\documentclass[twocolumn,letterpaper,sort]{article}
\pagestyle{myheadings} \usepackage{natbib}
\usepackage{../mcfnsSep2012}\usepackage{multicol}
%\usepackage[OpenPeerReview]{McfnsDraftcopy}
\usepackage{amsmath,bm,url,amssymb} \usepackage{graphicx,epstopdf,textcomp} \usepackage{setspace,lineno}
\usepackage{color} \usepackage[breaklinks]{hyperref}
\usepackage{rotating} % begin {sidewaystable} instead of begin {table}
\setcounter{topnumber}{2}
\setcounter{bottomnumber}{2}
\setcounter{totalnumber}{4}
\renewcommand{\topfraction}{0.85}
\renewcommand{\bottomfraction}{0.85}
\renewcommand{\textfraction}{0.15}
\renewcommand{\floatpagefraction}{0.7}
%\usepackage[breaklinks,pdfstartview= {FitH -32768},pdfborder= {0 0 0},bookmarksopen,bookmarksnumbered]{hyperref}
%\usepackage {ogonek} %\linenumbers %\pagewiselinenumbers
\widowpenalty=10000 \clubpenalty=10000
%______________________________________________________________________
% Define MCFNS variables:
% Moved to mcfns.sty like the \issueyear \publish	 {Feb.~00} 		%Still not sure if Issue Manuscript publication date
\setcounter {page} {135} \def\issueno {2}
\def\editors	 {Editor:~\href{mailto:editors@mcfns.com}{Randall~S~Morin}}
\def\submit 	 {Apr.~2,~2013} 	%Submission date can be different than the issue year \issueyear
\def\accept 	 {Jul.~9,~2013} 	%The works should be Accepted & Published in the year of the Current_Issue \issueyear
\def\lasterrata	 {Aug.~15,~2013} 	%Last Errata date can be different than the Issue-Year \issueyear
\def\citename	 {Tavernia} 		%"Author"
\def\citeemail	 {btavernia@gmail.com} 	% Use later: {\href{mailto://\citeemail} {FirstName \citename}}
\def\citeetal	 {~et~al.} % {} %or {} %for a single author; or

\author{
     {\href{mailto:\citeemail}{Brian~G~\citename}}$^1$, %Change only 1st name of 1st author
		 {\href{mailto:mdnelson@fs.fed.us} {Mark~D~Nelson}}$^2$,
		 {\href{mailto:goerndtm@missouri.edu} {Michael~E~Goerndt}}$^3$,
}
\affiliation {
%---------------
\large\scshape {
		 {\href{mailto:bfwalters@fs.fed.us} {Brian~F~Walters}}$^2$,
		 {\href{mailto:christoney@fs.fed.us} {Chris~Toney}}$^4$
} \\
%---------------
%
% the lines above are for multi-line multiple authors - braking \author causes problems
%
%---------------
\small $^1$\it{\href{http://harvest.cals.ncsu.edu/biology/}{Department of Biology, North Carolina State University, Raleigh, NC, USA}} \\
\small $^2$\it{\href{http://www.nrs.fs.fed.us/}{USDA Forest Service, Northern Research Station, Saint Paul, MN, USA}} \\
\small $^3$\it{\href{http://www.snr.missouri.edu/forestry/}{Department of Forestry, University of Missouri, Columbia, MO, USA}} \\
\small $^4$\it{\href{http://www.fs.fed.us/rmrs/}{USDA Forest Service, Rocky Mountain Research Station, Missoula, MT, USA}}
}
\def\yourtitle
 {{
Changes in forest habitat classes under alternative climate and land-use change scenarios in the northeast and midwest, USA
}} %need double {{for \\ e.g.: {{Title \\ Subtitle}}
\def\yourkwords
 {
Wildlife Habitat; Bioenergy; Biomass Harvest; Climate Change; Young Forest; Early Successional Habitat; FIA; Forest Projections.
}
\def\yourabstract
 {
Large-scale and long-term habitat management plans are needed to maintain the diversity of
habitat classes required by wildlife species. Planning efforts would benefit from assessments of potential
climate and land-use change effects on habitats. We assessed climate and land-use driven changes in
areas of closed- and open-canopy forest across the Northeast and Midwest by 2060. Our assessments
were made using projections based on A1B and A2 future scenarios developed by the Intergovernmental
Panel on Climate Change. Presently, forest land covers 70.2 million ha and is evenly divided between
closed- and open-canopy habitats. Projections indicated that total forest land would decrease by 3.8 or
4.5 million ha for A2 and A1B, respectively. Within persisting forest land, the balance between closed and
open-canopy habitats depended on assumed harvest rates of woody biomass. Standard harvest rates
led to closed-canopy habitat attaining a slight majority of total forest land area. Intensive harvest rates
resulted in the majority of forest land being in open-canopy habitat for A1B or maintained the even
split between closed- and open-canopy habitats for A2. Ultimately, managers need to identify benchmark
habitat conditions informed by historical conditions and wildlife population dynamics and plan to meet
these benchmarks in dynamic forest landscapes.
} %----------------------------------------------------------------------
% \usepackage[english,russian,polish]{babel}
% THE REST SHOULD BE AUTHOMATIC Go To the first Section
\title {\Large\bf\uppercase\yourtitle}
\begin{document} \markright {\hfil {{{\href {mailto://\citeemail} {\citename} \citeetal}} ~(\issueyear)/\mcfnshead}}
\twocolumn[
\begin{@twocolumnfalse}
\hypersetup {pdftitle= {\mcfnshead},pdfauthor= {\citename~(\issueyear)},pdfsubject= {\yourtitle},pdfkeywords= {\yourkwords}}
\maketitle
\hrule\begin{abstract} \yourabstract\\\\ {\bf Keywords:} \yourkwords
%\\\\ {{\bf S\l {} owa kluczowe:} Polskie slowa kluczowe.}
\end{abstract} \hrule\vspace{.3truein}
\end{@twocolumnfalse}
]

\section{Introduction}
 A significant challenge in natural resources management is providing sufficient
habitat for wildlife species that have diverse and sometimes conflicting habitat
needs (Magules and Pressey, 2000; Noon et al., 2009). Suites of species are associated
with particular forest habitat classes characterized by different compositions, ages,
and structures (Hagan et al., 1997; Patton, 2011). For example, some species (e.g.,
Cerulean warbler, \textit{Setophaga cerulea}) are associated with mature, deciduous
forests whereas others (e.g., Kirtland's warbler, \textit{Setophaga kirtlandii})
are found in disturbance-dependent early successional, coniferous habitat. Successful
conservation and management of species with different habitat associations requires
management plans that are large-scale and long-term in scope; such plans are necessary
to ensure that diverse habitat needs are simultaneously met and maintained through
time (Hamel et al., 2005).



 Changing climate and land-use conditions are expected to drive, in part,
the large-scale dynamics of forest habitat and wildlife distributions over the coming
decades and centuries (Iverson and Prasad, 1998; Matthews et al., 2004; Schwartz
et al., 2006). Ecologists have long recognized large-scale associations between distributional
limits of forest types and wildlife and climate conditions (Booth, 1990; Newton,
2003; Prentice et al., 1992). As climate changes, some forest ecosystems and forest-associated
species might shift their distributions to track hospitable climate conditions, and
others might adapt to new climates (Iverson and Prasad, 1998; Matthews et al., 2004;
Parmesan, 2006). When forest ecosystems and forest-associated species are unable
to move or adapt, their geographic ranges may shrink, or they may become extirpated
from portions of their former range (Parmesan, 2006; Thomas et al., 2004). To aid
long-term conservation and management planning, researchers often model the potential
distributions of forest types and wildlife species under alternative climate change
scenarios (Iverson and Prasad, 1998; Matthews et al., 2004; Schwartz et al., 2006).
The direction and magnitude of modeled distributional changes are often scenario-specific,
and accounting for uncertainty caused by scenario selection is a significant challenge
(Beaumont et al., 2008).



 Within large-scale patterns established by climate, land-use decisions
further modify the extent and configuration of forest types and wildlife species
diversity (Opdam and Wascher, 2004; Pearson et al., 2004). Forest conversion (e.g.,
to urban lands) reduces forest area and potentially fragments forest ecosystems;
this can result in small, extirpation-prone wildlife populations that are too isolated
to be rescued or re-established by immigrants from the surrounding landscape (Robinson
and Wilcove, 1994; Verboom et al., 1991). Land-use conversion may also place remaining
forest habitat in close proximity to anthropogenic land-uses, including agricultural
and urban areas. This proximity can alter food availability, ecological processes,
and biotic interactions in ways that hasten the decline of wildlife populations (e.g.,
via increased nest predation pressure, Donovan et al., 1995). Researchers and managers
recognize the importance of simultaneously assessing climate and land-use change
effects (Pearson et al., 2004). Despite this, few such assessments exist due, in
part, to a lack of climate and land-use projections that share common assumptions
about future demographic, economic, and technological conditions (Bierwagen et al.,
2010).



 Quantity and quality of habitat can be affected by increases in woody biomass
utilization for bioenergy. Woody biomass currently accounts for the greatest share
of bio-energy generation in the U.S. at about 53\% (U.S. Energy Information Administration,
2012). Annual woody biomass consumption for electricity generation is projected to
increase over the next 20 years (U.S. Energy Information Administration, 2012). Energy
markets for woody biomass may lead to the harvesting of stands previously viewed
as being non-commercial (e.g., due to poor wood quality) and to shorter rotation
times between harvests (Janowiak and Webster, 2010). Forest harvest can increase
the area of early successional or young forest habitat, benefiting wildlife dependent
on this habitat (e.g., Annand and Thompson, 1997). Plantations of short-rotation
woody crop species (e.g., \textit{Salix, Populus}) might serve as a source of biomass,
and plantation establishment may have negative or positive effects on wildlife, depending
on the land-use type that is converted. Studies have generally found that plantations
support fewer species of wildlife than unmanaged forest (Moore and Allen, 1999),
but the conversion of non-forest land-use types (e.g., agricultural fields) to plantations
might benefit forest wildlife by increasing total area and connectivity of habitat
(Cook and Beyea, 2000). Ultimately, the value of plantations to wildlife species
depends on how they are managed (Hartley, 2002).



 Energy markets for woody biomass may also provide sufficient incentives
to remove small-diameter woody material in addition to processing logging residues.
Integrated harvest, which includes removal of both logging residues and small diameter
trees, is one of several biomass procurement regimes that may be used to supply woody
biomass for co-firing in electrical plants (Aguilar et al., 2012). The removal of
woody residue previously left behind might negatively affect the abundance or quality
of important microhabitat features, including downed woody material and snags, and
the wildlife that depend on them, although more research is needed (Riffell et al.,
2011). These concerns have led to several states adopting Best Management Practices
(BMPs) specifically designed to minimize impacts of woody biomass removal on water
quality, soils, biodiversity and wildlife habitat (Shepard, 2006; Skog and Stantkurf,
2011).



 Efforts to conserve diverse wildlife communities would benefit from assessments
of current habitat conditions and from projections of climate and land-use change
effects on a suite of forest habitat classes. The Northern Forests Futures Project
(NFFP), a joint effort by the USDA Forest Service and several partners, is projecting
and assessing the potential impacts of climate and land-use changes on forest extent,
composition, and structure across 20 U.S. states in the Northeast and Midwest. These
projections are being made under common sets of assumptions about future demographic,
economic, and technological conditions. For NFFP projections, the USDA Forest Service
is capitalizing on forest composition and structure data provided by its Forest Inventory
and Analysis (FIA) program (Woudenberg et al., 2010). Past studies have used FIA
data to estimate status and trends of coarse-scale habitat characteristics, like
young hardwood forest area, or area of old softwood forest (Schmidt et al., 1996;
Trani et al., 2001). Finer scale habitat information for many forest-associated vertebrate
species can be obtained from more detailed FIA data on tree species, size, and condition,
for both live and dead trees (Nelson et al., 2011). One challenge in using FIA data
for habitat assessments is relating FIA data to habitat classes contained in wildlife
species-habitat matrices. For example, tree canopy cover thresholds are used to characterize
NatureServe (2011) forest habitat domains (hereafter, ``classes''), but historical
and current FIA data do not include estimates of tree canopy cover.



Multi-species management planning is often based on coarse-filter assessments
of the structure, function, and composition of habitat mosaics (Noon et al., 2009;
Schulte et al., 2006). The types and areas of coarse habitat classes in a region
can be used to broadly define the amount of habitat potentially available to broad
suites of species (Beaudry et al.,\textit{ }2010). As part of the NFFP effort, we
used projections of forest conditions in 2060 and ancillary data sets to assess potential
changes in areas of forest habitat classes. The primary objective of this study was
to assess potential changes to wildlife habitat classes over time under a suite of
future scenarios assuming different trajectories for climate, land-use, and biomass
utilization. We defined our habitat classes using thresholds in canopy cover, providing
us with the ability to crosswalk our habitat classes to a wildlife-habitat matrix
created by NatureServe and to report species richness by taxonomic group and conservation
status for each class.




\section{Methods}
\subsection{Region of Interest }
 Our study area encompasses the USDA Forest Service's Eastern Region (Fig. 1). The Eastern Region is heavily forested relative to the whole U.S. (42\% vs. 33\%, respectively) and contains 32\% of the nation's timberland (Shifley et al., 2010).
Approximately 5 million private forest owners hold the majority (55\%) of the region's
forest land and mostly adopt a low intensity management approach to their lands.
The region supports 124 million people (41\% of U.S. population) who depend on forests
to supply a wide variety of ecosystem services (Shifley et al., 2010). Among a variety
of forest resource issues, stakeholders in the region are concerned about the ability
of forest habitat to support diverse wildlife communities (Dietzman et al., 2011).

\begin{figure}[htb] \vspace{-.2in}
\hspace{-.45in}\includegraphics*[width=.575\textwidth]{Fig1.eps}
\caption{Map showing the location of states across the Northeast and Midwest,
USA, an area corresponding to the USDA Forest Service's Eastern Region. States include:
Connecticut (CT), Delaware (DE), Illinois (IL), Indiana (IN), Iowa (IA), Maine (ME),
Maryland (MD), Massachusetts (MA), Michigan (MI), Minnesota (MN), Missouri (MO),
New Hampshire (NH), New Jersey (NJ), New York (NY), Ohio (OH), Pennsylvania (PA),
Rhode Island (RI), Vermont (VT), West Virginia (WV), and Wisconsin (WI).}
\label{fig1}
\end{figure}


 Well-informed
forest and policy management decisions depend on assessments of the potential effects
of alternative decisions on a suite of forest resources and services. As part of
the 2010 Forest and Rangeland Renewable Resources Planning Act (RPA) Assessment,
the USDA Forest Service used alternative future scenarios of climate change, land-use
change, and human population growth (described below) to project forest and rangeland
conditions in 2060 for the Eastern Region (USDA Forest Service, 2012a). We used FIA
data (described below) to estimate current conditions and to project potential future
forest conditions. We applied alternative RPA scenarios to project potential changes
in forest habitat classes needed to sustain regionally diverse wildlife communities.




\subsection{FIA Data}
 FIA's definition of forest land includes components of both land cover
and land use. FIA forest land is defined as having ``\dots at least 10 percent cover
(or equivalent stocking) by live trees of any size, including land that formerly
had such tree cover and that will be naturally or artificially regenerated'' (Woudenberg
et al., 2010, p. 47). Forest land is not developed for a non-forest use such as agriculture,
residential, or industrial use, and includes commercial timberland, some pastured
land with trees, forest plantations, unproductive forested land, and reserved, noncommercial
forested land. FIA forest land requires a minimum area of 0.405 ha and minimum continuous
canopy width of 36.58 m (Woudenberg et al., 2010). FIA sample plots follow a nationally
consistent configuration comprised of a cluster of four fixed-radius circular subplots,
on which land use (e.g., proportion forest cover), tree (e.g., species, height, and
diameter at breast height: DBH, 1.37 m) and other site variables are collected. At
least one FIA plot is selected for each 2400-ha hexagon from a nationally consistent
hexagonal sampling frame. Field crews install, monument, and measure ground plots
if any portion of a plot contains a forest land use (Bechtold and Scott, 2005; Reams\textit{ }et al., 2005). FIA began collecting tree canopy cover data only recently; such data
are absent from historical FIA inventories.

 FIA data from 2004-2008 were used to produce estimates of current conditions,
assigned the decadal label of `2010' and referred to as `baseline'. These same FIA
data were also used to model future projections, by decade, as described below. Estimates
of baseline and future conditions were produced using estimators within PC-EVALIDator
tools in the Northern Forest Futures Database (NFFDB) (Miles et al., 2013).\textit{}


\subsection{Future Scenarios}
 The RPA Assessment used climate (Coulson and Joyce, 2010; Coulson et al.
2010), land-use (Wear, 2011), and population (Zarnoch et al., 2010) projections consistent
with greenhouse gas emission scenarios developed by the Intergovernmental Panel on
Climate Change (IPCC) (USDA Forest Service, 2012a). The analyses by RPA represented
an adaptation of the broad IPCC scenarios to a regional-scale through downscaling
of climate change, economical, and population projections (USDA Forest Service, 2012a;
USDA Forest Service, 2012b; Wear et al., 2013). Emissions scenarios were consistent
with IPCC storylines that assumed different trajectories of change for global populations
and gross domestic product (Tab. 1). For the RPA Assessment, the USDA Forest Service
elected to use IPCC's A1B, A2, and B2 storylines because these captured a range of
potential futures likely to drive variation in natural resources (USDA Forest Service,
2012a). These storylines also had marker emission scenarios that used common assumptions
about driving forces in storylines, were intended to illustrate their respective
storylines, and were subjected to greater scrutiny (USDA Forest Service, 2012a).
While capturing a range of potential futures, these storylines are not tied to specific
policy or management actions. We chose to use emission scenarios from the A1B and
A2 storylines for our forest habitat assessments. We eliminated the B2 storyline
because recent observations of greenhouse gas emissions (Raupach et al., 2007) suggest
that projected emissions under this storyline may underestimate actual emissions.


\begin{table}
\centering
\caption{Projections of global population and global gross domestic product (GDP) associated
with Intergovernmental Panel on Climate Change storylines. Source: USDA Forest Service
(2012a).}
\vspace{3pt}
\begin{tabular}{lrrrr} \hline\hline
Storyline & 2010 & 2020 & 2040 & 2060 \\ \hline\\
\multicolumn{5}{c}{Global Population (millions)} \\
A1 & 6,805 & 7,493 & 8,439 & 8,538 \\
A2 & 7,188 & 8,206 & 10,715 & 12,139 \\
B2 & 6,891 & 7,672 & 8,930 & 9,704 \\\\
\multicolumn{5}{c}{Global GDP (2006 trillion USD)} \\
A1 & 54.2 & 80.6 & 181.8 & 336.2 \\
A2 & 45.6 & 57.9 & 103.4 & 145.7 \\
B2 & 67.1 & 72.5 & 133.3 & 195.6 \\ \hline
\end{tabular}
\label{tab1}
\end{table}



 There are many sources of uncertainty when assessing future changes in
natural resources conditions (Beaumont et al., 2008). The A1B and A2 storylines capture
some uncertainty by representing a range of likely future climate, land-use, and
population conditions. Projected changes in climate for each storyline's emission
scenario depend on the general circulation model (GCM) used to simulate future climate
conditions. For the RPA, the USDA Forest Service projected future climate change
using projections from three GCMs: CGCM 3.1 MR (T47) developed by the Canadian Centre
for Climate Modeling, and Analysis; CSIRO MK 3.5 (T63) developed by Australia's Commonwealth
Scientific and Industrial Research Organization; and MIROC 3.2 MR (T42) developed
jointly by Japan's National Institute for Environmental Studies, Center for Climate
System Research, University of Tokyo, and Frontier Research Center for Global Change.
These GCMs had average or above average sensitivity to greenhouse gas emissions (Randall
et al., 2007), showed a reasonable degree of accuracy when simulating present-day
mean climate conditions (Reichler and Kim, 2008), and produced a range of future
climate conditions. To address uncertainty resulting from choice of IPCC storylines
and GCMs, we assessed potential changes of forest habitat conditions in 2060 under
six scenarios representing unique combinations of A1B and A2 IPCC storylines and
CGCM, CSIRO, and MIROC GCMs. Table 2 summarizes projected changes in climate, land-use,
and population under each scenario. Maps of current and projected changes in climate
conditions can be found in Tavernia et al. (2013).




\subsection{Forest Projections}
 Estimation of future forest conditions relied on the Forest Dynamics Model
(FDM) developed by Wear et al. (2013) (Fig. 2). The FDM is a set of interlinked
submodels which take an existing forest inventory and produce predictions of future
inventories, given assumptions about climate, timber market conditions, and land
use change. Climate, market, and land use assumptions link the forest forecasts to
the IPCC storylines (Wear et al., 2013). The FIA Database (FIADB) provided the foundational
data for the FDM. FIA inventories for each analysis unit were summarized at the plot
level, and only plots classified as forest were maintained so that the Forest Dynamics
Database of beginning inventories reflects the forest land base for 2010. For each
plot for each inventory i) major forest type group was assigned based on forest type,
ii) variables for physical characteristics (slope, aspect, etc.) were retained, and
iii) biophysical attributes (basal area, growing stock volume, number of trees, etc.)
at the population (expanded) and per-acre scale were calculated. Transition, partitioning,
and imputation sub-models were used to predict change in forest plot conditions through
time.

\begin{figure*}[htb]
\vspace{-.6in}
\centering\includegraphics*[width=.75\textwidth]{Fig2.eps} \vspace{-1in}
\caption{Modeling process used to project changes in the areas of forest habitat classes
by 2060. Population, technological, economic, and climate projections served as input
to Forest Dynamics and Land Use Models and drove changes in the extent and composition
of forests from 2010 to 2060 at 5-year increments. Input projections represented
future scenarios resulting from the combination of Intergovernmental Panel on Climate
Change (IPCC) storylines and General Circulation Models. Data and trends from the
USDA Forest Service's Forest Inventory and Analysis Database established 2010 forest
conditions and informed projections of forest change out to 2060. Projected land-use
conditions and forest inventories were summarized in the Northern Forest Futures
Database (NFFDB). A canopy cover algorithm assigned canopy cover estimates to forested
conditions in the NFFDB, enabling forest projections to be translated into forest
habitat classes found within a NatureServe habitat classification system. Green nodes
represent input data, pink nodes indicate modeling steps, and purple nodes are output
products. Flowchart adapted from Wear et al. (2013).}
\label{fig2}
\end{figure*}

\begin{table*}[htb]
\centering
\caption{Mean monthly temperature (T) and precipitation (PPT), land-use
conditions (\%), and population under baseline conditions and six future scenarios
for the Northeast and Midwest, USA. Future scenarios were defined using unique combinations
of Intergovernmental Panel on Climate Change storylines and General Circulation Models
(GCM). Mean climate conditions were calculated using area-weighted means of county-level
values. Baseline and future climate data from Coulson and Joyce (2010) and Coulson
et al. (2010), land-use from Wear (2011), and population projections from Zarnoch
et al. (2010).}\vspace{3pt}
\begin{tabular}{llrrrrrrr} \hline\hline
Storyline & GCM & T (${}^\circ$C) & PPT (mm) & Urban & Forest & Crop & Pasture & Population (mill.) \\ \hline
Baseline & Historical & 9.1 & 80.5 & 9.4 & 41.4 & 39.3 & 9.9 & 124.1 \\
A1B & CGCM & 11.4 & 84.4 & 15.5 & 38.7 & 36.5 & 9.3 & 157.6 \\
 & CSIRO & 11.5 & 79.8 & 15.5 & 38.7 & 36.5 & 9.3 & 157.6 \\
 & MIROC & 13.1 & 72.6 & 15.5 & 38.7 & 36.5 & 9.3 & 157.6 \\
A2 & CGCM & 11.8 & 83.1 & 14.2 & 39.2 & 37.2 & 9.4 & 178.0 \\
 & CSIRO & 11.2 & 86.2 & 14.2 & 39.2 & 37.2 & 9.4 & 178.0 \\
 & MIROC & 12.4 & 75.0 & 14.2 & 39.2 & 37.2 & 9.4 & 178.0 \\ \hline
\end{tabular}
\label{tab2}
\end{table*}



 The transition submodel projects changes in forest type, forest age, and harvesting and
the exogenous climate models project changes in key climate variables. The transition
submodel selects a specific outcome (condition) from among all possible outcomes
through combinations of forest type, forest age, harvest history and climate variables
represented in a probability matrix. The partitioning submodel groups plots from
the 2010 baseline record (we call these donor plots) based upon a set of biophysical
attributes. The plot characteristics that defined the partitioned groups included
forest attributes such as stand age, slope, and ownership, as well as climate variables
such as average temperature and precipitation. Given the conditions for each future
plot, the imputation submodel draws a random donor plot (with replacement) from the
plot's appropriate group as defined by the partitioning submodel. For example, if
the transition probabilities state that a 50-year-old oak-hickory plot will become
a 55-year-old oak-hickory plot (instead of an elm-ash-cottonwood plot or some other
forest type group), then the model imputes what this 50-year-old plot will look like
in five years by randomly picking an oak-hickory plot from the group of existing
55-year-old plots that have similar ecological characteristics. The time step for
imputation was set as the span of years between re-measurements for individual FIA
plots. Hence, the time step for imputation in the Northcentral and Northeast states
was five years (Wear et al., 2013). FDM projections were validated by applying the
model to past FIA inventories for select states and comparing the projections to
present-day FIA inventories. Additionally, calibrations were made to the FDM at the
state-level following reviews by USDA Forest Service FIA and state foresters and
planners to account for past trends in forest change (Moser and Shifley, 2012; Wear
et al., 2013).

 The transition and imputation submodels used probabilistic (Monte Carlo)
methods to simulate variance associated with different model components. Note that
the algorithm for the imputation submodel was run 26 times with random selection
for donor plots in each time step resulting in 26 ``inventories''. The stochastic
nature of the imputation model implies that no two inventories were exactly alike.
Aggregate validation was performed by comparing 95\% confidence intervals for trees
per acre, total biomass, and sawtimber biomass for both hardwoods and softwoods using
all 26 inventories (Wear et al., 2013). Ultimately, only one inventory was used to
estimate forest attributes for each future scenario in the NFFDB. The selected inventory
was the one with the greatest ``central tendency,'' defined as minimum proportional
distance of total growing stock volume, trees per acre and sawtimber volume for softwoods
and hardwoods from their means over the 50-year projection period (Wear et al., 2013).
The completed database summarizes the results of the FDM for future decades by summarizing
plot conditions for a projected future date in the same way that one would summarize
current or past forest conditions (Miles, 2013).



 Projections of forest conditions for combinations of IPCC storylines and
CGCM included variations which accounted for higher assumed increases in woody biomass
utilization in the future. While not tied to specific forest management policies,
these alternative scenarios (designated as ``BIO'' scenarios) show substantial growth
in harvesting reflecting the expansion in demands for forest biomass in bioenergy
production associated with each scenario. These expansions were applied to the scenarios
by adjusting harvest probabilities to reflect the harvests predicted by the U.S.
Forest Products Model (Ince et al., 2011). Probability of harvest was adjusted for
these scenarios in the FDM by increasing the probability of harvest by a pre-determined
percentage via a scale parameter. Consequently, for each BIO scenario, the probability
of harvest was increased by the same proportion across the region. Note that global
assumptions regarding plantation woody biomass were included in the RPA projections;
however, plantation biomass does not play a significant role in the projections from
the FDM (Ince et al., 2011; USDA Forest Service, 2012a). Harvest projections for
the BIO scenarios deviated from the original scenarios starting in 2020 for the A1B
and A2 storylines. For the A1B-BIO storyline, harvest levels were projected to reach
approximately 3 times the 2010 level by 2060. Similarly, for the A2-BIO storyline,
harvest levels were projected to be roughly 2.5 times the 2010 level by 2060 (Wear
et al., 2013).

 Land-use change was a major consideration when developing models to project
forest conditions for the selected IPCC storylines and GCMs. Land-use change was
projected using econometric models developed by Wear (2011). These models were linked
to historical land-use data to ensure that land-use change estimates are fairly consistent
with trends in urbanization intensity and urban land-use change. The most important
components used in the econometric models pertained to urbanization and allocation
of rural land. Urbanization projections were driven by population and personal income
projections for the IPCC storylines (Wear, 2011). The allocation of rural land (land
not converted to urban) was greatly based on the existing distribution of different
non-urban land classifications from historical data. All federal land, water area,
enrolled Conservation Reserve Program lands, and utility corridors were held constant
for all projections (Wear, 2011).

 Output from the forest projection process described above was combined
with data from the FIADB (Woudenberg et al., 2010) to produce the NFFDB (Fig. 2)
(Miles, 2013; Miles et al., 2013).

\subsection{Habitat and Species Richness Assessments}
 Within the NFFDB, we assigned FIA forested conditions to six different
habitat classes defined to match classes in a wildlife-habitat matrix created by
NatureServe (2011) and purchased by the USDA Forest Service for the NFFP (Tab. 3).
Using canopy cover thresholds, we arrayed habitat classes along two dimensions addressing
differences in structure and composition. With respect to structure, we identified
classes as being either closed- (\underbar{$>$} 66\% total canopy cover) or open-canopy
(10 to 66\%). Our closed-canopy definition is consistent with NatureServe's (2011)
definition of `forest' habitat class whereas our open-canopy definition encompasses
both NatureServe's `woodland' (40 to 66\% canopy cover) and `savanna' (10 to 40\%
canopy cover) habitat classes.
\vbox % Put part of the text in vbox to force balancing of the text at the end of the page
{
Following consultation with NatureServe staff (J.
McNees, pers. comm., NatureServe, December 19, 2011), we included NatureServe `savanna'
in our open-canopy class because regenerating forest with sparse canopy is not synonymous
with a savanna ecosystem, and because actual savanna habitat is very rare in our
study area. To avoid confusion with FIA's definition of forest land, which encompasses
all three NatureServe (2011) habitat classes -- `forest', `woodland', and `savanna',
we refer to closed- and open-canopy habitat classes. Closed- and open-canopy classes
were further refined based on differences in composition. We labeled areas as hardwood
or conifer when $>$ 66\% of the canopy consisted of hardwood or conifer tree species,
respectively. Habitats were labeled as mixed when neither hardwood nor conifer tree
}
cover exceeds 66\% of the total canopy cover, consistent with NatureServe's definitions.

\begin{table*}[htb]
\centering
\caption{Forest habitat classes (adapted from NatureServe Habitat Classes, 2011).}
\vspace{3pt}
\begin{tabular}{p{1.0in}p{1.0in}p{4.5in}} \hline\hline
Class Name & NatureServe Habitat Class & Description \\ \hline
Closed-canopy forest & Forest & Woody vegetation at least 6 m tall (usually much
taller) with a fairly continuous and complete (two-thirds or greater) canopy closure. \\ \hline
Closed-canopy
hardwood & Forest-hardwood & Angiosperms comprise over two-thirds of the canopy. \\ \hline
Closed-canopy
conifer & Forest-conifer & Gymnosperms comprise over two-thirds of the canopy. \\ \hline
Closed-canopy
mixed & Forest-mixed & Composed of both hardwood and conifer trees, neither dominating
as much as two-thirds of the canopy. \\ \hline
Open-canopy forest & Woodland \& Savanna & Crowns often not interlocking; tree canopy
discontinuous (often clumped), averaging between 40 and 66 percent overall cover
(Nature Serve Woodland), or, mosaic of trees or shrubs and grassland; between 10
and 40 percent cover by trees and shrubs (Nature Serve Savanna). \\ \hline
Open-canopy hardwood & Woodland-hardwood & Angiosperms comprise over two-thirds of
the canopy. \\ \hline
Open-canopy conifer & Woodland-conifer & Gymnosperms comprise over two-thirds of
the canopy. \\ \hline
Open-canopy mixed & Woodland-mixed & Stand composed of both hardwood and conifer
trees, neither dominating as much as two-thirds of the canopy. \\ \hline
\end{tabular}
\label{tab3}
\end{table*}


\vspace{.28in}
 Using the NatureServe matrix, we tabulated numbers of terrestrial vertebrate
species within the study area, by major taxon (amphibians, birds, mammals, reptiles)
associated with each of six habitat classes, and by global rank (NatureServe, 2011).
Habitat associations reflected species' entire annual cycles, i.e., a species could
be associated with a habitat type during any season. Rank is defined as follows:
1 = critically imperiled; 2 = imperiled; 3 = vulnerable to extirpation or extinction;
4 = apparently secure; 5 = demonstrably widespread, abundant, and secure. A small
number of records had ``T'' ranks (infraspecific taxon: subspecies or varieties);
these were combined with ``G'' ranks (global ranks), and all results were labeled
as ranks (G1-G5). For the purposes of our coarse-filter assessment, we summarized
numbers and global ranks to characterize wildlife communities that might be affected
by projected changes in habitat classes. Projections of changing habitat associations
or global ranks for individual species fell outside the scope of our study. Such
species-specific assessments might be important and appropriate if the objective
is to inform species-level conservation objectives, but our assessment focused on
changes in coarse habitat classes.

\vspace{.28in}
 FIA does not provide estimates of canopy cover, so we used a computer algorithm
to derive estimates of canopy cover from FIA data, enabling us to crosswalk NFFDB
area projections to habitat classes (Fig. 2). A canopy cover modeling approach (Toney
et al., 2009) was used to estimate canopy cover for trees (\underbar{$>$} 5 in. d.b.h.,
on subplots), if present, or saplings (1-4.9 in. d.b.h., on microplots) on forested
FIA conditions within 20 states of USDA Forest Service's Eastern Region, during the
inventory period 2004-2008. Canopy cover estimation was based on tree species-specific
predicted crown dimensions, and tree stem location coordinates recorded by field
crews within FIA subplots and microplots. Tree and sapling crown width predictions
are based on Bechtold (2003) and Bragg (2001). An optional spatial statistic (Ripley's
K) included as a predictor in Toney et al. (2009) was not utilized for canopy cover
modeling in the present study. Because FIA plots may contain multiple conditions,
tree and sapling canopy cover estimates were weighted based on condition proportion
and appended to the CONDITION table in the NFFDB.

A small number of forested FIA conditions contained no trees or saplings.
Thus, no canopy cover estimates were available for these conditions, and canopy cover
could not be used to assign habitat classes to those conditions. During a plot visit,
a field crew can look beyond subplot boundaries to determine some condition attributes
via visual interpretation, including those conditions containing no trees at the
time of field data collection. For conditions with no trees or saplings (e.g., prior
to regeneration, or with only small seedlings; i.e., estimated canopy cover = 0),
habitat classes were recoded to valid classes using other FIA condition attributes,
described in Nelson et al. (2012). Of 52,860 forested conditions within the database,
51,398 (97.2\%) contained trees and/or saplings, from which canopy cover was predicted
and subsequent habitat classes were assigned. For the remaining 2.8\% of forested
conditions, 1.6\% were assigned to one of the six habitat classes based on other
condition attributes and the remaining 1.2\% of conditions were labeled as `no data'
because they had neither canopy cover, nor ancillary condition data. Plots associated
with the `no data' conditions were excluded from further analyses.\textbf{}

\section{Results}
 Birds were the most numerous terrestrial vertebrate species associated
with closed- or open-canopy forest habitat classes within the study area at 189 species,
followed by mammals (85), reptiles (52), and amphibians
(50). For every one of the six individual habitat classes, birds
and mammals had most species (Fig. 3). Amphibian species outnumbered reptiles in
all three closed-canopy classes; reptile species outnumbered amphibians in all three
open-canopy classes (Fig. 3).

\begin{figure}[hbt!]
\centering\includegraphics*[width=.48\textwidth, trim = 5 5 5 5, clip]{Fig3.eps}\textbf{}
\caption{{Number of terrestrial vertebrate species associated with closed- and open-canopy
conifer, hardwood, and mixed forest, Midwest and Northeast USA, by major taxon. (Adapted
from NatureServe, 2011)}}
\label{fig3}
\end{figure}

\begin{figure}[b!]\vspace{-.2in}
\centering\includegraphics*[width=.48\textwidth, trim = 5 5 5 5, clip]{Fig4.eps} \vspace{-.2in}
\caption{Number of terrestrial vertebrate species associated with closed- and open-canopy
conifer, hardwood, and mixed forest, Midwest and Northeast USA, by global rank. (Adapted
from NatureServe, 2011)}
\label{fig4}
\end{figure}

 Overall, 25 of 376 species (6.6\%) were listed within one of the three
most at-risk ranks (G1-G3), ranging from a low of 1.1\% for birds, to a high of 14.1
\% for mammals. Amphibians and reptiles were intermediate, with 12.0\% and 9.6\%,
respectively. Figure 4 presents numbers of species by global rank within each habitat
class. The habitat classes with highest and lowest percentages, respectively, of
at-risk species (G1-G3) were closed-canopy hardwood (7.3\%), and open-canopy mixed
(1.9\%). Note that many species were associated with multiple habitat classes, so
it is not valid to sum species counts across habitat classes in Figure 3 or 4.

 Per-plot estimates of canopy cover were used to assign habitat classes.
Because almost all FIA forested conditions were assigned habitat labels, total area
of habitat classes was essentially equivalent to FIA forest land area for the study
area (Nelson et al., 2012). Mean canopy cover of forest land across the study area
was 60.4\%, ranging from lows of 41.6 -- 55.3\% in Minnesota, Maine, Wisconsin and
Michigan, to highs of 74.2 -- 75.8\% in Rhode Island, Massachusetts, and Connecticut.


\begin{table*}[htb!]
\centering
\caption{Area (millions of ha) and percent change of six closed- (CC) and
open-canopy (OC) forest habitat classes across the Northeast and Midwest. Estimates
are provided for 2010 baseline conditions and for six 2060 scenarios representing
unique combinations of two Intergovernmental Panel on Climate Change storylines (IPCC)
and three General Circulation Models (GCM). Two 2060 scenarios assuming intensive
biomass utilization for bioenergy (A1B-BIO, A2-BIO) are also included. Changes in
habitat classes between 2010 and 2060 were driven by projected climate and land-use
changes, forest succession, and forest harvest. See Table 3 for explicit definitions
of forest habitat classes.}
\vspace{3pt}
\begin{tabular}{p{0.6in}p{0.6in}p{0.6in}p{0.6in}p{0.6in}p{0.6in}p{0.6in}p{0.6in}p{0.6in}}
\hline\hline
\textbf{IPCC } & \textbf{GCM} & \textbf{Total \newline Habitat} & \textbf{CC
Hardwood} & \textbf{CC\newline Conifer} & \textbf{CC\newline Mixed} & \textbf{OC
Hardwood} & \textbf{OC\newline Conifer} & \textbf{OC\newline Mixed} \\ \hline
Baseline & Historical & 70.2 & 28.5 & 1.7 & 4.4 & 24.1 & 6.7 & 4.7 \\
A1B-BIO & CGCM & 65.7    \newline (-6.4\%) & 22.8 \newline (-20.0\%) & 1.4 \newline (-17.6
\%) & 3.2 \newline (-27.3\%) & 27.6     (14.5\%) & 6.4\newline (-4.5\%) & 4.4\newline (-6.4
\%) \\
A1B & CGCM & 65.7    \newline (-6.4\%) & 29.6\newline (3.9\%) & 2.1 (23.5\%) & 4.6
(4.5\%) & 19.6\newline (-18.7\%) & 5.8\newline (-13.4\%) & 4.0\newline (-14.9\%) \\
 & CSIRO & 65.7\newline (-6.4
\%) & 29.5\newline (3.5\%) & 2.0\newline (17.6\%) & 4.7\newline (6.8\%) & 19.6\newline (-18.7
\%) & 5.8\newline (-13.4\%) & 4.2\newline (-10.6\%) \\
 & MIROC & 65.7\newline (-6.4\%) & 29.4\newline (3.2\%) & 2.1\newline (23.5\%) & 4.5\newline (2.3
\%) & 19.7\newline (-18.3\%) & 5.8\newline (-13.4\%) & 4.3\newline (-8.5\%) \\
A2-BIO & CGCM & 66.4
(-5.4\%) & 26.4 \newline (-7.4\%) & 1.6 \newline (-5.8\%) & 3.9\newline (-11.4\%) & 24.0
(-0.4\%) & 6.2\newline (-7.5\%) & 4.4\newline (-6.4\%) \\
A2 & CGCM & 66.4     (-5.4\%) & 30.1 \newline (5.6\%) & 2.0 (17.6\%) & 4.7\newline (6.8
\%) & 19.7\newline (-18.3\%) & 5.9\newline (-11.9\%) & 4.1\newline (-12.8\%) \\
 & CSIRO & 66.4\newline (-5.4\%) & 29.9\newline (4.9\%) & 2.0\newline (17.6\%) & 4.6\newline (4.5
\%) & 19.8\newline (-17.8\%) & 5.8\newline (-13.4\%) & 4.3\newline (-8.5\%) \\
 & MIROC & 66.4\newline (-5.4
\%) & 29.9\newline (4.9\%) & 2.0\newline (17.6\%) & 4.7\newline (6.8\%) & 19.9\newline (-17.4
\%) & 5.9\newline (-11.9\%) & 4.0\newline (-14.9\%) \\ \hline
\end{tabular}
\label{tab4}
\end{table*}

Across the Northeastern and Midwestern U.S., total area of all forest land currently stands
at 70.5 million ha. Of 0.6 million ha in nonstocked conditions, 0.3 million ha were
assigned to habitat classes and 0.3 million ha were omitted from the habitat classification
(0.4\% of total forest land area), resulting in 70.2 million ha assigned to six habitat
classes (Tab. 4). The region is dominated by the closed-canopy hardwood (40.6\%
of forest habitat) and open-canopy hardwood (34.3\%) habitat classes with no other
class exceeding 10\% of forest habitat. Forest land is approximately evenly split
between the groups of closed- and open-canopy habitat classes (49.3\% and 50.7\%,
respectively).

Sampling errors associated with baseline per-state estimates of 2010 forest
land area ranged from 0.4\% (Michigan) to 4.2\% (Delaware), with a median value of
1.1\% across 20 states. Per-county estimates of total forest land area resulted in
considerably larger sampling errors than for per-state estimates. Delaware sampling
errors for per-county baseline estimates of total forest land area ranged from 6.0
-- 21.1\% (median = 9.8\%), and per-county sampling errors for Michigan baseline
estimates ranged from 1.4 -- 20.1\% (median = 5.6\%). Due to the smaller
sampling errors for per-state estimates of baseline conditions, and the
additional (but unknown) uncertainty introduced in the projections modeling process,
all subsequent results are reported only for per-state or region-wide scales.

Assuming standard forest harvest levels, loss of habitat area was projected
under both IPCC storylines with the magnitude of loss ranging from 3.8 million ha
(5.4\%) under A2 to 4.5 million ha (6.4\%) under A1B (Tab. 4). While projected losses
for total habitat area did not differ among GCMs for either storyline, choice of
GCM did affect projected changes for individual habitat classes, but these effects
were relatively minor and varied across habitat classes. For example, areas for the
open-canopy mixed habitat class ranged from 4.0 (CGCM) to 4.3 million ha (MIROC)
whereas areas for open-canopy conifer did not vary under the A1B storyline. Patterns
of change for habitat classes were consistent across both IPCC storylines (Tab. 4). All three closed-canopy forest habitat classes gained area; percent gains were
greatest for closed-canopy conifer and least for either closed-canopy hardwood or
mixed, depending on the GCM. Conversely, all three open-canopy habitat classes lost
area; percent losses were greatest for open-canopy hardwood and least for open-canopy
conifer or mixed, depending on the GCM. Closed-canopy habitat classes (54.7 to 55.3\%) were projected to increase relative to open-canopy habitat classes (44.7 to 45.3\%) as a percent of total habitat regardless of the scenario considered.

 Under the high biomass utilization scenarios, loss of habitat area was
projected under both IPCC storylines, with one exception: A1B-BIO-CGCM open-canopy
hardwood class gained 3.5 million ha (Tab. 4). Across the other eleven classes,
A1B-BIO-CGCM closed-canopy mixed displayed the greatest percent loss, and A2-BIO-CGCM
open-canopy hardwood displayed the least (Tab. 4). Thus, patterns of change were
mostly consistent, but not in magnitude across IPCC storylines (Tab. 4). Under the
A1B-BIO-CGCM scenario, closed-canopy habitat classes were in the minority (41.6\%
versus 58.4\% for open-canopy classes) whereas, under the A2-BIO scenario, the closed-
and open-canopy classes remained relatively balanced (48.0\% versus 52.0\%, respectively).

\begin{figure*}[htb!]
\centering\includegraphics*[width=.75\textwidth]{Fig5.eps}
\caption{Percent change in area of closed- and open-canopy habitat classes, 2010-2060, by state and future scenarios. Percent of 2010 forest land area within each habitat
class is shown for reference (left column). Future scenarios involved two Intergovernmental
Panel on Climate Change storylines (A1B, A2) and the CGCM 3.1 MR (T47) general circulation
model. Scenarios assumed either standard harvest rates (A1B, A2) or intensive harvest
rates for high biomass utilization (A1B-BIO, A2-BIO).}
\label{fig5}
\end{figure*}

 The greatest spatial contrasts seen in Figure 5 pertain to states with well-established
timber industries, such as Minnesota, Wisconsin, and Maine. These states showed general
increases in forest land area for all closed canopy classes for the original scenarios,
but displayed some of the greatest losses in closed canopy classes for the BIO scenarios.

\section{Discussion}
 Adopting
a coarse-filter approach, we used climate and land-use projections sharing a common
set of assumptions to assess potential changes in forest habitat classes across the
Northeast and Midwest from 2010 to 2060. For all scenarios considered, our assessments
suggest that the total area of forest habitat classes will decrease, and this loss
in total habitat area has the potential to negatively affect wildlife populations.
For an individual species, the degree of these effects may depend, in part, on the
spatial pattern of habitat loss. Although we do portray regional variation in habitat
trends among states, we did not directly assess spatial patterns of habitat loss
at a fine scale. Overall reduction in habitat area can lead to smaller and more isolated
forest patches. These patches support fewer individuals and are less likely to receive
immigrants from other areas, increasing the likelihood of local extirpation and decreasing
likelihood of recolonization or population rescue (Hanski, 1999). Habitat in smaller
forest patches in this region of North America is also more exposed to negative ecological
influences (e.g., nest predators, Donovan et al., 1995) from surrounding non-forest
land-uses, contributing to local population declines. Land-use and climate changes
may have synergistic effects on species. For example, reduced connectivity among
forest patches might influence the ability of a species to locate and occupy climatically
suitable environments as these shift in response to changing climate conditions (Hannah,
2008; Opdam and Wascher, 2004). If habitat loss is widespread, regional declines
and extirpations may result.

 Our assessments suggest that uncertainty about future demographic, economic,
technological, and climate conditions (as represented by different IPCC-GCM scenarios)
contributes to uncertainty about the extent of habitat loss. While we did not quantify
it, additional uncertainty arises from the unknowable possibility that future forest
and land-use management actions might greatly depart from historically observed actions.
Policy (e.g., promoting growth near existing urban centers) and financial mechanisms
(e.g., tax deductions resulting from conservation easements) might be used to limit
negative effects of land-use change on forest wildlife.



 The number of terrestrial vertebrate species varied among major taxa and
among closed- and open-canopy hardwood, conifer, and mixed forest habitat classes.
Birds and mammals dominated species richness. The number of species at-risk rank
was relatively low (6.6\%), with the largest percentages observed for mammals and
amphibians. While numbers of species were not projected for future conditions, consideration
for at-risk species may be needed for habitat  classes projected to decline in future decades.

 Researchers have reported decades-long declines in the area of early successional
forest habitat across the Northeast and Midwest (Trani et al., 2001). These declines
have been attributed to a number of different causes including forest maturation
of abandoned farmland, altered forest management practices, forest ownership patterns
that discourage harvest, disrupted natural disturbance regimes (e.g., fire suppression),
and land-use conversion (Askins, 2001; Lorimer and White, 2003; Trani et al., 2001).
Assuming that early successional forests can be characterized as having more open
canopies, projections of open-canopy habitat classes in our assessment suggested
that declines of this habitat type may continue into the near future. With the exception
of intensive biomass utilization scenarios, we found that all open-canopy habitat
classes declined and that regional habitat became dominated by closed-canopy habitat
classes. These projected declines may negatively affect not only open-canopy associated
species but also species typically associated with closed-canopy habitats that depend
upon open-canopy areas during certain times of the year (e.g., Streby et al., 2011;
Vitz and Rodewald, 2006). Ultimately, the future status of wildlife species dependent
on young forests or open-canopy habitat will depend on the scale, type, and frequency
of anthropogenic and natural disturbances occurring in landscapes across the Northeast
and Midwest.



 The harvest of woody biomass for bioenergy is perceived as having the potential
to mitigate climate change by alleviating, to a degree, dependence on traditional
fossil fuel sources (White, 2010). Climate change mitigation policies promoting biomass
harvest might increase the profitability of harvesting in stands previously seen
as being non-commercial (e.g., due to poor wood quality) and lead to shorter rotation
times (Janowiak and Webster, 2010). Harvest of woody biomass has the potential to
open up forest canopies and turn back succession, influencing the balance between
closed- and open-canopy habitat classes. Our intensive biomass utilization scenarios
led to smaller decreases for open-canopy habitat classes relative to the other scenarios
considered; one class (open-canopy hardwood) even displayed an increase under the
A1B-BIO-CGCM scenario. Under the intensive biomass utilization scenarios, the percent
cover of forest land in the open-canopy habitat classes remained stable or increased
relative to current conditions. This contrasts with our other scenarios in which
percent cover of open-canopy habitat classes declined and the closed-canopy habitat
classes attained a slight majority. Policies and tactics associated with woody biomass
harvest will partly determine the degree to which wildlife species dependent on open-canopy
habitat classes might benefit. Biomass harvest for bioenergy might incentivize the
removal of woody residue, or woody materials typically left behind after harvest
(e.g., tops, dead wood). These materials contribute to important microhabitat conditions
that can influence the habitat quality of an area. Several states have adopted BMPs
specifically designed to minimize impacts of woody biomass removal on water quality,
soils, biodiversity and wildlife habitat (Shepard, 2006; Skog and Stantkurf, 2011).
We did not examine changes in microhabitat features as a result of intensive biomass
utilization due to limitations of available FIA data and the projection technique.



 Recall
that some of the stark contrasts between the original scenarios and the BIO scenarios
regarding canopy cover classes occurred in northern states with relatively high current
levels of forest products utilization, such as Minnesota, Wisconsin, and Maine. This
was greatly due to the high current probabilities of harvest within these states
resulting in greater increases in harvest probability within the FDM for the BIO
scenarios. This is logical, as the probability of harvest was increased by the same
proportion across the region when the FDM was adjusted for higher biomass utilization
(Wear et al., 2013). Consequently, many of these same states also show lower decreases
(and sometimes increases) in forest area for the open canopy classes when compared
to the original scenarios A1B and A2. While the variability in direction and magnitude
of change among scenarios cautions against over-interpretation, these results suggest
that future trends in forest habitat conditions will vary across states presenting
unique challenges to wildlife managers in different areas.



 Interpreting the significance of projected shifts in the representation
of closed- and open-canopy habitat classes is difficult without appropriate ecological
context. One viewpoint is that the historical balance between closed- and open-canopy
habitat classes should be the standard because these are the conditions under which
organisms evolved (Askins, 2001; Litvaitis, 2003; Lorimer, 2001; Thompson and DeGraaf,
2001). Estimating the frequency and extent of historical disturbance events and open-canopy
habitat is difficult for a variety of reasons, including difficulty differentiating
natural from anthropogenic disturbances and spatiotemporal variation in disturbance
rates (Lorimer, 2001). To cope with temporal variability in disturbance rates, researchers
have suggested managing habitat classes to maintain a balance that falls within the
range of historical variability (Thompson and DeGraaf, 2001). With respect to wildlife
management, it is important to consider the minimal amount of open-canopy (or other
habitat class) required to support viable populations (Askins, 2001; Lorimer, 2001).
Studies have indicated that species dependent on open-canopy habitat might respond
to decreasing habitat areas in non-linear, threshold fashions although these thresholds
might occur at relatively low levels of habitat cover (e.g., Betts et al., 2010).
Identifying appropriate benchmarks for habitat management remains an active field
of research.



 It can be difficult to associate FIA data with habitat classes in established
wildlife-habitat matrices. The method presented here provides an operational approach
to predicting per-condition tree canopy cover from FIA tree data, with resulting
classifications used to assign FIA conditions to closed- and open-canopy habitat
classes, for which population estimates were produced. Although FIA's forest land
definition requires a minimum of 10 percent canopy cover, a small area of FIA forest
land was characterized by canopy cover below this threshold. Such conditions likely
occur shortly after full canopy removal (e.g., harvest, wildfire, etc.), but before
regenerating seedlings have established significant canopy. Tree canopy cover predictions
allowed FIA data to be used with NatureServe's (2011) wildlife-habitat matrices to
summarize species distribution across habitat classes. Because choice of habitat
classification systems can affect resulting estimates of habitat abundance, work
continues to link FIA data with a variety of habitat classifications systems, including
the National Vegetation Classification System (Federal Geographic Data Committee,
2008).



 A significant challenge for wildlife managers is developing and implementing
large-scale and long-term plans aimed at maintaining a suite of habitat conditions
suitable for diverse wildlife communities. To maintain wildlife communities in the
future, wildlife managers will need to cope with landscape dynamics driven by changes
in climate and land-use. Our assessments suggest that the overall area of forest
habitat might decline and the balance between different habitat classes might shift
in the future. The influence of assumptions about biomass utilization on the balance
between closed- and open-canopy habitat classes highlights the importance of policy
and management decisions in determining habitat conditions in the future. Ultimately,
managers will need to identify benchmark habitat conditions informed by historical
conditions and wildlife population dynamics and to develop plans to meet these benchmarks
in dynamic forest landscapes.


\section*{Acknowledgements}
 The authors thank NatureServe for producing a terrestrial vertebrate species-habitat
matrix for forest-associated species. The authors also acknowledge the assistance
of J.M. Reed, J. Stanovick, S. Oswalt, and three anonymous reviewers whose suggestions
improved the manuscript.

\begin{thebibliography} {99}

\bibitem[a(2012)]{a}
 Aguilar, F.X., M. Goerndt, N. Song, and S. Shifley. 2012. Internal, External
and Location Factors Influencing Cofiring of Biomass with Coal in the U.S. Northern
Region.\textit{ }Energ. Econ.\textit{ }34:1790-1798.

\bibitem[a(2012)]{a}
 Annand, E.M., and F.R. Thompson, III. 1997. Forest bird response to regeneration
practices in central hardwood forests. J. Wildlife Manage. 61:159-171.

\bibitem[a(2012)]{a}
 Askins, R.A. 2001. Sustaining biological diversity in early successional
communities: the challenge of managing unpopular habitats. Wildlife Soc. B. 29:407-412.

\bibitem[a(2012)]{a}
 Beaudry, F., A.M. Pidgeon, V.C. Radeloff, R.W. Howe, D.J. Mladenoff, and G.A. Bartelt. 2010.
Modeling regional-scale habitat of forest birds when lands management guidelines
are needed but information is limited. Biol. Conserv. 143:1759-1769.

\bibitem[a(2012)]{a}
 Beaumont, L.J., L. Hughes, and A.J. Pitman. 2008. Why is the choice of
future climate scenarios for species distribution modeling important? Ecol. Lett.
11:1135-1146.

\bibitem[a(2012)]{a}
 Bechtold, W.A. 2003. Crown-diameter prediction models for 87 species of
stand-grown trees in the Eastern United States. South. J. Appl. For. 27: 269-278.

\bibitem[a(2012)]{a}
 Bechtold, W.A., and C.T. Scott. 2005. The Forest Inventory and Analysis plot design. P. 27-42
in The Enhanced Forest Inventory and Analysis Program -- National Sampling Design
and Estimate Procedures. W.A. Bechtold and P.L. Patterson (Eds.). USDA For. Serv.
Gen. Tech. Rep. SRS-GTR-80.

\bibitem[a(2012)]{a}
 Betts, M.G., J.C. Hagar, J.W. Rivers, J.D. Alexander, K. McGarigal, and
B.C. McComb. 2010. Thresholds in forest bird occurrence as a function of the amount
of early-seral broadleaf forest at landscape scales. Ecol. Appl. 20:2116-2130.

\bibitem[a(2012)]{a}
 Bierwagen, B.G., D.M. Theobald, C.R. Pyke, A. Choate, P. Groth, J.V. Thomas,
and P. Morefield. 2010. National housing and impervious surface scenarios for integrated
climate impact assessments. P. Natl. Acad. Sci. USA. 107:20887-20892.

\bibitem[a(2012)]{a}
 Booth, T.H. 1990. Mapping regions climatically suitable for particular
tree species at the global scale. Forest Ecol. Manag. 36:47-60.

\bibitem[a(2012)]{a}
 Bragg, D.C. 2001. A local basal area adjustment for crown width prediction.
North. J. Appl. For. 18:22-28.

\bibitem[a(2012)]{a}
 Cook, J., and J. Beyea. 2000. Bioenergy in the United States: progress
and possibilities. Biomass Bioenerg. 18:441-455.

\bibitem[a(2012)]{a}
 Coulson, D.P., and L.A. Joyce, 2010. Historical Climate Data (1940-2006)
for the Conterminous United States at the County Spatial Scale Based on PRISM Climatology.\textit{ }Available
online at \href{http://www.fs.usda.gov/rds/archive/Product/RDS-2010-0010}
{http://www.fs.usda.gov/rds/archive/Product/RDS-2010-0010}; last accessed Feb. 25, 2012.

\bibitem[a(2012)]{a}
 Coulson, D.P., L.A. Joyce, D.T. Price, D.W. McKenney, R.M. Siltanen, P.
Papadopol, and K. Lawrence. 2010. Climate Scenarios for the Conterminous United States
at the County Spatial Scale Using SRES Scenarios A1B and A2 and PRISM Climatology.\textit{ }Available
online at \href{http://www.fs.usda.gov/rds/archive/Product/RDS-2010-0008}
{http://www.fs.usda.gov/rds/archive/Product/RDS-2010-0008}; last accessed Feb. 25, 2012.

\bibitem[a(2012)]{a}
 Dietzman, D., K. LaJeunesse, and S. Wormstead. 2011. Northern Forest Futures
Project: scoping of issues in the forest of the Northeast and Midwest of the United
States. Version 3.0, June 2011. USDA For. Serv. 42 p. Available online at
\href{http://www.nrs.fs.fed.us/futures/local-resources/downloads/NFFPScopingDoc.pdf}{http://www.nrs.fs.fed.us/futures/local-resources/downloads/NFFPScopingDoc.pdf}; last
accessed Aug. 15, 2013.

\bibitem[a(2012)]{a}
 Donovan, T.M., F.R. Thompson, III, J. Faaborg, and J.R. Probst. 1995. Reproductive
success of migratory birds in habitat sources and sinks. Conserv. Biol. 9:1380-1395.

\bibitem[a(2012)]{a}
 Federal Geographic Data Committee. 2008. National Vegetation Classification Standard. Vegetation
Subcommittee, Federal Geographic Data Committee, FGDC Secretariat. U.S. Geological
Survey, Reston, VA. 55 p.

\bibitem[a(2012)]{a}
 Hagan, J.M., P.S. McKinley, A.L. Meehan, and S.L. Grove. 1997. Diversity
and abundance of landbirds in a northeastern industrial forest. J. Wildlife Manage.
61:718-735.

\bibitem[a(2012)]{a}
 Hamel, P.B., K.V. Rosenberg, and D.A. Buehler. 2005. Is management for
golden-winged warblers and cerulean warblers compatible? P. 322-331 in Bird conservation
implementation and integration in the Americas: proceedings of the third international
Partners in Flight conference; 2002 March 20-24, Asilomar, CA, Volume 1, C.J. Ralph
and T.D. Rich (Eds.). USDA For. Serv. Gen. Tech. Rep. PSW-GTR-191.

\bibitem[a(2012)]{a}
 Hannah, L. 2008. Protected areas and climate change. Ann. N.Y. Acad. Sci.
1134:201-212.

\bibitem[a(2012)]{a}
 Hanski, I. 1999. Metapopulation ecology. Oxford University Press. 328 p.

\bibitem[a(2012)]{a}
 Hartley, M.J. 2002. Rationale and methods for conserving biodiversity in plantation forest.
Forest Ecol. Manag. 155:81-95.

\bibitem[a(2012)]{a}
 Ince, P.J., A.D. Kramp, K.E. Skog, H.N. Spelter, and D.N. Wear. 2011. U.S.
Forest Products Module: a technical document supporting the Forest Service 2010 RPA
Assessment. Research Paper FPL-RP-662. Madison, WI: U.S. Department of Agriculture,
Forest Service, Forest Products Laboratory. 61 p.

\bibitem[a(2012)]{a}
 Iverson, L.R., and A.M. Prasad. 1998. Predicting abundance of 80 tree species
following climate change in the eastern United States. Ecol. Monogr. 68:465-485.

\bibitem[a(2012)]{a}
 Janowiak, M.K., and C.R. Webster. 2010. Promoting ecological sustainability in woody biomass
harvesting. J. Forest. 108:16-23.

\bibitem[a(2012)]{a}
 Litvaitis, J.A. 2003. Are pre-Columbian conditions relevant baselines for
managed forests in the northeastern United States? Forest Ecol Manag. 185:113-126.

\bibitem[a(2012)]{a}
 Lorimer, C.G. 2001. Historical and ecological roles of disturbance in eastern North American
forests: 9,000 years of change. Wildlife Soc. B. 29:425-439.

\bibitem[a(2012)]{a}
 Lorimer, C.G., and A.S. White. 2003. Scale and frequency of natural disturbances
in the northeastern US: implications for early successional habitats and regional
age distributions. Forest Ecol Manag. 185:41-64.

\bibitem[a(2012)]{a}
 Margules, C.R., and R.L. Pressey. 2000. Systematic conservation planning.
Nature. 405:243-253.

\bibitem[a(2012)]{a}
 Matthews, S., R. O'Connor, L.R. Iverson, and A.M. Prasad. 2004. Atlas of
climate change effects in 150 bird species of the eastern United States. USDA For.
Serv. Gen. Tech. Rep. NE-GTR-318. 340p.

\bibitem[a(2012)]{a}
 Miles, P.D. 2013. Forest Inventory EVALIDator web-application version 1.5.1.05.
Available online at \href{http://apps.fs.fed.us/Evalidator/evalidator.jsp}{http://apps.fs.fed.us/Evalidator/evalidator.jsp}; last accessed
Jun. 20, 2013.

\bibitem[a(2012)]{a}
 Miles, P.D., R.J. Huggett, and W.K. Moser. 2013. Northern Forest Futures
database and reporting tools. USDA For. Serv. Gen. Tech. Rep. In press.

\bibitem[a(2012)]{a}
 Moore, S.E., and H.L. Allen. 1999. Plantation forestry. P. 400-433 in Maintaining
Biodiversity in Forest Ecosystems, M.L. Hunter (ed.). Cambridge University Press,
city, ST.

\bibitem[a(2012)]{a}
 Moser, W.K., and S.R. Shifley. 2012. The Northern Forest Futures Project:
A forward look at forest conditions in the northern United States. P. 44-54 in Environmental
Futures Research: Experiences, Approaches, and Opportunities. D.N. Bengston (Ed.).
USDA For. Serv. Gen. Tech. Rep. NRS-GTR-P-107.

\bibitem[a(2012)]{a}
 NatureServe. 2011. L48\_vert\_hab\_data\_022011.accdb -- a relational Access
database. Lists of Vertebrate Species in the Contiguous U.S.; February 17, 2011.

\bibitem[a(2012)]{a}
 Nelson, M.D., B.G. Tavernia, C. Toney, and B. Walters. 2012. Relating FIA data to habitat
classifications via tree-based models of canopy cover. P. 254-259 in Moving from
Status to Trends: Forest Inventory and Analysis Symposium\textit{ }2012. Morin, R.S.,
and G.C. Liknes (eds.). USDA For. Serv. Gen. Tech. Rep. NRS-GTR-P-105.

\bibitem[a(2012)]{a}
 Nelson, M.D., M. Brewer, C.W. Woodall, C.H. Perry, G.M. Domke, R.J. Piva,
C.M. Kurtz, et al. 2011. Iowa's Forest 2008. USDA For. Serv. Resour. Bull. NRS-RB-52.
48 p.

\bibitem[a(2012)]{a}
 Newton, I. 2003. Speciation and biogeography of birds. Academic Press. 656 p.

\bibitem[a(2012)]{a}
 Noon, B.R., K.S. McKelvey, and B.G. Dickson. 2009. Multispecies conservation
planning on U.S. federal lands. P. 51-83 in Models for planning wildlife conservation
in large landscapes. J.J. Millspaugh and F.R. Thompson, III (Eds.). Academic Press,
city, ST.

\bibitem[a(2012)]{a}
 Opdam, P., and D. Wascher. 2004. Climate change meets habitat fragmentation:
linking landscape and biogeographical scale levels in research and conservation.
Biol. Conserv. 117:285-297.

\bibitem[a(2012)]{a}
 Parmesan, C. 2006. Ecological and evolutionary responses to recent climate
change. Annu. Rev. Ecol. Evol. S. 37:637-669.

\bibitem[a(2012)]{a}
 Patton, D.R. 2011. Forest wildlife ecology and habitat management. CRC
Press, city, ST. 272 p.

\bibitem[a(2012)]{a}
 Pearson, R.G., T.P. Dawson, and C. Liu. 2004. Modelling species distribution
in Britain: a hierarchical integration of climate and land-cover data. Ecography.
27:285-298.

\bibitem[a(2012)]{a}
 Prentice, I.C., W. Cramer, S.P. Harrison, R. Leemans, R.A. Monserud, and
A.M. Solomon. 1992. A global biome model based on plant physiology and dominance,
soil properties and climate. J. Biogeogr. 19:117-134.

\bibitem[a(2012)]{a}
 Randall, D.A., R.A. Wood, S. Bony, R. Colman, T. Fichefet, J. Fyfe, V.
Kattsov, et al. 2007. Climate Models and Their Evaluation. P. 589-662 in Climate
Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change, Solomon, S.,
D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor, and H.L. Miller
(eds.). Cambridge University Press, city, ST.

\bibitem[a(2012)]{a}
 Raupach, M., G. Marland, P. Ciasis, C. Quéré, J. Canadell, G. Klepper,
and C.B. Field. 2007. Global and Regional Drivers of Accelerating CO${}_{2}$${}^{ }$Emissions.
P. Natl. Acad. Sci. USA. 104:10288-10293.

\bibitem[a(2012)]{a}
 Reams, G.A., W.D. Smith, M.H. Hansen, W.A. Bechtold, F.A. Roesch, and G.G.
Moisen. 2005. The Forest Inventory and Analysis sampling frame. P. 11-26 in The Enhanced
Forest Inventory and Analysis Program -- National Sampling Design and Estimate Procedures.
W.A. Bechtold and P.L. Patterson (eds.). USDA For. Serv. Gen. Tech. Rep. SRS-GTR-80.

\bibitem[a(2012)]{a}
 Reichler, T., and J. Kim. 2008. How well do coupled models simulate today's climate? B. Am.
Meteorol. Soc. 89:303-311.

\bibitem[a(2012)]{a}
 Riffell, S., J. Verschuyl, D. Miller, and T.B. Wigley. 2011. Biofuel harvests,
coarse woody debris, and biodiversity -- a meta-analysis. Forest Ecol. Manag. 261:878-887.

\bibitem[a(2012)]{a}
 Robinson, S.K., and D.S. Wilcove. 1994. Forest fragmentation in the temperate zone and its
effects on migratory songbirds. Bird Conserv Int. 4:233-249.

\bibitem[a(2012)]{a}
 Schmidt, T.L., J.S. Spencer, Jr. and M.H. Hansen. 1996. Old and potential
old forest in the Lake States, USA. Forest Ecol. Manag. 86:81-96.

\bibitem[a(2012)]{a}
 Schulte, L.A., R.J. Mitchell, M.L. Hunter, Jr., J.F. Franklin, R.K. McIntyre,
and B.J. Palik. 2006. Evaluating the conceptual tools for forest biodiversity conversation
and their implementation in the U.S. Forest Ecol. Manag. 232:1-11.

\bibitem[a(2012)]{a}
 Schwartz, M.W., L.R. Iverson, A.M. Prasad, S.N. Matthews, and R.J. O'Connor.
2006. Predicting extinctions as a result of climate change. Ecology. 87:1611-1615.

\bibitem[a(2012)]{a}
 Shepard, J.P. 2006. Water quality protection in bioenergy production: The US system of forestry Best Management Practices. Biomass Bioenerg. 30:378-384.

\bibitem[a(2012)]{a}
 Shifley, S.R., F.X. Aguilar, N. Song, S.I. Stewart, D.J. Nowak, D.D. Gormanson,
W.K. Moser, et al. 2012. Forests of the northern United States. USDA For. Serv. Gen.
Tech. Rep. NRS-GTR-90. 202 p.

\bibitem[a(2012)]{a}
 Skog, K.E, and J.A. Stanturf. 2011. Forest biomass sustainability
and availability. P. 3-25 in Sustainable Production of Fuels, Chemicals, and Fibers
from Forest Biomass, Zhu, J., X. Zhang, and X. Pan (eds.). American Chemical Society
Symposium Series Volume 1067, city, ST.

\bibitem[a(2012)]{a}
 Streby, H.M., S.M. Peterson, T.L. McAllister, and D.E. Andersen. 2011.
Use of early-successional managed northern forest by mature-forest species during
the post-fledging period. Condor. 113:817-824.

\bibitem[a(2012)]{a}
 Tavernia, B.G., M.D. Nelson, P. Caldwell, and G. Sun. 2013. Water stress
projections for the northeastern and Midwestern United States in 2060: anthropogenic
and ecological consequences. J. Am. Water Resour. As. DOI: 10.1111/jawr.12075

\bibitem[a(2012)]{a}
 Thomas, C.D., A. Cameron, R.E. Green, M. Bakkenes, L.J. Beaumont, Y.C.
Collingham, B.F.N. Erasmus, et al. 2004. Extinction risk from climate change. Nature.
427:145-148.

\bibitem[a(2012)]{a}
 Thompson, F.R., III, and R.M. DeGraaf. 2001. Conservation approaches for
woody, early successional communities in the eastern United States. Wildlife Soc.
B. 29:483-494.

\bibitem[a(2012)]{a}
 Toney, C., J.D. Shaw, and M.D. Nelson. 2009. A stem-map model for predicting
tree canopy cover of Forest Inventory and Analysis (FIA) plots. P. 1-19 in 2008 Forest
Inventory and Analysis (FIA) Symposium, McWilliams, W., G. Moisen, and R. Czaplewski
(eds.). USDA For. Serv. Proc. RMRS-P-56CD.

\addtolength{\textheight}{-4.85truein}

\bibitem[a(2012)]{a}
 Trani, M.K., R.T. Brooks, T.L. Schmidt, V.A. Rudis, and C.M. Gabbard. 2001.
Patterns and trends of early successional forests in the eastern United States. Wildlife
Soc. B. 29:413-424.

\bibitem[a(2012)]{a}
 U.S. Energy Information Administration. 2012. Annual Projections. Available
online at \href{http://www.eia.gov/analysis/projection-data.cfm#annualproj} {http://www.eia.gov/analysis/projection-data.cfm\#annualproj}; last accessed
Oct. 8, 2012.

\bibitem[a(2012)]{a}
 USDA Forest Service. 2012a. Future scenarios: a technical document supporting
the Forest Service 2010 RPA Assessment. USDA For. Serv. Gen. Tech. Rep. RMRS-GTR-272.
34 p.

\bibitem[a(2012)]{a}
 USDA Forest Service 2012b. Resources Planning Act (RPA) Assessment Online.
Available online at \href{http://www.fs.fed.us/research/rpa/}{http://www.fs.fed.us/research/rpa/}; last accessed Jul. 27, 2012.

\bibitem[a(2012)]{a}
 Verboom, J., A. Schotman, P. Opdam, and J.A.J. Metz. 1991. European nuthatch metapopulations
in a fragmented agricultural landscape. Oikos. 61:149-156.

\bibitem[a(2012)]{a}
 Vitz, A.C., and A.D. Rodewald. 2006. Can regenerating clearcuts benefit
mature-forest songbirds? An examination of post-breeding ecology. Biol. Conserv.
127:477-486.

\bibitem[a(2012)]{a}
 Wear, D.N. 2011. Forecasts of county-level land uses under three future
scenarios: a technical document supporting the Forest Service 2010 RPA Assessment.
USDA For. Serv. Gen. Tech. Rep. SRS-GTR-141. 41 p.

\vspace{.045in}
\bibitem[a(2012)]{a}
 Wear, D.N., R. Huggett, R. Li, B. Perryman, and S. Liu. 2013. Forecasts
of forest conditions in regions of the United States under future scenarios: a technical
document supporting the Forest Service 2010 RPA Assessment. USDA For. Serv. Gen.
Tech. Rep. SRS-GTR-170. 101 p.

\vspace{.045in}
\bibitem[a(2012)]{a}
 White, E.M. 2010. Woody biomass for bioenergy and biofuels in the United
States -- a briefing paper. USDA For. Serv. Gen. Tech. Rep. PNW-GTR-825. 45 p.

\vspace{.045in}
\bibitem[a(2012)]{a}
 Woudenberg, S.W., B.L. Conkling, , B.M. O'Connell, E.B. LaPoint,J.A. Turner,and
K.L. Waddell. 2010. The Forest Inventory and Analysis Database: Database description
and users manual version 4.0 for Phase 2. USDA For. Serv. Gen. Tech. Rep. RMRS-GTR-245.
336 p.

\vspace{.045in}
\bibitem[a(2012)]{a}
 Zarnoch, S.J., H.K. Cordell, C.J. Betz, and L. Langner. 2010. Projecting
county-level populations under three future scenarios: a technical document supporting
the Forest Service 2010 RPA assessment. USDA For. Serv. Gen. Tech. Rep. SRS-GTR-128.
16p.

\label{docend} \addtolength{\textheight}{-2truein}

\end{thebibliography}

\end{document}