Spatial and temporal distribution of bioclimatic potential for the Codling moth and the Colorado potato beetle in Norway: model predictions versus climate and field data from the 1990s

Authors


Trond Rafoss. Tel: +47 64 94 94 00; fax: +47 64 94 92 26; e-mail: trond.rafoss@planteforsk.no

Abstract

Abstract 1 Based on climate data from a network of agrometeorological stations in Norway, the effects of current and future climate regimes on the spatial and temporal distribution of the Codling moth (Cydia pomonella) and the establishment potential of the Colorado potato beetle (Leptinotarsa decemlineata) were investigated.

2 The study was accomplished using climex, a dynamic climate matching- and climate response estimation model, which predicts potential distribution of an organism based on its known geographical distribution.

3 Validation of the climex model predictions for C. pomonella against field data on spatial distribution of the species in Norway resulted in a refined set of climate response parameters for C. pomonella. Temporal occurrence of C. pomonella seems to be affected by climate (temperature) and insecticide treatment against the Apple fruit moth (Argyresthia conjugella) in the previous season.

4 Climate change scenarios (0.1 °C increase per degree in latitude in daily maximum and minimum temperatures) indicated an extension of the potential geographical range for C. pomonella, and 23 new locations were found favourable for its long-term survival. The abundance and pest status of C. pomonella could increase dramatically in those locations where the species is already established.

5 Leptinotarsa decemlineata would only temporarily find suitable climate conditions in Norway and hence only be able to establish interim populations in a few regions under current climate conditions. Climate change scenarios for L. decemlineata indicated that the species would be able to establish as far north as 64°N, mainly in the inland of eastern Norway.

6 In general, the methods applied support the process of decreasing the uncertainty both in our knowledge about the pests themselves and about the environment, which are crucial elements in predicting whether a species is able to establish in a new area.

Introduction

It has long been recognized that the distribution of poikilothermic animals is largely determined by climate (Andrewartha & Birch, 1954). Also, to a large extent, climate determines the distribution of host plants, and thereby indirectly influences the distribution of phytophagous insects. Assessment of the bioclimatic potential for pest organisms in a country or region may provide important information either on the establishment potential of exotic pests or for the management of native or immigrant pest species. Although the latter may contain a hidden potential of modelling in pest management, the former issue is increasingly important as international trade in agricultural commodities is steadily increasing. For countries trying to maintain protection from new biological invasions following international trade commodities, the World Trade Organization, through the agreement on sanitary or phytosanitary measures (the SPS agreement), now clearly requires member countries not to apply sanitary and phytosanitary measures to trade for purposes other than ensuring food safety and animal and plant health. In order to justify such sanitary and phytosanitary measures, the agreement also requires governments to provide an assessment of risk underpinning the appropriateness of their trade restrictions. These risk assessments are required to follow certain international standards.

Scientists working with plant health and quarantine problems are asked by the authorities to provide answers to questions often involving large uncertainties. The information needed to provide answers may be sparse, difficult to locate, or unavailable. In pest risk analysis, the second stage of the analysis, called pest risk assessment, is the most important and difficult stage in the process. This stage seeks to answer several questions, and involves the risk that a pest would enter, establish, spread and cause economic damage in a given country or area. Answers to several of these questions may be equally interesting for the risk management of non-quarantine pests. Regarding establishment, two sources of uncertainty are involved, our knowledge about the pest organism and our knowledge about the environmental conditions of the study area.

The present study simulates and validates the process involved in a pest risk assessment, taking into account the sources of uncertainty. In order to aid this analysis, we apply climex, a model commonly used as a tool in pest risk assessment (Worner, 1988; Sutherst et al., 1991; Baker et al., 1996; Hughes & Evans, 1996; Baker et al., 2000). The objective of this study is to assess the establishment potential for an exotic insect pest in Norway, the Colorado potato beetle, Leptinotarsa decemlineata (Say) (Coleoptera: Chrysomelidae), and the potential of further spread of a non-native pest, the codling moth, Cydia pomonella L. (Lepidoptera: Tortricidae), to new areas, as well as to evaluate this type of model generally for risk assessment.

The two insect species selected for this study are important pests in agriculture world-wide. They are relatively well studied and pose different threats to the study area. Apples are grown in most lowland municipalities of southern Norway and C. pomonella is a common pest in these areas. L. decemlineata has never been established or found in the field in Norway. However, it has been intercepted several times (Fjelddalen, 1991), and a total of 48 specimens have been reported from Norway. The last case was in 2000.

The establishment potential for L. decemlineata in Norway was predicted based on climate data from agricultural areas. The results from field surveys of the geographical distribution of C. pomonella in Norway (Sæthre & Edland, 2001) were compared with the potential geographical distribution predicted by the climex model. We also investigated temporal variations in model predictions with temporal variations in climate and the subsequent variations in model predictions for both species. Temporal variations in model predictions were again compared with field data available from a surveillance programme for occurrence of C. pomonella in Norwegian apple orchards. This kind of uncertainty may not only be present on a short temporal scale, but is also highly relevant on a longer temporal scale because of effects of potential climate change. The latter issue was addressed by running the model under different climate change scenarios (Sutherst, 1991). The general potential of climex to predict the impact of climate change, on invasive species, has been reviewed by Sutherst (2000). After validation of the model predictions for an established pest (C. pomonella), we wanted to analyse the establishment potential of an exotic pest (L. decemlineata) in light of that experience, including an analysis of both short and long-term variations in climate.

Finally, we addressed a common problem with climate response estimation models, the problem of estimating climate response parameters for pests. When estimating climate response parameters based on the observed distribution of a pest, it is uncertain whether its geographical distribution is an expression of its climatic tolerance, or whether it is only an expression of how far it has had time to spread or adapt. On the other hand, we consider this type of parameter estimation process as an interesting approach to knowledge updating regarding our information base on quantitative pest biology.

Materials and methods

Assessment of bioclimatic potential for an organism in an area (i.e. its innate capacity to maintain a population) (Meats, 1989) can be carried out in many ways by different methodologies. Meats (1989) provides a discussion of the two basic approaches that have been used: bioclimatic analysis and laboratory assessment. Laboratory assessment relies on measurements of survival, development and reproduction conducted under laboratory conditions to predict the species response to the natural environment. In its original form, bioclimatic analysis was based on a comparison of climatic conditions in areas where the pest was known to occur with conditions in the uninfested area under study (Cook, 1929; Nash, 1933). The latter approach has evolved to a whole class of models for bioclimatic analysis (Sutherst et al., 1995), and includes climex, a generic model of organisms response to climate (Sutherst & Maywald, 1985; Maywald & Sutherst, 1989, 1991; Sutherst et al., 1991; Skarratt et al., 1995). The model generates the response of an individual species to a given climate as an annual growth index (GI), describing ‘the overall potential for population growth’, generated by weekly indices of temperature (TI), moisture (MI), daylength (LI) and diapause (DI), summed and averaged to give the annual GI. The annual GI is combined with four stress indices (representing hot, cold, dry and wet weather) to produce an ecoclimatic index (EI), describing ‘the overall suitability of the location for the propagation and persistence of the species’ (Skarratt et al., 1995; Sutherst et al., 1995).

CLIMEX input

For each individual species, the inputs required in climex are one set of climate response parameter values (see Appendix). For the region to be studied, inputs are climate data as monthly values of five variables: average maximum daily temperature, average minimum daily temperature, average monthly rainfall, and average daily relative humidity at 09.00 and 15.00 hours. The model relies on inferring the species' requirements from its observed geographical distribution, supported by field observations on the phenology of populations. The two indices predicted by climex are scaled between 0 and 100, where 100 expresses optimal conditions for the growth and ecoclimatic potential of the species, and 0 expresses the lower limit, where no growth can occur and where ecoclimatic conditions do not support persistence of a population of the species. As the climate of a given area is hardly constant and ideally suited to any species, the EI very rarely approaches a value of 100. An EI of more than 50 represents a very favourable climate for the long-term survival of the species (Skarratt et al., 1995).

Table Appendix. 
Climatic response parameters used for Cydia pomonella and Leptinotarsa decemlineata. Climatic response parameters are obtained from the climex manual (Skarratt et al., 1995), with parameters adjusted by the authors (Adjusted).
  Cydia pomonella
Leptinotarsa
  CLIMEXAdjusteddecemlineata
  • a

    SM = 0 indicates no soil moisture, SM = 0.5 indicates soil moisture content is 100% of capacity, SM = 1 indicates that the soil moisture content is 100% of capacity, SM > 1 indicates a water content greater than the soil holding capacity (i.e. runoff). Maximum soil moisture capacity is set at 100 mm for all calculations.

  • b

    Each stress function has two parameters. A threshold parameter which determines the level above or below which stress is accumulated, and a rate parameter which determines the rate of accumulation of stress. The effective weekly stress is defined as the weekly stress multiplied by the number of weeks since the stress was zero. V(effective stress value) = S × t, where S is weekly stress and t is the time in weeks since zero stress. The yearly stress index = inline image.

Moisture parametersa
 SM0Lower soil moisture threshold for population growth  0.2  0.1 0.15
 SM1Lower optimal soil moisture for population growth  0.5  0.3 0.35
 SM2Upper optimal soil moisture for population growth  1.5  1.5 0.80
 SM3Upper soil moisture threshold for population growth  2.5  2.5 1.50
Temperature parameters
 DV0Lower temperature threshold for population growth  2.0 °C  9.0 °C12.0 °C
 DV1Lower optimal temperature for population growth 17.0 °C 15.0 °C18.0 °C
 DV2Upper optimal temperature for population growth 22.0 °C 27.0 °C28.0 °C
 DV3Upper threshold temperature for population growth 25.0 °C 30.0 °C35.0 °C
 PDDMinimum day-degrees 750 580400
Stress indicesb
 SMDSSoil moisture dry stress  0.15  0.15 0.1
 HDSRate of accumulation of dry stress  0.01  0.01 0.02
 SMWSSoil moisture wet stress  2.0  2.0 1.5
 HWSRate of accumulation of wet stress  0.002  0.002 0.0002
 TTCSThreshold of cold stress  5.0 °C  1.0 °C 1.0 °C
 THCSRate of accumulation of cold stress  0.00001  0.00001 0.00013
 TTHSThreshold of heat stress 29.0 °C 35.0 °C35.0 °C
 THHSRate of accumulation of heat stress  0.006  0.006 0.02
 DPD0Diapause induction daylength 12  015
 DPT0Diapause induction temperature 10.0 °C  5.0 °C 6.0 °C
 DPT1Diapause termination temperature  0.0 °C  0.0 °C 0.0 °C
 DPDDiapause development days, defaults to 0, i.e. facultative DP  0 90 0

In this study, we use climatic response parameter values for L. decemlineata originally estimated by Worner (1988) and later adjusted by Sutherst et al. (1991) (see Appendix). The parameter values initially used for C. pomonella were available in the manual of the system (see Appendix) (Skarratt et al., 1995). In the present study, we utilized these parameters as a starting point. However, when comparing the predictions made by climex with the observed distribution of the species in southern Norway (Sæthre & Edland, 2001), major discrepancies were observed. As a consequence, the climate parameters were adjusted to fit the observed distribution of the species in Norway.

Climate response parameter fitting for C. pomonella

Parameter fitting was performed by consulting different sources of biological and ecological information on C. pomonella, such as published reports (Glenn, 1922; Shel'deshova, 1967; Riedl & Croft, 1978; Richardson et al., 1982; Blago, 1992; Gottwald, 1996; Mani et al., 1997; Howell & Neven, 2000), information available on the internet (Anon, 2000, 2001), and the results from experiments recently conducted in Norway (Sæthre & Hofsvang, 2002). On this basis, the effect of parameter adjustments from the initial parameter values provided by Skarratt et al. (1995) were assessed by visual comparison of the predicted geographical distribution and the survey results (Sæthre & Edland, 2001). We also assessed the effects of parameter adjustments on the four different indices constituting the growth index and the eight stress indices.

Rainfall and humidity are reported to have only minor effects on any stage in the life cycle of C. pomonella (Anon, 2000). However, we found that the two soil moisture parameters, SM0 and SM1, lowered the overall MI for most locations in the generally drier south-east parts of the country, whereas the value for this index was 100 for the locations in the other regions. As we had no reason to believe that moisture limitations would limit the growth potential for populations of C. pomonella in this region, the soil moisture parameters SM0 and SM1 were adjusted to a level where the MI did not affect the growth potential (GI) (see Appendix). This adjustment must be seen in light of how moisture calculations are treated in climex where the moisture index is based on the assumption that soil moisture is the dominant factor determining microclimatic conditions and the condition of the vegetation (Skarratt et al., 1995). The soil moisture balance is calculated from the stored soil moisture, the rainfall and the evapotranspiration. Two parameters form an important part of these calculations (i.e. the maximum soil moisture storage capacity and the evapotranspiration coefficient). The maximum soil moisture storage capacity is set at 100 mm. Although this level is representative of a global soil moisture model, it will in reality vary with soil type. The evapotranspiration coefficient is set at 0.8 (Fitzpatrick & Nix, 1969). In the current climex implementation, these parameters cannot be adjusted by the user.

For many insect species, temperature will be the most critical climate parameter at higher latitudes. For C. pomonella, the critical climex parameters are the lower temperature thresholds for development and lower optimal temperature for population growth (see Appendix). The lower threshold for development commonly used for C. pomonella is 10 °C (Glenn, 1922; Riedl & Croft, 1978; Richardson et al., 1982; Blago, 1992; Gottwald, 1996; Mani et al., 1997; Anon, 2000; Howell & Neven, 2000; Anon, 2001), whereas 15–27 °C is suggested as optimal temperature conditions for population growth (Anon, 2000; Howell & Neven, 2000), and 30–34 °C as an upper temperature threshold for population growth (Anon, 2000; Howell & Neven, 2000). Shel'deshova (1967) reported that the lower temperature threshold for development of C. pomonella should be below 10 °C, and Sæthre & Hofsvang (2002) suggested that 9 °C is a more appropriate threshold for northern populations of C. pomonella. It was therefore decided to investigate 9 °C and 10 °C as threshold temperatures in the climex model. The stress indices were only slightly changed, to better fit the parameters adjusted above (See Appendix). Diapause induction day-length is not important in areas where the codling moth has only one generation annually (Shel'deshova, 1967; Sæthre & Hofsvang, 2002), and the parameter was disabled (see Appendix). Shel'deshova (1967) suggested that the isoline of 600 degree-days (DD) above 10 °C in Eurasia and America coincides very closely with the actual limit of the codling moth but, because of the individual variability in naturally occurring populations, 750 DD might be necessary for complete development of the first generation. Blago (1992) reported that only a partial first generation of C. pomonella could be observed during 2 years of field trials conducted in Svelvik, Norway. Minimum DD requirements between 500 and 750 were therefore used as the input in the climex model at two threshold temperatures.

Climate data

Climate data were obtained from the Norwegian Crop Research Institute (NCRI). The Institute has been running a network of automatic weather stations since 1987. The network has expanded into a system currently recording weather observations from more than 50 locations (Sivertsen, 2000) and the objective of these agrometeorological stations is to monitor the weather in the most important agricultural areas of Norway. Although these observations are being intensively used as input to pest warning systems (Seem et al., 1991; Sivertsen, 2000), little attention has been paid to use these data as a source of climatological information. As most of these stations now have been in operation for over a decade, a considerable amount of information has accumulated.

Weather observations obtained by the national meteorological institutes in different countries are often obtained from stations located at exposed places (e.g. at lighthouses at the coast and at airports) and they may therefore be less useful for representing the agroclimate in the region. Thus, we found it appropriate to invest the effort needed in collecting and systematizing the weather observations obtained by the NCRI-network into a source of information for agroclimate in Norway. The weather observations were compiled from raw data containing hourly mean values for the different parameters, year and station. These data were imported and stored in a relational database (MySQL version 3.23, MySQL AB, Uppsala, Sweden) from which the data necessary for input into the climex model could be extracted.

Biological observation data

Data on the geographical distribution of C. pomonella in southern Norway were obtained from a recently published paper (Sæthre & Edland, 2001). The investigation covered 93 locations in 54 municipalities and represents detailed information on the distribution of C. pomonella. The parameter fitting described above was partly done against the known distribution of the species in Norway.

Data on seasonal occurrence of C. pomonella were obtained from the surveillance programme run by the Plant Protection Centre of NCRI since 1994. The surveillance programme uses delta traps baited with codling moth sex pheromone, codlemone, to monitor the codling moth populations in selected apple orchards. During the years 1997–2000, some of the delta traps were replaced with traps that timed the catches of individual moths. In the analysis of data from the surveillance programme, we used data from nine locations: one location in western Norway and eight in eastern Norway. The observations series from two of the eastern locations were average numbers from several pheromone traps at each location, whereas the data from the other locations were catch counts from one trap at each location.

It is important to emphasize that pheromone trap catches are not accurate measurements of population densities (either absolute or relative) at a chosen location. There are a number of factors influencing the number of male moths caught in a pheromone trap, such as moth density, immigration, temperature, moonlight, wind speed, trap and lure placement and maintenance, and competition between traps and calling females (Blomefield & Knight, 2000). However, the trapping method used in this study was standardized within and between years and locations, and used with awareness of the limitations of the trapping method.

Spatial analysis

The geographical distributions of bioclimatic potential for C. pomonella and L. decemlineata were predicted by climex for both species for 44 agro-meteorological stations of NCRI. The predictions were compared with the geographical distribution of C. pomonella in Norway, and climate response parameter values for the species were adjusted (see Appendix).

Temporal analysis: short-term population fluctuations and long-term effects of climate change

For the subsequent analysis, we ran the model under three climate scenarios named ‘greenhouse’, ‘increased rainfall’ and ‘greenhouse and increased rainfall’. In the greenhouse scenario the minimum and maximum temperatures were increased by 0.1 °C for each degree of latitude, whereas the increased rainfall scenario added 10 mm to the average weekly rainfall in the summer-half of the year. This is in accordance with the expected climate change scenarios for temperature, i.e. stronger warming at higher latitudes (IPPC, 2001a) and increased rainfall and drought in northern and southern Europe, respectively (IPPC, 2001b).

In order to analyse the short-term effects of temporal variations in climate, both for exploring the potential of possible interim establishments of L. decemlineata and the potential effects of climate on population dynamics of C. pomonella, series of annual EIs and GIs were predicted for both species. The indices were compared with data from the running surveillance programme for C. pomonella to study possible effects of climate on temporal population fluctuations. In addition, information from annual pest warnings disseminated by the Plant Protection Centre of NCRI in terms of recommendations of spraying with insecticides against the Apple fruit moth (Argyresthia conjugella Zell.), the key insect pest of apples in Norway, was included as another possible effect on temporal population changes of C. pomonella.

Results

Spatial distribution for C. pomonella

Model predictions for C. pomonella gave different ecoclimatic indices (EI) for the species, depending on the lower temperature threshold for population growth (DV0) and minimum day-degrees (PDD) used (Fig. 1). The predictions made by climex are shown both under the current climate (Fig. 1A–C) and under an increased temperature climate change scenario (Fig. 1D). When using the original climate response parameters given by climex (see Appendix), many of the locations predicted as suitable for C. pomonella were far out of the observed geographical range of the species (Fig. 1A). Adjusting moisture parameters (see Appendix) gave only minor changes to the predicted EIs and did not exclude any locations. On further adjustment of the climate response parameters (See Appendix), and especially by setting DV0 = 9 °C and PDD = 580, the predictions made by climex fitted the observed geographical range quite well (Fig. 1B). By setting DV0 = 10 °C and PDD = 600, only three locations were predicted as suitable for the species in Norway (Fig. 1C). The final set of climate parameters adjusted for C. pomonella is given in the Appendix (column 4). Under the greenhouse climate scenario (DV0 = 9 °C and PDD = 580), the predicted range of C. pomonella was extended to include the coastal areas from Stavanger (58.6°N, 5.5°E) in the south to Bodø (67.2°N, 14.3°E) in the north (Fig. 1D). The area in eastern Norway also increased towards the north. In addition, the magnitude of the EIs increased for locations where the species is already established (Figs 1B.D, Table 1).

Figure 1.

Ecoclimatic indices and observed geographical distribution for Cydia pomonella in Norway. (A) Climatic response parameters from the climex manual (Skarratt et al., 1995). (B) Adjusted parameters from Table 1 (DV0 = 9 °C, PDD = 580). (C) Adjusted parameters from Table 1 (except DV0 = 10 °C, PDD = 600). (A–C) Under the current climate. (D) Adjusted parameters from Table 1 (DV0 = 9 °C, PDD = 580) under the greenhouse climate scenario.

The bioclimatic potential (EI) for C. pomonella in Norway (DV0 = 9 °C and PDD = 580) (i.e. predictions based on average climate for the respective periods of operation of the 44 weather stations under different climate scenarios) are listed in Table 1. The predicted EIs for C. pomonella were in the range 0–35 under the current climate, where 23 locations were in the range 27–35, indicating a suboptimal climate for the species. Under a greenhouse scenario, the EIs were in the range 35–64 and included all the 44 locations (Table 1). The combined greenhouse and increased rainfall scenario gave approximately the same EIs as the greenhouse scenario alone, whereas the increased rainfall only scenario gave EIs almost identical to current climate (Table 1). It should be noted that for the locations where EI became zero, this was because the DD requirements in the model were not fulfilled. The four stress factors (heat, cold, wet and dry stress) were zero or very limited at all locations, and the indices for population growth (GI) were similar between nearby locations. By closer examination of the four indices used by climex to generate GI (TI, MI, LI and DI), it became evident that the major of these factors affecting GI (and thereby EI) was TI. At most locations investigated, MI, LI and DI were 100 or close to 100, making TI, GI (and EI) nearly identical within the respective locations and under the respective climate scenarios.

Table 1.  Ecoclimatic index (EI) for Cydia pomonella and Leptinotarsa decemlineata for 44 locations in the Norwegian agrometeorological network under current climate (CU), greenhouse (GR) and increased rainfall (IR) climate scenarios. Adjusted climate response parameters from Table 1 are used for C. pomonella. Years in operation (Years), latitude (°N), longitude (°E) and altitude above sea level (Asl) are shown for each location
     Cydia pomonellaLeptinotarsa decemlineata
LocationYears°N°EAslCUGRGR+IRIRCUGRGR+IRIR
Ås859.610.88927.449.249.227.4037.723.24.4
Alvdal862.110.6485034.135.60018.618.92.5
Apelsvoll1160.710.9255046.246.20031.717.33.8
Balestrand361.26.51531.657.557.531.6025.29.43.8
859.49.010028.544.746.628.5029.734.814.3
Etne459.76.01529.657.65729.507.42.41.7
Fåvang1161.510.7195041.943.40027.8286
Frosta763.510.7100049.649.6003713.33
Fureneset961.35.07059.358.4007.12.81
Gausdal861.210.3367041.541.50029.118.12.7
Gjerpen659.29.64130.250.450.430.2039.5284.8
Gran960.410.6240044.844.80028.932.39.1
Gvarv259.49.24634.852.552.634.816.240.720.45.7
Hjelmeland1059.26.160055.355.2009.441.6
Hønefoss960.110.312628.848.148.128.8036.635.910.2
Hokksund859.89.91530.141.646.831026.63817.2
Ilseng1060.811.2177042.943.10028.330.48.1
Kise1360.810.811827.7474727.703634.74.9
Kvam460.36.21333.160.959.83307.52.72
Kvithamar1363.510.940049.849.80034.29.22.7
Landvik1358.38.5530.754.554.530.7031.68.53.9
Lier959.810.36029.749.149.129.7038.327.85.2
Linge462.37.23532.163.863.832.1032.4104.2
Løken1161.19.1521040.340.30026.1262.3
Lyngdal558.17.1526.555.154.326.308.64.11.8
Mære863.911.456045.845.80034.8102.5
Njøs961.26.93526.952.452.426.9039.218.13.9
Rakkestad1059.411.4100046.547.10033.324.34.7
Ramnes1059.410.23028.550.150.128.5037.229.44.7
Rissa863.610.018048.648.60012.66.51.7
Roverud1060.312.115026.746.14726.7034.734.49.1
Rygge459.410.82931.8535331.8040.912.25.1
Særheim1158.85.787057.857.80013.582
Sande1059.610.25529.249.649.629.2038.730.94.9
Skjetlein863.310.380047.647.60035.68.82.1
Skogmo764.512.035042.542.50025.25.51.4
Surnadal863.08.710046.546.50018.74.71.6
Svelvik459.610.41030.750.950.930.703211.14.8
Tingvoll662.98.220053.353.30026.88.72.6
Tjølling959.010.1528.651.751.728.603815.44.6
Tjøtta965.812.410050.250.20020.17.71.5
Tomb1059.310.82028.750.350.528.7037.227.85.3
Ullensvang560.36.71332.458.658.632.403714.14.9
Vågønes567.314.540047.147.100115.90.8

Temporal variations in ecoclimatic indices for C. pomonella

The general trend in the period 1990–2000 was that model predictions (using adjusted parameters from Appendix) of EI and GI showed increasing values towards the end of the decade. Some of the observed data indicated a decreasing trend in population densities of C. pomonella from 1995 to 2001 (Fig. 2). In Fig. 2, the annual trap catches of C. pomonella at four locations were plotted with the respective EIs and GIs, but with a 1-year time lag. This time lag was imposed because there is evidence that the weather conditions the previous year (i.e. the conditions for mating-flight, oviposition and development of the larval stage of C. pomonella) were important for the population dynamics between years (Mani et al., 1997). In addition to the general trend described above, changes in the annual observations of C. pomonella to some extent coincided with changes in the predicted EIs (Fig. 2). For some years, the predicted EIs became zero, again caused by lack of fulfilment of the DD requirements of the model (see Appendix). However, by consulting the GIs, it can be seen that changes in climate between years were not as dramatic as the EI values indicated (Fig. 2). Years with recommendations of spraying against A. conjugella were added (with a 1 year time lag), to investigate the influence of spraying as a factor influencing changes in population densities of C. pomonella. For most locations studied, spraying against A. conjugella seemed to affect the abundance of C. pomonella the following season, and appears to over-rule the possible effects of climate.

Figure 2.

Temporal variations in ecoclimatic indices, growth indices (right axis) and pheromone trap catches in the number of specimens (Observations) (left axis) for Cydia pomonella at Sogndal, Hønefoss, Darbu and Lier. Years when spraying against Argyresthia conjugella was recommended (pillar) (Spraying). Ecoclimatic indices, growth indices and spraying with a time lag of 1 year.

Spatial distribution for L. decemlineata

The bioclimatic potentials (EI) for L. decemlineata in Norway predicted by climex under different climate scenarios are listed in Table 1. Under the current climate, the predicted EIs were zero for all locations, except Gvarv, indicating that if the pest entered Norway, it would not be able to establish. The predicted EIs were in the range 7–41 under a greenhouse scenario (Table 1, Fig. 3). The increase in EIs was due to both a predicted increase in growth index and a decrease in cold stress. The results indicated that establishment should be expected in eastern Norway and some fjord-districts in the west (Fig. 3), but only locations with EIs above 30 are considered favourable for L. decemlineata populations in the long term. The greenhouse and increased rainfall scenarios combined gave EIs in the range 6–36 (Table 1), indicating a less favourable climate compared to increase in temperature only. The increased rainfall scenario gave EIs in the range 1–17 (Table 1), but establishment of the species in the long term is not likely to occur.

Figure 3.

Ecoclimatic indices for Leptinotarsa decemlineata in Norway under a greenhouse climate scenario.

Temporal variations in bioclimatic potential for Colorado potato beetle

Model predictions based on annual climate data revealed interim periods with positive establishment potential for L. decemlineata (Table 1). In 1997, 56% of the locations had EIs in the range 5.4–19.9 and in 1999, 16% of the locations had EIs in the range 6.7–17.5. The annual EIs for L. decemlineata gave only sporadically non-zero values for the years 1992, 1994, 1995 and 2000 (Table 1).

Discussion

By adjusting the climate response parameters for C. pomonella, the spatial distribution predicted by climex fitted the observed distribution of the species very well (Fig. 1B). The distribution of apples, its major host plant, in Norway also fitted rather well with the predicted distribution of the species (Edland, 1997; Sæthre & Edland, 2001), when the adjusted set of parameters was used.

Shel'deshova (1967) reported that the codling moth was of significant importance as a pest in regions where the effective temperature sum exceeds 750 degree-days (threshold temperature 10 °C) (i.e. where complete development of the population is assured). According to climex, no locations fulfil these requirements in Norway (Fig. 1C). Blago (1992) reported that only a partial first generation of C. pomonella could be observed during 2 years of field trials conducted in Svelvik, Norway. Combining these two statements and the observed distribution of C. pomonella (Sæthre & Edland, 2001) with the predictions made by climex in this study (Fig. 1B, Table 1), it is reasonable to conclude that the species only occasionally is able to complete a full first generation at most locations in Norway. This means that the 580 degree-days suggested as input in the climex model is not enough heat summation for C. pomonella to complete its life-cycle, but describes the eco-climatic conditions at which the species is able to maintain a population in the long term. Further studies on a Norwegian population of the species will be needed to estimate the correct number of degree-days required for a complete generation of the species.

In some of the surveyed municipalities C. pomonella were not found. Climate is the major reason for this observed absence, particularly in Eastern Norway. In Western Norway, however, the situation might be different because migration of adult codling moths is likely to be inhibited by natural barriers such as fjords, valleys and high mountains. Cydia pomonella is considered to be a rather sedentary species, and therefore natural barriers are likely to impact its distribution in Western Norway (Sæthre & Edland, 2001).

TI, GI and EI for C. pomonella became nearly identical within the 44 locations, if the requirements of 580 DD were fulfilled. This clearly suggests that temperature is the major factor affecting the distribution of C. pomonella in Norway. EI values of more than 50 were not achieved for C. pomonella at any of the study locations under the current climate (Table 1), suggesting that the climate in southern Norway is only suboptimal for C. pomonella compared to the optimal target value of EI = 100. However, under the greenhouse scenario, most locations had EIs near or even above 50 (Table 1). This would have a major impact on the distribution of the species in Norway. The heat-summations obtained under a greenhouse scenario suggest that a second or partial second generation of the species could occur at several locations. An increase in temperature could cause a major change in the abundance and pest status of C. pomonella in the most favourable areas. This effect could be enhanced by the fact that the light conditions in Norway favour flight of the adult moths during the entire night but, under the current climate, they are limited by low temperatures (Sæthre & Hofsvang, 2002).

We suspected that the weather conditions the previous year (i.e. the conditions for mating-flight, oviposition and development of the larval stage of C. pomonella) were important for the population dynamics between years, and this was partly supported by the findings in this study. Mani et al. (1997) reported that population changes in C. pomonella from one year to another can largely be explained by summer weather. However, the number of years involved in the present study, and the method for measuring population changes used, were both major sources of uncertainty.

We also suspected that recommended spraying against A. conjugella, which was followed by almost all of the commercial growers in the years that attack was forecasted, was a very important factor affecting population densities of C. pomonella. The spraying in most years was conducted at a time that highly affected the egg stage and newly hatched larva of the codling moth before entering the fruit. Because of the suboptimal climate conditions at most locations, the effect of spraying was enhanced. The effect of climate would therefore be better estimated in unsprayed orchards. Worner (1991) reported that studies using time series analysis and multivariate techniques to analyse the influence of climate on insect populations continue to confirm that weather factors, far before population events of interest, have a significant influence on population fluctuations. Changes in the population dynamics of insects are caused by a number of factors but, at least in temperate climates, temperature is considered to be the most important factor, affecting developmental rate, fecundity and mortality (Worner, 1992). Due to the lack of basic data in the present study, the importance of other factors is difficult to assess.

The predicted results on spatial bioclimatic potential for L. decemlineata showed that the climate is not suitable for long-term establishment of the species. Model predictions based on annual climate data revealed interim periods with positive establishment potential for L. decemlineata (Table 1), in accordance with the results reported by Hofsvang (1996).

The climate change scenarios clearly indicated that a temperature increase could provide a shift in the establishment conditions from non-favourable to favourable for L. decemlineata at several locations. The main host plant, potato, is available in all rural districts of Norway (Hofsvang, 1996), and host availability will not be a limiting factor.

On a long-term temporal scale, the effects of a possible climate change represent an interesting aspect of study for which the climex model has been used. Sutherst et al. (1995) provided a broad discussion of different modelling approaches, and they demonstrated the use of climex for estimating impacts of climate change on the potential geographical distribution of L. decemlineata in Europe. Worner (1988) used climex as a means of integrating information about the response of an insect to its environment which, in its raw state, was unmanageable and incomplete. In the present study, an initial approach that proved useful was to investigate the spatial pattern of the model results and validate them against observed data on the geographical distribution of C. pomonella. Further analysis of climate response for use in the parameter fitting process, and the subsequent analysis of model scenarios of climate change and of temporal variations in bioclimatic potential, are adjustments of parameter values of the model. In that context, all variability in model inputs could be interpreted as sensitivity analyses of the model. In general, the results showed that this modelling framework could successfully be used even for species that tend to be more single climate factor dependent such as C. pomonella.

The value of this kind of analysis depends on the quality of available data and our understanding of the biological system. By comparing environmental knowledge of an area (e.g. climate data) with biological information of a species, we investigate the potential effects of climate on the spatial and temporal abundance of the species in an area. However, the effects of climate will at best constitute only a partial explanation of the patterns that can, or will, be observed in species abundance. Hence, it may be helpful to sort out the sources of uncertainty involved. In the field of risk analysis, a distinction is commonly made between two kinds of uncertainty, epistemic and stochastic uncertainty (Hoffman & Hammonds, 1994). Applying this distinction to the current problems reveals that stochastic uncertainty (e.g. the natural weather variations) can result in short-term fluctuations in population sizes of a pest species such as C. pomonella, and even in the potential for interim population establishments of exotic pests such as L. decemlineata. Quantification of such natural variation may provide information of even greater value (e.g. for risk management of extremes in pest population sizes) than information only concerning how things perform on average. The other kind of uncertainty, epistemic uncertainty, is present both in our lack of knowledge about the pest response to climate and in our knowledge of the environmental conditions of the study area. That is the primary source of uncertainty behind questions such as ‘given that a pest has entered an area, is the species likely to become established?’ In the present work, our efforts in systematizing weather data from the study area and in the parameter fitting process for climate response of C. pomonella are both activities that reduce epistemic uncertainty.

A key problem with models estimating climate response parameters based on the observed geographical distribution of the species (e.g. climex) is that the parameters will be dependent of these observations. This makes the model predictions difficult to validate when observations from new areas are obtained. The problem arises when a survey in a new area reveals a species to be present there, whereas the model has predicted the environmental conditions to be unsuitable for that species in that area. Two alternative explanations exist in that case. Either our biological knowledge of the species has been limited or we may have revealed an instance where the species recently has adapted to the local environmental conditions. However, regardless of which explanation is true, our biological knowledge of the species has been updated by the new findings. In models such as climex, the question will always remain when it comes to fitting climate tolerance parameters for a species whether, or to what extent, a species has reached its maximum geographical distribution. Although this is an inherent weakness for the use of such models for prediction, their use could positively be considered as a common framework, and a continuous process, for updating our biological information about the species. Although the latter enables studies of temporal evolution of our knowledge of the species, we should be aware of the risk of confounding our knowledge development with environmental adaptation and the possible evolution of the species itself.

Acknowledgements

We are grateful to Håkon Magnus, Trond Hofsvang, Nina Svae Johansen and Leif Sundheim for their critical reading and helpful comments on a previous version of the manuscript. This work was partly funded by the Research Council of Norway.

Accepted 12 June 2002

Ancillary