Concentrating anthropogenic disturbance to ... - Wiley Online Library

Ecological Applications, 22(4), 2012, pp. 1268–1277 Ó 2012 by the Ecological Society of America

Concentrating anthropogenic disturbance to balance ecological and economic values: applications to forest management REBECCA TITTLER,1,3 CHRISTIAN MESSIER,1 1

AND

ANDREW FALL2

Centre d’e´tude de la forêt, Universite´ du Que´bec a` Montreál, C.P. 8888, Succ. Centre-Ville, Montreál, Que´bec H3C 3P8 Canada 2 School of Resource and Environmental Management, Simon Fraser University, Burnaby, British Columbia V5A 1S6 Canada

Abstract. To maintain healthy ecosystems, natural-disturbance-based management aims to minimize differences between unmanaged and managed landscapes. Two related approaches may help accomplish this goal, either applied together or in isolation: (1) concentrating anthropogenic disturbance through zoning (with protected areas and intensive management); and (2) emulating natural disturbances. The purpose of this paper is to examine the effects of these two approaches, applied both in isolation and in combination, on the structure of the forest landscape. To do so, we use a spatially explicit landscape simulation model on a large fire-dominated landscape in eastern Canada. Specifically, we examine the effects of (1) increasing the maximum size of logged stands (cutblocks) to better emulate the full range of fire sizes in a fire-dominated landscape, (2) increasing protected areas, and (3) adding aggregated or dispersed intensive wood production areas to the landscape in addition to protected areas (triad management). We focus on maximizing the amount and minimizing the fragmentation of old-growth forest and on reducing road construction. Increasing maximum cutblock size and adding protected areas led to reduced road construction, while the latter also resulted in less fragmentation and more old growth. Although protected areas led to reduced harvest volume, the addition of an intensive production zone (triad management) counterbalanced this loss and resulted in more old growth than equivalent scenarios with protected areas but no intensive production zone. However, we found no differences between aggregated and dispersed intensive wood production. Our results imply that differences between unmanaged and managed landscapes can be reduced by concentrating logging efforts through a combination of protected areas and intensive wood production, and by creating some larger cutblocks. We conclude that the forest industry and regulators should therefore seek to increase protected areas through triad management and consider increasing maximum cutblock size. These results add to a growing body of literature indicating that intensive management on a small part of the landscape may be better than less intensive management spread out over a much larger part of the landscape, whether this is in the context of forestry, agriculture, or urban development. Key words: conservation; cutblock size; forest management; old growth; road effects; SELES; triad zoning; Vermillion Landscape Model.

INTRODUCTION To maintain functioning managed ecosystems, there has been a move toward natural-disturbance-based management (Hunter 1993, Haila 1994). Assuming that native biodiversity is adapted to the unmanaged landscape, the aim of natural-disturbance-based management is to minimize the differences between managed and unmanaged landscapes and thus reduce the anthropogenic footprint (Hunter 1990, Attiwill 1994). In boreal forest, differences between managed and unmanaged landscapes include fragmentation and loss of old growth (e.g., DeLong and Tanner 1996) and Manuscript received 14 September 2011; revised 12 December 2011; accepted 20 December 2011; final version received 21 January 2012. Corresponding Editor: V. C. Radeloff. 3 E-mail: [email protected]

extensive road construction. These may negatively affect biodiversity. Just as loss of early successional forest is of concern in temperate forests of the northeastern United States (e.g., Hunter et al. 2001), loss and fragmentation of old growth is of concern in the boreal because it may negatively impact species that are common in this habitat (e.g., bryophytes and lichens [Boudreault et al. 2002]; some birds [Drapeau et al. 2003]), including many of conservation concern (e.g., Yezerinac and Moola 2006, Lohmus and Lohmus 2009). Although fragmentation is less important than habitat loss (Fahrig 2003), it may become more important when there is little habitat on the landscape (Fahrig 1998), as is the case for old growth in many managed boreal forest landscapes. Roads may also negatively impact biodiversity, possibly resulting in habitat loss, mortality and genetic isolation, increased dispersal of exotic plants, increased predator movement, and increased access to the forest (reviewed

1268

June 2012

CONCENTRATING ANTHROPOGENIC DISTURBANCE

in Forman and Alexander 1998). Although there may be other important differences between managed and unmanaged landscapes, to reduce deleterious effects, forest management should at least aim to increase old growth, limit fragmentation, and reduce road construction. One way to reduce fragmentation and road construction may be through cutblocks (logged stands) that emulate natural disturbances (Hunter 1993). Many stand-level management activities are designed to emulate natural disturbance (e.g., Fenton et al. 2009), with varying success (McRae et al. 2001), but this is not the subject of the current work. At the landscape scale, natural-disturbance-based management aims to reduce the differences in landscape structure between managed and unmanaged forests, even if there are differences at the stand scale. Specifically, in areas dominated by wildfire, these differences might be reduced if the size distribution of cutblocks better followed that of fires (Hunter 1993), making for fewer cutblocks overall and some larger than those currently allowed in most places. This method of concentrating logging has been mandated in the province of Ontario (OMNR 2001), but it is unclear what the landscape-level effects will be in forests where fire is still omnipresent (e.g., Payette 1992). Increasing protected areas may also reduce differences between unmanaged and managed landscapes by reducing fragmentation, loss of old growth, and road construction. Widely recognized as crucial to the conservation of biodiversity (e.g., CBD 1992), protected areas allow natural age-class structures to develop and are thus key to maintaining old growth. If they are relatively large and logging is thus concentrated in a smaller area of the forest, protected areas may also reduce overall fragmentation. Finally, protected areas may reduce road construction, since they require that industrial development be concentrated in the remaining landscape. We term the setting aside of large protected areas dyad zoning because it involves two zones: a biodiversity conservation zone (protected areas) and, in our case, an integrated forest management zone (the rest of the productive forest). Integrated forest management implies the management of the forest to satisfy various uses, be they economic, environmental, recreational, or spiritual, on the same piece of land. Protected areas may come with an economic and social cost in terms of lost jobs and revenue, but this cost may be reduced by further concentrating timber extraction through triad zoning. Under triad zoning, the forest is divided into (1) a biodiversity conservation zone (protected areas) where no logging occurs; (2) an intensive production zone, where silviculture is used to maximize timber yield; and (3) an integrated forest management zone, where the needs of the timber industry are balanced with the needs of other users (Seymour and Hunter 1999, Messier et al. 2010). The addition of the intensive production zone under triad management may allow for more conservation

1269

without sacrificing the viability of the forest industry. Coˆ te´ et al. (2010) found triad zoning to provide adequate levels of timber yield while allowing for more protected areas, more old growth, and less fragmentation. However, in Côte´ et al.’s study, 50% of the harvest volume in the integrated forest management zone was assumed to come from partial cuts, which were not modeled in a spatially explicit way. Unlike the situation for clearcuts, no roads were built to access partial cuts, and when a stand was partially cut, its overall age remained unchanged. It is unclear to what extent the observed benefits of triad zoning would hold without these assumptions. It is also unclear how the spatial distribution of zones in triad management may affect landscape configuration and road construction. Although large protected areas are generally thought to be better than small ones, no one has investigated the effect of large aggregated vs. small dispersed intensive production areas in triad management. In previous work (Coˆ te´ et al. 2010, Messier et al. 2010), wood production areas have been relatively small and spread out, located where production was likely to be high to maximize harvest volume. However, this may increase road construction. Thus, it may be better to have one large agglomerated area rather than many small, dispersed wood production areas, further concentrating logging. In general, a trend is emerging in favor of concentrating intensive rather than spreading out less intensive anthropogenic disturbances over a larger area. For example, for the same amount of housing, clustered housing is not as detrimental to native biodiversity as urban sprawl (Gagne´ and Fahrig 2010a, b). This also holds for road development. Since roads fragment habitat, keeping animals from accessing resources on the other side and potentially leading to population isolation (e.g., Keller and Largiade`r 2003), fewer large, high-traffic roads leave a more intact landscape than more small, dispersed roads (Jaeger et al. 2007), thus reducing road effects on native biodiversity. Furthermore, some animals are more likely to avoid high-traffic than lower-traffic roads (e.g., Eigenbrod et al. 2009), and are thus less likely to be killed when the same amount of traffic is concentrated on a few large roads than when it is spread among many smaller roads. Finally, a triad-type management scenario with protected areas, conventional (intensive) agriculture, and organic farming may be better for British butterfly populations than a more extensive organic farming scenario without protected areas and accompanying conventional agriculture (Hodgson et al. 2010). Similarly, a system where part of the landscape is protected while high-yield agriculture is carried out on the rest (land-sparing) may lead to higher crop yields and higher densities of birds and trees than a system where both production and conservation objectives are integrated over the landscape (Phalan et al. 2011). By concentrating anthropogenic disturbance, we may limit the spatial extent of the human footprint on the landscape.

1270

Ecological Applications Vol. 22, No. 4

REBECCA TITTLER ET AL.

METHODS Study area

FIG. 1. The province of Quebec (in gray, left) with the study area, the Vermillion landscape (right).

However, it is unclear to what extent anthropogenic disturbance can be concentrated. In urban development, it may be better to avoid urban sprawl, but it would not be economically viable for the entire population of a country to live in one city. Similarly, one road could not serve the entire population of any city. In forestry and agriculture, there may be a trade-off between resource availability (mature trees, fertile soil) and spatial agglomeration. We therefore investigate this issue with the present case study in natural-disturbance-based forest management. More specifically, we use a spatially explicit landscape simulation model to explore the effects of four measures of concentrating anthropogenic disturbance and minimizing the differences between managed and unmanaged forests: (1) logging some relatively large cutblocks, in line with the size distribution of fires, thus reducing the number of cutblocks and the spatial extent of logging operations; (2) setting aside protected areas, thus concentrating logging in the rest of the forest (dyad management); (3) dispersed triad management, in which intensive wood production areas are laid out across the landscape to counterbalance protected areas; and (4) agglomerated triad management, in which intensive wood production is concentrated in one area. We examined (1) in combination with (2), (3), and (4), but also in the absence of any zoning. Our hypotheses were that all four options would result in decreased road construction and fragmentation; dyad in more old growth than no zoning and triad in more than dyad; triad in greater timber harvest volume than dyad; and agglomerated triad in less road construction than dispersed triad. The combination of agglomerated triad with cutblock sizes more closely resembling fire sizes should minimize differences between managed and unmanaged landscapes.

This study is based on a 430 000-ha, largely forested management area on the Canadian Shield, fairly typical of the boreal mixedwood forests of south-central Quebec (Fig. 1). Over 390 000 ha of the area is forested; the rest is largely water and wetlands. The area straddles the yellow birch–balsam fir and white birch–balsam fir ecoregions of the boreal mixedwood forest and is dominated by white birch (Betula papyrifera), black spruce (Picea mariana), balsam fir (Abies balsamea), jack pine (Pinus banksiana), and trembling aspen (Populus tremuloides). The main natural disturbance is stand-replacing fire. Most fires are relatively small, but large fires (.10 000 ha) play an important role in structuring the landscape, generally accounting for almost 90% of the total area burnt. The fire cycle is ;250 years (Bergeron et al. 2001). The forest of the area has been heavily exploited since the mid 1800s, first for ship-building and more recently for pulp and paper. Although there is some partial cutting and selective logging, clearcutting has been the norm for the past century or so. The logging rotation is 100 years or less. There are some plantations, but most stands are left to regenerate naturally. (For more information, see Fall et al. 2004, Didion et al. 2007, James et al. 2007, and Côte´ et al. 2010.) In the North American boreal forest in which the study area is located, old growth has been variously defined. Most definitions include the onset of death in the initial cohort, accompanied by an increase in deadwood, the creation of canopy gaps, and the subsequent growth to canopy height of individual trees that were previously in the understory (reviewed in Kneeshaw and Gauthier 2003). In this part of the world, this occurs between 100 and 200 years (Kneeshaw and Gauthier 2003). Since preliminary analyses indicated no difference in results, whether old growth was defined as .100 or .200 years, we chose to define old growth as .100 years in this study. Model description We used the Vermillion landscape model (VLM) (Fall et al. 2004, Didion et al. 2007, James et al. 2007) built using the Spatially Explicit Landscape Event Simulator (SELES [Fall and Fall 2001]). The landscape layers used as input for this grid-based model are largely from the third decadal SIFORT data set, which was produced by photointerpretation, dates from 1997 for the study area, and has a polygon resolution of ;4 ha (Pelletier et al. 2007). The spatial resolution of the model was 0.25 ha per pixel and the temporal resolution, five years. The VLM consists of a series of submodels that interact in a spatially explicit way, simulating the effects of fire, logging and its associated road-building, and growth and succession on the structure and composition of the landscape. Roads are built when necessary to

June 2012


access cutblocks not otherwise accessible. New roads are the shortest line between cutblock and existing roads. Aging involves incrementing stand age with every consecutive time step after disturbance. Succession involves a semi-Markov chain of species state transitions (see James et al. 2007). After logging or fire, a new successional trajectory is computed for each cell based on information about strata and the trajectories of neighboring cells affected by the same disturbance. Fire size is based on a negative exponential distribution (as per Van Wagner 1978) with a mean of 2500 ha and a maximum of 30 000 ha. With a mean number of 6.7 fires per decade, the resulting fire cycle is ;250 years (as per Bergeron et al. 2001). Ignition and final fire size are independent of stand composition and age, making resulting fires more regular in shape than they might otherwise be, but also limiting the complexity of the model. No fire skips are modeled; each pixel is either burnt or not, and fire cannot spread across bodies of water wider than 25 m. To emulate the stand-replacing fires modeled here, the logging submodel simulates clear-cutting rather than partial cutting. Cutblock sizes are selected from a negative exponential distribution like that of fires, only truncated at 150, 300, 500, 1000, or 2000 ha (max 150– max 2000). This means that just over 80% of the cutblocks are set to be 50 ha, 5% are 50–100 ha, and the remaining 15% are selected from the rest of the permissible range. Unlike fire, logging was limited to pixels at or above minimum harvest age, as defined by the provincial authorities. These ages are defined per strata (a combination of species, drainage, and soil type) and vary from 47 to 94 years, with an average of 72 years. Logging priority is based on proximity to roads. The annual allowable cut is 2% in the intensive production zone, 0% in the biodiversity conservation zone, and 1% in the integrated management zone. In the intensive production zone, trees are assumed to provide the same volume as in the integrated management zone in half the time (i.e., growth rate is assumed to be double due to silvicultural enhancements, as has been found to be the case, for example, by Paquette and Messier [2010]). Thus, the rotation time is artificially halved in the intensive production zone. The number of cutblocks logged per time step varies with cutblock size, more cutblocks being logged when maximum cutblock size is lower. Note that, in the model, this is the only difference between the intensive production and integrated forest management zones. In both zones, volumes are estimated from regression analysis of plot data (see Côte´ et al. 2010). Since the temporal resolution of the model is five years, cutblocks could effectively be logged over a fiveyear period. Resulting cutblocks are similar in shape to modeled fires. Scenarios modeled We modeled 40 management scenarios and one unmanaged (fire-only) scenario. All scenarios included

1271

FIG. 2. The Vermillion model landscape with the four zoning options examined: (a) no zoning, (b) dyad, (c) dispersed triad, and (d) aggregated triad zoning.

fire, aging, and succession. The management scenarios also included logging and road building, and were composed of five different cutblock size distributions (see Model description) under four different zoning options: no zoning (logging throughout the forest), dyad (20% protected areas), and two types of triad (both with 20% protected areas and 20% intensive wood production), one with dispersed intensive wood production areas, the other with one large, aggregated intensive wood production area (Fig. 2). For the dyad and triad scenarios, three protected areas were selected to be as large as possible, but to be relatively dispersed to reduce the probability that all three would be struck by a single

1272



fire (see Côte´ et al. 2010). There was no logging or roadbuilding in the biodiversity conservation zone. In the dispersed triad, intensive wood production areas were close to roads and mills, and included existing plantations (Côte´ et al. 2010). In the aggregated triad, one large, intensive wood production area was located between protected areas (Fig. 2). We ran each scenario for 150 years, corresponding to the planning horizon in Quebec and elsewhere. We ran all scenarios on two initial landscapes: (1) the actual landscape, as described by the SIFORT data; and (2) a more ‘‘natural’’ landscape (henceforth called the simulated natural landscape), created by running only the fire submodel for 1000 years on the actual landscape to create a landscape representative of the fire regime modeled. Although we had aimed for a maximum cutblock size as large as the largest fires (30 000 ha), we were limited to 2000 ha because of a lack of available stands larger than the mean fire, regardless of initial landscape conditions. Statistical analysis We used general linear models and SPSS version 16.0 (SPSS 2007) to examine the effects of maximum cutblock size, zoning (no zoning, dyad, aggregated triad, and dispersed triad) and initial landscape conditions (actual and simulated natural) on percentage and configuration of old growth (.100 years), total harvest volume (in cubic meters), and total road construction (kilometers). When we found significant effects of zoning, we did Tukey’s post hoc analyses to identify significant differences between the four different zoning options (aggregated triad, dispersed triad, dyad, and no zoning; alpha ¼ 0.05). To measure landscape configuration, we used effective mesh size (meff ) of old growth, defined as follows: meff ¼ 1=At 3

n X

A2i

i¼1

where At is the total area of the study landscape, Ai is the area of patch i, and n is the total number of patches (Jaeger 2000). Effective mesh size can vary from 0 ha to the size of the study area, in this case 430 000 ha. Here, 0 ha would indicate a landscape completely devoid of old growth, 430 000 ha would indicate a landscape entirely made up of old growth (no fragmentation), and values in between would indicate a landscape with some old growth divided into patches of various sizes (e.g., 107 500 ha [a quarter of the total 430 000] would indicate a landscape half covered by one large patch of old growth). According to Van Wagner’s (1978) negative exponential model of stand age structure and a fire cycle of 250 years, we might expect an average of 67% of the forest to be .100 years old. If this old growth were all in one patch, we would have an effective mesh size of ;190 000 ha, whereas if it were split into 1000 patches of equal size, we would have an effective mesh size of 191 ha (R. Tittler, unpublished data). Effective mesh size is

less sensitive to small patches than mean patch size, reflecting the assumption that larger patches are of greater ecological value (Jaeger 2000). Although we ran all scenarios 20 times, we used the means of all runs of each scenario rather than the results of individual runs as the sample unit because of the artificial nature of the runs as replicates. In modeling, we choose the number of runs. With a large enough number of runs, statistical significance is almost guaranteed because of reduced variation around the mean. It is therefore difficult to choose an appropriate number of runs. By instead using the mean of the runs as the sample unit, we are able to choose the number of runs based on how representative the mean of these runs is likely to be of a theoretical true mean. We chose 20 runs because the mean calculated over all runs leveled off at this point. For analyses involving all responses but old growth, we included maximum cutblock size, zoning, and initial landscape as predictors and examined all two-way interactions except zoning 3 maximum cutblock size. We did not have sufficient replication to examine the latter two-way or the three-way interaction. Since effective mesh size is generally correlated with habitat amount, we also ran the effective mesh size analyses with controls for old growth to see whether differences in configuration were above and beyond differences in habitat amount. To control for percentage old growth, we added it as an independent variable in the analysis, examining the effects of interest only after the effects of percentage of old growth. Because we had no hypotheses about effects of cutblock size on amount of old growth, we included only zoning and initial landscape conditions in old growth analyses. To see if differences in old growth were due to differences in harvest volume (less area harvested leading to more old growth), we also ran these analyses with a control for harvest volume, adding it as an independent variable in the analysis in the same way as we controlled for the effect of percentage old growth on effective mesh size. RESULTS Although all management scenarios resulted in far less old growth than the fire-only scenarios, percentage old growth was significantly affected by zoning. In the fireonly scenario, mean percentage of old growth was .60%, whereas percentages varied from ;2% to 15% in the management scenarios. In all cases, zoning (triad and dyad) resulted in significantly more old growth than no zoning (Table 1, Fig. 3). When harvest volume was added to the analysis as an independent variable, the scenarios run on the simulated natural landscape resulted in more old growth than those run on the actual landscape. However, there was no significant interaction between initial landscape and zoning regardless of whether or not harvest volume was included in the analyses (Table 1).

June 2012


1273

TABLE 1. The results of general linear models examining the effects of maximum cutblock size (max. block, 150–2000 ha), zoning (aggregated triad [AT] vs. dispersed triad [DT] vs. dyad [D] vs. no zoning [No]), and initial landscape conditions (actual [A] and simulated natural [SN]) on various characteristics of the managed landscape. F

df

P

Partial R 2

Direction

Old growth (%) Initial Zoning Initial 3 zoning

1.692 71.914 0.199

1, 47 3, 47 3, 47

0.201 ,0.001 0.897

0.007 0.836 0.002

AT ’ D ’ DT . No

Old growth (%), harvest volume controlled Initial Zoning Initial 3 zoning

13.921 66.527 0.244

1, 39 3, 39 3, 39

,0.001 ,0.001 0.865

0.0391 0.560 0.002

SN . A AT ’ DT . D . No

Effective mesh size (ha) Initial Zoning Max. block Initial 3 zoning Initial 3 max. block

11.338 86.529 0.586 1.259 0.512

1, 3, 1, 3, 1,

39 39 39 39 39

0.002 ,0.001 0.450 0.306 0.480

0.037 0.849 0.002 0.012 0.002

SN . A D ’ DT ’ AT . No

Effective mesh size (ha), old growth controlled Initial Zoning Max. block Initial 3 zoning Initial 3 max. block

84.763 63.066 0.066 1.882 0.859

1, 3, 1, 3, 1,

39 39 39 39 39

,0.001 ,0.001 0.798 0.155 0.362

0.265 0.592 ,0.001 0.018 0.003

SN . A D ’ DT ’ AT . No

Road construction (km) Initial Zoning Max. block Initial 3 zoning Initial 3 max. block

37.107 10.280 135.293 0.151 0.191

1, 3, 1, 3, 1,

39 39 39 39 39

,0.001 ,0.001 ,0.001 0.928 0.665

0.159 0.132 0.578 0.002 ,0.001

SN . A No . DT ’ D ’ AT

Harvest volume (m3) Initial Zoning Max. block Initial 3 zoning Initial 3 max. block

992.168 116.356 291.736 3.649 0.528

1, 3, 1, 3, 1,

39 39 39 39 39

,0.001 ,0.001 ,0.001 0.024 0.473

0.593 0.208 0.174 0.007 ,0.001

SN . A No . DT ’ AT . D à

Response and predictor

Direction (þ or ) or order of significant effects. For zoning, these are results of Tukey’s post hoc analyses at alpha ¼ 0.05. à Differences are greater for the simulated natural landscape, but the trends are the same.

Effective mesh size was also greater for the fire-only scenarios than for any of the management scenarios, but was also affected by both initial landscape and zoning, regardless of whether or not percentage of old growth was statistically controlled for. In the fire-only scenario, mean effective mesh size was just over 14 000 ha under actual and 32 000 under simulated natural initial landscape conditions. It varied from ;7 to .6200 in the management scenarios. Scenarios run on the actual initial landscape generally resulted in lower effective mesh sizes (Table 1, Fig. 4a). Zoning (dyad and triad) resulted in larger effective mesh sizes than did no zoning, although there were no significant differences among zoning scenarios (between dyad and triad or aggregated and dispersed triad; Table 1, Fig. 3). There were also no significant interactions (Table 1, Fig. 3). Road construction was significantly affected by all main effects but no interactions. There was less road construction under actual than simulated natural initial landscape conditions. Fewer roads were constructed in the zoning scenarios (dyad and triad) than in the nozoning scenarios, although there were no significant

FIG. 3. The effects of zoning (aggregated triad, dispersed triad, dyad, and no zoning) and initial landscape (actual vs. simulated natural) on the percentage of old growth resulting from 150 years of simulation. Replicates represent different cutblock size distributions (maximum 150 ha to maximum 2000 ha).

1274



FIG. 4. The effect of maximum cutblock size, zoning, and initial landscape on (a) effective mesh size, (b) road construction, and (c) harvest volume resulting from 150 years of simulation.

differences among the three zoning scenarios. Finally, increasing maximum cutblock size reduced road construction (Table 1, Fig. 4b). Harvest volume was significantly affected by initial landscape conditions, zoning, and maximum cutblock size, but there was also a significant interaction effect between initial landscape and zoning. Harvest volume was highest for scenarios run on the simulated natural initial landscape, and was higher under no zoning than under any of the zoning options. Although there were no significant differences between the aggregated and dispersed triad options, they both resulted in significantly higher harvest volumes than did dyad. Like road construction, harvest volume was negatively affected by maximum cutblock size. Finally, there was a significant interaction between initial landscape conditions and zoning: differences between zoning scenarios were greater under simulated natural than under actual initial conditions, although the trends were in the same direction (Table 1, Fig. 4c). DISCUSSION In general, concentrating logging (some large cutblocks and protected areas counterbalanced by intensive

wood production in triad zoning) reduced the differences between the managed and unmanaged landscapes, lending support to the idea that it may be best to concentrate anthropogenic disturbance on the landscape. As predicted, both zoning and increased maximum cutblock size had positive effects on the landscape, although it did not seem to matter whether there was one large or many small intensive wood production areas in triad management. Zoning in general led to less road construction, higher levels of old growth, and less fragmentation, and triad zoning compensated somewhat for the loss in harvest volume associated with protected areas. As predicted, increasing maximum cutblock size also led to decreased road construction, since fewer roads had to be built to access larger cutblocks. On the other hand, apart from the effect on roads, increasing maximum cutblock size did not have the expected significant effects on configuration of old growth, perhaps because, in the end, there was relatively little variation in cutblock size among scenarios. Regardless of the specified distribution, 85% of all cutblocks were 100 ha or less; any variation was only in the distribution of the 15% of cutblocks that were .100 ha. In addition, the model could not consistently find

June 2012


large enough mature stands to log cutblocks .2000 ha, and we were thus unable to examine the effects of cutblock size distributions that emulated more closely the distribution of larger fire sizes. This may be because of the topographical structure of the landscape; as is typical of the Canadian Shield, the area is rich in rivers and lakes, which limited formation of contiguous blocks of harvestable stands. However, it likely also has to do with fires, which still structure the actual landscape. Note that the 2000-ha maximum corresponded fairly well to the average fire size (2500 ha). It may be beneficial to increase maximum cutblock size to reduce road construction, but the issue of residual tree retention in these cutblocks needs to be further explored. Some research indicates that cutblocks should have variable levels of retention in the form of riparian buffers and groups of live standing trees and snags. Such residual retention not only makes the cutblocks more similar to stand-replacing fires, which are often characterized by skips or patches of unburned trees, but may also provide habitat for wildlife and seed sources for regeneration (Gustafsson et al. 2010), and will likely make large cutblocks more socially acceptable. Without such residual, Greene et al. (2002) suggest that rates of natural regeneration may decline with increasing cutblock size. If they are to be implemented, guidelines for residual tree retention should be based on fire patterns; in this area, ;7–19% of the area is typically left unburned after fire (Dragotescu 2008), so similar levels of retention could be considered. However, these recommendations come from a stand-level perspective. From a landscape-scale perspective, if 7–19% of the area of cutblocks was left on site, more roads would have to be built and more cutblocks logged to make up for this loss of timber unless productivity was further increased in the intensive zone. To resolve the issue of cutblock size and retention, future research should weigh the costs of retention at the landscape scale against the benefits at the stand scale. The lack of predicted differences between aggregated and dispersed triad management may be due to the particular situation modeled here. Although in general there should be less road construction if wood production areas are concentrated, this effect may not occur here because the small, intensive wood production areas were specifically chosen to be near the existing road network (therefore requiring no or little road construction to access), whereas the large wood production area in the aggregated triad scenarios was located in a part of the landscape that was not particularly roaded. Further research should investigate this issue. Although there were almost no significant interactions between initial landscape and any of the other main effects, there were significant main effects of initial landscape conditions on the variables of interest. When the landscape was allowed to age for 1000 years, affected only by fire, it is to be expected that the resulting simulated natural initial landscape would have more

1275

mature and old stands available for harvest than the actual landscape. This would explain the higher percentages of old growth, effective mesh size, and harvest volumes resulting from scenarios run on simulated natural initial conditions, and the more road construction necessary to access this increased harvested area. Differences in harvest volume between zoning options were greater under simulated natural initial conditions than under actual initial conditions. This interaction implies that zoning, and particularly the setting aside of large protected areas, may come at a greater harvest volume cost in relatively untouched forests than in managed forests. While from an industrial perspective, it might be more profitable not to set aside large protected areas in relatively untouched forest ecosystems, from an ecological and biodiversity conservation perspective, the rare case of the relatively untouched forest is exactly when we should set aside large protected areas; such forest is arguably of much greater value as habitat for biodiversity and as a control by which to measure the effects of management than is forest that has already been affected by management. Although the modeled triad scenarios are not actually in place, the Mauricie triad project, just to the east of the study area, is similar to the dispersed triad scenario modeled here, involving relatively large protected and dispersed intensive wood production areas. Established in 2003, results of this pilot project so far support the idea that triad management may be a good compromise, incorporating biodiversity conservation values, timber production, and various other multiple uses of the forest (Messier et al. 2010). Unlike the triad scenarios modeled here, the Mauricie triad project involves a diversity of silvicultural treatments, not just clearcuts and intensive management. This is true in the integrated forest management zone (or ecosystem management zone, as it is called in the Mauricie project), where many forms of partial cuts are being applied. It is also true in the intensive zone, where plantations using both native and hybrid tree species in mono- and polyculture provide a large spectrum of methods to increase productivity. Since various forms of partial-cutting (with levels of retention varying from 10% to 75% or more) are becoming increasingly common in boreal systems, future modeling studies should incorporate the effects of such logging. This must be done in a spatially explicit way if landscape configuration and road building are to be examined. Furthermore, if we are to look at this finerscale process, we should also incorporate finer-scale natural disturbance processes, such as insect outbreaks and windthrow, so as to be able to model more realistic unmanaged landscapes for comparison (e.g., James et al. 2010). To assess ecological effects, it will also be critical to improve understanding of habitat–species relations in partial-cut stands.

1276



Although ours is the first study to jointly examine the effects of one large vs. many small intensive production areas, road construction, different cutblock size distributions, and natural disturbance in the managed landscape, our results are consistent with those of other modeling studies on triad zoning in terms of old growth and financial viability. Krcmar et al. (2003) compared triad with dyad zoning and found that the former allowed for a more viable forest industry. Ranius and Roberge (2011) suggest triad zoning as a way of decreasing extinction debts for insects dependent on dead wood. Montigny and MacLean (2006) and Côte´ et al. (2010) found increases in old growth with increasing area set aside for biodiversity conservation in triad management. Although it is unclear to what extent Côte´ et al.’s results can be attributed to their assumptions about partial-cuts, the fact that our results were similar indicates that the pattern observed is likely to hold. Conclusions and management implications According to our results, the differences between unmanaged and managed landscapes may be reduced by concentrating logging, i.e., by setting aside protected areas and counterbalancing this with an intensive production zone (triad). Although there is little evidence that one large is significantly better than many smaller intensive wood production areas, on the basis of these results, we recommend triad zoning in general as a viable option to reduce the differences between the unmanaged and the managed forest while limiting economic losses. As for cutblock size, we recommend a few cutblocks be allowed to be as large as 2000 ha in this landscape, or approximately equal to mean fire size in fire-dominated areas with different fire regimes, but that residual trees and patches be maintained, and that the effects of such a move be studied at the stand and landscape levels. Associated socioeconomic issues should also be examined. Note that the possibility of agglomerating smaller cutblocks should also be considered; in this model, cutblocks are effectively logged over five years, and could thus be considered agglomerations. In general, our results add to a growing body of literature supporting the idea that intensive management on a small part of the landscape is better than less intensive management spread out over a much larger part of the landscape or over the landscape as a whole. Further research should be done in various areas of landscape management to elucidate the extent to which this is true, and to examine thresholds above which such clustering is not possible or desirable. ACKNOWLEDGMENTS This project was funded by grants to C. Messier and collaborators from the Sustainable Forest Management Network, Abitibi-Bowater Inc., and the National Science and Engineering Research Council of Canada. The manuscript benefited from the comments and suggestions of two anonymous reviewers and the members of the lab of C. Messier.

LITERATURE CITED Attiwill, P. M. 1994. The disturbance of forest ecosystems. Forest Ecology and Management 63:247–300. Bergeron, Y., S. Gauthier, V. Kafka, P. Lefort, and D. Lesieur. 2001. Natural fire frequency for the eastern Canadian boreal forest. Canadian Journal of Forest Research 31:384–391. Boudreault, C., Y. Bergeron, S. Gauthier, and P. Drapeau. 2002. Bryophyte and lichen communities in mature to oldgrowth stands in eastern boreal forests of Canada. Canadian Journal of Forest Research 32:1080–1093. CBD. 1992. Convention on biological diversity. http://www. cbd.int/doc/legal/cbd-un-en. pdf Côte´, P., R. Tittler, C. Messier, D. D. Kneeshaw, A. Fall, and M.-J. Fortin. 2010. Comparing different forest zoning options for landscape-scale management of the boreal forest. Forest Ecology and Management 259:418–427. DeLong, C., and D. Tanner. 1996. Managing the pattern of forest harvest: lessons from wildfire. Biodiversity Conservation 5:1191–2105. Didion, M., M.-J. Fortin, and A. Fall. 2007. Forest age structure as indicator of boreal forest sustainability under alternative management and fire regimes: a landscape level sensitivity analysis. Ecological Modelling 200:45–58. Dragotescu, I. 2008. E´tude comparative des peuplements forestiers apre`s feux et apre`s coupes dans la forêt boreále mixte en Mauricie et au Te´miscamingue. Thesis. De´partement des sciences biologiques, Universite´ du Que´bec a` Montreál, Montreal, Canada. Drapeau, P., A. Leduc, Y. Bergeron, S. Gauthier, and J.-P. L. Savard. 2003. Les communaute´s d’oiseaux des vieilles forêts de la pessie`re a mousses de la ceinture d’argile. Forestry Chronicle 79:531–540. Eigenbrod, F., S. Hecnar, and L. Fahrig. 2009. Quantifying the road effect zone: threshold effects of a motorway on anuran populations in Ontario, Canada. Ecology and Society 14:24. Fahrig, L. 1998. When does fragmentation of breeding habitat affect population survival? Ecological Modelling 105:273– 292. Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology, Evolution and Systematics 34:487–515. Fall, A., and J. Fall. 2001. A domain specific language for models of landscape dynamics. Ecological Modelling 141:1– 18. Fall, A., M.-J. Fortin, D. D. Kneeshaw, S. H. Yamasaki, C. Messier, L. Bouthillier, and C. Smith. 2004. Consequences of various landscape-scale ecosystem management strategies and fire cycles on age-class structure and harvest in boreal forests. Canadian Journal of Forest Research 34:310–322. Fenton, N., M. Simard, and Y. Bergeron. 2009. Emulating natural disturbances: the role of silviculture in creating evenaged and complex structures in the black spruce boreal forest of eastern North America. Canadian Journal of Forest Research 14:258–267. Forman, R. T. T., and L. E. Alexander. 1998. Roads and their major ecological effects. Annual Review of Ecology and Systematics 29:207–231. Gagne´, S. A., and L. Fahrig. 2010a. The trade-off between housing density and sprawl area: minimizing impacts to Carabid beetles (Coleoptera: Carabidae). Ecology and Society 15:12. Gagne´, S. A., and L. Fahrig. 2010b. The trade-off between housing density and sprawl area: minimising impacts to forest breeding birds. Basic and Applied Ecology 11:723–733. Greene, D. F., D. D. Kneeshaw, C. Messier, V. Lieffers, D. Cormier, R. Doucet, G. Grover, K. D. Coates, A. Groot, and C. Calogeropoulos. 2002. Modelling silvicultural alternatives for conifer regeneration in boreal mixedwood stands (Aspen/ White spruce/Balsam fir). Forestry Chronicle 78:1–15.

June 2012


Gustafsson, L., J. Kouki, and A. Sverdrup-Thygeson. 2010. Tree retention as a conservation measure in clearcut forests of northern Europe: a review of ecological consequences. Scandinavian Journal of Forest Research 25:295–308. Haila, Y. 1994. Preserving ecological diversity in boreal forests. Annales Zoologici Fennici 31:203–217. Hodgson, J. A., W. E. Kunin, C. D. Thomas, T. G. Benton, and D. Gabriel. 2010. Comparing organic farming and land sparing: optimization yield and butterfly populations at a landscape scale. Ecology Letters 13:1358–1367. Hunter, M. L., Jr. 1990. Wildlife, forests, and forestry. PrenticeHall, Englewood Cliffs, New Jersey, USA. Hunter, M. L., Jr. 1993. Natural fire regimes as spatial models for managing boreal forests. Biological Conservation 65:115– 120. Hunter, W. C., D. A. Buehler, R. A. Canterbury, J. L. Confer, and P. A. Hamel. 2001. Conservation of disturbancedependent birds in eastern North America. Wildlife Society Bulletin 29:440–455. Jaeger, J. A. G. 2000. Landscape division, splitting index, and effective mesh size: new measures of landscape fragmentation. Landscape Ecology 15:115–130. Jaeger, J., R. Bertiller, and C. Schwick. 2007. Degree of landscape fragmentation in Switzerland: quantitative analysis 1885–2002 and implications for traffic planning and regional planning. Federal Office for the Environment, Swiss Confederation, Neuchaˆtel, Switzerland. James, P. M. A., M.-J. Fortin, A. Fall, D. D. Kneeshaw, and C. Messier, editors. 2007. The effects of spatial legacies following shifting management practices and fire on boreal forest age structure. Ecosystems 10:1261–1277. James, P. M. A., M.-J. Fortin, B. Sturtevant, A. Fall, and D. D. Kneeshaw. 2010. Modelling spatial interactions among fire, spruce budworm, and logging in the boreal forest. Ecosystems 14:60–75. Keller, I., and C. R. Largiade`r. 2003. Recent habitat fragmentation caused by major roads leads to reduction of gene flow and loss of genetic variability in ground beetles. Proceedings of the Royal Society of London B 270:417–423. Kneeshaw, D. D., and S. Gauthier. 2003. Old growth in the boreal forest. Environmental Review 11:S99–S114. Krcmar, E., I. Vertinsky, and G. C. van Kooten. 2003. Modelling alternative zoning strategies in forest management. International Transactions in Operational Research 10:483–498. Lohmus, P., and A. Lohmus. 2009. The importance of representative inventories for lichen conservation assess-

1277

ments: the case of Cladonia norvegica and C. parasitica. Lichenologist 41:61–67. McRae, D. J., L. C. Duchesne, B. Freedman, T. J. Lynham, and S. Woodley. 2001. Comparisons between wildfire and forest harvesting and their implications in forest management. Environmental Review 9:223–260. Messier, C., R. Tittler, D. D. Kneeshaw, N. Ge´linas, A. Paquette, K. K. Berninger. H. Rheault, P. Meek, and N. Beaulieu, editors. 2010. Triad zoning in Quebec. Forestry Chronicle 85:885–896. Montigny, M. K., and D. A. MacLean. 2006. Triad forest management. Forestry Chronicle 82:496–511. OMNR. 2001. Forest management guide for natural disturbance management emulation. Ontario Ministry of Natural Resources, Toronto, Ontario, Canada. Paquette, A., and C. Messier. 2010. The role of plantations in managing the world’s forests in the Anthropocene. Frontiers in Ecology and the Environment 8:27–34. Payette, S. 1992. Fire as a controlling process in the North American boreal forest. Pages 144–169 in H. H. Shugart, R. Leemans, and G. B. Bonan, editors. A systems analysis of the global boreal forest. Cambridge University Press, Cambridge, UK. Pelletier, G., Y. Dumont, and M. Be´dard. 2007. Syste`me d’information forestie`re par tesselle. Ministe`re des Ressources naturelles et de la faune, Gouvernement du Que´bec, Que´bec, Canada. Phalan, B., M. Onial, A. Balmford, and R. E. Green. 2011. Reconciling food production and biodiversity conservation: land sharing and land sparing compared. Science 2:1289– 1291. Ranius, T., and J.-M. Roberge. 2011. Effects of intensified forestry on the landscape-scale extinction risk of dead-wood dependent species. Biodiversity and Conservation 20:2867– 2882. Seymour, R. S., and M. L. Hunter, Jr. 1999. Principles of ecological forestry. Pages 22–61 in M. L. Hunter, Jr., editor. Maintaining biodiversity in forest ecosystems. Cambridge University Press, Cambridge, UK. SPSS. 2007. SPSS version 16.0. SPSS, Chicago, Illinois, USA. Van Wagner, C. E. 1978. Age-class distribution and the forest fire cycle. Canadian Journal of Forest Research 8:220–227. Yezerinac, S., and F. M. Moola. 2006. Conservation status and threats to species associated with old-growth forests within the range of the Northern Spotted Owl (Strix occidentalis caurina) in British Columbia, Canada. Biodiversity 6:3–9.