Northern Prairie Wildlife Research Center
Variables included in the models not only explain relatively large portions of the variation in number of wet basins, they also are intuitively sensible. Percentage of basins holding water is inversely related to temperature and directly related to precipitation and the number of basins that held water the previous year. That spring and fall temperatures proved more important than temperatures at other times of the year is likely related to formation and disintegration of the frost seal. Precipitation that falls after the ground is frozen in the fall and before spring thaw is more likely to remain in the wetland, rather than percolating into the soil (Shjeflo, 1968; Woo and Winter, 1993). The grassland models suggest that April is the critical spring period in these areas, with a smaller temperature spread favoring basin flooding.
The similarity among the three regions in response to precipitation may be explained by the physical similarities of the basins themselves. Northern prairie wetlands are depressions formed by glacial action during the Wisconsin glaciation. The Prairie Pothole Region as a whole contains similar geomorphology, a mixture of glacial till and outwash (Eisenlohr, 1972; National Wetlands Working Group, 1988). Basins are poorly integrated and lack overland drainage between basins. Water enters the basins primarily as precipitation falling directly on the flooded portion of the basin, or as within-basin runoff (Eisenlohr, 1972). Thus, the shape and depth of the basin, rather than complex drainage systems, determine the amount of water that accumulates.
Recent work (Karl, et al., 1991b) has documented that observed warming trends are occurring in minimum rather than maximum temperatures. The importance of this observation with respect to wetland dynamics cannot be addressed directly with these models. However, mean minimum monthly temperatures were evaluated in model development and found to fit the data less well than the mean maximum temperatures ultimately included. This is not to say that minimum temperatures are unimportant. Extension of the frost-free period that would accompany an increase in minimum temperatures would result in a longer growing season and thus a longer period during which plants would transpire. Daily evapotranspiration might not be as high if minimum rather than maximum temperatures increased, however. The ultimate effect on water balance would depend on the relative magnitudes of possibly lower daily evapotranspiration summed over a longer season (with increased minimum temperatures) and potentially higher daily evapotranspiration over a relatively shorter season (with increased maximum temperatures).
The greater vulnerability of parkland than grassland basins to increased temperature has important implications for waterfowl that rely on these wetlands for breeding habitat. Basin density in parkland is two to three times as high as that in the grassland regions (Table 1), representing a significant reservoir of breeding habitat. Studies of settling patterns suggest that many duck species enter the breeding grounds from the south, stopping at the first available habitat (Johnson and Grier, 1988). Birds not finding suitable habitat in the grasslands move northward into parkland habitat. Other species, such as canvasbacks (Aythya valisineria), use parkland as their primary breeding area, dispersing north or northwest if wetland conditions in the parkland are not suitable for nesting (Johnson and Grier, 1988). This response will not be universally acceptable if wetlands respond to climate change as I have modelled. One important consequence of this geographical difference in wetland response to temperature is the decreased probability of finding better conditions as waterfowl migrate farther north (Figure 4).
Anticipated changes in climate will affect not only the overall number of wet basins, but also are likely to change the vegetative structure within the wetlands. Poiani and Johnson (1993) simulated vegetation development in a semipermanently flooded wetland in North Dakota using climate change scenarios of 0 to +4 °C and -20 to +20% change in precipitation. With a 2 °C rise in temperature and an 11-year simulation period, they found that all precipitation levels except the 20% increase produced drawdown conditions in which seedlings of emergent plant species spread throughout the wetland. With a 4 °C rise in temperature the wetland was dry 38% of the time. The models of the current study and those of Poiani and Johnson suggest that anticipated climate changes may result in both fewer and lower-quality wetlands for waterfowl production. Neither study addresses the potential loss of temporary wetlands, which provide important habitat for migrating waterfowl and shorebirds.
Landscape-level models such as those I present would benefit from a better understanding of soil and groundwater interactions with surface water. For example, most of the soils in the glaciated prairie are built on till comprised of silt and clay, and are thus quite impermeable when moist but crack readily upon drying (Winter, 1989). The number of fractures in the soil may vary both geographically as a function of clay content, and in response to varying lengths of drought. The relationships may or may not be linear; a significant interaction seems likely. Terms that summarize these relationships would help refine model projections.
In all likelihood, the models described here are conservative in that groundwater depletion is not considered. Because wetland numbers are bounded at the lower end by zero, a continued moisture deficit after drying of the basin is not taken into account. If subsequent precipitation replenishes the local water table before adding to surface water, number of wet basins could be less than projected.
Although the temperature and precipitation values used in these simulations are within the range predicted by general circulation models for the northern Great Plains, projections can be only as good as the data used in making them, and regional climate models are still in their infancy (Grotch, 1988; Cushman and Spring, 1989). In addition, these wetland models are empirically-based; projections outside the range of values used to develop the models cannot include factors that only come into play at these extreme values. Some studies have suggested extensive shifts in vegetation resulting from climate-induced changes in fire and disturbance regimes (Overpeck et al., 1990; Hogenbirk and Wein, 1991); disruption of parkland and grassland vegetation types would certainly influence wetland dynamics in ways not anticipated by models dealing solely with climate. Other parameters associated with the accumulation of greenhouse gases, such as carbon dioxide fertilization of plants and increasing ultraviolet B radiation, and their influence on wetland dynamics (Larson, 1994), are also beyond the scope of these models.
The value of models such as these lies less in the accuracy of their projections than in the sensitivities they reveal. These models can suggest areas of concern in the future, independent of the actual magnitude of climatic change. In particular, the models suggest the importance of conserving wetlands in grassland areas of the Prairie Pothole Region that may be less affected by climatic changes. Empirical studies of wetlands throughout the northern Great Plains are needed to discern the reasons for geographic differences in wetland sensitivities to climate.