Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A spatial simulation model for evaluating the response of rare and endangered species to conservation… Demarchi, Donald Andrew 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1998-0218.pdf [ 6.73MB ]
Metadata
JSON: 831-1.0074818.json
JSON-LD: 831-1.0074818-ld.json
RDF/XML (Pretty): 831-1.0074818-rdf.xml
RDF/JSON: 831-1.0074818-rdf.json
Turtle: 831-1.0074818-turtle.txt
N-Triples: 831-1.0074818-rdf-ntriples.txt
Original Record: 831-1.0074818-source.json
Full Text
831-1.0074818-fulltext.txt
Citation
831-1.0074818.ris

Full Text

A Spatial Simulation Model for Evaluating the Response of Rare and Endangered Species to Conservation Strategies and Forest Practices: A Case Study on the Northern Spotted Owl by Donald Andrew Demarchi B.Sc, The University of British Columbia, 1992 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard The University of British Columbia April 1998 © Donald Andrew Demarchi 1998 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available "for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of t^jg>oL-oCh V The University of British Columbia Vancouver, Canada Date gL£> ^ 9 % DE-6 (2/88) Abstract A spatially explicit forest harvesting and individual-based population simulation model has been developed. This model was used to assess the response of rare or endangered forest wildlife to forest harvesting policies and conservation options. A specific focus was on the examining the efficacy of proposed management options for the northern spotted owl population in British Columbia. The forest harvest simulation component was used to construct spatial and temporal patterns of logging activities in the Fraser and Soo timber supply areas. Forest simulations were based on a forest inventory database of 25-ha resolution, and simulated harvests were subject to constraints imposed by the Forest Practices Code and wildlife conservation options. The delineation of breeding territories over this database was done using a submodel that maps likely territory areas following geographic constraints of the landscape, resulting in a "mosaic" of variously shaped territories. This allowed for aggregation of the finer scale forest-state data into simple indices of habitat quality for owls. These indices were updated on a yearly basis as the forest either aged or was harvested. The individual-based population simulator used "coarser" territory/habitat maps to simulate dispersal, location and fates of individual female owls (i.e., breeding pairs). Simulations suggest that a major limiting factor is the low survival rate of dispersing juvenile owls. The proposed management plans do not adequately address this problem and may, in fact, be detrimental to the owl population. These small conservation areas increase logging pressure on the surrounding areas. As a result, juveniles are forced to disperse from their natal habitats into marginal habitats where the risk of mortality is increased. Reduction in the annual allowable cut is more important than any proposed configuration of Spotted Owl Conservation Areas. n TABLE OF CONTENTS Abstract .... ii List of Figures v List of Tables '. . vii Acknowledgments viii 1. INTRODUCTION , : '. 1.1 Background : • ...1 1.2 Population Simulation Models 4 1.3 Individual-Based Models.... 6 1.4 Northern Spotted Owl Range 8 1.5 Northern Spotted Owl Biology '. !.8 1.5.1 Home Range Size • 9 1.5.2 Spotted Owl Habitat Requirements '. 10 1.5.3 Survival, Reproduction and Dispersal 11 1.6 The Trade-Off Hypothesis .': 13 1.7 Purpose • 13 2. METHODS : • .15 2.1 Modelling Approach ; 15 2.1.1 Forest Harvesting Component '. 15 2.1.1.1 Forest Inventory Database • 16 2.1.1.2 Dynamic Programming Calculation of Hauling Routes and Costs .....24 2.1.1.3 Forest Dynamics 25 2.1.1.4 Economics (Cost/Production) 26 2.1.1.5 Harvest Scheduling .27 2.1.2 Spotted Owl Model Component..... ..30 2.1.2.1 Territory Size...:... ". : .....32 2.1.2.2 Proportion of Suitable Habitat Within a Territory 32 2.1.2.3 Territory Formation 33 2.1.2.4 Cross-Scale Linkage 34 2.1.2.5 Juvenile Dispersal '. • 36 2.1.2.6 Biological Parameters 40 2.1.3 Calculation Sequences 42 2.1.3.1 The Forest Harvesting Model Calculation Sequence .42 2.1.3.2 The Spotted Owl Model Calculation Sequence : 43 2.2 Population Simulations 43 iii 2.2.1 Historical Forest Disturbance and the Trade-off Hypothesis... ; 44 2.2.2 AAC Levels, Conservation Options and the Trade-Off Hypothesis 46 3. RESULTS , 52 3.1 Territory Mosaic : 52 3.2 Juvenile Dispersal Patterns 55 3.3 History and the Trade-Off Hypothesis :55 3.4 Conservation Options : 59 4. DISCUSSION ....69 4.1 Population Responses to Historical Harvesting and Disturbance 69 4.2 Implications for Future Conservation Strategies 72 4.3 Problems with the Proposed Conservation Strategies 72 4.4 Problems with Academic Institution and Government Partnerships......... 78 5. CONCLUSIONS....... '. , 80 5.1 The Future of the Spotted Owl 80 5.2 The Future of the Model 81 Literature Cited • • 83 Appendix I - Conceptual structure of the model.... 90 List of Figures Figure 1 Location of the Fraser and Soo TSAs in British Columbia. These two TSAs contain almost the entire distribution of spotted owls in BC. Most of the Special Resource Management Zones (SRMZ) have been delineated over high priority spotted owl habitat 3 Figure 2 Current age of forested areas in the Fraser and Soo TSAs '. 18 Figure 3 Site productivity indices used for calculating potential forest yield 19 Figure 4 Elevation contour map. Elevation is a key parameter used to simulate hauling costs and direction .20 Figure 5 Forest management type used to simulate adjacency restrictions (e.g., green-up requirements). ; 21 Figure 6 Location of current parks and reserves in the Fraser and Soo TSAs. No simulated harvesting occurred in these areas 22 Figure 7 Current road access in the Fraser and Soo TSA's. This spatial information is used to calculate the costs associated with building roads to harvestable timber stands and hauling wood to the nearest mill 23 Figure 8 Net volume of wood in m3-ha"' produced by a stand based on its age and site productivity index. The volume is based on the Ministry of Forest's Variable Density Yield Prediction Program (VDYP v. 6.3) 28 Figure 9 Example of 15 Monte Carlo simulations of the owl population. These simulations exhibit the variation and uncertainty of the path that the population size takes. When 100 simulations are complete then an average number of owls per year, plus a standard deviation, can be calculated .31 Figure 10 Juvenile survival per dispersal step in relation to the proportion of suitable habitat (i.e., forest > 120 years old) the juvenile encounters while dispersing. This represents increasing juvenile mortality risk (e.g., predation, starvation) as the proportion of suitable habitat encountered declines. The curves represent the range of values of survivorship that can be set within the model. The "maximum" line, which can be set to any level, limits the maximum juvenile survival per dispersal step to that level (i.e., any survival probabilities above this level are reset back to this level). If the user wishes to remove the predation risk then a constant survival probability across all suitable habitat proportions can be set by setting the curve value to a level of 1500 or greater (i.e., the curve value always exceeds the maximum, no matter what the proportion of suitable habitat encountered, and is therefore reset to the "maximum" level). 38 Figure 11 Annual adult survivorship and juvenile fledging rate in relation to the proportion of mature and old-growth forest found within a breeding territory. The sections of the curves where the slopes are zero indicate the maximum levels used in the model 41 Figure 12 Area of forest disturbed (i.e., undocumented logging and development) and the area of forest logged in the Fraser and Soo TSAs from 1896 to 1995 (based on current forest cover data and the map files used within the model) 45 Figure 13 Annual Allowable Cut levels used to assess the efficacy of each conservation strategy at recovering the spotted owl population in BC 49 Figure 14 The Option C SOC A configuration (one of three options used in the model to assess the response of the spotted owl). The blue areas are proposed SOCAs. The green areas are comprised of parks present at the time the SORT Management Options Report (SORT 1994) was published. There is no harvesting in the parks or SOCAs. The second and third management options modelled are current existing parks only (Figure 6) and the current management plan (Figure 1). The current management plan is a combination of the Option C policy and current park system. This plan includes all parks plus 67% retention of forest >100 years old in all SOCAs (see the Special Resource Management Zones in Figure 1) .50 Figure 15 Owl territory "mosaic" laid out over the Fraser and Soo Timber Supply Areas (Figure 1). The land is delineated into potential owl territories for which suitabilities are contingent on forest conditions. Note the small territories that effectively remove a significant portion of the land base from use by breeding owl pairs. (Colours provide contrast between individual territories) ..; .53 Figure 16 Cross-scale territory map used to visualize of the temporal and spatial patterns of habitat change, female owl locations, and juvenile dispersal. 54 Figure 17 Typical dispersal patterns exhibited by simulated juvenile female owls. Black lines indicate the dispersal path and the red boxes indicate that the bird has located a potential breeding territory. Black lines without a red box attached represent unsuccessful dispersal attempts. The map on which the dispersal patterns are shown is the territory scale representation of the Fraser and Soo TSAs (see Figure 16)..... 56 Figure 18 Ending owl population sizes for historical disturbance regime simulations that varied juvenile dispersal survivorship and relative habitat quality (elevations > 1200 m were varied in quality relative to lower elevation sites). See section 2.1.2.5 for an explanation of the "curve index" .57 Figure 19 Female owl population responses to the historical disturbance regime reconstruction of the Fraser and Soo TSAs for the two extreme parameter combinations leading to 100 owl pairs (± 1 standard deviation, n-l00 simulations). Initial increase is a result of "seeding" the starting population at a level lower than was likely to have existed 100 years ago. 60 Figure 20 Owl population simulation results from the combinations of MELP and MOF AAC predictions and conservation options. The trendlines indicate yearly population level averages (n=100 simulations) and the error bars represent ± 1 standard error. 66 Figure 21 Simulated responses of the spotted owl to the immediate cessation of harvesting in the Fraser and Soo TSAs and to the maintenance of a high harvest (i.e., 2.335xl06 m3/year) under Option C conservation strategy. The average population levels per year of 100 simulations are plotted at each year (error = ± 1 standard deviation). It can be expected that the response of the owl to the current management plan will fall somewhere in between these two population trajectories. 67 vi List of Tables Table 1 Conservation strategy, forest harvesting level and trade-off value combinations tested with the model for effectiveness at recovering the northern spotted owl population in BC. 47 Table 2 Spotted owl population size at the end of the 100 year simulations for all the management options modelled (mean ± 1 standard deviation)... ....61 Table 3 Mean time-to-recovery in years. Each conservation strategy's mean time (standard deviation in parentheses) for the spotted owl population to recover to a level of 125 pairs. Mean time-to-recovery is based only on the simulation trials that reach 125 pairs (i.e., n = number of simulations out of 100 that reached 125 pairs). High/Low - high-elevation quality = 100% relative to low-elevation habitat and juvenile survival curve index = 17.5 Low/High - high-elevation quality = 67% relative to low-elevation habitat and juvenile survival curve index = 54. 62 Table 4 Results table from the ANOVA testing for the effects that AAC level, Conservation Strategy, Trade-off values, and their interactions have on the owl population's mean time-to-recovery.f = significant (p < 0.05) 63 Table 5 Results table from the ANOVA testing for the effects that AAC level, Conservation Strategy, Trade-off values, and their interactions have on the owl's mean population size at the termination of simulation (i.e., 100 years from the present). f significant (p < 0.05). : .'. .... 64 vii Acknowledgments This thesis would not have been possible without the financial support of the Forest Renewal BC Program, the Science Council of BC, and International Forest Products Ltd. I truly appreciate the teachings and guidance of my supervisor, Dr. Carl Walters. I am also grateful for the invaluable instruction I have received from my committee members, Dr. Fred Bunnell, Dr. Lee Gass, Dr. Tony Sinclair, and Dr. Jamie Smith. I am especially grateful to Bill Rosenburg, International Forest Products Ltd., who arranged my initial financial support. I also wish to thank Glen Loveng, Timberline Forest Inventory Consultants, who provided the forest database on which this model relies, Jim Goudie, who provided valuable technical assistance with the Ministry of Forests' Variable Density Yield Prediction Program, and Leonardo Huato who provided much of the modelling code base which launched the current developments. Early manuscripts of this thesis were edited by Carol Hartwig, Mike Demarchi and Ray Demarchi; their knowledge and contributions have substantially improved this thesis. Special thanks to Lisa Beaulac, Marilyn Oliver and Gerry Oliver for their encouragement. viii 1. INTRODUCTION 1.1 Background In the early 1980s it was thought that there were no longer any northern spotted owls (Strix occidentalis caurind) in British Columbia, because none were recorded between the mid 1970s and 1985 (Campbell and Campbell 1986). Then the Ministry of Environment, Lands and Parks (MELP) began inventory surveys over the owl's1 historic range with the intention of determining the status of the species in Canada. Since that time at least 52 spotted owl activity centers2 have been detected (SORT 1994, D. Dunbar pers. comm., I. Blackburn pers. comm.). There have even been two recent occurrences of spotted owls in the Lillooet Forest District (MELP 1997), outside their previously described historic range. Public concerns for biodiversity, particularly rare or endangered species like the spotted owl, are dramatically altering natural resource management. Most people in BC and the Pacific Northwest are familiar with the controversy over the northern spotted owl and harvesting old-growth forest. The conflict exists because both require large areas of mature and old-growth coniferous forest. The controversy has become a watershed that has changed resource and environmental policy in the United States, and Canada. It represents a change from the management of local species to management on a large-scale landscape or ecosystem basis (Yaffee 1994). In this perspective, conservation strategies must balance biological feasibility with the economic and social needs of society. 1 The confirmed historic range of the northern spotted owl in Canada occurs exclusively in BC. Specifically, north to Whistler, south to the international border, west to the Squamish River and east to Spuzzum (SORT 1994). 2 A spotted owl activity center refers to an area where the presence of a single owl, or pair of owls, has been detected. 1 Conservation measures proposed for the spotted owl (SORT 1994) could have major impacts on the forest industry in southwestern British Columbia. These impacts include major changes to BC's forest practices and reductions in the timber supply for the Fraser and Soo timber supply areas (TSAs) (MOF/MELP 1995) in the Vancouver Forest Region (Figure 1). The potential economic costs of conserving the spotted owl are very high. Therefore, it is critical to develop tools that incorporate the best available knowledge of ecological processes, to make the best possible assessments of policy alternatives before they are implemented. Is the spotted owl population declining in BC? Evidence from the US suggests that there is a minimum total population of 3500 pairs, and numbers are declining (USDI 1992, Thomas et al. 1990, Forsman et al. 1984). Although the historic and current range of the spotted owl in BC are similar (SORT 1994) it is widely assumed, based on habitat alteration over the last 100 years, that the population is declining (SORT 1994; Dunbar et al. 1991; Campbell and Campbell 1986). Even though there is, however, no direct evidence to support this conclusion (SORT 1994). Regardless, the British Columbia Spotted Owl Recovery Team (SORT 1994) has developed several policy alternatives that range from "doing nothing" (i.e., continue forest harvesting based on Ministry of Forests (MOF) projections) to stopping all forest harvesting in the owl's historic range (i.e., the Fraser and Soo TSAs (Figure 1)). These alternatives have been ranked by their "... likelihood that [they] would result in a particular COSEWIC [Committee on the Status of Endangered Wildlife in Canada] status category...in the future." A "Biological Assessment Team"3 were asked the following question: "Given the likely condition of habitat under a specific [management] option, 3 The Biological Assessment Team was comprised of 21 professional biologists and foresters from resource management agencies, the academic community and the forest industry (SORT 1994) 2 Figure 1 Location of the Fraser and Soo TSAs in British Columbia. These two TSAs contain almost the entire distribution of spotted owls in BC. Most of the Special Resource Management Zones (SRMZ) have been delineated over high priority spotted owl habitat. 3 what will be the [COSEWIC] status of the spotted owl in 100 years?" The ranking procedure consisted of votes cast by the Biological Assessment Team. SORT then published "Management Options for the Northern Spotted Owl in British Columbia" (SORT 1994). That report concluded (on the basis of the ranking procedure and not on biological data or research) that continuation of current forest practices would be the worst possible alternative for the spotted owl — even with impending reductions to the annual allowable cut (AAC) and implementation of the Forest Practices Code (FPC). The aim of my research was to provide more objective and quantitative tools for comparing policy options for the spotted owl and other rare species. I developed an individual-based population dynamics model for the owl using available information on recruitment rates, survival rates, and'dispersal patterns. I linked this population model to a detailed spatial simulation model for forest harvesting, to represent as accurately as possible the forest landscape changes that the owl has faced in the past and will face in the future. Using this model, I compared a variety of policy options for future forest management aimed at protecting and conserving the species. 1.2 Population Simulation Models . * ' Computer simulation models are one way to make predictions regarding large and complex systems. Many such models have been developed for the spotted owl and its habitat (Bart 1995a; McKelvey et al. 1992; Lamberson et al. 1994; Lamberson et al. 1992; Lahaye et al. 1994; Carroll et al. 1995; Doak 1989; Lande 1987 and 1988; Andersen and Mahato 1995; Holthausen et al. 1995). All these models attempt to estimate future components of spotted owl sub-populations (e.g., local extinction risk, persistence time, population trends). 4 However, most of those models do not account for changing forest harvesting practices and conservation strategies. The models that do simulate changes in habitat do so on very coarse spatial scales that oversimplify forest landscape processes. Those models either: 1) employ difference equations4 based on specific population rates (e.g., birth, death and immigration/emigration rates); 2) aggregate habitat state data across the range of a population then apply this aggregation to the population as a whole (i.e., without spatial variation in habitat quality); or 3) use a combination of these two approaches. A major limitation of past habitat models is their coarse spatial scales and the resulting inclusion of habitat not normally used by owls (i.e., the habitat quality indices include parts of the landscape, including non-forested land such as alpine and water, that would not normally be used by spotted owls). In reality, it is nearly impossible to calculate accurate indices of habitat quality at scales coarser than typical forest stand polygon scales (i.e., between 10 - 100 hectares). Existing models that represent owl territories or discrete habitats do so as symmetrical shapes (e.g., squares, hexagons) (Holthausen etal. 1995; Harrison et al. 1993; Lamberson et al. 1992). Animals do hot choose their habitat in this way, and pretending that they do can lead to problems. First, subdividing the landscape up into small symmetric shapes increases the risk of including habitat that is not normally used by, or is inaccessible to, the animals (e.g., non-forested land, territories that cover movement barriers). This is similar to the problem of using coarse spatial scales. Second, polygonal patterns allow more animals to fit into the landscape than is actually possible, resulting in an overestimate of carrying capacity. 4 Difference equations are used to simplify processes that change smoothly over time (i.e., differential equations) by employing discrete time intervals. In ecology, changes from year-to-year or season-to-season are often more important than changes on a continuum. 5 This overestimate results from the systematic placement of territories on the landscape. That is, the total area of the population's range divided by the size of an average territory equals the maximum number of owls that could exist if each territory was occupied. This does not allow for variation in the shape of habitats due to geographic features and constraints. The fundamental weakness of the habitat models is their inability to realistically represent changes to the habitat base. The goal of those models was to assess the effects that changes in habitat have on specific sub-populations of the northern spotted, owl. However, habitat change in those models was either random (e.g., Lamberson et al. 1992) or abrupt with new habitat maps being inserted at predetermined intervals (e.g., Holthausen et al. 1995). The most realistic model in the literature (Bart 1995a) based changes in habitat on past logging history and predicted future trends from management proposals in the US's spotted owl recovery plan (Bart et al. 1993). However, other components of this model are unrealistic (e.g., dispersing juvenile birds "jump" from their natal habitat to vacant territories without being exposed to the risks of dispersing through the habitats between them). 1.3 Individual-Based Models Population simulation methods typically incorporate difference equations based on population size. This requires the assumption that the single state variable representing population size can be determined by aggregating individual members of the population. It has been claimed that this necessarily assumes that each individual is not unique and the models average over the full distribution of individual types in the population (DeAngelis and Gross 1992). It may be more correct, however, to state that population modelers recognize that individuals are unique, but they can be described by statistical parameter 6 averages from the entire population. DeAngelis and Gross (1992) also pointed out that models without spatial-temporal dependence assume that the environment and every member of the population affects every other member of the population equally. These assumptions violate the tenets that every individual is unique in its physiology and behaviour, and that a given individual is primarily affected by other organisms in its spatial-temporal neighbourhood (DeAngelis and Gross 1992). One response to this perceived inadequacy has been the development of models that focus on age classification or size-class structure and dynamics. A second approach, that avoids aggregation altogether, describes populations by simulating large numbers of individuals simultaneously. For example a model can imitate the movements and fates of individual animals based on the habitat that they encounter. Both of these alternative approaches have been referred to as individual-based modelling but have been distinguished as individual state (i-state) distribution and i-state configuration models, respectively (Metz and Diekmann 1986 cited in DeAngelis and Gross 1992). The use of individual-based models to make or defend management decisions is becoming increasingly common in conservation.biology. All models have limitations, but they can still be useful tools for developing a deeper understanding of the biology of a species, especially individual-based models. Individual-based models represent a reductionist application of ecology (L0rnnicki 1992). Reductionism implies that the properties of a system can be derived from its individual components. As a result, individual-based models allow a greater understanding of population behaviour and underlying mechanisms (Botsford 1992; L0mnicki 1992). 7 However, Botsford (1992) cautions that these models must be realistic and that their structure be comparable to the population itself, through direct observation. 1.4 Northern Spotted Owl Range The northern spotted owl is resident along the Pacific coast from northern California to the extreme southwest portion of BC (Campbell et al. 1990). In the United States, they range east from the Pacific Ocean to the Palouse Prairie in Washington and the Great Basin shrub steppe in Oregon and California. In British Columbia, spotted owls have been detected west from the Cascade Mountain range to the Pacific Ocean and north from the international boundary to Birkenhead provincial park. In BC, spotted owls are generally found from sea level to 1400 m in elevation and occur almost exclusively in the Southern Pacific Ranges and Eastern Pacific Ranges Ecosections of the Pacific, and Cascade Ranges Ecoregion (Demarchi et al. 1990). Within these Ecosections, the owls are found predominantly in the Coastal Western Hemlock (CWH) arid Interior Douglas-fir (IDF) Biogeoclimatic Zones5. 1.5 Northern Spotted Owl Biology Information on owl biology used in this model was based primarily on research from the United States. Although the owl has been top priority for inventory in BC, little 5 The Biogeoclimatic Ecosystem Classification system (Krajina 1959, Pojar et al. 1987, British Columbia Ministry of Forests 1988) describes the plant and animal communities which reflect the climatic processes and landform parameters bounded within a terrestrial region 8 information has been gathered on specific demographic attributes. This may lead to management problems because owl biology in the US may differ from that in BC. Until detailed information is available for the species in BC, the model must assume that its biology does not differ from that in the US. Where possible, information obtained from northwestern Washington state has been used6. 1.5.1 Home Range Size The spotted owl is a territorial bird with obligate juvenile dispersal7 (Lamberson et al. 1994). Hanson et al. (1993) summarized home range data from radio-telemetry studies of spotted owls in Washington State. In the Western Cascade Physiographic Province of northwestern Washington, the owl exhibited wide variation in territory size (mean = 4198 ha. ± 2155 ha (95% CI; p = 0.05, n=7); range: 1302-7258 ha). Territory overlap within a breeding pair is high (75%) and territory overlap among breeding pairs is typically low, but may be as high as 28% in fragmented Douglas fir forests with low prey biomass (Carey and Peeler 1995). Prey biomass is inversely proportional to territory size (Carey et al. 1992). In BC, nearly 50% of the spotted owl's diet consists of northern flying squirrels (Glaucomys sabrinus) compared to 25% in the western Cascade region of Washington State (SORT 1994). Owl territory sizes generally increase with latitude (Thomas et al. 1990), and tend to be largest where owls have one primary prey source (Carey et al. 1992) and where flying squirrels dominate the diet (Zabel et al. 1995). The territory sizes of spotted owls in BC 6 The centres of abundance between northern Washington state and BC are about 200 km and 1° 45' of latitude apart. 7 An obligate disperser is a species that, for one reason or another (e.g:, defense of breeding territories or food resources), forces juveniles from their natal habitat to locate unoccupied habitat. should thus be as large or larger than those found in Northern Washington. 1.5.2 Spotted Owl Habitat Requirements Sufficient suitable habitat within the home range of a breeding pair must exist such that they can reproduce successfully. Limited information is available in this area and is generally focused on nest site and roost tree characteristics and on general descriptions of stand attributes (e.g., uneven aged, large amounts of coarse woody debris, high snag density) (USDI 1992). Because of this lack of specific information, the area of suitable owl habitat within a territory is generally based on forest age and the proportion of time owls spend in each age type relative to its availability within the territory (SORT 1994; Hanson et al. 1993). Spotted owls tend to nest and roost in forest stands with high vertical canopy layering (Mills etal. 1993). In the recent past it has been assumed that each owl territory must be composed of between 44-49% mature or old-growth forest (SORT 1994). In this context, old-growth and mature includes all forest that is at least 120 years old and exhibits stand characteristics similar to those of true old-growth (i.e., age class 9+ or 180+ years). Hanson et al. (1993) calculated the amount of suitable habitat found within each home range in the northwestern Cascades. They found that each territory contained an average of 1807 ha ± 1011 ha (95% CI; p = 0.05; n=7; range:694-4024) of suitable habitat. This translates into an average of 43% forest greater than 120 years of age in each home range. This level is close to 44-49% range, but it is obvious from the range of values that there is considerable variation. In that data set the minimum amount of mature/old-growth found within a territory was 20.6%. 10 1.5.3 Survival, Reproduction and Dispersal Barrowclough and Coats (1985) estimated yearly adult survival at 0.85 and the draft United States Forest Service's Environmental Impact Statement (EIS) (USFS 1986) used the same value. However, the final EIS (USFS 1988) reported annual adult survivorships of 0.96 and 0.97. Marcot and Holthausen (1987), Lande (1988), and Noon and Biles (1990) reported similarly high annual adult survival rates of 0.96, 0.942, and 0.942 respectively. Bart (1995b) re-analyzed the adult survival data from 102 owl sites in Oregon (Bart et al. 1992). He found that adult survivorship varied linearly with the proportion of suitable habitat found within each territory. Although the data do not appear to exhibit a threshold, above which there is little or no increase in survivorship, the maximum annual adult survival rate given by Bart (1995b) is approximately 0.90. The following equation (Bart 1995b) describes this relationship: S a =0.63 + 0.39/? where: S a = annual adult survivorship p - proportion of suitable habitat within the territory Clutch size ranges from 1 to 4 eggs, but most are <2 (SORT 1994). Bart (1995b) also re-analyzed the productivity data from Bart and Forsman (1990). This data showed that fledging rate also varied linearly with the proportion of suitable habitat found within each territory. Similarly, Ripple et al. (1997) reported preliminary results that suggest that reproductive success is related to the proportion of old coniferous forest in the landscape. 11 The following equation (Bart 1995b) describes the relationship between fledging success and habitat quality: ' / = 0.32 + 0.54/7 where: / = fledging rate of juvenile owls (owls/clutch) p = proportion of suitable habitat within the territory Juvenile spotted owl dispersal and survivorship have been the subject of intensive study in the United States. Dispersal begins in the fall, where fledged juveniles leave their natal habitats, searching for unoccupied suitable habitat in which they can establish breeding territories. The initial direction of dispersal is random and the distance traveled during the first year ranges between 14 km and 48 km (Gutierrez et al. 1985, Miller 1989 cited in SORT 1994). Juvenile females tend to disperse 12.5 km further, on average, than juvenile males (SORT 1994). This is likely an adaptive behaviour ensuring that siblings locate separate' breeding territories which prevents the potential genetic problems associated with inbreeding. Miller et al. (1997) found that juvenile owls did not exhibit selection among three levels of forest fragmentation during dispersal. However, Miller et al. (1997) found that the probability of mortality increased with increased use of clear-cut areas during dispersal. This increase in mortality has been attributed to a greater vulnerability to avian predators (Dawson et al. 1987) and to decreased foraging activity resulting in starvation (Miller 1989 cited in Miller et al. 1997). Forest fragmentation also leads to increased mortality because juvenile owls are required to disperse greater distances during a period of high vulnerability (Gutierrez et al. 1985). Reported survival rates of spotted owls during the first year.of life range from 0.11 to 0.34 (Noon and Biles 1990). 12 1.6 The Trade-Off Hypothesis Earlier owl population simulations, research and discussions with MELP personnel have led to the formation of a "trade-off hypothesis. The MELP contends that high-elevation habitat (i.e., habitat above 1200 m) is not capable of supporting spotted owls at the same densities as habitat below this elevation (e:g., high-elevation sites produce lower prey biomass than lower elevation sites and appropriate nest sites are lacking). The hypothesis states that if high-elevation habitat is inferior in quality to low-elevation habitat then there must be a corresponding increase in some biological parameter to counteract this "loss" of habitat from the land base. The trade-off is realized when the model is used deterministically to predict the current population size from past changes to the forested landscape. That is, if the quality of high-elevation habitat is decreased relative to low-elevation habitat then there must necessarily be an increase in some other biological parameter in order to achieve current population levels. This also allows for "tuning" of parameter estimates that are not very well understood against parameters that are much better understood. 1.7 Purpose The forest harvesting and individual-based spotted owl population simulation model that was developed for this thesis was a response to a MELP request. It was also a response to concerns expressed by the forest industry about the validity of the proposed conservation strategies and the necessity of implementing a conservation strategy over and above the FPC requirements for identified wildlife8. There was a clear need for a more accurate approach to 8 Identified wildlife refers to species that the Deputy Minister of Environment, Lands and Parks and the Chief Forester agree will receive special management attention (Forest Practices Code Managing Identified Wildlife Guidebook). The northern spotted owl is an identified wildlife species. 13 accounting for changes to the owl's habitat. The research on which this project is based has been aimed at developing simulation models to predict the long-term impacts of forest management practices on rare animal species in general. The purpose of this study was to: 1) develop an accurate simulation model that could be used to assess the long-range implications of forest management and conservation strategies for rare or endangered species; and 2) use the model to assess the implications of forest harvesting and the validity of the conservation strategies described by SORT (1994) for the northern spotted owl in BC: The model described in this thesis, which consists of a spatially explicit forest harvest simulator and an individual-based spotted owl population model, was also developed to allow resource managers to analyze forest harvest policy alternatives by performing an unrestricted number of "virtual experiments". These experiments, which are far too large to perform manually, can test the relative effect of certain forest harvesting practices and conservation options on the population dynamics of the spotted owls in reference to the population's recovery (i.e., the likelihood that the COSEWIC status of the spotted owl will be improved from its current "endangered" status). Is 14 2. METHODS 2.1 Modelling Approach The following description includes details of the model's structure, organization and the sequence of steps used in making specific predictions. The model parameters used and the assumptions made are discussed where applicable. The model consists of two main programs: 1) a forest harvesting component, where the spatial consequences of harvest scheduling are simulated; and 2) an individual-based spotted owl model, where simulations are run to assess the effects of different scenarios of forest harvesting and conservation plans on spotted owls. 2.1.1 Forest Harvesting Component The forest harvesting component of this model is temporal and spatial in nature. The purpose of spatial modelling is to: 1) exhibit site dynamics; 2) allow for a linkage structure (i.e., adjacency); 3) represent patches (i.e., aggregate spatial state data); and 4) store model-specific attributes in an array that can be changed if necessary. This component is the first step in assessing forestry impacts on spotted owls. Using current forest inventory data for the Fraser and Soo TSAs, the model mimics future harvesting based on a number of criteria. Harvesting simulations and associated changes in owl habitat availability can then be used by the model to assess the relative response of the owl population over time. The most important aspect of the forest harvesting model is that simulated changes, that occur spatially and temporally on the forested landbase, reflect reality to the accuracy level of the provincial forest inventory data. 15 2.1.1.1 Forest Inventory Database Early versions of the model (Walters and Huato, UBC, pers. comm.) lacked precision with respect to the spatial aspect of forest inventory data. Harvest scheduling was limited to clearcut logging on a 2500 ha basis. This was clearly an unreasonable representation of actual harvesting, so changes were made to represent the forest using a spatial resolution of 25 ha. This allowed greater flexibility in harvest practices over earlier versions. For example, this permits simulations of forest harvest and conservation strategies at scales similar to the landscape scales of actual clear-cut logging practices and management restrictions such as protection of wintering areas for ungulates. Simulation of forest harvesting policies by the model requires a number of parameters. These are: 1) latitude of the forested cell; 2) longitude of the forested cell; 3) current road access ; 4) forest management type (e.g., operable forest, retention areas, partial retention areas, black-tailed deer (Odocoileus hemionus columbianus) winter range, inoperable forest); 5) site productivity indices; 6) elevation; 7) forest age; 8) reserve status (i.e., parks and conservancy areas); 9) known spotted owl activity centres; and 10) potential log hauling routes. Information on most of these parameters was compiled at 100-ha cell resolution by Timberline Forest Inventory Consultants, Ltd. of Vancouver, BC, from current forest cover data. Information on the forest age of each 100-ha cell was compiled at a 25-ha resolution. Remotely sensed data were used to estimate missing parameters for forested. areas without forest cover data (e.g., provincial parks, Greater Vancouver Regional District watersheds). 16 The data were acquired by overlaying a grid system of 1km by 1km (100 ha) squares over both the Fraser and Soo TSAs and calculating either averages, area-weighted averages or absolute areas, depending on the precision required, for the various forest and landscape attributes. The outcome of these calculations was compiled and stored as an array in an ASCII file. The total map area of the Fraser and Soo TSAs is 23988 km2 and of this approximately 14460 km2 are forested or potentially forested land. Forest area by age was compiled for each 100-ha spatial cell which allowed for detailed age data at the 25-ha scale (Figure 2). Site productivity indices (Figure 3) were calculated using an area-weighted average for forests found within each 100-ha grid cell; elevation (Figure 4) was taken to be the average within each 100-ha cell; the management type (Figure 5) applied to each cell was taken to be the most dominant type; a cell was considered to be a park or reserve (Figure 6) if more than 50% of the cell overlaid park areas (Note: Spotted Owl Conservation Areas (SOCAs), as outlined in the SORT Management Options Report (SORT 1994), were added to the parks and reserves attribute only when assessing the spotted owl population's response to each conservation strategy); the general locations9 of known spotted owl pairs were obtained from the Surrey office of the MELP and were added to the database accordingly; road access (Figure 7) for each cell was determined by the presence or absence of primary and secondary roads (i.e., paved roads and logging roads); and optimum log hauling routes are determined by the model's "dynamic programming calculation of hauling routes and costs" routine (see section 2.1.1.2). The structure of the forest inventory array file, referred to as a map file, consists of an x- and y-coordinate plus the eight other attributes described above for each 25-ha grid cell. 9 Specific locations were classified confidential by the MELP, Surrey, and were not made available for this study. 17 Figure 2 Current age of forested areas in the Fraser and Soo TSAs. Figure 3 Site productivity indices used for calculating potential forest yield. Figure 4 Elevation contour map. Elevation is a key parameter used to simulate hauling costs and direction. 20 Figure 5 Forest management type used to simulate adjacency restrictions (e.g., green-up requirements). 21 Figure 6 Location of current parks and reserves in the Fraser and Soo TSAs. No simulated harvesting occurs in these areas. 22 Figure 7 Current road access in the Fraser and Soo TSA's. This spatial information is used to calculate the costs associated with building roads to harvestable timber stands and hauling wood to the nearest mill. 23 The information within this database can be manipulated through a user-friendly interface that allows manual changes. This enables the user to define multiple conservation strategies and to change the attribute values of individual cells. These changes can be used immediately or can be saved to a new map file. This manipulation of data can occur at anytime throughout the forest harvesting simulation allowing for predicted policy changes (e.g., new parks, newly operable forest). 2.1.1.2 Dynamic Programming Calculation of Hauling Routes and Costs Minimizing costs is an essential part of any logging company's plans. By finding optimal log hauling routes from each forest cell to the nearest mill, the model can locate and cut the timber in that cell based not only on its stand characteristics, but on whether or not it is economical to harvest. The procedure is relatively simple in concept. Starting out with known mill locations, the model recursively finds, for each cell, the cheapest way to build road access to and haul wood from that cell to a mill. This process continues until all the forest cells in the database are directly linked to a mill. When a cell without a direction attributed to it goes through this process, the minimum cost hauling directions for all surrounding cells are checked. The selection of connections without regard to geographical contours or mill location could result in hauling routes that are illogical (e.g., over top of high-elevation passes, long and/or adverse hauling routes). When adjacent cells with hauling directions applied to them are located the new cell is linked to the cell that has the lowest cost-to-mill (i.e., to minimize road building and transport costs). Road building costs are linked to elevation changes, so that optimal haul routes follow elevational contours that minimize road-building costs associated with elevational changes. Once the dynamic 24 program has attributed a hauling direction to all forest cells, the linkage structure is stored as the last of ten parameters found in the map file. This eliminates the need to run the dynamic program each time a map-file is loaded into the model (except to update the road linkage each time a previously unroaded cell is "roaded"), thereby improving the speed of the simulations. 2.1.1.3 Forest Dynamics Forest dynamics are not simple to model. Interdependent processes occur at many space-time scales. A general approach to this problem was chosen by addressing two large-scale processes, forest harvesting and fire generation. In the model, forest harvesting is represented mainly through clearcut logging at the 25-ha. scale. At this scale, each forest stand is treated as homogenous in age and species type. Volume-age tables (provided by the MOF's Variable Density Yield Prediction program (BCMOF 1995)) are used to represent the net effects of growth and natural mortality as each stand (cell) ages after cutting or burning. Stand age can be reset to zero to account for immediate silvicultural practices or to an age less than zero to account for regeneration delays (succession). To represent partial cutting, the model resets the age of the harvested cell to a lower value, to reflect the associated reduction in mean age and volume. The wood volume removed from a partially cut cell is calculated as the volume at the current age minus the volume from the reset age. At present, the model does not address harvest practices that are intended to mimic or accelerate development of old-growth characteristics. The partial-cut strategy creates a problem when the results of a forest harvest simulation are passed to the owl model. If, in reality, a forested cell is partially cut and still retains old-growth characteristics, then spotted owls are likely to use it. However, the model does not accurately 25 mimic this by simply resetting forest age to account for volume reduction. A cell is unavailable for owl use if its age is reset to less than 120 years. Future versions of the model may be to simulate the owl population response to stand characteristics other than age alone (see section 1.5.2). Generation of simulated forest fires relies on a user input of the probability of ignition and a minimum age of forest that will burn. This probability is used to trigger one or more fires each simulated year. Once a list of fire ignition cells is generated, each potential fire is spread by assigning adjacent cells a probability that they will catch fire and burn. For each cell burned, stand age is reset to zero with the assumption that the fire returns the forest to an early successional stage. The result is irregular patches of variously sized burns. The locations of fires are not constant from simulation to simulation due to the stochastic10 patterns of fire ignition. 2.1.1.4 Economics (Cost/Production) The cost and production values (in dollars) that result from harvesting are predicted with a few key parameters. Using results from a road and haul routing and cost calculation routine (see section 2.1.1.2), each forested cell has costs associated with harvesting it and hauling the logs to the nearest mill. Once a forested cell is chosen for logging, these costs plus cutting costs and net production value (price/m3 x volume in m3) are calculated. The cost of harvesting a cell is then the cost/km of building a road to that cell from the nearest 1 0 Stochastic refers to the involvement of random probabilities associated with the ignition of forested cells and to the variability, from simulation to simulation, due to that randomness. 26 road, plus the cost of cutting the timber, plus basic silviculture cost, plus the cost of hauling wood back to the nearest mill location (cost/km x distance). Production value is the net volume of wood obtained during one simulated year of harvest multiplied by the mean price/m3 of wood. During a simulation, the total area logged and the cost and production totals for all cells harvested within one year, are plotted against the simulation year for visual interpretation of patterns such as "fall-down" and increasing cost as less accessible stands are taken. 2.1.1.5 Harvest Scheduling Stand selection can be simulated using three alternative scheduling rules. The first is to log the oldest (i.e., old-growth) areas first. The second is to harvest cells that produce the maximum net volume each year. The third is to harvest cells that produce the maximum net value (i.e., maximize production minus cost). For each year, all cells/stands are ranked by the selected criterion, and top ranking cells are harvested until the A A C for that year is reached (each year the model calculates the volume, cost and production for each cell in the map then it continues to harvest cells, based on the chosen stand-selection criteria, until the A A C is reached). The net wood volume harvested from each cell is obtained from a volume-age table that lists net volumes by age and site index (Figure 8). This array was compiled using the MOF's Variable Density Yield Prediction Program (VDYP v.6.3b (BCMOF 1995)) with the following assumptions: 1) most forested areas in the model region are comprised of Douglas fir (Pseudptsuga menziesii); 2) average crown closure was 61%; 3) the forest inventory zone was C; 4) the average site index was between 22 and 25; 5) PSYU special cruise number (for waste and breakage) was 156; and 6) utilization level 1 was 17.5 cm. 27 Forest Age ure 8 Net volume of wood in m3-ha"' produced by a stand based on its age and site productivity index. The volume is based on the Ministry of Forest's Variable Density Yield Prediction Program (VDYP v. 6.3). 28 these assumptions were made in order to realize the observed average net-cut volume of between 600-700 mVha (J. Goudie, pers. comm.). As mentioned above there are five different forest management types (Figure 5). Each cell can be classified as either: 1) operable forest; 2) retention visual quality objective (VQO) areas; 3) partial retention VQO areas; 4) Glass A black-tailed deer winter range; • 5) inoperable forest; and * 6) permanent reserves (parks). Each of these types (except inoperable and permanent reserves) can be individually set for clearcutting or partial cutting. Also, some areas require extended green-up periods and/or minimized visual impacts (e.g., retention areas). For these areas the model can set "temporary reserve status" to adjacent cells, within a user-specified radius (e.g., 1 or more cells), until the green-up period is reached. The same system for creating temporary reserves applies to permanent ones (i.e., parks). Setting up temporary reserve status consists of applying a "flag" representing the number of years the cell is to be held in reserve. This "flag" is then counted down each year until the value reaches zero and the cell is again available for cutting. To create a permanent reserve, this "flag" is simply set to a point where it will never reach zero during a simulation (e.g., 1000). At anytime during a harvest simulation the model can be paused for the user to change the values of particular parameters. By displaying a "Forest Harvesting Parameter Form" in the model, the user can dynamically manipulate such factors as Cut Selection Criteria, A A C level, green-up requirements and economic variables. Simulations of allowable cut, cell selection criterion, and management type by cell create a rich array of alternative harvesting strategies and local tactics. Except as noted 29 below, for simulations of owl responses reported here the AAC was set to near historical levels, cell selections for each year was by the maximum net economic value criterion and management types were mapped from current MOF policies. 2.1.2 Spotted Owl Model Component The spotted owl population model developed here is an i-state configuration model (see section 1.3). That is, the movements and fates of individual female owls (representing breeding pairs) are simulated according the local habitat configuration in light of the best available data on this subject (Bart 1995b). Although individuals are simulated, this does not imply that the model avoids the use of parameter averages entirely, but the changes in habitat over time and the spatial location of each individual determine its fate. The model is necessarily stochastic and the results obtained using this approach vary from simulation to simulation (e.g., each owl survives each year if a 0-1 uniform random number is less than the annual survival rate). To compare policies, at least 100 trials were run (Monte Carlo simulations), to allow visualization of variability and uncertainty about the population trajectory (see Figure 9). The model keeps track of, and writes to a text file, the total number of owls each year for each simulation trial (e.g., population levels from 100 years multiplied by 100 trials), it also keeps track of the average number of owls per year (e.g., the sum of the population size at each year divided by the number of trials), plus a standard deviation, and writes this information to a text file. These files are created in the event the user wants to graph and/or compare simulation results. Simulated population changes result from the following individual-scale processes: 1) annual survival of age 1+ owls already present; 2) immigration of new owls into the 30 20 --0 -I 1 1 1 1 1 1 1 1 H 1 1 1 1 1 1 1 1 1 H 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70. 75 80 85 90 95 Year of simulation Figure 9 Example of 15 Monte Carlo simulations of the owl population. These simulations exhibit the variation and uncertainty of the path that the population size takes. When 100 simulations are complete then an average number of owls per year, plus a standard deviation, can be calculated. 31 population from US areas to the south (generally assumed to be 1-3 individuals/year); 3) stochastic production of new juvenile females by each territorial female; and 4) stochastic dispersal and survival of these juvenile females over their first year of life. At any moment in time, the 1+ population state consists of a number of females distributed over potential breeding territories, plus a number of dispersing ("floater") juvenile females seeking breeding territories. Net recruitment is a function of the number of juvenile female owls, including immigrants from the US, surviving the dispersal process minus adult female mortality. 2.1.2.1 Territory Size The model can incorporate any user-defined territory size, but with no size variation or overlap among territories. It has been suggested that the median of the data from Hanson et al. (1993) (section 1.5.1) should be used to estimate home range size, rather than the mean, to reduce the effects of abnormally high or low values (SORT 1994). There does not appear to be a reason to do this because there are no outliers in the data. Therefore, for simulation purposes the territory size was set to the Western Cascade Physiographic Province average of 4200 ha (i.e., 168 twenty-five hectare forest map cells) (Hanson et al. 1993). 2.1.2.2 Proportion of Suitable Habitat Within a Territory Because Bart (1995b) found that fecundity and annual adult survivorship were linearly related to the proportion of old-growth found within a territory, the minimum proportion of old-growth required for breeding within any territory was set at 0.206 (see section 1.5.2). Each year, the model sums the number of cells within each territory that have 32 ' old (120+) stand ages, then divides this number by the number of cells in the territory. Breeding occurs (i.e., the probability of fledging a juvenile female) if this ratio exceeds 0.206. 2.1.2.3 Territory Formation A General weakness of models that represent territory patterns, is that regularly shaped territories are systematically placed over the landscape. To avoid that problem, this model delineates territories such that they follow logical geographic contours and local restrictions, resulting in a "mosaic" of variously shaped territories. The "mosaic" is created, first by defining territories around currently known owl locations. Starting at the cell that contains the owl pair and concentrically adding cells to the simulated territory, the model can "sketch" a territory of any size. Once this is accomplished, the remainder of the habitat is partitioned into "potential" territories the same way. The concentric search and cell addition process produces irregular shaped territories in areas where natural boundaries prevent concentric addition of cells (e.g., along river bottoms bounded by non-forested alpine areas). Each territory is assigned a mean column and row number based on the maximum size of a territory. That is, the center of the territory is converted from the small scale row and column array of the forest database map to a new array that is dimensioned according to territory size. This new array system allows for finer-scale forest dynamics to be represented at the "coarser" territory scale as discussed in section 2.1.2.4. This new territory array is used to create a map of territory quality that is used in the model's interface to show the temporal and spatial patterns associated with habitat changes, breeding female owl location and juvenile dispersal. 33 2.1.2.4 CrossLScale Linkage We sought new ways to deal with the spatial cross-scale problem of how to simultaneously represent forest dynamics on a fine spatial scale while representing owl population dynamics on a coarser territory "grid". The key to this cross-linkage methodology has been to develop a method for mapping potential animal home ranges over the detailed forest landscape data, thus providing a coarse-scale mosaic of animal-use areas. The model then aggregates forest-state data (forest age in the case of the spotted owl) over these animal-use areas to provide simple indices of habitat quality. Each territory is assigned a quality index by adding up habitat index values over the cells contained in the territory (i.e., forest > 120 years has an index of 2; forest < 120 years has an index of 1). If the entire territory contains old-growth forest, then its suitability index would be two times the number of cells in a territory. If all cells have ages less than 120 years, then the territory habitat index would be exactly equal to the number of cells per territory. For the territory to meet the breeding habitat requirements mentioned above, then its index must meet or exceed the required user-defined index. The required index is calculated by multiplying the proportion of old-growth required for breeding by the number of cells per territory and then adding the number of cells per territory to that product. For example if a territory is comprised of 168 cells (168 x 25 ha = 4200 ha.) and a minimum of 20.6% old-growth is required for breeding (Hanson et al. 1993) then the minimum suitability index would be 202.6 (i.e., 168 cells x 0.206 + 168 cells = 202.6). This calculation effectively omits the contribution of young forested cells to the index. It ensures the index is based solely on the contribution of old forest and that breeding only occurs in territories with a suitability index >202.6. 34 The territory suitability index is updated each year during the course of a simulation. The logging/conservation scenario that was passed from the forest harvesting component to the owl population component is used to change the forest ages over time. Territory scale suitability indices are updated each year to reflect these changes. One further element regarding habitat quality needed to be incorporated into the territory suitability index. In the United States, spotted owls concentrate their activities in habitat that is <1200 m above sea level. Although owls regularly forage and occasionally nest above 1200 m, most spotted owl activity and nests occur below this elevation (Hays, unpublished data.; Campbell and Campbell, 1986). This led to the hypothesis that habitats >1200 m in elevation are less suitable for breeding than habitats <1200 m. Prey production, availability of nest trees, or some other stand quality attribute may decrease with increasing elevation. To represent this, we assume an owl requires a greater total area of low-quality habitat within its territory to procure the resources that it needs for maintenance and reproduction. This hypothesis was incorporated into the model by attributing a lower suitability index to forest cells found at high elevation. This was done by multiplying each cell's suitability index (i.e., 1 for young forest and 2 for old-growth) by a relative proportion. That is, habitat above 1200 m can be attributed a suitability value anywhere from 0 to l.Ox the lower elevation habitat. Territories above and below 1200 m would still require a minimum index of 202.6 to be considered breeding habitat, but territories that contain habitat above 1200 m would require more cells to be in old-growth condition to meet or exceed that suitability index. That is, if high-elevation habitat was 50% as suitable as low-elevation habitat then 2 ha of high-elevation old-growth habitat would be needed to contribute as much to the suitability index as 1 ha of low-elevation old-growth. 35 2.1.2.5 Juvenile Dispersal Juvenile spotted owls are obligate dispersers. Current understanding of dispersal by the spotted owl, and most other faunal species, is limited (Miller et al. 1997; Taylor 1990). Juvenile dispersal has been modelled such that survival is represented by the probability that the bird will survive a sequence of dispersal steps, with a probability of mortality during each step. The probability of survival over dispersal is assumed to depend on two parameters; first, the number of dispersal steps each juvenile makes before it locates a suitable breeding territory; and second, the proportion of suitable habitat (i.e., forest >120 years of age) in the spatial cells over which it disperses. Using the parameter for the proportion of old-growth required for dispersal across each potential territory/habitat, a suitable dispersal habitat index is calculated over the territory-scale map in the same way as the breeding habitat index (section 2.1.2.4). Two methods can then be used to model dispersal. The first method involves restricting juvenile dispersal steps to moves across territories that contain at least 20% old-growth, or some other constant proportion, and maintaining a constant survival probability per dispersal step. The second method, which stems from field research and modelling, models juvenile survivorship per step such that steps over clearcuts and open forest during dispersal increase the probability of mortality (Miller et al. 1997; McKelvey 1996; Gutierrez et al. 1985; Miller and Meslow 1985). This method allows dispersal over any habitat, regardless of the proportion of old-growth, but incorporates a specific predation or mortality risk by varying the probability of survival with the amount of suitable habitat encountered at each dispersal step (i.e., the lower the proportion of suitable habitat within a territory the lower the probability of survival to the next dispersal step). The general 36 assumptions are that the less old or mature forest cover each habitat contains, the higher the rates of predation (i.e., due to a lack of escape habitat) (Dawson et al. 1987) and starvation (i.e., the owls die due to the combination of inexperience at catching prey and low prey densities) (Miller 1989 cited in Miller et al. 1997). Also, dispersal success decreases in a fragmented landscape because juvenile owls are required to disperse greater distances through potentially unsuitable habitat during a period of relatively high vulnerability (Gutierrez et al. 1985). Unlike breeding territories, which must be as large as specified in the model's parameters, dispersal territory "cells" can be of any size, and include the areas determined to be too small to sustain a breeding pair. Using Holling's disc equation (Holling 1959) the probability of survival over any dispersal step has been modelled to rapidly decrease towards zero when owls cross habitat that contains less suitable habitat (Figure 10). The rate at which the curve bends can be varied by changing an index value (i.e., the "curve index" parameter) within the equation. The equation is: ^ probability of A juvenile survival to the next V. dispersal step ) (curve index) x proportion of old - growth in dispersal habitat / proportion of old - growth in 1 + (curve index) x ' . ,, , . V dispersal habitat Reasonable estimates of the curve index parameter were determined by using the model to reconstruct the history of the owl population (see section 2.2.1). 37 Proportion of suitable habitat encountered by a dispersing juvenile Figure 10 Juvenile survival per dispersal step in relation to the proportion of suitable habitat (i.e., forest > 120 years old) the juvenile encounters while dispersing. This represents increasing juvenile mortality risk (e.g., predation, starvation) as the proportion of suitable habitat encountered declines. The curves represent the range of values of survivorship that can be set within the model. The "maximum" line, which can be set to any level, limits the maximum juvenile survival per dispersal step to that level (i.e., any survival probabilities above this level are reset back to this level). If the user wishes to remove the predation risk then a constant survival probability across all suitable habitat proportions can be set by setting the curve value to a level of 1500 or greater (i.e., the curve value always exceeds the maximum, no matter what the proportion of suitable habitat encountered, and is therefore reset to the "maximum" level.). oo 00 This method of modelling dispersal survival is likely more realistic than a simple survival cut-off at one particular habitat quality proportion, but the latter is still an option in the model. The curve index method allows juveniles to disperse through landscapes with little or no old-growth but there is a higher mortality risk associated with doing so. To eliminate the possibility of calculating unrealistically high survival probabilities (i.e., greater than 0.95) a maximum survival per dispersal step can be set. This maximum dispersal survival can be set in the model's parameters to account for the "true" maximum dispersal survival' regardless of habitat quality. Also, the effects of predation or starvation can be eliminated by setting the parameter for the proportion of old-growth required for dispersal to a level that eliminates the steep portion of the curve from the set of areas "tested" by dispersing juveniles. The simulated pattern of juvenile dispersal (moves in successive steps) is not completely random. A random movement direction is initially chosen (i.e., to any one of eight adjacent forested territories) but successive dispersal moves are limited such that the juvenile cannot initially turn back in the direction from which if came. This is insured by keeping track of the last movement direction. This does allow for the eventual possibility of the juvenile returning to its initial cell by circling back, only i f its movements are consistently to the left or to the right (i.e., a spiral pattern). A maximum number of dispersal steps is set for each year and the probability of survival to the next step decreases as a function of the number of steps already taken (i.e., the probability of survival decreases as the dispersal distance increases). For simulation purposes, the maximum number of dispersal steps per year, was set at 20. Each dispersal step is the length of a territory and is approximately the distance a juvenile can cover in one week. This translates into a maximum dispersal distance of 130 km/year (i.e., 20 steps multiplied by the width of a territory in km) although most birds either "die" or locate a suitable 39 breeding habitat before that distance is reached. If a dispersing bird reaches the maximum number of dispersal steps within a year then the bird is considered a "floater" and continues its dispersal in the next year of the simulation until it either "dies" or locates a breeding territory. It is also assumed that the juvenile "dies" if it disperses outside the boundaries of the territory mosaic (i.e., outside the current range). 2.1.2.6 Biological Parameters The following are descriptions of the biological parameters used in the model. The population simulator is based on female birds only. Therefore, parameters such as clutch size and fledging rate are expressed as the number of female eggs laid and the number of female owls fledged by a breeding female. Modified versions of the survival and reproduction relationships described by Bart (1995b) are used to calculate the fledging rate and annual adult survivorship for each female owl (i.e., "pair") within each territory (Figure 11). The proportion of suitable habitat within a territory is calculated as described above in section 2.1.2.4. Annual adult survivorship in the highest quality habitat is set to a maximum of 0.90 to avoid implausibly high rates. The northern spotted owl is a long-lived species and, within the model, an owl living in a territory comprised solely of old-growth has about a 10% chance of living to the age of 22 years. Also, the relationship between fledging rate and suitable habitat has been halved because only females are modeled. This requires the assumption that half of the fledglings are females. Clutch size is assumed to be 1 female egg per female per year and, similar to the adult survival relationship, the fledging rate in high-quality habitat is limited to a maximum of 0.30 females fledged per territory per year. This maximum rate is higher than the values used in some demographic analyses 40 ja •a a 3 O 1 T 0.9 0.8 0.7 0.6 -f 0.5 0.4 + 0.3 0.2 0.1 + 0 Adult survival rate • Fledging rate -+-—-| 1 1 1 1 —( 1 4 — 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0!8 0.9 1 Proportion of mature and old growth forest within a breeding terrirory Figure 11 Annual adult survivorship and juvenile fledging rate in relation to the proportion of mature and old-growth forest found within a breeding territory. The sections of the curves where the slopes are zero indicate the maximum levels used in the model. 41 (Noon and Biles 1990), but is consistent with the information given by Bart (1995b) and allows for greater fledging success in high quality habitats. 2.1.3 Calculation Sequences An important aspect of model evaluation is a description of the sequence of calculations leading to the output (Bart 1995a). The following lists show the calculation sequences for the forest harvesting and owl model components. The forest harvesting and biological parameters for each calculation are described above. 2.1.3.1 The Forest Harvesting Model Calculation Sequence 1) The model uses the user-defined AAC level as its target then it "cuts" cells one-by-one until the target is reached. A cell will be chosen for harvesting if it meets the criteria set by the user, that is, if it is not set as a permanent or temporary reserve and if it has reached the minimum age for cutting. 2) The selection criteria (i.e., oldest first, maximum volume or maximum net . value) also limits which cells will or will not be cut. a) If the selection criteria is "oldest first" then cells with the oldest aged trees will be cut until the AAC is reached. b) If the criteria is "maximum volume" then the cells with a combination of the oldest trees and highest site productivity index will be cut. c) If the criteria is "maximum net value" then the cells that have the highest production less cost will be cut in priority. 3) As each cell is cut, the forest age of than cell is reset to a user-specified level. 4) If a cell is not cut, then its age is increased by one year. 5) The model then proceeds to the next year and continues until it reaches a predetermined number of years to simulate. 6) The model maintains a registry of the specific cells cut each year along . with the total number of cells cut per year during the course of a forest harvest simulation. The cell numbers and corresponding number of cells cut per year are output to a "forest harvest scenario file" for use by the owl model. 42 2.1.3.2 The Spotted Owl Model Calculation Sequence 1) The forest map file and the location and timing of forest harvesting are read into memory from a "forest harvesting scenario" file. 2) Next, the owl model delineates territories around current owl locations. 3) The model then proceeds to make "potential" territories in the remaining area. 4) Initial suitability indices for each territory are calculated based on the current age structure of the forest. 5) The initial locations of individual spotted owls are set. 6) The model then proceeds into its 100 year simulation. 7) For each breeding owl, the model adds fledglings, and sets their initial location as floaters, to be used as the starting position for dispersal. 8) The model then adds immigrant floater birds from the U.S. into the row of territories adjacent to the Canada/US. border if a randomly generated number is less than the user-specified probability of receiving an immigrant. 9) All floaters are then dispersed across the landscape until they either "die" or find a suitable breeding territory and become "breeders". If a juvenile makes the maximum number of dispersal steps and is still alive, it is considered a "floater" and added to next year's floater population. 10) Based on the timing and location of forest harvesting (read into the model from saved forest harvesting scenario files), the forest age of each 25-ha cell is changed by either adding one year to the age if it was not cut or reset to the user specified age if it was cut. 11) The suitability indices of the territories and dispersing habitat are updated based on the changes in forest age according to the forest harvesting scenario. 12) At the end of the year, surviving, owls are identified and mortalities removed from the population. 13) The model returns to step 6 and starts the next year of the simulation. It proceeds until the end of the forest harvest scenario is reached. 2.2 Population Simulations A key objective of this model is to evaluate relative responses of the BC component of the northern spotted owl population to various proposed forest harvesting and conservation strategies. When work on this model began, there was no spotted owl management plan in place and information for the BC population of spotted owls was lacking. The latter is still true, however, a management plan is now in place (LUCO 1997). 43 2.2.1 Historical Forest Disturbance and the Trade-off Hypothesis Of all the spotted owl's population dynamic parameters that have been studied, survival of dispersing juveniles appears to be the least well known. The trade-off hypothesis predicts that for owls to be present in BC at current population levels, then either: 1) juvenile dispersal survivorship is high and there is a corresponding decrease in the quality of high-elevation habitat (compared to low-elevation habitat) for typical home ranges; or 2) juvenile survivorship is low and the relative habitat quality must be similar to that of lower elevation sites. There could be, of course, some intermediate combination of these two parameters. Juvenile survivorship was chosen for intensive analysis and estimation because it is likely a major limiting factor for the spotted owl population. The model was used to identify parameter combinations that would allow the simulated owl population to terminate at current population estimates, given the history of forest disturbance in BC. A reconstruction of the historical forest disturbance in the Fraser and Soo TSAs was used to simulate owl population dynamics over the last 100 years (Figure 12). Using the historical changes in habitat configuration, the model can then be used to determine whether the model's predicted outcomes (e.g., population level trends, current population size) match the actual response of the owl. Such "reconstructive simulations" can be a powerful check on model parameter values. For example, setting too high a simulated sensitivity to reductions in old-growth area results in the prediction that owls should have disappeared long ago. The historical disturbance regime was inferred from the current forest age data by taking the current age of each cell and calculating when it was disturbed by development, burning, logging or agriculture. For example, if a forest stand is 36 years old then it was estimated to have been "cut" 36 years ago. This information was compiled for each stand disturbed within 44 Disturbance History Logging History Figure 12 Area of forest disturbed (i.e., undocumented logging and development) and the area of forest logged in the Fraser and Soo TSAs from 1896 to 1995 (based on current forest cover data and the map files used within the model). 45 the last 100 years and converted to a forest "harvest" scenario file which could be used in the owl model. Bart (1995a) suggests that this type of "reality check" is perhaps the best evaluation of a model's predictions. The most uncertain biological parameter is juvenile survival. Further, the relationship between the quality of high-elevation habitat relative to low-elevation habitat is not well understood. Simulations were performed by varying juvenile dispersal survivorship (i.e., changing the response of survival to habitat quality by varying the curve index, described above, between 1 (low survival) and 1096 (constant, high survival)) (Figure 10) and varying the relative habitat quality index of high-elevation sites (i.e., >1200 meters) between 40% and 100% of the quality of habitat <1200 m. 2.2.2 AAC Levels, Conservation Options and the Trade-Off Hypothesis The model was used to assess the effectiveness of the combinations of: 1) conservation strategy; 2) level of forest harvesting; and 3) trade-off values between juvenile dispersal survivorship and high-elevation habitat quality at maintaining and recovering the owl population within 100 years. The trade-off hypothesis predicts that when high-elevation habitat is equal to low-elevation habitat, then juvenile dispersal survivorship is necessarily low and vice-versa. These two combinations were determined by reconstructing the historical forest disturbance of the Fraser and Soo TSAs (see section 3.2). Twenty-four different combinations of strategy and parameter estimates were defined to cover the range of proposed strategies and uncertainty, including the strategy adopted by the current Provincial Government (LUCO 1997). Table 1 shows the combinations of harvesting 46 High AAC (maintenance of current AAC level - 2.335xl06nr7yr) MOF AAC (predicted decline over the next 100 years) Impact AAC (predicted impact of the current management plan MELP AAC (predicted impact of SORT Option C) High-elevation habitat A -juvenile dispersal survival trade-off values High/. • . Low Low,/- • ' High ; High/ Low • -Low/ High, High/. Low Low / • High'' -High/ Low LOW/ ; High Current Parks Only High AAC/ Current Parks High AAC/ Current Parks MOF AAC/ Current Parks MOF AAC/ Current Parks Impact AAC/ Current Parks Impact AAC/ Current Parks MELP AAC/ Current Parks MELP AAC/ Current Parks LUCO (1997) Management Plan High AAC/ Management Plan High AAC/ Management Plan MOF AAC/ Management Plan MOF AAC/ Management Plan Impact AAC/ Management Plan Impact AAC/ Management Plan MELP AAC/ Management Plan MELP AAC/ Management Plan SORT Option C High AAC/ Option C High AAC/ Option C MOF AAC/ Option C MOF AAC/ Option C Impact AAC/ Option C Impact AAC/ Option C MELP AAC/ Option C MELP AAC/ Option C Table 1 Conservation strategy, forest harvesting level and trade-off value combinations tested with the model for effectiveness at recovering the northern spotted owl population in BC. level, owl conservation strategy (SORT 1994; LUCO 1997), and owl parameters that were evaluated. A very extreme scenario, in which all forest harvesting within the Fraser and Soo TSAs was halted, was also included to provide a baseline for which the effectiveness at recovering the spotted owl by the other strategies could be compared. AAC levels used in the model to assess each conservation strategy are shown in Figure 13. The "High AAC" is maintenance of the AAC at 2.335xl06 mVyear; the "MOF AAC" is the predicted level from recent Timber Supply Reviews (MOF/MELP 1995); the "MELP AAC" is the predicted impact that the Option C SOC A plan (Figure 14) will have on the timber supply (MOF/MELP 1995); and the "Management Plan AAC" is the predicted impact that the current spotted owl management plan (Figure 1) will have on the timber supply (LUCO 1997). The management options modelled to assess the response of the spotted owl are shown graphically in Figure 1, Figure 6 and Figure 14. Figure 1 is a map of the current management plan and exhibits the locations of the Special Resource Management Zones (SRMZ), which were adapted from the proposed SOCAs; it appears to be very similar to SORT'S Option H plan. The current management plan stipulates that there will be 67% retention of forest greater than 100 years old in all SRMZs (LUCO 1997). The other two options modelled, shown in Figure 6 and Figure 14, are the current park system, from the Lower Mainland Protected Areas Strategy, and SORT'S Option C (i.e., no harvesting on 316,660 hectares, which includes current parks and non-forested land within 21 SOCAs/parks), respectively. The areas shown in blue are proposed SOCAs and the provincial parks are shown in green. Areas that are black are either operable or inoperable 48 3 o co I 1 --ra 3 C C < 0.5 ---0 High AAC (2.335x10A6 mA3/year) - - - - Minstry of Forest AAC prediction for the next 100 years -X—Suspected impact due to current management plan (LUCO) - • ^ M i n i s t r y of Environment's predicted impact on AAC due to Option C —+— 20 -+-30 -+- -+- -+-10 40 50 60, Years from present 70 80 90 100 Figure 13 Annual Allowable Cut levels used to assess the efficacy of each conservation strategy at recovering the spotted owl population in BC. 49 Figure 14 The Option C S O C A configuration (one o f three options used in the model to assess the response of the spotted owl). The blue areas are proposed S O C A s . The green areas are comprised o f parks present at the time the S O R T Management Options Report (SORT 1994) was published. There is no harvesting in the parks or S O C A s . The second and third management options modelled are current existing parks only (Figure 6) and the current management plan (Figure 1). The current management plan is a combination of the Option C policy and current park system. This plan includes all parks plus 67% retention o f forest >100 years old in all S O C A s (see the Special Resource Management Zones in Figure 1). 50 forest, and are subject to the harvesting constraints on each forested cell's management type imposed by the Forest Practices Code. The model attempts to simulate the harvesting constraints imposed by the Forest Practices Code as closely as possible. All harvested areas that were retention, partial retention, deer winter range and SRMZ areas were subject to green-up requirements before adjacent cells could be logged. Green-up periods of 25 years was used in areas of retention, 15 years in partial retention, and 35 years in deer winter range and SRMZs. For all simulations the forest fire ignition probability parameter was set to zero so as to maintain consistency between owl conservation policy evaluations. This removes the uncontrolled variable of fire which would complicate the simulation results unnecessarily. This reasonable in light of the effective forest fire suppression activities that take place in southern BC. The two population response measures that were generated for each proposed strategy, are the population size at the end of a 100-year simulation and the mean time-to-recovery. Time-to-recovery was measured relative to an arbitrary population level that was both above the current level and such that population inventories would detect the increase. This recovery level was set at 125 pairs or roughly 25% more than current population estimates. 51 3. RESULTS 3.1 Territory Mosaic The method of creating a territory "mosaic" compared to creating territories by the systematic placement of regular shapes has implications for the amount of habitat available to owls and the maximum number of owls that can be present at any given time. The simulated potential territory mosaic (Figure 15) indicates that at any given time there is a relatively large proportion of habitat that is unavailable for use by breeding spotted owls. For example, applying the model in the Fraser and Soo TSAs shows that approximately 15% of the total owl range is contained in habitat patches that are too small to serve as functional owl territories. The mosaic model predicts that portions of the landscape falling into these areas are effectively isolated by adjacent territories. That is, habitat in the upper portions of a watershed may be isolated by an occupied territory at its lower reaches. This suggests that the number of breeding owls that the current range can effectively support is far fewer (about 45% less) than estimates based on total area alone (e.g., Ward et al. 1991). Although isolated habitat patches are not large enough to be used by breeding pairs, they are used by dispersing juveniles in the simulations. Each territory in the mosaic, including the small habitat patches, is represented on a coarse territory grid/map (Figure 16). This grid is displayed in the model interface so that the user can visualize the temporal and spatial patterns associated with habitat quality changes, breeding female locations and juvenile dispersal patterns. 52 Figure 15 Owl territory "mosaic" laid out over the Fraser and Soo Timber Supply Areas (Figure 1). The land is delineated into potential owl territories for which suitabilities are contingent on forest conditions. Note the small territories that effectively remove a significant portion of the land base from use by breeding owl pairs. (Colours provide contrast between individual territories). 53 Figure 16 Cross-scale territory map used for the visualization of the temporal and spatial patterns of habitat change, female owl location and juvenile dispersal. 54 3.2 Juvenile Dispersal Patterns Similar to habitat changes and breeding female location, the patterns of juvenile female dispersal are visually displayed by the model. Figure 17 shows a single year example of the paths taken, distance traveled and whether the dispersal event was a success (i.e., location of an unoccupied breeding territory). 3.3 History and the Trade-Off Hypothesis Simulations of population response to historical disturbance indicate that as the habitat quality/juvenile survivorship curve11 and relative habitat quality of high-elevation sites increase, the resulting estimate of the current population also increases (Figure 18). To predict the current measured abundance, it is necessary to trade-off between juvenile dispersal survivorship and relative habitat quality of high-elevation sites. It is important to note that low dispersal survivorship/low relative habitat quality parameter combinations would not have allowed the owls to persist at current levels in BC. Further, parameter combinations with high dispersal survivorship/high habitat quality predict that there would be many more owls in BC than are currently estimated. Figure 18 shows that there are levels where the population ceases to respond to increases in both habitat quality and juvenile survivorship. The model assumes that space is the limiting factor (via potential territories), therefore, these plateau levels correspond directly to the carrying capacity of the habitat. " As the juvenile survivorship curve index increases so does the juvenile dispersal survival rate which is dependent on the index and on the dispersal habitat quality . 55 Figure 17 Typical dispersal patterns exhibited by simulated juvenile female owls. Black lines indicate the dispersal path and the red boxes indicate that the bird has located a potential breeding territory. Black lines without a red box attached represent unsuccessful dispersal attempts. The map on which the dispersal patterns are shown is the territory scale representation of the Fraser and Soo TSAs (see Figure 16). 56 Figure 18 Ending owl population sizes for historical disturbance regime simulations that varied juvenile dispersal survivorship and relative habitat quality (elevations > 1200 m were varied in quality relative to lower elevation sites). See section 2.1.2.5 for an explanation of the "curve index". The actual parameter combinations exhibited by the birds will likely have major implications on -SOCA configuration, other conservation strategies and corresponding impacts on forest harvesting. Therefore, the simulation model has been used to predict the relative effectiveness of various proposed strategies (see section 3.3) for a range of parameter combinations along the trade-off relationship. It has. been the opinion of MELP personnel and the Spotted Owl Recovery Team (SORT) that high-elevation sites, or inoperable timber areas, have relatively low habitat quality. The trade-off hypothesis predicts that this is possible, but it is quite possible that the opposite is true. Current locations of spotted owls suggest that many are in territories that contain large proportions of high-elevation sites (i.e., habitat that is possibly sub-optimal). This appears to be common across the distribution of owls in BC and cannot be explained by the MELP and SORT estimates of habitat quality requirements and dispersal survivorship. The implications of this are dealt with in the Discussion. . As mentioned, the historical disturbance regimes for the Fraser and Soo TSAs from the last 100 years was used to determine which parameter combinations would result in an ending population of approximately 100 owl pairs (roughly equal to current population estimates in BC). The parameter trade-off combinations leading to this population state were determined while maintaining breeding territory sizes of 4200 ha, a minimum breeding habitat index requirement of 202.6 and a maximum adult annual survival of 90%. The model indicates that if high-elevation habitat has 100% of the quality of low-elevation habitat, then the curve index of juvenile survivorship per dispersal step must be 17.5 (Figure 18). Conversely, the lowest the relative habitat quality that would result in a current population level of 100 pairs is 67%, at a juvenile dispersal survivorship curve index of 54. The average trajectory that the simulated population followed during reconstruction of the disturbance 58 history under the 100%/17.5 combination is shown in Figure 19A. A similar trajectory was described by the 67%/54 combination (Figure 19B). It should be pointed out that the owl population is initially "seeded" with 115 owls and quickly increases to the level that the landscape can sustain. Over time, the simulated population declines to current levels as logging and disturbance activities increase (Figure 12) and expand beyond the Lower Mainland and Fraser River areas. 3.4 Conservation Options Simulations of conservation options mimic experiments that are impossible to conduct in reality. The effect that three factors have on two aspects of the population's response have been tested. The three factors are: 1) Annual Allowable Cut levels; 2) conservation strategy; and 3) alternative estimates of the juvenile survivorship rate and the quality of high-elevation habitat (along the trade-off that results in current abundance). The third factor was taken at the extremes of the estimates that resulted in 100 pairs (see section 3.2, Figure 18, and Figure 19). The mean number of female owls at the end of a 100-year simulation and the mean time-to-recovery are shown in Table 2 and Table 3. Using SYSTAT v.5.2, a 3-factor analysis of variance (ANOVA) was performed to test for the effects of conservation option, AAC levels, and the high-elevation quality/juvenile survival trade-off, and/or their interactions, on the spotted owl's mean time-to-recovery and on the mean number of owls at the termination of the simulation. The ANOVA results for testing the effects on mean time-to-recovery are shown in Table 4 and the ANOVA results testing the effects on mean population size at simulation end are given in Table 5. 59 ure 19 Female owl population responses to the historical disturbance regime reconstruction of the Fraser and Soo TSAs for the two extreme parameter combinations leading to 100 owl pairs (± 1 standard deviation, n=100 simulations). Initial increase is a result of "seeding" the starting population at a level lower than was likely to have existed 100 years ago. 60 High AAC (maintenance of current AAC level - 2.335xl06m7yr) MOF AAC (predicted decline over the next 100 years) Impact AAC (predicted impact of the current management plan MELP AAC (predicted impact of SORT Option C) High-elevation habitat level /Juvenile survival level • High / Low Low./ "High High/ . Low ' Low/ High •. High/ Low Low / High High/ Low Low/ High Current Parks Only -98.0 (15.8) 94.0 (12.9) 147.1 (15.5) 138.6 (17.9) 152.4 (15.5) 143.3 (18.1) 165.6 (14.0) 160.3 (13.8) Management Plan (SORT Option H) 93.7 (15.8) 90.2 (15.0) 146.2 (16.0) 138.5 (17.9) 149.0 (16.4) 141.7 (14.9) 166.6 (12.5) 160.0 (14.8) SORT Option C 91.3 (14.5) 89.1 (13.1) 148.3 (13.7) 139.1 (16.1) 151.6 (13.9) 147.3 (15:7) 164.9 (12.6) 160.6 (16.8) Table 2 Spotted owl population size at the end of the 100 year simulations for all the management options modelled (mean ± 1 standard deviation). High/Low - high-elevation quality = 100% relative to low-elevation habitat and juvenile survival curve index = 17.5 Low/High - high-elevation quality = 67% relative to low-elevation habitat and juvenile survival curve index = 54. High AAC MOF AAC (predicted Impact AAC (predicted MELP AAC (predicted (maintenance of current AAC decline over the next impact of the current impact of SORT Option C) level- 2.33 5x106  m3 /yr) 100 years) management plan High-elevation habitat level High/ Low / High/ Low / High/ Low/ High/ Low /' / Juvenileisurvival level Low High Low i : High Low High Low ni»h : 88.6 88.0 80.5 84.0 78.9 84.2 67.3 73.8 Current Parks Only (12.8) (15.7) (13.9) (10.3) (13.0) (11.3) (13.1) (12.6) n= 10 n = 3 n = 98 n = 84 n = 98 n = 90 n=100 n= 100 94.8 84.0 80.0 86.5 79.4 83.6 67.2 73.6 Management Plan (SORT (4.1) (28.6) (14.2) (11.9) (13.9) (11-4) (14.3) (12.6) Option H) n = 4 n = 6 n = 96 n = 90 n = 95 n = 90 n= 100 n= 100 92.2 91.0 79.1 84.3 74.0 81.0 68.6 73.3 SORT Option C (5.7) (N/A) (12.9) (8.6) (15.6) (10.9) (14.4) (14.9) n = 6 n= 1 n = 98 . n = 92 . n = 97 n = 98 n=100 n = 98 Table 3 Mean time-to-recovery in years. Each conservation strategy's mean time (standard deviation in parentheses) for the spotted owl population to recover to a level of 125 pairs. Mean time-to-recovery is based only on the simulation trials that reach 125 pairs (i.e., n = number of simulations out of 100 that reached 125 pairs). High/Low - high-elevation quality = 100% relative to low-elevation habitat and juvenile survival curve index = 17.5 Low/High - high-elevation quality = 67% relative to low-elevation habitat and juvenile survival curve index = 54. Source of Sum-of- Degrees of Mean- F-Ratio Probability Variation Squares Freedom . Square AAC level • 49409.862 3 16469.954 97.804 0.000f Conservation Strategy 19.668 2 9.834 0.058 0.943 Trade-off Values 648.031 ' 1 648.031 3.848 0.050f AAC Level * Conservation Strategy 1378.095 6 229.683 1.364 0.226 AAC Level * Trade-off Values 438.179 3 146.060 0.867 0.457 Conservation Strategy * trade-off Values-; 107.384 2 53.692 0.319 0.727 AAC Level * Conservation Strategy * Trade-off Values 488.046 6 81.341 0.483 0.821 Error 291326.742 1730 . 168.397 Table 4 Results table from the ANOVA testing for the effects that AAC level, Conservation Strategy, trade-off values, and their interactions have on the owl population's mean time-to-recovery.f = significant (p < 0.05) 63 Source of Sum-of- Degpes of Mean-' F-Ratio Probability Variation Squares •;"5iiiiom Square . AAC Level 1670507.783 3 556835.928 2379.840 O.OOO1 Conservation Strategy 1111.548 2 555.774 2.375 0.093 1 rade-off Values 21528.060 1 21528.060 92.008 0 .000 f AAC Level * Conservation Strategy 4358.289 6 726.382 3.104 0.0.05* AAC Level * Trade-off Values 2207.563 3 735.854 3,145 0 .024 f Conservation Strategy .: * Trade-off Values 308.747 2 154.374 0.660 0.517 AAC Level * (\mservation Strategy * Trade-off Values 543.869 6 90.645 0.387 0.887 Error 555937.500 2376 233.980 Table 5 Results table from the ANOVA testing for the effects that AAC level, Conservation Strategy, Trade-off values, and their interactions have on the owl's mean population size at the termination of simulation (i.e., 100 years from the present). f significant (p < 0.05) 64 When the responses of the owl population to the 24 conservation policy alternatives are examined a striking feature is revealed (Figure 20). All of the simulations, except for those that are continued at a high AAC level, allow the population to recover. Also, it can be seen, from the 3-factor ANOVA results (Table 4 and Table 5), that the level of cut has a significant effect on the differences in mean time-to-recovery (p < 0.000) and population size at simulation end (p < 0.000). As the cut is reduced the mean ending population size increases and the average time-to-recovery is reduced. The effect that the configuration of conservation areas has on mean time-to-recovery and mean population size are not significant (p = 0.943 and p = 0.093 respectively) indicating that current and proposed management plans may have little impact on the owl population. The trade-off values also have a significant, but modest, impact on the mean time-to-recovery (p = 0.050) and ending population size (p < 0.000). The results of the simulations where the high-elevation habitat is of poor quality and the juvenile survival is high are consistently lower than the results from good quality high-elevation habitat and lower juvenile survival. Interaction effects are significant only on the number of owls at the end of the simulation. The dependence between the AAC levels and conservation strategies (p = 0.005) as well as the dependence between AAC levels and the trade-off values (p = 0.024) on the final population size of the simulations are significant. These interactions likely stem from the changes in spatial aspects of the forest between conservation strategies and from response of juvenile owl survivorship associated with those changes. The simulation that involves the stopping of forest harvesting in the Fraser and Soo TSAs completely is plotted in Figure 21 to provide a baseline which allows visual assessment of all other harvesting and conservation strategies. In contrast, the worst case scenario, the 65 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year c n c E 200 j 150 I 50 I High AAC/Option C 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year 1995 2015 2035 2055 2075 Year Figure 20 Owl population simulation results from the combinations of MELP and MOF AAC predictions and conservation options. The trendlines indicate yearly population level averages (n=100 simulations) and the error bars represent ± 1 standard error. Figure 21 Simulated responses of the spotted owl to the immediate cessation of harvesting in the Fraser and Soo TSAs and to the maintenance of a high harvest (i.e., 2.335x106 mVyear) under Option C conservation strategy. The average population levels per year of 100 simulations are plotted at each year (error = ± 1 standard deviation). It can be expected that the response of the owl to the current management plan will fall somewhere in between these two population trajectories. 67 potential response of the owl population to the maintenance of high harvest levels (i.e., 2.335xl06 nrVyear) under the Option C conservation strategy, is also shown in Figure 21. Even though forest harvesting has ceased there is still an initial decrease in population size and it does not recover to a level above the current one for approximately 25 years on average. 68 4. DISCUSSION 4.1 Population Responses to Historical Harvesting and Disturbance Using the historical forest disturbance regime for the Fraser and Soo TSAs (which was compiled from current forest cover data) to drive the population simulator has provided better population parameter estimates than had originally been expected from using only direct field data on survival, recruitment, and movement. As a result, more precise hypotheses about future population responses to management have been developed. Obtaining better parameter estimation was realized by utilizing the model in a deterministic way. If a target of the current population estimate was set then it was easy to find those parameter combinations that would be necessary to reach that target (e.g., Figure 18). This allows the user to locate a range of likely values for other parameters that are less certain. Simulations can then proceed with a narrower focus to predict alternative future population trends. Based on juvenile survivorship research, it is not likely that the combination of high juvenile survivorship and low habitat quality existsi in the natural population. Juvenile survivorship is typically low. In Washington State only 2 of 40 radio tagged juveniles survived to their second year (Dale Herter, pers. comm.). Researchers have found that survivorship of tagged females was lower than untagged females, but they found no difference between tagged and untagged male survivorship (Paton et al. 1991). Higher death rates of tagged females suggests that radio telemetry data must be interpreted cautiously with respect to survival rates. However, there appears to be no doubt that the survival rate of juvenile female spotted owls is a major factor limiting population size. 69 Simulations of the owl response to the historical forest disturbance regime have indicated that the current population size would not be possible if the high-elevation habitat had less than 67% of the quality of low-elevation habitat (Figure 18), even if juvenile survival rates were extremely high. This suggests that high-elevation habitat is an important component of owl habitat. Most territories contain both high-elevation and low-elevation habitat due to the topography of the landscape within the spotted owl's range in BC (i.e., steep, narrow watersheds, high mountain ranges). Flexibility to use high elevation areas may allow owls to disperse, locate nest sites below and forage above 1200 m as indicated by the nest site and radio-telemetry data from the Mt. Baker area of Washington State (Hays, unpub. data). If in fact high-elevation habitats are inferior to low-elevation habitats then the effect might be expressed through an increase in territory size. However, that aspect has already been acknowledged by having large territory sizes in the model as predicted by the trend of increasing size from south to north (Thomas et al. 1990.). Either the owl uses high-elevation habitat (without it being of much lower quality than low-elevation habitat), or its use of habitats is more plastic than previously believed (SORT 1994). Further, it is possible that owls are utilizing sub-mature forest areas in their territories. The possibility that the owl is more plastic than assumed in MELP assessments arises from the simulation results obtained using parameter values from the literature, not from the information that was received directly from the MELP (i.e., 43% of the territory must be mature/old-growth forest for successful breeding). The simulated response of the spotted owl to the historical forest disturbance regime, using the parameters obtained from the MELP (SORT 1994), indicated that the owl could not be present at the current levels in BC and, in fact, should already have already been extirpated from significant portions of its BC range. 70 Strategically, an argument that high-elevation habitat is unsuitable would require increased protection of low-elevation habitats. This would then support the MELPs "best" proposed conservation strategy . The simulations suggest that high-elevation habitat is very important to the dispersal survival of juvenile spotted owls. Juvenile spotted owl survivorship is very low, and without the ability to use high-elevation habitat then the spotted owl would not likely be present at current population levels and its distribution would be restricted to the GVRD watersheds and adjacent parks. The MELP has, to date, successfully used arguments about low suitability of high-elevation habitat to combat the forest industry's argument that there is plenty of owl habitat contained in inoperable forest areas (i.e., mostly high-elevation timber). Results from the model support the hypothesis that high-elevation habitats have been, and are, an important component contributing to the persistence of the spotted owl in BC. The low-elevation habitats have been highly fragmented by past development activities. Without the fortuitous and naturally occurring refuge of the undeveloped high-elevation habitats, the spotted owl would likely have been extirpated from BC. Simulated response of the spotted owl to historic forest disturbance suggests that the rate of population decline has slowed slightly in recent years (Figure 19). This may correspond to the decrease in the amount of forest disturbed (Figure 12). The slowing of the population decline is a good indication that the owl may respond positively to further reductions in the Annual Allowable Cut. The basic reason that even the most pessimistic scenarios in Figure 20 indicate population stability, rather than further decline, is that a minimum population level will be supported in existing reserves/inoperable areas, no matter what happens with logging. In 71 other words, the bulk of the damage that could be done through old-growth harvesting in the Fraser and Soo TSAs has already been done. . 4.2 Implications for Future Conservation Strategies Simulation results indicate that existing inoperable forest and reserve areas in southwestern BC are going to be essential to maintenance of the spotted owl population. They further indicate that socially and economically feasible harvest policies and conservation strategies will have little additional benefit for population viability. Even with the maintenance of high AAC levels, the population stabilizes at approximately 75 to 80 pairs and there is even a slight recovery near the end of the simulation, which is likely attributable to the recent addition of parks resulting in protection of forest that will become mature/old-growth during the simulation period. Also, it can be seen from the 3-factor ANOVA results (Table 4 and Table 5) that the level of cut is the main factor responsible for differences in mean time-to-recovery and owl population size, not the configuration of conservation areas. Even the lowest AAC level proposed by the MELP does not result in a much greater number of owls. 4.3 Problems with the Proposed Conservation Strategies Kennedy (1997) reviewed the scientific literature for evidence of a decline in northern goshawk (Accipiter gentilis atricapillus) populations. Although US government petitions claimed that goshawks have declined significantly in the US (due to forest harvesting), to the point of being threatened with extinction, Kennedy (1997) found no evidence of a decline. 72 On the contrary, the species is beginning to expand its range, perhaps recolonizing areas where they were historically present. My results hint that we should expect similar findings for the BC spotted owl. There are risks in advocating a conservation strategy without concrete evidence for the need to do so. Not only must economic and social impacts be considered, but the strategy could negatively affect the target population. It is entirely possible that the species may act negatively to a well-intentioned policy, if the realm of possible responses is not adequately considered. As results of this research have indicated, that is a distinct possibility with the spotted owl in BC. Simulations have helped identify potential flaws in SORT'S proposed Option C management plan. While a management plan presently exists (LUCO 1997), the need to properly evaluate it remains. Models such as this one can help, but frequent assessments of the conservation plan's progress are required so that changes can be made in a timely manner. This approach constitutes "adaptive management", or the use of scientific experiments to assess the outcome of management actions (Walters 1986). In addition to the model developed here, other models suggest that the viability of the spotted owl is sensitive to dispersal dynamics (Doak 1989; Lande 1991; Lamberson et al. 1992; Lamberson et al. 1994). As a result, the owl population may not respond as expected to government-proposed conservation strategies. Increasing the amount of protected area, without decreasing the AAC, intensifies forest harvesting on unprotected areas. This could be detrimental to the conservation of the spotted owl in BC. The small SOCAs and SRMZs developed by the MELP may force female juveniles to disperse into areas of intensified forestry and increased fragmentation surrounding the protected zones. This in turn, may increases the mortality risk for dispersing juveniles. This risk is not likely to be reduced by 73 owls avoiding high-risk habitats. Miller et al. (1997) showed that juvenile spotted owls did not exhibit avoidance or selection among three levels of forest fragmentation during dispersal. They also concluded that the probability of mortality increased with increased use of clearcuts during dispersal. This risk has been addressed in US conservation plans as one of the most important factors affecting the viability of the spotted owl: Thomas et al. (1990) suggest that the primary focus of a conservation plan should be the efficient design of a reserve system for the owl recognizing that it is a species with obligate dispersal. This should be the primary focus of spotted owl conservation in BC but, unfortunately, it is not. The SORT Management Options Report asserts that Management Option C (i.e., no harvesting on 316,660 hectares, which includes current parks and non-forested land within 21 SOC As/parks) is the best option for spotted owl recovery (SORT 1994). When this option is simulated in the model for 100 years, it does not allow the population to recover as expected, but may actually be detrimental to the recovery of the population under current AAC levels (Table 2 and Table 3). Under this option, the SOCA configuration reduces habitat quality between SOCAs by increasing the logging pressure in the remaining areas, and as a result, the juveniles have low dispersal survivorship when displaced into the surrounding areas. If the SOCAs and SRMZs were increased in size to make them large enough to support a metapopulation12 structure as suggested by the work of Harrison et al. (1993) and Lamberson et al. (1994), then the owl would likely respond with a recovery response related to the amount of protected area. This, in combination with a reduction in the AAC, would likely provide the best chance of recovery for the owl. 1 2 A metapopulation is a term applied to a population that consists of small sub-populations that are isolated in space (Shaffer 1985). 74 The simulations indicated that reducing the AAC immediately would slow the rate of habitat loss. This in turn, would slow the decline of the owl population. However, a "recovery" of the spotted owl population, under the current management plan, will not likely be detected for 75 years (see Table 3). It will take this long before the reduction in cut will allow previously disturbed habitats to provide the old-growth qualities necessary for owl use. The level of cut proposed by MELP results in a modestly greater number of owls compared to that expected under the MOF's predicted AAC (Table 2). Based on the confidence intervals usually obtained from, and required for, wildlife management inventories (Krebs 1989) it is doubtful that a statistically significant difference would be detected. A key prediction of the model is that, regardless of the management policy implemented, there will be an unavoidable population bottleneck approximately 30-60 years from now due to past forest practices. . Lamberson et al. (1994) modeled a spotted owl population and concluded that once reserves reach a size that includes 20 to 25 territories, occupancy rates no longer increase effectively. Preserving connectivity and expanding the geographic context of the conservation areas become more important for insuring long-term viability of the owl. In addition, the "Thomas Strategy" model predicts that the larger the reserves, or clusters of territories, the longer the persistence of the spotted owl (Harrison et al. 1993). In that model juvenile dispersal was found to occur more successfully within territory clusters than between. It also suggested that these clusters be comprised of enough area to contain 5-20 territories (i.e., similar to Lamberson et al. 1994) and to be spaced no more than 20 km apart. Both of these models suggest that size and connectivity should play an important role any 75 spotted owl conservation plan. The SORT report recognized this information but their conservation plans do not reflect it (SORT 1994). The recent spotted owl management plan for British Columbia (LUCO 1997) recognizes the importance of large reserve sizes, yet the reserve patches and Special Resource Management Zones (SRMZs) that will play a key role in spotted owl management are not nearly as large as some researchers have recommended (Harrison et al. 1993; Lamberson et al. 1994). In fact, the average SOC A, excluding current parks, includes enough forested area for only 3.5 territories. Also, the spotted owl management plan does not include a deliberate connectivity component; this is a potentially disastrous flaw. With continued logging at any proposed level, these SOCAs will become increasingly insular in an already fragmented landscape. Conversely, if SOCAs are not put in place then the logging pressure will be spread out over the entire operable landscape. This minimizes pressure on currently occupied habitats that might otherwise fall outside the SOCA system. As a result, habitat quality would be less affected at the territory scale. Connectivity concerns must be adequately addressed before too much confidence is placed in the current plan. The current management plan indicates that areas specific for spotted owl management "...were designed to maintain multiple breeding pairs...and were located to provide a reasonable probability for successful movement between SRMZs and protected areas." It is probable that areas of special management were delineated from recent locations of spotted owls without regard to size or connectivity of habitat patches. Managers simply laid out conservation areas where they found owls. Over time these areas became viewed as the best alternative for spotted owl conservation. Proposed management plans included 76 various levels of habitat protection within the SOCAs, but no policy alternatives were presented that included fewer, larger conservation areas and connectivity corridors. The BC population of spotted owls is in danger of being completely isolated from the US population. Urbanization and deforestation of the Fraser Valley has narrowed the connection between the two population segments, and a large proportion of what remains of this connection is comprised of unsuitable alpine habitat. Simulations involving immigration of spotted owls from the US indicate that immigration will not be adequate to maintain the BC population in the absence of suitable habitat. Immigration may be important, however, for the maintenance of genetic variablity. Results of this model should not be used to argue that the proposed conservation strategies, while they may be detrimental to the spotted owl, may in fact help conserve many other species. This may be the case but it is very important to not lose sight of the fact that, at this time, this project has provided insight into the biology and conservation of the spotted owl exclusively. The model could be adapted to assess other species' responses to forest harvesting if there is occasion to do so. The model has been constructed with a visual map interface so that the user can see the changes to the habitat and can see the locations and movements of spotted owls during a simulation. Although it has not been quantified, this visualization has indicated that the Greater Vancouver Regional District' (GVRD) watersheds, in conjunction with Cypress, Seymour, Indian Arm, Pinecone/Burke, Garibaldi and Golden Ears Parks, have, and will, play a vital role in the persistence of the owl population in BC. There appear to be two main components to the BC metapopulation: 1) the population in the north shore mountains, which is an area that is relatively unfragmented and acts as a source; and 2) the population 77 occupying the upper Fraser River valley area and associated watersheds, which acts as a sink. This second component appears to be highly fragmented and as a consequence is the first area to lose its population under high levels of forest harvesting even when accounting for immigration from the US. Therefore, it appears to be appropriate to develop a conservation strategy that identifies the north shore as a population source from which dispersal, to appropriately sized conservation areas throughout the rest of its historic range, can maintain the owl population. 4.4 Problems with Academic Institution and Government Partnerships In 1995 the MELP solicited us at UBC to develop this model to assess their spotted owl conservation plans. They stated a desire for a close partnership to develop the best modelling tool possible. However, when results of the model were released to the MOF, MELP and local forest companies a surprising response was given by the MELP. The results did not provide a favourable assessment of SORT'S conservation plans, and the MELP then refused to cooperate on any further changes to the model. Initially this work involved the development of the model only. To strengthen the model we proposed that field work be conducted to validate some of the parameters that were being used in the model. UBC requested the MELP's information on current owl nest sites to conduct the field work. These data were essential to the study because nest site locations are difficult, time consuming and costly to obtain and were well beyond the UBC budget. However, MELP staff verbally refused to divulge nest site information, citing that it is considered highly confidential and that they did not want outside agencies using this information (D. Dunbar, pers. comm.). We made a formal written request for the nest site 78 locations but, received no written response. However, MELP staff indicated verbally that a decision regarding access to the nest site information would be forthcoming (D. Dunbar, pers. comm). No response, however, has been given and it was apparent that they had made a decision to withdraw their co-operation in the project. To preserve future relations between the University of British Columbia and the Government of British Columbia, the issue was pressed no further. Academic institutions have an obligation to be objective as possible. When governmental agencies with specific agendas become involved, the application of proper science can be compromised. The interference of politicians in two of Canada's fisheries, the Atlantic cod and Pacific salmon, illustrates this point very clearly (Hutchings et al. 1997). Government scientists have been stifled by their own agency when their results do not coincide with the political agenda. The case of the spotted owl is very similar with respect to the intentional suppression of information regarding proposed conservation strategies. The results of this thesis indicate a need for further research and a re-analysis of what constitutes an adequate conservation plan. MELP's agenda may be to use the spotted owl as a tool (i.e., an umbrella species) to protect biodiversity in BC. When they discovered that the owl may not be an ideal tool, they restricted access to information. A full audit of all spotted owl research and management that has occurred in BC should be requested by the public of BC. Over the past few years alone the budget available for spotted owl research has been several million dollars. This audit would determine whether the MELP has been deliberately withholding information regarding spotted owl habitat requirements in BC. 79 5, CONCLUSIONS 5.1 The Future of the Spotted Owl When 100-year simulations are performed at high AAC levels (i.e., 2.335xl06 mVyear), existing inoperable forest plus reserve areas and the other restricted logging areas in southwestern BC are critical to the maintenance of spotted owls in BC. In these simulations the population stabilized at approximately 75 to 80 owl pairs and was concentrated in the southwestern portion of the BC range. The risk of extirpation under this scenario was negligible due to the large amount of protected habitat in this area. This implies that alternative future harvest policies and conservation strategies, while having various effects on the population level, will have little impact on the persistence of the population in BC during the next 100 years. The currently operable forest has been highly fragmented from past logging practices, and the owl would likely already have been extirpated from BC if it were not for the current and de facto wilderness areas and, perhaps, the owl's adaptability to a changing habitat. However, a recovery of the owl is desired, and it appears that the best possible alternative would be to reduce the timber cut to a level that is economically and socially acceptable. The MOF's current plans for reduction in the Fraser and Soo TSAs AAC may already be adequate. An option that the MELP should consider is a relationship with the US to maintain a corridor system along the international boundary. This could ensure that the BC population does not become completely isolated from the U.S. population. The design of corridors should be mindful of the size of the owl's home range, social structure, diet and foraging 80 patterns (Lindenrnayer and Nix 1993). This would help maintain the owl population's plasticity to habitat changes as a whole by maintaining genetic variability in the population. 5.2 The Future of the Model Opponents of this thesis can point out the difficulties of modelling complex ecosystems. The fact remains, however, that models, no matter how simple they are in comparison to reality, are useful tools to generate alternative hypotheses. It is very important to remember that data, not models, are used to test these hypotheses. The model developed here is not only useful for evaluating the spotted owl's response to forest harvesting and conservation measures. It could be easily adapted to many other species that lend themselves to individual-based modelling. At present there are plans to use the model for pileated woodpeckers on southern Vancouver Island and for the black-backed woodpecker and three-toed woodpecker in the boreal forests of northern Alberta. The model may also be modified to accommodate any species of mammal, amphibian or reptile that is directly affected by forest harvesting and can be modelled using an individual-based system. Spotted owl managers can now use a very sophisticated population model. It must be used with discretion and only as a tool to test the feasibility of policy alternatives. By no means should its results be relied upon without sound scientific testing in the field. Well planned, large scale experiments and proper monitoring are needed to assess the impact that forest harvesting has on spotted owls. In particular there is a vital question that needs to be answered: Is there a difference in juvenile survival rates between contiguous and fragmented habitat? The relatively large areas of contiguous habitat remaining in the GVRD watersheds, Nathatlatch River drainage and South Anderson River drainage would allow for a replicated 81 experimental design. The key components that the experiment should focus on are: 1) measurement of juvenile dispersal patterns and dispersal success in fragmented and unfragmented habitat; and 2) manipulation of habitat structure within home ranges, with concurrent measurement of bird use and breeding success, to determine the required proportion of old-growth habitat and whether this habitat needs to be low-elevation forest rather than higher elevation inoperable forest. There is a good opportunity to adaptively manage forest harvesting activities that occur throughout the range of the spotted owl in BC. This opportunity should not be missed. 82 Literature Cited Andersen, M.C. and D. Mahato. 1995. Demographic models and reserve designs for the California spotted owl. Ecological Applications. 5(3): 639-647. Barrowclough, G.F. and S.L. Coats. 1985. The demography and population genetics of owls, with special reference to the conservation of the spotted owl (Strix occidentalis). Pages 74-85 In R.J Gutierrez and A.B. Carey (eds.). Ecology and management of the spotted owl in the Pacific Northwest Forest and Range Experimental Station. General Technical Report PNW-GTR-185, Portland, Oregon. Bart, J. 1995a. Acceptance criteria for using individual-based models to make management decisions. Ecological Applications. 5(2):411-420. Bart, J. 1995b. Amount of suitable habitat and viability of northern spotted owls. Cons. Biol. 9(4):943-946. Bart, J., R.G. Anthony, M. Berg, J.H. Beuter, W. Elmore, J. Fay, R.J. Gutierrez, H.T. Heintz, Jr., R.S. Holthausen, K. Lathrop, K. Mays, R. Nafziger, M. Pagel, C. Sproul, E.E. Starkey, and J.C. Tappeiner. 1993. Recovery plan for the northern spotted owl - final draft. United States Government Printing Office, Washington, DC, USA. Bart, J., R.G. Anthony, M. Berg, J.H. Beuter, W. Elmore, J. Fay, R.J. Gutierrez, H.T. Heintz, Jr., R.S, Holthausen, K. Starkey, and J.C. Tappeiner. 1992. Recovery plan for the northern spotted owl - final draft. Department of the Interior, Washington, D.C. Bart, J. and E.D. Forsman. 1990. Dependence of the northern spotted owls on old-growth forests. Biological Conservation. 62:95-100. Blackburn, Ian. Personal Communication. Wildlife Biologist, Ministry of Environment, Lands and Parks, Surrey, BC. Botsford, L.W. 1992. Individual state structure in population models. Pages 213-236 in Individual-based models and approaches in ecology. DeAngelis, D.L. and L.J. Gross (eds.). 1992. Chapman and Hall, New York and London. British Columbia Ministry of Forests. 1995. ' Variable Density Yield Prediction interactive application (v. 6.3b May 1995). BC Ministry of Forests, Resources Inventory Branch, Victoria, BC. 83 British Columbia Ministry of Forests. 1988. Biogeoclimatic zones of British Columbia 1988. British Columbia Ministry of Forests, Victoria. Map. Campbell, R.W., N.K Dawe, I. McTaggart-Cowan, J.M. Cooper, G.W. Kaiser, M.C.E. McNall. 1990. The Birds of British Columbia. Volume 2. Nonpasserines. Diurnal Birds of Prey Through Woodpeckers. UBC Press, Vancouver. 636pp. Campbell, E.C. and R.W. Campbell. 1986. Status report on the spotted owl in Canada (Strix occidentalis caurina). Committee on the Status of Endangered Wildlife in Canada: 54pp. Carey, A. and K. Peeler. 1995. Spotted owls: resource and space use in mosaic landscapes. Journal of Raptor Research. 19(4):223-239. Carey, A., S. Horton and B. Biswell. 1992. Northern spotted owls influence of prey base and landscape character. Ecological Monographs. 62(2):223-250. Carroll, J.E., R.H. Lamberson, and S. Roe. 1995. Sinks, sources, and spotted owls: a territorial population model with continuous mortality and discrete birth. Mathematical Biosciences. 129:169-188. Dawson, W.R., J.D. Ligon, J.R. Murphy, J.P. Myers, D. Simberloff and J. Verner. 1987. Report of the scientific advisory panel on the spotted owl. Condor. 89:255-229. DeAngelis, D.L. and L.J. Gross (eds.) 1992. Individual-based models and approaches in ecology. Chapman and Hall, New York and London. Demarchi, Dennis A., R.D. Marsh, A.P. Harcombe and E.C. Lea. The Environment. Pp 55-142 in: Campbell, R.W., N.K Dawe, I. McTaggart-Cowan, J.M. Cooper, G.W. Kaiser, M.C.E. McNall. 1990. The Birds of British Columbia. Volume 1. Nonpasserines. Introduction, Loons Through Waterfowl. UBC Press, Vancouver. 514 pp. Doak, 1989. Spotted owls and old-growth logging in the Pacific Northwest. Conservation Biology. 3:389-396. Dunbar, Dave. Personal Communication. Wildlife Biologist, Ministry of Environment, Lands and Parks, Surrey, BC. Dunbar, I., B.P. Booth, E.D. Forsman, A.E. Hetherington and D.J. Wilson. 1991. Status of the spotted owl (Strix occidentalis) and barred. owl (Strix varia) in Southwestern British Columbia. Canadian Field-Naturalist. 105(4):464-468. 84 Forsman, E.D., E.C. Meslow and H.M. Wight. 1984. Distribution and biology of the spotted owl in Oregon. Wildlife Monographs. 87:1-64. Goudie, Jim. Personal communication. Growth and yield biometrician, Research Branch, Ministry of Forests, Vicotria, BC. Gutierrez, R.J., A.B. Franklin, W. LaHaye, V.J. Meretsky and J.P. Ward. 1985. Juvenile spotted owl dispersal in northwestern California: preliminary results. Pages 60 - 65 In R J Gutierrez and A.B. Carey (eds.). Ecology and management of the spotted owl in the Pacific Northwest Forest and Range Experimental Station. General Technical Report PNW-GTR-185, Portland, Oregon. Hanson, E., D.W. Hays, L. Hicks, L. Young and J. Buchanan. 1993. Spotted owl habitat in Washington. Washington Forest Practices Board. Washington. 115 pp. Harrison, S., A. Stahl and D. Doak. 1993. Spatial models and spotted owls: exploring some biological issues behind recent events. Conservation Biology. 7(4):950-953. Hays, D. 1994. Conservation Biologist, Nongame Program, Department of Wildlife, Olympia, Washington. Unpublished Data in a letter to Dave Dunbar, Ministry of Environment, Lands and Parks, Surrey, BC. Dated February 8, 1994. Herter, Dale R. Wildlife biologist, Raedeke Associates Scientific Consulting, Seattle, WA. Holling, C.S. 1959. Some characteristics of simple types of predation and parasitism. Canadian Entomologist. 91:385-398. Holthausen, R.S., M.G. Raphael, K.S. McRelvey, E.D. Forsman, E.E. Starkey and D.E. Seaman. 1995. The Contribution of Federal and Non-Federal Habitat to the Persistence of the Northern Spotted Owl on the Olympic Peninsula, Washington: Report of the Reanalyis Team. General Technical Report PNW-GTR-352. Portland, Oregon: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 68pp. Hutchings, J., C. Walters and R. Haedrich. 1997. Is scientific inquiry incompatible with government information control? Canadian Journal of Fisheries and Aquatic Sciences. 54(5):1198-1210. Kennedy, P.L. 1997. The northern goshawk (Accipiter gentilis atricapillus): Is there evidence of a population decline? Journal of Raptor Research. 31(2):95-106. 85 Krajina, V.J. 1959 Biogeoclimatic zones in British Columbia. University of British Columbia Botanical Series No. 1, Vancouver. 47 pp. Krebs, C.J. 1989. Ecological Methodology. Harper and Row, Publishers, New York. 654 pp. Lahaye, W.S., R.J. Guitierrez and H. Resit Akcakaya. 1994. Spotted owl metapopulation dynamics in southern California. Journal of Animal Ecology. 63:775-785. Lamberson, R.H., R. McKelvey, B.R. Noon and C. Voss. 1992. A dynamic analysis of northern spotted owl viability in a fragmented forest landscape. Conservation Biology. 6(4):505-512. Lamberson, R.H., B.R.Noon, C.Voss and K.S. McKelvey. 1994. Reserve design for territorial species: the effects of patch size and spacing on the viability of the northern spotted owl. Conservation Biology. 8(1):185-195. Lande, R. 1991. Population dynamics and extinction in heterogeneous environments: the northern spotted owl. Pages 566 - 580 in C M . Perrins, J.D. Lebreton, and G.J.M. Hirons (eds.). Bird population studies. Oxford University Press, Oxford, U.K. Lande, R. 1987. Extinction thresholds in demographic models of territorial populations. American Naturalist. 130:624-635. Lande, R. 1988. Demographic models of the northern spotted owl (Strix occidentalis caurina). Oecologia. 75:601-607. Lindenmayer, D. and H. Nix. 1993. Ecological principles for the design of wildlife corridors. Conservation Biology. 7(3):627-630. L0mnicki, A. 1992. Population ecology from the individual perspective. Pages 3 - 17 in Individual-based models and approaches in ecology. DeAngelis, D.L. and L.J. Gross (eds.) 1992. Chapman and Hall, New York and London. LUCO - Land Use Coordination Office. 1997. Spotted owl management plan. Summary report. Victoria, BC. Marcot, B.G. and R. Holthausen. 1987. Analyzing population viability of the spotted owl in the Pacific Northwest. Transactions of the North American Wildlife and Natural Resources Conference. 52:333-347. 86 McKelvey, R. 1996. Viability analysis of endangered species: a decision-theoretic perspective. Ecological Modelling. 92(2-3):193-207. McKelvey, K.S., B.R. Noon, R.H. Lamberson. 1992. Conservation planning for species occupying fragmented landscapes: the case of the northern spotted owl. In: P.M. Karieva, J.G. Kingsolver, and R.B. Huey (eds.). Biotic interactions and global change. Sinauer Associates, Inc. Sunderland, MA. pp. 424-450. Metz, J.A.J, and O. Diekmann (eds.) 1986. The dynamics of physiologically structured populations. Lecture Notes in Biomathematics 68. Springer-Verlag, Berlin. Miller, G.S., R.J Small, and E.C. Meslow. 1997. Habitat selection by spotted owls during natal dispersal in western Oregon. Journal of Wildlife Management. 61(1): 140-150. Miller, G.S. 1989. Dispersal of juvenile northern spotted owls in western Oregon. M.S. thesis, Oregon State University, Corvallis, Oregon. 139pp. Miller, G.S. and E.C. Meslow. 1985. Dispersal data for juvenile spotted owls: the problem of small sample sizes. Pages 69-73 in R.J Gutierrez and A.B. Carey (eds.). Ecology and management of the spotted owl in the Pacific Northwest Forest and Range Experimental Station. General Technical Report PNW-GTR-185, Portland, Oregon. Mills, L., R. Fredrickson and B. Moorhead. 1.993. Characteristics on old-growth forests associated with northern spotted owls in Olympic National Park. Journal of Wildlife Management. 57(2):315-321. Ministry of Environment, Lands and Parks (MELP). 1997. Request for proposals for spotted owl inventory in the Lillooet Forest District. Ministry of Environment, Lands and Parks, Southern Interior Region, Kamloops, BC. 11 pp. Ministry of Forests and Ministry of Environment, Lands and Parks. 1995. Economic evaluation of the northern spotted owl management options for British Columbia. 35pp. Noon, B.R. and C M . Biles. 1990. Mathematical demography of spotted owls in the Pacific Northwest. Journal of Wildlife Management. 54(1): 18-27. Paton, P., C. Zabel, D. Neal, G. Steger, N. Tilghman, and B. Noon. 1991. Effects of radio tags on spotted owls. Journal of Wildlife Management. 55(4):617-622. 87 Pojar, J., K. Klinka and D.V. Meidinger. 1987. Biogeoclimatic classification in British Columbia. Forestry Ecology and Management. 22:119-154. Ripple, W.J., P.D. Lattin, K.T. Hershey, F.F. Wagner and E.C. Meslpw. 1997. Landscape composition and pattern around northern spotted owl nest sites in southwest Oregon. Journal of Wildlife Management. 61 (1): 151-158. Shaffer, M.L. 1985. The metapopulation and species conservation: the special case of the northern spotted owl. Pages 86-89 In R.J Gutierrez and A.B. Carey (eds.). Ecology and management of the spotted owl in the Pacific Northwest Forest and Range Experimental Station. General Technical Report PNW-GTR-185, Portland, Oregon. SORT (Canadian Spotted Owl Recovery Team). 1994. Management Options for the Northern Spotted Owl in British Columbia. 180pp. Taylor, A.D. 1990. Metapopulations, dispersal, and predator-prey dynamics: an overview. Ecology. 71:429-433. Thomas, J.W., E.D. Forsman, J.B. Lint, E.C. Meslow, BR. Noon and J. Verner. 1990. A Conservation Strategy for the Northern Spotted Owl. Report of the Interagency Scientific committee to Address the Conservation of the Northern Spotted Owl (Portland, Oregon). 427pp. USDI. 1992. Recovery Plan for the Northern Spotted Owl - DRAFT. US Department of the Interior. Washington, D.C. 662pp. U.S. Forest Service. 1986. Draft supplement to the environmental impact statement for an amendment to the Pacific Northwest regional guide. Two volumes. U.S. Forest Service, Portland, Oregon. 390pp. U.S. Forest Service. 1988- Final supplement to the environmental impact statement for an amendment to the Pacific Northwest regional guide. Two volumes. U.S. Forest Service, Portland, Oregon. 897 pp. Walters, CJ . 1986. Adaptive Management of Renewable Resources. MacMillan Publishing Company, New York. 374 pp. Walters, C. and L. Huato. Personal Communication. University of British Columbia. 88 Ward, J.P. Jr., A.B. Franklin and R.J. Gutierrez. 1991. Using search time and regression to estimate abundance of territorial spotted owls. Ecological Applications. 1(2):207-214. Yaffee, S.L. 1994. The Wisdom the Spotted Owl: Policy Lessons for a New Century. Island Press. Washington, D.C./Covelo, California. 430pp. Zabel, C, K. McKelvey and J. Ward Jr. 1995. Influence of primary prey on home-range size and habitat use patterns of northern spotted owls (Strix occidentalis caurina). Canadian Journal of Zoology. 73(3):433-439. 89 Appendix I - Conceptual structure of the model Flow chart of the model and connections between components 90 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0074818/manifest

Comment

Related Items