Content uploaded by Anne M. Trainor
Author content
All content in this area was uploaded by Anne M. Trainor on Mar 31, 2022
Content may be subject to copyright.
Evaluating the effectiveness of a Safe Harbor Program for
connecting wildlife populations
A. M. Trainor1, J. R. Walters2, D. L. Urban3& A. Moody1,4
1 Department of Geography, University of North Carolina, Chapel Hill, NC, USA
2 Department of Biological Sciences, Virginia Tech, Blacksburg, VA, USA
3 Nicholas School of the Environment, Duke University, Durham, NC, USA
4 Curriculum of Ecology, University of North Carolina, Chapel Hill, NC, USA
Keywords
conefor sensinode; connectivity; dispersal
behavior; graph theory; Picoides borealis;
urban growth modeling.
Correspondence
Anne M. Trainor. Current address: School of
Forestry & Environmental Studies, 370
Prospect Street, Yale University, New
Haven, CT 06511, USA.
Fax: +1 203 432 3929
Email: anne.trainor@yale.edu
Editor: Darren Evans
Associate Editor: James Austin
Received 27 May 2012; accepted 5
February 2013
doi:10.1111/acv.12035
Abstract
Conversion of lands to agriculture and development within remaining natural
habitats have fragmented ecosystems and reduced wildlife populations. The US
Fish and Wildlife Service has adopted an incentive-based conservation strategy
known as the Safe Harbor Program (SHP) to reduce the vulnerability of federally
protected species located on private properties. In addition to protecting popula-
tions, the SHP also strives to enhance species viability by devising strategies to
(re)connect populations among habitat patches. We empirically evaluated the
effectiveness of the initial Safe Harbor agreement, developed for Red-cockaded
Woodpeckers (Picoides borealis, hereafter RCW) in the North Carolina Sandhills,
in enhancing connectivity for that species. According to our results, breeding
territories located on private properties enrolled in the SHP promoted dispersal of
RCWs and enhanced overall population connectivity relative to otherwise similar
territories located on non-SHP properties. Moreover, the SHP created extensive
stepping-stone corridors throughout the region. Our analysis also showed that
RCW connectivity will be negatively impacted directly and indirectly by encroach-
ing urban growth. By combining an urban growth model with estimated connec-
tivity, managers and conservation planners can identify which properties critical
for connectivity are most threatened by urban encroachment. These results can
help conservation planners develop strategic actions on private land based on the
species specific movement ability, current landscapes and projected urban growth.
Introduction
Human activities are causing high rates of biodiversity loss
by reducing, degrading and fragmenting natural ecosystems
globally (Hoekstra et al., 2005; Millennium Ecosystem
Assessment, 2005). In the US, efforts to conserve biological
diversity are complicated because >90% of federally pro-
tected species habitat is located on private property (GAO,
1994; Miller & Hobbs, 2002), which is especially threatened
by adverse land transformations (Dale et al., 2000; Fahrig,
2003). Owners of private lands that harbor federally pro-
tected species are not required to manage habitat to enhance
species persistence and recovery unless the owners proposed
activities might detrimentally affect habitat, in which case
they are required to develop habitat conservation plans
(HCPs) designed to compensate for species’ habitat losses
(Beatley, 1996; Bingham & Noon, 1998; U.S. Fish and
Wildlife Service, 2003a; Wilcove, 2004). Thus, protecting
biodiversity and ensuring the persistence of endangered
species will rely to a great extent on private landowners
voluntarily participating in implementing conservation
actions (Leopold, 1991; Beatley, 1996). To this end, many
government agencies have adopted incentive-based conser-
vation strategies that reward private landowners for restor-
ing and enhancing habitat on their properties (Bean, 1998).
One of these voluntary, incentive-based conservation
strategies in the US is the Safe Harbor Program (SHP; U.S.
Fish and Wildlife Service, 2004) introduced by the US Fish
and Wildlife Service (USFWS) in 1995. SHP participants
agree to manage their properties for federally protected
species by enhancing, restoring or maintaining habitat in
order to maintain the current or ‘baseline’ level of those
species present at the time of the agreement (Bonnie, 1997;
US Department of the Interior, 2006). In exchange for these
actions, the USFWS guarantees that participating property
owners incur no additional liability due to increases in the
federally protected species on their property (Bonnie, 1997;
Zhang & Mehmood, 2002; Wilcove & Lee, 2004). The SHP
has grown to include over 35 endangered species, with
habitat restoration projects covering nearly 2 million
acres (Wilcove, 2004). In addition to preserving existing
habitat, the SHP is intended to enhance species viability by
bs_bs_banner
Animal Conservation. Print ISSN 1367-9430
Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London 1
connecting or reconnecting populations (US Fish and
Wildlife Service, 2004). Despite growing participation in the
program, it has yet to be demonstrated that the SHP ben-
eficially impacts connectivity.
Lack of knowledge about the effectiveness of the SHP in
promoting connectivity results largely from the difficulty of
acquiring the necessary data and the duration of SHP agree-
ments. Most species have been integrated into SHP agree-
ments relatively recently, making it too early to directly
examine population responses to conservation actions moti-
vated by SHP (Wilcove & Lee, 2004). In addition, the move-
ment behavior of most protected species and their responses
to different landscape configurations are poorly understood
(Cantwell & Forman, 1993; Urban & Keitt, 2001; Urban
et al., 2009), making it difficult to reliably estimate connec-
tivity. Our aim is to take the necessary first step in evaluat-
ing the impact of SHP on connectivity using its inaugural
agreement, for the Red-cockaded Woodpecker (Picoides
borealis, hereafter RCW) in the North Carolina Sandhills
(US Department of the Interior, 1995). For this initial agree-
ment, sufficient time has elapsed to quantitatively evaluate
the response to conservation actions using extensive records
collected in the region. RCW populations in the Sandhills
ecoregion have been extensively researched with several
long-term monitoring projects evaluating habitat require-
ments, demography, group composition and dispersal
behavior (Costa & Daniels, 2004). For instance, an ongoing
30-year monitoring program is located within the boundary
of the original Safe Harbor agreement. Moreover, extensive
mark-recapture and radio telemetry movement data in this
area were recently integrated into RCW habitat connectivity
estimates (Trainor et al., in press). These monitoring efforts
provide the necessary demographic and movement data
needed to empirically evaluate the impact of the SHP on
habitat connectivity. As data on movement behavior
become available for more federally protected species, the
approach applied in this study can be implemented for many
species exposed to a wider variety of landscapes and threats.
The goal of this study was to assess the utility of the SHP
by determining whether the presence and quality of RCW
breeding sites on SHP properties improves the population’s
connectivity. To increase the effectiveness of restoration
projects on improving the region’s connectivity, we assessed
the benefit of restoring inactive breeding sites by determin-
ing the potential for natural recolonization, which would
avoid costly translocation efforts. Finally, breeding sites on
private property are threatened by encroaching urbaniza-
tion and, thus, should be the focus for future enrollment in
the SHP. To this end, we assessed whether breeding sites on
Safe Harbor properties adequately mitigate detrimental
impacts of projected urban growth.
Methods
Study species
RCWs are endemic to mature longleaf pine ecosystems
(Conner et al., 2002; Rudolph, Conner & Schaefer, 2002;
Walters et al., 2002), which have been reduced to less than
3% of their original extent due to clearing, logging, fire
suppression and urbanization (Frost, 2006). RCWs are ter-
ritorial, cooperative breeders (Walters, 1990), in which a
single-family group defends a home territory that includes
multiple nesting and roosting cavities excavated in living
pines (Walters, Doerr & Carter, 1988). In the Sandhills,
dispersal movements have been observed as far as 31 km
(Walters, 1990; Kesler, Walters & Kappes Jr, 2010). Despite
potentially strong dispersal ability, connectivity can be
affected by a variety of natural and human-modified land-
scape features such as the presence of fire-suppressed
longleaf pine forests, open deforested areas and urbaniza-
tion (Conner & Rudolph, 1991; Trainor et al., in press).
Based on these studies, highly fragmented longleaf pine
forest intermixed with urban and agricultural development
could impede movements between disconnected territories.
Study area
The impact of SHP on connectivity was evaluated for a
3721-km2area (79°12′W 35°7′N; Fig. 1) in the Sandhills
ecoregion in North Carolina (Griffith, Omernik &
McGinley, 2007). This region contains rolling topography
at an average elevation of 103 m. Historically, the Sandhills
ecoregion was covered by fire-dependent longleaf pine Pinus
palustris woodlands, characterized by an open canopy with
minimal hardwood midstory and dense herbaceous under-
story vegetation (Provencher et al., 2001; Frost, 2006).
Much of the forest has been transformed into cropland,
pasture, urban land-use activities and mixed woodlands.
The remaining forests are primarily mixed-pine species
(longleaf, loblolly P.taeda, shortleaf P.echinata and pond
pine P.serotina) in second-growth forest with varying
amounts of hardwood understory and midstory (Griffith
et al., 2007).
The remaining longleaf pine forests are managed by
several types of landowners, including extensively managed
large tracts on state properties (e.g. Game Lands) and two
military installations, Fort Bragg and Camp Mackall
(Britcher & Patten, 2004). In total, the government-
managed properties cover <25% of the study area but
contain >85% of the established RCW territories (Table 1).
In contrast, >65% of the study area is privately owned but
contains <10% of the active RCW territories. As of 2010,
116 landowners were enrolled in the SHP, with 47 active
woodpecker territories on properties that varied in size from
0.15 to 16 km2(median =0.76 km2) and with varying land
uses including private forest, residential properties, horse
farms, golf courses and conservation lands. Enrolled
properties contain, on average, 2.71 (se =0.60) and 1.84
(se =0.32) active and inactive territories, respectively.
Private forest and residential land-use properties contribute
most to the number of SHP agreements, land mass and
quantity of territories (Supporting Information Table S1).
In contrast, horse farms and nongovernment conservation
properties enrolled in the SHP contribute the smallest
parcels and the fewest territories.
Evaluating Safe Harbor Program A. M. Trainor et al.
2Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London
Movement data
Long-term nest monitoring and banding of RCW has been
administered by the Sandhills Ecological Institute and the
Endangered Species Branch of the Department of Natural
Resources on Ft. Bragg. RCWs are marked with a unique
combination of bands following established protocols
(Walters et al., 1988) and monitored each breeding season.
Recorded dispersal events (n=259) of banded juvenile
females born between 2004 and 2007 were used in our analy-
ses. Those years were selected to ensure that movement data
Figure 1 A map of natal dispersal events observed during 2004 to 2007 involving juvenile female Red-cockaded Woodpecker Picoides borealis
born on and/or dispersing to territories on properties enrolled in the Safe Harbor Program (SHP) in the Sandhills region of North Carolina with
the percent distribution of such dispersal events among land-owner types (enclosed loops represent dispersers that dispersed to a territory with
the same land-ownership classification).
Table 1 The number of Red-cockaded Woodpecker Picoides borealis territories grouped by landownership in the North Carolina’s Sandhills
ecoregion
Ownership
Area Number of territories
Total (km2) % Area Active Inactive
Government
Federal 660 17.4 442 105
State 298 8.0 147 78
Municipal 9 0.2 0 2
Conservation NGO 74 2.0 18 3
Safe Harbor Properties 190 5.1 47 51
Non-SHP-enrolled private land 2448 65.8 15 85
NGO, nongovernmental organization; SHP, Safe Harbor Program.
A. M. Trainor et al. Evaluating Safe Harbor Program
Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London 3
were collected under similar environmental conditions and
landscape composition to the landscape analysis.
Models
Graph networks
We used a graph-theory approach to quantify the effects of
landscape features on movement ability and thus to deter-
mine if active and inactive territories on SHP properties
enhance connectivity for the entire RCW population. In this
approach, graph networks are composed of nodes (points),
which represent habitat patches (or territory centers for this
study), and edges (links), which connect these nodes and
represent friction-weighted distances (Adriaensen et al.,
2003) between nodes. Friction-weighted distances were
determined using a resistance surface map. The resistance
surface was derived from LiDAR-based (light detection and
ranging) landscape structure attributes in combination with
empirical data on the prospecting behavior of 34 radio-
tagged juvenile female RCWs (Trainor et al., in press). Fric-
tion values ranging from 1 to 100 were assigned to landscape
features, with 1 representing the most suitable prospecting
habitat and 100 representing the least suitable (most avoided)
habitat. This resistance surface was then used to calculate the
shortest accumulated travel cost, or friction-weighted dis-
tance (dij), from territory ito territory j(for all iand j) using
a cost distance model (ArcInfo Workstation; Environmental
Systems Research Institute, Redlands, CA, USA). A matrix
representing all combinations of territories iand jwas popu-
lated with friction-weighted distances from the center coor-
dinates of each pairwise combination of all 670 active
territories on the landscape.
Resistance along network edges can be represented using
a binary or probabilistic approach. We used a probabilistic
approach, as a binary approach may not be relevant for
observed dispersal events (Saura & Pascual-Hortal, 2007).
We converted the friction-weighted distances between terri-
tories into a probabilistic model. When movement data are
limited, a generic negative-exponential dispersal function is
commonly used to characterize skewed dispersal processes
(Urban & Keitt, 2001; Bodin & Saura, 2010; Baranyi et al.,
2011). An extensive dataset of observed dispersal events
from natal territories to each breeding territories (dij) was
used to estimate the probability of dispersal between all
territory pairs (pij) by fitting a distribution function. The pij
ranged from 0.0 to 1.0, with values near 1 indicating a strong
connection between two territories because they are func-
tionally close together, and values close to 0 indicating a
weak connection.
Analysis
Connectivity was derived from the probability of connectiv-
ity (PCnum) index (Saura & Pascual-Hortal, 2007). PCnum is
a network-level index that integrates habitat availability,
inter-patch dispersal probability and graph structure for all
nodes to quantify the probability that two individuals
randomly placed within the landscape fall into two points
(breeding territories for this study) that are reachable from
each other (interconnected) given the set of nhabitat patches
and the connections (pij) among them (Saura et al., 2011a,b).
The index is given by
PCnum === ∑∑ aa p
ijij
j
n
i
n*,
11 (1)
where pij* is the maximum product probability of all possi-
ble paths between territories iand j, including single-step
paths [Conefor Sensinode, Version 2.5.8; Saura & Torne,
2009). The variables aiand ajare defined as average habitat
quality in the core area of each territory (defined as a circu-
lar area with radius of 174 m from the territory center).
Because forest structure is an important determinate of
habitat quality for RCWs (Conner & Rudolph, 1991),
breeding territory quality was based on six LiDAR-derived
forest structure variables (Supporting Information Appen-
dix S1): (1) maximum vegetation height; (2) skewness of the
vegetation height distribution; and (3–6) percent cover at
four vegetation height classes (1–8, 8–13, 13–20 and greater
than 20 m). Using these six variables and locations of 670
active territories (Fig. 1), we estimated habitat quality with
Maxent (Version 3.3.1), a machine-learning algorithm for
predicting habitat suitability based on a maximum entropy
approach (Elith et al., 2006; Phillips, Anderson & Schapire,
2006). The breeding territories were randomly divided into
training (75%) and testing (25%) points. To account for
variation in training and testing datasets, bootstrapping
with 10 replicate samples (with replacement) was employed.
The final habitat suitability map was an average of 10
habitat quality models. Habitat suitability values near one
indicate high quality habitat, while poor quality is indicated
by values close to zero.
Active territories on government-managed and privately
owned land were prioritized according to their ability to
connect the entire population. Using Conefor software,
nodes (i.e. active breeding territories) were iteratively
removed from the network and PCnum was recalculated for
each iteration (Saura & Pascual-Hortal, 2007; Saura et al.,
2011a,b). The percent importance of losing each active ter-
ritory (dPC) is calculated according to
dPC PC PC
PC
remove
num remove
num
=−×100 (2)
where PCremove is the connectivity value after each active
territory was removed from the network. Similarly, the
percent importance of restoring inactive territories on
private land was calculated by iteratively adding each inac-
tive territory from the network and recalculating the PCnum
according to
dPC PC PC
PC
add
add num
num
=−×100, (3)
where PCnum are the connectivity values with only active
territories and PCadd is the connectivity value after an
Evaluating Safe Harbor Program A. M. Trainor et al.
4Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London
inactive territory was added (Saura & Pascual-Hortal, 2007;
Saura et al., 2011a,b).
Recently, Saura & Rubio (2010) developed an approach
for node addition or removal analysis that partitions
the dPC connectivity metric into three fractions that
evaluate how each node contributes to connectivity
(dPC =dPCintra +dPCflux +dPCconnect). dPCintra repre-
sents the habitat quality within a single patch (since
i=jai¥aj=a2). dPCconnect is dependent on the node’s posi-
tion within the network and reflects a node’s importance as a
stepping stone. dPCflux incorporates both the habitat
quality and the juxtaposition of the habitat network by
quantifying a habitat quality-weighed dispersal flux value for
each habitat patch. dPCflux estimates how well a patch is
connected to other patches in the landscape based on the
potential flux. Partitioning the overall connectivity into three
separate metrics for each active and inactive territory allows
a common currency to quantify how each breeding territory
contributes to different roles in overall habitat connectivity.
The degree to which territories located on SHP properties
facilitate connectivity when facing encroaching urbaniza-
tion was estimated using the SLEUTH model to project
urban growth (Clarke, Hoppen & Gaydos, 1997; Support-
ing Information Appendix S2). Projected urban growth
for 2050 and 2100 was used to examine changes in RCW
connectivity by increasing friction values of cells in the
resistance surface with >50% probability of undergoing
urbanization to match the current landscape’s urban fric-
tion value estimates for RCW (average =53.39, sd =29.3).
Friction-weighted distances between all territories were then
recalculated (Supporting Information Appendix S2). These
updated friction-weighted distances for 2050 and 2100 sce-
narios were converted into dispersal probabilities (see the
Graph networks section) and inserted into new graph net-
works. The square root of the PCnum index, the equivalent
connectivity metric (hereafter referred to as EC), was used
to express the overall connectivity value for the entire land-
scape. This metric allows a direct comparison of the changes
in connectivity with changes in the amount of habitat in the
landscape by quantifying the patch attribute (e.g. area or
quality) that would provide the same probability of connec-
tivity as the actual habitat pattern in the landscape. The
urban-growth-modified networks were used to calculate EC
and dPCremove for the 2050 and 2100 scenarios.
Results
Dispersal events observed with mark-recapture banding
data provided insight into the SHP contribution to RCW
connectivity (Fig. 1). Overall, 6% of the 259 dispersing juve-
nile females dispersed to and from territories on SHP prop-
erties. Nineteen birds dispersed from a natal territory
located on a SHP property to a breeding territory on par-
ticipating SHP properties. Twelve additional individuals
born on SHP properties dispersed to nine territories on
government-managed properties and three territories on
private properties not enrolled in the SHP. A similar
number of juvenile females (n=14) born on non-SHP
territories obtained breeding status on SHP properties.
According to pairwise t-tests with a Bonferroni adjust-
ment, both active (P=0.011) and inactive (P<0.001) terri-
tories on SHP private properties had significantly greater
habitat quality than territories on non-SHP participant
properties (Fig. 2a) with habitat quality on SHP properties
Figure 2 The average (⫾SE) habitat quality for territories (a) by landowner and (b) for private properties enrolled in the Safe Harbor Program (SHP)
only, by the participants’ land-use activities.
A. M. Trainor et al. Evaluating Safe Harbor Program
Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London 5
varying by status (active vs. inactive, P<0.001) and land-
use activities (P=0.016; Fig. 2b). Surprisingly, active terri-
tories located on SHP with the greatest average habitat
quality were in private forests and not conservation
properties (Fig. 2b).
Active territories with the greatest contribution to con-
nectivity throughout the entire region were situated on
federal government properties (Table 2). These territories
also experienced a wide range of connectivity values
(dPCremove: 0.04–1.18) that were unevenly distributed
throughout the properties (Fig. 3a). On state-owned prop-
erty, the average connectivity for active territories was lower
than for territories on federal and private properties
(Table 2). In addition, over 73% of the active territories
located on state properties contributed very little to the
population’s overall connectivity (Fig. 3a). Beyond
government-managed properties, the relative importance of
active territories to overall connectivity did not significantly
differ between territories on and off SHP private properties
(P=0.842). The key territories on private lands with the
greatest contribution to connectivity were located in
the center of the population near military installation
borders (Fig. 3a). Sixteen active territories on SHP
private properties strongly contributed to connectivity
Figure 3 The distribution and relative contribution to connectivity (dPC) for active (a) and inactive (b) Red-cockaded Woodpecker Picoides
borealis territories on privately owned land in North Carolina’s Sandhills region and the percent change in active territories dPCremove from current
conditions to projected urban growth in 2100 (warmer colors are an increase in connectivity while cooler colors represent a decrease in
connectivity) (c).
Table 2 The average (⫾SE) relative contribution to connectivity for
active (dPCremove) and inactive (dPCadd) Red-cockaded Woodpecker
Picoides borealis territories grouped by landownership in North
Carolina’s Sandhills ecoregion
Ownership dPCremove dPCadd
Federal 0.426 (0.025) 0.272 (0.022)
State 0.099 (0.006) 0.064 (0.007)
Safe harbor properties 0.201 (0.018) 0.109 (0.132)
Non-SHP-enrolled private land 0.196 (0.044) 0.064 (0.009)
SHP, Safe Harbor Program.
Evaluating Safe Harbor Program A. M. Trainor et al.
6Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London
(dPCremove >50%), while only four active territories not par-
ticipating in the SHP were estimated as high priority for
overall connectivity, all of which were located within 1 km
of the Fort Bragg boundary (Fig 3a).
The number of inactive territories was equally distributed
among government-managed properties and private lands
(Table 1), but the potential contribution of these territories
to connectivity of inactive territories varied greatly among
landowners. Similar to active territories, inactive territories
on federal properties have the greatest potential to contrib-
ute to connectivity. Inactive territories on non-SHP private
properties had the least potential contribution to connectiv-
ity (Table 2). Inactive territories on SHP-enrolled private
properties were predicted to contribute more to connectivity
than inactive territories on state-owned properties (Table 2).
As would be expected, inactive territories on the periph-
ery of the network are expected to contribute the least to
connectivity. However, inactive territories with the greatest
probability of being recolonized, and thus improve connec-
tivity, were located northwest of Fort Bragg and between
the Fort Bragg and Camp Mackall military installations
(Fig. 3b). Of the 34 inactive territories in the top 25th per-
centile of connectivity, 18 were located on 10 SHP proper-
ties with a variety of land-use activities (private forests =4,
residential properties =5, and horse farm =1).
Within dPC metrics, there are three main factors that
contribute to a territory’s overall connectivity: (1) habitat
quality (dPCintra); (2) how well a territory is connected to
other territories in terms of flux (dPCflux); (3) importance as
a stepping-stone (dPCconnect). A ternary plot was used to
disentangle how each factor contributes to overall connec-
tivity for territories located beyond government-managed
properties (Fig. 4). For active territories on SHP properties,
most of the connectivity was equally distributed among the
three fractions, suggesting that SHP territories with high
habitat quality are centrally positioned within the network
in a way that increases direct flux and provides stepping-
stone habitats (Fig. 4a). In contrast, most of the connectiv-
ity in active territories on non-SHP properties came from
their contribution as stepping-stones (dPCConnect).
The distribution of connectivity fractions varied among
inactive territories. Few inactive territories on private prop-
erties, regardless of enrollment in the SHP, contributed
equally to all three fractions of connectivity (Fig. 4b). The
few inactive territories with high overall connectivity
(dPCadd) were located on SHP properties north-west of the
Fort Bragg boundary (Fig. 3b) and had the greatest contri-
bution from habitat quality and flux. The remaining inactive
territories on private properties would largely contribute as
stepping-stones due to their poor habitat quality (low
dPCintra).
We observed both direct and indirect detrimental impacts
to connectivity due to projected urban growth in the region.
According to the 2050 projection, 7.5% of the region would
be converted into urban development, subsuming 7 active
territories and 13 inactive on private properties. Most of
these territories in danger of urban encroachment were esti-
mated to contribute greatly to the population’s overall con-
nectivity (dPC >50%). The 2100 forecasted urban growth
suggests that 19.5% of the region will be converted into urban
land-use activities, with an additional 11 active and 16 inac-
tive territories likely eliminated by urban encroachment. If
SHP participants remain in the program beyond 2050, they
could prevent 22.2 km2of the area region from being con-
verted to urban development and secure the safety of five
active territories that strongly contribute to the population’s
connectivity and five inactive territories with strong potential
of increasing connectivity. Moreover, participants remaining
in the program beyond 2100 would protect an additional 13
active and 11 inactive territories throughout a 54-km2area
covering 82% (n=95) of the SHP participants’ properties.
In addition to the direct impact of urban development on
territories, connectivity is expected to decrease with increas-
ing urban encroachment throughout the Sandhills region.
Based on the predicted EC for the entire region, urban
growth is expected to produce a 1.2% reduction in connec-
tivity by 2050 and a 5% reduction by 2100. Interestingly, the
relative importance (dPCremove) of active territories within
government-managed property is also expected to shift as a
result of urban growth. For instance, by 2100, territories on
Active Territory
45.56
11.39
34.17
34.17
45.56
45.56
dPCConnect
dPCIntra
dPCFlux/100
0.12
0.58
0.12
0.25
0.38
0.25
0.38
0.25
0.38
0.51
0.12
0.51
Inactive Territory
Territory
SHP
NonSHP
dPCConnect
dPCIntra
dPCFlux/100
Figure 4 A ternary plot showing the relative
contribution of a territory’s habitat quality
(dPCintra), position within the network
(dPCconnect), and habitat-weighted disper-
sal flux (dPCflux) to the overall connectivity
(dPC =dPCintra +dPCconnect +dPCflux)
for territories located beyond government-
managed properties. The relative size of the
dot is correlated with the overall connectiv-
ity (dPC), that is, larger dots represent ter-
ritories with greater connectivity.
A. M. Trainor et al. Evaluating Safe Harbor Program
Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London 7
the western edge of Fort Bragg are expected to decrease
their contribution to overall connectivity and the territories
on the eastern portion of Fort Bragg are expected to
increase their contribution to connectivity (Fig. 3c).
Discussion
The SHP has improved relations among federal government
agencies and private landowners. For instance, as of 2006,
there are no records of private landowners withdrawing
from the program or altered restored habitat, even though
participants are permitted to terminate their contract at the
end of the agreement (Wilcove & Lee, 2004; US Department
of the Interior, 2006). Based on growing participation and
retention of private landowners in the SHP program, the
USFWS views the program successful in increasing habitat
quality for many federally protected species (Wilcove & Lee,
2004). Our results show that private properties enrolled in
SHP can play an important role in species’ persistence by
connecting otherwise isolated habitat patches with a volun-
tary conservation program. In the North Carolina San-
dhills, this partnership has enhanced effective conservation
beyond the boundaries of protected areas.
Conservation planning outside of protected areas is best
serve by locating interventions to maximize species persist-
ence with limited financial resources (Saura et al., 2011a,b).
Similar to many voluntary programs, the USFWS con-
ducted a broad campaign throughout the Sandhills region
to initially enroll landowners into SHP (P. Campbell, pers.
comm.). These efforts have been rewarded with high enroll-
ment, albeit with an uneven distribution conservation ben-
efits. Our analysis enables a spatially explicit approach for
targeting future enrollment. Using remotely sensed data and
graph network models to estimate connectivity, agencies
can allocate limited resources toward where conservation
efforts will provide the greatest benefits.
While RCW can fly long distances, their dispersal and
prospecting movements can still be impeded by certain
land-cover types (Kesler et al., 2010; Trainor et al., in
press). As a result, some breeding sites may remain inactive
due to the inability of birds to reach them. To mitigate this
problem, managers have expanded the spatial distribution
of active territories and increased genetic variation by trans-
locating juvenile RCWs to territories beyond barriers
(Rudolph et al., 1992; Allen, Franzreb & Escano, 1993;
U.S. Fish and Wildlife Service, 2003b). However, the
USFWS has not permitted translocation of RCWs within
the Sandhills (J. R. W., pers. comm.); thus, restored terri-
tories have been solely dependent on natural recoloniza-
tion. Our approach can be used to locate barriers that can
potentially inhibit recolonization throughout the region
to develop strategic conservation planning beyond
government-managed properties.
Once locations for SHP activities are identified, managers
need to determine the specific actions required to enhance
the appropriate abiotic and biotic community elements.
Many SHP participants are provided technical guidance
and cost-share assistance to restore or improve habitat
(Bonnie, 1997; Bonnie, Campbell & Cantrell, 2004; Wilcove
& Lee, 2004). These funds have assisted SHP participants
with maintaining active territories and restoring inactive
territories, primarily by constructing artificial cavities
(Walters, 1991; U.S. Fish and Wildlife Service, 2003b;
Walters et al., 2009). By evaluating each of the three factors
influencing connectivity within the dPC metric in conjunc-
tion with LiDAR-derived forest structure measurements, we
are able to identify what type of management action would
be most appropriate for each territory to enhance move-
ments and increase the population’s overall connectivity.
This could indicate where management activities, such as
improving habitat quality with prescribed fire, could be
directed.
Conservation planning on private land needs to consider
future threats to biodiversity, such as urban growth (Wilcove
et al., 1998; Miller & Hobbs, 2002). The development of
urban growth models has allowed researchers to forecast
threat of urban encroachment for many cities throughout the
world (Silva & Clarke, 2002; Jantz, Goetz & Shelley, 2003;
Yang & Lo, 2003). We showed that the SHP program is
important for conserving the RCW population in the San-
dhills given projected urban growth in the region likely will be
negatively impacted with encroaching urban growth.
Although we assumed that government-managed properties
would be excluded from direct urban growth, a decline in
connectivity is still predicted for territories located on federal
land due to a decrease in movements beyond protected prop-
erties. This loss of connectivity throughout the region will
likely cause breeding sites on government-managed proper-
ties to become disconnected and increasingly isolated. By
using a graph-theoretic approach, we were able to empiri-
cally estimate the location of breeding sites on private lands
that are most vulnerable to urbanization, and this informa-
tion can now be used to develop strategic conservation plans
to mitigate the threats of urban encroachment. Specifically,
six separate SHP properties containing 18 territories are
expected to be lost due to urban growth could instead be
protected by voluntary efforts of private landowners.
Due to the large number of imperiled species on private
land, and the lack of management regulations (Wilcove
et al., 1998), conservation efforts on nonfederal land are
becoming vital for the survival and recovery of federally
protected species (U.S. Department of the Interior, 1999).
The USFWS initiated two conservation policies (HCP and
SHP) specifically address to private landowners. HCP
policy was created to resolve potential conflicts between
private development and endangered species protection by
allowing the private landowner to negatively impact feder-
ally protected species or their habitat in exchange for con-
servation measures that compensate for unavoidable loss
(Beatley, 1996; Bean & Wilcove, 1997; Kishida, 2001).
Widespread human modification and conversion of land has
motivated a broader geographic distribution of HCPs than
SHPs. SHPS provide a better outcome: before harming
species, private landowners voluntarily enter into SHP
agreements to proactively promote the persistence of the
listed species on the property (Kishida, 2001).
Evaluating Safe Harbor Program A. M. Trainor et al.
8Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London
Conclusion
Wildlife management typically focuses on habitat patches
located on protected or managed properties, and minimal
consideration is given to the surrounding private lands. Even
less attention is paid to connectivity among habitat patches
on private and government-managed properties. However,
degraded landscapes between these properties can restrict
movements and reduce genetic diversity while increasing the
probability of local populations becoming extinct (Fahrig &
Merriam, 1994). We have shown that SHP agreements, in
which private landowners voluntarily manage RCW breed-
ing sites and longleaf pine forest, increase connectivity
beyond the managed properties. In addition, we suggest that
species persistence can be increased by focusing voluntary
conservation programs on lands with the greatest impact on
overall habitat connectivity. While many connectivity plan-
ning efforts have focused on connecting protected areas, they
have been primarily developed for large mammals (Beier,
Majka & Spencer, 2008).
The growing success of SHP for a wide range of taxa and
ecosystems along with and our results show that SHP has the
potential to enhance connectivity between protected areas
for a wide variety of protected species. However, these efforts
have an uneven geographic distribution. Even through most
of the land in the eastern US is privately held, most SHP
agreements are located in the western US where extensive
systems of public lands are set aside to protect biodiversity
conservation exist. Greater outreach should be conducted in
the eastern US to enable a stronger and more connected
network of SHP properties with many different stakeholders.
Most funding and conservation activities within the US
are directed toward single-species conservation (Lambeck,
1997). Furthermore, connectivity is usually viewed as a
species-centric concept because dispersal and movement
behavior across landscapes can vary greatly between species
(Wiens, 2006). In addition to the physical environment, the
integrity of an ecosystem depends on a complex assortment
of species from multiple taxonomic groups. It is often argued
that conservation policy would be more effective if focused at
the ecosystem level (Hutto, Reel & Landres, 1987). There-
fore, it is important to expand single-species conservation
policy and actions to multiple-species conservation plans
that incorporate multiple species. The advent of remotely
sensed data, GIS technology and long-term movement data-
sets, will all the methods deployed in this paper, will become
more accessible a wider range of species to conduct a broad
scale comparison of SHP throughout the country. For
example, many species are strongly associated with multiple
vegetation structure characteristics (MacArthur & Horn,
1969; Erdelen, 1984; Walters et al., 2002) that are not com-
monly available in regional land-cover maps. With over 150
projects in 30 states spanning almost 800 000-km2area
[USGS Center for LiDAR Information Coordination and
Knowledge (CLICK) http://lidar.cr.usgs.gov/], future col-
laborative partnerships employing with LiDAR data sharing
and acquisition are becoming possible to extrapolate vegeta-
tion structure metrics for a wide species impacted by a single
SHP. As movement behavior becomes available for more
federally protected species, the approach applied in this study
can provide greater confidence in connectivity estimates for
other species. Finally, by combining biologically infused
connectivity model with a forecasted urban growth model,
we identified strategic locations for these future management
actions on private land. As movement behavior becomes
available for more federally protected species, the approach
applied in this study can be expanded to further evaluate the
impact of the SHP on connectivity of other species and
identify strategic locations for future management actions on
private land.
Acknowledgments
We are grateful to S. Miller and P. Campbell of the USFWS
for useful discussions and data related to the Sandhills SHP.
We would like to thank the Sandhills Ecological Institute
and Fort Bragg’s Endangered Species Branch for their
efforts in collecting and organizing the extensive monitoring
data. Z. I. Cleveland and two anonymous reviewers pro-
vided comments on earlier drafts of the paper which greatly
improved clarity. Funding for this project was provided by
the U.S. Department of Defense Strategic Environmental
Research and Development Program (RC-1471).
References
Adriaensen, F., Chardon, J.P., De Blust, G., Swinnen, E.,
Villalba, S., Gulinck, H. & Matthysen, E. (2003). The
application of ‘least-cost’ modelling as a functional land-
scape model. Landsc. Urban Plan. 64, 233–247.
Allen, D.H., Franzreb, K.E. & Escano, R.E.F. (1993). Effi-
cacy of translocation strategies for Red-cockaded
Woodpeckers. Wildl. Soc. Bull. 21, 155–159.
Baranyi, G., Saura, S., Podani, J. & Jordán, F. (2011).
Contribution of habitat patches to network connectivity:
redundancy and uniqueness of topological indices.
Ecol. Indic. 11, 1301–1310.
Bean, M.J. (1998). The Endangered Species Act and private
land: four lessons learned from the past quarter century.
Environ. Law Report. 28, 10701–10710.
Bean, M.J. & Wilcove, D.S. (1997). The private-land
problem. Conserv. Biol. 16, 1–2.
Beatley, T. (1996). Habitat conservation planning: endan-
gered species and urban growth. 2nd edn. Austin: Univer-
sity of Texas Press.
Beier, P., Majka, D.R. & Spencer, W.D. (2008). Forks in
the road: choices in procedures for designing wildland
linkages. Conserv. Biol. 22, 836–851.
Bingham, B.B. & Noon, B.R. (1998). The use of core areas
in comprehensive mitigation strategies. Conserv. Biol. 12,
241–243.
Bodin, Ö. & Saura, S. (2010). Ranking individual habitat
patches as connectivity providers: integrating network
analysis and patch removal experiments. Ecol. Model.
221, 2393–2405.
A. M. Trainor et al. Evaluating Safe Harbor Program
Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London 9
Bonnie, R. (1997). Safe harbor for the Red-cockaded
Woodpecker. J. Forest. 95, 17–22.
Bonnie, R., Campbell, P.V. & Cantrell, M.A. (2004). North
Carolina Sandhills Red-cockaded Woodpecker Safe
Harbor Program: current status and lessons learned. In
Red-cockaded Woodpecker: road to recovery: 174–179.
Costa, R. & Daniels, S.J. (Eds). Blaine: Hancock House
Publishers.
Britcher, J.J. & Patten, J.M. (2004). Red-cockaded Wood-
pecker management on Fort Bragg: then and now. In
Red-cockaded Woodpecker: road to recovery: 116–126.
Costa, R. & Daniels, S.J. (Eds). Blaine: Hancock House
Publishers.
Cantwell, M.D. & Forman, R.T.T. (1993). Landscape
graphs: ecological modeling with graph theory to detect
configurations common to diverse landscapes. Landsc.
Ecol. 8, 239–255.
Clarke, K.C., Hoppen, S. & Gaydos, L.J. (1997). A self-
modifying cellular automaton model of historical urbani-
zation in the San Francisco Bay area. Environ. Plan. B:
Plan. Design 24, 247–261.
Conner, R.N. & Rudolph, D.C. (1991). Forest habitat loss,
fragmentation, and Red-cockaded Woodpecker popula-
tions. Wilson Bull. 103, 446–457.
Conner, R.N., Shackelford, C.E., Schaefer, R.R., Saenz, D.
& Rudolph, D.C. (2002). Avian community response to
southern pine ecosystem restoration for Red-cockaded
Woodpeckers. Wilson Bull. 114, 324–332.
Costa, R. & Daniels, S.J. (2004). Red-cockaded Wood-
pecker: road to recovery. Blaine: Hancock House
Publishers.
Dale, V.H., Brown, S., Haeuber, R.A., Hobbs, N.T.,
Huntly, N., Naiman, R.J., Riebsame, W.E., Turner,
M.G. & Valone, T.J. (2000). Ecological principles and
guidelines for managing the use of land. Ecol. Appl. 10,
639–670.
Elith, J., Graham, C.H., Anderson, R.P., Dudík, M.,
Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F.,
Leathwick, J.R., Lehmann, A., Li, J., Lohmann, L.G.,
Loiselle, B.A., Manion, G., Moritz, C., Nakamura, M.,
Nakazawa, Y., McC. Overton, J., Peterson, A.T. &
Phillips, S.J. (2006). Novel methods improve prediction
of species’ distributions from occurrence data. Ecography
29, 129–151.
Erdelen, M. (1984). Bird communities and vegetation struc-
ture: I. correlations and comparisons of simple and
diversity indices. Oecologia 61, 277–284.
Fahrig, L. (2003). Effects of habitat fragmentation on biodi-
versity. Annu. Rev. Ecol. Evol. Syst. 34, 487–515.
Fahrig, L. & Merriam, G. (1994). Conservation of frag-
mented populations. Conserv. Biol. 8, 50–59.
Frost, C. (2006). History and future of the longleaf pine
ecosystem. In The Longleaf pine ecosystem: ecology, silvi-
culture, and restoration: 9–42. Shibu, J., Jokela, E.J. &
Miller, D.L. (Eds). New York: Springer.
GAO (Government Accountability Office). (1994). Endan-
gered Species Act: Information on Species Protection on
Nonfederal Lands. GAO/RCED-95-16, 1–29.
Griffith, G.E., Omernik, J.M. & McGinley, M. (2007).
Ecoregions of North Carolina and South Carolina. In
Encyclopedia of earth. Cleveland, C.J. (Ed.). Washington,
DC: Environmental Information Coalition, National
Council for Science and the Environment. Available from
http://www.eoearth.org/article/Ecoregions_of_North_
Carolina_and_South_Carolina (accessed 12 March
2013).
Hoekstra, J.M., Boucher, T.M., Ricketts, T.H. & Roberts,
C. (2005). Confronting a biome crisis: global
disparities of habitat loss and protection. Ecol. Lett. 8,
23–29.
Hutto, R.L., Reel, S. & Landres, P.B. (1987). A critical
evaluation of the species approach to biological conserva-
tion. Endangered Spec. Upd. 4, 1–4.
Jantz, C.A., Goetz, S.J. & Shelley, M.K. (2003). Using the
SLEUTH urban growth model to simulate the impacts of
future policy scenarios on urban and land use in the
Baltimore-Washington metropolitan area. Environ. Plan.
B: Plan. Design 30, 251–271.
Kesler, D.C., Walters, J.R. & Kappes, J.J., Jr (2010). Social
influences on dispersal and the fat-tailed dispersal distri-
bution in Red-cockaded Woodpeckers. Behav. Ecol. 21,
1337–1343.
Kishida, D. (2001). Safe Harbor agreements under the
Endangered Species Act: are they right for Hawaii?
U. Haw. L. Rev. 23, 507–553.
Lambeck, R.J. (1997). Focal species: a multi-species
umbrella for nature conservation. Conserv. Biol. 11, 849–
856.
Leopold, A. (1991). The river of the mother of God and
other essays. In Conservation economics: 384. Flader, S.L.
& Callicott, J.B. (Eds). Madison: University of Wisconsin
Press.
MacArthur, R.H. & Horn, H.S. (1969). Foliage profile by
vertical measurements. Ecology 50, 802–804.
Millennium Ecosystem Assessment. (2005). Ecosystems and
human well-being: synthesis. Washington: Island Press.
Miller, J.R. & Hobbs, R.J. (2002). Conservation where
people live and work. Conserv. Biol. 16, 330–337.
Phillips, S.J., Anderson, R.P. & Schapire, R.E. (2006).
Maximum entropy modeling of species geographic distri-
butions. Ecol. Model. 190, 231–259.
Provencher, L., Herring, B.J., Gordon, D.R., Rodgers,
H.L., Tanner, G.W., Hardesty, J.L., Brennan, L.A. &
Litt, A.R. (2001). Longleaf pine and oak responses to
hardwood reduction techniques in fire-suppressed San-
dhills in northwest Florida. Forest Ecol. Manag. 148,
63–77.
Rudolph, D.C., Conner, R.N., Carrie, D.K. & Schaefer,
R.R. (1992). Experimental reintroduction of Red-
cockaded Woodpeckers. Auk 109, 914–916.
Evaluating Safe Harbor Program A. M. Trainor et al.
10 Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London
Rudolph, D.C., Conner, R.N. & Schaefer, R.R. (2002).
Red-cockaded Woodpecker foraging behavior in relation
to midstory vegetation. Wilson Bull. 114, 235–242.
Saura, S. & Pascual-Hortal, L. (2007). A new habitat avail-
ability index to integrate connectivity in landscape con-
servation planning: comparison with existing indices and
application to a case study. Landsc. Urban Plan. 83,
91–103.
Saura, S. & Rubio, L. (2010). A common currency for the
different ways in which patches and links can contribute
to habitat availability and connectivity in the landscape.
Ecography 33, 523–537.
Saura, S. & Torne, J. (2009). Conefor Sensinode 2.2: a soft-
ware package for quantifying the importance of habitat
patches for landscape connectivity. Environ. Model.
Softw. 24, 135–139.
Saura, S., Estreguil, C., Mouton, C. & Rodriguez-Freire,
M. (2011a). Network analysis to assess landscape connec-
tivity trends: application to European forests (1990–
2000). Ecol. Indicators 11, 407–416.
Saura, S., Vogt, P., Velázquez, J., Hernando, A. & Tejera,
R. (2011b). Key structural forest connectors can be
identified by combining landscape spatial pattern and
network analyses. For. Ecol. Manage. 262, 150–160.
Silva, E.A. & Clarke, K.C. (2002). Calibration of the
SLEUTH urban growth model for Lisbon and Porto,
Portugal. Comput. Environ. Urban Syst. 26, 525–552.
Trainor, A.M., Walters, J.R., Morris, W.F., Sexton, J.O. &
Moody, A. (in press). Empirical estimation of dispersal
resistance surfaces: a case study with Red-cockaded
Woodpeckers. Landsc. Ecol. doi: 10.1007/s10980-013-
9861-5
U.S. Fish and Wildlife Service. (2004). Safe Harbor Agree-
ments for Private Landowners. Endangered Species
Program, Available from http://www.fws.gov/endangered/
landowners/safe-harbor-agreements.html (accessed 22
April 2012).
U.S. Department of the Interior. (1995). Availability of an
environmental assessment and an application for
an incidental take permit to implement the Red-cockaded
Woodpecker ‘Safe Harbor’ Program in the Sandhills
region of North Carolina. Fed. Reg. 60, 10400–10401.
U.S. Department of the Interior. (1999). Announcement of
final Safe Harbor policy. Fed. Reg. 64, 32717–32726.
U.S. Department of the Interior. (2006). Proposed program-
matic statewide Red-cockaded Woodpecker Safe Harbor
agreement, North Carolina. Fed. Reg. 71, 29350–29351.
U.S. Fish and Wildlife Service. (2003a). Endangered Species
Act of 1973 as amended through the 108th congress.
Washington, DC: U.S. Fish and Wildlife Service.
U.S. Fish and Wildlife Service. (2003b). Recovery plan for
the Red-cockaded Woodpecker (Picoides borealis): second
revision. Atlanta: U.S. Fish and Wildlife Service.
Urban, D.L. & Keitt, T.H. (2001). Landscape connectivity:
a graph-theoretic perspective. Ecology 82, 1205–1218.
Urban, D.L., Minor, E.S., Treml, E.A. & Schick, R.S.
(2009). Graph models of habitat mosaics. Ecol. Lett. 12,
260–273.
Walters, J.R. (1990). Red-cockaded Woodpeckers: a ‘primi-
tive’ cooperative breeder. In Cooperative Breeding in
Birds: 69–101. Stacey, P.B. & Koenig, W.D. (Eds). Cam-
bridge: Cambridge University Press.
Walters, J.R. (1991). Application of ecological principles to
the management of endangered species: the case of the
Red-cockaded Woodpecker. Annu. Rev. Ecol. Evol. Syst.
22, 505–523.
Walters, J.R., Doerr, P.D. & Carter, J.H. (1988). The
cooperative breeding system of the Red-cockaded
Woodpecker. Ethology 78, 275–305.
Walters, J.R., Daniels, S.J., Carter, J.H. & Doerr, P.D.
(2002). Defining quality of Red-cockaded Woodpecker
foraging habitat based on habitat use and fitness.
J. Wildl. Mgmt. 66, 1064–1082.
Walters, J.R., Burst, K., Carter, J.H., III & Anchor, S.
(2009). Red-cockaded Woodpecker demographic moni-
toring in the North Carolina Sandhill: Is Safe Harbor a
biological success? USFWS No. 1448-40181-01-G-218,
1–48.
Wiens, J.A. (2006). Introduction: connectivity research-what
are the issues? Connectivity conservation: 23–28. Crooks,
K.R. & Sanjayan, M. (Eds). New York: Cambridge
University Press.
Wilcove, D.S. (2004). The private side of conservation.
Front. Ecol. Environ. 2, 326–327.
Wilcove, D.S. & Lee, J. (2004). Using economic and regula-
tory incentives to restore endangered species: lessons
learned from three new programs. Conserv. Biol. 18,
639–645.
Wilcove, D.S., Rothstein, D., Dubow, J., Phillips, A. &
Losos, E. (1998). Quantifying threats to imperiled species
in the United States. Bioscience 48, 607–615.
Yang, X. & Lo, C.P. (2003). Modeling urban growth and
landscape change in the Atlanta metropolitan area.
Int. J. Geogr. Inf. Sci. 17, 463–488.
Zhang, D. & Mehmood, S.R. (2002). Safe harbor for the
Red-cockaded Woodpecker: private forest landowners
share their views. J. Forest. 100, 24–29.
Supporting information
Additional Supporting Information may be found in the
online version of this article at the publisher’s web-site:
Figure S1 Safe Harbor Program (SHP) properties showing
the duration of each agreement by land-use activity
Table S1 Summary statistics for numbers of territories on
Safe Harbor Program (SHP) properties grouped by land-use
activity
Appendix S1. LiDAR data details.
Appendix S2. Urban growth model details.
A. M. Trainor et al. Evaluating Safe Harbor Program
Animal Conservation •• (2013) ••–•• © 2013 The Zoological Society of London 11