REFINING AN OLD FOREST QUALITY RANKING SYSTEM FOR THE PRINCE GEORGE FOREST DISTRICT

Transcription

1 REFINING AN OLD FOREST QUALITY RANKING SYSTEM FOR THE PRINCE GEORGE FOREST DISTRICT Report Prepared by: Manning, Cooper and Associates Ltd. Northern Interior Office # Quebec Street Prince George, B.C. V2L 1W2 CONTACT: Karl Bachmann Prepared For: Canadian Forest Products Ltd Northwood Pulpmill Road Prince George, B.C. V2L 4W2 ATTN: Janine Gervais

2 EXECUTIVE SUMMARY We present refinements made to a previous project on defining and ranking old forest in the Prince George Forest District by relative quality. Six intrinsic indicators and five additive indicators were used to rank old forest polygons (>120 years old or >140 years old) into six quality classes. This ranking was done at both the forest district scale and the BEC Zone scale. The model was run with only the intrinsic indicators, intrinsic plus some additive indicators, and a full model with all intrinsic and all additive indicators included. We report the percentage of old forest by quality rank within each merged NDU/BEC unit, for the full model run at the BEC Zone scale. Finally, we offer recommendations on the application of this model to forest management.

3 ACKNOWLEDGMENTS Janine Gervais with Canfor Ltd. provided digital data and gave technical input into the project structure. The Public Advisory Group Old Forest Subcommittee provided valuable input to the project and comments on the finished product. Matt Wheatley conducted all GIS and statistical analyses, organized digital data from a variety of sources into useable formats, and ranked and mapped the old forest polygons. Karl Bachmann and Matt Wheatley wrote this report. This project was funded by the Forest Investment Account (FIA) and administered by Canadian Forest Products Ltd.

4 TABLE OF CONTENTS EXECUTIVE SUMMARY... 2 ACKNOWLEDGMENTS... 3 TABLE OF CONTENTS... 4 LIST OF TABLES...5 LIST OF FIGURES... 5 LIST OF APPENDICIES INTRODUCTION Project Background Project Objectives and Parameters STUDY AREA METHODS Sample Population Model Refinements Old Forest Quality Mapping RESULTS DISCUSSION AND MANAGEMENT RECOMMENDATIONS Methodology and Management Implications Recommendations LITERATURE CITED... 21

5 LIST OF TABLES Table 1. Biogeoclimatic Ecosystem Classification areas in the Prince George Forest District Table 2. Measures of central tendency used to partition the final rank-sum scores into quality ranks for each model iteration. The six quality ranks correspond to the traditional six standard deviation partitions, three on either side of the mean Table 3. Areas of old forest quality ranks by merged NDU/BEC for the model run with intrinsic indicators only for the Prince George Forest District. Red numbers indicate the highest amounts of old forest for a particular merged NDU/BEC Table 4. Areas of old forest quality ranks by merged NDU/BEC for the model run with intrinsic indicators and all additive indicators for the Prince George Forest District. Red numbers indicate the highest amounts of old forest for a particular merged NDU/BEC LIST OF FIGURES Figure 1. Prince George Forest District in central British Columbia (map from Ministry of Forests webpage) Figure 2. Merged Biogeoclimatic Ecosystem Classification / Natural Disturbance Unit classification for the Prince George Forest District (map generated by Canfor Ltd.) Figure 3. Standard deviations used to derive old forest quality ranks in the Prince George Forest District LIST OF APPENDICIES Appendix 1. Histograms for 6 old forest quality indicators for old forest polygons at the Forest District and BEC zone scales in the Prince George Forest District. Y axis is number of polygons Appendix 2. Intrinsic indicator ranks at the Forest District and BEC Zone scales in the Prince George Forest District Appendix 3. Old forest quality categories mapped at the Forest District and BEC Zone scale for the Prince George Forest District. Darker shades of red indicator higher quality ranks for old forest Manning, Cooper and Associates Ltd. 5

6 1.0 INTRODUCTION 1.1 Project Background As public awareness and valuation of old-growth forests has increased, strategies to identify and retain old-growth have taken a central role in forest management in British Columbia (B.C.). The emphasis placed on old-growth forests reflects to a large part its inherent ecological importance. Un-managed areas, including old-growth, act as a benchmark through which the success of adaptive forest management can be judged (Kremsater et al. 2003). Given that either the habitat requirements or existence of many invertebrate species are largely unknown, retaining old-growth forest in an unmanaged state can act as a coarse filter approach to maintaining these species (Bunnell et al. 2003). Many species of vascular plants, fungi, lichen, vertebrates and invertebrates are either restricted to old-growth forests, or reach their highest population densities there (Carey 1989). Reflecting this importance, several documents have been developed in B.C. to direct planning for old-growth: the Old-growth Strategy for British Columbia (BC Ministry of Forest 1992), the Biodiversity Guidebook (BC Ministry of Forests and BC Ministry of Environment 1995), and the Landscape Unit Planning Guide (BC Ministry of Forests and BC Ministry of Environment, Lands and Parks 1999). This guidance, however, is broad scale and coarse in the sense that it offers a dichotomous old-growth or not old-growth definition to be applied to provincial scale forested areas in B.C. Currently, a standard method for defining and/or ranking old-growth areas and quality at a finer scale or in more detail is not available in B.C. There are many values implicit in and definitions surrounding the term old-growth forest. Wells et al. (1998) outlined 3 broad ways in which old-growth can be defined: economic, aesthetic, or ecological. Within the ecological category, Wells et al. (1998) further classify definitions into 3 categories: conceptual, quantitative, and working. A conceptual definition describes old-growth in a qualitative matter (e.g., a forest in which natural processes predominate), while a quantitative definition ascribes numbers or values to specific elements of old-growth (e.g., number of large live trees per hectare). Finally, working definitions are quantitative definitions that use some surrogate measure, usually forest inventory based, to define old forest in the absence of feasibly applicable quantitative definitions. Each of these 3 categories of ecological definitions can be based on forest demographic processes, forest structure and composition criteria, or gap dynamics criteria (Wells et al. 1998). Structural or compositional definitions are the most commonly used quantitative definitions, as they are readily measured and comparable across different stands (e.g., Quesnel 1996). The structural method classifies stands as old-growth once they attain some given threshold for structural attributes such as density of large live trees, m 2 basal area, density of large snags and other characteristics (e.g, MacKillop and Holt 2003, MacKillop and Holt 2004). This allows a very precise definition of which stands are old-growth versus which are not. It can be, however, costly to define stands in this manner, as many of the most useful structural attributes are not available at an appropriate accuracy in the ecological data commonly used by forest managers. Manning, Cooper and Associates Ltd. 6

7 Earlier work on old-growth stands tended to use a dichotomous old-growth or not old-growth approach to classify stands according to threshold values associated with a set of indicators (e.g., Franklin et al. 1981). The limitations of this approach become clear, however, when trying to classify the diversity of stands on the landscape. Many of the stands that would be considered old-growth do not share all or even many characteristics in common. From an ecological perspective, it is likely that all the old-growth stands existing on a landscape are not of equal value or quality. Thus, using an index of old-growthness that characterizes stands based on a continuum of old-growth attributes is probably most appropriate from an ecological perspective (Kimmins 2003). Additionally, this allows a judgment of old-growth quality to be made based on the sum of relative rankings of each indicator of old-growth. Because of the varying perceptions of what the term old-growth forest defines, for the purposes of this report we use the term old-forest. Old-forest denotes a working definition based on strictly a stand age, with the implicit assumption that old-growth characteristics are linked with age. This stand age definition often has varying threshold values by ecological unit (e.g. B.C. Ministry of Forests and B.C. Environment 1995). Defining old-growth temporally provides a simple and quantitative method for managing it in a forestry context. Applying somewhat arbitrary age thresholds to large ecological or geographic areas, however, can lead to the inclusion of stands that do not functionally act as old-growth forests, and vise versa (Harrison et al. 2002). Canadian Forest Products Ltd. (Canfor) Prince George Division has achieved Canadian Standards Association (CSA) certification under the auspices of its Sustainable Forest Management Plan (SFMP). The SFMP uses a criteria and indicators framework to assess the progress of sustainable forest management. Criterion 1 focuses on the conservation of biological diversity, requiring that the forest conserve biological diversity by maintaining integrity, function, and diversity of living organisms and the complexes of which they are part. Given the research indicating the importance of old forest as a component of biodiversity, effective identification, planning, and monitoring of old forest is crucial. 1.2 Project Objectives and Parameters The goal of this project was to refine the first iteration of a model to rank old forest polygons by their relative quality in the Prince George Forest District (MCA 2007). This involved incorporating Public Advisory Group (PAG) Old Forest Subcommittee and Licensee recommendations for model refinement. Two major issues were addressed in this refinement process: 1. Apply the model at 2 different observational scales to qualitatively assess how robust model results are to changes of scale 2. Run the model with different combinations of intrinsic and additive indicators to assess if individual indicators are driving the predicted distribution of old forest Manning, Cooper and Associates Ltd. 7

8 2.0 STUDY AREA Our study area was the Prince George Forest District (PGFD, Figure 1) with an area of 3.3 million ha., 2.4 million ha. of which are forested. Figure 1. Prince George Forest District in central British Columbia (map from Ministry of Forests webpage). The district encompasses portions of five Biogeoclimatic Ecosystem Classification (BEC) zones (Table 1, Meidinger and Pojar 1991). The majority of the area (72.9%) is dominated by low elevation (valley bottoms to m) Sub-boreal Spruce (SBS). This zone is characterized by long, cold winter and short, warm summers. Climax species are hybrid white spruce (Picea glauca x engelmannii) and subalpine fir (Abies lasiocarpa); lodgepole pine (Pinus contorta var. latifolia) is a common seral species. Higher elevation areas (> 900 m) are occupied by the Engelmann Spruce Subalpine Fir (ESSF) zone. Long, cold winters with a short growing Manning, Cooper and Associates Ltd. 8

9 season during the cool, moist summers predominate. Engelmann Spruce (Picea engelmannii) and subalpine fir are the climax species; lodgepole pine is the most common seral species. The Interior Cedar-Hemlock (ICH) zone is one of the wettest interior B.C. zones, occurring between 400 and 1500 m. Western red-cedar (Thuja plicata) and western hemlock (Tsuga heterophylla) and climax species, with a wide variety of other tree species occurring as seral species, or in low densities in climax stands. The Alpine Tundra (AT) and Sub-boreal Pine Spruce (SBPS) zones both occupy a relatively small portion of the PGFD. Table 1. Biogeoclimatic Ecosystem Classification areas in the Prince George Forest District. % FORESTED BEC VARIANT 1 AREA OF THE AREA (HA.) PG FD AT unp ESSFmm ESSFmv ESSFmv ESSFmv ESSFwc ESSFwk ESSFwk ICH vk ICH wk ICH wk ICH wk SBPSdc SBS dw SBS dw SBS dw SBS mc SBS mc SBS mh SBS mk SBS mw SBS vk SBS wk SBS wk Total See below for definition of BEC zone codes To better represent similarities and differences in natural disturbance processes and landscape pattern, DeLong (2002) classified the Prince George Forest Region into 9 different Natural Disturbance Units (NDUs). This is a refinement of the provincial Natural Disturbance Type (NDT) system. The PGFD encompasses portions of the Omineca, Boreal Foothills, McGregor Plateau, Wet Mountain, Wet Trench, and Moist Interior NDUs. Canfor has used the BEC and NDU system to classify the PGFD into 25 merged BEC/NDU units (Figure 2). The large white polygons in the centre and bottom edge of Figure 3 are Tree Farm Licenses (TFL) 30 and 53, respectively. Manning, Cooper and Associates Ltd. 9

10 Figure 2. Merged Biogeoclimatic Ecosystem Classification / Natural Disturbance Unit classification for the Prince George Forest District (map generated by Canfor Ltd.). 3.0 METHODS 3.1 Sample Population The sample population for this project was unchanged from MCA (2007): forests > 140 years old in the PGFD, with the exception of areas in the Moist Interior NDU (all BEC variants), the Omenica NDU (SBSdk, SBSdw3, BWBSdk1, SBSmc2, SBSmk1), and the McGregor NDU (SBS mk1 and SBSmh), where we adjusted the definition of old forest to > 120 years old based on corresponding biological circumstances of the tree species and moisture regimes of these areas. We used the same data sources and analysis criteria as MCA (2007) to define our landbase. Updated road and depletion layers to 2007 were used to update the results from 2006 (MCA 2007). 3.2 Model Refinements The methods used generally followed MCA (2007) but with some key refinements. First, we ranked old forest polygons relative to each other using two different observational scales: Manning, Cooper and Associates Ltd. 10

11 1. Within the entire Prince George Forest District, and; 2. Within each of 3 BEC zones (ESSF, SBS, and ICH) Within each observational scale, the relative importance of different polygons changed as a function of the area considered and the corresponding number and physical-ecological characteristics of the polygons contained therein. For the second observational scale, all old forest in the PGFD was ranked within three BEC zones: the ICH, the SBS, and the ESSF. There are in total five BEC zones within the PGFD, however, we excluded AT and SBPS because they were either non-forested (AT) or represented too small of an area (SBPS <0.1 % by area). Thus, we constructed four histograms for each indicator, one for the entire PGFD and one for each of the three BEC zones, effectively changing the relative abundance of polygons at each scale, as well as changing their quality rank or score accordingly. In doing so, we were able to account for key broad-scale ecological differences among BEC zones, which in turn incorporated these differences into the model iterations. Second, we adjusted the rank scores for each intrinsic indicator to represent abundance and rarity relative to all polygons included in each observational scale. This fundamentally changed the rank sums contingent on the observational scale used. For instance, to acquire a rank of 10 in volume for the PGFD observational scale required a much lower value in m 3 compared to a within-bec scale. Differences among observational scale were characteristic of volume, polygon age, core forest, and the number of tree species. Thus, ranking systems for these indicators are entirely different among observational scales. Differences were not apparent for polygon size and road densities, so ranking schemes for these indicators were kept consistent among observational scales (Appendix 1). Finally, the ranks assigned to different indicator values were based on the abundance of different values within each scale; that is, high ranks were assigned based on rarity in a frequency distribution and lower ranks were assigned based on commonness. Consequently, this method assigns high quality ranks to polygons that contain the most old-forest indicators relative to whatever other polygons greater than a certain age (either 120 or 140 yrs) contain. In all models, therefore, users are able to easily identify and quantify relatively important old forest polygons. Third, final quality categories (1 through 6) were assigned to polygons based on the central tendencies inherent in each area s frequency histogram of all final rank-sum scores. A quality category (1 through 6; good through best respectively) was assigned to each successive standard deviation on either side of the mean (Figure 3) based on each model s rank-sum standard deviation. Again, similar to the preceding individual indicator ranking system, this method assigned higher ranks based on both attribute scores and rarity of abundance. Manning, Cooper and Associates Ltd. 11

12 Old Forest Quality Rank 1 - good best Figure 3. Standard deviations used to derive old forest quality ranks in the Prince George Forest District. Each standard deviation represents a fixed percentage of the data (e.g., 34 % of the data falls within 1 stand deviation to the right of the mean), as defined by each observational scale s resulting frequency histogram. This system allowed comparison of old forest quality between different BEC zones despite having used different ranking values contingent on the relative values within each zone. Fourth, different iterations of the resulting model are presented here and are incorporated into the final mapping.shp file for use in GIS. After initial model development (MCA 2007) there was interest in understanding the relative effect of adding different additive features and addressing ecological differences among BEC zones on the final quality ranks appearing on the final map product. Further, there was indication that some additive indicators should be weighted heavier than others in future model iterations. To address these issues, at each observational scale we produced three final models, each consisting of different combinations of observational scale or combinations of additive polygon features (Table 2). These 6 model iterations began by looking at just intrinsic indicators at the 2 observational scales, then progressively adding either 3 or 5 additive indicators to these 2 separate scales. Finally, we weighted three of the additive features heavier than previous model iterations (Table 2). Manning, Cooper and Associates Ltd. 12

13 Table 2. Measures of central tendency used to partition the final rank-sum scores into quality ranks for each model iteration. The six quality ranks correspond to the traditional six standard deviation partitions, three on either side of the mean. Quality Rank Metrics b good best Model Iteration Model 1 Just intrinsic, whole FD level mean=32; sd=8 Model 2 Just intrinsic, BEC-level ranks mean=35; sd=6 Model 3 Intrinsic FD level, plus 3 additives a mean=37; sd=11 Model 4 Intrinsic FD level plus 5 additives mean=38; sd=12 Model 5 Intrinsic BEC-level plus 3 additives a mean=40; sd=10 Model 6 Intrinsic BEC-level plus 5 additives mean=41; sd=10 < >50 < >49 < >61 < >64 < >62 < >63 a initial 3 additives include caribou habitat, red- and blue-listed species, and ecosystem representation. b based on central tendencies (mean and standard deviation) of each model. Model iterations were as follows (outlined in Table 2). Model 1 included forest polygons ranked using only intrinsic indicators at the PGFD scale. Model 2 included forest polygons ranked by intrinsic indicators but with quality scores adjusted to their relative importance within each BEC zone. For example, forest age structure differs significantly among BEC zones, with the ICH showing some of the region s oldest forests (Appendix 1). Thus, by ranking these zones individually, quality scores become different within each. In other words, to obtain a high score for age, a forest polygon must be much older in the ICH relative to the ESSF. Model 3 used the intrinsic indicators at the PGFD scale plus 3 additive indicators including caribou, red-and-blue listed species, and ecosystem representation. Moreover, the rank weights for these were increased from 5 (MCA 2007) to 10 to reflect their importance regarding threatened or rare species management. Model 4 added the last 2 additive features to model 3, namely ungulate winter range and protected areas. Because they possess their own management strategies in other Manning, Cooper and Associates Ltd. 13

14 sectors of government, and are not representative of declining or rare components of crown land, these final 2 additive features retained their original rank-weight of 5 (MCA 2007). Model 5 included the intrinsic indicators ranked at the BEC scale, plus the first three additive features (caribou, red-and-blue, and ecosystem representation; all ranked at 10), and Model 6 added the last 2 additive features (ungulate winter range and parks) to Model 5. Finally, there was interest in using the model refinements to integrate these findings into a management context to demonstrate how the model might be used on-the-ground. To address this, we present the quality predictions for Model 6 (quality categories 1 through 6) by the 25 merged BEC/NDU units and compare these with Canfor s old forest area target summaries for the same units. 3.3 Old Forest Quality Mapping We used ArcGIS ver9.x and ArcView ver3.3 (ESRI, Redlands, CA) to complete initial spatial data management and summaries. Mapping was completed using BC Ministry of Environment digital data standards ( and follow protocols as defined by FIA Project Spatial data were obtained from Canfor (Prince George Division) and the Provincial Land and Resource Data Warehouse ( All analyses presented herein were based on provincial Forest Cover polygons. Mapping and associated digital products resulting from this project are provided in the BC Albers standard geographic projection. 4.0 RESULTS Observational scale rank adjustments. Frequency distributions for intrinsic indicators differed among observational scales for some (age, volume, old interior forest, number of tree species), but was similar for others (road density, polygon area; Appendix 1). Based on these differences, we employed a different set of rank criterion (from 1 to 10) to reflect rarity and abundance of different metric scores (Appendix 2). For example, the average volume for forest polygons in the ESSF is less than those in the ICH (Appendix 1). Accordingly, to achieve a rank of 10 for volume (one of the 6 intrinsic indicators) in the ESSF, a lower volume in m 3 must be obtained relative to the ICH (i.e., 310m 3 versus 460m 3 ). For metrics with similar distributions among observational scale (e.g., polygon area), we used the same rank system for all observational scales (Appendix 1, Appendix 2). Visual interpretations of resulting maps. In this report we present resulting rank-quality maps at the forest-district scale (Appendix 3), therefore, local-level changes in rank quality among smaller areas are not directly evident. These smaller changes can easily be quantified, however, using the associated digital.shp file containing all 6 model iterations produced as part of this project. Here, we provide only a broad-scale forest district interpretation of how each model adjusted forest-quality rankings. An important note must be made regarding rank qualities: although some polygons are ranked higher than others, all of them captured in this analysis still represent old forest in some context. Areas ranked lower and colored light-yellow on the maps Manning, Cooper and Associates Ltd. 14

15 should not be interpreted as poor-quality forest, just ranked differently relative to other polygons. Intrinsic Indicators PGFD observational scale. At the PGFD observational scale, there is a clear prevalence of higher-ranked old forest in the eastern portion of the forest district. Because data at the PGFD scale were ranked irrespective of BEC zone, there is an abundance of high-ranked old forest in the ICH areas, likely driven by both volume and age ranks. As such, an abundance of polygons in the 1 through 3 rank categories covers the western portion of the forest district. BEC Zone observational scale. When ranks are adjusted within each BEC zone, though not strikingly apparent on the forest-district level maps, more old forest ranked in the 3 through 6 categories is present in the western portion of the study area. BEC zone related changes in these ranks are more apparent at smaller observational scales than in the maps included in this report, and will become more evident using the digital.shp file of quality ranks in GIS. Adjusting ranks to account for ecological differences among BEC zones creates more higher-ranked old forest in the western portion of the forest district. Interpretation of forests ranked at the BEC observational scale should fully acknowledge the ecological differences of the forests among BEC zones; some of these are apparent in the histograms provided here (Appendix 1), and will potentially drive large differences in quality ranks for some areas at smaller spatial extents (i.e., management level areas). Additive indicators The effect of additive indicators can be seen mostly in the eastern portion of the forest district, and their effect is similar for both the BEC and PGFD observational scales. Adding the first three additive indicators (caribou, red-and-blue, and ecosystem representation) creates the most obvious visible effect on the maps, creating deep-red areas in abundance throughout the eastern portions of the district. The addition of the final 2 additive features does not alter the general appearance of the map, however, changes in rank-qualities do occur, but are apparent only at smaller scales below those presented here. Of note, however, is the reduction in red-ranked polygons (categories 3 through 6) in the north tip of the district when the last 2 additive features are included in a full model. These alter the relative quality ranks such that northern polygons are of relatively lower quality (e.g., Model 3 versus Model 4). Addition of the first 3 additives resulted in a shift in quality from west to east, while the addition of the last 2 additives resulted in a quality shift from north to southwest. It is important to note that the addition of additive features not only changes the rank-sum totals for corresponding polygons, it also lowers the relative ranks of polygons not containing additive features. Therefore, while additive features make some areas turn deeper red, they also de facto make other areas turn lighter yellow. Merged BEC unit management interpretation. The models developed here can be partitioned into different administrative or management boundaries as required. By doing so, managers can acquire a per-area representation of different forest qualities within different management units and compare these to their old-forest management targets. Within the PGFD, there are different by-area targets for old forest retention in forest-management contexts, presented as total hectares by merged BEC/NDU units (Table 3, Table 4). These targets allow a direct comparison between what we calculated here as total hectares of old forest present in each NDU and what are set as Manning, Cooper and Associates Ltd. 15

16 management-based targets (highlighted columns in Table 3 and Table 4). When comparing total hectares for what we calculated as present within each NDU for each old-forest quality rank to the set targets for forest management, in every case the total existing in each merged BEC/NDU unit is greater than the set retention targets (Table 3 and Table 4, Anonymous 2006). What we present in Table 3 and Table 4 is based on the full model (Model 6) using the BEC observational scale and including all additive indicators. Shifts in the amounts of old forest will be evident as different models are used and compared to old-forest retention targets. Manning, Cooper and Associates Ltd. 16

17 Table 3. Areas of old forest quality ranks by merged NDU/BEC for the model run with intrinsic indicators only for the Prince George Forest District. Red numbers indicate the highest amounts of old forest for a particular merged NDU/BEC. Natural Merged Old Forest Target Disturbance NDU/BEC Total CFLB Unit (NDU) (ha) % Hectares Boreal Foothills A1 7,255 33% 2,394 McGregor McGregor McGregor Moist Interior A5 12,396 Moist Interior A6 16,417 Moist Interior A7 5,928 Moist Interior A8 9,145 Moist Interior A9 33,443 Moist Interior A10 39,088 Moist Interior A11 128,566 Moist Interior A12 179,032 Moist Interior A13 370,589 Wet Mountain A14 154,009 Wet Mountain A15 27,832 Wet Mountain A16 33,914 Wet Mountain A17 114,673 Wet Trench A18 33,997 Wet Trench A19 65,010 Wet Trench A20 98,712 Wet Trench A21 114,753 Wet Trench A22 27,176 Wet Trench A23 145,660 Wet Trench A24 131,802 Wet Trench A25 152,701 NTA Totals (ha.) 2,203,482 A2 10,349 26% 2,691 A3 71,779 12% 8,613 A4 219,256 26% 57,007 29% 3,595 29% 4,761 17% 1,008 12% 1,097 12% 4,013 17% 6,645 12% 15,428 12% 21,484 12% 44,471 50% 77,005 84% 23,379 26% 8,818 50% 57,337 80% 27,198 48% 31,205 80% 78,970 48% 55,081 53% 14,403 53% 77,200 30% 39,541 46% 70,242 Very Low % area of old forest within rank category Low Marginal Medium High Very High Total ha. 0.3% 1.0% 2.6% 23.3% 58.3% 14.4% 3, % 2.8% 42.4% 32.3% 22.5% 0.0% 8, % 9.2% 29.7% 43.1% 17.1% 0.4% 33, % 5.5% 28.5% 43.4% 20.5% 1.8% 83, % 1.1% 18.3% 46.4% 27.1% 7.1% 7, % 2.1% 16.7% 41.7% 35.4% 4.2% 10, % 33.5% 41.2% 19.6% 4.5% 1.1% 3, % 8.3% 66.9% 24.6% 0.0% 0.0% 5, % 11.1% 41.1% 41.0% 5.8% 0.8% 9, % 4.6% 23.2% 43.7% 26.2% 2.1% 22, % 13.7% 47.1% 32.8% 5.1% 0.3% 71, % 11.0% 39.4% 38.1% 10.4% 0.5% 80, % 10.9% 34.3% 40.9% 12.4% 0.9% 154, % 4.3% 24.5% 37.0% 28.4% 5.8% 142, % 15.1% 42.1% 29.2% 11.2% 2.4% 35, % 2.5% 20.5% 53.3% 19.6% 4.0% 15, % 1.6% 16.5% 45.7% 33.3% 2.8% 90, % 6.7% 27.4% 43.4% 20.9% 1.8% 50, % 1.1% 10.0% 30.7% 43.5% 14.8% 61, % 2.0% 13.6% 34.4% 44.3% 5.8% 102, % 0.8% 6.6% 32.8% 51.8% 7.9% 70, % 5.1% 22.5% 38.4% 33.0% 0.7% 19, % 2.7% 24.0% 40.9% 26.7% 5.5% 100, % 4.8% 19.4% 39.8% 31.6% 4.3% 44, % 3.2% 19.6% 43.3% 31.9% 1.8% 82, % 3.9% 7.9% 88.3% 0.0% 0.0% % 733,586 3, , , , , , ,312, Manning, Cooper and Associates Ltd. 17

18 Table 4. Areas of old forest quality ranks by merged NDU/BEC for the model run with intrinsic indicators and all additive indicators for the Prince George Forest District. Red numbers indicate the highest amounts of old forest for a particular merged NDU/BEC. Natural Merged Old Forest Target Disturbance NDU/BEC Total CFLB Unit (NDU) (ha) % Hectares Boreal Foothills A1 7,255 33% 2,394 McGregor McGregor McGregor Moist Interior A5 12,396 Moist Interior A6 16,417 Moist Interior A7 5,928 Moist Interior A8 9,145 Moist Interior A9 33,443 Moist Interior A10 39,088 Moist Interior A11 128,566 Moist Interior A12 179,032 Moist Interior A13 370,589 Wet Mountain A14 154,009 Wet Mountain A15 27,832 Wet Mountain A16 33,914 Wet Mountain A17 114,673 Wet Trench A18 33,997 Wet Trench A19 65,010 Wet Trench A20 98,712 Wet Trench A21 114,753 Wet Trench A22 27,176 Wet Trench A23 145,660 Wet Trench A24 131,802 Wet Trench A25 152,701 NTA Totals (ha.) 2,203,482 A2 10,349 26% 2,691 A3 71,779 12% 8,613 A4 219,256 26% 57,007 29% 3,595 29% 4,761 17% 1,008 12% 1,097 12% 4,013 17% 6,645 12% 15,428 12% 21,484 12% 44,471 50% 77,005 84% 23,379 26% 8,818 50% 57,337 80% 27,198 48% 31,205 80% 78,970 48% 55,081 53% 14,403 53% 77,200 30% 39,541 46% 70,242 % area of old forest within rank category Very Low Low Marginal Medium High Very High Total ha. 0.0% 0.0% 0.5% 7.1% 48.9% 43.5% 3, % 0.6% 17.0% 61.1% 18.4% 2.8% 8, % 6.0% 48.3% 38.1% 7.4% 0.1% 33, % 5.7% 49.0% 33.4% 10.2% 1.6% 83, % 0.1% 13.0% 32.4% 41.2% 13.3% 7, % 0.1% 2.6% 17.5% 35.3% 44.5% 10, % 19.7% 59.5% 19.6% 1.1% 0.0% 3, % 0.7% 21.4% 71.6% 6.4% 0.0% 5, % 13.2% 63.4% 19.6% 3.8% 0.0% 9, % 3.6% 35.7% 39.0% 17.2% 4.4% 22, % 18.0% 67.9% 13.2% 0.6% 0.0% 71, % 12.8% 61.3% 23.0% 2.1% 0.5% 80, % 9.7% 47.2% 34.4% 8.5% 0.1% 154, % 1.5% 16.5% 35.7% 29.7% 16.6% 142, % 0.7% 24.2% 38.2% 25.7% 11.2% 35, % 4.1% 62.1% 28.7% 3.4% 1.7% 15, % 1.0% 26.0% 39.5% 28.1% 5.4% 90, % 0.3% 5.2% 23.5% 46.5% 24.6% 50, % 0.3% 5.5% 21.5% 44.6% 28.2% 61, % 0.1% 3.9% 16.1% 40.0% 39.8% 102, % 0.1% 2.0% 14.9% 34.7% 48.3% 70, % 6.3% 40.8% 44.4% 8.2% 0.2% 19, % 1.9% 28.6% 41.6% 23.4% 4.5% 100, % 5.7% 41.3% 38.8% 11.5% 2.6% 44, % 2.4% 28.0% 41.3% 23.9% 4.4% 82, % 0.0% 5.7% 18.9% 75.4% 0.0% % 733, , , , , , ,312, Manning, Cooper and Associates Ltd. 18

19 5.0 DISCUSSION AND MANAGEMENT RECOMMENDATIONS 5.1 Methodology and Management Implications Additive indicators change the relative distribution of quality ranks within the forest district. The increasing concentration of orange and red (quality ranks 3 through 6) in the eastern portion of the district is largely driven by the additive indicators of caribou and red and blue listed species occurrences. These 2 additive indicators are concentrated in the eastern portion of the forest district, have large spatial extents (MCA 2007) and are both ranked relatively heavily (i.e., 10 versus 5). As both caribou and red and blue listed ecosystems are a management focus in the PGFD, we felt it justifiable to assign the heavier weighting to these additive indicators. The final rank-sums upon which quality ranks are based will require significant changes to affect final rank qualities. Because the model framework incorporates several indicators of old forest quality which together tally to upwards of 65 or 70 in total scores, and because each indicator is only one-sixth of this final total, the models are somewhat robust to changes in one single indicator. Thus it is important to use an a priori approach to constructing the model, rather than manipulating the list of included indicators or indicator weightings to fit pre-conceived ideas of how high quality old forest should be distributed. We have attempted to facilitate this process by revising the original model into 6 different iterations, which can be selected from depending on the management context. Users of these models are encouraged to explore the applicability of the different indicator combinations when applying them regardless of spatial extent. Notable differences are apparent in the frequency histograms when compared among BEC zones. Therefore, it is our opinion that classification of old forest into different quality categories is best done at the BEC Zone observational scale. BEC zones incorporate significant differences between plant communities and other large-scale physiographic properties among ecologically distinct areas, each containing unique components of biodiversity. If old forest quality rankings are compared to each other across BEC zones, then zones that naturally have lower values for some or all indicators will have less high quality old forest identified. For example, ICH stands in the PGFD have a higher mean volume than do SBS stands. If all old forest polygons in the PGFD are ranked against each other, then relatively more ICH stands will fall into the right-hand (higher ranked) side of the distribution. In a case where all other indicators were equal, this would result in ICH stands being ranked higher than SBS stands. A relative adjustment of these ranks incorporating BEC-zone differences (i.e., models 2, 5, and 6) is encouraged. Users of these models must be acutely aware of the scale issue (Marceau 1999) and the modifiable area unit problem (MAUP; Oppenshaw 1984) when applying and interpreting these models. These 2 issues essentially state that sampling a bigger area with less detail looks different than sampling a smaller portion of the same area with more detail. In this context, one may find less differentiation between uncommon polygons within each BEC zone if the ranking is done at a forest district scale versus a BEC scale. For example, when compared to low volume SBS stands, relatively more ICH stands will be included in the top rankings for volume. Looking at volumes for polygons only within ICH stands, however, a stand has to have a relatively higher volume to be classed into a higher rank. This spreads out the fewer values of an Manning, Cooper and Associates Ltd. 19

20 indicator across the same 1 to 10 ranking, giving the truly rare values at either end of the distribution reflectively lower or higher ranks. A fundamentally different observation would occur depending on the spatial extent addressed when assigning ranks, because this extent changes the relative importance of the polygons included therein (i.e., MAUP). Finally, it is important to remember that this model is entirely based on provincial forest inventory data. Thus, the accuracy with which the model identifies high quality old forest will be dependant on the accuracy of the inventory data. As this model is practically applied, limitations or inaccuracies will be become apparent. For example, there will be instances (effectively unavoidable) where old-forest polygons identified using these models are in fact post-harvest cutblocks on-the-ground; an effect of Forest Cover update schedules, not the accuracy of the modeling process. Where possible, Forest Cover updates should be addressed in subsequent iterations. Additionally, incorporating new inventories or modeling initiatives (e.g., large-scale snag or coarse-woody debris modeling) will allow more precise identification a resolution of high quality old forest. 5.2 Recommendations Based on the results of our revised ranking of old forest in the PGFD, we recommend: 1. Using the model at BEC Zone observational scale. 2. Use a model iteration incorporating both intrinsic and additive indicators. Additive indicators are not based on purely ecological criteria, in contrast to intrinsic indicators which are defined on a range of existing values. They do, however, allow other values to be identified concurrently to identifying higher quality old forest. 3. Update this model with new information as it becomes available. Results presented in this paper represent the best attempt to rate high quality old forest given the data available. Future iterations of this model should be informed by: (1) feedback from practical application of the model, and (2) updated or new inventories. 4. Avoid interpreting yellow-ranked forest (i.e., quality ranks 1 through 3) as poor quality. These polygons still represent significant attributes of old forest, even though they may contain less relative to other polygons. Manning, Cooper and Associates Ltd. 20

21 6.0 LITERATURE CITED 1. Anonymous Prince George Sustainable Forest Management Plan Arsenault, A A note on the ecology and management of old-growth forests in the Montane Cordillera. The Forestry Chronicle 79(3): Attiwill, P.M The disturbance of forest ecosystems: the ecological basis for conservation management. Forest Ecology and Management 63: BC Ministry of Forests Old growth strategy for British Columbia. Ministry of Forests. Victoria, British Columbia. 74 p. 5. BC Ministry of Forests and BC Environment Biodiversity guidebook forest practices code of British Columbia. Ministry of Forests and Ministry of Environment. Victoria, British Columbia. 99 p. 6. BC Ministry of Forests and BC Ministry of Environment, Lands and Parks Lanscape unit planning guide. Ministry of Forests and Ministry of Environment, Lands and Parks. Victoria, British Columbia. 101 p. 7. BC Sustainable Forest Management Plan Working Group Working paper: developing a sustainable forest management plan. 8. Botting, R.S. and A.L. Fredeen Contrasting terrestrial lichen, liverwort, and moss diversity between old-growth and young second-growth forest on two soil textures in central British Columbia. Canadian Journal of Botany 84: Bunnell, F., G. Dunsworth, D. Huggard, and L. Kremsater Learning to sustain biological diversity on Weyerhaeuser s costal tenure. Unpublished report. 10. Burton, P.J., D.D. Kneeshaw, and K.D. Coates Managing forest harvesting to maintain old growth in boreal and sub-boreal forests. The Forestry Chronicle 75(4): Carey, A.B Wildlife associated with old growth forests in the Pacific Northwest. Natural Areas Journal 9(3): Center for International Forestry Research Guidelines for developing, selecting and testing criteria and indicators for sustainable forest management; criteria and indicators toolbox series #1. CIFOR Jakarta, Indonesia DeLong, C Natural disturbance units of the Prince George Forest Region: guidance for sustainable forest management. Unpublished report. Manning, Cooper and Associates Ltd. 21

22 14. DeLong, S.C Implementation of natural disturbance based management in northern British Columbia. The Forestry Chronicle In Press. 15. Demarchi, D.A An introduction to the ecoregions of British Columbia. Wildlife Branch, Ministry of Environment, Lands and Parks. Victoria, British Columbia. 16. Franklin, J.F., K. Cromack Jr., W. Denison, A. McKee, C. Maser, J. Sedell, F. Swanson, and G. Juday Ecological characteristics of old-growth Douglas-fir forests. United States Department of Agriculture, Forest Service, Pacific Northwest Forest and Range Experimental Station. General Technical Report PNW Harrison, M., C. Delong, and P.J. Burton A comparison of ecological characteristics in stands of differing age in the ICHwk3, ESSFwk2, ICHmm, and ESSFmm: development of an index to assess old-growth features. Report prepared for Robson Valley Enhanced Forest Management Project. 18. Hunter, M.L. and A.S. White Ecological thresholds and the definition of old-growth forest. Natural Areas Journal 17: Kimmins, J.P (Hamish) Old-growth forest: an ancient and stable sylvan equilibrium, or a relatively transitory ecosystem condition that offers people a visual and emotional feast? Answer it depends. The Forestry Chronicle 79(3): Klenner, W., R. Walton, and W. Kurz Habitats for tomorrow: understanding the consequences for today s decisions and natural disturbances on future habitat condition. In: L. M. Darling, editor Proceedings of a Conference on the Biology and Management of Species and Habitats at Risk, Kamloops, B.C., Feb.,1999. Volume One. B.C. Ministry of Environment, Lands and Parks, Victoria, B.C. and University College of the Cariboo, Kamloops, B.C. 490pp. 21. Kneeshaw, D.D. and P.J. Burton Assessment of functional old-growth status: a case study in the sub-boreal spruce zone of British Columbia, Canada. Natural Areas Journal 18(4): Kneeshaw, D. and S. Gauthier Old growth in the boreal forest: a dynamic perspective at the stand and landscape level. Environmental Review (Supplement 1): Kremsater, L., F. Bunnell, D. Huggard, and G. Dunsworth Indicators to assess biological diversity: Weyerhaeuser s coastal British Columbia forest project. The Forestry Chronicle 79(3): MacKillop, D. and R.F. Holt Describing old-growth forests in the Boreal Foothills natural disturbance unit. Report prepared for Canadian Forest Products Ltd. and Chetwynd Forest Industries, a division of West Fraser Mills, Ltd. Manning, Cooper and Associates Ltd. 22

23 25. MacKillop, D. and R.F. Holt Mountain pine beetles (Dendroctonus ponderosae) and old-growth forest characteristics in the Moist Interior Plateau, Vanderhoof Forest District. Research report prepared for West Fraser Sawmills, Fraser Lake, B.C. 26. Marceau, D.J The scale issue in social and natural sciences. Canadian Journal of Remote Sensing 25: MCA Ltd Old forest quality indicators and ranking for the Prince George Forest District. FIA report prepared for Canadian Forest Products Ltd., Prince George, B.C. 28. Meidinger, D. and J. Pojar Ecosystems of British Columbia. B.C. Ministry of Forests Special Report Series 6, Victoria, B.C. URL: Newmaster, S.G, R.J. Belland, A. Arsenault, and D.H. Vitt Patterns of bryophyte diversity in humid coastal and inland cedar-hemlock forests of British Columbia. Environmental Review 11: S159-S Openshaw, S., The Modifiable Areal Unit Problem. Concepts and Techniques in Modern Geography (CATMOG), No. 38, 40 p. 31. P. Beaudry and Associates Ltd Aquatic indicator development, Project no: Prepared for Slocan Forest Products Ltd. Plateau Division. Vanderhoof, B.C. 32. Quesnel, H.J Assessment and characterization of old-growth stands in the Nelson Forest Region. Final Report for E.P Technical Report TR-013. Forest Sciences Section, Ministry of Forests. 33. Spies, T.A Forest structure: a key to the ecosystem. Northwest Science 72 (Special Issue No. 2): Wells, R.W., K.P. Lertzman, and S.C. Saunders Old-growth definitions for the forests of British Columbia, Canada. Natural Areas Journal 18(4): Manning, Cooper and Associates Ltd. 23

24 Appendix 1. Histograms for 6 old forest quality indicators for old forest polygons at the Forest District and BEC zone scales in the Prince George Forest District. Y axis is number of polygons. Age (x axis is age in years) Forest District ESSF Manning, Cooper and Associates Ltd. 24

25 ICH SBS Manning, Cooper and Associates Ltd. 25

26 Volume (x axis is volume in m 3 ) Forest District ESSF Manning, Cooper and Associates Ltd. 26

27 ICH SBS Manning, Cooper and Associates Ltd. 27

28 Patch Area (x axis in hectares) Forest District ESSF Manning, Cooper and Associates Ltd. 28

29 ICH SBS Manning, Cooper and Associates Ltd. 29

30 Road Density (x axis in m/m 2 ) Forest District ESSF Manning, Cooper and Associates Ltd. 30

31 ICH SBS Manning, Cooper and Associates Ltd. 31

32 Old Interior Forest (x axis is % area by polygon) Forest District ESSF Manning, Cooper and Associates Ltd. 32

33 ICH SBS Manning, Cooper and Associates Ltd. 33

34 Number of Tree Species Forest District ESSF Manning, Cooper and Associates Ltd. 34

35 ICH SBS Manning, Cooper and Associates Ltd. 35

36 Appendix 2. Intrinsic indicator ranks at the Forest District and BEC Zone scales in the Prince George Forest District. Age (years) Forest District Rank Value 1 < >230 ICH Rank Value 1 < >320 ESSF Rank Value 1 < >230 SBS Rank Value 1 < >220 Manning, Cooper and Associates Ltd. 36

37 Volume (m 3 ) Forest District Rank Value 1 < >400 ESSF Rank Value 1 < >310 ICH Rank Val ue 1 < >460 SBS Rank Value 1 < >460 Manning, Cooper and Associates Ltd. 37

38 Patch Area (hectares) Forest District, ESSF, ICH, and SBS Rank Val ue 1 < >50 Road Density (m/m 2 ) Forest District, ESSF, ICH, and SBS Rank Value 1 > < Manning, Cooper and Associates Ltd. 38

39 Old Interior Forest (% of polygon area) Forest District Rank Value 1 < >0.9 ICH Rank Value 1 < >0.9 ESSF Rank Value 1 < >0.98 SBS Rank Val ue 1 < >0.62 Manning, Cooper and Associates Ltd. 39

40 Number of tree species Forest District Rank Value ,6 ICH Rank Value ESSF Rank Value ,5,6 SBS Rank Value ,6 Manning, Cooper and Associates Ltd. 40

41 Appendix 3. Old forest quality categories mapped at the Forest District and BEC Zone scale for the Prince George Forest District. Darker shades of red indicator higher quality ranks for old forest. MODEL 1. Forest District Scale, All Intrinsic Indicators Old Forest Quality 1 - good best Manning, Cooper and Associates Ltd. 41

42 MODEL 2. BEC Zone Scale, All Intrinsic Indicators Old Forest Quality 1 - good best Manning, Cooper and Associates Ltd. 42

43 MODEL 3. Forest District Scale, All Intrinsic Indicators and three Additive Indicators (Caribou, Red and Blue Listed Species, Ecosystem Representation) Old Forest Quality 1 - good best Manning, Cooper and Associates Ltd. 43

44 MODEL 4. Forest District Scale, Intrinsic Indicators and All Additive Indicators Manning, Cooper and Associates Ltd. 44 Old Forest Quality 1 - good

45 Manning, Cooper and Associates Ltd. 45

46 MODEL 5. BEC Zone Scale, All Intrinsic Indicators and three Additive Indicators (Caribou, Red and Blue Listed Species, Ecosystem Representation) Old Forest Quality 1 - good best Manning, Cooper and Associates Ltd. 46

47 MODEL 6. BEC Zone Scale, Intrinsic Indicators and All Additive Indicators Old Forest Quality 1 - good best Manning, Cooper and Associates Ltd. 47