Impact of Climate Change on Developmental Dynamics of Thrips tabaci

PHYSIOLOGICAL ECOLOGY

Impact of Climate Change on Developmental Dynamics of Thrips tabaci (Thysanoptera: Thripidae): Can It Be Quantified? KLEMEN BERGANT,1,2 STANISLAV TRDAN,2 DRAGAN ZˇNIDARCˇICˇ,2 ZALIKA CˇREPINSˇEK,2 LUCˇKA KAJFEZˇ-BOGATAJ2

AND

Environ. Entomol. 34(4): 755Ð766 (2005)

ABSTRACT We attempt to quantify the impact of future climate change on the developmental dynamics of onion thrips in Slovenia. Monthly averaged results of simulations of future climate from four different general circulation models (GCMs) were projected to local scale by empirical downscaling. The GCM simulations were based on two emission scenarios (IPCC SRES A2 and B2). Local estimates of monthly averaged air temperatures for Þve locations in Slovenia were adjusted for an additional four emission scenarios (SRES A1T, A1F1, A1B, and B1) using a pattern scaling technique. They were further transferred to a daily scale using a Þrst-order autoregressive model. A simple degree-day model, based on data reported in the literature, was used to relate the development of onion thrips to temperature. Potential changes in the period with favorable developmental conditions for onion thrips (i.e., temperatures above the lower developmental threshold) and in the number of generations per season were estimated with regard to the expected future climate change in Slovenia. The changes are inßuenced by the magnitude of temperature increase, its asymmetry within the year, and present climate conditions. Using this approach, one can obtain quantitative estimates of the impact of climate change on the developmental dynamics of an insect pest, but one must be fully cognizant of all the assumptions made in the procedure, which introduce uncertainties in the Þnal results. Further research is needed to evaluate the plausibility of such simpliÞed projections. KEY WORDS climate change, downscaling, degree-days, developmental dynamics, Thrips tabaci

ONION THRIPS (Thrips tabaci Lindeman; Thysanoptera: Thripidae) are a polyphagous pest that causes serious damage on vegetables and ornamentals all over the world (Murai 2000). It is especially important as a pest on Alliaceae plants such as onions (Allium cepa L.) and leeks (Allium porrum L.) (McKenzie et al. 1993, Theunissen and Schelling 1998), as well as on Brassicaceae plants such as cabbage (Brassica oleracea L.) (Shelton et al. 1998). The nymphs and the adults feed mostly on green leaf tissue, causing direct damage by destroying epidermal cells. They feed by piercing surface tissues and imbibing exuded cellular contents (Koschier et al. 2002). The empty cells on attacked plants create silvery-white spots, referred to as silver damage, that make the plants less marketable. Onion thrips are also an important vector of several plant viruses such as tomato spotted wilt virus (Kritzman et al. 2002, Jenser et al. 2003), tobacco streak virus (Sdoodee and Teakle 1987), and sowbane mosaic virus (Hardy and Teakle 1992). The reproduction and the development of onion thrips are strongly related to environmental conditions, especially temperature (Edelson and Magaro 1 Corresponding author: Centre for Atmospheric Research, Nova Gorica Polytechnic, Vipavska 13, SI-5000 Nova Gorica, Slovenia (email: [email protected]). 2 Department of Agronomy, Biotechnical Faculty, University of Ljubljana, Jamnikarjeva 101, SI-1111 Ljubljana, Slovenia.

1988). Photoperiod does not induce diapause in the development of onion thrips, but the adults exhibit temperature-induced reproductive quiescence (Jenser and Sze´ na´si 2004). In temperate zones, cold weather delays the production of eggs, and breeding continues when warm weather arrives in the spring (Sites and Chambers 1990). In agriculturally important areas of Slovenia, under the present climate, onion thrips usually become active in the spring (March or April) and are present until autumn (October or November), with the population peak in the summer (June and July). They survive the winter in the adult stage, concealed in plant debris. Because warming temperatures inßuence population growth rates of insects through decreased cold-related mortality (Bentz and Mullins 1999) and shorter generation times (Ungerer et al. 1999), the population dynamics of onion thrips in Slovenia might change in the future. Determining developmental and reproductive responses to different temperatures is vital for understanding and predicting the life history and population dynamics of onion thrips (Murai 2000) in various climate conditions. A common approach in predicting developmental dynamics and migration of insects in relation to weather conditions involves the use of degree-day models (Bryant et al. 1998, Roltsch et al. 1999). In agriculture, cumulative degree-days are used as a mea-

0046-225X/05/0755Ð0766$04.00/0 䉷 2005 Entomological Society of America

756

ENVIRONMENTAL ENTOMOLOGY

Vol. 34, no. 4

Table 1. Selected meteorological stations in Slovenia together with geographical data (␭g, longitude; ␸g, latitude; zg, altitude), climate type, average air temperature in the warm and cold halves of the year, and the period from which data was drawn for use in this study Meteorological stations Ljubljana Novo mesto Murska Sobota Ratee Bilje

Geographical data

␭g

␸g

zg

14.52⬚ E 15.18⬚ E 16.18⬚ E 13.72⬚ E 13.63⬚ E

46.07⬚ N 45.80⬚ N 46.65⬚ N 46.50⬚ N 45.90⬚ N

299 m 220 m 188 m 864 m 55 m

Period

Climate

1951Ð2002 1951Ð2002 1951Ð2002 1951Ð2002 1962Ð2002

Temperate continental Temperate continental Temperate continental Sub-Alpine Sub-Mediterranean

Average air tempa Warm half

Cold half

16.1⬚ C 15.7⬚ C 15.7⬚ C 11.9⬚ C 17.4⬚ C

3.5⬚ C 3.2⬚ C 2.7⬚ C ⫺0.4⬚ C 6.2⬚ C

a

Averages are related to the reference period 1961Ð1990.

sure of accumulated heat energy over a selected period and can be related to developmental and growth processes (McMaster and Wilhelm 1997, Bonhomme 2000, Cesaraccio et al. 2001). A linear relationship between developmental rate of insects and temperature is presumed in degree-day models within the range of ecologically relevant temperatures (Honeˇ k 1996, Bonhomme 2000). However, the relationship is nonlinear over large temperature variations (Briere and Pracros 1998). Different degree-day models for onion thrips (Edelson and Magaro 1988, Murai 2000, Stacey and Fellowes 2002, Khani et al. 2004) indicate a possible capacity for adaptation to local climate conditions (Stacey and Fellowes 2002), and consequently to changed climate. The major inßuence of a warming climate apparently will be changed population growth rates, and this needs to be assessed on an individual species basis (Logan et al. 2003). The expected increase of air temperature (Houghton et al. 2001) will speed up the life cycle of onion thrips and prolong the period of favorable developmental conditions, perhaps resulting in a greater number of generations during the growing season. This insect may also emerge as a pest in areas where current environmental conditions are not favorable for growth of their host plants. Quantitative estimates of regional temperature change are needed to use a degree-day model for estimating the impact of climate change on onion thrips development. Such temperature change estimates are usually based on the results of simulations with general circulation models (GCMs). On a large scale, GCMs reliably simulate the most important mean features of global climate (Zorita and von Storch 1999). The direct application of GCM results is limited because the Þnest spatial scale used for global climate simulation is too coarse for meaningful ecological applications (Grotch and MacCracken 1991, von Storch 1995, Logan et al. 2003). However, the results of GCM can be linked to local climate characteristics if the local climate is affected by large-scale features (Benestad et al. 2002). Empirical downscaling (e.g., Wilby and Wigley 1997, Zorita and von Storch 1999) has been widely used to bridge this gap between the large-scale (results of GCMs) and local-scale (input data needed in impact studies). The basic idea behind empirical downscaling is to use the observed relationships between the large-scale climate parameter (predictor) and the local-scale climate or climate-dependent parameter (predictand) to project GCM results to a regional or a local scale.

This paper presents an attempt to quantify the potential impact of climate change on the developmental dynamics of onion thrips, based on available climate change simulations with GCMs, their projections to climatologically heterogeneous locations in Slovenia, and a simple degree-day model. Large-scale monthly averages of near-ground air temperature (tmnt) and sea level pressure (pmnt) derived from GCM simulations were projected to local air temperature (lmnt). The estimated lmnt data were used as input for a simple stochastic generator of daily air temperatures (lday). The latter were fed into the degree-day model to estimate the number of generations of onion thrips per year (ngen). A change in the number of generations was used as a qualitative assessment of potential climate change impact on the harmfulness of onion thrips in the 21st century. The change in the length of the period (PER) with favorable temperature conditions for onion thrips development also was estimated to assess the potential impact of climate change on the developmental dynamics of onion thrips. Materials and Methods Monitoring of Onion Thrips (Thrips tabaci Lindeman). To illustrate population dynamics of onion thrips over a year, we used monitoring results of this pest in an onion Þeld in Ljubljana (see Table 1 for detailed information on location). The monitoring was performed using light-blue sticky boards (11 by 14 cm, Unichem d.o.o., Sinja Gorica, Slovenia), the most common tool for detecting economically important Thysanoptera species (Brødsgaard 1989, Trdan et al. 2005). The Þrst boards were set up in early April 1999, which is slightly later than the usual Þrst emergence of adults of onion thrips in Ljubljana. Four boards were placed at selected locations in the Þeld and were changed approximately twice a month until the end of December 2000. The boards were refrigerated before counting thrips under a stereo-microscope (Olympus SZ30) at ⫻15 magniÞcation. The average daily catch (cday) among all four boards was used as a measure of population size. Large-scale Climate Variables. Average monthly near-ground air temperature (tmnt) and sea level pressure (pmnt) were used as predictor data for empirical downscaling. In the case of observed and most modeled temperature data in this paper, the term of near ground refers to 2 m above ground level. The only exception is the HadCM model, where the near ground

August 2005

BERGANT ET AL.: ONION THRIPS AND CLIMATE CHANGE

757

Table 2. General circulation models, the results of which were used in this study: model label, country of development, period for which the data were used, the approximate horizontal resolution of data, and some references for the models and/or simulations Model

Country

Period

Resolution

References

CSIRO/Mk2 UKMO/HadCM3 DOE-NCAR/PCM ECHAM4-OPYC3a

Australia United Kingdom United States Germany

1961Ð2100 1951Ð2099 1961Ð2099 1951Ð2100

5.6 ⫻ 3.2⬚ 3.8 ⫻ 2.5⬚ 2.8 ⫻ 2.8⬚ 2.8 ⫻ 2.8⬚

Gordon and OÕFarrel 1997 Pope et al. 2000, Gordon et al. 2000 Washington et al. 2000 Roeckner et al. 1996, Stendell et al. 2000

a

The entire name of the model includes the label of development centers (MPI-DMI/ECHAM4-OPYC3).

air temperature refers to 1.5 m above ground level. Observed tmnt and pmnt data for the period 1951Ð2002 were extracted from the dataset of a joint project of the U.S. National Centers for Environmental Prediction (NCEP) and the National Center for Atmospheric Research (NCAR) (Kalnay et al. 1996, Kistler et al. 2001). The data set is usually referred to as the NCEP/NCAR reanalysis data set (http://ingrid.ldgo.columbia.edu/SOURCES/ .NOAA/.NCEP-NCAR/.CDAS-1/.MONTHLY/). The horizontal resolution of observed tmnt is ⬇1.9 by 1.9⬚ and of pmnt is 2.5 by 2.5⬚. Modeled tmnt and pmnt data were generated in simulations with four different GCMs (results of simulations are available from IPCC Data Distribution Center: http://ipcc-ddc.cru.uea.ac.uk/dkrz/ dkrz_indeks.html). Details about selected GCMs (common label, country of development, approximate horizontal resolution, and some references) are given in Table 2. For the simulation of future global climate, the GCMs were forced by A2 and B2 marker emission scenarios developed by the Intergovernmental Panel on Climate Change (IPCC) and reported in the Second Report on Emission Scenarios (SRES; Nakicénovic´ et al. 2000, Swart et al. 2002). The SRES scenarios are based on different assumptions of future social, economic, and technological development (Nakicénovic´ et al. 2000). Because of the methodology used for empirical downscaling, all large-scale data were interpolated to a common grid of 2.5 by 2.5⬚ by a simple bilinear interpolation method (Press et al. 2001). The predictor data were extracted over the area from 35⬚ N to 65⬚ N and from 10⬚ W to 30⬚ E. The predictor area was chosen as a compromise between the skillful scale of GCMs, at which their results are reliable (Grotch and MacCracken 1991, von Storch et al. 1993), and the quality of empirical models, which usually decreases as the predictor area gets larger (Benestad 2001a, b). Local-Scale Climate Variable. Near ground air temperature was used as a local climate variable. Average daily values (lday) were used for constructing the Table 3. in Slovenia

temperature generator and estimating degree-days, and the average monthly values (lmnt) were used as predictor values in empirical downscaling. Five locations were included in our study (Table 1). All local temperatures were collected and quality checked by the Environmental Agency of the Republic of Slovenia. Empirical Downscaling. The regression approach to empirical downscaling was used to relate average monthly near ground air temperature (lmnt) at selected locations with large-scale near ground air temperature (tmnt) and sea level pressure (pmnt) Þelds. Simultaneously observed predictor values tmnt and pmnt were organized in matrixes Tmnt and Pmnt, where rows represent observations at different times and columns at different grid points. The columns of Tmnt and Pmnt subsequently were combined in a common matrix Xmnt. Multi-way partial least squares regression (e.g., Bro 1996, Smilde 1997, de Jong 1998) was used to estimate weight matrix W to extract important features from Xmnt, represented by the matrix S ⫽ XmntW, and to estimate the regression coefÞcients b relating S to the time series of lmnt as lmnt ⫽ Sb. A detailed description of the downscaling procedure (preprocessing of the data and EM development) can be found in Bergant and Kajfezˇ-Bogataj (2005). The regression models were developed separately for single months using available observations for large-scale and local-scale data. The variability of lmnt explained by regression models is shown in Table 3. The models were used afterward for the projection of available large-scale GCM data (Table 1) based on SRES A2 and B2. The variability of lmnt not described by the regression model was treated as white noise, with zero mean and variance of the residuals between observations and model estimates of lmnt (von Storch 1999). Such randomized estimates of lmnt exhibit similar temporal variability as that observed for lmnt that is only partially controlled by the large-scale dynamics.

Coefficients of determination (R2) of empirical downscaling models for average monthly air temperature (lmnt) at five locations

Location/mo

1

2

3

4

5

6

7

8

9

10

11

12

Ljubljana Novo mesto Murska Sobota Ratee Bilje

0.87 0.86 0.84 0.81 0.92

0.83 0.89 0.85 0.85 0.82

0.93 0.88 0.86 0.92 0.92

0.95 0.91 0.92 0.91 0.76

0.91 0.91 0.93 0.93 0.98

0.96 0.95 0.93 0.98 0.98

0.90 0.91 0.90 0.92 0.82

0.93 0.94 0.93 0.93 0.77

0.94 0.94 0.91 0.96 0.90

0.90 0.91 0.81 0.95 0.73

0.82 0.78 0.78 0.90 0.92

0.76 0.79 0.74 0.76 0.82

Models are based on large-scale average monthly sea level pressure (pmnt) and air temperature (tmnt) and were developed separately for single months.

758 Table 4.


Vol. 34, no. 4

Autocorrelation coefficients (r) for local daily air temperature (lday) at five locations in Slovenia

Location/mo

1

2

3

4

5

6

7

8

9

10

11

12

Ljubljana Novo mesto Murska Sobota Ratee Bilje

0.68 0.72 0.63 0.70 0.56

0.66 0.69 0.65 0.68 0.52

0.68 0.70 0.71 0.70 0.48

0.60 0.62 0.62 0.67 0.50

0.62 0.60 0.62 0.67 0.50

0.66 0.64 0.64 0.72 0.59

0.67 0.62 0.58 0.68 0.54

0.66 0.66 0.65 0.70 0.52

0.71 0.73 0.70 0.72 0.52

0.74 0.74 0.72 0.75 0.59

0.70 0.70 0.66 0.71 0.65

0.69 0.73 0.66 0.69 0.58

Values were estimated separately for single months from data for the period of 1951Ð2002.

Pattern-Scaling. A pattern-scaling technique (Mitchell 2003) was used to adapt the results of GCM simulations and their local projections based on A2 and B2 marker SRES to other marker SRES. The local response of average monthly near ground air temperature (lmnt) to a selected SRES scenario was estimated by assuming a linear response between the amount of global warming and local climate response. Knowing the estimate of local warming under a reference emission scenario (e.g., SRES A2 or SRES B2; ⌬lref), together with the estimate of global warming under a reference (⌬gref) and a selected emission scenario (⌬gsel), an estimate of local warming under a selected scenario (⌬lsel) can be obtained (⌬lsel/⌬lref ⫽ ⌬gsel/ ⌬gref). In this manner, downscaled results for lmnt under SRES A2 and B2 scenarios were adapted to SRES A1T, A1B, A1Fl, and B1 using global warming ratios reported in Bergant et al. (2005). Daily Temperature Generator. A simple stochastic temperature generator (Wilks and Wilby 1999), describing average daily temperature (lday) as a Þrstorder autoregressive process, was used to relate the monthly and daily statistics of local near ground air temperature (Yu et al. 2002). Because lday is strongly autocorrelated (Table 4), it can be estimated based on its value for a previous day (lprv), the autocorrelation coefÞcient between lday and lprv (r), and the monthly average and SD of lday (lmnt and smnt) as lday ⫺ lmnt ⫽ r(lprv ⫺ lmnt) ⫹ (1 Ðr2)1/2N(0,smnt). The partition of the variability not explained by the autoregressive model was treated as white noise, with zero mean and smnt SD [N(0,smnt)]. Synthetic lday under climate change conditions can be generated using empirical downscaling results for lmnt, assuming that the autocorrelation of lday, the weak relationship between lmnt and smnt (smnt ⫽ a1lmnt ⫹ a0) (Yu et al. 2002) and the contribution of noise (von Storch 1999) to the lmnt and smnt will remain the same in the future. Derived lday values can be used for the estimation of degree-days as a measure of pest developmental rate. Thirty different realizations of stochastic part in the temperature generator were used to estimate synthetic lday, which were further used for the estimation of degree-days and length of period with favorable conditions for development of onion thrips (PER). A more detailed description of the daily air temperature generator can be found in Bergant et al. (2005). Degree-Day Model. Temperature is an important factor of developmental time in insects. Over a range of ecologically signiÞcant temperatures, develop-

mental rate (d)Ñthe reciprocal of developmental time (in days)Ñincreases linearly with temperature (l) (Honeˇ k 1996). Such a linear relationship (d ⫽ b0 ⫹ b1l) forms the basis for calculating the thermal constants for development: the lower developmental threshold (LDT; the temperature at which the development ceases) and the sum of effective temperature (SET; the cumulative degree-days above LDT needed to complete development). The LDT can be estimated from the linear relationship as the ratio Ð b0/b1, and SET can be estimated as 1/b1 (e.g., Honeˇ k 1999, Jarosˇ õ´k et al. 2004). Given SET and LDT estimates, development time is easily predicted under a variety of conditions, including the natural course of weather (Honeˇ k 1996). Different approaches can be used to calculate degree-days (Roltsch et al. 1999). In our case, positive differences between average daily air temperature and LDT (lday ⫺ LDT) were used for the cumulative degree-days in a selected year. Different estimates of SET and LDT for the life cycle of onion thrips can be found in the literature. Experimental data reported by Edelson and Magaro (1988), van Rijn et al. (1995), Murai (2000), Stacey and Fellowes (2002), and Khani et al. (2004) were used to estimate a hybrid model (LDT, SET). This model was used to estimate the number of generations per season (ngen) by dividing the cumulative annual degree-days with the SET requirement for one life cycle of onion thrips. The effect of potential climate change on the length of the period (PER) with lday higher than LDT also was investigated, in addition to the change in ngen, to asses the potential impact of climate change on developmental dynamics of onion thrips under more favorable conditions. Results The impact of climate change on PER and cumulative degree-days is symbolically shown in Fig. 1 for different types of warming: symmetric warming over the entire year and more/less intensive warming in the cold/warm seasons. The estimated times for the total life cycle of onion thrips and corresponding developmental rates at different temperatures reported in the literature were used to develop a hybrid degree-days model (Fig. 2). Because the linear relationship between d and l is valid only in the range of temperatures where the complex of life functions and Þtness of the organism are not dramatically reduced (Honeˇ k 1999), the values of d at

August 2005


Fig. 1. Impact of warming on the length of the period (PER) with conditions favorable for onion thrips development, and cumulative degree-days (DD) above the lower developmental threshold: present state (thick solid line), more intensive warming in cold half of the year (dotted line), more intensive warming in warm half of the year (dashed line), and symmetric warming over the entire year (thin solid line).

the lowest (⬍15⬚) and the highest values of l (⬎28⬚C) were not used for estimating LDT and SET. Thus, the estimated thermal constants of the onion thrips life cycle based on a linear model (r2 ⫽ 84%) are SET ⫽ 249.9⬚C above an LDT ⫽ 7.9⬚C. Although it is difÞcult to draw any conclusions about thermal constants based on population monitoring, an example of onion thrips abundance in Ljubljana (Fig. 3) alongside observed temperatures suggest that 7.9⬚C is a reasonable LDT value for the temperate climatic region of Slovenia. In 1999 and 2000 in Ljubljana, the periods when the air temperature exceeded the LDT coincided with the occurrence of onion thrips in the spring and their disappearance in the autumn. The reason for a large, almost d = 0.004l −0.0316, r2 = 84%, SET = 249.9°C, LDT = 7.9°C

0.12 0.1

Edelson & Magaro, 1988 Khani et al., 2004 Murai, 2000 van Rijn et al., 1995 Stacey & Fellowes, 2002

d [day−1]

0.08 0.06 0.04 0.02 0 5

10

15

20 l [°C]

25

30

35

Fig. 2. Linear relationship between mean developmental rate (d) and temperature (l), together with estimated lower developmental threshold (LDT) and sum of effective temperatures (SET) needed for completion of one lifecycle of onion thrips based on data reported in literature.

759

10-fold difference in onion thrips population in 1999 and 2000 in Ljubljana (Fig. 3) is not only the difference in air temperature but also the difference in precipitation, because abundant precipitation can affect mortality of insects (Patterson et al. 1999, Norris et al. 2002). Average air temperature from the beginning of April through the end of June, when the population becomes abundant, was 15.8⬚C in 1999 and 17.2⬚C in 2000. Because of the higher temperatures, the development of onion thips in late spring and early summer was more rapid in 2000 than in 1999. However, the difference in cumulative precipitation in the same period of the year was very large: 461 mm in 1999 and 261 mm in 2000. High precipitation in 1999 inßuenced the survival of onion thrips and reduced the population, whereas 2000 was a dry year with more favorable conditions for onion thrips development. To use the degree-days model to estimate the potential number of onion thrips generations per season (ngen) under conditions of climate change, reasonable estimates of daily near ground air temperatures (lday) are needed. They were produced by using empirical downscaling models (EM) and stochastic generators. Because the EMs based on the NPLS method explain a great part of lmnt variability at selected locations (Table 3), they can be used to project the results of GCMs to a local level. The projections of future changes in lmnt can be assessed using different GCMs, SRESs, and EMs (Fig. 4). With the exception of Bilje, a similar increase of air temperature (⌬l) over that of the period 1961Ð1990 was projected for the selected locations (Fig. 4). The projections of ⌬l for the cold half of the year (October to March) are higher than for the warm half (April to September). In the case of Bilje, the situation is reversed. Different projections of ⌬l for Bilje compared with the other locations could be related to a different regional response of the Vipava valley to large scale climate changes. Nevertheless, the results for Bilje should be treated with some caution, because the meteorological station was moved at the beginning of the 1990s and was transformed from a climatological to a synoptic system. Also, the period of available temperature observations is shorter for Bilje than for the other locations (Table 2), which also can affect the quality of EMs and their projections. An increase of air temperature will inßuence the length of the period with temperatures favorable for onion thrips development (PER) and the number of generations of onion thrips per season (ngen) related to cumulative degree-days. The effect of warming on changes in PER and degree-days depends on its asymmetry over the year and on present climate conditions. Different intensities of warming in the cold and warm season result in different projections of PER and degree-days changes (Fig. 1). The situation for Bilje is close to that represented by the dashed line in Fig. 1, where there is more intensive warming in the warm half of the year and less in the cold half. Less intensive warming projected for the cold half of the year results in a less distinctive prolongation of PER compared with that projected for the other locations (Ljubljana,

760


Vol. 34, no. 4

Fig. 3. Mean daily catch of onion thrips at Ljubljana, Slovenia in 1999 (left) and 2000 (right) together with mean daily air temperatures, a 7-d moving average of temperatures and estimated lower developmental threshold (LDT) of 7.9⬚C, and cumulative daily precipitation.

Novo mesto, Murska Sobota, and Ratecˇ e; Fig. 5). Asymmetry in warming over the year at these latter locations is similar to that represented by the dotted line in Fig. 1. The projections of PER changes are very similar for Ljubljana, Novo mesto, and Murska SobotaÑlocations with similar present temperature conditions (Table 1). In contrast, the projections of PER changes for Ratecˇ e show a wider range of uncertainty, despite a similar expected warming. Apart from the impact on PER, present climate conditions and the asymmetry of warming also impacted predictions of cumulative annual degree-days and consecutive estimates of ngen (Fig. 6; Table 5). The largest increase in degree-days and ngen was estimated for Bilje, which has the warmest climate among the selected locations and the highest projection of ⌬l in the warm half of the year. However, the lowest increase in degree-days and ngen was projected for Ratecˇ e, which has the coldest climate among the selected locations. The projections are similar for Ljubljana, Novo mesto, and Murska Sobota, because their present climate and expected warming are comparable. Discussion Using results from simulations of future climate change, projecting the results to a regional level and

feeding them into models of insect development and population dynamics can provide quantitative estimates of the impact of climate change on pest insects. However, the question is how reliable are such estimates, given that many assumptions must be made in this procedure. Our results show that the expected increase in air temperature will result in a larger accumulation of degree-days and a longer period with temperatures above the estimated lower threshold for onion thrips development. However, we have to be aware that such quantitative estimates are burdened with a large amount of uncertainty related to differences in basic assumptions about future emissions of greenhouse gasses and aerosols and their impact on climate on different scales, as well as the relationship between climate and pest development (see Bergant et al. 2005). It is even less certain how the changes in cumulative degree-days and favorable development period will affect the harmfulness of onion thrips. Relating the development of onion thrips and their population dynamics only to temperature conditions is a major simpliÞcation of reality. Temperature does not act in isolation to inßuence pest status, and it is important to consider interactions with other variables, such as rainfall, humidity, irradiance, and carbon dioxide concentrations (Harrington et al. 2001). This was conÞrmed by the results of our monitoring of onion thrips in two climatologically different years,

August 2005


761

Fig. 4. Projections of air temperature change in the 21st century from that observed in 1961Ð1990 at the indicated locations in Slovenia. Changes are presented as 30-yr averages with steps of 10 yr. Dashed and dotted lines with different markers correspond to different marker SRES scenarios averaged across four GCM; the solid line indicates their average, and the gray area represents the range of changes across all SRES and GCM.

762


Vol. 34, no. 4

Fig. 6. Projected annual cumulative degree-days (DD) above the lower developmental threshold (LDT) of 7.9⬚C together with estimated number of generations (ngen) of onion thrips at Þve locations in Slovenia. Data are presented as in Fig. 4. Fig. 5. Projected length of the period (PER) of temperatures favorable for onion thrips development (above LTD ⫽ 7.9⬚C) at Þve locations in Slovenia. Data are presented as in Fig. 4.

August 2005


763

Table 5. Estimated average (Ave.) no. of generations of onion thrips per year (ngen) at five locations in Slovenia under the present climate (1961–1990), and in future climates (2021–2050 and 2051–2080), together with a minimum (Min.) and maximum (Max.) no. under different emission scenarios Period Location Ljubljana Novo mesto Murska Sobota Ratee Bilje

1961Ð1990

2021Ð2050

2051Ð2080

Ave.

Ave.

Min.

Max.

Ave.

Min.

Max.

6.6 6.6 6.3 3.5 7.5

8.2 8.1 8.0 4.9 9.7

7.3 7.1 7.1 5.1 8.5

9.5 9.5 9.3 5.9 11.4

9.3 9.4 9.3 5.8 11.2

7.6 7.5 7.4 4.4 9.0

12.0 12.4 12.0 8.2 14.4

1999 and 2000. Additionally, thrips in the Þeld can be exposed to signiÞcantly higher temperatures than those measured by the shaded meteorological instruments. Using standard meteorological data for the estimation of degree-days can result in underestimation of insect development. It is also known that the developmental rate is a nonlinear function of temperature (e.g., Logan et al. 1976, Sharpe and DeMichele 1977, Briere and Pracros 1998). Linear approximation for this function could result in underestimated development at low temperatures and overestimated development at high temperatures. If an expected increase in temperature results in a nonlinear response, the increase in the number of generations of onion thrips per year will be less extensive than the increase predicted by the change in cumulative degree-days based on the linear model. Additionally, not only changes in mean temperature, which were considered in our study, but also changes in variability (e.g., diurnal range, frequency of extremes) can have major impacts on insect populations (Patterson et al. 1999, Harrington et al. 2001). When using average daily temperatures, we underestimated development on days when the maximum temperature was above the developmental threshold but the mean temperature was below it. However, without estimates of future changes in the diurnal temperature range, additional assumptions are required that will not necessarily decrease the uncertainty in the Þnal results. Although temperature is the main factor affecting onion thrips development, it is an open question of whether the thermal constants are truly constant in different climatic conditions. Because of their mobility and fast reproduction, insect pests are likely to adapt to different (new) environmental conditions (Harrington et al. 2001). Human-induced local climate change, as has occurred within urban areas, already provides evidence of how insects can adapt to changes in their environment (Patterson et al. 1999). Different values for thermal constants of onion thrips, reported in the literature (Edelson and Magaro 1988, Murai 2000, Stacey and Fellowes 2002, Khani et al. 2004), could be partially related to the adaptation capacity of onion thrips. However, because the amount and quality of available food also have an important impact on population dynamics of onion thrips (Milne and Walter 1998, Murai 2000, Stacey and Fellowes 2002), this could be another reason for differences in reported values.

Nevertheless, we can expect an increase in the number of generations of onion thrips per season because of the big predicted increase in cumulative degreedays and the prolongation of the period with favorable developmental conditions. More generations could lead to larger populations and consequently more damage on host plants. In Slovenia, especially the areas already grappling with the problem of onion thrips (i.e., southwest Slovenia represented by Bilje, northeast Slovenia represented by Murska Sobota, and central Slovenia represented by Ljubljana and Novo mesto) could be affected. The onion thrips also could become a serious threat in areas where it is not currently a problem (i.e., northwest Slovenia represented by Ratecˇ e). A decrease in longevity of onion thrips with increasing temperature may compensate to some extent for the positive effect of increased air temperature on growth of onion thrips populations with regard to its harmfulness (Murai 2000). The problem of the impacts of climate change on insects is in general very complex. Climate change could alter patterns of disturbance from pest insects through direct effects on their development and survival, adaptation capability, availability of host plants and physiological changes in host defenses, and indirect effects from changes in the abundance of natural enemies, mutualists, and competitors (Porter et al. 1991, Patterson et al. 1999, Ayres and Lombardero 2000). Attempts to quantify the impacts of climate change are usually very simpliÞed, as in our case, when only some of the direct effects were considered. Misleading results can be obtained in such studies (Davis et al. 1998), but in view of a common lack of data and understanding, simpliÞed approaches may be the only choice available (Baker et al. 2000). Despite the caveats in climate change studies, there is already evidence that changes in pest ecology are occurring, which are consistent with predictions of climate change impacts (Harrington et al. 2001). Ayres and Lombardero (2000) suggest some priorities in future research programs that would help us to better understand the impact of climate change on insect herbivores and the damage they cause. Among the priorities, continual surveys are needed of the abundance and impact of insect herbivores, as well as improved understanding of environmental effects on them and their hosts. We see these priorities also as suggestions on how to evaluate the results of our study in the future. A continuous monitoring of onion thrips

764


in the Þeld should be performed at climatologically different locations in Slovenia for at least several years. This would include some extreme years (2000 was such example) that may become more frequent in the future. Additionally, some laboratory experiments should be performed to conÞrm or reject the use of using simple degree-day models over the range of temperatures that the insects are expected to be exposed to in the future. In general, new experiments and measurements will lead to better understanding and modeling of the impact of climate change on the harmfulness of pest insects. Better descriptions of subgrid processes and better horizontal resolution of GCMs will reduce intermodel differences and lead to more reliable estimates of the response of climate systems to changed boundary conditions, even on a regional level. However, the problem of an unpredictable future will always remain, at least in making assumptions about changes in future climate boundary conditions (i.e., changes in concentrations of greenhouse gases and aerosols). Thus, any quantitative assessment of impact of climate change is sentenced at least to that uncertainty. Acknowledgments The authors thank B. Zupancˇ icˇ from the Environmental Agency of the Republic of Slovenia for kindly providing us with the meteorological data and I. Zˇ ezˇlina from the Chamber of Agriculture and Forestry of Slovenia for help with the Þeld experiments. Additionally, the authors thank the reviewers and the topical editor for comments on the manuscript.

References Cited Ayres, M. P., and M. J. Lombardero. 2000. Assessing the consequences of global change for forest disturbance from herbivores and pathogens. Sci. Total. Environ. 262: 263Ð286. Baker, R.H.A., C. E. Sansford, C. H. Jarvis, R.J.C. Cannon, A. MacLeod, and K.F.A. Walters. 2000. The role of climatic mapping in predicting the potential geographical distribution of non-indigenous pests under current and future climates. Agr. Ecosyst. Environ. 82: 57Ð71. Benestad, R. E. 2001a. The cause of warming over Norway in the ECHAM4/OPYC3 GHG integration. Int. J. Climatol. 21: 371Ð387. Benestad, R. E. 2001b. A comparison between two empirical downscaling strategies. Int. J. Climatol. 21: 1645Ð1668. Benestad, R. E. 2002a. Empirically downscaled multi-model ensemble temperature and precipitation scenarios for Norway. J. Climate. 15: 3008 Ð3027. Benestad, R. E. 2002b. Empirically downscaled temperature scenarios for northern Europe based on a multimodel ensemble. Climate Res. 21: 105Ð125. Benestad, R. E., E. J. Førland, and I. Hanssen-Bauer. 2002. Empirically downscaled temperature scenarios for Svalbard. Atmos. Sci. Lett. 3: 71Ð93. Bentz, B. J., and D. E. Mullins. 1999. Ecology of mountain pine beetle cold hardening in the Intermountain West. Environ. Entomol. 28: 577Ð587. Bergant, K., and L. Kajfezˇ-Bogataj. 2005. N-PLS regression as empirical downscaling tool in climate change studies. Theor. Appl. Climatol. (in press).

Vol. 34, no. 4

Bergant, K., L. Kajfezˇ-Bogataj, and S. Trdan. 2005. Uncertainties in regional modelling of climate change impact in future: an example of onion thrips (Thrips tabaci Lindeman) in Slovenia. Ecol. Model. (in press). Bonhomme, R. 2000. Bases and limits to using Ôdegree.dayÕ units. Eur. J. Agronomy. 13: 1Ð10. Briere, J. F., and P. Pracros. 1998. Comparison of temperature dependent growth models with the development of Lobesia botrana (Lepidoptera: Tortricidae). Environ. Entomol. 27: 94 Ð101. Bro, R. 1996. Multi-way calibration. Multi-linear PLS. J. Chemometr. 10: 47Ð 62. Brødsgaard, H. F. 1989. Coloured sticky traps for Frankliniella occidentalis (Pergande) (Thysanoptera, Thripidae) in glasshouses. J. Appl. Entomol. 107: 136 Ð140. Bryant, S. R., J. S. Bale, and C. D. Thomas. 1998. ModiÞcation of the triangle method of degree-day accumulation to allow for behavioural thermoregulation in insects. J. Appl. Ecol. 35: 921Ð927. Cesaraccio, C., D. Spano, P. Duce, and R. L. Snyder. 2001. An improved model for determining degree-day values from daily temperature data. Int. J. Biometeorol. 45: 161Ð 169. Davis, A. J., L. S. Jenkinson, J. H. Lawton, B. Shorrocks, and S. Wood. 1998. Making mistakes when predicting shifts in species range in response to global warming. Nature (Lond.). 391: 783Ð786. de Jong, S. 1998. Regression coefÞcients in multilinear PLS. J. Chemometr. 12: 77Ð 81. Edelson, J. V., and J. J. Magaro. 1988. Development of onion thrips, Thrips tabaci Lindeman, as a function of temperature. Southwest Entomol. 13: 171Ð176. Gordon, H. B., and S. P. O’Farrel. 1997. Transient climate change in the CSIRO coupled model with dynamic sea ice. Mon. Weather Rev. 12: 875Ð907. Gordon, C., C. Cooper, C. A. Senior, H. Banks, J. M. Gregory, T. C. Johns, J.F.B. Mitchell, and R. A. Wood. 2000. The simulation of SST, sea ice extents and ocean heat transports in a version of Hadley Centre coupled model without ßux adjustments. Clim. Dynam. 16: 147Ð168. Grotch, S., and M. MacCracken. 1991. The use of general circulation models to predict regional climate change. J. Climate. 4: 286 Ð303. Hardy, V. G., and D. S. Teakle. 1992. Transmission on sowbane mosaic virus by Thrips tabaci in the presence and absence of virus-carrying pollen. Ann. Appl. Biol. 121: 315Ð320. Harrington, R., R. A. Fleming, and P. Woiwod. 2001. Climate change impacts on insect management and conservation in temperate regions: can they be predicted? Agr. Forest. Entomol. 3: 233Ð240. Honeˇk, A. 1996. Geographical variation in thermal requirements for insect development. Eur. J. Entomol. 93: 303Ð 312. Honeˇk, A. 1999. Constraints on thermal requirements for insect development. Entimol. Sci. 2: 615Ð 621. Houghton, J. T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell, and C. A. Johnson. 2001. Climate change 2001: the scientiÞc basis. Cambridge, Cambridge University Press. Jarosˇı´k, V., L. Kratochvill, A. Honeˇk, and A.F.G. Dixon. 2004. A general rule for the dependence of developmental rate on temperature in ectothermic animals. Proc. R. Soc. Lond. 271(suppl.): S219 ÐS221. Jenser, G., and A´ . Szeńa´ si. 2004. Review of the biology and vector capability of Thrips tabaci Lindeman (Thysanoptera: Thripidae). Acta. Phytopathol. Entomol. Hung. 39: 137Ð155.

August 2005


Jenser, G., R. Ga´ borja´ nyi, A. Szeńasi, A. Alma´ asi, and M. Grasselli. 2003. SigniÞcance of hibernated Thrips tabaci (Thysan., Thripidae) adults in the epidemic of tomato spotted wilt virus. J. Appl. Entomol. 127: 7Ð11. Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell, S. Saha, G. White, J. Wollen, Y. Zhu, M. Chelliah, W. Ebisuzaki, W. Higgins, J. Janowiak, K. C. Mo, C. Ropelewski, J. Wang, A. Leetmaa, R. Reynolds, R. Jenne, and D. Joseph. 1996. The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc. 77: 437Ð 471. Khani, A., S. Moharramipour, S. M. Hosseini, Y. Fathipour, and A. A. Talebi. 2004. Developmental threshold and degree-day requirements of onion thrips, Thrips tabaci Lindeman, in laboratory conditions. J. Agr. Sci. Nat. Res. (in press). Kistler, R., E. Kalnay, W. Collins, S. Saha, G. White, J. Wollen, M. Chelliah, W. Ebisuzaki, M. Kanamitzu, V. Kousky, H. van den Dool, R. Jenne, and M. Fiorino. 2001. The NCEP/NCAR 50-year reanalysis: monthly means CD-ROM and documentation. BAMS. 2: 247Ð267. Koschier, E. H., K. A. Sedy, and J. Novak. 2002. Inßuence of plant volatiles on feeding damage caused by onion thrips Thrips tabaci. Crop. Prot. 21: 419 Ð 425. Kritzman, A., A. Gera, B. Raccah, J.W.M. van Lent, and D. Peters. 2002. The route of tomato spotted wild virus inside the thrips body in relation to transmission efÞciency. Arch. Virol. 147: 2143Ð2156. Logan, J. A., D. J. Wollkind, S. C. Hoyt, and L. K. Tanigoshi. 1976. An analytic model for description of temperature dependent rate phenomena in arthropods. Environ. Entomol. 5: 1133Ð1140. Logan, J. A., J. Re´gnie`re, and J. A. Powell. 2003. Assessing the impacts of global warming on forest pest dynamics. Front. Ecol. Environ. 1: 130 Ð137. McKenzie, C. L., B. Cartwright, M. E. Miller, and J. V. Edelson. 1993. Injury to onions by Thrips tabaci (Thysanoptera: Thripidae) and its role in the development of purple blotch. Environ. Entomol. 22: 1266 Ð1277. McMaster, G. S., and W. W. Wilhelm. 1997. Growing degree-days: one equation, two interpretations. Agr. Forest Meteorol. 87: 291Ð300. Milne, M., and G. H. Walter. 1998. Host species and plant part speciÞcity of the polyphagous onion thrips, Thrips tabaci Lindeman (Thysanoptera: Thripidae), in an Australian cotton-growing area. Aust. J. Entomol. 37: 115Ð119. Mitchell, T. D. 2003. Pattern scaling: an examination of the accuracy of the technique for describing future climates. Clim. Change. 60: 217Ð242. Murai, T. 2000. Effect of temperature on development and reproduction of onion thrips, Thrips tabaci Lindeman (Thysanoptera: Thripidae), on pollen and honey solution. Appl. Entomol. Zool. 35: 499 Ð504. Nakic´ enovic´ , N., O. Davidson, G. Davis, A. Gru¨ bler, T. Kram, E.L.L. Rovere, B. Metz, T. Morita, W. Pepper, H. Pitcher, A. Sankovski, P. Shukla, R. Swart, R. Watson, and Z. Dadi. 2000. Emissions scenariosÑsummary for policymakers. Intergovernmental Panel on Climate Change, Geneva, Switzerland. Norris, R. J., J. Memmott, and D. J. Lovell. 2002. The effect of rainfall on the survivorship and establishment of a biocontrol agent. J. Appl. Ecol. 39: 226 Ð234. Patterson, D. T., J. K. Westbrook, R.J.V. Joyce, P. D. Lingren, and J. Rogasik. 1999. Weeds, insects, and diseases. Clim. Change. 43: 711Ð727. Pope, V. D., M. L. Gallani, P. R. Rowntree, and R. A. Stratton. 2000. The impact of new physical parameterizations

765

in the Hadley Centre climate model (HadAM3). Clim. Dynam. 16: 123Ð146. Porter, J. H., M. L. Parry, and T. R. Carter. 1991. The potential effects of climatic change on agricultural insect pests. Agr. For. Meteorol. 57: 221Ð240. Press, W. H., A. A. Teukolosky, W. T. Vetterling, and B. F. Flannery. 2001. Numerical recipes in Fortran 77: the art of scientiÞc computing, 2nd ed. Cambridge University Press, Cambridge, MA. Roeckner, E., K. Arpe, L. Bengtsson, M. Christoph, M. Claussen, L. Du¨ menil, M. Esch, M. Giorgetta, U. Schlese, and U. Schulzweida. 1996. The atmospheric general circulation model ECHAM4: model description and simulation of present-day climate. Max-Planck-Institut fu¨ r Meteorologie, Hamburg, Germany. Roltsch, W. J., F. G. Zalom, A. J. Strawn, J. F. Strand, and M. J. Pitcairn. 1999. Evaluation of several degre-day estimation methods in California climates. Int. J. Biometeorol. 42: 169 Ð176. Sdoodee, R., and D. S. Teakle. 1987. Transmission of tobacco streak virus by Thrips tabaci: a new method of plant virus transmission. Plant Pathol. 36: 377Ð380. Sharpe, P.J.H., and D. W. DeMichele. 1977. Reaction kinetics of poikilotherm development. J. Theor. Biol. 64: 649 Ð 670. Shelton, A. M., W. T. Wilsey, and M. A. Schmaedick. 1998. Management of onion thrips (Thysanoptera: Thripidae) on cabbage by using plant resistance and insecticides. J. Econ. Entomol. 91: 329 Ð333. Sites, R. W., and W. S. Chambers. 1990. Initiation of vernal activity of Frankliniella occidentalis and Thrips tabaci on the Texas South Plains. Southwestern Entomol. 15: 339 Ð 343. Smilde, A. K. 1997. Comments on multilinear PLS. J. Chemometr. 11: 367Ð377. Stacey, D. A., and M.D.E. Fellowes. 2002. Temperature and the development rates of thrips: evidence for constraint on local adaptation? Eur. J. Entomol. 99: 399 Ð 404. Stendell, M., T. Schmith, E. Roeckner, and U. Cubasch. 2000. The climate of the 21st century: transient simulations with a coupled atmosphere-ocean general circulation model. Danmarks Klimacenter, København. Swart, R., J. Mitchell, T. Morita, and S. Raper. 2002. Stabilization scenarios for climate impact assessment. Global Environ. Change. 12: 155Ð165. Theunissen, J., and G. Schelling. 1998. Infestation of leek by Thrips tabaci as related to spatial and temporal patterns of undersowing. BioControl. 43: 107Ð119. Trdan, S., N. Valicˇ , I. Zˇ ezˇlina, K. Bergant, and D. Zˇ nidarcˇ icˇ . 2005. Light blue sticky boards for mass trapping of onion thrips, Thrips tabaci Lindeman (Thysanoptera: Thripidae), in onion crops: fact or fantasy? Z. Pßanzenkr. Pßanzenschutzs. 112: 173Ð180. Ungerer, M. J., M. P. Ayres, and M. J. Lombardero. 1999. Climate and the northern distribution limits of Dendroctonus frontalis Zimmermann (Coleoptera: Scolytidae). J. Biogeogr. 26: 1133Ð1145. van Rijn van, P.C.J., C. Mollema, and G. M. Steenhuis-Broers. 1995. Comparative life history studies of Franklinella occidentalis and Thrips tabaci (Thysanoptera: Thripidae) on cucumber. Bull. Entomol. Res. 85: 285Ð297. von Storch, H. 1995. Inconsistencies at the interface of climate impact studies and global climate research. Meteorol. Z. 4: 72Ð 80. von Storch, H. 1999. On the use of “inßation” in statistical downscaling. J. Climate. 12: 2505Ð2506. von Storch, H., E. Zorita, and U. Cubasch. 1993. Downscaling of global climate change estimates to regional scales:

766


an application to Iberian rainfall in wintertime. J. Climate. 6: 1161Ð1171. Washington, W. M., J. W. Weatherly, G. A. Meehl, A. J. Semtner, T. W. Beetge, A. P. Craig, W. G. Strand, J. Arblaster, V. B. Wayland, R. James, and Y. Zhang. 2000. Parallel climate model (PCM) control transient simulations. Clim. Dynam. 16: 755Ð774. Wilby, R. L., and T.M.L. Wigley. 1997. Downscaling general circulation model output: a review of methods and limitations. Prog. Phys. Geog. 21: 530 Ð548. Wilks, D. S., and R. L. Wilby. 1999. The weather generation

Vol. 34, no. 4

game: a review of stochastic models. Prog. Phys. Geog. 23: 329 Ð357. Yu, P.-S., T.-C. Yang, and C.-K. Wu. 2002. Impact of climate change on the water resources in southern Taiwan. J. Hydrol. 260: 161Ð175. Zorita, E., and H. von Storch. 1999. The analog method as a simple statistical downscaling technique: comparison with more complicated methods. J. Climate. 12: 2474Ð2489. Received for publication 10 November 2004; accepted 10 May 2005.