Postfire seed rain of black spruce, a ... - Canadian Science Publishing

1575

Postfire seed rain of black spruce, a semiserotinous conifer, in forests of interior Alaska Jill Johnstone, Leslie Boby, Emily Tissier, Michelle Mack, Dave Verbyla, and Xanthe Walker

Abstract: The availability of viable seed can act as an important constraint on plant regeneration following disturbance. This study presents data on seed quantity and quality for black spruce (Picea mariana (Mill.) B.S.P.), a semiserotinous conifer that dominates large areas of North American boreal forest. We sampled seed rain and viability for 2 years after fire (2005–2007) in 39 sites across interior Alaska that burned in 2004. All sites were dominated by black spruce before they burned. Structural equation modeling was used to assess the relative importance of prefire spruce abundance, topography effects, canopy fire severity, and distance to unburned stands in explaining variations in black spruce seed rain. Prefire basal area of spruce that remained standing after fire was a significant predictor of total seed rain, but seed viability was more strongly related to site elevation, canopy fire severity, and distances to unburned stands. Although positive relations between tree basal area and the size of the aerial seed bank may place a first constraint on seed availability, accurate prediction of postfire viable seed rain for serotinous conifers also requires consideration of the effects of abiotic stress and canopy fire severity on seed viability. Re´sume´ : La disponibilite´ de graines viables peut constituer une contrainte importante pour la re´geńe´ration des ve´ge´taux apre`s une perturbation. Cette e´tude pre´sente des donneés sur la quantite´ et la qualite´ des graines d’e´pinette noire (Picea mariana (Mill.) B.S.P.), un conife`re a` coˆnes semi-se´rotineux qui domine de vastes zoes de la foreˆt boreále de l’Ame´rique du Nord. Nous avons ećhantillonne´ la pluie de graines et estime´ la viabilite´ des graines pendant 2 ans (2005–2007) dans 39 stations bruˆleés en 2004 a` l’inte´rieur de l’Alaska. Toutes les stations e´taient domineés par l’e´pinette noire avant d’eˆtre bruˆleés. Une mode´lisation par e´quation structurelle a e´te´ utiliseé pour estimer l’importance relative de l’abondance de l’e´pinette avant le feu, des effets de la topographie, de la se´ve´rite´ du feu dans la canopeé et de la proximite´ de stations non bruˆleés sur la variation de la pluie de graines d’e´pinette noire. La surface terrie`re avant le feu des e´pinettes noires qui sont demeureés sur pied apre`s le feu e´tait une variable pre´dictive significative de la pluie de graines totale, mais la viabilite´ des graines e´tait plus fortement relieé a` l’altitude de la station, a` la se´ve´rite´ du feu dans la canopeé et a` la proximite´ des peuplements non bruˆle´s. Quoique des relations positives entre la surface terrie`re des arbres et la taille de la banque ae´rienne de graines puissent constituer une premie`re contrainte sur la disponibilite´ des graines, on doit aussi tenir compte des effets des stress abiotiques et de la se´ve´rite´ du feu dans la canopeé sur la viabilite´ des graines pour pre´dire avec exactitude la pluie de graines viables des conife`res se´rotineux apre`s un feu. [Traduit par la Re´daction]

Introduction Cone serotiny is a reproductive adaptation found worldwide among woody plants that live in habitats with frequent fire (Lamont et al. 1991; Tapias et al. 2004). Serotinous cones provide prolonged storage of seeds within an aerial seed bank, allowing the release of seeds after heating and supplying seed for postfire recruitment (Lamont et al. 1991). In the boreal forest, fire is common, and cone serotiny is found among several dominant conifers from the Pi-

naceae family (Zasada et al. 1992; Agee 1998). Black spruce (Picea mariana (Mill.) B.S.P.) is a semiserotinous conifer that is widespread in the boreal forests of North America, where it forms the canopy dominant in areas with particularly cool and nutrient-poor soils (Rowe 1972; Van Cleve and Viereck 1981). Stands of black spruce in western North America typically burn at fire return intervals of 80– 150 years (Viereck 1983; Larsen 1997; Cumming 2001). The growth form of black spruce, with its low-level branches that connect ground and canopy fuels, leads to

Received 4 July 2008. Accepted 8 May 2009. Published on the NRC Research Press Web site at cjfr.nrc.ca on 18 August 2009. J. Johnstone.1 Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK S7N 5E2, Canada; Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, AK 99775, USA. L. Boby and M. Mack. Department of Botany, University of Florida, Gainesville, FL 32611, USA. E. Tissier. Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, AK 99775, USA. D. Verbyla. School of Natural Resource and Agriculture Sciences, University of Alaska Fairbanks, Fairbanks, AK 99775, USA. X. Walker. Department of Biology, University of Saskatchewan, 112 Science Place, Saskatoon, SK S7N 5E2, Canada. 1Corresponding

author (e-mail: [email protected]).

Can. J. For. Res. 39: 1575–1588 (2009)

doi:10.1139/X09-068

Published by NRC Research Press

1576

fires that readily crown and cause high levels of stand mortality (Viereck 1983; Johnson 1992). As is common with woody species having high flammability (Schwilk and Ackerly 2001), black spruce relies heavily on the postfire recruitment of seedlings to maintain stand dominance across multiple fire cycles (Viereck 1983). For obligate seeders like black spruce, the availability of postfire seed may strongly constrain patterns of stand recovery after fire (Greene et al. 1999; Ne’eman et al. 1999; Schoennagel et al. 2003). Restrictions on seed availability can interrupt the recovery of conifer forests and lead to replacement by forests with unusually low densities of conifer trees (Jasinski and Payette 2005; Johnstone and Chapin 2006b; Verkaik and Espelta 2006). Not only the amount of seed, but the timing of seed availability is important. Boreal forest seedbeds are most receptive to seeds in the first few years after fire, after which seedbed quality declines and remains low for several decades (Simard et al. 1998; Charron and Greene 2002; Peters et al. 2005). Viable seeds that arrive at a site in the initial years after fire are likely to have a much greater potential for successful recruitment than seeds that arrive later (Peters et al. 2005). Interactions between seed availability and temporal changes in seedbed quality are believed to be the primary mechanisms that control the density and age structure of boreal forest regenerating after stand-replacing fire (Greene et al. 1999; Peters et al. 2005). Understanding the factors that control early postfire seed availability is an important element of predicting future patterns of secondary succession and stand structure in postfire forests (e.g., Ne’eman et al. 1999; Borchert et al. 2003; Dovciak et al. 2003). However, seed dispersal is often highly variable over space and time, and empirical data on dispersal are typically difficult and costly to collect (e.g., Clark et al. 2003). In boreal forests, there have been useful advances in modeling seed availability as a function of plant growth and other factors (Greene and Johnson 1999; McIntire et al. 2005), but empirical data to test these models remain sparse. The size of the aerial seed bank of serotinous conifers has been shown to be positively related to the sizes or total basal area of the parent trees in a stand (Greene and Johnson 1999; Ne’eman et al. 1999; Turner et al. 2007). This relationship can provide a first estimate of the reproductive potential of stands of serotinous conifers (Greene and Johnson 1999). However, several other factors may exert additional control on postfire seed availability and modify the predicted relationship between prefire stand structure and postfire seedling recruitment (Ne’eman et al. 1999). In forests near tree line, severe climate conditions are likely to reduce the potential production of viable seeds (Black and Bliss 1980; Sirois 2000). Fire characteristics may also influence postfire seed rain and viability. Seed viability is likely to decline when canopy fires are sufficiently intense to kill or consume seeds in the aerial seed bank (Zasada et al. 1979; Arseneault 2001; de Groot et al. 2004). Fire may also have indirect effects on seed rain by determining the distance of a burned stand to unburned trees and, thus, the potential for unburned trees to provide a source of seeds to the site (Greene and Johnson 2000). This study measured postfire seed rain across a range of

Can. J. For. Res. Vol. 39, 2009

burned spruce stands in interior Alaska to investigate key factors controlling variations in postfire seed quantity and quality in northern boreal forests. We hypothesized that prefire species basal area and climate conditions would act as the primary controls over total seed production and that fire characteristics would have effects on seed viability and seed input from unburned stands. First, measurements of prefire stand structure and postfire seed rain were used to assess whether prefire species basal area alone was a strong predictor of observed seed rain. Second, we fit a multivariate model that allowed us to test the relative importance of prefire factors, such as species basal area, topographic effects on local climate, and fire effects in determining postfire viable seed rain.

Methods Study sites The study area consisted of three large burn complexes located in interior Alaska (Fig. 1). The burns originated from multiple fire starts that occurred during the hot and dry summer of 2004, when Alaska experienced its largest fire year on record (Todd and Jewkes 2006). The burned areas were located within a large region of continuous boreal forest with extensive coverage of black spruce forests (Calef et al. 2005). Other important forest types in the region were dominated by white spruce (Picea glauca (Moench) Voss), trembling aspen (Populus tremuloides Michx.), and Alaskan paper birch (Betula neoalaskana Sarg.) (Viereck et al. 1986; Calef et al. 2005). The climate in this region is continental, with cold winters and relatively warm summers, and the area is underlain by discontinuous permafrost (Shulski and Wendler 2007). In the town of Fairbanks, near the center of the region, annual average temperature is –3 8C, with an annual average precipitation of 254 mm. Minimum daily temperatures in January average –28 8C, and maximum temperatures in July average +23 8C (Shulski and Wendler 2007). The fires of 2004 burned into late August, and study sites were selected in the following spring, using three selection steps. First, we conducted an initial survey of all road-accessible portions of the three burned areas in May 2005, immediately after snowmelt. The roads we used to access the burns were the Taylor and Dalton Highways, and the Steese Highway and associated side roads (Fig. 1). These roads were paved or gravel, one- or two-lane highways that crossed river valleys and mountainous terrain to reach small towns in rural Alaska. Because the road network intersected large portions of the 2004 fires within wilderness areas of interior Alaska, they provided an exceptional opportunity to feasibly access large areas of burned forest (Fig. 1). Within each burned area, we selectively identified *30 sites (90 sites total) that encompassed the full range of fire severities, site moisture, topographic positions, and geographic dispersion that we observed in black spruce forests in each burn. We selected our sampling sites strategically to obtain maximum representation of the full range of variation in each burn, rather than by random sampling, which would have concentrated samples in the most common set of preand post-fire conditions. Sites were located 100–600 m Published by NRC Research Press

Johnstone et al.

1577

Fig. 1. Map of the study sites, with the location of the study area in interior Alaska shown as a square in the inset. Areas burned in 2004 are shown as grey-filled polygons. Sample sites were located in burn complexes along the Dalton, Steese, and Taylor Highways, and are shown as black circles (note that dots overlap for adjacent sites).

from the road, beyond the range of any visible road effects, including dust (Auerbach et al. 1997), and away from any signs of other human disturbance. For each site, we collected initial measurements of fire, stand, and site characteristics. All sites were dominated by black spruce when they burned in 2004. At some sites, white spruce, paper birch, or trembling aspen were also present in the prefire stand, but at lower densities than black spruce. The sampled community types included moist, black spruce forests with a prefire understory of Eriophorum vaginatum L. tussocks or Sphagnum moss, mesic black spruce forests with a feathermoss understory, and dry or elevational black spruce woodlands with an open canopy and lichen or moss understory (Hollingsworth et al. 2006). From the initial set of 90 potential sites, we used stratified selection to identify 39 sites that represented different combinations of prefire conditions (Fig. 2). We randomly selected eight sites to represent each category in a factorial combination of high and low site moisture, and high and low fire severity (n = 32). We also included seven highelevation sites to represent burned tree-line forests from each area. These 39 sites were used for intensive study of postfire seed rain. The study sites were divided approximately equally among the three burn areas, with 14 sites along the Dalton highway (65.98–66.38N, 149.88– 150.48W), 12 sites along the Steese highway and side roads (65.18–65.38N, 146.78–147.58W), and 13 sites along the Taylor highway (63.48–64.18N, 142.08–142.58W;

Fig. 2. A scatterplot showing the observed range of site elevations and prefire basal area of black spruce observed for the set of 39 intensive study sites (&), in relation to the original set of 90 surveyed sites (all * and &).


1578


Fig. 3. Summary of observed seed rain, shown as (A) total seedsm–2, (B) seed percent viability, and (C) viable seedsm–2. Values are shown per collection period, except for the grey boxes, which represent mean annual values, calculated using all available data points. Note that A and C are plotted on a log10 scale. Boxes encompass the 25%–75% quartiles of the data, with the median indicated by the line through the centre of the box. Whiskers extending from the box encompass the 95% quartiles, and extreme observations are shown as filled circles. The number shown on the inside of each box represents the sample size (sites).

Fig. 1). Each site was sampled as a 30 m 30 m area (*0.01 ha). Plot boundaries were permanently marked with stakes to delimit a 30 m 30 m square plot with sides aligned to E–W and N–S compass orientations. To minimize trampling and provide space for different sets of field measurements, all sampling took place along parallel, 30 m transects laid out at fixed positions within the plot boundaries. Seed rain measurements We collected seed rain in 16 *0.1 m2 (52 cm 22.5 cm) greenhouse trays that were attached to the ground with metal spikes. Eight trays were spaced evenly along two, 30 m transects that were 15 m apart. The trays were lined with a plastic, synthetic ‘‘grass turf’’ to provide a rough surface to catch the seeds and prevent them from being blown out of the shallow (5 cm deep) tray, as has been observed in other postfire seed studies (Zasada et al. 1979; J. Johnstone, personal observation). Seed traps were established at the 39 sites in mid-June 2005. The contents of the traps were collected in late August 2005, late May 2006, early September 2006, and early June 2007. Collections in August–September and May–June represented summer and winter seed rain, respectively. In some cases, persistent rainy weather prevented accurate sample collection because trap contents could not be reliably extracted when the traps were wet. Observer error also caused the loss of some trap samples. Consequently, our analysis did not include seed trap samples from the Taylor Highway sites for summer and winter samples in 2006, and from Dalton Highway sites for the summer 2006 sample (Fig. 3). In total, approximately 1800 trap samples were collected during the 2 years of the study. During field collections, the seed traps were removed from the ground and emptied into large plastic bags. Usually, four adjacent traps were collected into the same bag, resulting in four composite samples per site. Occasionally, seed traps were located directly below the cone clusters of fallen black spruce and were collected separately. Some traps located under cone clusters had seed counts more than 10 times higher than those in other traps from the same site. These were considered outliers and were excluded from the site-level summaries and analyses (see Results). In the lab, the contents of the seed trays were sifted to remove leaf litter and other coarse debris, and then sorted by hand to collect all of the spruce seeds in the sample. The seeds of black and white spruce overlap in size and are similar in morphology, and we tallied all spruce seeds in our samples as those of black spruce. Black spruce represented the majority of conifers in the prefire stands (>95% in forest stands, and >70% in tree line stands), so most of the seed Published by NRC Research Press

Johnstone et al.

we collected is likely to have been black spruce. White spruce is a nonserotinous species, and any white spruce seed included in our sample probably originated from surrounding unburned stands (Greene and Johnson 2000; Purdy et al. 2002). Consequently, contributions of white spruce seed would likely have impacted our analyses by increasing our estimate of the importance of dispersal from unburned stands. Once sorted, seeds were stored frozen until germination trials could be completed to assess seed viability (Leadem et al. 1997). Seed viability was measured as the proportion of seeds that germinated when seeds were placed on moist filter paper in a sterile Petri dish (11 cm diameter) for 21 days (Leadem et al. 1997). All samples from a collection period were tested at the same time. Each sample was tested in a separate dish, and samples with >100 seeds were separated into multiple dishes. The germination trials took place at room temperature in a lab under 10–12 hday–1 of natural and fluorescent light. Seeds were regarded as ‘‘germinated’’ when the hypocotyl was twice the length of the seed coat (Leadem et al. 1997). Counts of total and viable seeds were translated into areabased estimates of seed rain (seedsm–2) by summing the total seed count for a site and collection period, and then dividing by the total seed trap area. Viability estimates were averaged across composite trap samples to obtain a site-level estimate for each collection period. We then calculated average annual seed rain (seedsm–2year–1) as the sum of averaged summer and winter seed counts. All sites had samples from at least one summer and one winter collection. Average percent viability of seeds was calculated as the average across collection periods. These average data were used as the dependent variables in the analyses described below (n = 39 sites). Other site measurements Measurements of vegetation and environmental characteristics of the sites were made in summer 2005. To characterize the prefire stand, we measured the density (stemsha–1) and basal area (m2ha–1) of all trees that were alive prior to the fire. Trees that were already dead at the time of the 2004 burn were distinguished based on deep charring of the stem bole. The species and diameter at breast height (DBH, height = 1.4 m) was measured for each tree >1.4 m height that was rooted in a 2 m 30 m belt transect. We also noted whether the tree was currently live or dead, and standing or fallen. DBH was measured on fallen trees at 1.4 m above the apparent root collar. Tree measurements indicated that most sites (38/39) had at least 95% canopy mortality from the fire, while one site near tree line had 50% canopy mortality. We used several metrics to estimate canopy fire severity, defined here as a measure of canopy biomass consumption that occurred as the direct result of a fire (Lentile et al. 2006). These metrics were as follows: (i) a stand-level ranking of canopy consumption based on progressive levels of needle, twig, and branch consumption; (ii) a summation of tree-based estimates of percent biomass consumption derived from allometric equations; and (iii) the overstory portion of the composite burn index (CBI), a semiqualitative index used to score consumption within forest canopies (Key and Benson 2005). All of these severity rankings implicitly contain information on canopy mortality, as trees that survived the fire with green needles have the lowest

1579

consumption scores. However, each metric estimates canopy severity in slightly different ways, and requires different amounts of user effort to measure in the field. Stand-level ranking of canopy consumption was made on a qualitative, ordinal scale by assigning each stand to a canopy severity rank based on the dominant pattern of canopy fuel consumption: 1, low consumption, with many brown needles and most small twigs remaining; 2, low to moderate, with few needles but most small twigs remaining; 3, moderate, with few small twigs remaining but many branches; 4, moderate to high, with most small twigs and many branches consumed; and 5, high, with most of the aboveground canopy except the central trunk and branch stubs consumed. To estimate the proportion of canopy biomass consumed, we recorded a visual estimate of the percentage of needles, fine twigs, coarse branches, and cones that were consumed by the fire for each tree encountered in the 2 m 30 m belt transect described above. Measurements of visual percent consumption were translated into percent biomass consumption using previously developed allometric equations relating canopy biomass to trunk diameter (Mack et al. 2008). Finally, we measured the composite burn index (CBI) for the stand overstory as the sum of stand-level scores of visual percent cover of the tree canopy within categories of green (live), brown (scorched), and black (consumed) components, and percent canopy mortality (Key and Benson 2005). As the amount of seed dispersed into burns has been shown to be proportional to the distance from an unburned edge (Greene and Johnson 2000), we used distance from unburned stands to provide an indicator of the potential contribution of seeds from unburned trees. We estimated the distance (m) from the edge of a site to the nearest live stand (>20 live trees) of black spruce. Ground estimates were based on paced distances for distances up to 200 m, and visual estimates were used for greater distances. We also estimated the distance of a site to the nearest unburned pixel using classified postfire Landsat satellite images, where unburned pixels were identified by low values of the normalized burn ratio (NBR; Epting et al. 2005). Ground-based measurements have the advantage of allowing us to be sure that measured distances were to the nearest black spruce, rather than other vegetation types, which were not distinguished in the NBR classification. However, remote sensing allowed detection of unburned patches not visible from the ground, and likely generated more accurate estimates of longer distances. To capitalize on both measures, we developed a synthetic measure of distance to live stands that used ground-based measures for distances up to 200 m, and remote sensing measures for greater distances. In the complex topography of interior Alaska, slope, aspect, and elevation can all have strong effects on local climate conditions (Van Cleve et al. 1991). We measured slope and aspect at each site with a hand-held clinometer and compass. Elevation was measured as metres above sea level with a GPS. The relative amount of solar energy received at a site was calculated from slope and aspect as an insolation index (I) for the summer solstice, as, I ¼ cosðSÞ cosðL 23Þ sinðSÞ cosðAÞ sinðL 23Þ where S is slope angle in degrees, A is slope aspect in dePublished by NRC Research Press

1580

grees, and L is degrees North latitude, adjusted to the solstice sun position (e.g., Bennie et al. 2006). We also estimated the expected log heat load of a site based on an empirically derived equation for 308–608N latitude that assumes greater temperatures on SW than NE facing slopes (McCune and Keon 2002). Data analysis We used three metrics to describe black spruce seed rain at the sites: (i) total seed rain, measured as the total number of seedsm–2year–1; (ii) seed viability, the percentage of seeds that germinated in laboratory trials; and (iii) viable seed rain, the number of viable seedsm–2year–1 (Fig. 3). Although viable seed rain is arguably the most biologically important metric of seed rain, it is determined by the two potentially independent factors of total seed number and seed viability. Consequently, we focused the first steps in our analysis on total seed rain and seed viability, and then tested our understanding of direct and indirect effects on viable seed rain using a structured statistical model. Prior to analysis, we examined our data graphically to look for outliers and assess departures from normality and homoscedasticity. We used log transfomations to correct for problems of skew and increasing variance that were common with many of our continuous variables. For variables containing zero values, we used a (log10+ c) transformation, where c is a constant equal to the smallest non-zero value observed in the data. One site from the Taylor highway was identified as a strong outlier with very low seed counts. This site had reburned after a short fire interval of 0.05). Consequenly, all sites were included together in the model analyses. Relationships in the models were assumed to be linear (Arbuckle 2007). However, the bivariate relationship between log seed viability and elevation showed a significantly better fit when modeled as a quadratic rather than linear relationship (ANOVA model comparison, F = 8.55, p = 0.006). To accommodate this nonlinear relationship, we included a quadratic term (elevation squared) in the multivariate model. We assessed the overall fit of a model by comparing the observed covariance structure to that expected from the proposed model, using a c2 test of goodness-of-fit (Grace Published by NRC Research Press

Johnstone et al.

2006). Failure to reject the null hypothesis of no difference indicated adequate fit of the model to the observed data. We assessed the significance of individual pathways using z tests, estimated as the ratio of the parameter estimate divided by its standard error, to test for path coefficients significantly different from zero at a = 0.05. Undirected correlations that were not significant were removed from the model. We compared the relative support for alternative models based on R2 values for the dependent variables, and the Akaike information criterion (AIC), an information-theoretic statistic that incorporates a penalty for increases in model complexity (Anderson et al. 2000).

Results Seed rain estimates Total seed rain of black spruce observed at a site in any single collection period ranged from 0 to 193 seedsm–2season–1. Estimates of seed rain did not vary in a consistent fashion among collection periods (Fig. 3), and collection periods were pooled to estimate median annual, total seed rain at 37 seedsm–2year–1 (range 4.8–276). Seed viability averaged near 10% and rarely exceeded 30%; consequently, estimates of viable seed rain were considerably lower than the total seed rain (Fig. 3). Median viable seed rain was estimated at 3.4 viable seedsm–2year–1, with maximum values near 50 seedsm–2year–1. Total and viable seed rain counts for all collection periods were positively skewed, with frequent low values and a small number of observations with very high values (Fig. 3). Our study included 17 trap samples from six sites ( 0.1). Site elevation was unrelated to total seed rain (r = –0.11, p = 0.53), but was negatively correlated with seed viability (Figs. 6A and 6B). The negative relationship between elevation and seed viability was largely due to a steep decrease in log seed viability at high elevations, a relationship that was best fit with a quadratic polynomial (R2adj = 0.40, p < 0.001). Estimates of site insolation and heat load were strongly correlated with each other (r = 0.67, p < 0.01), but showed no relationship with seed rain or viability (r < 0.2, p > 0.1). Distance to the nearest unburned stand was unrelated to total seed rain (Fig. 6C), but was negatively correlated with seed viability (Fig. 6D). The three metrics of canopy fire severity (canopy severity rank, percent consumption of canopy biomass, and overstory CBI) were strongly intercorrelated (r = 0.60–0.70, p < 0.001). None of these metrics of canopy severity were significantly associated with total seed rain (p > 0.1). However, canopy severity rank was strongly correlated with seed viability (r = –0.54, p < 0.001, Fig. 6E), and overstory CBI showed a weaker negative correlation with viability (r = –0.36, p = 0.02, Fig. 6F). Canopy biomass consumption was not significantly related to seed viability (r = –0.26, p = 0.11). Relative importance of factors to variations in viable seed rain The conceptual variables of local climate, dispersal from live stands, fire severity, and prefire abundance in our multivariate structural equation model (Fig. 4) were represented by variables of site elevation, distance to the nearest live stand, canopy severity rank, and standing prefire basal area, respectively (Fig. 7). Initial examination of the model indicated that including site elevation as a squared value (elevation2) improved model fit but also made any causal pathways from untransformed elevation redundant and nonsignificant. Consequently, the model was fit using only squared elevation to represent elevation effects. The model structure showed no significant lack of fit with the observed data set (c2 test p > 0.1; Fig. 7A). The strongest paths in the model represented negative effects of elevation and canopy fire severity on seed viability, and a positive effect of standing basal area on total seed rain. The high significances of paths from viability or total seed to viable seed were expected, because the combination of those two factors comPublished by NRC Research Press

1582


Fig. 5. Bivariate relationships between prefire basal area of black spruce (m2ha–1) and total or viable seed rain (seedsm–2year–1). Plots show either the total (A, B) or only the standing portion (C, D) of prefire spruce basal area. Points represent individual sites (n = 38) and the plot axes are on a log10 scale.

pletely determines the estimate of viable seed rain. Unexplained error in the predictions of viable seed rain were due to small errors introduced through the rounding, sample averaging, and transformation steps that were implemented prior to model analysis. The hypothesized model predicted a substantial amount of the variation among sites in seed viability (R2 = 0.76), as a function of canopy severity, elevation, and distance to the nearest live stand (Fig. 7A). Of these factors, canopy severity and elevation showed a similar strength of effects, where changes in one standard deviation of either predictor variables were associated with a half deviation change in viability. The unstandardized path coefficients indicated that, when all other factors were held constant, a one-unit increase in canopy severity rank would lead to a decrease of 1.5% viability. Increases in elevation had a negative effect on seed viability, and use of squared elevation in the model indicated that an elevation change of 100 m would have a larger impact on seed viability at high elevations compared with low elevations. The model also indicated that as distance to the nearest live stand decreased, viability increased, but this effect was weaker than the effects of elevation or canopy severity. Total seed rain was most strongly predicted by standing basal area of prefire black spruce (Fig. 7). Although paths from elevation and distance to live stands contributed to the predicted variation in total seed (R2 = 0.46), z tests of individual path coefficients indicated that these paths were not significantly different from zero (p > 0.05). Path coefficients in our initial model indicated that there

were significant unanalyzed correlations between distance to the nearest unburned stand and canopy severity rank or site elevation (Fig. 7A). Examination of bivariate plots for these variables indicated that high elevation sites generally had greater distances to unburned stands. This relationship may represent a real effect of elevation on the size of burn patches, or may be a spatial artifact of our sampling. Of the four exogenous variables in the initial model, distance to unburned stands had the weakest effects on seed rain variables. As an alternative to this model, we assessed the fit of a second model (Fig. 7B) that did not include distance to unburned stands in the model. This simpler model adequately fit the covariance structure of the data (c2 test p > 0.05). The lower value of AIC for this second model suggested that it was an improvement over our original model, despite some small reductions in the amount of variance explained for total seed rain and viability (Fig. 7B). The total effect of an independent variable on viable seed rain can be calculated for each model by multiplying the standardized path coefficients connecting each variable to viable seed rain (Grace 2006; Table 1). When all the variables were included in the model (Fig. 7A), elevation, distance to unburned stands, canopy severity, and spruce basal area all had similar strength of effects on viable seed rain (Table 1). Removal of distance to unburned stands as a factor in the second model (Fig. 7B) increased the influence attributed to elevation effects, and elevation was estimated to have the strongest effect on viable seed rain. Spruce basal area had a consistent magnitude of effect on viable seed rain across both models. Published by NRC Research Press

Johnstone et al.

1583

Fig. 6. Selected bivariate relationships of site variables with total seed rain or seed percent viability. Total seed rain is plotted on a log10 scale against (A) site elevation, and (C) distance to the nearest unburned stand (log10 scale). Average seed viability is plotted on a log10 scale against (B) site elevation, (D) distance to the nearest unburned stand (log10 scale), (E) ranked index of stand canopy severity, and (F) overstory score of the composite burn index (CBI). Each point in a scatterplot represents a site (n = 38, except n = 36 in F). Linear correlation coefficients are given for each bivariate pair, and best-fit lines are shown for significant correlations, except for (B), where the best-fit line is a quadratic polynomial.

Discussion The results from this study provide partial support for a positive relationship between prefire species abundance (basal area) and postfire seed availability of serotinous conifers such as black spruce (Greene and Johnson 1999;

Ne’eman et al. 1999; Turner et al. 2007). However, our analyses indicate that the impact of this relationship can be strongly modified by fire behaviour. In the organic-rich soils of black spruce forests, variations in surface fire severity influence the number of trees that remained standing after fire, Published by NRC Research Press

1584


Fig. 7. Fitted structural equation models used to assess effects of prefire vegetation, site conditions, and fire characteristics on postfire seed rain (n = 38). (A) A multifactor model representing the hypothesized effects shown in Fig. 4. (B) A revised multifactor model in which the effects of distance to unburned stands have been removed. For both models, causal paths are indicated by directed arrows, and adjacent numbers are the standardized path coefficients with unstandardized coefficients in parentheses. Double-pointed, curved arrows indicate unanalyzed correlations, and adjacent numbers are the correlation coefficients. All coefficients are significant (a < 0.05) unless noted (ns, not significant). The strength of the causal path is indicated by the width of the path arrow, and nonsignificant directed paths are shown as broken lines. Nonsignificant correlations were removed from the model. R2 values indicate the amount of variation in endogenous variables that was explained by the model. Statistics summarized below each model indicate overall model fit (c2 test, where p < 0.05 indicates a significant lack of fit, and Akaike information criterion (AIC), where smaller values indicate improved fit). Squared elevation in the model was represented in units of 1000 m.

and thereby mediate basal area effects on total seed rain. The data also clearly indicate that variations in canopy fire severity can have important direct effects on seed viability.

We observed a negative relationship between seed viability and the level of canopy consumption by fire, which suggests that sites with high canopy combustion experienced suffiPublished by NRC Research Press

Johnstone et al.

1585 Table 1. Summary of standardized total effects of causal variables on viable seed rain (m–2year–1), based on the two models shown in Fig. 7. Model A (original, 4-factor model)

B (revised, 3-factor model)

Exogenous variable Elevation2 Distance to unburned Canopy severity rank Standing spruce basal area Elevation2 Canopy severity rank Standing spruce basal area

Standardized total effect on viable seed rain –0.429 –0.281 –0.290 0.380 –0.566 –0.342 0.395

Note: Viable seed rain, standing spruce basal area, and distance to unburned stands were log-transformed prior to analysis.

cient heating to kill seeds in the aerial seed bank (Zasada et al. 1979; Arseneault 2001; de Groot et al. 2004). In the absence of information about fire effects, prefire stand data were unable to predict observed patterns of viable seed rain (Fig. 5). This is consistent with other observations of serotinous conifers that suggest relations between prefire basal area and seedling recruitment differ between levels of canopy fire severity (Greene et al. 2004), and that variations in canopy severity can cause strong differences in seedling recruitment in the absence of variations in prefire stand structure (Arseneault 2001). Landscape factors also clearly played a role in determining postfire seed rain in black spruce forests found in the complex topography of interior Alaska. Seed viability declined significantly with site elevation, and high elevation sites (>850 m above sea level) showed consistently low rates of seed viability. Similar reductions in seed quality, but not quantity, have been found for conifer populations near latitudinal (Black and Bliss 1980; Sirois 2000) and elevational tree line (van Mantgem et al. 2006), and have been attributed to low heat sums near tree line (Sirois 2000). Current or future warming trends that increase temperatures at tree line are likely to increase the production of viable spruce seed, which may be a key step in permitting spruce expansion in woodland or treeless environments. It remains unclear how other concurrent changes in climate may affect black spruce seed production across its range, perhaps causing an increased frequency of drought stress similar to that observed for white spruce in upland forests in Alaska (Barber et al. 2000). Because of covariance between elevation and distance to the nearest live stand, our data were unable to clearly distinguish between the effects of these two factors on seed rain. Seed viability appeared to decline with increased distances from live stands, with no concurrent effect on seed quantity. Seeds dispersed from nearby, unburned stands may have had higher rates of viability because they would have avoided any negative effects of heating during the fire. Nevertheless, our analyses suggest that the effects of seed dispersal from nearby live stands had a relatively minor role in determining viable seed rain, as removal of distance from live stands as a variable in the models had little impact on explanatory power. Most studies of seed availability for black spruce report on seeds still held within cones or dispersed in unburned

stands, and there are few studies to which we can compare our estimates of postfire seed rain. For three burned black spruce stands in Alaska, Zasada et al. (1979) report an average of 289–386 total seedsm–2year–1 dispersed during the first year after fire, and 41–77 total seedsm–2year–1 in the second and third year. Their estimates of seed rain for the second and third year fully overlap our observed range of total, annual seed rain. However, their estimates of viable seed rain for the second and third year after fire (28–47 viable seedsm–2year–1) are in the highest part of the range observed in this study. Historical fire records indicate that 2004 was an extreme fire year in Alaska (Todd and Jewkes 2006). High levels of canopy mortality and fuel consumption in the 2004 burns may account for the generally lower level of seed viability observed in this study compared with seed from black spruce stands burned in 1971 (Zasada et al. 1979). The sites used in this study were restricted to areas near existing roads and were subjectively sampled to capture a broad range of site conditions found in the 2004 burns. Consequently, average values should not be assumed to represent ‘‘typical’’ conditions within interior Alaska forests. Estimates of the importance of different factors in driving variations in black spruce seed rain may also be conditional on the range of characteristics found in the 2004 burns. Extrapolation of our results to other regions of boreal forest, or other burn years, should be done with caution and incorporate data from additional empirical studies. When interpreting our data, it is important to note that most of our sites burned in June–August of 2004, and our seed collections were not initiated until the following summer of 2005. Consequently, we are likely to have missed the initial and largest pulse of black spruce seed released in the first year after the fire (Zasada et al. 1979). However, we believe our data on seed rain are biologically relevant for two reasons. Firstly, our observations of seedling establishment at the same sites (J. Johnstone, unpublished data) show that most (63%) of the spruce seedlings observed by 2008 germinated in 2006. Given low rates of seed dormancy in black spruce (Zasada et al. 1992), these seedlings would have arisen from seeds dispersed in 2005– 2006, the period covered by our seed collection. Secondly, comparison of our data with other observations of black spruce suggests that we were able to capture a large portion of the postfire seed dispersal in our stands. If we use the equations provided by Greene and Johnson (1999) to predict the total size of the aerial seed bank of black spruce using Published by NRC Research Press

1586

stand basal area, from our observed basal area range of 0.1– 28.3m–2ha–1 we predict a potential total seed rain of 2.9– 620 seedsm–2 across our study sites. For the 13 sites for which we had complete data for the 2 years, the observed total seed rain over the 2 years was 15–309 seedsm–2, approximately half of the estimated potential. Careful age structure analysis in boreal forests indicates that most seedling establishment is concentrated within the first decade after fire (St-Pierre and Gagnon 1992; Gutsell and Johnson 2002; Peters et al. 2006), and permanent plots indicate that recruitment likely peaks within the first 3 years after fire (Charron and Greene 2002; Johnstone et al. 2004). Consequently, our measurements of seed rain 2 and 3 years after fire are likely to have captured a substantial portion of the biologically relevant seed rain. The results of this study provide a new insight into the relative roles of prefire vegetation, site factors, and fire effects in determining seed availability and potential regeneration of a dominant boreal conifer. This knowledge is a key component in improving our ability to predict patterns of stand regeneration after fire and other disturbances in the boreal forest. However, the consequences of variation in seed rain to future forest structure and composition will be dependent on the degree to which seed rain influences seedling establishment. Existing research suggests that large variations in seedbed quality (Charron and Greene 2002; Johnstone and Chapin 2006a; Greene et al. 2007), and in some situations, seed predation by granivores (Peters et al. 2003; Greene et al. 2004) may be as or more important than variations in seed rain in driving tree recruitment patterns in boreal forests. However, observations of very low conifer recruitment following disturbances that reduce seed availability in the aerial seed bank (Payette et al. 2000; Jasinski and Payette 2005; Johnstone and Chapin 2006b) and low rates of seed production at species range limits (Black and Bliss 1980; Sirois 2000) strongly suggest that there are conditions where variations in seed rain can have important impacts on the postfire recovery of boreal conifers. The results from this study are an important step in understanding and predicting variations in postfire seed rain, and provide a basis for assessing situations where seed rain is likely to limit conifer regeneration after fire.

Acknowledgements This study was made possible by funding from the US Joint Fire Science Program (project 05-1-2-06) and in-kind support from the Bonanza Creek Long-Term Ecological Research program. We gratefully acknowledge assistance in the field and lab from Emily Bernhardt, Michelle Bifelt, Rachel Bifelt, Carissa Brown, Terry Chapin, Julie Cossette, Teresa Hollingsworth, Dana Nossov, Janet Pritchard, and Jayme Viglas. We also appreciate the discussions with Terry Chapin and thoughtful comments from two anonymous reviewers that helped us improve the quality of this paper.

References Agee, J.K. 1998. Fire and pine ecosystems. In Ecology and biogeography of Pinus. Edited by D.M. Richardson. Cambridge University Press, Cambridge. pp. 193–218. Anderson, D.R., Burnham, K.P., and Thompson, W.L. 2000. Null

Can. J. For. Res. Vol. 39, 2009 hypothesis testing: problems, prevalence, and an alternative. J. Wildl. Manage. 64(4): 912–923. doi:10.2307/3803199. Arbuckle, J.L. 2007. AMOS 16.0 user’s guide. SPSS, Inc., Chicago, Ill. Arseneault, D. 2001. Impact of fire behavior on postfire forest development in a homogeneous boreal landscape. Can. J. For. Res. 31: 1367–1374. doi:10.1139/cjfr-31-8-1367. Auerbach, N.A., Walker, M.D., and Walker, D.A. 1997. Effects of roadside disturbance on substrate and vegetation properties in arctic tundra. Ecol. Appl. 7(1): 218–235. doi:10.1890/10510761(1997)007[0218:EORDOS]2.0.CO;2. Barber, V.A., Juday, G.P., and Finney, B.P. 2000. Reduced growth of Alaskan white spruce in the twentieth century from temperature-induced drought stress. Nature, 405: 668–673. doi:10.1038/35015049. PMID:10864320. Bennie, J., Hill, M.O., Baxter, R., and Huntley, B. 2006. Influence of slope and aspect on long-term vegetation change in British chalk grasslands. J. Ecol. 94: 355–368. doi:10.1111/j.13652745.2006.01104.x. Black, R.A., and Bliss, L.C. 1980. Reproductive ecology of Picea mariana (Mill.) Bsp., at tree line near Inuvik, Northwest Territories, Canada. Ecol. Monogr. 50(3): 331–354. doi:10.2307/ 2937255. Borchert, M., Johnson, M., Schreiner, D.S., and Wall, S.B.V. 2003. Early postfire seed dispersal, seedling establishment and seedling mortality of Pinus coulteri (D. Don) in central coastal California, USA. Plant Ecol. 168(2): 207–220. doi:10.1023/ A:1024447811238. Calef, M.P., McGuire, A.D., Epstein, H.E., Rupp, T.S., and Shugart, H.H. 2005. Analysis of vegetation distribution in interior Alaska and sensitivity to climate change using a logistic regression approach. J. Biogeogr. 32: 863–878. doi:10.1111/j.13652699.2004.01185.x. Charron, I., and Greene, D.F. 2002. Post-wildfire seedbeds and tree establishment in the southern mixedwood boreal forest. Can. J. For. Res. 32: 1607–1615. doi:10.1139/x02-085. Clark, J.S., Lewis, M., McLachlan, J.S., and HillRisLambers, J. 2003. Estimating population spread: what can we forecast and how well? Ecology, 84(8): 1979–1988. doi:10.1890/01-0618. Cumming, S.G. 2001. Forest type and wildfire in the Alberta boreal mixedwood: what do fires burn? Ecol. Appl. 11(1): 97–110. doi:10.1890/1051-0761(2001)011[0097:FTAWIT]2.0.CO;2. de Groot, W.J., Bothwell, P.M., Taylor, S.W., Wotton, B.M., Stocks, B.J., and Alexander, M.E. 2004. Jack pine regeneration and crown fires. Can. J. For. Res. 34: 1634–1641. doi:10.1139/x04-073. Dovciak, M., Reich, P.B., and Frelich, L.E. 2003. Seed rain, safe sites, competing vegetation, and soil resources spatially structure white pine regeneration and recruitment. Can. J. For. Res. 33: 1892–1904. doi:10.1139/x03-115. Epting, J., Verbyla, D.L., and Sorbel, B. 2005. Evaluation of remotely sensed indices for assessing burn severity in interior Alaska using Landsat TM and ETM+. Remote Sens. Environ. 96: 328– 339. doi:10.1016/j.rse.2005.03.002. Grace, J.B. 2006. Structural equation modelling and natural systems. Cambridge University Press, New York. Greene, D.F., and Johnson, E.A. 1999. Modelling recruitment of Populus tremuloides, Pinus banksiana, and Picea mariana following fire in the mixedwood boreal forest. Can. J. For. Res. 29: 462–473. doi:10.1139/cjfr-29-4-462. Greene, D.F., and Johnson, E.A. 2000. Tree recruitment from burn edges. Can. J. For. Res. 30: 1264–1274. doi:10.1139/cjfr-30-81264. Greene, D.F., Zasada, J.C., Sirois, L., Kneeshaw, D., Morin, H., Charron, I., and Simard, M.-J. 1999. A review of the regeneraPublished by NRC Research Press

Johnstone et al. tion dynamics of North American boreal forest tree species. Can. J. For. Res. 29: 824–839. doi:10.1139/cjfr-29-6-824. Greene, D.F., Noel, J., Bergeron, Y., Rousseau, M., and Gauthier, S. 2004. Recruitment of Picea mariana, Pinus banksiana, and Populus tremuloides across a burn severity gradient following wildfire in the southern boreal forest of Quebec. Can. J. For. Res. 34: 1845–1857. doi:10.1139/x04-059. Greene, D.F., Macdonald, E.S., Haeussler, S., Domenicano, S., Noel, J., Jayen, K., Charron, I., Gauthier, S., Hunt, S., Gielau, E.T., Bergeron, Y., and Swift, L. 2007. The reduction of organic-layer depth by wildfire in the North American boreal forest and its effect on tree recruitment by seed. Can. J. For. Res. 37: 1012–1023. doi:10.1139/X06-245. Gutsell, S., and Johnson, E.A. 2002. Accurately ageing trees and examining their height-growth rates: implications for interpreting forest dynamics. J. Ecol. 90: 153–166. doi:10.1046/j.00220477.2001.00646.x. Hollingsworth, T.N., Walker, M.D., Chapin, F.S., III, and Parsons, A.L. 2006. Scale-dependent environmental controls over species composition in Alaskan black spruce communities. Can. J. For. Res. 36: 1781–1796. doi:10.1139/X06-061. Jasinski, J.P.P., and Payette, S. 2005. The creation of alternative stable states in the southern boreal forest, Quebec, Canada. Ecol. Monogr. 75(4): 561–583. doi:10.1890/04-1621. Johnson, E.A. 1992. Fire and vegetation dynamics: studies from the North American boreal forest. Cambridge University Press, Cambridge. Johnstone, J.F., and Chapin, F.S., III. 2006a. Effects of soil burn severity on post-fire tree recruitment in boreal forests. Ecosystems (N.Y., Print), 9: 14–31. doi:10.1007/s10021-004-0042-x. Johnstone, J.F., and Chapin, F.S., III. 2006b. Fire interval effects on successional trajectory in boreal forests of Northwest Canada. Ecosystems (N. Y., Print), 9: 268–277. doi:10.1007/s10021-0050061-2. Johnstone, J.F., Chapin, F.S., III, Foote, J., Kemmett, S., Price, K., and Viereck, L. 2004. Decadal observations of tree regeneration following fire in boreal forests. Can. J. For. Res. 34: 267–273. doi:10.1139/x03-183. Key, C.H., and Benson, N.C. 2005. Landscape assessment sampling and analysis methods. In FIREMON: fire effects monitoring and inventory system. Edited by D.C. Lutes, R.E. Keane, J.F. Caratti, C.H. Key, N.C. Benson, S. Sutherland, and L.J. Gangi. USDA For. Serv. Gen. Tech. Rep. RMRS-GTR-164-CD. pp. LA1–51. Lamont, B.B., Lemaitre, D.C., Cowling, R.M., and Enright, N.J. 1991. Canopy seed storage in woody-plants. Bot. Rev. 57(4): 277–317. doi:10.1007/BF02858770. Larsen, C.P.S. 1997. Spatial and temporal variations in boreal forest fire frequency in northern Alberta. J. Biogeogr. 24: 663–673. Leadem, C.L., Gillies, S.L., Yearsley, H.K., Sit, V., Spittlehouse, D.L., and Burton, P.J. 1997. Field studies of seed biology. B.C. Minist. For. Land Manage. Handb. No. 40. Lentile, L.B., Holden, Z.A., Smith, A.M.S., Falkowski, M.J., Hudak, A.T., Morgan, P., Lewis, S.A., Gessler, P.E., and Benson, N.C. 2006. Remote sensing techniques to assess active fire characteristics and post-fire effects. Int. J. Wildland Fire, 15: 319– 345. doi:10.1071/WF05097. Mack, M.C., Treseder, K.K., Manies, K.L., Harden, J.W., Schuur, E.A.G., Vogel, J.G., Randerson, J.T., and Chapin, F.S., III. 2008. Recovery of aboveground plant biomass and productivity after fire in mesic and dry black spruce forests of interior Alaska. Ecosystems (N. Y., Print), 11: 209–225. doi:10.1007/ s10021-007-9117-9. McCune, B., and Keon, D. 2002. Equations for potential annual direct incident radiation and heat load. J. Veg. Sci. 13: 603–606.

1587 McIntire, E.J.B., Duchesneau, R., and Kimmins, J.P. 2005. Seed and bud legacies interact with varying fire regimes to drive long-term dynamics of boreal forest communities. Can. J. For. Res. 35: 2765–2773. doi:10.1139/x05-187. Ne’eman, G., Fotheringham, C.J., and Keeley, J.E. 1999. Patch to landscape patterns in post fire recruitment of a serotinous conifer. Plant Ecol. 145: 235–242. doi:10.1023/A:1009869803192. Payette, S., Bhiry, N., Delwaide, A., and Simard, M. 2000. Origin of the lichen woodland at its southern range limit in eastern Canada: the catastrophic impact of insect defoliators and fire on the spruce-moss forest. Can. J. For. Res. 30: 288–305. doi:10. 1139/cjfr-30-2-288. Peters, S., Boutin, S., and Macdonald, E. 2003. Pre-dispersal seed predation of white spruce cones in logged boreal mixedwood forest. Can. J. For. Res. 33: 33–40. doi:10.1139/x02-153. Peters, V.S., Macdonald, S.E., and Dale, M.R.T. 2005. The interaction between masting and fire is key to white spruce regeneration. Ecology, 86(7): 1744–1750. doi:10.1890/03-0656. Peters, V.S., Macdonald, E.S., and Dale, R.T. 2006. Patterns of initial versus delayed regeneration of white spruce in boreal mixedwood succession. Can. J. For. Res. 36: 1597–1609. doi:10. 1139/X06-020. Purdy, B.G., Macdonald, S.E., and Dale, M.R.T. 2002. The regeneration niche of white spruce following fire in the mixedwood boreal forest. Silva Fenn. 36(1): 289–305. Rowe, J.S. 1972. Forest regions of Canada. Canadian Forestry Service, Department of the Environment, Ottawa, Ont. Schoennagel, T., Turner, M.G., and Romme, W.H. 2003. The influence of fire interval and serotiny on postfire lodgepole pine density in Yellowstone National Park. Ecology, 84(11): 2967–2978. doi:10.1890/02-0277. Schwilk, D.W., and Ackerly, D.D. 2001. Flammability and serotiny as strategies: correlated evolution in pines. Oikos, 94(2): 326– 336. doi:10.1034/j.1600-0706.2001.940213.x. Shipley, B. 1997. Exploratory path analysis with applications in ecology and evolution. Am. Nat. 149(6): 1113–1138. doi:10. 1086/286041. PMID:18811266. Shulski, M., and Wendler, G. 2007. The climate of Alaska. University of Alaska Press, Fairbanks, Alaska. Simard, M.-J., Bergeron, Y., and Sirois, L. 1998. Conifer seedling recruitment in a southeastern Canadian boreal forest: the importance of substrate. J. Veg. Sci. 9: 575–582. doi:10.2307/3237273. Sirois, L. 2000. Spatiotemporal variation in black spruce cone and seed crops along a boreal forest – tree line transect. Can. J. For. Res. 30: 900–909. doi:10.1139/cjfr-30-6-900. St-Pierre, H., and Gagnon, R. 1992. Re´geńe´ration apre`s feu de l’e´pinette noir (Picea mariana) et du pin gris (Pinus banksiana) dans la foreˆt boreále, Que´bec. Can. J. For. Res. 22: 474–481. doi:10.1139/x92-062. Tapias, R., Climent, J., Pardos, J.A., and Gil, L. 2004. Life histories of Mediterranean pines. Plant Ecol. 171(1–2): 53–68. Todd, S.K., and Jewkes, H.A. 2006. Fire in Alaska: a history of organized fire suppression and management in the last frontier. University of Alaska Fairbanks. USDA Agric. For. Exp. Station Bull. 113. Turner, M.G., Turner, D.M., Romme, W.H., and Tinker, D.B. 2007. Cone production in young post-fire Pinus contorta stands in Greater Yellowstone (USA). For. Ecol. Manage. 242: 119– 126. doi:10.1016/j.foreco.2006.12.032. Van Cleve, K., and Viereck, L.A. 1981. Forest succession in relation to nutrient cycling in the boreal forest of Alaska. In Forest succession, concepts and application. Edited by D.C. West, H.H. Shugart, and D.B. Botkin. Springer-Verlag, New York. pp. 184– 211. Published by NRC Research Press

1588 Van Cleve, K., Chapin, F.S., III, Dryness, C.T., and Viereck, L.A. 1991. Element cycling in taiga forest: state-factor control. Bioscience, 41: 78–88. doi:10.2307/1311560. van Mantgem, P.J., Stephenson, N.L., and Keeley, J.E. 2006. Forest reproduction along a climatic gradient in the Sierra Nevada, California. For. Ecol. Manage. 225: 391–399. doi:10.1016/j.foreco. 2006.01.015. Verkaik, I., and Espelta, J.M. 2006. Post-fire regeneration thinning, cone production, serotiny and regeneration age in Pinus halepensis. For. Ecol. Manage. 231(1–3): 155–163. doi:10.1016/j. foreco.2006.05.041. Viereck, L.A. 1983. The effects of fire in black spruce ecosystems of Alaska and northern Canada. In The role of fire in northern circumpolar ecosystems. Edited by R.W. Wein, and D.A. MacLean. Wiley, Chichester. pp. 201–220. Viereck, L.A., Van Cleve, K., and Dyrness, C.T. 1986. Forest eco-

Can. J. For. Res. Vol. 39, 2009 system distribution in the taiga environment. In Forest ecosystems in the Alaskan taiga. Edited by K. Van Cleve, F.S. Chapin, III, P.W. Flanagan, L.A. Viereck, and C.T. Dyrness. Springer-Verlag, New York. pp. 22–43. Westoby, M. 1984. The self-thinning rule. Adv. Ecol. Res. 14: 167–225. doi:10.1016/S0065-2504(08)60171-3. Zasada, J.C., Viereck, L.A., and Foote, M.J. 1979. Quantity and quality of dispersed seed. In Ecological effects of the Wickersham Dome fire near Fairbanks, Alaska. Edited by L. Viereck and C.T. Dyrness. USDA For. Serv. Gen. Tech. Rep. PNW-90, Portland, Oreg. pp. 45–49. Zasada, J.C., Sharik, T.L., and Nygren, M. 1992. The reproductive process in boreal forest trees. In A systems analysis of the global boreal forest. Edited by H.H. Shugart, R. Leemans, and G.B. Bonan. Cambridge University Press, Cambridge. pp. 85–125.
