Forestry Futures Trust
Enhanced Forest Productivity Science Program
Project Work Report
A: PROJECT INFORMATIONProject Title:
MULTI-COHORT FOREST MANAGEMENT IN NORTH-EASTERN ONTARIO: COHORT CLASSIFICATION, ASSOCIATED WILDLIFE COMMUNITIES, AND PROJECTED STAND DYNAMICS / Project Number:
035-2-R1
Reporting Period:
16March 2007 to31 March 2008 / Duration of Project (Yrs)
3
Interim Report (15 October):
15 October 2007 / Annual Report (15 March):
15 March 2008 / Final Report (for all years):
August 2009 (see cover letter)
BUDGET REPORTING SUMMARY
Fiscal Year / EFPS Funding Approved
(Notice of Project Approval) / EFPS Funding Spent (Table 1 Total) / EFPS Funding Invoiced
(Total to Date) / Matching/In-Kind Funding Provided
(Table 2 Total)
2005/2006 / 102,000 / 97,300 / 97,300 / 218,500
2006/2007 / 102,000 / 101,108 / 101,108 / 199,250
2007/2008 / 101,000 / 91,182 / 91,182 / 181,500
TOTAL / 305,000 / 289,590 / 289,590 / 599,250
Project Leader (Name, Title and Organization)
Jay R. Malcolm, Associate Professor, Faculty of Forestry, University of Toronto
Mailing Address:
33 Willcocks St., Faculty of Forestry, University of Toronto, Toronto, ON M5S 3B3
Phone:
416-978-0142 / Fax:
416-978-3834 / Email:
B: SUMMARY OF WORK COMPLETED
Research Progress:
- Please summarize project progress for this fiscal year.
- Please summarize project progress since the project commenced.
- Highlight milestones as listed in the original project proposal and explain how they have been accomplished.
- Explain any deviations from the original project proposal.
- Provide revised milestones if necessary.
- Research results need not be included.
The pursuit of multi-cohort management(MCM) in Ontario entails significant research needs, including the development of cohort classification systems that can be applied at stand and management unit levels. Specific objectives of this subproject (Ben Kuttner, Ph.D.Candidate) are to: 1) develop a regionally-based, foresttype specific system of cohort structure classification that focuses on live tree diameters and height diversitiesas measured in growth and yield plots (MW2, SF1, SB1) and 2) test the utility of remotely sensed light detection and ranging (LiDAR) data to recover structure classes and extend the classification approach to the stand and management unit levels. Key questions include: Do different forest types require separate cohort classification approaches? How do forest age, productivity, and disturbance history influence structure-based cohort classes? Do LiDAR data have the potential to re-create, and possibly improve upon, ground based cohort classifications?
The focus of Ben's research over the past year has been twofold: 1) extension, refinement, and testing of a ground-based multi-cohort forest structure classification system that has been under development since the beginning of the projectand 2) development of LiDAR/GIS data management system to provide a first step in addressing remaining project objectives. Using data from the provincial network of growth-and-yield monitoring plots located across Ecoregion 3E in northeastern Ontario, a total of 416 sites in three forest types (SB1, SF1, & MW2 standard forest units) were included in the final ground-based multi cohort structural classification. The first objective of this study was realized in that an objective and quantitative structural classification approach was developed for boreal spruce (SB1) and mixedwood (SF1 and MW2) analysis groups that separates relatively dense, even-sized stands (cohort I) from increasingly irregular and complex uneven-sized stands (cohorts II and III) in accord with multi-cohort forest management concepts.
Moreover, following the classification, relationships between cohort class and stand age, productivity, and disturbance history were examined in detail and results of these post-classification analyses were presented at a IUFRO conference in July, 2007 (Kuttner et al. 2007). The novel methods that Ben developed for structural classification, and his classification results, are now in draft form as chapter one of his Ph.D. dissertation. Post-classification analyses, including the development of productivity indices and logistic regression models to explain relative influences of stand age and productivity on structure class membership, will constitute a second chapter. Manuscripts based on these chapters will be prepared and submitted for publication in peer-reviewed journals in 2008-2009.
Simultaneously, Ben has made considerable progress towards achieving the second project objective. A proposal to the Ontario Centres of Excellence to further develop software tools and applications to test and enable LiDAR-based multi-cohort structural classification was accepted in 2007. The OCE funded project: “Multi-Cohort Forest Classification using LiDAR” (proponents Malcolm, Durst, Kuttner, Parton, Rudy, and Woods) provides three years of funding (from April 2007 to March 2010) to further enhance the LiDAR component of our EFPS research by providing the opportunity to expand the scope and accelerate the application of our LiDAR goals. The GIS system under development is designed to support the examination of vegetation height density, variability, and grain (semivariance) in relation to cohort structure classes at multiple spatial scales, by partitioning the continuous LiDAR data into 400m2 raster cells. These cells are centered on the classified, individual growth plots and are arranged in a grid that extends to the FRI stand boundaries surrounding each study site. Development of this custom GIS software is ongoing. To date, it has incorporated outputs from multiple commercial LiDAR processing and GIS software packages. It also has required the development of custom software scripts and tools, including plug-ins, visualization tools, and several user interfaces to manage and analyze the LiDAR data in relation to cohort classified ground plots. Data preparation and management is not a trivial task given the huge amounts of LiDAR data involved. Analysis of the LiDAR dataset to address the second EFPS project objective is ongoing (anticipated completion data: fall 2008).
SORTIE subproject:
Principal objectives of this subproject (Mark Vanderwel, Ph.D. Candidate) are to: 1) use the stand dynamics model SORTIE to determine the degree to which various partial harvesting treatments can maintain structural complexity in boreal mixedwood stands in the short and long term and 2) develop an individual-based model to simulate the manner in which a putative old-growth specialist, the southern red-backed vole (Clethrionomys gapperi), responds to stand-level habitat structure created by partial harvesting.
To date, Mark has made excellent progress on both of these objectives. Collaborators at l’Université du Québec à Montréal have completed a SORTIE parameter file that simulates growth, mortality, and succession of mixedwood stands in the boreal claybelt region. To date he has undertaken initial explorations of the model’s behaviour under a preliminary parameterization, and hasconducted tests both with and without partial harvesting.
To examine stand structure dynamics within SORTIE, Mark added new code to compute a suite of indices relating to the abundance, condition, and spatial distribution of structures within simulated mixedwood stands. In a test of the model’s utility, he ran simulations wherein changes in these structural indices were followed over a 250 year period. Multivariate analysis of model output recovered four distinct stages of stand development that corresponded well to current understanding of boreal mixedwood stand dynamics and provide good verification of model performance.
Building further upon the current capacities of the model, Mark has focused model development efforts towards enhancing SORTIE’s ability to simulate the dynamics of dead wood. He has written, tested, and distributed a new SORTIE module to track inputs and changing accumulations of downed woody debris in five decay classes, two size classes, and two species groups over time. Using maximum-likelihood techniques, he also has completed a data analysis to estimate model parameter values governing decay class transitions of woody debris in the boreal claybelt region. These developments enable the model to simulate changes in the quality and quantity of downed wood, which is considered to be an important component of wildlife habitat.
Using data collected on the companion SFMN-funded project ("Dynamics of woody debris in eastern boreal forests: implications for carbon and wildlife management"), Mark has begun scale-dependent, maximum likelihood modelling of local (within-grid) responses of red-backed voles to variation in key habitat features such as: downed wood of different decay classes, shrub cover, and moisture distributions (as revealed through compositional changes in shrub communities).
Small mammal subproject:
Objectives of the small mammal sub-project (Charlotte Sharkey, M.Sc.F.) were to:
1) based on the distribution of live stem diameters, classify black spruce (SB1) and mixedwood (MW2) stands according to their multicohort structure and examine the stability of such classifications as a function of sampling intensity;
2) explore the utility of such diameter distributions (as described by 2-parameter Wiebull functions) in succinctly describing: a) other stand structural features (including stand 3-dimensional structure and standing and downed coarse woody debris) and b) small mammal communities and compare the performance of the Weibull functions against other potential drivers of variation such as stand age, productivity, and tree species composition;
3) undertake the above analysis for two different stem diameter cutoffs (2.5- and 10-cm) and compare the performance of the two cutoffs.
Charlotte, who successfully defended her thesis in January of 2008, studied two broadly-defined forest types in boreal northeastern Ontario in the summer of 2006: upland mixedwood (n = 18 sites) and lowland black spruce (n = 14 sites). Sites were selected according to their tree cohort structure, stand composition/productivity, and age.
As a measure of stand cohort structure, she focused on parameters of a 2 parameter Weibull distribution fit to the diameter distribution of live trees. She also measured a wide variety of additional stand structural features that have been shown to be important to a wide variety of flora and fauna, including basal area of snags (by decay class), horizontal and vertical distribution of foliage, volume by decomposition class of DWD, and shrub stratum openness. Age since last stand-replacing disturbance was determined from the Forest Resource Inventory (FRI). Site productivity was based on Northeastern Ontario Forest Ecosystem Classification vegetation types (FEC). Not surprisingly, black spruce sites exhibited much less variation in Weibull parameters than mixedwood sites. The scale parameter, in particular, showed much less variation, which is to be expected given that black spruce on lowland sites tend to have smaller trees than upland sites.
Charlotte collected detailed information on diameter distributions at a relatively large scale (75 × 75 m), which allowed her to examine relationships between cohort class and sampling intensity. To investigate this relationship, Weibull-based cohort types were determined from randomly-picked samples of 2, 4, and 8 of the 16 prism sweeps per site and compared against the classification as determined from the complete (16-prism sweep) sample. Shapes of cohort 2 and 3 mixedwoods were misclassified in about a quarter of sub-samples with sampling intensity of 2 sub-plots; accuracy improved when four sub-plots were sampled, and with 8 sub-plots although some misclassification still occurred, it was only for a small proportion of sub-samples. In black spruce sites, misclassification of the more structurally-complex cohort types was very high. Misclassification of about a quarter of these sites persisted at a sub-sample size of 8 prism sweeps.
To explore whether the Weibull parameters were capable of summarizing other structural features, a variety of statistical techniques were used. Weibull parameters at both diameter limits performed as succinct descriptors of various subsets of structure variables, as did stand age. Weibull shape and scale of stems ≥2.5 cm were the best predictors, however. In mixedwoods, Weibull parameters of stems ≥2.5 cm explained a greater amount of the variance in the structure data than did Weibull parameters of larger stems (≥10 cm dbh) or age, although both sets were significant. In black spruce stands, Weibull parameters at the 2.5 cm diameter limit explained a greater amount of structural variance than did the 10 cm limit or age, and variance explained by the 10 cm limit was not significant.
Whether due to harvesting per se or to an age effect, natural origin mixedwood sites (n = 3) ordinated separately from logged sites; however, wide overlap was shown between mechanized- and horse-logged sites. For black spruce stands, in comparison to natural origin sites, harvested sites were impoverished with respect to a number of structure variables, including snag basal area, volume of DWD, heterogeneous overstories and canopies, vertical foliage complexity, and relatively large trees.
Small mammals were trapped by use of both ground and arboreal live-traps and pitfall traps. Two trapping cycles, each for 3 consecutive nights, took place in each stand: the first in spring/early summer (late May to mid-June), and the second in late summer (August to early September). Pitfalls were set for a week in spring and a week in late summer. Pitfall specimens (mostly shrews) were stored in alcohol and later identified to species based on dentition. Sixteen small mammal species were captured (2,241 individuals), including shrews (5 species), voles (3 species), bog lemmings, deermice, jumping mice (2 species), and squirrels (3 species). In the analyses detailed below, habitat relationships were more prominent for the spring than the summer sample, indicating that habitat relationships were likely masked by dispersal.
In mixedwoods, Weibull shape and scale of stems ≥2.5 cm were significant predictors of small mammal assemblages. Weibull shape and scale of stems ≥10 cm predicted assemblages less strongly. Decomposition of variance revealed that Weibull parameters and canopy height variables explain much of the same variance in small mammal community composition. This was expected because both characterise the cohort structure of a forest. Dead wood variables (volume of early decomposition woody debris, late decomposition woody debris, and basal area of snags) did not significantly explain small mammal community composition. In black spruce in spring, Weibull scale of stems ≥2.5 cm significantly predicted small mammal assemblages. Neither diameter limit, however, clearly outperformed the other. As was the case in mixedwoods, Weibull parameters of stems ≥2.5 cm and canopy height variables explain much of the same variance in small mammal community composition, and Weibull parameters of stems ≥2.5 cm and dead wood structure shared very little explained variance, and dead wood structure was not a significant predictor. Live tree basal area was a significant predictor. When Weibull parameters of stems ≥2.5 cm were analyzed with basal area as a covariate, they did not significantly explain additional variance in either spring or late summer. Of the measured habitat structure variables, basal area thus appeared to be the best predictor.
Decomposition of variance was used to determine effects of forest structure (as measured by Weibull paraneters) per se in comparison to age, site productivity (FEC vegetation type), or percent deciduous composition. For the mixedwood sites, neither stand age nor percent deciduous composition significantly explained spring or late summer mixedwood small mammal community composition. Some of the variance explained by cohort structure (Weibull of stems ≥2.5 cm) was shared with age in spring, which is not surprising because age and aspects of multicohort structure are related. The weighted rank of FEC herb richness, a surrogate for site productivity, significantly explained variation in spring mixedwood small mammal community composition, but explained considerably less variation in late summer. Structure as measured by Weibull parameters thus appeared to be a better overall descriptor of small mammal community composition compared to the other variables. In black spruce stands, neither stand age nor percent deciduous composition significantly explained spring or late summer black spruce small mammal community composition, although in late summer, age was nearly significant (P = 0.073).
Bird subproject:
This subproject (Mike Burrell, M.Sc.F. Candidate) has two major objectives: 1) to investigate the utility of multi-cohort classifications, and in particular the distribution of live tree diameters, in describing boreal bird community variation in the Romeo Mallette Forest and 2) examine the utility of LiDAR in assessing multi-cohort structure and bird habitats.
Sites were selected during early May, 2007, and to the extent possible, were in stands that had OMNR permanent growth plots. Two standard forest units were included (as defined from FRI data):MW2 (mixedwoods with an abundant hardwood component) and SF1 (mixedwoods with a higher softwood component) and all had ages >30 years. An even mix of SF1 and MW2 sites was selected as was a mixture of ages within each. In addition, based on previous classifications by Kuttner (in litt.) and visual assessments, a mix of cohort types was sampled in each forest unit. In total, 45 sites, of which 23 and 22 were in stands identified as MW2 and SF1, respectively, were selected for sampling.
Birds were sampled in May through early July, 2007,via two techniques: playbacks (conducted for earlyseason, cavitynesting species) and passive point counts. Playback sampling for cavity nesters (Yellow-bellied Sapsucker, Downy Woodpecker, Hairy Woodpecker, Pileated Woodpecker, Northern Flicker, Black-capped Chickadee, Boreal Chickadee, Brown Creeper, and Red-breasted Nuthatch) was conducted in late May, 2007. Sites were visited between dawn and 11:00 h on calm, nonraining days and playbacks were conducted using a portable compact disc player attached to a portable Pignose guitar amplifier. The volume was set such that an observer could easily hear the recorded Hairy Woodpecker calls from 100 m away. The playback recording consisted of sequential recordings of calls and drumming (for woodpeckers) of the various specieswith a period of silence between each. Individuals observed or heard during the playback were recorded, as were their estimated locations (i.e., within or >100 m from the speaker). Site order was randomized, with the proviso that sites nearby to each other were sampled in clusters. Point counts were conducted between early June and early July, 2007. Each site was visited three times between dawn and 9:30 h on calm mornings with no precipitation. In order to avoid sampling biases associated with the timing of sampling, each site was visited once during: 1) each third of the sampling season, 2) each of early, mid, and late parts of the morning sampling session, and 3) each of the early, mid, and late sampling parts of the one-third-season sampling periods. Point counts consisted of two, backtoback, fiveminute listening periods during which time individuals seen or heard were identified to species and their distance from the point count centre estimated (i.e., within or >100m from the point count centre).