Waves and synchrony in Epirrita autumnata /Operophtera brumata outbreaks. II. Sunspot activity cannot explain cyclic outbreaks

Authors


A. C. Nilssen, Zoology Department, Tromsø Museum, University of Tromsø, N-9037 Tromsø, Norway. E-mail: arnec.nilssen@tmu.uit.no

Summary

  • 1In recent studies, it has been argued that sunspot activity forces the Epirrita autumnata 9–10-year outbreak periodicity in the mountain birch forest of Fennoscandia. For the following reasons, we challenge this conclusion.
  • 2With a 10-year outbreak cycle of E. autumnata and the 11-year sunspot cycle, it is expected that the cycles will run in-phase, out-of-phase and in-phase within 10 × 11 years. Hence, given such cycle lengths, sunspot activity should not affect outbreak periods. For a test, the E. autumnata series should be at least 110 years in length.
  • 3A well-documented E. autumnata outbreak series of 81 years (1888–1968; outbreak periods IV–XII) exists. This series is here lengthened to 114 years by adding outbreak frequencies for three decades (1969–2001).
  • 4By lengthening the series, three more E. autumnata/Operophtera brumata periods (XIII, XIV, XV) are identified. Period XV, like several earlier periods, was of the moving type, i.e. outbreaks moved in a wavelike manner from northern Fennoscandia to southern Norway.
  • 5As with several earlier outbreak periods in central northern Fennoscandia, the main timing of periods XIII–XV centred at the middle of the decades. In contrast, outbreaks at the extreme north-western coast of Norway centred at the decadal shifts, i.e. about 1979, 1989 and 1999. Supported by historical documents, we explain the 1979 and 1999 outbreaks as the final expressions of east–west outbreak waves that branched off from the main waves which moved southward during periods XIII and XV. These side-waves in the north are new observations. Outbreaks at the decadal shift 1989/1990 may have been of a more complex nature.
  • 6We find that sunspot activity does not explain outbreak waves. Furthermore, a test of our 114-year long E. autumnata series against the contemporaneous sunspot series shows that the two series run in-phase and out-of-phase. The observed interval between the two cycles coming in-phase agrees with the expected interval. This challenges the hypothesis of sunspot synchronization of the E. autumnata (and O. brumata) outbreaks.

Introduction

Large-scale spatial synchrony in population density fluctuations has been described for several forest insects with periodic outbreaks (for a survey, see Myers 1998). Synchrony here means that local outbreaks in a region gather into periods separated by periods without outbreaks (e.g. Myers 1998). Spatial synchrony may also occur among guild members, as Klimetzek (1979, 1990) has shown for insects feeding on Scots pine (Pinus sylvestris L.) needles. This synchronization occurs over distances of several hundred kilometres up to 1200 km (Liebhold, Koenig & Bjørnstad 2004; Johnson et al. 2005). Operophtera brumata populations, for example, have fluctuated in synchrony over England, Norway and Germany in the latter half of the 20th century (Roland 1998). These patterns should be due to an equally widespread synchronizing force. As such, sunspot cycles have been suggested as an ultimate force (first for forest insects: Eidmann 1931) and were found by Myers (1998) to have a significant association with insect outbreaks. Proximate synchronizing effects, elicited by sunspot activity, may be associated with climate and weather or indirectly, e.g. through changes in the quality of hosts plants as food for insects and/or mobile natural enemies (e.g. Ruohomäki et al. 2000; Haukioja 2005).

Historical data on Epirrita autumnata (Bkh.) and O. brumata L. outbreaks on birch (Betula pubescens Ehrh.) forests in the Fennoscandian mountain chain (the Scandes) have been reviewed for the period 1862–1968 (Tenow 1972). They revealed 12 outbreak periods (periods I–XII) which occurred approximately every 9–10th year on an entire Fennoscandian scale. During some periods, the local outbreaks occurred more or less contemporaneously from northern Norway, Sweden and Finland via the Scandes to southern Norway. During other periods, they were of the ‘moving type’ (Tenow 1972), i.e. they moved over years like a wave from north to south, or the reverse. However, the nature of these wave-like ‘moves’ cannot be specified.

Periodicity in E. autumnata outbreaks in northern Fennoscandia (> 65°30′ N) has been cyclic (Haukioja et al. 1988; Neuvonen 1988) and regional synchronization of local population density peaks has been ascribed to an association with variation in sunspot activity (Myers 1998; Ruohomäki et al. 2000; Selås et al. 2004). This notion seems to have become an established fact (Haukioja 2005; Økland & Kobro 2005). We lengthen an existing time-series on outbreak frequencies extending from 1888 to 1968 (Tenow 1972) to also include 1969–2001 and test this proposed associate with sunspot cycles on the scale of the whole of Fennoscandia. The outbreaks of O. brumata seem to be coupled with those of E. autumnata with a delay of 1–2 years (Tenow et al. 2007), and the results may also apply to O. brumata.

From the aspect of sunspot forcing, it is of interest that outbreaks have sometimes moved like a wave from north to south (e.g. outbreak periods V and VII) or the reverse (period X), and sometimes in northern Fennoscandia from east to west (Tenow et al. 2007). We therefore complement the previous survey (Tenow 1972) by mapping the spatio-temporal patterns of local outbreaks during 1969–2001. Conclusions in previous studies on sunspot influence are discussed in the light of presented results.

Materials and methods

the scandes

The Scandes has a NNE–SSW extension of approximately 1700 km (71° N −58° N), in the north with midnight sun in summer and polar night in winter and a vegetative season (daily air mean temperatures = +3 °C) about 2 months shorter than in the south (Wallén 1960). Being a sharp climatic divide, the Scandes causes a maritime climate on its western (Atlantic) side and an intermediate to continental climate on its eastern side. Average winter air temperatures vary from about ± 0°C at the Norwegian coast to about −15 °C in inland areas (Wallén 1960). One of the coldest regions is the Finnmarksvidda highland plain (200 × 400 km) in north-eastern Norway, with winter air temperatures usually falling below −40 °C.

species

Moths of E. autumnata and O. brumata mate and lay eggs in autumn. The eggs are placed singly on branches and twigs of the birch exposed to the extremes of weather. Both sexes of E. autumnata can fly, whereas in O. brumata the female moth has stunted wings and must therefore climb the tree for oviposition (Tenow 1972). The eggs of both species overwinter and hatch at budburst in spring, when the folivorous caterpillars start feeding. When fully grown in July, the caterpillars leave the trees to pupate in the litter. The pupal stage of both species lasts until autumn, when the moths appear and start egg-laying (Haukioja et al. 1988; Peterson & Nilssen 1998). Both species are found over most of Fennoscandia but the distribution of their outbreaks differs, O. brumata's along the western side of the Scandes and E. autumnata's mainly along the eastern side.

historical documentation

Reports on outbreaks and field data on population density peaks (for a definition, see Tenow et al. 2007) were used to determine outbreak periods. Information on outbreaks in Norway during 1969–87 were gathered from reports published in Anonymous 1971–89, Oslo, from unpublished Norwegian reports including samples sent to the Norwegian Institute for Forest Research (NISK) and for 1988–2000 from http://www.nisk.no/skogskade/report. Interviews with residents in the counties of Troms and Nordland, NW Norway, in 1990 (A.C. Nilssen & K. Engelskjøn) added information about the late 1980s. From the 1970s, the detailed annual reporting by Swedish forest officers in the surveyed area, important for exploring the time-period 1862–1968 (see Tenow 1972), gradually ceased. Instead, information for Sweden was collected through personal communications with biologists and earth scientists, tourist guides and national wildlife guards, visiting in the outbreak area. Observations by the authors and information in newspapers provided complementary data. An inquiry sent out to Norwegian and Swedish forest officers in 1976 (O. Tenow, unpublished data) provided additional information for the 1970s. Outbreak and peak years have been surveyed thoroughly by Finnish ecologists working in the outbreak area during the studied time-period. Information for Finland has therefore been taken exclusively from published Finnish studies. For uncertainties in historical documents, see Tenow (1972) and Neuvonen, Bylund & Tømmervik (2005).

The methods used to determine outbreak periods in the 1970s, 1980s and 1990s were the same as for the time-period 1862–1968 (cf. Tenow 1972). First, to elucidate the spatio-temporal extent of outbreaks in 1969–2001, reported outbreak localities were plotted (collectively for species) in the same time–latitude diagram complemented by a similar time–longitude diagram. In a second analysis, outbreaks of E. autumnata only were quantified in a frequency diagram for 1969–2001. This was conducted by marking reported outbreak/peak localities (and years) on a map of Fennoscandia covered with a grid of squares with the side = 1° of latitude (= 111 km) and counting attacked squares per year.

The historical documentation is available for consultation at http://www.blackwell-synergy.com. In this material, reports on outbreaks are listed separately for E. autumnata, O. brumata and unidentified geometrids.

outbreaks and sunspot activity

It is relevant for comparison with sunspot frequencies that if an approximately 10-year animal population cycle, e.g. that of E. autumnata, runs unaffected by and alongside the approximately 11-year sunspot cycle, the two cycles will run from in-phase via out-of-phase to in-phase again (cf. Hoyt & Schatten 1997) within about 10 × 11 years. Thus, to test whether such shifts occur or not, the animal series should be at least 110 years in length. Of the outbreak periods I–XII, periods I–III (1862–1884) are known only fragmentally (Tenow 1972). The remaining better-known periods IV–XII span 81 years (1888–1968). In the present paper, we lengthen this E. autumnata series by surveying documents and adding information on outbreaks during the period 1969–2001. This completed outbreak series, 114 years in length, should be sufficient for a comparison with the contemporary sunspot series (http://www.wdcb.ru/stp/data/solar.act/sunspot/YEARLY).

Results

outbreak periods 1969–2001

In Fig. 1a,b, the different outbreak/peak localities are plotted in a time–latitude and a time–longitude coordinating system. In northern Fennoscandia, local outbreaks appear to have been distributed randomly over time. Spatio-temporally, however, two main categories of local outbreaks seem to have existed, one with outbreaks (or peaks) in inland areas (including inner parts of fjords) around the middle of decades, and one at decadal shifts with outbreaks on the Ofoten–Vesterålen–Lofoten island chain and along the north-westernmost coast (68°−70° N/14°−19° E). In Fig. 1, outbreaks on the island chain and along the NW coast are marked under their own symbol.

Figure 1.

(a, b) Outbreaks of Epirrita autumnata, Operophtera brumata and unidentified geometrids (collectively) on birch in the Scandes and neighbouring regions in 1971–2001, placed (a) in a time–latitude and (b) in a time–longitude coordinating system. Outbreaks in extreme NW position are marked (∼).

If the outbreaks at the extreme north-west at decadal shifts are excluded, inland outbreaks gather into three separate periods for the entire Scandes and northern Finland (Fig. 1). The first period began in the same year (1972–73) in northern Fennoscandia and southern Norway (Fig. 1a). In spite of that, the period seems mainly to have progressed from north to south from about 1974–77 (regression analysis: Y = −0·703X + 1453·3; R2 = 0·089; P = 0·073; n = 38). The second period occurred more or less contemporaneously all over the outbreak area in about 1984–87. The third period moved as a wave from north to south from about 1992–2001 (Y = −0·795X + 1653·4; R2 = 0·28; P < 0·0001; n = 84). Due to the few reports from the middle part of the Scandes, this outbreak wave seems divided into one northern and one southern part. However, in its east–west progress (Fig. 1b) the period is coherent. On the other hand, the first and second periods have a gap in records in an east–west direction (26°−22° E).

The three outbreak periods are also evident in the number of 111 × 111-km squares attacked by E. autumnata each year (Fig. 2), although the frequency distribution for the 1992–2001 period is bimodal due to the few records from the middle part of the Scandes.

Figure 2.

The frequency of 111 × 111-km squares attacked per year by E. autumnata in the entire Scandes and northern Fennoscandia in 1971–2001.

The three outbreak periods, centring around the middle of the 1970s, 1980s and 1990s, link up well with previously documented periods (Tenow 1972) and are, accordingly, numbered XIII–XV.

outbreaks and sunspot activity

The hypothesis that sunspot activity is driving the E. autumnata outbreak cycle implies that outbreaks should coincide consistently over time with either sunspot minima or maxima, or with either of the two transitional phases between these extremes. However, this has not been the case. In Fig. 3, the frequency curve for outbreaks in the entire Scandes and northern Finland during 1888–2001 is compared with the sunspot curve for the same time-period. The two curves appear to have been out-of-phase in the last decades of the 19th century; thereafter they were significantly in-phase up to approximately 1930, when they drifted gradually relative to each other until the two curves again swung significantly out-of-phase during the last four decades of the 20th century (Pearson's r tests, see figure legends).

Figure 3.

Sunspot fluctuations and the frequency of Epirrita autumnata outbreaks in 1888–2002. On top of the diagram Pearson's correlation (r) coefficient is plotted for each maximum and minimum of sunspot activity. Correlations between yearly sunspot number and number of attacked 111 × 111-km squares were tested for each sunspot cycle (from minimum to minimum as well as from maximum to maximum) along the time-line; n = 10–13 for each individual test. Correlations are significant (•) at r = ±0·6. At the bottom of the diagram, the numbers of outbreak periods are given.

The average interval between sunspot maxima was significantly larger than that between E. autumnata maxima, 10·7 (SE = 0·3; n = 10) and 9·5 (SE = 0·3; n = 11) years, respectively (t-test: t = 2·723; P = 0·014), as measured between the highest peak of each cycle. According to the independence of the E. autumnata curve, the two curves should have run from one phase to the next same phase in 10·7 × 9·5 = 102 (compound SE = 6; Hansen, Hurwitz & Madow 1953) years (see ‘Outbreaks and sunspot activity’) over the studied time-period. In Fig. 3, the two curves seem to begin running in-phase in about 1900 and again in about 2000, i.e. at an interval of about 100 years, in good agreement with the expected interval. The mean interval (9·5 years) between outbreak periods in the extended series of periods does not differ from a previous calculation (Tenow 1972).

Discussion

outbreak waves in northern fennoscandia

In northern Fennoscandia, the combined outbreak areas of E. autumnata and O. brumata has a wide east–west extension, i.e. from north-eastern Fennoscandia, over northern Norway and Sweden to north-western Norway, where it reaches into the Ofoten–Vesterålen–Lofoten island chain, i.e. the extreme western offset of northern Scandes. It has been shown (Tenow et al. 2007) that the interperiodic outbreaks in north-western Norway in 1998–2000 were the westernmost endpoints of an outbreak wave moving along this stretch and which branched off from the main wave moving down the Scandes during period XV.

Can the interperiodic outbreaks around 1979 and 1989 be explained in the same way? In midwinter, the average E. autumnata egg freezes at −36 °C (Nilssen & Tenow 1990), which has prevented outbreaks on the cold Finnmarksvidda and in northernmost Sweden (Tenow 1972). In 1987–93, however, the winters were unusually mild in northern Norway (Vaer and Klima, 1989–1995), and hence outbreaks had time to develop on the Finnmarksvidda during period XV (Tenow 1996). This made it possible to track the east–west wave of outbreaks (or population peaks) from north-eastern Fennoscandia via the Finnmarksvidda to north-westernmost Norway (Tenow et al. 2007). In contrast, winters during periods XIII and XIV were normally cold, with egg-killing air temperatures Anonymous 1979–87. Hence, outbreaks could not develop and be tracked on the Finnmarksvidda (cf. lack of reported outbreaks between 22° and 26° E in Fig. 1b). However, with the documented east–west wave in the 1990s (Tenow et al. 2007) as a representation, we interpret the interperiodic outbreaks around 1979 mainly as the outer ripples of a similar western side-wave branched off from the main outbreak wave down the Scandes. In the 1980s, the spatio-temporal pattern differed, i.e. outbreaks at the north-westernmost coast occurred not only at the decadal shift (1988–89), but also in the middle of the decade (1985–86). Presumably, the 1988–89 outbreaks were locally delayed outbreaks or the result of a prolonged outbreak period over most of the north-western coastal area or, even, local populations independent from a common periodicity (see Ims, Yoccoz & Hagen 2004).

synchrony over the scandes

The Scandes and neighbouring areas offer a great variety of living conditions for E. autumnata, O. brumata and other birch-defoliating geometrids (see Materials and methods). Despite these differences, historical documentation shows that outbreaks have been synchronized in distinct outbreak periods along the entire range of the Scandes, and in northern Scandes approximately in the middle of each decade during the period 1895–1968 (outbreak periods V–XII: Tenow 1972). Three types of E. autumnata/O. brumata outbreaks were recognized (see Introduction). Outbreak periods XIII–XV were congruent with this recognition of types. Of these, the ‘moving type’ seems similar to what has been named recently ‘travelling waves’, e.g. outbreak waves of the lepidopteran Zeiraphera diniana Gn. in the European Alps (Bjørnstad et al. 2002).

We could show that the interperiodic local outbreaks at the north-western coast of Norway in 1998–2000 belonged to outbreak period XV (Tenow et al. 2007). Consequently, the interperiodic outbreaks around 1979 and 1989 in the same area should have belonged to periods XIII and XIV, respectively. In contrast, none of the few interperiodic local outbreaks south of about 67° N (Fig. 1) can be referred to any outbreak period, preceding or succeeding it.

outbreak waves and sunspot activity

Are outbreak periods of the moving type compatible with forcing by sunspot activity? None of the studies claiming a forcing of E. autumnata outbreaks by sunspot activity (see Introduction) has posed this question. This may be explained by the fact that these studies were either world-wide surveys concerning several defoliating insects (including E. autumnata: Myers 1998) or studies restricted to parts of the Scandes. The study of Ruohomäki et al. (2000) was limited to outbreaks north of 64° N and that of Selås et al. (2004) to population changes at one single locality, Budal, in central Norway. However, the compatibility of moving population densities and sunspots has been discussed by Williams (1954) concerning the 10-year periodicity in numbers of North American grouse. He argued that if there is an extra-terrestrial forcing, there should be no interregional time lag (of 4–7 years) as there appears to be in grouse occurrence between the Maritime Provinces, the rest of Canada, and Alaska. We agree with this conclusion and apply it to E. autumnata: if there is sunspot forcing of E. autumnata numbers there should be no waves from north to south (or the reverse) in outbreaks as happened, for instance, during period XV. Further, in cases when local outbreaks occur synchronized along the Scandes about mid-decade (in theory possibly synchronized by sunspot activity) there should be no delayed, prolonged or isolated outbreaks in the north at the subsequent decadal shift, as happened during period XIV.

outbreak periods and sunspot activity

We have shown that a data set covering at least 110 years is needed for a reliable test on the dependence of E. autumnata outbreaks on sunspot activity (see ‘Outbreaks and sunspot activity’). During the time-period 1888–2001, i.e. 114 years, outbreaks of E. autumnata were out-of-phase, in-phase and out-of-phase with sunspot numbers at the scale of the entire Scandes (Fig. 3). In addition, the observed interval between two successive in-phase positions of the two cycles, about 100 years, agrees with the expected interval. This incompatibility of the sunspot cycle with the E. autumnata outbreak cycle indicates that the E. autumnata (and probably O. brumata) outbreak periods are not related to sunspot activity. A similar incompatibility of the sunspot series with the 9–10-year cyclic Canadian lynx (Lynx canadensis Desm.) series has been demonstrated (Lindström, Kokko & Ranta 1996; see also Elton & Nicholson 1942: 242).

This observation is at variance with the results of Selås et al. (2004) and Ruohomäki et al. (2000). Selås et al. (2004) compared an E. autumnata data set from Budal (1972–2000) with the four latest sunspot cycles (1971–2000). As in our study, they found a strong correlation between caterpillar maxima and solar activity minima during this (relatively) short time-period. A causal connection was also suggested. During sunspot minima, ultraviolet-B (UV-B) radiation is maximal and the metabolic costs of producing UV-B protective pigments in leaves should reduce the resistance of trees to herbivores.

Ruohomäki et al. (2000) analysed the same set of 12 E. autumnata outbreak periods as in the present study (but with non-quantitative data on E. autumnata occurrence) against the same suite of 11 sunspot cycles. They found that the first year of the 12 outbreak periods were associated significantly with the decline phases of sunspot activity, but causal connections remained unclear. We interpret the results of Ruohomäki et al. (2000) so that the outbreak periods coincided more or less with sunspot minima, i.e. in essence the same conclusion as that of Selås et al. (2004). Because the surveyed time-period (1885–1999) embraced twice as many outbreak starts during sunspot decrease phases as during increase phases (cf. Ruohomäki et al. 2000; Fig. 3), the test method used (G2-test) gave a significant, but spurious, relation between outbreak periods and sunspot minima. If outbreak starts are ordered chronologically, the temporal shifts of the first year of outbreaks will parallel almost exactly the out-of-phase, in-phase and out-of-phase shifts of outbreak and solar cycles over time, as presented here (cf. Pearson's r values in Fig. 3). Hence, outbreaks > 64° N, such as the outbreak over the entire Scandes, seem not to have been forced by sunspot activity.

Eidmann (1931) compared sunspot frequencies and outbreaks of folivorous Lepidoptera on Scots pine in Germany during the period 1830–1925. He concluded that sunspot maxima, particularly since 1870, coincided with outbreak maxima. Myers (1998) analysed the same relation for the same type of insects from a diverse of trees (including conifers) around the Northern Hemisphere. She surveyed the period from the 1930s to the 1990s. Her conclusion was the reverse of Eidmann's, i.e. outbreaks occur during the troughs of sunspot frequencies. In these cases, opposite results were reached depending on which time-period was considered.

Thus, shifts between phases have occurred as predicted for cycles of different periods. Besides, the E. autumnata long-term population fluctuation does not show any apparent sign corresponding to the intermittent periods of phase lock in the 10-year cycle of the Canadian snowshoe hare (Lepus americanus Erxl.), which may indicate a synchronization by sunspot forcing (see Sinclair et al. 1993). Rather, the E. autumnata cycle has drifted continuously (Fig. 3).

A synchronization by sunspot activity demands a climatic link to biosystems. One important cause for interannual and perhaps interdecadal climate variability in Europe, especially during boreal winters, is the North Atlantic Oscillation (NAO), an index of the strength of the westerly airflow across the sea (Hurrell & Dickson 2004). However, the NAO does not relate to the number of sunspots (Selås et al. 2004). It covarys with spring air temperatures in Norway (Selås et al. 2004), but has small or insignificant effects on budburst in B. pubescens (Post & Stenseth 1999). This may imply little or no influence on optimal timing of budburst with egg hatch and, hence, little or no influence on the performance of neonate caterpillars in spring. On the other hand, eggs freeze during cold winters, these in turn being correlated with a low NAO (Klemola et al. 2003; cf. Yoccoz et al. 2002).

Altogether, sunspot activity seems not to be a synchronizer of E. autumnata (and O. brumata) cycles, but NAO could be a candidate. The most commonly suggested mechanisms to explain spatial synchrony include dispersal and regional stochasticity, i.e. the ‘Moran effect’ (Peltonen et al. 2002). Dispersal of E. autumnata and O. brumata seems less likely (Tenow et al. 2007). In addition, interspecific synchrony cannot be explained by dispersal, while weather which is correlated over large areas clearly could (Liebhold et al. 2004 and references therein), possibly in interaction with density-dependent dynamics of similar kind among populations.

Selås et al. (2004) asked: can sunspot activity explain cyclic outbreaks of forest moth pest species? In some systems there may be a significant congruence between population fluctuations and sunspot numbers. That was demonstrated by Klvana et al. (2004) for a 130-year series of indices (feeding scars on tree trunks) of porcupine (Erethizon dorsatum L.) occurrence in a particular part of the Canadian conifer forest region (cf. the snowshoe hare cycle: Sinclair et al. 1993; Selås 2006) Solar–climatic mechanisms are complex and still unclear (e.g. Raspopov, Dergachev & Kolström 2004). A definitive answer must therefore wait. However, according to the data available we answer: sunspot activity cannot explain cyclic outbreaks of E. autumnata and O. brumata.

Acknowledgements

E. Christiansen kindly gave us information on outbreaks in Norway collected by the Norwegian Institute for Forest Reseach (NISK) in the 1970s and 1980s and K. Engelskjøn explored outbreak localities in northern Norway in 1989. Several other people in Norway, Sweden and Finland informed about individual local outbreaks. M. Ovhed supported with climate data and D. Sundkvist informed on ‘space weather’. We thank these people and institutions. Part of the study was funded by the Swedish Environmental Protection Agency (project contract no. 27312, O. Tenow & A. C. Nilssen).

Ancillary