Fire Regime and the Abundance of Red Pine

Transcription

1 Int. J. Wildland Fire 3(4): O IAWF. Printed in U.S.A. Fire Regime and the Abundance of Red Pine M.D. Flannigan Natural Resources Canada- Canadian Forest Service, Petawawa National Forestry Institute, Chalk River, Ontario, KOJ IJO, Canada Tel : Fax Abstract. Red pine (Pinus resinosa Ait.) is a firedependent species. This study examines the relationship between the fire regime and the abundance of red pine. The fire regime is represented by components of the Canadian Fire Weather Index System and outputs from the Canadian Fire Behavior Prediction System as well as the average area burned and the percentage of conifers of each forest section. Extreme as well as averages values were used in this analysis as a large forest fire is a rare event that can occur on only a few days of the year under extreme fire weather conditions. Results from a forwardstepwise regression explained about 70% of the variance in red pine volume (abundance) data. Variables selected in the regression analysis included extreme headfire intensity, area burned and average drought code. These results suggest that abundance of red pine and other fire affected tree species is directly related to the aspects of the fire regime such as fire intensity. Keywords: Red pine; Pinus resinosa; Fie regime; Fire intensity; Fire behavior. Introduction Red pine (Pinus resinosa Ait.), also sometimes known as northern pine, is found in eastern North America extending from southeastern Manitoba eastward to Newfoundland and as far south as West Virginia (Figure 1) (Burns and Honkala 1990). The species occurs in the Great Lakes - St. Lawrence Forest Region and in southern sections of the Boreal Forest Region (Rowe 1972). Red pine survived the most recent glaciation in refugia in the Appalachians (Wright 1964) and probably reached a peak in abundance and northern extent during a hypsithermal period between 7000 and 3000 years B.P. (Liu 1990). The fire regime1 is an integral part of the ecology of red pine. Recent studies have confirmed that red pine is a fire-dependent species (Bergeron and Brisson 1990, Engstrom and Mann 1991). This dependence on fire is due to several factors. First, the species is intolerant of shade and requires the removal of the canopy for regeneration. Fire is the most effective means of removing the canopy. Second, red pine requires mineral soil exposure for seedling establishment, which means that the organic layer must be removed. Removal of the organic layer is often best achieved by a moderate to intense fire. Finally, fire is an agent that removes competitive vegetation. Fire is the most likely natural agent to fulfill all of these requirements for red pine establishment over the present range of the species. Other natural agents such as windthrow can fulfill the regeneration requirements but in terms of area affected over the species' range these agents are of minor importance when compared to fire. For regeneration red pine requires survivors, which puts the species at a competitive disadvantage compared with species that have serotinouslsemi-serotinous cones or species that can reproduce vegetatively. One important characteristic of red pine is that mature red pine have a thick bark and are able to survive all fires except crown fires and very intense surface fires. In contrast to the protection from fire provided by the bark, the needles are extremely flammable and, in pure stands, red pine ecosystems may represent one of the most flammable species in eastern North America. This flammability promotes frequent fires and fires with lower than extreme intensities, which is an ideal fire regime in terms of red pine regeneration (Van Wagner 1970). Fires that are too intense kill all the red pine and any chance of regeneration. Fires that are not intense ' The Fire regime has four components: fire intensity, fire frequency, timing (season) of the fire and fire size (Malanson 1987).

2 Flannigan, M.D. Figure 1. Native range of red pine. intense enough or too infrequent, on the other hand, will not provide a suitable physical environment for regeneration. The objective of this study was to examine the relationship between the fire regime and the abundance of red pine. Given that the autecology of red pine is closely tied to the fire regime, it is hypothesized that the fire regime is strongly linked to the abundance of red pine. Specifically, abundance of red pine should be related to fire intensity and frequency as described above. Fire intensity can be calculated using the Canadian Forest Fire Weather Index (FWI) System in tandem with the Canadian Fire Behavior Prediction (FBP) System. Data and Methods Three data sets were employed in this study - forest inventory data, climatic data (required to calculate components of the FWI and FBP systems) and area burned by wildfire data. The forest inventory data were obtained from Natural Resources Canada -Canadian Forest Service and the USDA Forest Service (USFS). Data from 234 sites were used in this study (Figure 2). Red pine volume and ground area coverage by county were obtained from the USFS. In Canada, red pine volume was obtained for Rowe's forest sections2 (Rowe 1972). After standardizing the units of measurement each inventory site included centroid latitude and longitude and the red pine volume (m3ha-i). Also Rowe (1972) divided Canadian forests into eight forest regions. These forest regions were subdivided on the basis of distribution patterns of vegetation and physiography into 90 forest sections. obtained for Rowe's forest sections was the percentage of conifers by volume. Climatic data were obtained from the Environment Canada's Atmospheric Environment Service. Thirtyseven climate stations were used in this study (Figure 3). The data included air temperature and relative humidity recorded inside a Stevenson screen 1.5m above ground, wind speed measured at 10 m above ground and the 24 h precipitation at 1200 LST for the period April 1 to September 30 which bounds the fire season in Canada. This information was obtained for the period inclusive, except for stations that did not have records covering the entire period. Area burned data were obtained from Harrington (1982). Data for 28 years ( ) were used to calculate a mean area burned for Manitoba, western Ontario3, eastern Ontario, Quebec and the combined Atlantic provinces. Climatic variables were selected to calculate the components of the Canadian FWI System4 (Van Wagner The province of Ontario was split into two regions. Western Ontario includes everything west of Lake Nipigon. The Fire Weather Index System comprises three fuel moisture codes and three fire behaviour indexes. The three moisture codes represent the moisture content of the fine fuels (Fine Fuel Moisture Code), loosely compacted organic matter (Duff Moisture Code), and the deep layer of compact organic matter (Drought Code). The three fire behaviour indexes, which are derived from the moisture codes and the surface wind, indicate the rate of initial fire spread (Initial Spread Index), total available fuel (Buildup Index), and the intensity of spreading fire (Fire Weather Index).

3 - Fire Regime and the Abundance of Red Pine - Figure 2. Centroid location of forest inventory data. 1987). The Canadian FBP System (Forestry Canada 1992) was also employed to quantify the intensity of forest fires. The FBP System provides quantitative estimates of headfire spread rate, fuel consumption, fire intensity and fire description for 16 different fuel types. The headfire intensitys (kw-m-i) was calculated for three standard fuel types in the FBP System (C6 (red pine plantation), M1 (mixedwood leafless), M2 (mixedwood green)). These fuel types were selected because they represent a range of fuel types where red pine could be found in natural settings. The headfire intensity was calculated using the average annual maximum and the extreme maximum of the fine fuels moisture code (FFMC), build-up index (BUI) and wind speed. The dependent variable in this study was the volume of red pine per hectare, which represented abun- Figure 3. Location of climate stations. Fire Intensity was originally defined by Byram (1959) as: I = HwR where I is the intensity of the fire (kw m-i); His the low heat of combustion for the fuel (lcj kg-1) a standard value of 18,000 kj kg-' is given for H, w is the weight of the fuel consumed per unit area in the active fire front (kg m-i); and R is the rate of forward spread (m sl). dance. The volume of red pine obtained from the forest inventory data was interpolated from the 234 inventory sites to the 37 climate stations using a thin-plate cubic spline (Duchon 1976, Thiebaux and Pedder 1987). A thin-plate cubic spline fits a surface to the existing data, in this case latitude, longitude and red pine volume. The volume of red pine was estimated at the 37 climate stations from this cubic spline surface. The selected

4 244 Flannigan, M.D. independent variables included the average, average annual maximum and extreme maximum for the six components of the FWI System and the average annual maximum and extreme maximum headfire intensity for the three standard fuel types in the FBP system (C6,Ml and M2)6. Other independent variables used included the annual average area burned over regions the size of provinces (Harrington 1982) and the percentage of conifers in each forest section which was interpolated to the climate stations using a thin plate cubic spline. The percentage of conifers was included not because red pine is a conifer but because conifers usually give rise to higher intensity fires and more frequent fires as opposed to broadleaved trees. Extremes were used in this study because only a few days with extreme fire weather conditions are responsible for most of the area burned (Flannigan and Harrington 1988). A study relating components of the FWI System to monthly provincial area burned by wildfire revealed that extreme values of the components explained the most variance, particularly when looking at regions the size of provinces (Harrington et al. 1983). A forward-stepwise regression (SAS Institute Inc. 1990) was used to investigate the partitioning of the variance in the red pine volume data in terms of he fire regime as represented by the components of the FWI System and FBP System. Terms were accepted only if they met the 0.05 significance level, which corresponds to a minimum F-value for entrance of 4.0; terms were removed if they failed to meet the 0.05 significance level. Results Two stepwise linear regressions were preformed using the following equations 1 and 2: Vol' = (percent conifer) lo5- (area burned) (1) In equation 1 the percentage of conifers was selected first and, as a single variable, it explained the most variance (54%). Because the abundance of red pine should be directly linked with the percentage of conifers, the second forward stepwise linear regression was performed with the percentage of conifers excluded (Equation 2). In equation 1 the percentage of conifers and the average area burned explained a total of 68% of the variance. Intuitively, a strong relationship between the abundance of red pine and the percentage of conifers was expected as red pine is a conifer. However, as illustrated in Figure 4, the relationship between the percentage of conifers and the volume of red pine is negatively correlated; i.e., the lower the percentage of conifers the greater the volume of red pine. Figure 5 shows a plot of the area burned versus the volume of red pine. Once again, there is a negative correlation; i.e., the greater the area burned the less likely red pine will be found Percentage of Conifers Figure 4. Volume of red pine (m3 ha-') versus percentage of conifers. Variance = 54% + 14% Explained = 68% Vol = (extreme headfie intensity fuel type M2) -8.31*105*(areaburned) *(DC) (2) Variance = 48% 14% 10% Explained 72% ' Vol is an abbreviation for the volume of red pine (m3 ha-') Headfire intensities calculated from the average values of FFMC, BUI, and wind speed were not used because the resulting intensities were so low that the fire would be of no consequence ZOml Area Burned (ha) Figure 5. Volume of red pine versus area burned.

5 - Fire Regime and the Abundance of Red Pine In equation 2, three terms were selected: the extreme headfire intensity for standard fuel type M2, the area burned and the mean drought code which together explained 72% of the variance. Figure 6 shows a negative correlation between the extreme headfire intensity for fuel type M2 and the volume of red pine suggesting, that the more intense the fire, the less likely red pine will be found. Figure 7 illustrates a positive correlation between he mean drought code and the volume of red pine indicating that red pine occurs in regions more prone to drought. Table 1 presents the correlation coefficients and the variance explained for the volume of red pine volume by each variable. Correlation coefficients range from to 0.51 and the variance explained ranges from 0 to 54% for the individual variables. Table 1. Correlation coefficients and variance explained for red pine volume by individual variables. -- Variable Correlation Variance Explained (%I Fine fuel moisture code Fine fuel mositure code* Fine fuel moisture code** Duff moisture code Duff moisture code* Duff moisture code** Drought code Drought code* Drought code** Initial spread index Initial spread index* Initial spread index** Build up index Build up index* Build up index** Fire weather index Fire weather index* Fire weather index** Headfire intensity C6* Headfire intensity C6** Headfire intensity MI* Headfire intensity MI** Headfire intensity M2* Headfire intensity M2** Area burned Percentage of conifers * Average annual maximum ** Extreme maximum recorded A variable with no '*' indicates an average value. Discussion J Headfire Intensity (1000 kw/rn) Figure 6. Volumeof redpineversus extreme headfire intensity for fuel type M Drought Code Figure 7. Volume of red pine versus average drought code (W. The inverse relationship between the percentage of conifers and the volume of red pine was intriguing. However, this result is weakened by the size of the final data set (only 37 points). There were very few data points south of the southern limit of red pine where a low percentage of conifers with little or no red pine would be expected. To the north, there were several stations beyond the northern limit of red pine that had a high percentage of conifers with no red pine volume, which is typical of the boreal forest with its large stands of conifers, such as jack pine (Pinus banksiana Lamb.) and black spruce (Picea mariana (Mill.) B.S.P.), that flourish in a stand-replacing crown fire regime. This relationship between the percentage of conifers and volume of red pine is consistent with the forest inventory data, which indicates that the volume of red pine is highest just north of its southern limit which borders the decidous forest mannigan 1993). The relationship between the extreme headfire intensity for fuel type M2 and the volume of red pine suggests that there is a strong link between fire intensity and abundance. High-intensity crown fires, which would kill all red pine, are associated with little or no red pine volume. However, locations with moderateintensity fires (10,000-40,000 kw-m-') are capable of supporting relatively high volumes of red pine. This

6 246 Flannigan, M.D. interpretation is supported by the high proportion of variance explained by the headfire intensities for FBP System fuel types M1 and M2 (Table 1). The fact that the headfire intensity for fuel type C6 (red pine plantation) is not well correlated with the volume of red pine is interesting. The reason for the poor correlation could be that the C6 fuel type, which represents fire intensities for a red pine plantation, is not representative of the natural mixed forest. Table 1 shows that fuel types M1 and M2 have similar correlation coefficients and percentages of variance explained, although the selection of fuel type M2 (mixedwood green) over fuel type M1 (mixedwood leafless) might be explained by the fact that in eastern Canada most of the area burned by wildfire is burned in June (Harrington 1982) when the mixedwoods have flushed. The relationship between area burned and the volume of red pine is also as one would expect (Figure 5). Data on the annual average area burned for regions the size of provinces were averaged over a relatively short period (28 years). However, the relationship is clear that as the area burned increases the volume of red pine decreases. Because there were no regions in which the annual average area burned was low, there is no way of proving that, if fire is nearly absent, red pine would be absent as well. Data for smaller regions, preferably covering a longer period, would be required to derive conclusive results. Average and extreme drought codes have a positive correlation with the volume of red pine (Table 1 and Figure 7). This is an interesting result because it would be expected that locations experiencing the most intense fires would have the highest drought code values (headfire intensity, on the other hand, is negatively correlated to the volume of red pine ). However, this is not the case for two reasons. First, the drought code is a measure of long term drought as represented by the fuel moisture of deep compacted organic fuels and large downed wood. Intense crown fires occur under weather conditions that are conducive to fire ignition and fire spreading and require only a short period of dry weather (Flannigan and Harrington 1987). Second, the period used in this study to compute the drought code was April 1 to September 30, which is longer than the actual fire season for the northern climate stations. Therefore, the drought code being used may not be representative of the drought code during the actual fire season. Extreme headfire intensity values are strongly correlated with the volume of red pine (Table 1). A large forest fire is a rare event and is usually responsible for a large proportion of the total area burned in a region for that year. These events can occur on only a few critical days. Therefore, it is expected that the extreme values associated with these critical days should be correlated with the fire regime and the volume of red pine. It has been established that the fire regime varies with the changes in average climate (Clark 1988). However, results from the present study indicate that fire intensity may be more sensitive to changes in climatic variability (frequency of extreme conditions) than to changes in the average climate. The soils and site are important in the ecology of red pine particularly when considering individual trees or stands. These aspects have not been considered in this study and may be responsible for some of the unexplained variance in the data. In this study, the impact of soil and site are probably minimized because each data point represents a large and usually heterogenous area. The present day abundance of red pine is testimony to the diversity of the physical environment given the exacting requirements for the regeneration of red pine. Since the European settlement d North America, the fire regime has been altered due to a fire-exclusion policy. In Ontario, for example, the average area burned during the pre-suppression era was 701,400 ha compared with the present day value of 81,059 (Ward and Tithecott 1993). Thus, fire management policy can have a significant impact on the abundance of firedependent species like red pine. Also, since European settlement the geographical range and abundance of red pine have been altered by anthropogenic influences other than fire management policies such as logging and agricultural development. Many tree species associated with several ecosystems in North America are dependent on fire. The results from this study would suggest that the natural abundance of these species might be directly related to the fire regime. Therefore, models of forest composition should take into account the role of fire on the presence and abundance of tree species. This might be particularly important in trying to simulate the distribution of vegetation in the future under a different climate. Davis (1989) argues that a change in fire regime related to climatic warming might have a greater impact on the forest than the changes due to warming. Summary The ecology of red pine is intimately linked to the fire regime. This study has shown quantitatively that the abundance of red pine, as represented by volume, is related to the fire regime. This does not mean that other processes, such as competition and climate (drought, summer warmth, winter cold, extreme tem-

7 - Fire Regime and the Abundance of Red Pine peratures, etc.), are not playing a role in determining the range and abundance of red pine. Climatic processes may be integrated into some of the variables used in this study (fuel moisture codes). It is clear from this study that the fire regime, as represented by the extreme headfire intensity of mixedwoods, annual average area burned and the average drought code, plays a major role in determining the abundance of red pine. A study using a finer spatial and temporal scale might yield more information on the role of fire frequency, fire size and time of fire (season). Acknowledgements. I would like to thvlk Professor Ian Woodward, Rob McAlpine and Dr. Mike Weber for reviewing an earlier version of this paper. Thanks also go to Mike Wotton and Chris Magnussen for preparing the figures. References Bergeron, Y. and J. Brisson Fire regime in red pine stands at the northern limit of the species range. Ecology 71: Bums, R.M. andb.h. Honkala Silvics ofnorth America trees. 1. Conifers. USDA Forest Service, Washington, D.C., Agriculture Handbook 654. Byram, G.M Forest fire behavior. In Forest fie: Control and use. Edited by Davis, K.P. McGraw-Hill, New York. Clark, J.S Effect of climate change on Fie regimes in northwestern Minnesota. Nature Davis, M.B Lags in vegetation response to greenhouse warming. Climatic Change Duchon, J Interpolation des fonctions de deux variables suivant le principe de la flexion des plaques minces. Revue franais d'automatique, informatique, recherche oprationnelle: Analyse Numerique Engstrom, F.B. and D.H. Mann Fire ecology of red pine (Pinus resinosa) in northern Vermont, U.S.A. Canadian Journal of Forest Research 21: Flannigan, M.D. and J.B. Harrington Synoptic weather conditions during the Porter Lake Experimental Fire Project. Climatolcigical Bulletin 21:1940. Flannigan, M.D. and J.B. Harrington A study of the relationof meteorological variables to monthly provincial area burned by wildfie in Canada Journal of Applied Meteorology 27: Flannigan, M.D Environmental controls of red pine (Pinus resirwsa Ait.) distribution and abundance. Ph.D. Thesis. University of Cambridge, Cambridge, U.K. Forestry Canada Development and Structure of the Canadian Forest Fire Behavior Prediction System. Forestry Canada, Information Report ST-X-3, Ottawa. Harrington, J.B A statistical study of area burned by wildfire in Canada Environment Canada, Canadian Forestry Service, Petawawa National Forestry Institute, Information Report PI-X-16. Harrington, J.B., M.D. Flannigan, andc.e. Van Wagner A study of the relation of components of the Fire Weather Index to monthly provincial area burned by wildfiie in Canada Environment Canada, Canadian Forestry Service, Petawawa National Forestry Institute, Information Report PI-X-25. Liu, Kam-Biu Holocene paleoecology of the Boreal Forest and Great Lakes-St. Lawrence Forest in northern Ontario. Ecological Monographs 60: Malanson, G.P Diversity, stability and resilience: effects of fire regime. In The role of Fie in ecological systems. Edited by L. Trabaud. SPB Academic Publishing. Den Haag. Rowe, J.S Forest regions of Canada. Environment Canada, Ottawa. SAS INSTITUTE INC SAS User's Guide: Statistics, version 6. SAS Institute Inc., Cq, NC. Thiebaux. H.J. and M.A. Pedder Spatial objective analysis: with applications in atmospheric science. Academic Press, London. Van Wagner, C.E Fire andred pine. pp in 10th AnnualTall Timbers Fie Ecology Conference. TallTimbers Res. Station. Van Wagner, C.E Development and structure of the Canadian Forest Fire Weather Index System. Canadian Forestry Service, Technical Report 35. Ward, P.C. and A.G. Tithecott The impact of fire management on the boreal landscape of Ontario. Aviation. Flood and Fire Management Branch PublicationNo Wrigh~ H.E., Jr Aspects of the early postglacial forest succession in the Great Lakes region. Ecology 45: