Processes Associated with Afforestation near Public Lands

doi:10.4236/aasoci.2012.21001

Paper Menu >>

Journal Menu >>

Advances in Applied Sociology

2012. Vol.2, No.1, 1-6

Published Online March 2012 in SciRes (http://www.SciRP.org/journal/aasoci) http://dx.doi.org/10.4236/aasoci.2012.21001

Processes Associated with Afforestation near Public Lands

Zola K. Moon, Frank L. Farmer

University of Arkansas, Fayetteville, USA

Email: [email protected]

Received December 21st, 2011; revised January 18th, 2012; accepted February 4th, 2012

This paper empirically examines patterns of afforestation in vicinities immediately surrounding National

Park/National River and National Forest lands. The public lands (Ozark National Forest and Buffalo Na-

tional River) are found on the Ozark Plateau and represent different management mandates. A spatial lag

model is presented comparing two LANDSAT images in conjunction with sociodemographic measures

covering the same time period. The findings here make two important points. First, the public land

boundaries are shown to act as ecological switches. Second, results underscore the importance of under-

standing how publicly managed lands with different mandates function within the larger social as well as

geophysical landscape matrix. Empirical evidence demonstrates that public lands set aside for “preserva-

tion” (Buffalo National River) are associated with greater afforestation, whereas public lands set aside for

“conservation and use of natural resources” (Ozark National Forest) are surrounded by less afforestation.

Keywords: Landcover Change; Afforestation; LANDSAT; Spatial Lag Regression; Public Lands

Introduction

Public lands have been set aside around the world for a vari-

ety of reasons. The rationale for these publicly managed lands

can range from safeguarding a natural landscape or protecting

the habitat of a particular species to conserving forests for fu-

ture utilization. Human interaction with the landscapes in and

around these public lands is a matter of concern in that those

activities may threaten the very resources the public lands are

designed to protect. Stated differently, the very act of setting

aside land may serve to threaten those resources that originally

stimulated the protective effort.

Embedded in the discussion of land use near protected areas

is the impact of changes in human activities near protected

landscapes, especially forests. Mixed empirical results have

been found—in some cases, increases in population result in

more deforestation and development; in other cases, the role of

population is less clear or not important (see for examples

Frentz et al., 2004; Lambin & Geist, 2006). Previous research

on changes in land cover close to public land boundaries has

established that impacts differ by type of public land and oper-

ate in a complex manner (Moon & Farmer, 2010). The current

paper provides an empirical model demonstrating the complex-

ity of afforestation1 patterns resulting from the interaction of

human populations with the surrounding landscape. This re-

search purposefully builds on previous lines of inquiry (Moon

& Farmer, 2010) and specifically considers whether afforesta-

tion, a different type of land cover change, operates differently.

The specific focus of the current research is afforestation rather

than deforestation or forest loss. Deforestation is often the focus

of land cover research projects (see for examples Bhattarai &

Hammig, 2001; Carr, Suter, & Barbieri, 2005; Kok, 2004; Lam-

bin & Geist, 2003; Laurance et al., 2002; Perz, 2002; Radeloff

et al., 2001) because of its often deleterious environmental im-

pacts. However, the processes surrounding afforestation are of

import as well, on the assumption that afforestation at least

partially remediates negative consequences of forest loss, in-

cluding carbon sequestration, erosion control, and non-con-

sumptive use of forest products. The empirical evidence pre-

sented here underscores the importance of understanding how

publicly managed lands function within the larger landscape

matrix.

Analytic Framework

Two main concepts underlie the empirical orientation of this

research. The first is conceptualizing the public land boundaries

as an ecological switch (Wilson & King, 1995). A switch is

defined as a behavior or condition that results in changes in

vegetation such that those changes reinforce the behavior or

condition, forming a positive feedback loop. An example of an

ecological switch would be the processes surrounding a mown

pasture. If a person selectively mows a region in a pasture, the

next mower is likely to follow the evident outline of the previ-

ous mowing session. The “boundary” is provided by the taller

grass that has not been mown. The presence of the boundary

induces mowing up to the boundary but not beyond, maintain-

ing the behavior. With continuation of this pattern, the actual

mix of speciation of grasses will change as shorter grasses in-

crease in the mown area while taller grasses are more success-

ful in the unmown area. Thus the mowing process produces an

ecological switch.

1Afforestation is the establishment of a forest or stand in an area where the

preceding vegetation or land use was not forest while reforestation is the

reestablishment of previously existing forest cover. While some areas in the

study area may be reforested in that prior to 1993 these areas were forest

stands, the definition used in the study is technically the measurement o

those lands that were not classified forest in the beginning of the timeframe

(1993) but are by the end of the study timeframe (2004). Given a different

temporal framework, some of these areas could be considered reforesting.

The second concept in the grounding of this research relates

to forest transition theory. Some researchers have presented the

concept of a “forest transition”, in some ways similar to the

demographic transition (Caldwell, 1976), as an explanation for

the pattern of rapid deforestation followed by a slower affore-

Z. K. MOON ET AL.

station (Mather, 1992; Mather & Needle, 1998; Rudel, 1998;

Rudel et al., 2002). The theory proposes that the processes of

economic development, industrialization and urbanization im-

pact forests in foreseeable ways, suggesting a time sequence for

forest cover changes that are prompted by these socioeconomic

changes. According to forest transition theory, a relatively rapid

deforestation takes place initially but then the trend reverses

and a slower increase in forest cover occurs.

According to the theory, the initial stage is deforestation and

is generally linked to agricultural development and expanding

populations. Following this rapid deforestation, populations

stabilize, forest cover remains somewhat static, and then the

afforestation process begins slowly even if the population con-

tinues to grow. The reversal of the trend from deforestation to

recovery of forests is generally explained through two pathways

(Rudel et al., 2005). In the first, the rapid deforestation period is

driven by demands for agricultural lands and forest products,

especially for housing and transportation corridors. After this

first period of relatively rapid change, the local economy de-

velops and diversifies, farm labor moves into off-farm em-

ployment as agricultural production becomes less profitable

compared to other, non-farm economic activities. Some propor-

tion of the rural labor force may migrate to more densely popu-

lated areas. Marginal agricultural land is then abandoned as a

result and reverts to forest cover. Urbanization and economic

development, then, drive this transformation which in the latter

stages results in increasing forest cover. Purchasing abandoned

lands for parks or forest reserves reinforces the momentum of

this transformation. This pathway is exemplified by historic

patterns found in some developed countries, including the USA

(Rudel et al., 2005).

The second explanation is what Rudel et al. (2005) term the

“forest scarcity path”. This explanation relies less on transfor-

mation of the agricultural sector and attributes increases in the

price of scarce forest products as the primary driver for land-

owners to plant trees instead of crops or pasture grasses. The

state increases the momentum along this path through affore-

station programs for marginal lands. Evidence exists (Rudel et

al., 2005) that lower income countries with recent increases in

forest cover follow this pathway.

Regardless of the pathway by which the forest transition may

take place, evidence of a forest transition across the United

States has been found by other researchers (Evans & Kelley,

2008; Rudel, 1998; Rudel & Fu, 1996). More specifically, evi-

dence for either or both pathways to forest transition theory

may be extant in the study area (Rudel & Fu, 1996) as forest

stocks have expanded and changed across the region (Rosson &

Rose, 2010) in the face of significant socioeconomic changes

(Farmer et al., 2010).

In the current model, afforested lands and socio-environment

variables are identified that are associated with change over

approximately a ten year period of time. Population structures,

economic and transportation infrastructure, geophysical char-

acteristics and distance measures to urban centers and public

lands are considered. The primary hypothesis of this paper is

that the presence and location of the public lands has “spillover

effects” into the surrounding landscape. These effects are as-

sumed to be the result of the public land boundaries functioning

as ecological switches on surrounding privately held lands. A

secondary interest is to seek empirical support for the forest tran-

sition process in the study area. An examination of land cover

change in the following sections allows for empirical consid-

eration of these ideas.

Environmental History of the Study Area

The study area is eight counties in the Ozark Plateau Region

within the state of Arkansas and encloses the Buffalo National

River (BNR) and much of the Ozark-St. Francis National For-

est (NF). The Ozark National Forest covers a large percentage

of the study area’s southern portion while the Buffalo National

River lands bisect the study area along a roughly east-west axis.

Between 1906 and 1909, widespread and large-scale timber

cutting took place in the study area, which at that time was the

site of the last noteworthy stretches of virgin forest to be found

east of the Rocky Mountains. In addition to commercial lumber

production, small farmers cleared land during this time with no

intention of replanting trees. In reaction to the thin and highly

erodible soil, the farming strategy was to clear another location

when the first one gave out. Another common practice of the

time was burning woodlands in the fall or early spring, a prac-

tice that was highly damaging to the forest cover. Burning the

woodlands was done to create better forage and pastures for

stock and, according to the local folk wisdom, control ticks and

snakes.

As the federal government reacted to loss of timber reserves

across the entire country, broad areas of forest were withdrawn

from the public domain and placed in the nation’s forest re-

serves. Eventually, these lands became national forests, and

within the study area the Ozark National Forest itself was es-

tablished in 1908. The original understanding among local peo-

ple was that the withdrawal of these lands from the public do-

main into federal ownership to form the Ozark National Forest

would conserve the forests for future production needs of the

local citizens and as such was acceptable policy (Strausberg &

Hough, 1997). Within a year of the establishment of the na-

tional forest lands in Arkansas, however, the public’s attitude

changed dramatically. As the forest rangers began to curb now-

illegal but customary practices of using the land as a de facto

commons for stock forage, timber cutting, and burning and

clearing without regard to title, a public backlash occurred. Local

opposition to management of these lands by the US Forest Ser-

vice has continued from that time to the present, as local inter-

ests have continued to contend with the federally established

management objectives (Strausberg & Hough, 1997).

In 1972 the US House of Representatives established the

Buffalo River as the nation’s first National River. Management

of the Buffalo River fell under the purview of the National Park

Service and was designed to preserve the free-flowing nature of

the Buffalo River. With the establishment of the Buffalo Na-

tional River, the river was “preserved” and people living on

land adjacent to the river were removed over a period of time.

The establishment of the National River, while heralded by con-

servation and environmental activists, was met with “emotional

turmoil in the community regarding the disruption of life for

the Buffalo River residents, whether they were willing or un-

willing sellers” (Rogers, 2010). Administrative boundaries were

established to enforce only approved uses and limit access to

the Buffalo National River lands much as had happened with

US Forest Service lands.

The headwaters of the Buffalo River are included in US For-

est Service lands but most of the rest of the land outside the

National River’s boundaries is privately owned. Consequently,

ongoing concerns by environmental advocacy groups have been

focused on efforts to preserve the clarity and purity of the river

water by advocating restrictions on uses of private lands in the

Z. K. MOON ET AL.

river’s drainage system. Activists have also been involved in the

management plans of US Forest Service lands for the headwa-

ters and drainage of the Buffalo River (Rogers, 2010; Straus-

berg & Hough, 1997).

The current analysis considers these two different types of

public lands situated within the study area: the Buffalo National

River, under management by the National Park Service, and the

Ozark-St. Francis National Forest, under management by the

US Forest Service. The history of these two different public

lands captures many of the dilemmas surrounding the estab-

lishment of publicly managed lands. The different public land

management regimes also reflect different originating impulses

for their establishment—on the one hand, preservation of an

existing state is the rationale for the Buffalo National River

while conservation of a resource to be used for the broader

public good is evident for the Ozark-St. Francis National Forest.

Interactions along the boundaries of these two lands are hypothe-

sized to reflect these differences in management objectives.

Creation and continued management of these public lands are

not the only macro-level changes taking place in the study area.

Relatively rapidly growing human populations in the nearby

metropolitan areas have continued pressures on the landscape

matrix surrounding these two public lands through increased

infrastructure development, increased housing and development,

and increased demand for recreational uses. These changes

have impacted both US Forest Service and National Park lands

although not in identical ways (Moon & Farmer, 2010).

In order to understand afforestation patterns a spatial lag re-

gression model is provided to demonstrate the relationship of

socioeconomic, transportation, and geophysical variables with

afforestation at a region-wide level. Evidence is sought for the

hypotheses of a forest transition along with ecological switch-

ing behavior associated with the boundaries of the public lands.

Data and Methods

Data

The models presented below provide an examination of af-

forestation. The endogenous variable (AFFORESTED) is cre-

ated from satellite imagery from 1993 and 2004. The previously

processed imagery2 (Arkansas Natural Resource Conservation

Commission and the University of Arkansas’ Center for Ad-

vanced Spatial Technologies October 31, 2005; US Geological

Survey January 1, 1999) used for the scenes contain both pro-

jection and extent information. The 1993 scene has a resolution

of 29.34 meters/pixel while the 2004 imagery has a resolution

of 28.5 meters/pixel. These images are pre-processed and con-

verted into land cover classes by the original authors. For this

study, the images were clipped to the study area and reclassed

for use in a series of binary maps. The change matrix included a

category called “Afforested” defined as a change from any

other category to forest cover3.

A series of density surfaces were created to generate the ex-

ogenous variables. A random point was generated on the study

area map and a grid created from that point with 1-kilometer

spacing in each cardinal direction across the study area. The

resultant grid contains 11,576 one-kilometer square blocks. These

blocks become the individual records to be analyzed. Density

surfaces were created from socio-demographic data derived from

1990 and 2000 census block data (US Department of Com-

merce Washington DC: Bureau of the Census [producer] 1991,

1993, 2002a, 2002b) using a method developed earlier (see Moon

& Farmer 2001 for full details).

Additional locational information (a database of rural resi-

dential structures obtained from a state agency) was used to

convert areal census data into point data. Then, based on the

population density surface, other socioeconomic data were

allocated across the grid. A difference surface was created for

each of the socioeconomic variables and these differences were

then summed for each 1 square kilometer block and stored in

the data matrix. Direct line distance measures were calculated

for each of the distance variables, using the centroid of the 1

square-kilometer block under consideration to the centroids of

the closest metropolitan and micropolitan polygons (US Census

Bureau—Geography Division 2006), to the center of the near-

est segment of each type of public land (BNR and NF) bounda-

ries (AHTD, 2006) and the center of the nearest major highway

segment (AHTD, 2000). Thirty-meter digital elevation model

data (AHTD, 2001) provided the slope and aspect measures.

Road density was calculated by summing road length (in kilo-

meters, paved and unpaved) in each analytical unit and dividing

by the area of the unit. Thus, a matrix was created with one row

for each of the 1 km × 1 km blocks and columns for the 15

exogenous variables.

AFFORESTED, the measure of the extent of afforestation,

was calculated from a binary map of pixels that changed from

any other category to “Forest”. This binary map was overlaid

with the same 1 km × 1 km block grid. The area within each

block that changed into the AFFORESTED category was sum-

med, the percentage of the total block area calculated, and the

percentages stored. All data was converted to standardized

z-scores, eliminating problems in interpreting parameters in dif-

ferent metrics.

The variables population density (POPDEN), percentage of

working age people (WORKERS), percentage of in-migrants in

the previous five years (MOVERS), percentage of the em-

ployed in natural resource extraction activities such as forestry

or mining (NATRES), percentage of new homes constructed in

the last five years (NEWHOMES), percentage of homes using

electricity as the primary heating source (ELECTRICITY), and

the percentage of homes with phone service (PHONES) were

drawn from the 1990 and 2000 Census data (US Department of

Commerce Washington DC: Bureau of the Census [producer]

1991, 1993, 2002a, 2002b), calculated as change over the time

period. Additional variables include road density (RD_DENS)

calculated as the number of kilometers of road within the 1 km

× 1 km block.

2Imagery was retrieved from GeoStor, a public domain geospatial data re-

pository managed by the Arkansas Geographic Information Office and pilo-

ted by University of Arkansas’ Center for Advanced Spatial Technologies.

Available at http://www.geostor.arkansas.gov.

3As noted earlier, afforestation is used rather than reforestation. Transitional

areas and shrubland were classified as herbaceous and not included in affor-

estation. However, intentional planting (landscaping or plantation) was not

separable from the imagery obtained. Therefore whether the afforestation is

directly the result of natural re-growth or plantation/landscaping growth can

only be suggested by location.

Other distance variables include distance to nearest major,

heavily traveled highway (DIST_HWY), to nearest metropolis-

tan area (DIST_METRO), to nearest micropolitan area (DIST_

MICRO), to the border of the Buffalo National River park lands

(DIST_BNR), and to the Ozark-St. Francis National Forest lands

border (DIST_NF). Completing the variables are percentage of

a given block’s area with southern aspect (PCTSOUTH) and

the percentage of area that has little slope (PCTFLAT). These

Z. K. MOON ET AL.

two variables are taken from the DEM data. Table 1 contains

definitions, the first two moments, and Moran’s I for all vari-

ables.

Methods

As noted above, the extent of afforestation is analyzed using

a spatial lag model, which provides an empirical insight into

regional influences. Ordinary least squares regression analysis

was performed and is provided for comparison. Residuals from

the OLS model demonstrated significant spatial autocorrelation,

suggesting the violation of the assumption of independence.

Additionally, diagnostic techniques (Anselin et al., 2006), spe-

cifically the Lagrange Multiplier for the lag variable as well as

for the error term, indicate the appropriateness of using the

spatial lag model. Results for the OLS and spatial lag models

are presented in Table 2.

yafforestation = β1POPDEN + β2WORKERS + β3MOVERS +

β4RD_DENS + β5NATRES + β6NEWHOMES +

β7DIST_ΒNR + β8DIST_NF + β9PCTSOUTH +

β10PCTFLAT + β11DIST_HWY +

β12DIST_METRO + β13DIST_MICRO +

β14ELECTRICITY + β15PHONES +

β16y*lag + ε

Results

Spatial lag regression findings provide a regional overview

of the processes associated with afforestation in the study area.

The endogenous variable is “Afforested” or that percentage of

the area within each 1 square kilometer block analytical unit

that changed from any other type of land cover to “Forest”. As

would be expected from an examination of the Moran’s I values

for the variables, a regression model accounting for spatial

dependencies significantly improves the model fit over an or-

dinary least squares (OLS) approach. Inclusion of the spatial

lag variable substantively improved the adjusted R-square, AIC,

and reduced the spatial autocorrelation in the residuals (see

Table 2 for comparison of OLS and spatial lag regression mod-

els). The spatial lag model explains nearly 60% of the variation

in the model (R-square = 0.59). The lag variable is significant

and the strongest parameter in the model, demonstrating the

spatial association of changes in land cover due to afforestation.

With the exception of population density, the demographic

and economic factors included are not significant. Change in

population density is negative, indicating that increasing popu-

lation densities result in less afforestation. This is as anticipated

given the rapidly growing population levels in this area over the

time period in question.

The road density parameter is positive while the distance to a

major highway is negative. These seemingly contradictory in-

dicators, however, might be better understood as tied to devel-

opment corridors where land once cleared for farming is now

used for residential purposes, increasing the density of roads,

but also resulting in tree plantings around housing and with

nearby cleared lands reverting to forest cover. Housing devel-

opment is likely to take place near transportation arteries, re-

sulting in the relationship of increasing afforestation as distance

to a major highway decreases.

Table 1.

Univariates, Moran’s I, and definitions of variables.

Variable Mean

Standard

Deviation Variance Moran’s I Definition

POPDEN 2.649 15.078 227.35 0.144 Persons per square kilometer, change 1990-2000

MOVERS 0.136 0.596 0.36 0.573 Percentage who lived in different place 5 years ago, change 1990-2000

WORKERS –0.222 0.539 0.29 0.596 Percentage working age (16 - 64), change 1990-2000

NATRES –0.025 0.082 0.01 0.532

Percentage employed in natural resource activities (farming, forestry, mining), change

1990-2000

NEWHOMES 0.035 0.242 0.06 0.430 Percentage housing built within last 5 years, change 1990-2000

RD_DENS 45.757 41.072 1686.94 0.335 Kilometers per square kilometer

PCTFLAT 94.790 8.665 75.09 0.563 Percentage land with less than 20% slope

DIST_BNR 25.066 15.112 228.37 na Distance from centroid of analytical unit to nearest segment of Buffalo National River

boundary

DIST_NF 22.385 14.597 213.08 na Distance from centroid of analytical unit to nearest segment of Ozark National Forest

boundary

PCTSOUTH 36.099 15.362 236.00 0.176 Percentage land with southern aspect

ELECTRICITY 0.084 0.295 0.09 0.494 Percentage homes using electricity as heating source, change 1990-2000

PHONES 0.129 0.598 0.36 0.459 Percentage homes with telephones, change 1990-2000

DIST_METRO 94.125 35.261 1243.34 na Distance from centroid of analytical unit to centroid of nearest metropolitan statistical area

DIST_MICRO 35.785 15.769 248.67 na Distance from centroid of analytical unit to centroid of nearest micropolitan statistical area

DIST_HWY 2.379 1.864 3.47 na Distance from centroid of analytical unit to nearest segment of state highway

AFFORESTED 7.76 7.37 54.36 0.587 Change to forest from any other land cover category

Z. K. MOON ET AL.

Table 2.

Ordinary least squares and spatial lag models compared; standardized

coefficients.

OLS model Spatial lag model

Variable Coefficient Std. Error Coefficient Std. Error

W_REF 0.751** 0.008

CONSTANT 0.000** 0.008 0.000 0.006

RD_DENS 0.084** 0.009 0.104** 0.007

PCTFLAT 0.242** 0.008 0.090** 0.006

DIST_BNR –0.232** 0.012 –0.049** 0.010

DIST_METRO –0.291** 0.010 –0.071** 0.008

DIST_MICRO –0.046** 0.010 –0.007 0.008

DIST_NF 0.331** 0.011 0.073** 0.009

DIST_HWY –0.094** 0.008 –0.017* 0.006

PCTSOUTH –0.055** 0.008 –0.040** 0.006

POPDEN –0.022* 0.008 –0.025** 0.006

MOVERS 0.049* 0.016 –0.005 0.012

NATRES 0.006 0.009 0.005 0.006

NEWHOMES 0.009 0.010 0.012 0.008

ELECTRICITY

–0.037* 0.012 –0.015 0.009

PHONES –0.002 0.017 0.012 0.013

WORKERS 0.039* 0.015 0.014 0.011

Log likelihood –14580 –11939

Akaike info criterion 29,192 23,913

Schwarz criterion 29,310 24,038

R-squared 0.273 0.586

Sigma-square 0.72800 0.4143

S.E. of regression 0.85323 0.6437

*p < 0.05; **p < 0.01.

In conjunction with transportation infrastructure, distance

to business centers is considered. Distance to a micropolitan

area—a smaller business center—is not significant. However,

the parameter for distance to the nearest metropolitan center is

negative and significant. This parameter may also reflect the link

to afforestation that concurs with increasing urbanization of the

populace, suggested by forest transition theory.

Topographical features as measured by slope and aspect also

come into play. The greater the percentage of flat lands in an

area, the greater the afforestation. Conversely, the decreasing

amounts of land with southern exposure results in increasing

afforestation. These factors also may reflect development pref-

erences and/or the reversion of land cleared previously for ag-

ricultural uses now returning to forest cover.

Turning now to the relationship between afforestation and

the boundaries of public lands, the model supports the hypothe-

sized relationships. Distance to the Buffalo National River is

negative, meaning that more afforestation is present the closer

the land is to the boundary of the Buffalo National River. The

relationship between the boundary of the Ozark-St. Francis

National Forest lands and afforestation is, however, positive.

This indicates that there is less afforestation on lands closer to

the Ozark-St. Francis National Forest boundaries. These find-

ings suggest that the process of land cover change in this region

is driven by environmental and locational considerations.

Limitations

The methods and data presented here have some limitations.

Land cover images inherently have post-processing errors.

Census data itself has well-known difficulties (US Department

of Commerce Washington DC: Bureau of the Census [producer]

1991, 1993, 2002a, 2002b). However, the process used to dis-

aggregate areal data itself introduces little additional error (Moon

& Farmer, 2001). Measurements of distance between centroids

of polygons or polyline segments provided in publicly available

files may contain inaccuracies in the original data providing an

unknown source of error. Other measures, such as actual trans-

portation networks, may arguably be a better measure of acces-

sibility to metropolitan or micropolitan centers than the use of

polygon centroids used here. Intensity of computation combined

with substantive operationalization questions regarding the

appropriate pathways through a network argued for a simpler

approach.

Conclusion

The findings here make two important points. First, the pub-

lic land boundaries are shown to act as ecological switches.

Second, results underscore the importance of understanding

how publicly managed lands with different mandates function

within the larger social as well as geophysical landscape matrix.

The original approach was to analyze whether the boundaries

of public lands act as ecological switches, inducing some par-

ticular human activities in proximity to the boundary. The an-

swer is yes, and is seen in the empirical demonstration that each

type of public land functions differently in the model and is a

significant contributor in understanding the variation of affore-

station across the study area.

Proximity to the BNR boundary is associated with increases

in afforestation, while proximity to the NF lands is associated

with decreases in afforestation. These opposite influences can

be understood in the context of the different rationales for es-

tablishment of each of the public lands. The socially con-

structed objectives for each area—preservation in the case of

the Buffalo National River and conservation of renewable re-

sources for future use in the case of the Ozark-St. Francis Na-

tional Forest lands—provide the lens through which individuals

may view appropriate use of the land in close proximity to ei-

ther of the different types of public lands. Afforestation takes

place near the National River as persons view that area as ap-

propriate for a “return to a more natural state” whereas timber

is a resource to be used as needed, even if conserved for a pe-

riod of time, in areas associated with the US Forest Service

lands.

Some empirical support for using forest transition theory to

understand afforestation patterns in this region is provided. As

posited by forest transition theory, afforestation appears linked

to areas undergoing urbanizing and development. Beyond po-

pulation density, other socio-economic indicators were not sig-

nificant in explaining afforestation, but locational and topog-

raphical indicators associated with urbanization and develop-

ment were influential in explaining afforestation.

The empirical evidence here underscores the importance of

understanding how publicly managed lands function within the

larger social and environmental landscape matrix. The concept

of ecological switching, coupled with recognition of the so-

cially constructed goals for the publicly managed lands provid-

Z. K. MOON ET AL.

ing the context for decision-making, gives managers and policy

makers a method for anticipating land cover changes in the

future specific to the particular type of public land.

REFERENCES

Anselin, L., Syabri, I., & Kho, Y. (2006). GeoDa: An introduction to

spatial data analysis. Geographical Analysis, 38, 5-22.

doi:10.1111/j.0016-7363.2005.00671.x

Arkansas Highway and Transportation Department (AHTD). (2000).

Roads, all. Center for Advanced Spatial Technologies, University of

Arkansas.

Arkansas Highway and Transportation Department (AHTD). (2001).

Digital elevation model. Fayetteville, AR: University of Arkansas.

Arkansas Highway and Transportation Department (AHTD). (2006).

Public land boundaries (polygon). URL (last checked 1 October

2006). http://www.geostor.arkansas.gov

Arkansas Natural Resource Conservation Commission and the Univer-

sity of Arkansas’ Center for Advanced Spatial Technologies. (2005).

Land use land cover fall 2004 (raster).

Bhattarai, M., & Hammig, M. (2001). Institutions and the environ-

mental Kuznets Curve for deforestation: A crosscountry analysis for

Latin America, Africa and Asia. World Development, 29, 995-1010.

doi:10.1016/S0305-750X(01)00019-5

Caldwell, J. C. (1976). Toward a restatement of demographic transition

theory. Population and Development Review, 2, 321-366.

doi:10.2307/1971615

Carr, D. L., Suter, L., & Barbieri, A. (2005). Population dynamics and

tropical deforestation: State of the debate and conceptual challenges.

Population and Environment, 27, 89-113.

doi:10.1007/s11111-005-0014-x

Evans, T. P., & Kelley, H. (2008). Assessing the transition from defor-

estation to forest regrowth with an agent-based model of land cover

change for south-central Indiana (USA). Geoforum, 39, 819-832.

doi:10.1016/j.geoforum.2007.03.010

Farmer, F. L., Miller, W. P., Moon, Z. K., & Goforth, L. (2010). Rural

profile of Arkansas, 2009: Social and economic trends affecting ru-

ral Arkansas. Fayetteville, AR: Division of Agriculture, University

of Arkansas.

Frentz, I. C., Farmer, F. L., Guldin, J. M., & Smith, K. G. (2004). Pub-

lic lands and population growth. Society & Natural Resources, 17,

57-68. doi:10.1080/716100627

Kok, K. (2004). The role of population in understanding Honduran land

use patterns. Journal of Environmental Management, 72, 73-89.

doi:10.1016/j.jenvman.2004.03.013

Lambin, E. F., & Geist, H. J. (2006). Land-use and land-cover change,

global change—The IGBP series. Berlin: Springer.

Lambin, E. F., & Geist, H. J. (2003). Regional differences in tropical

deforestation. E nvironment, 45, 22-36.

Laurance, W. F., Albernaz, A. K. M., Schroth, G., Fearnside, P. M., Ber-

gen, S., Venticinque, E. M., & Da Costa, C. (2002). Predictors of de-

forestation in the Brazilian Amazon. Journal of Biogeography, 29,

737-748. doi:10.1046/j.1365-2699.2002.00721.x

Mather, A. S. (1992). The forest transition. Area, 24, 367-379.

Mather, A. S., & Needle, C. L. (1998). The forest transition: A theo-

retical basis. Area, 30, 117-124.

doi:10.1111/j.1475-4762.1998.tb00055.x

Moon, Z. K., & Farmer, F. L. (2001). Population density surface: A

new approach to an old problem. Society & Natural Resources, 14,

39-49. doi:10.1080/089419201300199545

Moon, Z. K., & Farmer, F. L. (2010). Human induced switches on

public lands boundaries: The emergence of ecological islands? Soci-

ety & Natural Resources, 23, 1-17. doi:10.1080/08941920802317606

Perz, S. G. (2002). The changing social contexts of deforestation in the

Brazilian Amazon. Social Science Quarterly, 83, 35-52.

doi:10.1111/1540-6237.00069

Radeloff, V. C., Hammer, R. B., Voss, P. R., Hagen, A. E., Field, D. R.,

& Mladenoff, D. J. (2001). Human demographic trends and land-

scape level forest management in the northwest Wisconsin Pine Bar-

rens. Forest Science, 47, 229-241

Rogers, S. (2010). Buffalo National River: The encyclopedia of Arkan-

sas history & culture. URL. (last checked 2 May 2010).

http://www.encyclopediaofarkansas.net/encyclopedia/entry-detail.aspx?

search=1&entryID=7

Rosson Jr., J. F., & Rose, A. K. (2010). Arkansas' forests, 2005. (S. R.

Station Ed.) Asheville, NC: US Department of Agriculture Forest

Service.

Rudel, T. K., Bates, D., & Machinguiashi, R. (2002). A tropical forest

transition? Agricultural change, out-migration, and secondary forests

in the Ecuadorian Amazon. Annals of the Association of American

Geographers, 92, 87-102. doi:10.1111/1467-8306.00281

Rudel, T. K. (1998). Is there a forest transition? Deforestation, refores-

tation, and development. Rural Sociology, 63, 533-552.

doi:10.1111/j.1549-0831.1998.tb00691.x

Rudel, T., & Fu, C. (1996). A requiem for the Southern Regionalists:

Reforestation in the South and the uses of regional social science.

Social Science Quarterly, 77, 804-820.

Rudel, T. K., Coomes, O. T., Moran, E., Achard, F., Angelsen, A., Xu,

J., & Lambin, E. (2005). Forest transitions: Towards a global under-

standing of land use change. Global Environmental Change, 15, 23-

31. doi:10.1016/j.gloenvcha.2004.11.001

Strausberg, S., & Hough, W.A. (1997). The Ouachita and Ozark-St.

Francis National Forests: A history of the lands and USDA Forest

service tenure. General Technical Report SO-I 21. Washington, DC:

US Department of Agriculture.

US Census Bureau—Geography Division. (2006). TIGER/line files, 2006

Second edition. US Department of Commerce. URL. (last checked 8

March 2006). http://www.census.gov/geo/www/tiger

US Department of Commerce Washington DC: Bureau of the Census

(producer). (1991). Census of population and housing, 1990 (UNITED

STATES): Summary Tape File 1A (Computer file). Ann Arbor, MI:

Inter-University Consortium for Political and Social Research (dis-

tributor), 1999.

US Department of Commerce Washington DC: Bureau of the Census

(producer). (1993). Census of population and housing, 1990 (UNITED

STATES): Summary Tape File 3A (Computer file). Ann Arbor, MI:

Inter-University Consortium for Political and Social Research (dis-

tributor), 1999.

US Department of Commerce Washington DC: Bureau of the Census

(producer). (2002a). Census of population and housing, 2000 (UNITED

STATES): Summary File 1, Final National (Computer file). Ann

Arbor, MI: Inter-University Consortium for Political and Social Re-

search (distributor), 2003.

US Department of Commerce Washington DC: Bureau of the Census

(producer). (2002b). Census of population and housing, 2000 (UNITED

STATES): Summary File 3, Arkansas (Computer file). Ann Arbor,

MI: Inter-University Consortium for Political and Social Research

(distributor). 2002.

US Geological Survey. (1999). Landcover 30 M national landcover da-

taset 1992 (raster).

Wilson, J. B., & King, W. M. (1995). Human-mediated vegetation switches

as processes in landscape ecology. Landscape Ecology, 10, 191-196.

doi:10.1007/BF00129253