natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

10
Natural regeneration of white spruce in aspen- dominated boreal mixedwoods following harvesting Jonathan Martin-DeMoor, Victor J. Lieffers, and S. Ellen Macdonald Abstract: In some boreal forests sites, there are considerable amounts of natural regeneration of white spruce (Picea glauca (Moench) Voss) after logging, even without silvicultural treatments to encourage establishment. We assessed the factors controlling the amount of this regeneration 8–15 years postharvest on previously aspen-dominated (Populus tremu- loides Michx.) boreal mixedwood sites. We surveyed 162 transects across 81 cutovers, exploring the effects of mast years, season of harvest, distribution of seed trees, weather conditions around the time of harvest, and abundance of grass or woody vegetation on white spruce regeneration. Substantial amounts of naturally regenerated white spruce were found; however, sites with no seed trees had virtually no spruce regeneration. Average stocking was 7% (percentage of 9 m 2 plots along a transect across a cutover that had at least one seedling), ranging from 0% to 62%. Stocking levels were higher in cutblocks that had been harvested in the summer, prior to seedfall of a mast year, and where there was a seed source within 60 m. Stocking was lower when conditions were cool and wet the year before and 2 years after harvest and when the site contained extensive cover of grass or woody vegetation. Re ´sume ´: Sur certaines stations forestie `res bore ´ales, il y a une quantite ´ conside ´rable de semis naturels d’e ´pinette blanche (Picea glauca (Moench) Voss) a ` la suite d’une coupe, me ˆme sans traitement sylvicole visant a ` promouvoir leur e ´tablisse- ment. Nous avons e ´value ´ les facteurs qui de ´terminent la quantite ´ de cette re ´ge ´ne ´ration sur des stations coupe ´es il y a 8 a ` 15 ans et pre ´ce ´demment occupe ´es par des peuplements mixtes bore ´aux domine ´s par le peuplier faux-tremble (Populus tremuloides Michx.). Nous avons e ´tabli 162 transects re ´partis dans 81 aires de coupe pour e ´valuer les effets des anne ´es se- mencie `res, de la saison durant laquelle la coupe a eu lieu, de la distribution des semenciers, des conditions climatiques a ` l’e ´poque de la coupe et de l’abondance de la ve ´ge ´tation herbace ´e et ligneuse sur la re ´ge ´ne ´ration d’e ´pinette blanche. Un nombre substantiel de semis naturels d’e ´pinette blanche a e ´te ´ inventorie ´. Cependant, les stations sans semenciers n’avaient pratiquement pas de re ´ge ´ne ´ration d’e ´pinette. Le coefficient de distribution de la re ´ge ´ne ´ration e ´tait en moyenne de 7 % (pourcentage de placettes de 9 m 2 le long d’un transect e ´tabli dans une coupe ou ` il y avait au moins un semis) et variait de 0 % a ` 62 %. Le coefficient de distribution de la re ´ge ´ne ´ration e ´tait plus e ´leve ´ sur les stations re ´colte ´es pendant l’e ´te ´, avant la pluie de graines d’une anne ´e semencie `re et ou ` il y avait une source de semences a ` moins de 60 m. Le coefficient de distribution de la re ´ge ´ne ´ration e ´tait plus faible lorsque les conditions climatiques durant l’anne ´e pre ´ce ´dant la coupe et deux ans apre `s la coupe e ´taient fraı ˆches et humides et lorsque la station e ´tait abondamment couverte de ve ´ge ´tation herba- ce ´e ou ligneuse. [Traduit par la Re ´daction] Introduction Boreal mixedwood forests composed of trembling aspen (Populus tremuloides Michx.) and white spruce (Picea glauca (Moench) Voss) are found on mesic and subhygric sites across large portions of North America from Alaska to Quebec. The spruce in these stands originated by means of natural regeneration from seed, usually following fire. The conditions for regeneration of spruce are exacting and there is a large volume of literature on this topic for postfire re- generation (e.g., Purdy et al. 2002; Peters et al. 2005; Kemball et al. 2006; Greene et al. 2007). There is less work describing conditions for natural regeneration after logging (but see Zasada 1986; Timoney and Peterson 1996; Greene et al. 1999; Wurtz and Zasada 2001). White spruce is a masting species with very large peri- odic seed crops synchronized across the landscape, with high interannual variation (Kelly 1994; Lamontagne and Boutin 2007). Mast years generally occur every 2–5 years (Waldron 1965; Peters et al. 2005; Stewart et al. 2005) but may be up to 12 years apart (Wurtz and Zasada 2001). The limited range of dispersal of white spruce seed limits the area in which successful natural regeneration can be ex- pected. The deposition of seed declines exponentially away from the seed source (Dobbs 1976; Greene and Johnson 1996), with maximum effective dispersal occurring at about 100–120 m (e.g., Youngblood and Max 1992; Zasada et al. 1992), although some longer-range dispersal does occur (Zasada and Lovig 1983; Stewart et al. 1998). Received 5 July 2009. Accepted 4 January 2010. Published on the NRC Research Press Web site at cjfr.nrc.ca on 15 March 2010. J. Martin-DeMoor, 1 V.J. Lieffers, and S.E. Macdonald. Department of Renewable Resources, University of Alberta, Edmonton, AB T6G 2H1, Canada. 1 Corresponding author (e-mail: [email protected]). 585 Can. J. For. Res. 40: 585–594 (2010) doi:10.1139/X10-016 Published by NRC Research Press

Upload: s-ellen

Post on 30-Mar-2017

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

Natural regeneration of white spruce in aspen-dominated boreal mixedwoods followingharvesting

Jonathan Martin-DeMoor, Victor J. Lieffers, and S. Ellen Macdonald

Abstract: In some boreal forests sites, there are considerable amounts of natural regeneration of white spruce (Piceaglauca (Moench) Voss) after logging, even without silvicultural treatments to encourage establishment. We assessed thefactors controlling the amount of this regeneration 8–15 years postharvest on previously aspen-dominated (Populus tremu-loides Michx.) boreal mixedwood sites. We surveyed 162 transects across 81 cutovers, exploring the effects of mast years,season of harvest, distribution of seed trees, weather conditions around the time of harvest, and abundance of grass orwoody vegetation on white spruce regeneration. Substantial amounts of naturally regenerated white spruce were found;however, sites with no seed trees had virtually no spruce regeneration. Average stocking was 7% (percentage of 9 m2 plotsalong a transect across a cutover that had at least one seedling), ranging from 0% to 62%. Stocking levels were higher incutblocks that had been harvested in the summer, prior to seedfall of a mast year, and where there was a seed sourcewithin 60 m. Stocking was lower when conditions were cool and wet the year before and 2 years after harvest and whenthe site contained extensive cover of grass or woody vegetation.

Resume : Sur certaines stations forestieres boreales, il y a une quantite considerable de semis naturels d’epinette blanche(Picea glauca (Moench) Voss) a la suite d’une coupe, meme sans traitement sylvicole visant a promouvoir leur etablisse-ment. Nous avons evalue les facteurs qui determinent la quantite de cette regeneration sur des stations coupees il y a 8 a15 ans et precedemment occupees par des peuplements mixtes boreaux domines par le peuplier faux-tremble (Populustremuloides Michx.). Nous avons etabli 162 transects repartis dans 81 aires de coupe pour evaluer les effets des annees se-mencieres, de la saison durant laquelle la coupe a eu lieu, de la distribution des semenciers, des conditions climatiques al’epoque de la coupe et de l’abondance de la vegetation herbacee et ligneuse sur la regeneration d’epinette blanche. Unnombre substantiel de semis naturels d’epinette blanche a ete inventorie. Cependant, les stations sans semenciers n’avaientpratiquement pas de regeneration d’epinette. Le coefficient de distribution de la regeneration etait en moyenne de 7 %(pourcentage de placettes de 9 m2 le long d’un transect etabli dans une coupe ou il y avait au moins un semis) et variaitde 0 % a 62 %. Le coefficient de distribution de la regeneration etait plus eleve sur les stations recoltees pendant l’ete,avant la pluie de graines d’une annee semenciere et ou il y avait une source de semences a moins de 60 m. Le coefficientde distribution de la regeneration etait plus faible lorsque les conditions climatiques durant l’annee precedant la coupe etdeux ans apres la coupe etaient fraıches et humides et lorsque la station etait abondamment couverte de vegetation herba-cee ou ligneuse.

[Traduit par la Redaction]

Introduction

Boreal mixedwood forests composed of trembling aspen(Populus tremuloides Michx.) and white spruce (Piceaglauca (Moench) Voss) are found on mesic and subhygricsites across large portions of North America from Alaska toQuebec. The spruce in these stands originated by means ofnatural regeneration from seed, usually following fire. Theconditions for regeneration of spruce are exacting and thereis a large volume of literature on this topic for postfire re-

generation (e.g., Purdy et al. 2002; Peters et al. 2005;Kemball et al. 2006; Greene et al. 2007). There is less workdescribing conditions for natural regeneration after logging(but see Zasada 1986; Timoney and Peterson 1996; Greeneet al. 1999; Wurtz and Zasada 2001).

White spruce is a masting species with very large peri-odic seed crops synchronized across the landscape, withhigh interannual variation (Kelly 1994; Lamontagne andBoutin 2007). Mast years generally occur every 2–5 years(Waldron 1965; Peters et al. 2005; Stewart et al. 2005) butmay be up to 12 years apart (Wurtz and Zasada 2001). Thelimited range of dispersal of white spruce seed limits thearea in which successful natural regeneration can be ex-pected. The deposition of seed declines exponentially awayfrom the seed source (Dobbs 1976; Greene and Johnson1996), with maximum effective dispersal occurring at about100–120 m (e.g., Youngblood and Max 1992; Zasada et al.1992), although some longer-range dispersal does occur(Zasada and Lovig 1983; Stewart et al. 1998).

Received 5 July 2009. Accepted 4 January 2010. Published onthe NRC Research Press Web site at cjfr.nrc.ca on 15 March2010.

J. Martin-DeMoor,1 V.J. Lieffers, and S.E. Macdonald.Department of Renewable Resources, University of Alberta,Edmonton, AB T6G 2H1, Canada.

1Corresponding author (e-mail: [email protected]).

585

Can. J. For. Res. 40: 585–594 (2010) doi:10.1139/X10-016 Published by NRC Research Press

Page 2: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

White spruce seed dispersed in the fall must germinatethe next spring, as the seed quickly loses viability in thesoil (Nienstaedt and Zasada 1990). Microsite quality de-creases within 2 years after disturbance (Arlidge 1967;Stewart et al. 2000; Purdy et al. 2002); therefore, time ofmasting relative to seedbed availability is key to the suc-cessful establishment of white spruce (Peters et al. 2005).Excellent mineral soil seedbeds are produced after fire(Greene et al. 2005; Kemball et al. 2005) and after loggingif site preparation is applied; the harvesting and skiddingmay also disturb the forest floor to create favourable seed-beds (Wurtz and Zasada 2001), especially in summer-harvested cutblocks (Berger et al. 2004).

White spruce germinants require continuous moistureavailability (Day 1963; Hogg and Schwarz 1997), althoughexcessive moisture can cause mortality due to floodingstress (Timoney and Peterson 1996) and create damp condi-tions conducive to damping-off fungi, such as Phytophy-thora spp. or Fusarium spp. (Hiratsuka 1987). Competingvegetation, especially the native grass Calamagrostis cana-densis (Eis 1981; Lieffers et al. 1993) and dense aspen re-generation (Kabzems and Lousier 1992; Lieffers and Stadt1994), negatively impact growth of spruce seedlings (Manet al. 2008).

While each of the above factors has been identified beforeas important for establishment of spruce, there have beenfew studies that explored simultaneously the impact of awide range of these variables on white spruce regenerationsuccess. Our objective was to quantify the levels of naturalwhite spruce regeneration in harvested boreal mixedwoodstands approximately a decade after logging in relation tomasting, season of harvest, distribution of seed trees, cli-matic conditions, and competing vegetation. Aspen-dominated stands were chosen for study, firstly, becausespruce regeneration in these stands is less studied, and sec-ondly, because we wanted to examine the potential of natu-ral regeneration to contribute to a basic level of stocking ofwhite spruce in these stands as an alternative to artificial re-generation (Lieffers et al. 2008) and as an approach thatcould help maintain the forest composition and stand struc-

tural attributes that have been linked to maintenance of bio-diversity in boreal mixedwood forests (Hobson and Bayne2000; Work et al. 2004; Buddle et al. 2006; Macdonald andFenniak 2007; Hart and Chen 2008).

Methods

Field samplingThe field sites were located in west-central Alberta be-

tween 52826’N and 54801’N and between 115814’W and117812’W, with elevations between 838 m and 1207 m a.s.l.All sites were within the Lower Foothills natural subregion.Mixedwood stands of aspen, balsam poplar (Populus balsa-mifera L), white spruce, and lodgepole pine (Pinus contortaLoudon) dominate this natural subregion (Beckingham et al.1996). The landscape is characterized by rolling topographywith ridges underlain by sandstone and shale till (NaturalRegions Committee 2006); most of the study sites, however,were generally flat.

Prior to harvesting, stands were aspen dominated, andwhite spruce abundance ranged from 0% to 25% canopycover. We sampled regenerated stands that were harvestedin spruce mast years and in years with low seed crops. Re-gional masting occurred in 1993 and 1998 (Stewart et al.2000; Peters et al. 2005; Lamontagne and Boutin 2007).We also assessed the effect of season of harvest by sam-pling regenerated cutblocks that were harvested either inthe summer prior to seedfall (July through early September)or in the winter (late November through March) when mostseed would already be on the forest floor (Table 1). To-gether these two factors resulted in four ‘‘treatments’’ repre-senting the time of harvest: summer before mast, winterafter mast, summer in non-mast year, and winter in non-mast year.

We selected only cutblocks that had passed the localstandard for regeneration of deciduous trees, i.e., at leastone deciduous tree in 80% of 0.001 ha plots. Sampled standshad received no silvicultural treatments to enhance regener-ation after logging and were 8–15 years postharvest at thetime that we conducted our sampling.

Table 1. Time line of treatments by season of harvest, year, and seed crop strength.

Date of harvest

Treatmenta Season Harvest year Fall seed crop From ToSbm Summer 1993 High 19 July 8 Sept.Wam Winter 1993–1994 21 Dec. 21 Mar.

Summer 1994 Very lowWim Winter 1994–1995 12 Jan. 2 Feb.Sim Summer 1995 Low 16 July 10 Sept.Wim Winter 1995–1996 5 Dec. 5 Mar.Sim Summer 1996 Low 21 Aug. 11 Sept.

Winter 1996–1997Summer 1997 LowWinter 1997–1998

Sbm Summer 1998 Very high 19 Aug. 4 Sept.Wam Winter 1998–1999 12 Nov. 31 Mar.

Note: Also shown is the range of harvest dates for each treatment.aSbm, harvested the summer before dispersal of a mast seed crop; Wam, harvested the winter

after dispersal of a mast seed crop; Sim, summer harvest in a non-mast (intermast) year; Wim,winter harvest in a non-mast (intermast) year.

586 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

Page 3: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

Transects of 6 m width were established across thesampled cutblocks in April and May of 2006. Two transectswere often established in a single large cutblock, but theywere always widely separated (see Fig. 1). Geographic coor-dinates of both ends of each transect were marked by globalpositioning system. To test for the effect of seed dispersalon white spruce regeneration, we tried to start or end tran-sects near mature white spruce along the block edge and,when possible, to direct transects past residual spruce withinthe block. However, many cutblocks had no seed treesnearby.

Surveyors recorded the number of white spruce on eachside of the centre-line of the transect and their distancealong the transect. Also, block boundaries and seed treeswere mapped, and the portions of the transects to be re-moved from analysis, e.g., in-block roads or unforested wet-lands, were recorded. We recorded only seedlings taller than10 cm because we felt that the error rate in finding all of thesmaller seedlings was unacceptably high and that smallerseedlings in these *10-year-old stands would also have amuch lower probability of contributing to the future stand.Seedling age was estimated by whorl counts, and the healthof the seedling was scored on a scale of 1–5. Only seedlingsconsidered to have germinated after the time of harvest wereincluded in the model construction. After completing eachtransect, the average cover of grass, short shrubs (<1.3 m),and tall shrubs (>1.3 m, including tree saplings) along theentire length of the transect was rated on a three-pointscale — low (0%–30% cover), medium (30%–60% cover),and high (60%–100% cover).

Additional data collectionIn addition to mapping the location of mature white

spruce while in the field, aerial photos were used to locateand position of residual spruce trees within and surroundingthe sampled cutblocks. The aerial photos were black andwhite, at scales of 1 : 15 000 to 1 : 30 000, and were takenwithin a year or two prior to harvest.

Data describing the forest stands surrounding the sampledcutblocks were obtained from digitized Alberta Vegetation In-ventory maps. These maps are derived from aerial photos at ascale of 1 : 20 000 and contain information on site moisture re-gime, crown closure, stand height, structure, decade of origin,species composition, and percent cover of the forest canopy.

A second measure of competition from aspen regenerationwas obtained from the deciduous regeneration surveys ofthese cutblocks; surveys of density, stocking rate, and heightof the deciduous regeneration had been done on these blocksto ensure that harvested areas met regeneration standards.

To explore the effects of climatic conditions 1 year beforeand for 2 years after harvest, multiple climate variables wereobtained from interpolations developed by the CanadianForest Service (McKenney et al. 2006). Climate variablesincluded monthly temperature and precipitation, a variety ofannual temperature and precipitation indices, and other bio-climatic variables derived from the original climate data,such as the number of growing degree-days per year (seeAppendix A, Table A1).

Digital elevation model data were obtained from NaturalResources Canada (2005), and these were used to calculateslope and aspect variables for each transect.

Fig. 1. Example of a study cutblock with two transects showing the locations of white spruce seedlings and potential seed sources. Theshaded buffers shown illustrate search radii used to calculate seed source density within the given distance from the transect centre-line.

Martin-DeMoor et al. 587

Published by NRC Research Press

Page 4: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

Calculation of additional variablesFor subsequent analyses, the transects were divided into

contiguous 9 m2 plots along their length. Transects were6 m wide; therefore, one 9 m2 plot was located on eachside of the transect centre-line every 3 m. If there were oneor more white spruce seedlings within a 9 m2 plot, the plotwas considered to be stocked. Transect-level stocking wascalculated as the percentage of the total number of 9 m2

plots on a given transect that were found to be stocked.Seed source strength was estimated by calculating the

density of mature white spruce within different distances ofthe transect centre-line (‘‘buffers’’ of 5, 10, 15, 30, 60, and100 m) (Fig. 1). Additionally, a second set of inner bufferrings was determined, and seed trees of the inner ring weresubtracted from the seed trees of the larger buffers. This wasdone in recognition of the fact that seed is not well depos-ited directly below the crown of seed trees (Greene andJohnson 1989; Stewart et al. 1998). Densities of seed treeswithin these rings were calculated for all combinations ofbuffer distances. We also calculated the average and maxi-mum distance-weighted total seed source strength for eachtransect. The distance-weighted seed source strength wascalculated by totalling the inverse log-weighted distancesfrom each 9 m2 plot to every seed source within 500 m ofeach plot centre. High values occurred when a few seedtrees were near the transect or when many seed trees werepresent at a greater distance.

To determine the effect of the surrounding forest on thelevel of white spruce regeneration, we calculated the dis-tance and direction (in both degrees and cardinal direction)from the centre of each 9 m2 plot to the nearest forest edge.The percentage of white spruce composition of the nearestforest edge (from the Alberta Vegetation Inventory data)was determined for each 9 m2 plot, and the average valueof this was calculated for each transect. To explore the rela-tionship between landscape-scale prevalence of white spruceand regeneration levels, the percentage of white sprucewithin 1 km and within 5 km of each transect was calculatedfrom the Alberta Vegetation Inventory stand compositiondata.

The monthly temperature and precipitation data were usedto calculate a climate moisture index (CMI) for each tran-sect for 1 year prior to harvest and 2 years postharvest. TheCMI used was as described by Hogg (1997). The CMI pro-vides an estimate of the annual accumulated moisture (centi-metres of water) and was calculated for 12 month periodsending 31 July, i.e., when the radial growth of aspen hascompleted for the year; this ‘‘growing year’’ is suitable forforestry applications (Ted Hogg, personal communication,2008). The CMI was also calculated for 12 month periodsending 31 December and for 9 month periods ending 30September. The number of growing degree-days (with abase temperature of 5 8C) (see McKenney et al. 2006) dur-ing the 1 year prior to harvest and the 2 years postharvestwas also used.

Statistical analysesTransect-level stocking rate (i.e., percentage of stocked

9 m2 plots on each transect) was used as the dependent var-iable in our construction of models of white spruce regener-ation. Because these data were highly skewed to the right,

with many zero values (see Fig. 2) a log base 10 transforma-tion was applied to make the data satisfy the assumption ofnormality. We also constructed models of white spruce seed-ling density (i.e., the number of seedlings per 100 m oftransect length), but the important independent variableswere similar to stocking, so we do not present these.

Models were constructed by means of Generalized LinearModels, using PROC GLM and PROC MIXED in SASv. 9.1.3 (SAS Institute Inc. 2003). Appendix A (Table A1)provides a list of all explanatory variables considered for in-clusion in the final model. Dummy variables were coded forall independent variables that were categorical. Only statisti-cally significant (P < 0.05) variables that increased the pro-portion of variation explained by the model (R2 value) wereincluded. In the case of groups of correlated variables, suchas the variables of different search radii for seed trees, sepa-rate analyses were conducted to determine which variableamong that subset would be tried for inclusion into the finalmodel. A combination of stepwise forward selection andmanual forcing of variables of interest was used. Stepwiseforward selection was used to identify an initial set of varia-bles that minimized the Akaike Information Criterion statis-tic to which additional variables of particular interest werethen added and assessed for their influence on the overallmodel fit.

Fig. 2. (A) Frequency distribution of transect-level stocking for the162 transects showing the positively skewed distribution due to lit-tle or no stocking of white spruce in most of the transects (note thebreak in y axis); and (B) distribution of measured white spruceseedling heights. Note that seedlings shorter than 10 cm were notincluded in the data set. All seedlings taller than 2 m were aged bywhorl count to determine that they had established postharvest.

588 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

Page 5: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

ResultsAlong the 162 transects, which covered 53.44 km of

transect length for a total of *32 ha of sampled area and31 670 9 m2 plots, we recorded a total of 5078 white sprucesaplings or seedlings of which 77% had regenerated sinceharvest. Only plots containing a postharvested regeneratedtree were considered stocked for purposes of the model con-struction. Seedlings were generally in good condition, with95.7% rated the top health code, and most were between0.3 and 0.7 m tall (Fig. 2B).

The mean stocking rate per transect was 6.8%, with a me-dian of 2.4% and a maximum of 61.6%. The density of ad-vance growth (understory white spruce present prior toharvest) was typically low, but including those trees, raisedthe mean transect-level stocking rate to 8.3%.

The final transect-level model included six variables, in-cluding a mast year by season of harvest interaction termrepresenting the four treatments. The model explained68.9% of the variation in the data. The model coefficientsfor the predictor variables, i.e., masting, season of harvest,density of seed trees, CMI, and grass and shrub cover arepresented in Table 2.

The interaction term describing the four combinations ofthe mast year and season of harvest had a statistically signif-icant effect on transect stocking rates (P = 0.0001, Table 2).Controlling for the influence of other factors in the model(using average seed tree density, average CMI, and lowcover of grass and shrubs), transects located in cutblocksthat were harvested during summer before seedfall in amast year had stocking rates of 6.72%, as compared withthe other three treatments for which stocking rate variedfrom 3.27 to 3.94 (Table 3).

The density of mature white spruce seed trees surroundinga transect had a significant positive effect on transect stock-

ing rates (P < 0.0001, Table 2). This effect was highly sig-nificant in both mast and non-mast years, and thecoefficients indicate that if cutting is done in a non-mastyear, more than three times as many seed trees are requiredto produce the same effect as harvesting during mast years(b = 0.054 in mast years, b = 0.198 in non-mast years). Ofthe buffer distances tested, the one that best explained thevariation in observed stocking was the number of seedsources within 60 m (double the approximate average heightof a full-grown white spruce) of the transect centre-line, ex-cluding the 5 m closest to the transect centre-line. We usedthis combination of buffer rings because in an independentanalysis, predictions of stocking were better if the inner5 m ring were excluded from the 60 m ring. The density ofseed trees within 100 m was second best at explaining thevariation in stocking, while the density within 200 m was adistant third place in terms of explanatory power, reaffirm-ing that trees far away do not contribute much seed.

Stocking rates were also affected by average CMI aver-aged over the year before and 2 years after cutting. Stockingdeclined when harvesting was done in wet (and cool) peri-ods (Table 2, P < 0.0001) (Fig. 3). Thirteen of the 15 tran-sects with a stocking rate greater than 20% had a CMI ofbetween 5 and 9 cm in the first growing year postharvest.

The final two variables in the model quantified abundanceof regenerating understory vegetation. The presence of highamounts of either grass or tall (>1.3 m) shrubs had a statisti-cally significant negative effect on transect stocking rate(P = 0.0008 and P = 0.0033, respectively, Table 2). Interest-ingly, our fairly coarse categorical estimates of tall shrubdensity were not correlated with the more detailed quantita-tive data collected during aspen regeneration surveys con-ducted 5 years postharvest, and the density of aspen

Table 2. Independent variables included in the final model of the (log transformed) percent stocking at the transect-level.Overall model R2 = 0.6886.

Degrees of freedom

Effect Numerator Denominator F value CoefficientStandarderror P

Mast year by season of harvest treatments 3 153 7.70 0.0001Mast year and summer harvest 1.3182 0.0996 <0.0001Mast year and winter harvest 1.0602 0.0812 <0.0001Non-mast year and summer harvest 0.8616 0.1266 <0.0001Non-mast year and winter harvest 0.8792 0.1385 <0.0001

Seed source density (5–60 m) (SSD) 2 153 30.13 <0.0001SSD in a mast year 0.0541 0.0121 <0.0001SSD in a non-mast year 0.1981 0.0305 <0.0001

Climate moisture index (Aug.–July) 1 153 42.07 –0.0458 0.0071 <0.0001Grass (dense) 1 153 11.67 –0.2099 0.0615 0.0008Tall shrub (dense) 1 153 8.89 –0.1794 0.0602 0.0033

Note: Model is as follows: Log(Stocking+1) = bMast*Season + bMast*SSD(Seed source density) + bCMI(CMI) + bGrass(Grass) + bShrub(Tall shrub),where ‘‘Stocking’’ is the transect stocking rate as a percentage; bMast*Season is the coefficient for the mast year by season of harvest treatmentinteraction (i.e. Sbm, Wam, Sim, Wim; see Table 1); mast years are dummy coded as 1 and non-mast years are coded 0; summer harvestedblocks are dummy coded as 1 and winter harvest blocks are coded 0; bMast*SSD is the coefficient for the mast year by seed source densityinteraction in a mast year or non-mast year; ‘‘Seed source density’’ is the number of seed trees per hectare between 5 m and 60 m of thetransect centre-line; bCMI is the coefficient for CMI; CMI is the climate moisture index (centimetres of accumulated moisture) (Hogg 1997)averaged for the year prior to and the 2 years after logging; bGrass is the coefficient for ‘‘Grass’’; ‘‘Grass’’ is equal to 1 for blocks with highgrass density and 0 for medium and low grass density; bShrub is the coefficient for ‘‘Tall shrub’’; and ‘‘Tall shrub’’ is equal to 1 for blockswith high tall shrub density and 0 for medium and low density of tall shrubs. Given are the degrees of freedom, F value, coefficient, stan-dard error, and significance (P) for the independent variables retained in the final model.

Martin-DeMoor et al. 589

Published by NRC Research Press

Page 6: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

regeneration determined by those surveys was not signifi-cantly related to white spruce stocking rates.

DiscussionThis study has shown that the highest levels of natural

spruce regeneration occurred when a site was logged in thesummer before seed cast in a mast year (Table 3). The im-portance of seed source strength on stocking is highlightedby the positive influences of mast year and the density ofseed trees within 5–60 m of the transect. Stocking levelswere higher if climate conditions in the time period immedi-ately before and after the harvesting were moderate (not too

cool or too wet) and if postharvest competition from grassand aspen regeneration was not overly severe. The resultsof this study identify many of the same key factors influenc-ing white spruce regeneration as identified by previous stud-ies conducted in postfire and postlogged stands (e.g., Zasada1986; Purdy et al. 2002; Peters et al. 2005); our study hasprovided insight into the influence of combinations of thesemultiple interacting factors on natural regeneration of whitespruce following logging in aspen-dominated mixedwoods.

The strong influence of masting and season of harvestsupports previous work showing a significant and long-lasting influence of masting on density of white spruce re-generation after fire and the important role of seedbed avail-ability in establishment (Peters et al. 2005). Whenharvesting occurred in years between mast years, whitespruce regeneration was uniformly low regardless of seasonof harvest. However, in mast years, the sites that were cutthe summer before seed cast had much higher levels ofstocking than sites that were cut the winter after masting(Table 3). White spruce seed mostly falls by mid-October(Rowe 1955; Youngblood and Max 1992); therefore, insummer-harvested cutblocks, the seed from trees on the pe-rimeter of the block or from scattered trees within the blockwill fall on substrates disturbed by the summer logging. Incontrast, winter-harvested blocks may have had a greatersupply of seed already on the ground at the time of loggingbecause of the fact that seed sources within the block wouldhave shed their seed in the months before they were cut.The poorer regeneration in these winter-harvested blocks islikely due to the fact that frozen soils and snow during the

Table 3. Predicted white spruce stocking (percentage of 9 m2 plots that have at least one seedling)under different sets of conditions of the effects included in the model, including mast year (yes (Y)or no (N)), season of harvest (summer (S) or winter (W)), seed source strength (SS/ha: seed treesper hectare within 5–60 m of the transect centre-line), climate moisture index (CMI, centimetres ofaccumulated moisture), grass and shrub abundance (low (L) or high (H)).

Scenario Mast Season SS/haa CMIa Grass ShrubStocking(%)

Best observed values Y S 12.4 3.6 L L 65.70Average valuesb

Sbm Y S 1.7 11.4 L L 6.72Wam Y W 1.7 11.4 L L 3.27Sim N S 1.7 11.4 L L 3.75Wim N W 1.7 11.4 L L 3.94

SS/haHigh (dense) Y S 8.0 3.6 L L 37.55Moderate Y S 6.0 3.6 L L 29.05Low Y S 3.0 3.6 L L 19.68Very low (sparse) Y S 1.0 3.6 L L 15.12

CMILow (dry) Y S 12.4 5.0 L L 56.54Moderate Y S 12.4 10.0 L L 32.94High Y S 12.4 15.0 L L 19.02Very high (wet) Y S 12.4 20.0 L L 10.81

Grass — high Y S 12.4 3.6 H L 40.14Shrub — high Y S 12.4 3.6 L H 43.13

aValues chosen are based on the range of values observed in this study: for seed source density 12.4/ha was thebest case scenario, 1.7/ha was the average; CMI: 3.6 was the best case scenario, 11.4 was the average.

bSbm, summer harvest in a mast year; Wam, winter harvest in a mast year; Sim, summer harvest in a non-mastyear; Wim, winter harvest in a non-mast year.

Fig. 3. Effect of average climate moisture index (averaged over1 year prior to harvest and 2 years postharvest) on transect-levelstocking (percentage of 9 m2 plots along a transect that had at leastone white spruce seedling).

590 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

Page 7: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

logging prevented the disturbance and mixing of the forestfloor necessary for creation of suitable seedbeds. Nienstaedtand Zasada (1990) demonstrated the importance of seedbedquality where 5–30 seeds are needed to generate a spruceseedling on an exposed mineral soil microsite versus 1000on undisturbed organic microsites within harvested areas.Additionally, the blocks sampled in this study were gener-ally small (average 15.5 ha) and had irregular geometriessuch that little area was far outside the dispersal range ofwhite spruce seed from the surrounding forest, if seed treeswere present in the vicinity. Indeed, 83% of the sampledtransect length was within 100 m of the forest edge and58% was within 100 m of the nearest mature seed source.

Stocking success was positively related to density of theseed source within 5–60 m of the transect (Table 2). How-ever, this effect varied between mast and non-mast years,and surprisingly, the influence of seed source density onstocking was greater in non-mast years (Table 2). This effectwas likely related to the fact that the number of seed treeswas not uniform across the various masting conditions. The5–60 m search area was selected as the distance that ex-plained the most variance in stocking after testing a varietyof search distances ranging up to 200 m. This result is con-sistent with Rowe (1955), Dobbs (1976), and Nienstaedt andZasada (1990) in that the dispersal of white spruce seed de-clines sharply with distance and is severely limited past100 m. The significance and explanatory power of the den-sity of seed sources within a given search radius increasedwhen the density of trees within 5 m of the transect centre-line were excluded. This is consistent with the work ofStewart et al. (1998) and Greene and Johnson (1989) whoreported very little seedfall under or very near to a whitespruce seed tree. Greene and Johnson (1996) attributed thisto the fact that a minimum wind speed is required to shakethe ripe seeds from the cone, and this wind will then carrythe released seed at least a few metres from the tree.

Another key factor influencing stocking rates was themeasure of annual moisture accumulation — the CMI in theyears immediately before and after cutting. Our resultsshowed that when CMI values were high (i.e., wet, cool cli-matic conditions), stocking levels were reduced. This rela-tionship was likely due to early seedling mortality causedby flooding or damping-off fungi (Hamm et al. 1990). Re-cent work (Curran, Leiffers, and Macdonald, unpublisheddata) has shown that even short periods of standing waterafter snowmelt on these flat sites can cause significant mor-tality in newly germinated spruce seedlings. Flooding aftersnowmelt seems to occur when the soil is not yet suffi-ciently thawed to allow drainage. High rainfall in the springmay also wash seeds off of preferred seedbeds, such as ex-posed mineral soil or elevated logs, into less suitable micro-sites (Timoney and Peterson 1996).

While the negative coefficient for the (linear) CMI termin the model indicates that transect stocking rates decreasedwhen CMI increased, caution must be used in interpretingthis trend, especially in terms of extrapolation to conditionsdrier than *4.5 cm CMI (Fig. 3). Hogg and Schwarz (1997)showed that white spruce regeneration levels declined sig-nificantly when CMI values were less than –5 cm and thatalmost no regeneration was observed at a CMI drier than

–15 cm. None of the transects sampled for this study had3 year average CMI values <3.5 cm.

Measures of grass and tall shrub abundances were the fi-nal two variables included in the final model. While bothvariables had a significant effect on stocking rate, their con-tributions to the model were modest compared with theother variables. The grass growing in the sampled cutblockswas mostly Calamagrostis canadensis and the tall shrubswere mostly aspen saplings, although some willows or othershrubs were included in this category. Heavy recruitment ofgrass and aspen after harvest can create low-light environ-ments (Lieffers et al. 1993) that likely reduce the growthrates of newly established white spruce seedlings, potentiallyreducing survival. Further, high levels of grass and decidu-ous leaf litter may crush small seedlings when they collapseunder a weight of snow (Lieffers et al. 1993; Wang andKemball 2005). Given that >95% of seedlings found alongthe sampled transects were healthy and growing well, how-ever, it seems that if seedlings can survive through the firstyear or two when mortality rates are very high, they will beable to thrive even in the low-light environment of well-stocked deciduous stands (Feng et al. 2006).

We completed a separate analysis using logistic regres-sion to construct models predicting stocking of individualplots. The same variables were tested, along with severalfine-grained spatial variables related to the distance and di-rection from each plot to potential seed sources. The finalplot-level model corroborated the findings presented herefor transect-level stocking, with seed source strength (mastyears, distance and density of seed sources) being very im-portant, along with CMI and growing degree-days in theyears surrounding harvest (refer to Martin-DeMoor (2009)for further details on this analysis). This analysis of individ-ual plots allowed us to test for the importance of direction toindividual seed trees; this variable was not selected for in-clusion in the final plot-level model, because presumablythere were usually seed trees in a variety of directions fromthe transects.

Despite the lack of control and the uncertainty of sam-pling in a retrospective studies such as ours, our resultsclearly show that white spruce can and will regenerate natu-rally in harvested deciduous-dominated mixedwood sitesthat have received no treatments to encourage spruce estab-lishment. The best-case scenario (summer logging before amast year, with 12 or more seed trees per hectare, in warmyears with moderate rainfall and on sites with low levels ofgrass or tall shrubs) resulted in a predicted 65% stocking forwhite spruce (Table 3). Stocking was negligible on siteswith no seed trees. This model could be used by forest man-agers to predict stocking of natural white spruce in siteswhere full stocking of conifer is not to be expected by natu-ral means. Further, our results suggest that natural regenera-tion can be used to defray the high silvicultural costs ofplanting spruce seedlings while achieving some level of par-tial stocking on aspen-dominated mixedwood forest stands.

AcknowledgementsWe are grateful to the Sustainable Forest Management Net-

work and the Mixedwood Management Association for fund-ing and to Lee Martens and Weyerhaeuser Company for helpin selecting the study sites and providing preharvest data on

Martin-DeMoor et al. 591

Published by NRC Research Press

Page 8: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

them. Jonathan Martin-DeMoor acknowledges with thanks anIndustrial Postgraduate Scholarship from the Natural Scien-ces and Engineering Research Council (Canada), a bursaryfrom the Canadian Forest Service, and additional scholarshipsand bursaries from the University of Alberta. Derek Bakker,Melanie Mattila, Ian Curran, Penny Wizniuk and Gina Sageassisted with the field work.

ReferencesArlidge, J.W.C. 1967. The durability of scarified seedbeds for

spruce regeneration. Research Note No. 42. British ColumbiaForest Service.

Beckingham, J.D., Corns, I.G.W., and Archibald, J.H. (Editors).1996. Field guide to ecosites of west-central Alberta. CanadianForest Service, Northwest Region, Northern Forestry Centre, Ed-monton, Alta.

Berger, A.L., Puettmann, K.J., and Host, G.E. 2004. Harvesting im-pacts on soil and understory vegetation: the influence of seasonof harvest and within-site disturbance patterns on clear-cut aspenstands in Minnesota. Can. J. For. Res. 34(10): 2159–2168.doi:10.1139/x04-097.

Buddle, C.M., Langor, D.W., Pohl, G.R., and Spence, J.R. 2006.Arthropod responses to harvesting and wildfire: implications foremulation of natural disturbance in forest management. Biol.Conserv. 128(3): 346–357. doi:10.1016/j.biocon.2005.10.002.

Day, R.J. 1963. Spruce seedling mortality caused by adverse summermicroclimate in the rocky mountains. Forest Research BranchPublication 1003, Canadian Department of Forestry, Ottawa, Ont.

Dobbs, R.C. 1976. White spruce seed dispersal in central BritishColumbia. For. Chron. 52: 225–228.

Eis, S. 1981. Effect of vegetative competition on regeneration ofwhite spruce. Can. J. For. Res. 11(1): 1–8. doi:10.1139/x81-001.

Feng, Z.L., Stadt, K.J., and Lieffers, V.J. 2006. Linking juvenilewhite spruce density, dispersion, stocking, and mortality to fu-ture yield. Can. J. For. Res. 36(12): 3173–3182. doi:10.1139/X06-192.

Greene, D.F., and Johnson, E.A. 1989. A model of wind dispersalof winged or plumed seeds. Ecology, 70(2): 339–347. doi:10.2307/1937538.

Greene, D.F., and Johnson, E.A. 1996. Wind dispersal of seedsfrom a forest into a clearing. Ecology, 77(2): 595–609. doi:10.2307/2265633.

Greene, D.F., Zasada, J.C., Sirois, L., Kneeshaw, D., Morin, H.,Charron, I., and Simard, M.-J. 1999. A review of the regenera-tion dynamics of North American boreal forest tree species.Can. J. For. Res. 29(6): 824–839. doi:10.1139/cjfr-29-6-824.

Greene, D.F., Macdonald, S.E., Cumming, S., and Swift, L. 2005.Seedbed variation from the interior through the edge of a largewildfire in Alberta. Can. J. For. Res. 35(7): 1640–1647. doi:10.1139/x05-080.

Greene, D.F., Macdonald, S.E., 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 oforganic-layer depth by wildfire in the North American borealforest and its effect on tree recruitment by seed. Can. J. For.Res. 37(6): 1012–1023. doi:10.1139/X06-245.

Hamm, P.B., Campbell, S.J., and Hansen, E.M. 1990. Growinghealthy seedlings: identification and management of pests innorthwest forest nurseries. Special Publication 19. Forest Re-search Laboratory, Oregon State University, Corvallis, Ore.

Hart, S.A., and Chen, H.Y.H. 2008. Fire, logging, and overstory af-fect understory abundance, diversity, and composition in borealforests. Ecol. Monogr. 78(1): 123–140. doi:10.1890/06-2140.1.

Hiratsuka, Y. 1987. Forest tree diseases of the Prairie Provinces.Northern Forestry Centre Informaiton Report NOR-X-286. Ca-nadian Forestry Service, Edmonton, Alta.

Hobson, K.A., and Bayne, E. 2000. Breeding bird communities inboreal forest of western Canada: consequences of ‘‘unmixing’’the mixedwoods. Condor, 102(4): 759–769. doi:10.1650/0010-5422(2000)102[0759:BBCIBF]2.0.CO;2.

Hogg, E.H. 1997. Temporal scaling of moisture and the forest–grassland boundary in western Canada. Agric. For. Meteorol.84(1–2): 115–122. doi:10.1016/S0168-1923(96)02380-5.

Hogg, E.H., and Schwarz, A.G. 1997. Regeneration of planted con-ifers across climatic moisture gradients on the Canadian prairies:implications for distribution and climate change. J. Biogeogr.24: 527–534. doi:10.1111/j.1365-2699.1997.00138.x.

Kabzems, R.D., and Lousier, J.D. 1992. Regeneration, growth anddevelopment of Picea glauca under Populus spp. canopy in theboreal white and black spruce zone. Forest Resource Develop-ment Agreement Report 176. British Columbia Forest Service.

Kelly, D. 1994. The evolutionary ecology of mast seeding. TrendsEcol. Evol. 9(12): 465–470. doi:10.1016/0169-5347(94)90310-7.

Kemball, K.J., Wang, G.G., and Dang, Q.L. 2005. Response of un-derstory plant community of boreal mixedwood stands to fire,logging, and spruce budworm outbreak. Can. J. Bot. 83(12):1550–1560. doi:10.1139/b05-134.

Kemball, K.J., Wang, G.G., and Westwood, A.R. 2006. Aremineral soils exposed by severe fire better seedbeds for coniferregeneration? Can. J. For. Res. 36(8): 1943–1950. doi:10.1139/X06-073.

Lamontagne, J.M., and Boutin, S. 2007. Local-scale synchrony andvariability in mast seed production patterns of Picea glauca. J.Ecol. 95(5): 991–1000. doi:10.1111/j.1365-2745.2007.01266.x.

Lieffers, V.J., and Stadt, K.J. 1994. Growth of understory Piceaglauca, Calamagrostis canadensis, and Epilobium angustifoliumin relation to overstory light transmission. Can. J. For. Res.24(6): 1193–1198. doi:10.1139/x94-157.

Lieffers, V.I., Macdonald, S.E., and Hogg, E.H. 1993. Ecology ofand control strategies for Calamagrostis canadensis in borealforest sites. Can. J. For. Res. 23(10): 2070–2077. doi:10.1139/x93-258.

Lieffers, V.J., Armstrong, G.W., Stadt, K.J., and Marenholtz, E.H.2008. Forest regeneration standards: Are they limiting manage-ment options for Alberta’s boreal mixedwoods? For. Chron. 84:76–82.

Macdonald, S.E., and Fenniak, T.E. 2007. Understory plant com-munities of boreal mixedwood forests in western Canada: nat-ural patterns and response to variable-retention harvesting. For.Ecol. Manage. 242(1): 34–48. doi:10.1016/j.foreco.2007.01.029.

Man, C.D., Comeau, P.G., and Pitt, D.G. 2008. Competitive effectsof woody and herbaceous vegetation in a young boreal mixed-wood stand. Can. J. For. Res. 38(7): 1817–1828. doi:10.1139/X08-032.

Martin-DeMoor, J. 2009. Natural white spruce regeneration inaspen-dominated boreal mixedwoods following harvest. M.Sc.thesis, Department of Renewable Resources, University of Al-berta, Edmonton, Alta.

McKenney, D.W., Pedlar, J.H., Papadopol, P., and Hutchinson, M.F.2006. The development of 1901–2000 historical monthly climatemodels for Canada and the United States. Agric. For. Meteorol.138(1–4): 69–81. doi:10.1016/j.agrformet.2006.03.012.

Natural Regions Committee. 2006. Natural regions and subregionsof Alberta. Compiled by D.J. Downing and W.W. Pettapiece.Publ. No. T/852. Government of Alberta.

Natural Resources Canada. 2005. Available from www.geobase.ca[accessed 5 January 2006].

592 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press

Page 9: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

Nienstaedt, H., and Zasada, J.C. 1990. White spruce. In Silvics ofNorth America. Vol. 1. Conifers. Edited by R.M. Burns andB.H. Honkala. U.S. Dep. Agric. Handb. 654. pp. 389–442.

Peters, V.S., Macdonald, S.E., and Dale, M.R.T. 2005. The interac-tion between masting and fire is key to white spruce regenera-tion. Ecology, 86(7): 1744–1750. doi:10.1890/03-0656.

Purdy, B.G., Macdonald, S.E., and Dale, M.R.T. 2002. The regen-eration niche of white spruce following fire in the mixedwoodboreal forest. Silva Fenn. 36: 289–306.

Rowe, J.S. 1955. Factors influencing white spruce reproduction inManitoba and Saskatchewan. Minister of Northern Affairs andNational Resources, Ottawa, Ont.

SAS Institute Inc. 2003. SAS version 9.1. SAS Institute Inc., Cary,N.C.

Stewart, J.D., Hogg, E.H., Hurdle, P.A., Stadt, K.J., Tollestrup, P.,and Lieffers, V.J. 1998. Dispersal of white spruce seed in ma-ture aspen stands. Can. J. Bot. 76(2): 181–188. doi:10.1139/cjb-76-2-181.

Stewart, J.D., Landhausser, S.M., Stadt, K.J., and Lieffers, V.J.2000. Regeneration of white spruce under aspen canopies: seed-ing, planting, and site preparation. West. J. Appl. For. 15: 177–182.

Stewart, J.D., Jones, T., Snedden, J., and Martin-DeMoor, J. 2005.White spruce regeneration in mixedwood forests of the EMENDproject (MDFP 15/99): final report. Natural Resources Canada,Canadian Forest Service, Northern Forestry Centre.

Timoney, K.P., and Peterson, G. 1996. Failure of natural regenera-tion after clearcut logging in Wood Buffalo National Park, Ca-nada. For. Ecol. Manage. 87(1–3): 89–105. doi:10.1016/S0378-1127(96)03831-5.

Waldron, R.M. 1965. Cone production and seedfall in a maturewhite spruce stand. For. Chron. 41: 316–329.

Wang, G.G., and Kemball, K.J. 2005. Balsam fir and white spruceseedling recruitment in response to understory release, seedbed

type, and litter exclusion in trembling aspen stands. Can. J. For.Res. 35(3): 667–673. doi:10.1139/x04-212.

Work, T.T., Shorthouse, D.P., Spence, J.R., Volney, W.J.A.,Morgantini, L.E., and Innes, J.L. 2004. Stand composition andstructure of the boreal mixedwood and epigaeic arthropods ofthe Ecosystem Management Emulating Disturbance (EMEND)landscape in northwestern Alberta. Can. J. For. Res. 61: 1498–1514.

Wurtz, T.L., and Zasada, J.C. 2001. An alternative to clear-cuttingin the boreal forest of Alaska: a 27-year study of regenerationafter shelterwood harvesting. Can. J. For. Res. 31(6): 999–1011.doi:10.1139/cjfr-31-6-999.

Youngblood, A., and Max, T.A. 1992. Dispersal of white spruceseed on willow island in interior Alaska. USDA For. Serv. Res.Pap. PNW-RP-443.

Zasada, J.C. 1986. Natural regeneration of trees and tall shrubs onforest sites in interior Alaska. In Forest ecosystems in the Alas-kan taiga: a synthesis of structure and function. Edited by K.Van Cleve, F.S. Chapin, III, P.W. Flanagan, L.A. Viereck, andC.T. Dyrness. Springer-Verlag, New York.

Zasada, J.C., and Lovig, D. 1983. Observations on primary disper-sal of white spruce, Picea glauca, seed. Can. Field Nat. 97:104–106.

Zasada, J.C., Sharik, T.L., and Nygren, M. 1992. The reproductiveprocess in boreal forest trees. In A systems analysis of the glo-bal boreal forest. Edited by H.H. Shugart, R. Leemans, and G.B.Bonan. Cambridge University Press, Cambridge, UK.

Appendix AAppendix Table A1 appears on the following page.

Martin-DeMoor et al. 593

Published by NRC Research Press

Page 10: Natural regeneration of white spruce in aspen-dominated boreal mixedwoods following harvesting

Table A1. Explanatory variables considered for inclusion inthe model of white spruce stocking. The response variable waslog % stocking at the transect level.

No. Variable description

Treatment variables1 Study region2 Masting year/season of harvest interaction3 Season of harvest4 Annual seed crop strength5 Treatment replicate6 Cutblock harvest date7 Age of block at time of survey8 East–west coordinate of transect centrepoint9 North–south coordinate of transect centrepoint

10 Transect bearing11 Average transect elevation12 Transect slope13 Transect aspect14 Transect aspect (cardinal directions)15 Competition rating — grass16 Competition rating — short shrubs (<1.3 m)17 Competition rating — tall shrubs and deciduous tree

saplings (>1.3 m)18 Binary competition rating — grass19 Binary competition rating — short shrubs (<1.3 m)20 Binary competition rating — tall shrubs (>1.3 m)21 Deciduous stocking rate — year 5 regen survey22 Deciduous stocking code — year 5 regen survey23 Deciduous density — year 5 regen survey24 Average deciduous height — year 5 regen survey25 Conifer stocking rate — year 5 regen survey26 Conifer density — year 5 regen survey27 Average conifer height — year 5 regen survey28 Density of residual seed trees within 5 m of transect

centre-linea

29 Density of residual seed trees between 5 and 60 mof transect centre-lineb

30 Average cumulative distance-weighted seed sourcestrength of all 9 m2 plotsc

31 Maximum cumulative distance-weighted seed sourcestrength of all 9 m2 plotsc

32 Average of spruce component of the forest edgenearest to each point along transect

33 Maximum of spruce component of the forest edgenearest to each point along transect

34 Average age of stands within 1 km of transect35 Average spruce component of stands within 1 km of

transect36 Average understory spruce component of stands

within 1 km of transect37 Average age of stands within 5 km of transect38 Average spruce component of stands within 5 km of

transect39 Average understory spruce component of stands

within 5 km of transect

Climatic variables40 Temperature (monthly min.)41 Temperature (monthly max.)42 Monthly precipitation43 Monthly evapotranspiration

Table A1 (concluded ).

No. Variable description

44 Monthly climate moisture index45 Potential evapotranspirationd,e

46 Climate moisture indexd,e

47 Mean diurnal ranged

48 Isothermalityd

49 Temperature seasonalityd

50 Max temperature of warmest periodd

51 Min temperature of coldest periodd

52 Temperature annual ranged

53 Mean temperature of wettest quarterd

54 Mean temperature of driest quarterd

55 Mean temperature of warmest quarterd

56 Mean temperature of coldest quarterd

57 Annual mean temperatured

58 Annual minimum temperatured

59 Annual maximum temperatured

60 Mean temperature for growing seasond

61 Temperature range for growing seasond

62 Annual precipitationd

63 Precipitation of wettest periodd

64 Precipitation of driest periodd

65 Precipitation seasonalityd

66 Precipitation of wettest quarterd

67 Precipitation of driest quarterd

68 Precipitation of warmest quarterd

69 Precipitation of coldest quarterd

70 Julian day number of start of growing seasond

71 Julian day number at end of growing seasond

72 Number of days of growing seasond

73 Total precipitation in 3 months prior to growing sea-sond

74 Total precipitation for growing seasond

75 Growing degree-days for growing seasond

Note: Variables included in the final model are in boldface.aDensity of residual trees was calculated within 5, 10, 15, 30, 60, 100,

and 200 m of the transect centre-line.bDensity of residual trees was calculated for all possible single

combinations of the above buffer widths.cDistances were weighted linearly, inverse squared, and inverse log.dCalculated for the year (Aug.–July) preharvest (GS0), and the first

and second years (Aug.–July) postharvest (GS1 & GS2).eCalculated for years starting 1 August (Growing Year), 1 January

(Calendar Year), and for the period between 1 January and30 September.

594 Can. J. For. Res. Vol. 40, 2010

Published by NRC Research Press