1624

Ecological Applications, 14(6), 2004, pp. 1624–1633q 2004 by the Ecological Society of America

HARVESTING DISRUPTS BIOLOGICAL CONTROL OF HERBIVORES IN ASHORT-ROTATION COPPICE SYSTEM

CHRISTER BJORKMAN,1 RICCARDO BOMMARCO, KARIN EKLUND, AND SOLVEIG HOGLUND

Department of Entomology, Swedish University of Agricultural Sciences, P.O. Box 7044, SE-750 07 Uppsala, Sweden

Abstract. Disturbances such as harvesting often interfere with the ecological processesthat lead to the biological control of insect pests. For willows, grown as short-rotationcoppice crops harvested every third to fifth year, it has been suggested that high plantquality in the resprouting shoots after harvesting may explain observed high densities ofherbivorous insects, especially leaf beetles (Coleoptera: Chrysomelidae), in the plantations.In this study, we show that generalist predators may be important as regulators of leafbeetle populations. All three leaf beetle species, which were studied for five years in 12plantations, showed a negative correlation between the population growth rate from springto fall and the abundance of the most common generalist predator Orthotylus marginalis(Heteroptera: Miridae). For the most abundant leaf beetle, Phratora vulgatissima, we alsofound a significant positive correlation between its population growth rate and egg survival,indicating an overall effect of predation on herbivore population growth. Harvesting, whichtakes place during the winter, had a negative effect on the abundance of leaf beetles andpredators. However, the first year after harvesting, all three leaf beetle species regainedthis loss with a very high population growth rate. A reason for the better ability of theherbivores to recover from the disturbance may be that they, unlike the predators, mainlyoverwinter outside the plantations. All three leaf beetles peaked in density three years afterharvesting whereas the density of generalist natural enemies increased or leveled off duringthe five-year period after harvesting. We conclude that predation by generalist predators ispotentially important for population control of leaf beetles in willow coppice, but that theintermediate disturbance regime of around five years between harvests, appears to be tooshort to avoid disruption of biological control. Alternatives for more efficient biologicalcontrol in short-rotation coppice systems may be a longer period between harvests thatenables the predators to fully respond numerically, to leave natural enemies refuges atharvest, or to harvest adjacent plantations asynchronously.

Key words: Anthocoridae; biological control; disturbance; generalist predators; leaf beetles;Miridae; Orthotylus marginalis; Phratora vulgatissima; recovery after harvesting; Salix viminalis;willow coppice; willows.

INTRODUCTION

Biological control of pest insects is affected by thefrequency of disturbances such as pesticide applica-tions or harvesting. Such disturbances often disruptbiological control, something that is particularly evi-dent in annual crops (Watt 1965, Smith et al. 1997).As a consequence, there are more examples of suc-cessful biological control from stable habitats such asorchards and forests than from annual crops (Waageand Greathead 1988, Wissinger 1997). Few, if any,studies have been performed in crops with an inter-mediate disturbance regime where more than one year,but less than the 30–80 years typical of forestry ro-tations, passes between harvests. One reason may bethat, at present, there are few such crops around.

Willow coppice (Salix spp.), grown to produce bio-mass for energy in Sweden and Great Britain, is anintermediately disturbed crop system where harvesting

Manuscript received 24 October 2003; revised 2 April 2004;accepted 17 April 2004. Corresponding Editor: J. A. Logan.

1 E-mail: Christer.Bjorkman@entom.slu.se

normally takes place every third to fifth year (Bates1995, Ledin and Willebrand 1995). It has been sug-gested that the regular harvesting results in high plantquality, which, in turn, may explain the often-observedhigh densities of herbivorous insects in the stands(Price and Martinsen 1994). In this paper we investi-gate to what extent harvesting affects the natural en-emies of the leaf beetles and ask whether negative ef-fects of harvesting on natural enemies may be an al-ternative explanation to the observed high densities ofherbivorous insects.

The willow coppice system provides an opportunityto explore the effect on herbivore dynamics of regulardisturbance at an intermediate time scale. Such studiesinform pest management decisions in the growing fieldof energy forests. It also offers an opportunity for ask-ing ecologically interesting questions about effects ontrophic interactions in an ecosystem where more timeis available for populations to recover and stabilizebetween disturbances. In annually harvested crop sys-tems there is an obvious risk that the disturbance is sofrequent that some patterns and hence mechanisms are

December 2004 1625HARVESTING DISRUPTS BIOLOGICAL CONTROL

PLATE 1. A stand of short-rotation coppice willows (Salixviminalis) in the spring after being harvested in the previouswinter. Resprouting shoots are in the foreground; a strip ofunharvested willows is in the background. Photo credit: C.Bjorkman.

more difficult to detect. For instance, there is little timefor numerical responses to be expressed in annual hab-itats, and ecological interactions result to a higher de-gree from functional responses and dispersal. Excep-tions are interactions between species with very shortgeneration times and high growth rates that may re-spond numerically even if they are disturbed repeatedlyin a growing season (Rauwald and Ives 2001).

Reports of insect outbreaks in willow plantations arecommon (Sage 1994, Sage and Tucker 1998). So far,the most severe problems have come from leaf beetles(Coleoptera: Chrysomelidae) of which three species arecommonly found in willow stands in Sweden: Phratoravulgatissima (L.), Galerucella lineola (F.), and Loch-mea caprea (L.). Defoliation can be severe, especiallyby the larvae of P. vulgatissima, and may reduce willowgrowth by up to 40% (Bjorkman et al. 2000a). Apply-ing insecticides is not an option in willow coppice be-cause of environmental concerns and small economicprofit margins. Biological control is therefore the onlyrealistic alternative for pest management.

In this paper we focus mainly on the role of naturalenemies. Among these we concentrate on three heter-opterans that have been identified as predators attack-ing eggs and/or larvae of P. vulgatissima and G. lineola(Bjorkman et al. 2000b, Bjorkman et al. 2003): Ortho-tylus marginalis (Reuter) (Miridae: Heteroptera), Clos-terotomus fulvomaculatus (DeGeer) (Miridae: Heter-optera), and Anthocoris nemorum (L.) (Anthocoridae:Heteroptera).

In the present study, we examined abundance esti-mates of three leaf beetle species, from 12 willow plan-tations in central Sweden that were followed for fourto five years. We assessed patterns of spatial synchronyand intraguild synchrony among the herbivores. Thedensities of the three potentially important heteropterannatural enemies were estimated, and we examined therelationship between herbivore population growth rateand natural enemy abundance. We, furthermore, in-vestigated the effects of egg predation on populationgrowth, utilizing data on survival of P. vulgatissimaeggs obtained from field experiments. Finally, we ex-plored how harvesting affected the herbivore and pred-ator populations and how they recovered after har-vesting.

METHODS

Study organisms

Willow coppice.—The most commonly used willowspecies in biomass production is Salix viminalis (L.).To grow willows for biomass production (and cleaningby removal of, for instance, heavy metals from soil andsewage) in short-rotation forestry is a recent devel-opment (Ledin and Willebrand 1995). The method issimple. Cuttings from re-sprouting current-year shootsare planted in the soil. The sprouting shoots grow vig-orously. Plants are harvested in the winter after three

to five years. The timing is determined by the stemdiameter. After about six harvests it is timely to replantwith new cuttings (see Plate 1).

For willows grown for biomass production, defoli-ation by herbivores is a problem because it may reducegrowth substantially. For example, heavy defoliationby P. vulgatissima may reduce biomass production byup to 40% (Bjorkman et al. 2000a).

Leaf beetles.—Three leaf beetle species are commonin willow coppice stands. The most abundant leaf beetlespecies in willow stands, P. vulgatissima, has receivedthe most attention. The sampling method we have usedfor estimating its density has, however, also been ef-ficient in collecting two other abundant leaf beetle spe-cies, G. lineola and L. caprea. Both of these speciesoccasionally reach outbreak densities in willow plan-tations (Larsson and Wiren 1982, Christersson et al.1993, Sage and Tucker 1998). All three species areunivoltine and overwinter as adults mainly outside thewillow plantations (Kendall et al. 1996b, Peacock etal. 1999, Sage et al. 1999), primarily in bark crevicesand hollow stems such as those of Phragmites sp. Ovi-position and larval feeding behavior differ among spe-cies. Phratora vulgatissima oviposits on willow leavesnear the base of willow shoots (Kendall et al. 1996a),whereas G. lineola oviposits on leaves at all levels ofwillow shoots (C. Bjorkman, K. Eklund, and S. Hog-lund, unpublished data). Oviposition in both these spe-cies is extended over a period of almost a month andthe first larvae normally emerge when there are stillunhatched eggs present. Phratora vulgatissima larvaefeed gregariously at first, but become more solitarywith time. Galerucella lineola larvae disperse beforefeeding. Lochmea caprea oviposits in the ground andits larvae, which occur a bit later than the larvae ofthe two other species, feed individually. More basic

1626 CHRISTER BJORKMAN ET AL. Ecological ApplicationsVol. 14, No. 6

information about the biology of the species can befound in Maisner (1974).

Natural enemies.—The most common and wide-spread (occurring in all surveyed plantations) naturalenemies, that attack eggs and larvae of P. vulgatissimain willow plantations, are three heteropteran predators,Orthotylus marginalis, Closterotomus fulvomaculatus,and Anthocoris nemorum (Bjorkman et al. 2003). In afew plantations we have in some years observed fairlyhigh densities of syrphid (Syrphis sp.) larvae attackingeggs and larvae of P. vulgatissima. Heteropterans, es-pecially A. nemorum, seem also to be important naturalenemies for G. lineola larvae (Bjorkman et al. 2000b).However, the surface of G. lineola eggs appears to betoo tough for heteropterans, and several other enemies,to penetrate. In a field trial in two willow plantations,egg survival was 80% for G. lineola compared to 40%for P. vulgatissima for eggs exposed to natural enemies,whereas survival among caged eggs was ;90% (C.Bjorkman, K. Eklund, and S. Hoglund, unpublisheddata). The heteropterans overwinter in different lifestages: O. marginalis and C. fulvomaculatus as eggs,A. nemorum as adults. Observations of high densitiesof all species in willow plantations very early in theseason (C. Bjorkman, K. Eklund, and S. Hoglund, un-published data) suggest that a substantial proportionof these predators overwinter within the willow plan-tations (see also Anderson 1962). High levels of par-asitoid attack are sometimes observed in G. lineola but,to our knowledge, never in P. vulgatissima.

The natural enemies of L. caprea are less well knownbut this species has eggs with a surface similar to thatof P. vulgatissima eggs, suggesting that they may alsobe vulnerable to heteropterans. However, because L.caprea deposits its eggs in the soil it is less likely thatheteropterans will find them. Larvae of L. caprea areprobably exposed to the same natural enemies as theother two leaf beetle species.

Study sites

The willow plantations were situated northwest ofStockholm in central Sweden (598409 N, 178309 E; Fig.1). They were established between 1990 and 1994 andhave thus been harvested once or twice since estab-lishment. The sizes of the plantations vary between 0.2and 11.8 ha (mean 5 3.2, SD 5 3.0). All plantationsare planted completely with S. viminalis except for onelarge plantation where there are a few rows of S. da-syclados.

Estimating density of herbivores and predators

The density of leaf beetles and natural enemies wasestimated by knockdown sampling twice a year, in latespring (mid May–early June) and early autumn (Au-gust) in all 12 plantations. Leaf beetles were sampledin all five years, 1998–2002, whereas natural enemies,due to buildup of knowledge, were sampled during2000–2002. Samples were taken at equal distances (15

m) along six transects within a plantation and alongthe entire edge of the plantation. Transects were evenlydistributed to obtain samples from the entire plantation.Transect length varied with plantation size. However,at least 30 samples were collected from each plantation,which is a sufficient number to arrive at a stable es-timate of density. In several independent previous stud-ies, we have found that 22–24 samples are sufficientto obtain stable estimates of mean and variance of den-sities in a plantation. Each knockdown sample was tak-en from the 40 cm top part of a shoot. Because ofdifferences in behavior of adults in spring and autumnwe used somewhat different methods to estimate abun-dance. In spring, beetles tend to occur at all heights,and we therefore multiplied the number of beetles ina sample with the number of 40-cm pieces of a stool(a stump with resprouting shoots) to estimate the num-ber per stool. The amount of leaves varies somewhatwith years after harvesting. This was accounted for byusing different correction values for different times af-ter harvesting. This means that the values of leaf abun-dance are adjusted for the amount of resource (willowleaves) available. The number of beetles per hectarewas obtained by multiplying with 10 3 103 (the averagenumber of stools per hectare). In autumn, adult beetlessit almost exclusively in the top shoots. We thereforemultiplied the number of beetles in a sample with thenumber of top shoots per stool and by 10 3 103 to getthe abundance per hectare.

Analysis of herbivore time series

Zero counts provide a problem in analyses of pop-ulation time series. We performed all tests with allzeroes set to half the minimum observed abundancefor each herbivore and predator species. This decisionis based on the assumption that the species were presentin each plantation each year, but that they sometimesoccurred at levels too low to be detected by our sam-pling method. All abundance data were log10 trans-formed prior to analysis.

Two types of analyses were performed: (1) withinspecies spatial synchrony and (2) intraguild synchrony.The first analysis was done to reveal if willow plan-tations could be viewed as independent units in thefurther analyses. The second analysis was done to findout to what extent the three leaf beetle species variedsynchronously. A high level of synchrony would in-dicate that they are affected by the same factors.

Within species spatial synchrony.—Using Mantelcorrelograms, we explored spatial synchrony for eachleaf beetle species separately (cf. Bjørnstad et al.1999a, b, Buonaccorsi et al. 2001, Peltonen et al.2002). First, we examined if the dynamics occurringin adjacent plantations was linked and therefore moresynchronized than distant plantations, by plotting thecorrelogram for time series of the log abundance,log(N ), and the population change, R, between the sea-sonal sampling occasions, Rt 5 log(Nt11) 2 log(Nt).

December 2004 1627HARVESTING DISRUPTS BIOLOGICAL CONTROL

FIG. 1. Density fluctuations (log10 N/ha) of three leaf beetle species (Phratora vulgatissima, Galerucella lineola, andLochmea caprea) in 12 willow coppice plantations over five years. The spring and autumn generations of beetle adults weresampled. In the center is a map showing the size of and distance between plantations. The cross is situated at 598379 N,178279 E. Arrows indicate when harvesting took place.

Second, we estimated the regional level of synchronyamong all plantations. Since we are interested in thelocal processes, we wish to remove potential large-scale trends over time prior to analysis (Koenig 1999).Such trends were detected by applying linear regressionto the time series for each species. If the regressionslope proved significant, the residuals were used toexamine spatial synchrony. In the cases where therewas a significant overall trend over time for a species,we instead used the residuals from the regression,log(N )r, and the difference of residuals (r) betweensampling occasions, Rtr 5 log(Nt11)r 2 log(Nt)r. Averageregional spatial synchrony was calculated for each spe-cies and the 95% confidence interval was obtained with1000 replacement bootstraps (Efron and Tibshirani1993).

Synchrony among herbivore species.—The level ofsynchrony among leaf beetle species was estimated byperforming pairwise correlations between time seriesof herbivore species in each location. Time series ofabundance as well as population growth rates between

sampling occasions were analyzed. The average cor-relation coefficient between time series among loca-tions was calculated. To obtain a measure of variabilitythe correlation coefficients were bootstrapped with re-placement 1000 times and the 95% confidence intervalwas calculated (Efron and Tibshirani 1993).

Egg survival

To estimate the overall predation pressure we mea-sured survival of P. vulgatissima eggs for five years insix of the plantations. The survival of eggs may beused as an indicator of total mortality up to pupationas egg and larval survival are very well correlated(Bjorkman et al. 2003). In each plantation 10 randomlydistributed egg batches were used to obtain an estimateof average egg survival in each plantation. Eggs wereobtained by keeping gravid females in sleeve cages onwillow shoots in the field for two days. The beetlesand the cages were then removed and the number ofeggs counted. The eggs were exposed to natural ene-mies for 11 days, except for an extended exposure of

1628 CHRISTER BJORKMAN ET AL. Ecological ApplicationsVol. 14, No. 6

FIG. 2. Centered Mantel correlation plots showing thespatial synchrony of the three leaf beetle species (Phratoravulgatissima, Galerucella lineola, and Lochmea caprea) fortime series of population change between subsequent seasonalsampling points.

14 days in 1998 due to cold weather. Survival equaledthe difference between the number of eggs at start andthe number of eggs not hatching, or that had disap-peared. To get an accurate estimate of eggs not hatch-ing, the eggs were brought to the laboratory and keptin petri dishes. After a week the number of unhatchedeggs, which are easy to distinguish from predated andhatched eggs, were counted under a stereomicroscope.

Predators, herbivores, and recovery after harvesting

To assess if herbivore population growth in the sum-mer is affected by predator occurrence, populationgrowth rate for each herbivore species was plottedagainst predator log abundance. Visual inspection in-dicated linear relationships, and linear equations werefitted to the data with least squares regression.

Predator and herbivore abundances, and herbivorepopulation growth from spring to fall, were plottedagainst number of years after harvesting in the willowplantations. Analysis of variance was performed to de-termine whether abundance or population growth dif-fered significantly among stand age classes in the sum-mer.

The statistical programming language R, version1.6.1 for Linux (Ihaka and Gentleman 1996) was usedfor analyzing spatial synchrony and for bootstrapping.For all other analyses we used SAS version 8.2 forLinux (SAS Institute 1999).

RESULTS

Density estimates

Population abundance time series, log(N ), and a mapof the 12 willow plantations are presented in Fig. 1.Note that harvesting, indicated by vertical arrows inFig. 1, is not synchronized among plantations: twoplantations were harvested in 1996, four in 1997, onein 1998, six in 1999, none in 2000, three in 2001, andone in 2002.

Among the leaf beetles, P. vulgatissima was the mostabundant. It had a negative regional trend in abundanceover the years (linear regression, log(N ) 5 5.03 2 0.223 year, Ps , 0.0001, n 5 117, s, slope, where 1997 isset to year one in the regression). G. lineola was thesecond most abundant leaf beetle and also showed anegative regional trend in abundance over time (log(N )5 4.37 2 0.29 3 year, Ps , 0.0001, n 5 117). L. capreawas the least abundant with no regional trend (log(N )5 2.89 2 0.03 3 year, Ps 5 0.44, n 5 117).

Among the natural enemies, O. marginalis was onaverage 0.69 (SE 5 0.07) and 1.4 (SE 5 0.13) ordersof magnitude more abundant than C. fulvomaculatusand A. nemorum, respectively (Fig. 5).

Analysis of herbivore time series

Within species spatial synchrony.—Analyzing spa-tial synchrony for population change, R, we found noclear pattern of spatial synchrony or asynchrony of any

leaf beetle species across the range of 15 km in thisstudy (Fig. 2), i.e., there was high variability in pop-ulation change among locations, also between adjacentplantations. A similar pattern as in Fig. 2 was foundfor all species when analyzing abundance, log(N ), inthe same way (not presented). These analyses wereperformed on the residuals after removing the regionalsynchrony. The average regional synchrony was sig-nificantly or nearly significantly positive for all threeherbivore species (Table 1), but was highly variable.

Synchrony among herbivore species.—G. lineolaand P. vulgatissima varied synchronously within lo-calities for spring abundance, log(N ), and for popu-lation growth, R, between spring and summer. The oth-er combinations of species were also significantly dif-ferent from zero correlation but showed a larger var-iability and a lower average degree of synchronyamong species (Table 2).

Predator abundance and herbivore growth rate

Population growth of all three beetle species betweenspring and fall was negatively correlated to the abun-

December 2004 1629HARVESTING DISRUPTS BIOLOGICAL CONTROL

TABLE 1. Regional synchrony in abundance [log(N )] and population growth rate (R) of threeleaf beetle species in willow plantations with 95% bootstrapped confidence limits.

Species log(N ) R

Phratora vulgatissimaGalerucella lineolaLochmea caprea

0.72 [0.56, 0.86]†0.65 [0.45, 0.83]†0.28 [20.05, 0.65]

0.61 [0.42, 0.78]†0.53 [0.31, 0.73]†0.67 [0.46, 0.83]

† Residuals from linear regression are used in the analysis (see Methods).

TABLE 2. Synchrony in spring abundance [log(N )] and summer population growth (R) amongthree leaf beetle species in willow plantations.

Compared species log(N ) R

Galerucella lineola, Phratora vulgatissimaGalerucella lineola, Lochmea capreaPhratora vulgatissima, Lochmea caprea

0.78 [0.70, 0.84]0.42 [0.24, 0.59]0.37 [0.18, 0.53]

0.57 [0.42, 0.71]0.29 [0.12, 0.43]0.34 [0.19, 0.48]

Note: Average level of correlation among locations and 95% bootstrapped confidence limitsare presented for pairwise comparisons.

dance of the predator O. marginalis (Fig. 3; P. vul-gatissima, r2 5 0.23, P 5 0.003, n 5 35; G. lineola,r2 5 0.25, P 5 0.002, n 5 35; L. caprea, r2 5 0.14,P 5 0.03, n 5 35). There was no significant correlationbetween the growth rate of any of the herbivore speciesand abundance of the predator A. nemorum.

Egg survival

In P. vulgatissima, there was a positive relationshipbetween egg survival (i.e., predation pressure) and pop-ulation growth both from spring to fall (Fig. 4; Pearsoncorrelation, r 5 0.39, P 5 0.03, n 5 33). The effectwas detectable throughout the year, and the relationshipremained significant when comparing egg survival topopulation change from spring to the next spring (Pear-son correlation, r 5 0.49, P 5 0.01, n 5 26).

Predators, herbivores, and recovery after harvesting

Harvesting reduced both leaf beetle and predatorabundance (Fig. 5a, b). The population growth ratefrom spring to fall of the two leaf beetles P. vulgatis-sima and G. lineola was very high the first year afterharvesting and differed (although not significantly inG. lineola) in relation to years after harvesting (Fig.6; general linear model, P. vulgatissima, F4,42 5 3.8, P5 0.008; G. lineola, F4,51 5 2.4, P 5 0.06). For L.caprea there was a significant difference between yearsafter harvesting with respect to population growth rate(F4,51 5 4.4, P 5 0.004) but for this species the ratewas very high the first two years after harvesting (Fig.6).

The density of leaf beetles peaked, in most planta-tions, some years after harvesting. For all three leafbeetle species the peak normally came two years afterharvesting. The abundance of all predator species in-creased after harvesting (Fig. 5b). For O. marginalisand possibly C. fulvomaculatus there was, after the firstyear, a tendency for a weak but continuous increase,

whereas the abundance of A. nemorum tended to leveloff.

DISCUSSION

Harvesting of willow coppice every fifth year seemsto disrupt biological control for all three leaf beetlespecies in the current study. The natural enemies arenegatively affected by harvesting (Fig. 5b) as theyprobably overwinter in the plantations. Herbivoreabundance also decreases after harvesting. However,the populations of the three leaf beetle species recoverrapidly (Figs. 5 and 6), something that can be explainedby low levels of predation (Figs. 3 and 4) as a con-sequence of the dip in natural enemy abundance im-mediately after harvesting (Fig. 5). After this the pred-ators regain in numbers and start reducing herbivorepopulation growth. As a result, the highest herbivorepopulation levels were generally reached three yearsafter harvesting, followed by a tendency for decliningherbivore populations until the next harvest (Fig. 5).

The drastic decrease in abundance of both herbivoresand predators after harvesting may partly be due to themethod used for estimating density; the newly sprout-ing shoots from harvested willows need to grow to acertain height before we can make a knockdown sam-ple. This means that the sampling in harvested standstakes place one to two weeks later than in the otherstands. A proportion of the adult leaf beetles may diewithin this time period. However, we normally stillobserve egg-laying beetles when the sampling is done,indicating that they are still active. For the natural en-emies the situation is different: new nymphs (andadults) of the anthocorid, which overwinters as anadult, have then normally started to occur whereasnymphs of the mirids, which overwinter as an egg, havereached their mid-nymphal stages. Thus, there is asmall risk that the abundance of leaf beetles has beensomewhat underestimated in the harvested stands. For

1630 CHRISTER BJORKMAN ET AL. Ecological ApplicationsVol. 14, No. 6

FIG. 3. Correlation between abundance (log10 N/ha) of thegeneralist predator species Orthotylus marginalis and popu-lation growth rates in three leaf beetle species (Phratora vul-gatissima, Galerucella lineola, and Lochmea caprea) in wil-low coppice plantations.

FIG. 4. Rate of log population growth (R) in relation toegg survival, estimated independently in field experiments,in the leaf beetle Phratora vulgatissima in willow plantations.

FIG. 5. Effect of harvesting on the abundance (mean 61 SE) of (a) three leaf beetle species and (b) three commongeneralist natural enemy species and in willow coppice plan-tations. Key for herbivores: solid diamonds, Phratora vul-gatissima; open squares, Galerucella lineola; 3, Lochmeacaprea. Key for predators: solid diamonds, Orthotylus mar-ginalis; 3, Closterotomus fulvomaculatus; open squares, An-thocoris nemorum.

the natural enemies there is no such risk. If this is true,one consequence could be that the population growthrate of leaf beetles directly after harvesting is over-estimated. However, the patterns of change in abun-dance two to five years after harvesting (Fig. 5) wouldstill hold, indicating an effect of the predators on theherbivores. Another consequence would be that the var-iability in population growth rate in the herbivoreswould decrease. It is not possible to judge whether thiswould lead to stronger or weaker relationships (e.g.,Figs. 3 and 4).

It has been suggested that herbivore populations onwillows are mainly affected by bottom-up factors suchas plant quality (Price and Martinsen 1994). Thus, adrastically improved plant quality could be an alter-

native to lowered predation for the observed patternsin leaf beetle abundance. Harvesting makes the willowsgrow more vigorously and one might, therefore, expectan improved plant quality for herbivores. However,studies investigating the effect of plant quality on leafbeetle preference and performance have not provided

December 2004 1631HARVESTING DISRUPTS BIOLOGICAL CONTROL

FIG. 6. Effect of harvesting on the log population growthrates (mean 6 1 SE) of three leaf beetle species (Phratoravulgatissima, Galerucella lineola, and Lochmea caprea) inwillow coppice plantations.

any evidence supporting the hypothesis that high plantquality results in high leaf beetle densities (Kelly andCurry 1991a, Bjorkman et al. 2000b). For example,Kelly and Curry (1991a) found no difference in pref-erence or performance in P. vulgatissima between cop-piced and uncoppiced Salix viminalis stands despitethere being twice the density of larvae in the coppicedstands. They did not study the impact of natural ene-mies. Furthermore, we demonstrated in a study thatdifferences in density of G. lineola in two adjacentwillow stands (of S. cinerea) were correlated with pre-dation rate and density of natural enemies in each stand,whereas no difference in plant quality could be detected(Bjorkman et al. 2000b). In the G. lineola study weactually observed a substantial difference in plant vigor(i.e., shoot length) between willow stands with low and

high leaf beetle densities but found no difference inadult oviposition or larval performance (Bjorkman etal. 2000b). Nevertheless, there is still a need for studieswhere the variation in plant quality in relation to har-vest is quantified more accurately. Recent findings thatleaf beetle defoliation may induce increased leaf hair-iness in willows (Zvereva et al. 1998, Dalin and Bjork-man 2003) are one of several confounding factors (alsocf. Kelly and Curry 1991b, Peacock et al. 2001) thatneed to be considered when evaluating the possible roleof plant quality in leaf beetle population dynamics.

The present study suggests instead that predation canbe important for determining population growth of her-bivores on willow. This is indicated by the positivecorrelation between survival of P. vulgatissima eggsand population growth rate found in the current study.The effect was significant both for summer populationgrowth measured from spring to fall (Fig. 4), and foryearly growth measured from spring to spring. Fur-thermore, the abundance of the heteropteran predatorO. marginalis was negatively correlated with popula-tion growth of all three leaf beetle species in the sum-mer (Fig. 3), again suggesting the potential for pred-ators to influence herbivore population growth.

Although A. nemorum is well known for being animportant predator especially on pear aphids (Scuta-reanu et al. 1999), we found no indications of it havinga regulatory effect on the leaf beetles. The lack ofrelationship between population growth rate in the leafbeetles and abundance of A. nemorum and C. fulvo-maculatus, may simply be an effect of too low densitiesof these predators. The low densities of A. nemorummay, in turn, partly be the result of strong negativeintraspecific interactions in this active and mobile spe-cies (C. Bjorkman and A.-S. Liman, unpublished man-uscript). The less mobile O. marginalis showed no neg-ative intraspecific interactions at similar densities (C.Bjorkman and A.-S. Liman, unpublished manuscript).Strong negative intraspecific interactions may lead tomore dispersal (Evans 1976) and/or stronger compe-tition, resulting in slower population growth. The lev-eling off of the postharvest increase in abundance ofA. nemorum seen in the third year indicates such aneffect.

Despite obvious differences in biology and behavioramong the three leaf beetle species, they displayed sim-ilar patterns in most analyses. This suggests that allthree species are affected by the same factors. For ex-ample, the similarity in the relationship between pop-ulation growth rate and abundance of the common pred-ator O. marginalis indicates that predation on larvaeis important. As mentioned in the Methods, we knowthat eggs of G. lineola are not predated by heteropter-ans. Nevertheless, egg survival in P. vulgatissima,which is easier to measure, seemed to give a reasonablygood indication of overall predation pressure in a plan-tation in a certain year.

1632 CHRISTER BJORKMAN ET AL. Ecological ApplicationsVol. 14, No. 6

A prerequisite for analyzing population data fromspatially separated willow plantations is that the dy-namics in each plantation, at least to some degree, varyindependently. Despite the very short distances be-tween some of the plantations, no strong spatial au-tocorrelation could be detected. One reason for thispattern may be low levels of dispersal of either her-bivores or predators between plantations. A real needfor dispersal of the herbivores seems unlikely, as foodvery seldom is in short supply. The willows grow con-tinuously during the whole season. However, all threeleaf beetle species demonstrated a certain degree ofregional synchrony (Table 2), which probably is a resultof weather which is unlikely to vary significantly atthe spatial scale of the current study. Nevertheless,there was a large amount of variation, in both herbivoreabundance and population growth, among plantationsthat can be attributed to local processes acting withinplantations. That the abundance of natural enemies alsomay vary considerably is illustrated by the fact that C.fulvomaculatus was more abundant than A. nemorumin this study whereas the opposite result was obtainedin a similar study based on 13 other willow plantations(Bjorkman et al. 2003).

The interaction between dispersal and disturbance isa current issue in applied ecology (e.g., Hastings 2003).To understand this interaction is essential when tryingto predict the outcome of predator–prey interactionsand the role of refuges. Spatial refuges with low levelsof predation may act as a stabilizing factor on the pop-ulation dynamics of prey (Luck 1990, Murdoch 1994).However, how such spatial effects transfer to temporaldynamics is not easy to predict (Murdoch 1994). In thepresent system harvesting is a disturbance that createsa temporal refuge for the leaf beetles with low predatorabundance, leading to instability and an increased like-lihood for high leaf beetle densities. A challenge forecologists is to identify how spatial and temporal ref-uges are linked and quantify their relative importance.

The intermediate disturbance regime in willow cop-pice plantations, of around five years between harvests,appears to be too short to avoid serious disruption ofbiological control of the leaf beetles. For efficient bi-ological control, a harvest period of six to eight years,as in poplar cultivation, could be an alternative givingthe predators more time to respond numerically. If theincreased herbivore population growth indeed is a re-sult of predators (which overwinter inside the planta-tions) being wiped out at harvesting during the winter,a pest management strategy could be to leave refugesfor natural enemies at harvest. Another approach wouldbe to harvest adjacent willow plantations asynchro-nously. Such experiments provide a next step for im-proving biological control in plantations of willowstogether with more careful evaluation of the populationdynamics of the predators.

ACKNOWLEDGMENTS

We would like to thank Adriana Jerlstrom-Maj, Peter Dalin,Einar Larsson, and Weronika Linkowski for assistance in the

field. The farmers Goran Eriksson, Peter Rickberg, Tina Ru-dolfsson, Nils Thel, Robert Tiblom, Sven Wretler, and LeifZetterberg kindly let us use their willow plantations. We thankHelena Bylund, C. M. Collins, Peter Dalin, Peter Hamback,Christer Solbreck, and an anonymous reviewer for valuablecomments on the manuscript. Ottar Bjørnstad kindly providedthe R-code used for analysing spatial synchrony. The SwedishNational Energy Administration financially supported thestudy. In addition, R. Bommarco was financially supportedby The Swedish Research Council for Environment, Agri-cultural Sciences and Spatial Planning (FORMAS).

LITERATURE CITED

Anderson, H. N. 1962. Bionomics of six species of Antho-coris (Heteroptera: Anthocoridae) in England. Transactionsof the Royal Entomological Society of London 114:67–95.

Bates, J. 1995. Alternatives in agriculture—growing cropsfor industrial, energy and other non-food purposes. Parlia-mentary Office of Science and Technology, London, UK.

Bjorkman, C., B. Bengtsson, and H. Haggstrom. 2000b. Lo-calized outbreak of a willow leaf beetle: plant vigour ornatural enemies? Population Ecology 42:91–96.

Bjorkman, C., P. Dalin, and K. Eklund. 2003. Generalist nat-ural enemies of a willow leaf beetle (Phratora vulgatissi-ma): abundance and feeding habits. Journal of Insect Be-haviour 16:747–764.

Bjorkman, C., S. Hoglund, K. Eklund, and S. Larsson. 2000a.Effects of leaf beetle damage on stem wood production incoppicing willow. Agricultural and Forest Entomology 2:131–139.

Bjørnstad, O. N., R. A. Ims, and X. Lambin. 1999b. Spatialpopulation dynamics: causes and consequences of spatialsynchrony in density fluctuations. Trends in Ecology andEvolution 14:427–431.

Bjørnstad, O. N., N. C. Stenseth, and T. Saitoh. 1999a. Syn-chrony and scaling in dynamics of voles and mice in north-ern Japan. Ecology 80:622–637.

Buonaccorsi, J. P., J. S. Elkinton, S. R. Evans, and A. M.Liebhold. 2001. Measuring and testing for spatial syn-chrony. Ecology 82:1668–1679.

Christersson, L., M. Ramstedt, and J. Forsberg. 1993. Pests,diseases and injuries in intensive short-rotation forestry.Pages 185–215 in C. P. Mitchell, J. B. Ford-Robertson, T.Hinkley, and L. Sennerby-Forsse, editors. Ecophysiologyof short rotation forest crops. SLU/repro, Uppsala, Sweden.

Dalin, P., and C. Bjorkman. 2003. Adult beetle grazing in-duces willow trichome defence against subsequent larvalfeeding. Oecologia 134:112–118.

Efron, B., and R. J. Tibshirani. 1993. An introduction to thebootstrap. Chapman and Hall, London, UK.

Evans, H. F. 1976. Mutual interference between predatoryanthocorids. Ecological Entomology 1:283–286.

Hastings, A. 2003. Metapopulation persistence with age-de-pendent disturbance of succession. Science 301:1525–1526.

Ihaka, R., and R. Gentleman. 1996. R: a language for dataanalysis and graphics. Journal of Computational andGraphical Statistics 5:299–314.

Kelly, M. T., and J. P. Curry. 1991a. The biology and pop-ulation density of the willow beetle (Phratora vulgatissima[L.]) on Salix viminalis in reclaimed cutaway peat. Journalof Applied Entomology 111:44–56.

Kelly, M. T., and J. P. Curry. 1991b. The influence of phenoliccompounds on the suitability of three Salix species as hostsfor the willow beetle Phratora vulgatissima. EntomologiaExperimentalis et Applicata 61:25–32.

Kendall, D. A., T. Hunter, G. M. Arnold, J. Liggit, T. Morris,and C. W. Wiltshire. 1996a. Susceptibility of willow clones(Salix spp.) to herbivory by Phyllodecta vulgatissima (L.)and Galerucella lineola (Fab.) (Coleoptera: Chrysomeli-dae). Annals of Applied Biology 129:379–390.

December 2004 1633HARVESTING DISRUPTS BIOLOGICAL CONTROL

Kendall, D. A., C. W. Wiltshire, and M. Butcher. 1996b. Phe-nology and population dynamics of willow beetles (Cole-optera: Chrysomelidae) in short-rotation coppiced willowsat Long Ashton. ETSU (DTI) Biofuels Study B/M4/00487/14/00. IAcr-Long Ashton Research Station, Long Ashton,Bristol, UK.

Koenig, W. D. 1999. Spatial autocorrelation of ecologicalphenomena. Trends in Ecology and Evolution 14:22–26.

Larsson, S., and A. Wiren. 1982. Leaf-eating insects in anenergy forest stand of Salix viminalis L. in central Sweden.Annales Entomologici Fennici 48:119–125.

Ledin, S., and E. Willebrand, editors. 1995. Handbook onhow to grow short rotation forests. IEA Bioenergy, SwedishUniversity of Agricultural Sciences, Uppsala, Sweden.

Luck, R. F. 1990. Evaluation of natural enemies for biologicalcontrol: a behavioural approach. Trends in Ecology andEvolution 5:196–199.

Maisner, N. 1974. Chrysomelidae, Blattkafer. Pages 202–236in W. Schwenke, editor. Die Forstschadlinge Europas, Bd2. Paul Parey, Hamburg, Germany.

Murdoch, W. W. 1994. Population regulation in theory andpractice. Ecology 75:271–287.

Peacock, L., S. Herrick, and P. Brain. 1999. Spatio-temporaldynamics of willow beetle (Phratora vulgatissima) in short-rotation coppice willows grown as monocultures or a ge-netically diverse mixture. Agricultural and Forest Ento-mology 1:287–296.

Peacock, L., M. Lewis, and S. Powers. 2001. Volatile com-pounds from Salix spp. varieties differing in susceptibilityto three willow beetle species. Journal of Chemical Ecology27:1943–1951.

Peltonen, M., A. M. Liebhold, O. N. Bjørnstad, and D. W.Williams. 2002. Variation in spatial synchrony among for-est insect species: roles of regional stochasticity and dis-persal. Ecology 83:3120–3129.

Price, P. W., and G. D. Martinsen. 1994. Biological pest con-trol. Biomass and Bioenergy 6:93–101.

Rauwald, K. S., and A. R. Ives. 2001. Biological control indisturbed agricultural systems and the rapid recovery ofparasitoid populations. Ecological Applications 11:1224–1234.

Sage, R. B. 1994. A review of the status and control strategiesof known and perceived insect pests on Salix and Populusin North West Europe. ETSU B/M3/00388/:10/REP. Har-well Laboratories, Oxford, UK.

Sage, R. B., D. Fell, K. Tucker, and N. W. Sotherton. 1999.Post hibernation dispersal of three leaf-eating beetles (Co-leoptera: Chrysomelidae) colonising cultivated willows andpoplars. Agricultural and Forest Entomology 1:61–70.

Sage, R. B., and K. Tucker. 1998. The distribution of Phra-tora vulgatissima (Coleoptera: Chrysomelidae) on culti-vated willows in Britain and Ireland. European Journal ofForest Pathology 28:289–296.

SAS Institute. 1999. SAS version 8.2 for Linux. SAS Insti-tute, Cary, North Carolina, USA.

Scutareanu, P., R. Lingeman, B. Drukker, and M. W. Sabelis.1999. Cross-correlation analysis of fluctuations in localpopulations of pear psyllids and anthocorid bugs. Ecolog-ical Entomology 24:354–362.

Smith, J. W., R. N. Wiedenmann, and F. E. Gilstrap. 1997.Challenges and opportunities for biological control inephemeral crop habitats: an overview. Biological Control10:2–3.

Waage, J. K., and D. J. Greathead. 1988. Biological control:challenges and opportunities. Philosophical Transaction ofthe Royal Society London B 318:111–128.

Watt, K. E. F. 1965. Community stability and the strategy ofbiological control. Canadian Entomologist 97:887–895.

Wissinger, S. A. 1997. Cyclic colonization in predictablyephemeral habitats: a template for biological control in an-nual crop systems. Biological Control 10:4–15.

Zvereva, E. L., M. V. Kozlov, and P. Niemela. 1998. Effectsof leaf beetle pubescence in Salix borealis on host-plantchoice and feeding behaviour of the leaf beetle, Melasomalapponica. Entomologia Experimentalis et Applicata 89:297–303.