This study examines the potential population-level impacts of climate change on Bicknell's Thrush, a vulnerable montane forest songbird. Using occupancy models, the researchers predict that a 1°C increase in mean July temperature would reduce available Bicknell's Thrush habitat by 67%, fragmenting it into fewer (29 vs. 128) and larger patches. This loss of habitat is expected to decrease the species' occupancy, colonization rates, and increase extinction rates. While some persistence may be possible in remaining high quality habitat patches under small amounts of warming, the authors conclude that over 1°C of warming could threaten the long-term persistence of Bicknell's Thrush in Vermont.
11. Multi-season Occupancy Model (MacKenzie et al. 2003 Ecology) Population Parameters Occupancy Colonization Detection Extinction Methods - Model Between seasons Colonization or extinction 1,2,3 1,2,3 1,2,3 1,2,3 Surveys Seasons 1 2 3 4 Occupancy & detection Within seasons
12. Habitat Covariates Local = boreal basal area (BBA) (Beckage et al. 2008 PNAS) Landscape = patch size (Lambert et al. 2005 Wilson Bull) Methods - Model
14. Landscape: + 1° C to MJT envelope Local (BBA): MJT inc. over time (amnt. of time for 1° C inc. = 7 yrs.) Northeast Climate Data (www.northeastclimatedata.org) A1 Scenario Methods – Warming Predictions
15. Current & future (+1°C) maps of predicted habitat amount (landscape) & suitability (patch) Compared amount and quality of habitat Estimated occupancy, colonization, and extinction now and with 1°C of warming Methods – Population Predictions
16. Occupancy (Boreal Basal Area * Patch Size) Colonization (Boreal Basal Area) Extinction (Boreal Basal Area + Patch Size) Detection (Survey + Patch Size) Methods - Model
17. No time lag – once pixel moves out of MJT envelope, it is no longer available as habitat Boreal basal area is a good indicator of local habitat quality Beckage et al. 2008 model realistically represents changes in the composition of local habitat over time Lambert et al. 2005 model is a realistic representation of current potential habitat availability at the landscape scale Model Assumptions
18. Results - Landscape # of patches decreased from 128 to 29 Total amount of habitat reduced by 67%
19. Results - Landscape # of patches decreased from 128 to 29 Total amount of habitat reduced by 67%
20. Results - Local Avg. patch size increased 44.6% (90.5 to 130.8 ha) Remaining patches = higher quality BBA 29.1 - 60.6 m2/ha Total BBA decreased 31.3%
21. Results - Local Avg. patch size increased 44.6% (90.5 to 130.8 ha) Remaining patches = higher quality BBA 29.1 - 60.6 m2/ha Total BBA/patch decreased 31.3%
27. Colonization least influenced by changes in landscape scale features, although total area available for colonization decreased by 66.8% Results - Population parameters
30. Population implications are primarily related to the total amount of habitat lost Composition of montane patches altered – less so than overall amount of habitat Loss of lower elevation sites likely has less of an impact - represent marginal habitat (?) Conclusions
31. Under small amounts of warming, Bicknell’s Thrush may be able to persist in the remaining high quality patches > 1°C of warming may call into question the long term persistence of this species in Vermont Conclusions
36. 70+ year dataset on Mt. Washington (NH) shows that warming at high elevations may be occurring at a slower rate than at lower elevations (Seidel et al. 2009 Arctic, Antarctic, and Alpine Research) Hope for Bicknell’s?
Montane systems in the NE are relatively uncommon and their existence depends on cooler climates at high elevations, making them particularly vulnerable to climate change.
Since BITH is a high-elevation specialist, moving up in elevation is not a long-term viable solution
Uses detection/non-detection data to estimate the following parameters. 88 sites across VT were surveyed 1-3 times in 2006 & 2007
Boreal basal area measured at 88 sites across VT. Patch size calculated from the BITH potential habitat model .Ran 65 models => every combination of local and landscape for each parameter (single, additive, interaction). Patch size calculated from potential habitat model (potential habitat identified by presence-absence of BITH, cutoff by elevation which changes by latitude, and then pixels of coniferous dominated forest (landcover map)
Raster of MJT for current BITH distribution
AICc weight =0.16. Top 6 models within 2 AICc points of top model all very similar. Used as input variables for the metapopulation model to estimate the following parameters currently and with one degree of warming
Graph of the % change in the avg. parameter value between current & predicted conditions
Further explanation 3rd bullet = there are smaller proportions of habitat conditions favored by this high elevation disturbance specialist