beyond Narvik to the south and beyond Leirbotn, 20 km NE of Alta, to the north-east (Fig. 3). Mostly the distribution of damage extended from sea level up to about 100 m, also in the extreme north-east (Leirbotn: 0–125 m a.s.l.), from where it gradually diminished upwards. Above 200 m a.s.l., most birch forests looked normally green and undamaged. No difference in attack intensity was discernible as to S, W, N and E aspects of damaged slopes.

Degree of attack

Assessment of damaged shoots at the different transect localities (Fig. 4) confirmed the results of the regional survey. In the W–E transect, the percentage of attacked shoots was high at the coast and low at the intermediate position (Melbu–Løbergsbukta contra Gratangen–Engmo: F = 252·54, d.f. = 1, P < 0·001) or nil inland (Frihetsli, Abisko). At three localities, the proportion of damaged shoots was highest in the lower parts of the slope (two-way anovas with year as cofactor): Løbergsbukta: F = 45·03, d.f. = 2, P < 0·001; Gratangen: F = 155·73, d.f. = 3, P < 0·001; Engmo: F = 5·13, d.f. = 3, P < 0·002; although at Engmo the intermediate altitude had the strongest attacks most years (Fig. 4a). At the divergent locality, Melbu farthest to the west, the attack intensity was highest close to the forest limit (230 vs. 120 and 20 m a.s.l: two-way anova with year as cofactor: F = 39·47, d.f. = 1, P < 0·001; Fig. 4a). In the N–S transect the highest incidence of damage was at the coast locality Alta (Fig. 4b). A. retinella was absent from the inland localities (Joatka, Maze, Suoppatroavvi) (Fig. 4b).

Figure 4.

Figure 4.

Degree of Argyresthia retinella damage at different altitudes in northern Norway 1993–96: (a) W–E transect (note different scales on x-axis); (b) N–S transect with longitudinal height profile of the Tverrelva valley. In 1995, Joatka was substituted by Suoluvuopmi in a similar regioclimate 24 km to the SW (cf. Fig. 2).

Figure 4.

Figure 4.

Degree of Argyresthia retinella damage at different altitudes in northern Norway 1993–96: (a) W–E transect (note different scales on x-axis); (b) N–S transect with longitudinal height profile of the Tverrelva valley. In 1995, Joatka was substituted by Suoluvuopmi in a similar regioclimate 24 km to the SW (cf. Fig. 2).

The most severe damage, >50% of the shoots wilting, was found in 1994 at the highest station at Melbu in the W–E transect (Fig. 4a). The next highest amount of damage appeared at Løbergsbukta in 1994 as >40% of the living shoots were attacked in the lower 110 m of the slope. At Alta, the degree of damage was >30% in 1993 (Fig. 4b).

Dynamics of attack

In Fig. 5, the degree of damage to birch is pooled for all altitudes at each locality in the W–E transect where A. retinella attacks occurred (cf. Figure 4a). The study covered an increase of damage from 1993, over a culmination and a decline to low levels again in 1996. The overall differences between years (i.e. Engmo, Gratangen, Løbergsbukta and Melbu combined) were significant (F = 62·03, d.f. = 3, P < 0·001, two-way anova with locality as cofactor). Post hoc pairwise comparisons (Tukey HSD) revealed that all years were significantly different from each other (P < 0·01), with 1994 as the peak year, see also Fig. 5. At Melbu, the forest appeared green and undamaged during the first visit in 1990 (photographic documentation). In 1991 and 1992, notations were made on wilting and brown leaves, and in 1993 the entire, upper part of the forest appeared brown (30% damage, cf. Figure 4a; also photographic documentation). The forest at Løbergsbukta already appeared blackish in 1991 and the attack by A. retinella had probably gone on for some years before 1991. In 1996, a regreening was evident. At Gratangen, attacked, wilting shoots were seen at the lowest station in 1991 (120 m a.s.l.). At Alta, in the N–S transect (Fig. 5), the degree of damage was high in 1993 and 1994 and decreased in 1995. Also here, a blackish appearance of the forest indicated a high intensity of attack prior to 1993. No estimate of damage was made in 1996 at Alta.

Figure 5.

Dynamics of Argyresthia retinella damage in northern Norway 1993–96. Data are pooled for all altitude stations at each locality (cf. Fig. 4).

Structural changes in birch

One measure of the accumulated damage is the sparsity of the foliage as expressed by the length of bare twig between leaf rosettes. At Løbergsbukta in 1993, the foliage sparsity was significantly higher (t-test: P < 0·001) in the lower than in the upper half of the forest (control), the average distance between widest spaced, neighbouring rosettes being 17 and 8 cm, respectively (Fig. 6a). In spite of that, the density of shoots was the same at all three stations, about 30 leafy shoots per metre of twig (Fig. 6b). Thus, in 1993 the birches still retained the capacity to compensate for shoots losses. However, under persistent attack the number of living shoots per length of twig decreased compared with undamaged twigs, as at Løbergsbukta 2 years later (Fig. 7b).

Figure 6.

Spacing (a) and density (b) of shoots on twigs at Løbergsbukta in 1993 in damaged (30 and 110 m a.s.l.) and undamaged (250 m a.s.l.) parts of the birch forest (cf. Fig. 4a). Mean ± 1 SE, and range are indicated.

Figure 7.

Observed and fitted negative binomial distribution of shoots per 20-cm branch length in 1995 at (a) Melbu and (b) Løbergsbukta, in undamaged and damaged parts of the birch forest. Note the different scales.

The compensation in shoot density of the damaged twigs combined with an increasing distance between shoots should imply an enhanced clumping of leafy shoots. This was illustrated by the study of shoot distributions per twig length at Løbergsbukta and Melbu (Fig. 7). Over the twig length used (20 cm), none of the distributions were significantly different from the corresponding negative binomial distribution, although the distributions were more clumped (k lower) where the forest was severely damaged over the years (Løbergsbukta lower, Melbu upper) than undamaged forest (Løbergsbukta upper, Melbu lower) (Fig. 7a,b).

There is no comparable information from the N–S transect. However, the birch trees at the Alta locality had sparse foliage similar to that at Løbergsbukta, and some birch trees in the valley of the river Tverrelva (Alta) seemed to die from A. retinella attacks (Fig. 1d).

Temperature climate

Figure 8 shows the yearly deviations from the long-term temperature average of the mean temperatures for the months of the occurrence of post-diapause egg, larva and adult stages. The scatter of the temperature deviations suggests division of the 1868–1996 period into two parts, the first from 1868 to 1919 and the second from 1920 onwards. The averages of the two parts differ by about 0·9 °C. The standard deviations for both series are similar, at 0·3–0·4 °C. During the period from 1920, the deviation of the 5-year running means comes close to the +1 °C limit on three occasions: in the middle of the 1930s, around 1960 and around 1990.

Figure 8.

Deviations of average air temperatures during months of the occurrence of post-diapause eggs, larvae and moths of Argyresthia retinella. Deviations are in relation to the long-term air temperature average (1868–1996) at Tromsø, northern Norway (cf. Fig. 3). When the value for each year (crosses) was calculated, the two periods (March–June, August–September) were given equal weight. This was done by calculating, for each period, the average deviation of individual months from the long-term averages and then averaging the two periods.

Given a stochastic distribution of the deviation points between 1920 to present (Fig. 8), the expected frequency for attaining the level of +1 °C for a 5-year period is about five times per century (which broadly matches the three periods observed in 75 years).

Discussion

Outbreak distribution

The distribution of the A. retinella damage was clearly coastal, i.e. with steeply increasing intensity westward and northward along the two transects and reducing with altitude at each locality, except for Melbu. Gratangen is located at the bottom of the Gratangen Fjord, and Engmo in the inland valley of Salangsdalen. The generally lower occurrence of A. retinella at Gratangen and Engmo than at the extreme coastal Melbu, Løbergsbukta and Alta (Fig. 5) may depend on shorter flight periods in autumn and/or later egg hatch in spring at the former localities (cf. Nasjonalatlas for Norge 1993). In addition, the low degree of attack at the lowermost stations at Engmo may indicate a lethal effect on the overwintering egg population of a cold air ‘lake’, accumulated over the valley floor, similar to that described for egg populations of Epirrita autumnata (Bkh.) (Tenow 1975; Tenow & Nilssen 1990). In the system of outbreak areas of A. retinella, Operophtera brumata L. and Epirrita autumnata, parallel to the coast, the A. retinella outbreak area was the most maritime (Tenow 1996). Why the vertical distribution of damage was reversed at Melbu is unknown. However, Melbu is in the extreme maritime part of the W–E transect where climatic conditions possibly allow life cycles and, hence damage patterns, to be different from those further inland.

Population dynamics

Each A. retinella larva probably mines more than one shoot during its development. A sample too early in the phenology may therefore underestimate the damage. However, because sampling was performed when most larvae were in late instars or even full-grown, this error should have been small.

The quantitative presence of A. retinella was expressed in percentage damaged shoots per year. At the beginning of an attack period, and as long as shoot mortality is compensated for by an enhanced production of new shoots, percentage damage should be approximately proportional to the number of larvae per foliage unit. However, as damage accumulates over years and shoot mortality comes to dominate, proportionality will disappear. Then, even if the degree of damage one year is higher than the year before, larvae may be fewer. Thus, it is not possible to state with certainty when a larval population peaked, but this may have occurred some years prior to the damage peak. Judged from the appearance of forests at the first visits in 1990, populations appeared to increase mainly from the last years of the 1980s. This is supported by reports on damage since 1989, resembling that caused by A. retinella (Johansen & Kobro 1996). After culminating in the first half of the 1990s, populations crashed in 1995–96. Contributing to this could have been parasitization by Bracon intercessor (see above) which locally killed >60% of the larvae at Tromsø in 1995 (F. Elverum & A.C. Nilssen, unpublished data).

Foliage structure

Generally, shoot distribution over twig length was more clumped in lower (damaged) parts than in higher (undamaged) parts of the forested slopes (Figs 1c and 9b). Thus, differential foliage structure could be a consequence of altitude rather than of damage by the insect. However, the association of enhanced clumping at Melbu and Løbergsbukta with the highest incidence of A. retinella damage at high and low altitude, respectively, clearly points to the damage as the main cause of clumping.

Figure 9.

(a) Damaged birch trees in the Narvik region on 28 May 1940. In the background is a British warship which dates the photograph. Beyond are the Taraldsvik mountains (photo supplied by Per Ilsaas). (b) Birch twigs damaged by Argyresthia retinella in 1993 (photo supplied by Tor Johansen). (c) Birch trees damaged by A. retinella, Løbergsbukta, Gullesfjord in Troms, Norway, 9 July 1991.

In polycormic birches, defoliation of current leaves triggers the growth of basal sprouts and the individual trees usually survive (e.g. Kallio & Lehtonen 1973). The A. retinella larvae kill the next year's bud in addition to current leaves and thereby destroy the basis for future foliage. Furthermore, attacks are on coastal birch, which are mainly monocormic with a lower suckering capacity and therefore may have reduced recovery capacity (cf. Kallio & Lehtonen 1973).

Earlier outbreaks?

Only a single outbreak of A. retinella has been reported and studied (this paper; Johansen & Kobro 1996; cf., e.g. Werner 1958; Friese 1969; Agassiz 1987). Was this outbreak unique in northern Norway (and elsewhere)? A photograph taken on 28 May 1940 in the Narvik region (Fig. 9) shows birch crowns and branches with severe damage very similar to that inflicted by A. retinella. The damage is different from that caused by geometrid caterpillars. In addition, the photo was taken too early in the summer for severe O. brumata or E. autumnata defoliations to have developed. ‘Frost’ damage, if ever occurring at the coast, primarily kills shoots on the tips of twigs and branches and in the top of the birch trees (Tenow 1996). Thus, this photo is the only indication of an earlier outbreak of A. retinella. If there was an outbreak in the years preceding 1940, the lack of reports may be due to its confounding with damage inflicted during O. brumata outbreaks which have occurred regularly along the coast, e.g. in the 1930s (Tenow 1972).

Role of climate

A. retinella occurs in most provinces of Fennoscandia. The restriction of the outbreak to an area with a marked maritime climate points to a climatic dependence of the population increase. All life stages may be involved in such a relationship. However, the change from larval overwintering in southern areas, where no outbreaks have been reported, to egg overwintering in the outbreak area suggests a possible connection between the outbreak and the change in overwintering strategy.

For biological reasons, life stages of the insect may vary in sensitivity to air temperature variations. The mean supercooling point (SCP) of A. retinella eggs in midwinter is ≈ −36 °C (F. Elverum & A.C. Nilssen, unpublished data). Such low temperatures never occur within the outbreak area. Extreme minimum temperatures are therefore not expected to be an important mortality factor for A. retinella eggs. However, many insects cannot withstand prolonged periods of winter cold even at temperatures above their SCP (Bale 1993). Elevated winter temperatures may imply shorter periods of cold which would increase survival and, hence, may be a releasing factor of outbreaks. To highlight this, laboratory experiments were performed with A. retinella eggs that had completed diapause but still retained the midwinter supercooling capacity (F. Elverum & A.C. Nilssen, unpublished data). After 14 and 10 days at −27 °C, 100 and 50% of the eggs had died, respectively. Similarly, after 21 and 15 days at −18 °C, 100 and 50% of the eggs died, respectively. Cold spells of these temperatures and durations frequently occur inland and may explain the absence or very sparse occurrence of A. retinella at Abisko, Frihetsli, Joatka, Maze and Suoppatroavvi. However, such cold spells are either exceptional or they do not occur within the outbreak area. Thus, the diapause egg will not be a critical stage in coastal areas. Similarly, the pupal stage, protected in the ground litter, may also not be critical. In addition, during the pupal stage temperatures were lower in the 1990s than in many of the preceding periods. The post-diapause egg, larva (benefiting from an early eclosure) and adult (egg-laying) stages, occurring during the periods March–June and August–September, became the combination for which temperature deviations were analysed.

In Fig. 8, the highest temperature deviation peaks are for the 1930s and around 1990. The A. retinella populations of the 1990s evidently built up contemporarily with the latter peak and culminated a few years after it. Thus, this peak may be related to the 1990s’A. retinella outbreak and the former peak to the possible outbreak before 1940.

Thus, if no marked change in the temperature climate occurs in the future, one may expect that outbreaks of A. retinella will occur from time to time in the region. If there is a lowering trend in temperature, the chance of outbreaks would diminish on the present assumptions. However, many General Circulation Models predict a marked increase of surface temperatures at high latitudes in winter (at a doubling of anthropogenic CO2 atmospheric concentrations: Schneider 1993; European Climate Support Network 1995). The effect on such insect populations of a future increasing trend of air temperature, natural or anthropogenic, is more difficult to predict. Initially, outbreaks may become more frequent and severe. Outbreaks may shift area. Several unknown influences might be involved, related to parasitoids, etc.

In conclusion, the coincidence of temperature peaks for the three life stages (the post-diapause egg, larva and adult) and the A. retinella outbreak in the 1990s is suggested as a possible explanation for a causative climatic factor. Additional evidence of an outbreak in the 1930s is required as well as further experiments on, for example, cold hardiness of eggs in spring, for confirmation.

Recommendations

The most important result of this study is the insight that previously innocuous forest insects may become serious outbreak species if climate changes. Thus, in addition to well-known outbreak species, one should be prepared to meet new ones. Once recognized, research on the biology and outbreak ecology of a new insect pest is a prerequisite for attaining control means. For A. retinella, the present study may serve as a starting point. The concentration of damage to coastal birch forests in northern Fennoscandia lends itself to monitoring. To begin with, therefore, it is recommended that foresters learn to identify and include observations on A. retinella damage in their annual reports.

Acknowledgements

Britta, Christian and Lina Tenow assisted in the field and in the laboratory. The Royal Swedish Academy of Sciences and Abisko Scientific Research Station offered lodging and working facilities. P. Ilsaas and T. Johansen kindly put photographs at our disposal. J. Bale, W. Block and an anonymous referee improved the manuscript. Our investigation was funded by the Swedish Environmental Protection Agency (project contract no. 27312, O. Tenow & A.C. Nilssen). We thank all these persons and institutions.

Received 17 November 1997; revision received 14 December 1998

Ancillary