Field datasets

Pacific Northwest Species Change

Focus species for the study, their codes and proportion of plots where the species is present or absence is shown in the table below.

For more information on the plot locations click on the species latin name.

Species	Common name	American species code	Presence	Absence	% presence	% absence	Total sites (N)
Pseudotsuga menziesii	Douglas-fir	202	6643	16128	29.2	70.8	22771
Thuja plicata	Western redcedar	242	3107	19664	13.6	86.4	22771
Tsuga heterophylla	Western hemlock	263	2106	20665	9.3	90.8	22771
Tsuga mertensiana	Mountain hemlock	264	473	22298	2.1	97.9	22771
Pinus contorta	Lodgepole pine	108	4742	18029	20.8	79.2	22771
Pinus ponderosa	Ponderosa pine	122	2393	20378	10.5	89.5	22771
Picea sitchensis	Sitka spruce	98	364	22407	1.6	98.4	22771
Pinus albicaulis	Whitebark pine	101	644	22127	2.8	97.2	22771
Chamaecyparis nootkatensis	Alaska yellow cedar	42	887	21884	3.9	96.1	22771
Abies procera	Noble fir	22	82	22689	0.4	99.6	22771
Picea breweriana	Brewer's spruce	92	6	22765	0.0	100	22771
Chamecyparis lawsoniana	Port Orford-cedar	41	23	22748	0.1	99.9	22771
Pinus jeffreyi	Jeffrey pine	116	350	22421	1.5	98.5	22771
Picea engelmannii	Engelmann spruce	93	1365	21406	6	94	22771
Larix occidentalis	Western larch	73	670	22101	2.9	97.1	22771
Juniperus occidentalis	Western juniper	64	456	22315	2.0	98.0	22771
Juniperus scopulorum	Rocky Mt. Juniper	66	301	9217	3.2	96.8	9518
Quecus garryana	Garry Oak	815	118	9400	1.2	98.8	9518
Acer macrophyllum	Big leaf maple	312	480	22291	2.1	97.9	22771
Calocedrus decurrens	Incense cedar	81	561	22210	2.5	97.5	22771
Abies lasiocarpa	Subapline fir	19	2369	20402	10.4	89.6	22771
Abies amabilis	Pacific silver fir	11	764	22007	3.4	96.6	22771
Abies grandis	Grand fir	17	598	22173	2.6	97.4	22771
Alnus rubra	Red alder	351	879	21892	3.9	96.1	22771

FIA:42 BC:Cy

FIA:312 BC:Mb

FIA:92

FIA:202 BC:Fd

FIA:93 BC:Se

FIA:815 BC:Qg

FIA:17 BC:Bg

FIA:81

FIA:116

FIA:108 BC:Pl

FIA:264 BC:Hm

FIA:22 BC:Bp

FIA:122 BC:Py

FIA:41

FIA:351 BC:Dr

FIA:66 BC:Jr

FIA:11 BC:Ba

FIA:98 BC:Ss

FIA:19 BC:Bl

FIA:263 BC:Hw

FIA:242 BC:Cw

FIA:64

FIA:73 BC:Lw

FIA:101 BC:Pa

Field Datasets

We have developed an innovative approach to modeling tree species distribution by taking advantage of the strengths of both the empirical and process-modeling approaches. Rather than utilizing climatic data directly in mapping species distribution, we interpret the physiological implications of seasonal variation in water availability, humidity deficits, excess or suboptimal temperatures, and the impact of frost on a generic tree species. By extracting information on how these environmental variables differentially affect the performance of selected species across their current ranges, we are able to predict where a species might be stressed under a changing climate.

To model the distributions, we utilize regression-tree modeling technique to rank dependent variables in order of importance based on least squares analysis (Melendez et al. 2006). This approach can be considered as a sequence of yes/no queries partitioning dependent variables for a given species into homogeneous sets of responses. Regression tree analyses are increasingly applied in ecological research (e.g., De’ath et al. 2000; Schwalm et al. 2006) as they are independent of statistical distributions and have proved to be particularly suited for dealing with collinear datasets, potentially insignificant predictors and outliers (Schwalm et al. 2006). A prerequisite of our generic modeling approach is that it must be able to predict, with reasonable accuracy, the current distribution of individual species, and to do so automatically.

The 3-PG model has generally proven able to predict maximum canopy LAI over the full range of forested environments in Oregon with reasonable

predictions in other Pacific NW states (Waring et al. 2002, Waring et al. 2005). There has been considerable effort to use remote sensing techniques

to estimate tree mortality caused by insects, disease, and fire. Quickbird, Landsat, and video imagery have all served to estimate mortality from bark

beetles and defoliating insects (Franklin et al. 1995). In this work, we will take advantage of a variety of remote sensing techniques, including

U.S. Forest Service aerial mapping and a MODIS-derived disturbance index, to confirm the extent of insect damage across the westernUnited States

to validate his and our model predictions of where climatic variation could induce stress sufficient to support a major outbreak (Nemani 2008).