0400-B3
J. Régnière, J.A. Logan, A. Carroll and L. Safranyik 1
We merged three fields of mathematical modelling (climate, insect phenology and landscape-level interpolation) to predict the impact of climate change on the geographical distribution of native and introduced insects in North America. We used the mountain pine beetle and the gypsy moth as examples. Our models predict range expansions northward and towards higher elevations in both cases. These changes of potential distribution resulting from plausible climate-change scenarios could have severe ecological and economic consequences because they would expose forest resources that have so far not been threatened by these insects. Our methods can be applied to a wide range of organisms from weeds to plant pathogens and insects, but basic knowledge of thermal responses, major climatic influences and seasonality control is required. Detailed information on basic biological characteristics is essential to predict changes in the impact of native and introduced species that are induced by global warming.
Insects and diseases are major sources of disturbance in long-lived forest ecosystems (Dale et al. 2001). These organisms should respond quickly to climate change by shifting their geographical distribution and population behaviour to take advantage of new climatically benign environments: rapid ecological and genetic adaptation by insects has already been documented in Europe although predicting the precise responses of specific insects is difficult because of differences between species in basic life history traits such as seasonality, thermal responses, mobility and host ranges (Logan et al. 2003).
While several modeling approaches to these questions have been used, we have relied on detailed knowledge of the most elementary insect responses to climate: the influence of temperature on their development (Régnière and Logan 2003). Thermal responses determine a species' phenology and seasonality in a given environment. In the absence of additional knowledge on the ecological relationships between an insect and its food, competitors and predators, it is possible to determine a species' potential geographical range and performance simply on the basis of its ability to achieve an adaptive seasonality in a given environment. An adaptive seasonality occurs when the various life stages appear at times when their essential resources are available to them.
The thermal environment in which insect populations develop is very strongly affected by landscape features such as latitude, elevation, exposure to sunlight, continentality, and air and water circulation patterns. Thus, to understand the probable range of a given indigenous or introduced species in response to climate change, modeling at the landscape scale must be brought to bear.
In this paper, we discuss the use of this modeling technology to predict changes in distribution resulting from climate change, using two major insect pests of North American forest ecosystems as examples: the mountain pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae), and the gypsy moth, Lymantria dispar (L.) (Lepidoptera: Lymantriidae).
The mountain pine beetle (MPB) is a native bark beetle that significantly impacts pine forests, particularly ponderosa pine, Pinus ponderosa, and lodgepole pine, Pinus contorta var. latifolia, throughout western North America, from Mexico to British Columbia (Logan and Powell 2001). MPB's distribution is not restricted by available host trees, but by climate (Safranyik 1978). Two critical factors determine the ability of MPB to overcome the defenses of its host tree and successfully reproduce: adequate seasonal timing and simultaneous attack by large numbers of beetles. In the MPB, there is no winter diapause, which implies that seasonality is entirely determined by seasonal weather regimes. Thus, complex spatial and temporal patterns of abundance can result from the details of local annual temperature fluctuations. A phenology model is available for MPB (see Logan and Powell 2001). This model can be used to determine if a given weather regime leads to an adaptive seasonality or not, as defined by voltinism and the stability of oviposition dates from one generation to the next.
Another model predicting the impact of weather on MPB outbreak likelihood is available (Safranyik et al. 1975). This model combines several critical attributes of climate (heat accumulation, temperature extremes during critical periods, seasonal precipitation, etc.) and their effects on the insect and host trees to generate a MPB infestation hazard index. This model accounts for more than the effect of temperature on development. However, it fails to consider the negative impact of partial multivoltinism that can result from excess summer heat.
We compared the output of these two models under plausible climate change scenarios (Price et al. 2001) in terms of the potential distribution and severity of MPB outbreaks in British Columbia and Alberta, Canada.
The gypsy moth (GM) has become a periodic pest of several deciduous forest trees in eastern North America since its introduction in 1869. Hypotheses about its rate of spread and eventual range on the continent, especially to the north, have focused on egg mortality during winter, or on forest susceptibility (Sharov et al. 1999). Limitations that the insect will encounter in establishing to the west and south of its current distribution are less understood.
A model of GM phenology was assembled from the literature (Régnière and Sharov 1998) and was refined to include a full description of GM egg diapause (Gray et al. 2001). The resulting model simulates the entire life cycle of the insect through successive generations. It was first used for monitoring and timing of aerial pesticide applications during an eradication program in British Columbia, Canada, and to determine the areas of southern British Columbia that were most likely to support the establishment of this exotic insect (Régnière and Nealis 2002). This analysis was based on whether or not the model predicted a biologically feasible life cycle for the insect in a given location under normal climatic conditions.
We focused on the likelihood of GM establishment in Utah, USA, under a plausible climate change scenario (Kittel et al. 1995). We also overlap this expected establishment probability with the current distribution of two major host plant groups in the state, aspen (Populus spp.) and oak (Quercus spp.), to predict the change in impact that the insect might have should it become established in Utah.
Several General Circulation Models (GCM) have been used to develop plausible climate change scenarios given the current trends in the accumulation of greenhouse gases in the earth's atmosphere. Two have become widely used in North America: the CGCM1 model, developed by the Canadian Centre for Climate Modeling and Analysis, and HADCM2SUL, developed by the Hadley Centre for Climate Prediction and Research. The coarse spatial resolution of these GCMs has been greatly improved through modeling and interpolation techniques that project their predictions to ecologically meaningful spatial and temporal scales in the United States and Canada. From these climate change scenarios, we created climate change normals databases for Canada and Utah.
Because of the pronounced non-linearity of insect thermal responses, daily or even hourly weather inputs are required to model insect phenology. Daily weather generation provides a general approach that can be applied to past, present and future climates. Although several daily weather generators have been developed (Wilks 1999 and citations therein), we prefer TempGen (Régnière and Bolstad 1994) because it uses monthly normals as inputs (30-year average and extreme minimum and maximum temperatures). This generator has recently been expanded to generate realistic daily rainfall, and validated (unpublished data).
Our modeling approach to landscape-wide phenology simulation is based on the methods of Régnière (1996) as implemented in the BioSIM© software (Régnière et al. 1995). In this approach, two inputs are needed: digital representations of the terrain and suitable weather data. The US Geological Survey provides high-resolution (_100 m) digital elevation models (DEM) for the United States, and low-resolution (_1 km) DEMs for the entire world. From this source we assembled a high-resolution DEM of the state of Utah (USA) and a low-resolution DEM of British Columbia and Alberta (Canada). Sources of weather information are usually scarce relative to the spatial resolution required for mapping the outcomes of biological phenomena. Spatial interpolation methods must be used to obtain air temperature and precipitation information for any number of unsampled points across a landscape from a limited source of geo-referenced weather stations. We use the GIDS algorithm, a local-regression approach (Nalder and Wein 1998).
Models are run for a few hundred randomly located points across a landscape, and results are interpolated between simulation points to estimate model output at other locations on the landscape. Two interpolation methods are especially useful: spatial regression using latitude, longitude, elevation, slope and aspect as predictors, and universal kriging with elevation as an external drift variable. Spatial regression is most appropriate over smaller and steeper landscapes; over larger areas, or where topography is less pronounced, kriging often performs better.
Using the MPB seasonality model, we generated a series of maps depicting the changes in probability of MPB achieving adaptive seasonality as defined by the three criteria listed earlier, as a function of climate normals at 10-year intervals from 1931-1961 to 2041-2070. Simulations were run for 500 randomly located points in British Columbia and Alberta. Universal kriging (with elevation as drift variable) was used for interpolation between simulation points. We used the same method to generate a series of maps from the output of the MPB infestation risk model.
A series of maps of the potential distribution of GM in Utah covering the period 1915 to 2085 (10 year increments) was prepared using the CGCM1 climate change scenario (Kittel et al. 1995). Spatial regression was used as the interpolation method. The resulting probability maps were overlaid with the current distribution of aspen and oak stands in the state to provide an estimate of the amount of forest resources potentially exposed should the GM become established in Utah.
Although overwintering MPB larvae are protected from freezing by antifreeze (glycerol) in their tissues, exposure to -40_C for a few hours is sufficient to kill a large proportion of the overwintering population (Safranyik and Linton 1998). The most plausible climate change scenario for western Canada indicates that the -40_C minimum air temperature isocline is likely to retract northward during the first half of the 21st century, except for the highest mountain ranges in southern British Columbia and Alberta (Fig. 1).
The MPB infestation risk model predicts an overall increase in the area infested by MPB, with a clear northward expansion of high risk climates, as winter minima warm up and precipitation decreases over much of the area (Fig. 2). The MPB seasonality model predicts a gradual northward shift in the insect's most suitable range in British Columbia and Alberta. The model also predicts a gradual restriction of the MPB to higher elevations and northern sections as the insect is increasingly forced into incomplete multivoltinism by the warming climate of the area (Fig. 3). However, direct temperature control of seasonality (no diapause) involves regional populations that are selected for regional climate (Logan and Bentz 1999), and southern populations have been shown to be adapted to warmer southern climates (Bentz et al. 2001). An alternative outcome could be a shift north of more southern populations to occupy warming, northern habitats.
Historically, MPB populations have been most common in southern British Columbia and southwestern Alberta. Non-forested prairies and the high elevations of the Canadian Rocky Mountains have contributed to confining it to that distribution. However, it is likely that the MPB will spill over the natural barrier of high mountains, given the increasing suitability of northern latitudes and higher elevations as global warming proceeds. In recent years, MPB populations have been detected in Alberta along the northeastern slopes of the Rockies (unpublished data). The northern half of Alberta and Saskatchewan is forested by jack pine, Pinus banksiana Lamb., a susceptible species (Cerezke 1995) that may soon come in contact with MPB. Thus, there is a clear risk of MPB becoming a problem in the vast expanses of jack pine forests of Canada east of the Rockies.
Our models of climate change and GM seasonality predict that the overall proportion of susceptible deciduous forests in the state of Utah overlapping with the potential distribution of GM did not vary significantly between 1915 and 1965, averaging 25%. However, between 1975 and 2065, this proportion is expected to increase to over 95% of the susceptible forest. The GM has not permanently become established in Utah, and pest management organizations in the state keep a sharp eye on the situation to prevent its establishment. However, the ecological and economic impacts of a potential range expansion of GM in Utah as a result of climate change could be very important should the insect become established in areas where abundant susceptible forest stands exist, such as in the Wasatch Mountains (Fig. 4).
Current modeling technology makes possible the analysis of climate change impacts on the distribution of native and introduced species as long as sufficiently detailed information on their thermal responses exists. It is possible to estimate the range changes that can be expected purely from the point of view of developmental physiology leading to adaptive seasonality.
Additional information on links between species performance and climate can also be taken into consideration through more complete or detailed models predicting not only seasonality, but also the degree of risk through climate-related impacts to critical aspects of pest organisms' life histories.
When potential geographic ranges are overlaid with susceptible forest types, it is possible to predict the ecological and economic impacts of climate change on insects and other sources of disturbance in forest ecosystems.
Bentz, B.J., J.A. Logan and J.C. Vandygriff, 2001. Latitudinal variation in Dendroctonus ponderosae (Coleoptera: Scolytidae) development time and adult size. The Canadian Entomologist 133: 375-387.
Cerezke, H.F., 1995. Egg gallery, brood production, and adult characteristics of mountain pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae), in three pine hosts. The Canadian Entomologist 127: 955-965.
Dale, V.H., L.A. Joyce, S. McNulty, R.P. Neilson, M.P. Ayres, M.D. Flannigan, P.J. Hanson, L.C. Irland, A.E. Lugo, C.J. Peterson, D. Simberloff, F.J. Swanson, B.J. Stocks and B.M. Wotton, 2001. Climate change and forest disturbances. BioScience 51: 723-734.
Gray, D.R., F.W. Ravlin and J.A. Braine, 2001. Diapause in the gypsy moth: a model of inhibition and development. Journal of Insect Physiology 47: 173-184.
Kittel, T.G.F., N.A. Rosenbloom, T.H. Painter and D.S. Schimel, 1995. The VEMAP integrated database for modelling United States ecosystem/vegetation sensitivity to climate change. Journal of Biogeography 22: 857-862.
Logan, J.A., and B.J. Bentz, 1999. Model analysis of mountain pine beetle (Coleoptera: Scolytidae) seasonality. Environmental Entomology 28: 924-934.
Logan J.A., and J.A. Powell, 2001. Ghost forests, global warming, and the mountain pine beetle (Coleoptera: Scolytidae). American Entomologist 47: 160-173.
Logan, J.A., J. Régnière and J.A. Powell. Assessing the impacts of global climate change on forest pests. Frontiers of Ecology (in press).
Nalder, I.A., and R.W. Wein, 1998. Spatial interpolation of climatic normals: test of a new method in the Canadian boreal forest. Agricultural and Forest Meteorology 92: 211-225.
Price, D., D.W. McKenney, D. Caya, M.D. Flannigan and H. Côté, 2001. Transient climate change scenarios for high resolution assessment of impacts on Canada's forest ecosystems. Report to Climate Change Action Fund. http://www.cics.uvic.ca/scenarios/index.cgi?Other_Data#transienthighres
Régnière, J. 1996. A generalized approach to landscape-wide seasonal forecasting with temperature-driven simulation models. Environmental Entomology 25: 869-881
Régnière, J., and P. Bolstad, 1994. Statistical simulation of daily air temperature patterns in eastern North America to forecast seasonal events in insect pest management. Environmental Entomology 23: 1368-1380.
Régnière, J., and J.A. Logan. Animal life cycle models. in M. Schwartz [Ed], Phenology, an integrative environmental science. Kluwer, New York (in press).
Régnière, J., and V.G. Nealis, 2002. Modelling seasonality of gypsy moth, Lymantria dispar (Lepidoptera: Lymantriidae), to evaluate the probability of its persistence in novel environments. The Canadian Entomologist 134: 805-824.
Régnière, J., and A. Sharov, 1998. Phenology of Lymantria dispar (Lepidoptera: Lymantriidae), male flight and the effect of moth dispersal in heterogeneous landscapes. International Journal of Biometeorology 41: 161-168.
Régnière, J., B. Cooke and V. Bergeron, 1995. BioSIM: A computer-based decision support tool for seasonal planning of pest management activities. User's Manual. Canadian Forest Service, Information Report LAU-X-116.
Safranyik, L., 1978. Effects of climate and weather on mountain pine beetle populations. pp. 77-84 in Kibbee, D.L., A.A. Berryman, G.D. Amman and R.W. Stark [Eds], Theory and practice of mountain pine beetle management in lodgepole pine forests: Symposium Proceedings, Washington State University, Pullman, 25-27 April 1978. Forest, Wildlife and Range Experiment Station, University of Idaho,.
Safranyik, L., and D.A. Linton, 1998. Mortality of mountain pine beetle larvae, Dendroctonus ponderosae (Coleoptera: Scolytidae) in logs of lodgepole pine (Pinus contorta var. latifolia) at constant low temperatures. Journal of the Entomological Society of British Columbia 95: 81-87.
Safranyik, L.D., D.M. Shrimpton and H.S. Whitney, 1975. An interpretation of the interaction between lodgepole pine, the mountain pine beetle and its associated blue stain fungi in western Canada. pp. 406-428 in Baumgartner, D.M. (Ed), Management of lodgepole pine ecosystems: Symposium Proceedings, Washington State University, Pullman, 9-11 October 1973. Washington State University Cooperative Extension Service.
Sharov, A.A., B.C. Pijanowski, A.M. Liebhold and S.H. Gage, 1999. What affects the rate of gypsy moth (Lepidoptera: Lymantriidae) spread: winter temperature or forest susceptibility? Agricultural and Forest Entomology 1: 37-45.
Wilks, D.S., 1999. Simultaneous stochastic simulation of daily precipitation, temperature and solar radiation at multiple sites in complex terrain. Agricultural and Forest Meteorology 96: 85-101.
Figure 1. Distribution of average winter extreme minimum temperature in British Columbia and Alberta under a plausible climate change scenario. Dark blue area: minimum temperature < -40_C. Pale blue: minimum temperature > -40_C. Points: normal-generating weather stations.
Figure 2. Likelihood of Mountain Pine Beetle outbreaks developing in British Columbia and Alberta under a plausible climate change scenario using the MPB infestation risk model (Safranyik 1975).
Figure 3. Likelihood of Mountain Pine Beetle achieving adaptive seasonality (univoltine and properly timed) in British Columbia and Alberta under a plausible climate change scenario using the MPB seasonality model (Logan and Powell 2001).
Figure 4. Expected change in the overlap between the distributions of susceptible deciduous forest (oak and aspen) and gypsy moth between 1965 and 2085 as a result of climate change in the Wasatch Mountains, east of Salt Lake City, Utah. Area in blue: city limits. Area in black: probability of GM establishment > 0.5. Area in white: probability of GM establishment < 0.5. Area in red: susceptible forest overlapping with high GM establishment probability. Area in green: susceptible forest not overlapping with high GM establishment probability.
1 Natural Resources Canada, Canadian Forest Service, 1055 du P.E.P.S., P.O. Box 3800, Sainte-Foy, QC, Canada G1V 4C7. [email protected]