1. Predicting Sapling Recruitment Following Partial Cutting in the Acadian Forest: Using Long-Term Data to Assess the Performance of FVS-NE David Ray 1, Chad Keyser 2, Robert Seymour 1 and John Brissette 3 1 School of Forest Resources, The University of Maine, Orono ME 2 Forest Management Service Center, USDA-FS, Fort Collins, CO 3 Northeastern Research Station, USDA-FS, Durham, NH

2. Outline Background –Motivation –Findings from past work Objectives Methods –Dataset –Analysis Results Conclusions

3. Creation of Stand Structures Over the Past 25-yrs in Maine

4. The Northeastern Variant (FVS-NE) Covers the 14 Northeastern States –Formerly NE-TWIGS (Teck and Hilt 1991) –Lacks a “full” establishment model Newly coded Beta version incorporates some major changes –Small tree height and diameter growth –Background and density dependent mortality –Growth modifier function

5. Assessing Recruitment Dynamics Partial cutting leads to cohort recruitment –Regeneration is prolific in this forest type (Brissette 1996) –Heavy cutting favors intolerant hardwoods; lighter cuts promote tolerant conifers Long-term forecasts require consideration of regeneration/recruitment dynamics Compare performance of the production and beta versions of the Northeastern Variant Provide feedback that can be used to improve model performance

6. Findings From Past Work

7. Penobscot Experimental Forest (PEF) US Forest Service Compartment Study –50 yrs of remeasurement data (numbered trees since the mid-70s) –Inventoried before and after harvests and at approximately 5-yr intervals between harvests –2 reps/treatment (~10 ha units) Tolerant Northern Conifers (BF, RS, EH) Range in silvicultural intensity –From 5-yr selection to commercial clearcutting

8. Live BA Following Partial Cutting at the PEF Observed vs. FVS Predictions Ray, Seymour, and Keyser (2006) Proc. ECANUSA Conference

9. Diameter class midpoint Summary of Net Growth Comparison based on ~25 yr Simulation Runs ~40% above observed production rates (0.5 cd/ac/yr) 2.5 10 14 >16 6

10. Methodology

11. CodeDescriptionCutting cycle (yrs) HarvestsPlot count FDLFixed diameter- limit 20*333 MDLModified diameter- limit 20332 S05Single- tree/small groups 51033 S10Single- tree/small groups 10535 S20Single- tree/small groups 20337 URHCommercial clearcut 30*241 NATUntreated control n/a 20 Characteristics of the Partial-Cut Treatments

12. Nested Plot Design

13. Simulation Run Details Focus on 5-yr runs at the plot level –250 plots; 1,182 plot/interval combinations Calibration of LT diameter growth (≥1-in dbh) Forest wide SI for balsam fir set at 55-ft Large regeneration only- issues with SDI max Regeneration specified by mid-point of height class interval –Beta model equations used to derive species specific heights for trees ≥ 4.5-ft tall but <0.5-in dbh Key in on saplings crossing the 0.5-in dbh threshold (1-in dbh class)

14. Performance Criteria Presence absence of new recruits Compare diameter distributions Rates of sapling recruitment and mortality (BA, ft 2 /ac/5-yr) Correlation analysis between residuals and plot attributes

15. Results

16. Large Tree Calibration Statistics

17. The Nested Plots Recruitment was observed on 55% (653/1,182) of the plot/interval combinations –Tall regeneration was present on 68% of plots where recruitment was observed (‘appeared’ on 32%) Simulated recruitment was limited to plots with large regeneration present (n=729) –Recruitment was observed on 61% of these plots –PROD FVS predicted recruitment on 35% –BETA FVS predicted recruitment on 68% Agreement between observed and predicted –For PROD FVS was 29% –For BETA FVS was 56%

18. more on The Nested Plots Backwards extrapolation of observed diameter growth –10% may have been smaller than large regeneration (URH- intolerant broadleafs) Sufficient abundance of large regeneration –96% of plot/intervals had more than enough to account for observed recruitment

19. Results- Recruitment Basal Area PROD FVS = -0.43 + 0.59 (OBS); r 2 =0.10 BETA FVS = 0.26 + 1.66 (OBS); r 2 =0.13

20. Results- Diameter Distribution Recruitment

21. Results- Recruitment & Mortality 4 X Recruitment BA Recruitment/Ingrowth

22. Results- Residual Analysis -0.67*** -0.18*** -0.49*** -0.83*** -0.08* -0.18*** 0.14*** -0.16*** 0.10** Recruitment

23. Summary of Findings

24. Conclusions I Difficult hoop for the model to pass through Large tree calibration statistics were closer for BETA FVS than PROD FVS Recruitment rates were underestimated by PROD FVS (~50%) and overestimated by BETA FVS (~100%) relative to that observed on partially cut plots at the PEF (~2 ft 2 /ac/5yr) Mortality rates were too high, particularly for BETA FVS

25. Conclusions II The changes implemented in BETA FVS should improve model performance Model biases were related to –Large regeneration density for BETA FVS (strong) –QMD, % SW regen, Harvests for PROD FVS (weak) Resetting GMOD to 0.5 (from 0.15), too high? –Shade tolerant saplings can just sit there in the understory (GMOD by shade tolerance?) The Northeastern Variant covers a large geographic range; the Acadian Forest Region represents a relatively small part

26. Acknowledgements US Forest Service –PEF Dataset –Support with FVS Northeastern States Research Cooperative (NSRC) UMO School of Forest Resources