ArticlePDF Available

Microtargeting for conservation

Authors:
  • Chesapeake Conservancy

Abstract and Figures

Widespread human action and behavior change is needed to achieve many conservation goals. Doing so at the requisite scale and pace will require the efficient delivery of outreach campaigns. Conservation gains will be greatest when efforts are directed toward places of high conservation value (or need) and tailored to critical actors. Recent strategic conservation planning has relied primarily on spatial assessments of biophysical attributes, largely ignoring the human dimensions. Elsewhere, marketers, political campaigns, and others use microtargeting-predictive analytics of big data-to identify people most likely to respond positively to particular messages or interventions. Conservationists have not yet widely capitalized on these techniques. To investigate the effectiveness of microtargeting to improve conservation, we developed a propensity model to predict restoration behavior among 203,645 private landowners in a 5,200,000 ha study area in the Chesapeake Bay Watershed (U.S.A.). To isolate the additional value microtargeting may offer beyond geospatial prioritization, we analyzed a new high-resolution land-cover data set and cadastral data to identify private owners of riparian areas needing restoration. Subsequently, we developed and evaluated a restoration propensity model based on a database of landowners who had conducted restoration in the past and those who had not (n = 4978). Model validation in a parallel database (n = 4989) showed owners with the highest scorers for propensity to conduct restoration (i.e., top decile) were over twice as likely as average landowners to have conducted restoration (135%). These results demonstrate that microtargeting techniques can dramatically increase the efficiency and efficacy of conservation programs, above and beyond the advances offered by biophysical prioritizations alone, as well as facilitate more robust research of many social-ecological systems.
Content may be subject to copyright.
Contributed Paper
Microtargeting for conservation
Alexander L. Metcalf ,1Conor N. Phelan,1Cassandra Pallai,2Michael Norton,2Ben Yuhas,3
James C. Finley,4and Allyson Muth4
1W.A. Franke College of Forestry and Conservation, University of Montana, 440 CHCB, 32 Campus Drive, Missoula, MT, 59812, U.S.A.
2Chesapeake Conservancy, 716 Giddings Avenue, Annapolis, MD, 21403, U.S.A.
3Yuhas Consulting Group, LLC, 121 Hawthorne Road, Baltimore, MD, 21210, U.S.A.
4Ecosystem Science and Management, The Center for Private Forests Pennsylvania State University, Penn State 333 Forest Resources
Building, University Park, PA, 16802, U.S.A.
Abstract: Widespread human action and behavior change is needed to achieve many conservation goals.
Doing so at the requisite scale and pace will require the efficient delivery of outreach campaigns. Conserva-
tion gains will be greatest when efforts are directed toward places of high conservation value (or need) and
tailored to critical actors. Recent strategic conservation planning has relied primarily on spatial assessments
of biophysical attributes, largely ignoring the human dimensions. Elsewhere, marketers, political campaigns,
and others use microtargeting—predictive analytics of big data—to identify people most likely to respond
positively to particular messages or interventions. Conservationists have not yet widely capitalized on these
techniques. To investigate the effectiveness of microtargeting to improve conservation, we developed a propen-
sity model to predict restoration behavior among 203,645 private landowners in a 5,200,000 ha study area
in the Chesapeake Bay Watershed (U.S.A.). To isolate the additional value microtargeting may offer beyond
geospatial prioritization, we analyzed a new high-resolution land-cover data set and cadastral data to identify
private owners of riparian areas needing restoration. Subsequently, we developed and evaluated a restora-
tion propensity model based on a database of landowners who had conducted restoration in the past and
those who had not (n=4978). Model validation in a parallel database (n=4989) showed owners with the
highest scorers for propensity to conduct restoration (i.e., top decile) were over twice as likely as average
landowners to have conducted restoration (135%). These results demonstrate that microtargeting techniques
can dramatically increase the efficiency and efficacy of conservation programs, above and beyond the advances
offered by biophysical prioritizations alone, as well as facilitate more robust research of many social–ecological
systems.
Keywords: conservation marketing, land-use planning, private lands, resource allocation, return on investment,
spatial planning, systematic conservation planning, triage
Micro-focalizaci´
on para la Conservaci´
on
Resumen: Se necesitan acciones humanas diseminadas y un cambio en el comportamiento para lograr muchos
objetivos de conservaci´
on. Que se logre esto a la escala y al ritmo requerido requerir´
adelaentregaeficientedelas
campa˜
nas de participaci´
on. Las ganancias de la conservaci´
on ser´
an mayores cuando los esfuerzos est´
en dirigidos
hacia sitios con un alto valor (o necesidad) de conservaci´
on y est´
en personalizados para los actores m´
as importantes.
La reciente planeaci´
on estrat´
egica de la conservaci´
on ha dependido principalmente de las evaluaciones espaciales
de los atributos biof´
ısicos, ignorando generalmente las dimensiones humanas. En otros ´
ambitos, los mercad´
ologos,
email alex.metcalf@umontana.edu
Article impact statement: Microtargeting boosts conservation impact by finding willing partners and individualizing behavior-change inter-
ventions.
Paper submitted June 27, 2018; revised manuscript accepted March 1, 2019.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution
and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
1141
Conservation Biology, Volume 33, No. 5, 1141–1150
C
2019 The Authors. Conservation Biology published by Wiley Periodicals, Inc. on behalf of Society for Conservation Biology.
DOI: 10.1111/cobi.13315
1142 Microtargeting
las campa˜
nas pol´
ıticas, as´
ı como otros, usan la micro-focalizaci´
on – el an´
alisis predictivo de datos masivos –
para identificar a las personas con mayor probabilidad de responder positivamente a mensajes o intervenciones
particulares. Los conservacionistas todav´
ıa no han capitalizado extensamente estas t´
ecnicas. Desarrollamos un
modelo de tendencia para predecir el comportamiento de restauraci´
on entre 203,645 terratenientes privados en
un ´
area de estudio de 5,200,000 ha en la cuenca de la Bah´
ıa de Chesapeake (E.U.A.) y as´
ı investigar la eficiencia
de la micro-focalizaci´
on en el aumento de la conservaci´
on. Para aislar el valor adicional que puede ofrecer la
micro-focalizaci´
on m´
as all´
a de la priorizaci´
on geoespacial, analizamos un nuevo conjunto de alta resoluci´
on de
datos sobre la cobertura del suelo y datos catastrales para identificar a los terratenientes privados de ´
areas ribere˜
nas
que necesitan restauraci´
on. Despu´
es de esto, desarrollamos y evaluamos el modelo de tendencia a la restauraci´
on
basado en una base de datos de terratenientes que han realizado restauraciones en el pasado y aquellos que no
(n=4,978). La validaci´
on del modelo en una base de datos paralela (n=4,989) mostr´
o que los terratenientes
con los puntajes m´
asaltosparalatendenciaalarestauraci
´
on (es decir, el decil superior) ten´
ıan el doble de
probabilidad de haber realizado acciones de conservaci´
on que el terrateniente promedio (135%). Estos resultados
muestran que las t´
ecnicas de micro-focalizaci´
on pueden incrementar dram´
aticamente la eficiencia y la eficacia de
los programas de conservaci´
on, m´
as all´
a de los avances ofrecidos s´
olo por las priorizaciones biof´
ısicas, as´
ı como
facilitar la investigaci´
on m´
as s´
olida sobre muchos sistemas socio-ecol´
ogicos.
Palabras Clave: asignaci´
on de recursos, mercadotecnia de la conservaci´
on, planeaci´
on espacial, planeaci´
on
sistem´
aticadelaconservaci
´
on, planeaci´
on del uso de suelo, protocolo de intervenci´
on, retorno de la inversi´
on,
tierras privadas
:,,
(),
,,
()

,(Chesapeake Bay Watershed) 5,200,000
, 203,645,
,
,,
(n =4,978) (n =4,989) ,
()(135%)
,,,
:;:
:,,,,,,,
Introduction
Conservation opportunities are distributed unevenly
across landscapes, and spatial prioritization can help max-
imize returns on investment, especially when they ac-
count for social and ecological variability. Yields will be
largest in places of high ecological value where people
are likely to take conservation action (Fig. 1). Existing
efforts to identify areas of high conservation value have
relied primarily on biophysical variables (Fig. 1), an im-
portant first step for guiding investments (Naidoo et al.
2008; Belote et al. 2017). Conservation, however, is a
social endeavor dependent on human decision making
and behavior (Mascia et al. 2003; Schultz 2011). Efforts
to enhance conservation prioritizations through rigorous
integration of social and ecological data have been frus-
trated by data availability and incompatible resolutions.
Biophysical prioritizations are proliferating due to a rev-
olution in remotely sensed data and analysis techniques,
and although they are generally limited to coarse spatial
scales (e.g., 30 m2) by cost or technology (Kwok 2018),
this is changing rapidly as technology advances (e.g., the
land-cover data used in the current study are at 1 m2
resolution). In contrast, estimates of social propensity
for conservation behavior (Fig. 1), if considered at all, are
generally made for entire populations, doing little to pre-
dict conservation action of individuals living in or owning
important places. These social data and analysis limita-
tions are especially challenging when prioritizing land-
scapes with substantial private land ownership where
conservation is a collective product of many landowners’
individual management actions.
Recent advances in consumer data availability and anal-
ysis techniques may provide opportunities to predict in-
dividuals’ propensities toward conservation behaviors,
helping to overcome these challenges and improve the
efficiency and efficacy of conservation programs. For
example, land trusts could identify landowners more
willing to sell or donate conservation easements, re-
ducing recruitment and development costs, or groups
Conservation Biology
Volume 33, No. 5, 2019
Metcalf et al. 1143
Figure 1. Conceptual
conservation-priority space
defined by conservation
value and social propensity
for conservation action
(prime prospects, areas
where conservation value
and social propensity are
both high; premium, areas
of high value where social
propensity is low; trivial,
areas with lower
conservation value despite
likely conservation action;
sinks, areas where
conservation value and
social propensity for action
are both low).
promoting proconservation behavior could identify peo-
ple for whom particular intervention strategies are most
likely to succeed, increasing program effectiveness. How-
ever, most human dimensions data (e.g., qualitative in-
terviews and quantitative surveys) and analyses, where
they exist at all, do not allow characterization of individ-
uals in a population, affording instead only estimates of
population-level trends or parameters. If data were avail-
able for every individual in a population, conservation-
ists could deliver individualized content, incentives, or
programming and likely increase return on conservation
investments (i.e., outcomes per unit invested).
Increasingly, data are in fact available for most in-
dividuals in society, but they do not fit the mold of
traditional conservation behavior models and thus have
largely been overlooked by conservationists. Behavioral
theories are extremely useful for understanding the why
behind proenvironmental behaviors and for designing ef-
fective behavior change interventions. For example, so-
cial norms can often strongly influence behavior (Schultz
et al. 2018); Ajzen (2012) reasoned action approach has
long shown that behaviors are more likely when attitudes
toward a behavior are positive (even more so if the behav-
ior is intended); and recent advancements in behavioral
economics suggest choice architecture and other biases
strongly determine environmental behavior (Sunstein &
Reisch 2014). Powerful interventions can be designed
using these theories, but despite increasing theoretical
sophistication, data are rarely available to identify who
within a population is prone to a particular behavior or
likely to respond a specific intervention. Consider that
even when theory suggests (and survey work or exper-
imentation confirm) landowners with positive attitudes
toward wildlife conservation are more likely to donate a
conservation easement, there remain no attitudinal data
to identify who those individuals are in any given popu-
lation. In contrast, publicly available consumer data are
widely available, and while they rarely include variables
relevant to behavioral models, they afford robust pattern-
detection analysis (i.e., analytics) to identify individuals
belonging to particular population segments, which can
be used to enhance myriad outreach efforts.
Consumer data are widely available and generally in-
clude purchasing history and identifying information, al-
though it is often combined with other streams of data
such as voting behavior, social media activity, or mo-
bile application use to enable robust pattern detection
analyses. Methods for analyzing publicly available con-
sumer data sets have been pioneered in other fields to
predict behavioral propensity for each (or most) individ-
uals in many different populations (Berry & Linoff 2004).
Marketers, charities, and political campaigns have used
such data to microtarget individuals (Fig. 2) and realized
profound efficiencies in their operations. For instance,
with a small group of past known participants, consumer
data can be mined to identify likely future participants
from large and otherwise nebulous populations ([i] in
Fig. 2). Or, if market segments have been identified
through prior analysis, consumer data sets can be mined
to estimate which segment each population member is
likely to belong ([ii] in Fig. 2). Specific applications of
these techniques in other fields include online adver-
tisers who employ “behavioral targeting” to understand
users’ preferences (Jaworska & Sydow 2008), political
campaigns that use “microtargeting” to differentiate and
motivate voters (Murray & Scime 2010), and energy sup-
pliers using “load profiling” to predict household energy
demand (Wang et al. 2015). Although microtargeting, or
big data analytics, is transforming marketing and other
fields (Erevelles et al. 2016), they have received little
Conservation Biology
Volume 33, No. 5, 2019
1144 Microtargeting
Figure 2. Examples of
microtargeting use of
consumer data to (i)
identify likely participants
in conservation programs
from lists of past known
participants and (ii)
identify likely members of
population segments based
on known market segments.
attention in the conservation realm, with exceptions fo-
cusing on consumers of environmentally related prod-
ucts, such as electric vehicles (Eppstein et al. 2011)
and green energy (Tabi et al. 2014). Some efforts have
moved toward this idea, including social marketing to
improve efficiencies in conservation-outreach program-
ing (McKenzie-Mohr et al. 2012) and numerous attempts
to build conservation-related typologies of the public
(Maibach et al. 2009) or landowners (Metcalf et al. 2016).
Still, these marketing-based approaches inform message
development for mass broadcast, contributing little to
customized appeals at the individual level.
Microtargeting may prove useful in many conservation
contexts, including in private land settings where strate-
gic conservation planning based on spatial analysis of
biophysical attributes can be directly coupled to indi-
vidual landowner decisions and behavior. Private lands
are critical to many global conservation efforts (Kamal
et al. 2015), and propensity models predicting landowner
receptivity toward conservation behaviors could help im-
prove return on investments made by outreach agencies
and organizations to promote stewardship. Data enabling
these analyses are increasingly available, yet often un-
tapped for such purposes. For example, spatial cadastral
parcel databases are proliferating in the United States and
can be intersected with spatial assessments of conserva-
tion value to enumerate owners of priority properties
(Ver Planck et al. 2016). Some analyses have used cadas-
tral data to begin generating parcel- or property-level pre-
dictive models, but rely solely on biophysical variables.
More specifically, some have investigated spatial patterns
of ownership and resource parcelization (Kilgore et al.
2013; Zipp et al. 2017), whereas others have ranked prior-
ity parcels based on resource location, cost, risk, or other
variables (Smith et al. 2016), and a few have predicted
land use change based on biophysical variables such as
parcel size, shape, proximity to landscape features, or
resource co-occurrence (Nelson et al. 2008). No studies
we are aware of have capitalized on widely available con-
sumer data to inform individual-level propensity models
of private lands conservation behavior. A few studies
have coupled social and biophysical variables to prior-
itize landscapes, either by summarizing social receptivity
based on qualitative data (Game et al. 2011) or by us-
ing coarse-resolution data over wide extents (Guerrero &
Wilson 2017; Karimi et al. 2017). Some have used agent-
based modeling to represent the complexity of social
and ecological interrelationships (Miyasaka et al. 2017),
and one study by Nielsen et al. (2017) coupled cadas-
tral and forest-cover data to a national registry of
conservation-program participation, the latter being a
data set unique to Denmark and unavailable elsewhere.
Microtargeting techniques adapted for conservation
contexts could enable more widespread consideration
of variation in social dimensions relevant to conser-
vation objectives and enhance strategic conservation
planning as well as implementation. Microtargeting is
not a substitute for other human dimensions research,
but can instead enhance practical applications of re-
search findings. For example, developing effective behav-
ior change interventions will always require theoretical
insight and empirical evaluation, but once effective out-
reach strategies are identified, microtargeting can help
deliver them to the people for whom they are most likely
to be effective. Coupling these approaches to spatial pri-
oritizations can help conservation outreach efforts reach
the right people in the right places for the least cost.
We explored the potential utility of microtargeting
for conservation. The private lands that dominate the
Pennsylvania portion of the Chesapeake Bay Watershed
(U.S.A.) (see Supporting Information) offered an
opportune study area where restoration of riparian areas
is needed to achieve water-quality goals in the Chesa-
peake Bay (Lowrance et al. 1997). More specifically,
Conservation Biology
Volume 33, No. 5, 2019
Metcalf et al. 1145
Pennsylvania (an area nearly double our study area)
seeks to revegetate nearly 55,000 ha of riparian areas to
achieve a 3,600,000 kg reduction in nitrogen pollution
(M. Keefer, personal communication). Based on the mean
size of past restoration projects (1.9 ha), this will require
participation of nearly 29,000 landowners across the
state. We sought to provide an approach that could en-
hance restoration efforts in the watershed; demonstrate
how consumer data-driven, individual-level propensity
models can be coupled to strategic conservation planning
efforts; begin to quantify the value microtargeting adds,
if any, to such geospatial prioritizations; and extend
discussions of how marketing methods may be more
effectively employed to advance conservation objectives.
Methods
Overview
To understand the value offered by microtargeting above
and beyond traditional geospatial prioritizations, we be-
gan by developing high-resolution (1 m2) land-cover and
stream-network data sets. We used these to assess the
condition of riparian buffers (i.e., areas within 11 m of
streams) (Natural Resources Conservation Service 2010).
We overlaid this biophysical inventory with spatially ex-
plicit cadastral data to enumerate the population of ripar-
ian buffer gap owners in our study area (i.e., owners of
riparian areas within 11 m of streams with nonforest and
nonshrubland land cover). We manually searched and
reviewed owner records to remove public ownerships
(because attribute data did not consistently identify all
public lands), thus completing the biophysical portion
of our prioritization. Then, to assess the additional value
offered by microtargeting, we used a database of known
past buffer-program participants and nonparticipants to
build a model that predicts buffer-restoration program
participation (i.e., participation propensity). We evalu-
ated model performance with a similar data set reserved
for validation. Finally, we generated a coupled ranking
score for 203,645 properties within the study area based
on biophysical (i.e., size of buffer gap) and social (i.e.,
landowner propensity score) variables. Field validation of
model performance is underway, including a social mar-
keting campaign (Metcalf et al. 2019) and on-the-ground
recruitment of landowners into buffer restoration pro-
grams. Both are being implemented under experimental
designs where response rates and buffer installations will
be measured as functions of model propensity scores.
High-Resolution Land-Cover Data
Guided by the Chesapeake Bay Program partnership, the
Chesapeake Conservancy led the creation of our land-
cover data set. The University of Vermont Spatial Anal-
ysis Laboratory (UVM) performed the supervised clas-
sification known as object-based image analysis based
on National Agricultural Imagery Program quarter-quad
images from 2013, PA orthophotos from 2003 to 2006,
and light detection and ranging (LiDAR) data from 2006
to 2008. Analysts created image stacks where these data
sets overlapped geographically and then divided them
into homogenous regions based on similarities between
neighboring pixels (Blaschke 2010). Using these image
objects, analysts identified distinguishing characteristics
of regions in a series of if–then rules. For example, water
features typically returned low values of near-infrared ra-
diation, so a rule assigned all image objects with low
near-infrared values to the water bodies class. In this
way, rules were used to classify image objects until a
desired level of accuracy was achieved. Postclassification,
analysts applied manual corrections to improve accuracy
and appearance, and incorporated local planimetric data
sets where available, such as edge of pavement or roads.
Twelve classes of land cover were created: water, wet-
lands, tree canopy, shrubland, low vegetation, barren,
structures, impervious surfaces, impervious roads, tree
canopy over structures, tree canopy over impervious sur-
faces, and tree canopy over impervious roads. The data
set is 90% accurate (Pallai & Wesson 2017).
To locate water bodies, and subsequently riparian ar-
eas, relative to this new, high-resolution land-cover data
set, we used a D8 algorithm on LiDAR-derived digital
elevation models to assign flow direction to each pixel ac-
cording to the direction of steepest descent (O’Callaghan
& Mark 1984). Flow was accumulated across the land-
scape, and pixels that drained more than 24.3 ha were
identified as part of the stream network (see Tarboton
and Ames [2001] for discussion of challenges with area-
based stream initiation assumptions). We estimated chan-
nel widths based on regional curves published by the U.S.
Geological Survey (Krstolic & Chaplin 2007). Widened
channels were merged with the high-resolution land-
cover water class to create an enhanced flow path net-
work. We defined riparian buffers as areas within 11 m
of the enhanced-flow path network with buffers com-
posed of tree canopy or shrubland land-cover classes,
and we defined gaps as all other land-cover types. In
total, the data set covered 25,899,881 ha (see Supporting
Information for full data set).
Microtargeting Analysis
To develop the predictive model of landowner propen-
sity for buffer restoration, we conducted an analysis of
both property and owner attributes with commercially
available data. We began with a list of known buffer-
restoration program participants acquired from the Penn-
sylvania Department of Environmental Protection. This
list contained landowners who had partnered with a
variety of NGOs and governmental agencies to restore
riparian buffer areas on their properties between 2003
Conservation Biology
Volume 33, No. 5, 2019
1146 Microtargeting
and 2015 (initial n=7440). To contrast these known
participants with nonparticipants, we intersected ripar-
ian gaps with cadastral data to enumerate the population
of gap-owning landowners, and extracted a simple ran-
dom sample of landowners with gaps of at least 0.04 ha
(initial n=9400). This was based on the assumption that
landowners who had participated in buffer-restoration
programs would no longer have buffer gaps of this size.
Instances where this assumption was false would sup-
press model performance (rather than overestimate). We
purchased consumer and property data from a reputable
vendor, TargetSmart, for approximately US$2,200 and
matched them to the riparian owner population and past
participant data sets based on combinations of name and
location (e.g., surname and ZIP code). These data con-
tained over 800 variables including demographics (e.g.,
age and gender), consumer behavior, property-level at-
tributes (e.g., size, price, and date of last purchase for
both land and structures), and many other variables about
individuals’ voting history, area of residence, and com-
mercial interests (see Supporting Information for data
dictionary). Match rates were 30% for known participants
(final n=2259) and 66% for assumed nonparticipants (fi-
nal n=7708). We merged these lists (summed n=9967)
before randomly dividing them in half to create training
(n=4978) and validation (n=4989) data sets.
We ran a developmental stepwise logistic regression
with the training data set to contrast known participants
with nonparticipants and then used predictive variables
from the output to calculate a propensity score for those
landowners in the training, validation, and full riparian-
owner-population data sets for whom consumer data
were available. Our goal for this developmental analysis
was to maximize variance explained by the model (with-
out overfitting), rather than to isolate predictive or causal
variables, such as those suggested by theory (final model
in Supporting Information). We evaluated model perfor-
mance with a second logistic regression model within the
validation data set to test the prediction that propensity
score could differentiate between participants and non-
participants. We grouped propensity scores into deciles
and tallied actual participation rates within each decile
for both training and validation data sets to facilitate
interpretation and communication. We used the prod-
uct of property riparian buffer gap area and propensity
score decile to calculate a coupled ranking score for each
parcel (Fig. 3). All research methods were reviewed and
approved by the Institutional Review Boards at Penn State
University and the University of Montana.
Results
The study area covered 5,529,100 ha and included over
82,000 km of stream network. All riparian buffer areas
totaled 136,028 ha and buffer gaps constituted 38,882 ha
(22.2%). Riparian buffer gaps were owned by a total of
201,647 landowners, with an inverse J-shaped distribu-
tion based on hectares of gaps on the property. A plu-
rality of buffer gap hectares were found on properties
with large gaps; 19% of all buffer gap hectares were on
properties with at least 2 ha of buffer gaps, whereas prop-
erties with small gaps (i.e., <0.04 ha) were numerous
(n=106,977) yet accounted for only 1.4% of all buffer
gap hectares.
Consumer data were successfully matched to nearly
two-thirds of nonpublic, riparian gap owners (60%).
These owners controlled 46% of all buffer gap hectares.
A small percentage of owners was classified as public
(2.4%) controlling 7.4% of buffer gap hectares; these
are minimum estimates because some public owners
were likely missed in the manual classification process.
Just over 9% of nonpublic owners were unscorable be-
cause owner names were businesses, limited liability
corporations, trusts, partnerships, associations, or other
types of owners that were not compatible with con-
sumer databases. These unscorable owners controlled
18% of buffer gap hectares. About 30% of nonpublic
riparian owners were unscored for some other reason
(e.g., names misaligned with consumer data due to mis-
spellings or errors, owners unrepresented in consumer
data); these unscored owners controlled 29% of buffer
gap hectares.
The developmental stepwise logistic regression suc-
cessfully differentiated between known buffer-program
participants and nonparticipants in the training data set,
correctly identifying 79.1% of participants with a cut
value of 0.500 (χ2=628.602; df =11; p<0.001;
Cox and Snell R2=0.119; Nagelkerke R2=0.184). The
model performed similarly well in the validation data
set, correctly identifying 78.6% of participants (B=
5.913, Wald =476.563; df =1; p<0.001; Cox and Snell
R2=0.104; Nagelkerke R2=0.157).
The overall participation rate (percentage of known
participants) in the validation data set was 23%. Participa-
tion rates were higher within each subsequent propensity
score decile, with a participation rate of 3% in the bottom
decile and 54% in the top decile (Fig. 4). Thus, the partic-
ipation rate in the top decile was 2.3 times higher than
the average participation rate in the overall validation
data set, and 18 times higher than the participation rate
in the bottom decile.
Discussion
Conservation requires behavior change (Cinner 2018),
yet resources to inspire such change are often limited.
Microtargeting can help identify individuals with
higher propensity toward particular behaviors, or
those for whom particular interventions are more
likely to succeed. Applications of these techniques to
conservation challenges may help increase the efficiency
and effectiveness of conservation programs. Toward
Conservation Biology
Volume 33, No. 5, 2019
Metcalf et al. 1147
Figure 3. Properties with
riparian gaps (i.e.,
nonforest and nonshrub
areas within 11 m of
streams) jointly ranked by
gap size and owner’s score
for propensity to participate
in conservation (jittered
deciles). Properties with
gaps >3.50 ha are clustered
at 3.50 ha in the figure for
clarity.
Figure 4. Percentage of known
buffer-restoration program
participants in the training and
validation data sets by score for
propensity to participate in
restoration.
efficiency, microtargeting may help conservationists
identify individuals most likely to participate in specific
programs, or engage in specific behaviors (Fig. 2).
To increase efficacy, microtargeting may help deliver
customized interventions designed for different market
segments of a target population (Fig. 2). Unlike most
approaches to conservation behavior change research
or interventions, microtargeting employs consumer data
that is widely available for most members of population at
relatively low cost, enabling individual-level predictions
rather than population-wide inferences. We found micro-
targeting successfully differentiated restoration-program
participants from nonparticipants, suggesting its applica-
tion to program recruitment could dramatically improve
conservation returns on investment. There may be
myriad opportunities to apply microtargeting techniques
to improve conservation outcomes while reducing cost,
although much more research is needed on a variety of
fronts.
One area where microtargeting may be particularly
helpful is strategic conservation planning in landscapes
dominated by private land. In these contexts, the man-
agement of valued resources is directly tied to individ-
ual landowner decisions. Previous studies have sought
coupled social–ecological landscape prioritizations, rec-
ognizing that the likelihood of conservation success is a
function of biophysical as well as human dimensions and
dynamics (Game et al. 2011; Guerrero et al. 2013; Karimi
et al. 2017; Nielsen et al. 2017). Adapting microtargeting
to this particular challenge presents a novel opportunity
to prioritize conservation investments by spatially cou-
pling fine-grain social and ecological data across large
extents. This approach is unique in that it enables a
social–ecological landscape prioritization for conserva-
tion investments by combining a biophysical resource as-
sessment, an enumeration of private resource ownership,
and analysis that identifies landowners with high propen-
sity for restoration-program participation. Our results
Conservation Biology
Volume 33, No. 5, 2019
1148 Microtargeting
showed that even after a large investment in biophysi-
cal data development and spatial analysis, microtargeting
still offered dramatic improvements to the landscape pri-
oritization through fine-scale consideration of variation
in human likelihood to take conservation action (i.e.,
Fig. 1). In other words, by using these techniques,
riparian-restoration outreach programs could reduce
their outreach budgets by over 50% (i.e., spend less to re-
cruit similar numbers of participants), or more than dou-
ble the impact of existing programs (i.e., spend the same
to recruit more participants). With these savings, budgets
could be more efficiently allocated toward achieving con-
servation outcomes (e.g., revegetation costs), rather than
recruiting program participants. Future research should
strive for a robust accounting of microtargeting costs,
savings, and benefits in different contexts.
Beyond efficiency and efficacy enhancement, micro-
targeting also enables robust evaluations of intervention
success. Conservation social science is in need of more
experimental design to objectively evaluate and quantify
impact (Ferraro & Pattanayak 2006; Baylis et al. 2016) and
learn from intervention successes and failures (Ver´
ıssimo
et al. 2018). Other social marketing efforts prove difficult
to evaluate. For example, how does one know whether
mass communication campaigns work or how to deter-
mine for whom they did not and evolve subsequent ef-
forts? Microtargeting operates at the individual level, so
interventions can be strategically deployed under exper-
imental designs with large sample sizes, affording robust
evaluation of successes and failures. Further, research of
coupled human and natural systems may benefit from the
spatially explicit prediction of individual actor behavior,
and resource consequences (e.g., in response to social-
ecological change), afforded by these methods.
More generally, marketing tools offer promise for
achieving conservation goals, but have been largely ig-
nored, eschewed, or their efficacy questioned (Ver´
ıssimo
& McKinley 2017). Exceptions do exist and are growing,
such as social marketing for environmental protection
(McKenzie-Mohr et al. 2012). Although these broader
applications of marketing are extremely valuable, our
methodology allows a finer scale assessment of target
audiences providing conservation guidance at the individ-
ual level without limiting geographic scope. Microtarget-
ing allows conservation interventions to be customized
for each person in a population, thus increasing conser-
vation return on investment. In our case study, micro-
targeted landowners were over twice as likely to have
participated in the restoration program. Alternative non-
parametric analysis techniques for likelihood estimation
(e.g., neural networks, discriminant analysis, machine
learning; Vellido 1999) may offer even higher model ac-
curacies and programmatic returns, but further research
is needed.
Still, legitimate questions remain about the value of
this approach compared with other methods, whether
results here will hold when more rigorously evaluated
(i.e., based on direct measures of behavior change), and
the ethical bounds of utilizing these types of publically
available data. There may be other methods for achieving
similar outcomes, but more research is needed. For ex-
ample, propensity models based solely on biophysical or
economic data (e.g., land or real estate value) may yield
similar results to consumer data analyses in some settings.
If and where these contexts exist, they should be evalu-
ated. In other instances, the cost and expertise required
to conduct microtargeting may not be worth the savings
produced, although consumer data acquisition is rela-
tively low cost and the statistical techniques for analysis
are generally accessible (e.g., logistic regression). Despite
low costs, the real outcomes generated by these tech-
niques will require field testing (underway in our study
area) that take time, resources, and commitment to ac-
complish. In an evaluation of an initial riparian restoration
outreach campaign, we recently demonstrated our micro-
targeting algorithm could increase engagement among
landowners by over half (66%), however continued test-
ing is needed as owners progress from initial engage-
ment to actual buffer installation (Metcalf et al. 2019).
Future research should explore the extent and impact
of nonindividual owners on the utility of microtargeting
application and ways to circumvent this limitation.
In addition to pragmatic questions about effectiveness,
conservationists must not ignore the ethical questions
raised by microtargeting techniques. For example, big
data are used by industry, political campaigns, and others
to understand, predict, and manipulate human behavior,
raising concerns about privacy, the appropriate methods
for inspiring behavior change, and what behaviors ought
to be promoted or discouraged. Should consumer data
be publically available in the first place? What types of
behavior change interventions are appropriate, if any?
What are the costs and benefits of methodological trans-
parency? What segments of the population are excluded
from these analyses and thus potentially excluded from
conservation outreach? The ethical standards for con-
servationists are unique from marketers, and best man-
agement practices for microtargeting should be articu-
lated and closely followed so as to not undermine trust
of the stakeholders with whom conservation requires
partnership. Behavioral economists view their interven-
tions as “libertarian paternalism,” arguing the beneficial
outcomes, and retention of choice, justify their actions
(Sunstein & Thaler 2003). Similarly, conservationists may
view their efforts as ‘facilitated altruism’ because they
encourage (not require) behaviors that produce public
goods or protect common resources, but more debate
is needed before ethical clarity is achieved (Ver´
ıssimo &
McKinley 2016).
Our results demonstrated how one conservation pro-
gram benefited from a coupled social and biophysi-
cal prioritization informed by microtargeting. Similar
Conservation Biology
Volume 33, No. 5, 2019
Metcalf et al. 1149
opportunities may exist in areas where digital and spa-
tially explicit parcel ownership data are available to
couple ecological to social data based on joint location
information. In other contexts, microtargeting could help
predict conservation-related behaviors of other actors
(e.g., recreationists, customers, and voters) or identify
people most likely to respond to interventions (Fig. 2).
More work is needed to understand what efficiencies can
be achieved by using these techniques and whether those
gains are worth the investments required to conduct con-
sumer data analysis. However these methods advance,
the prospects for microtargeting to improve conservation
programs are profound.
Acknowledgments
This work was generously supported by the RK Mellon
Foundation, the USDA Forest Service, the National Park
Service, the Chesapeake Bay Program Partnerships, and
the Pennsylvania Department of Conservation and Natu-
ral Resources.
Supporting Information
Study area map (Appendix S1), 1 m2resolution land-
cover data set (Appendix S2), consumer data dictionary
(Appendix S3), and output of developmental regression
model (Appendix S4) are available online. The authors
are solely responsible for the content and functionality
of these materials. Queries (other than absence of the
material) should be directed to the corresponding author.
Literature Cited
Ajzen I. 2012. Martin Fishbein’s legacy: the reasoned action approach.
The Annals of the American Academy of Political and Social Science
640:11–27.
Baylis K, Honey-Ros´
es J, B¨
orner J, Corbera E, Ezzine-de-Blas D, Ferraro
PJ, Lapeyre R, Persson UM, Pfaff A, Wunder S. 2016. Mainstream-
ing impact evaluation in nature conservation. Conservation Letters
9:58–64.
Belote RT, Dietz MS, Jenkins CN, McKinley PS, Irwin GH, Fullman TJ,
Leppi JC, Aplet GH. 2017. Wild, connected, and diverse: building
a more resilient system of protected areas. Ecological Applications
27:1050–1056.
Berry MJ, Linoff GS. 2004. Data mining techniques: for marketing, sales,
and customer relationship management. John Wiley & Sons, Inc.,
Hoboken, New Jersey.
Blaschke T. 2010. Object based image analysis for remote sensing. ISPRS
Journal of Photogrammetry and Remote Sensing 65:2–16.
Cinner J. 2018. How behavioral science can help conservation. Science
362:889–890.
Eppstein MJ, Grover DK, Marshall JS, Rizzo DM. 2011. An agent-based
model to study market penetration of plug-in hybrid electric vehi-
cles. Energy Policy 39:3789–3802.
Erevelles S, Fukawa N, Swayne L. 2016. Big data consumer analytics
and the transformation of marketing. Journal of Business Research
69:897–904.
Ferraro PJ, Pattanayak SK. 2006. Money for nothing? A call for empirical
evaluation of biodiversity conservation investments. PLoS Biology 4
(e105) https://doi.org/0040105.
Game ET, Lipsett-Moore G, Hamilton R, Peterson N, Kereseka J, Atu W,
Watts M, Possingham H. 2011. Informed opportunism for conserva-
tion planning in the Solomon Islands. Conservation Letters 4:38–46.
Guerrero AM, McAllister RRJ, Corcoran J, Wilson KA. 2013. Scale mis-
matches, conservation planning, and the value of social-network
analyses. Conservation Biology 27:35–44.
Guerrero AM, Wilson KA. 2017. Using a social–ecological framework
to inform the implementation of conservation plans. Conservation
Biology 31:290–301.
Jaworska J, Sydow M. 2008. Behavioural targeting in on-line advertising:
an empirical study. Pages 62–76 in International conference on web
information systems engineering. Volume 5175. Springer, Berlin,
Germany.
Kamal S, Grodzi´
nska-Jurczak M, Brown G. 2015. Conservation on pri-
vate land: a review of global strategies with a proposed classifica-
tion system. Journal of Environmental Planning and Management
58:576–597.
Karimi A, Tulloch AIT, Brown G, Hockings M. 2017. Understanding
the effects of different social data on selecting priority conservation
areas. Conservation Biology 31:1439–1449.
Kilgore MA, Snyder SA, Block-Torgerson K, Taff SJ. 2013. Challenges
in characterizing a parcelized forest landscape: why metric, scale,
threshold, and definitions matter. Landscape and Urban Planning
110:36–47.
Krstolic JL, Chaplin JJ. 2007. Bankfull regional curves for streams in
the non-urban, non-tidal coastal plain physiographic province, Vir-
ginia and Maryland. Scientific investigations report 2007–5162. U.S.
Geological Survey, Washington, D.C.
Kwok R. 2018. Ecology’s remote-sensing revolution. Nature 556:137–
138.
Lowrance R, et al. 1997. Water quality functions of riparian forest
buffers in Chesapeake Bay Watersheds. Environmental Management
21:687–712.
Maibach E, Roser-Renouf C, Leiserowitz, A. 2009. Global warming’s six
Americas 2009. George Mason University, Fairfax, Virginia.
Mascia MB, Brosius JP, Dobson TA, Forbes BC, Horowitz L, McKean MA,
Turner NJ. 2003. Conservation and the social sciences. Conservation
Biology 17:649–650.
McKenzie-Mohr D, Lee NR, Schultz PW, Kotler PA. 2012. Social market-
ing to protect the environment: what works. Sage, Thousand Oaks,
California.
Metcalf AL, Angle JW, Phelan CN, Muth BA, Finley JC. 2019. More “bank”
for the buck: microtargeting and normative appeals to increase so-
cial marketing efficiency. Social Marketing Quarterly 25:26–39.
Metcalf AL, Gruver JB, Finley JC, Luloff AE. 2016. Segmentation to
focus outreach: behavioral intentions of private forest landowners
in Pennsylvania. Journal of Forestry 114:466–473.
Miyasaka T, Bao Q, Toshiya L. 2017. Agent-based modeling of complex
social–ecological feedback loops to assess multi-dimensional trade-
offs in dryland ecosystem services. Landscape Ecology 32:707–727.
Murray GR, Scime A. 2010. Microtargeting and electorate segmenta-
tion: data mining the American National Election Studies. Journal of
Political Marketing 9:143–166.
Naidoo R, Balmford A, Costanza R, Fisher B, Green RE, Lehner B, Mal-
colm TR, Ricketts TH. 2008. Global mapping of ecosystem services
and conservation priorities. Proceedings of the National Academy
of Sciences 105:9495–9500.
Natural Resources Conservation Service. 2010. Conservation practice
standard: riparian forest buffer 391-1. U.S. Department of Agricul-
ture, Richmond, Virginia.
Conservation Biology
Volume 33, No. 5, 2019
1150 Microtargeting
Nelson E, Polasky S, Lewis DJ, Plantinga AJ, Lonsdorf E, White D, Bael D,
Lawler JJ. 2008. Efficiency of incentives to jointly increase carbon
sequestration and species conservation on a landscape. Proceedings
of the National Academy of Sciences 105:9471–9476.
Nielsen ASE, Strange N, Bruun HH, Jacobsen JB. 2017. Effects of pref-
erence heterogeneity among landowners on spatial conservation
prioritization. Conservation Biology 31:675–685.
O’Callaghan JF, Mark DM. 1984. The extraction of drainage networks
from digital elevation data. Computer Vision, Graphics, and Image
Processing 27:323–344.
Pallai C, Wesson K. 2017. Chesapeake Bay Program Partnership
high-resolution land cover classification accuracy assessment
methodology. Chesapeake Conservancy, Annapolis, Maryland.
Schultz PW. 2011. Conservation means behavior. Conservation Biology
25:1080–1083.
Schultz PW, Nolan JM, Cialdini RB, Goldstein NJ, Griskevicius V. 2018.
The constructive, destructive, and reconstructive power of social
norms. Psychological Science 18:429–433.
Smith JT, Evans JS, Martin BH, Baruch-Mordo S, Kiesecker JM, Naugle
DE. 2016. Reducing cultivation risk for at-risk species: predicting
outcomes of conservation easements for sage-grouse. Biological
Conservation 201:10–19.
Sunstein CR, Reisch LA. 2014. Automatically green: Behavioral eco-
nomics and environmental protection. Harvard Environmental Law
Review 38:127–158.
Sunstein CR, Thaler RH. 2003. Libertarian paternalism is not an oxy-
moron. The University of Chicago Law Review 70:1159–1202.
Tabi A, Hille SL, W¨
ustenhagen R. 2014. What makes people seal the
green power deal?—Customer segmentation based on choice ex-
periment in Germany. Ecological Economics 107:206–215.
Tarboton DG, Ames DP. 2001. Advances in the mapping of flow net-
works from digital elevation data. The World Water and Environ-
mental Resources Congress 111:166–175.
Vellido A. 1999. Neural networks in business: a survey of applications
(1992–1998). Expert Systems with Applications 17:51–70.
Ver Planck NR, Metcalf AL, Finley AO, Finley JC. 2016. Evaluation of the
USDA Forest Service National Woodland Owner Survey estimators
for private forest area and landowners: a case study of Montana.
Forest Science 62:525–534.
Ver´
ıssimo D, McKinley E. 2016. Introducing conservation marketing:
Why should the devil have all the best tunes? Oryx 50:14.
Ver´
ıssimo D, McKinley E. 2017. Understanding conservation marketing
and focusing on the best available evidence: a reply to Hobson. Oryx
51:22.
Ver´
ıssimo D, Schmid C, Kimario FF, Eves HE. 2018. Measuring the im-
pact of an entertainment-education intervention to reduce demand
for bushmeat. Animal Conservation 21:324–331.
Wang Y, Chen Q, Kang C, Zhang M, Wang K, Zhao Y. 2015. Load
profiling and its application to demand response: a review. Tsinghua
Science and Technology 20:117–129.
Zipp KY, Lewis DJ, Provencher B. 2017. Does the conservation of land
reduce development? An econometric-based landscape simulation
with land market feedbacks. Journal of Environmental Economics
and Management 81:19–37.
Conservation Biology
Volume 33, No. 5, 2019
... Recent advances in big data and artificial intelligence (AI) should enable the global assessment and prediction of bird population trends. Big data and machine-learning technologies have been used successfully in recent years in biodiversity discovery and conservation projects (Fang et al. 2019, Metcalf et al. 2019. Databases such as the International Union for Conservation of Nature (IUCN) Red List (IUCN 2019) and the Encyclopedia of Life (Smithsonian National Museum of Natural History 2019) have accumulated substantial detailed information for an unprecedented number of species, providing an excellent platform for researchers to conduct meta-analyses on a global scale. ...
Article
Birds are crucial for the functioning of Earth’s ecosystems but bird population declines have been documented worldwide in recent decades. A global assessment of potential causes of population declines is needed. Our goal here was to combine the power of big data and machine learning to identify predictors correlated with bird population declines and to predict population declines for species with unknown population trends on the IUCN Red List. From existing online databases, we gathered detailed species‐level data for 10 964 extant bird species around the world, featuring life history, ecology, distribution, taxonomy and categorical population trend information (i.e. decreasing or not decreasing). For the 10 163 species with known population trends, we split the data into a 75% training set to tune and train a machine‐learning model (Light Gradient Boosting Machine – ‘LightGBM’) and a 25% test set to evaluate the trained model. Our model predicted (i) bird population declines with an ROC AUC score of 0.828, F1 score of 0.748 and average accuracy of 0.747, and (ii) that 47% (n = 801) of bird species with currently unknown population trends are declining. Correlation analyses suggested that, globally, the top predictor associated with bird population declines was a severely fragmented population, with non‐migratory birds in South American and Southeast Asian tropical and subtropical forests being particularly vulnerable. Despite the lack of long‐term quantitative population trend data for all species worldwide, our study presents big data and machine learning as a useful tool for informing conservation priorities, lending insight, albeit imperfect, into bird population declines on the global scale for the first time.
... This trait might not be limited to environmental and/or energy issues, but applicable to general lifestyle; thus, it might be possible to understand which segments mostly identify with this attitude by utilizing data from other fields. While this will require efforts to guarantee privacy data protection, acquiring knowledge of this and other psychological traits may enable energy interventions to better utilize tools such as micro-targeting [109] (i.e. understanding and satisfying individual preferences). ...
Article
Full-text available
To guide effective energy policy-making towards a fundamental understanding of the mechanisms relevant to behavioral change, it is important not only to investigate whether energy interventions succeed or not, but also to explore the underlying reasons that shape each result. However, certain limitations are hindering a global consensus on the effectiveness of two popular types of energy interventions: the ones based on social influence (peer pressure) and the ones based on economic instruments (rewards and penalties). The aim of this paper is to provide a new perspective on the exploration of the factors that affect the effectiveness of such interventions. Based on a review of studies published during the last two decades, an agenda of six critical research questions is thus set up to identify new priority areas of research. The relevance of this agenda is illustrated via a survey that explores the potential of peer pressure and economic interventions designed to influence residential space cooling energy savings in an urban setting. The survey results provide evidence that such a potential can be affected by the type of targeted behavior (efficiency or conservation), by householder characteristics (openness to change and environmental awareness), and by the existence of past influence events. Interestingly, peer pressure is regarded as highly influential independently of the channel through which it is communicated, i.e. offline or online. These observations can assist public policy in countries with a growing emphasis on changing people's energy behavior to redefine the targeting scope of interventions, thus strengthening their potential.
... Rather than relying on intuition, campaigns grounded in psychological theories will provide a stronger foundation for supporting evidence-based decisions. Although attitudes (Nilsson et al. 2019) and norms (Kinzig et al. 2013;Metcalf et al. 2019) have received widespread attention, limited research has developed and tested values-based messaging (Lee & Jan 2015;Cetas & Yasué 2017), despite values being fundamentally important for providing guidance on how to promote CB in everyday life (Stern et al. 1999). Parks and protected areas are ideal settings to motivate CB through messaging because public lands facilitate intimate interactions between people and nature. ...
Article
Values are the fundamental reasons why people engage in conservation behaviors. Recent research has called for a more refined approach to studying values in a way that accounts for the concept of eudaimonia. However, the empirical properties for a eudaimonic value scale have not been tested given that previous investigations have remained at the theoretical level. Drawing from an on-site survey of visitors to Denali National Park and Preserve, Alaska, we used a latent profile analysis used a latent profile analysis to better understand the expression of multiple values of nature. Specifically, we segmented respondents by their value orientations with a particular focus on evaluating eudaimonic and hedonic values, alongside the established dimensions of altruistic, biospheric, and egoistic values. We identified 4 distinct subgroups defined by value orientations and validated these subgroups based on measures of conservation behavior and sociodemographic characteristics (e.g., age). These results indicated campaign messaging should harness a combination of eudaimonic, biospheric, and altruistic values to propel individual behavior. We also observed that hedonic and egoistic values defined how people related to nature and played a role, albeit less pronounced, in motivating them to take action. Our study is one of the first efforts to operationalize eudaimonia in a conservation context; thus, we have opened a new avenue for protected-area managers to align their strategies with the underlying values of stakeholders.
... The audience targeted in this study was broad. Ideally, future work would consider more specific audience segmentation for each of the behaviors selected and their potential interventions (Metcalf et al., 2019). Effective audience segmentation and targeting involves different interventions or messaging strategies for each audience segment (Kidd et al., 2019), as individuals have different motivations and barriers to participation (Asah & Blahna, 2012). ...
Article
Full-text available
The conservation profession is increasingly seeking effective ways to reduce societal impact on biodiversity, including through targeted behavior change interventions. Multiple conservation behavior change programs exist, but there is also great uncertainty regarding which behaviors are most strategic to target. Behavioral prioritization is a tool that has been used effectively to support behavior change decision-making in other environmental disciplines and more recently for a small subset of biodiversity behavior change challenges. Here, we use behavioral prioritization to identify individual behaviors that could be modified to achieve biodiversity benefits in the state of Victoria, Australia. We use an adapted nominal group technique method to identify potential biodiversity behaviors and, for each behavior, estimate the corresponding plasticity (or capacity for change) and positive impact on biodiversity outcomes. We elicited 27 behaviors that individuals could undertake to benefit or reduce their negative impact on biodiversity. This list was then used to prioritize 10 behaviors as determined by their likely effect(s) on biodiversity, plasticity, and current prevalence in Victoria. We take a first step in outlining a list of behaviors that can direct Victorian decision-makers toward increasing positive and reducing negative impacts of society on biodiversity, guide motivated individuals to reduce their own biodiversity footprint, and more broadly, develop a behavior change research agenda for behaviors most likely to benefit biodiversity. K E Y W O R D S behavior adoption, behavior change, behavior prioritization, biodiversity conservation, community-based social marketing, conservation psychology, nominal group technique, Victoria
... More recently, researchers have begun using a hyper-segmentation method called microtargeting, which employs "markets-of-one" or "one-to-one marketing" principles, to improve the efficacy of conservation initiatives. A study by Metcalf et al. (2019) applied big data analytics (i.e., statistical and algorithmic processes that analyze large and varied data sets to reveal patterns and correlations) to identify and predict individuals who were the most likely to respond positively to certain messages or interventions aimed at recruiting conservation easement participants. ...
Article
Full-text available
Waterfowl hunter numbers and waterfowl populations were closely correlated until the past 2 decades when hunter numbers declined despite near‐record breeding population estimates for ducks in North America. As a result, efforts to recruit, retain, and reactivate (R3) waterfowl hunters have been promoted by the North American Waterfowl Management Plan community because hunters are a key source of funding for wetland‐wildlife habitat conservation and management. Increasing access and opportunity for hunting appears to be the primary R3 strategy in North America. We suggest that hunt quality is an equal, if not more important, facet of waterfowl hunter R3 that is substantially overlooked and undervalued by current R3 initiatives. We contend that providing abundant access and opportunity to hunt waterfowl alone, especially if it jeopardizes hunting quality, is inadequate. We urge the R3 community to integrate principles and methods used within the human dimensions field to focus on the relationships between quality, motivation, and satisfaction across various audiences and market segments. Such an approach will help R3 initiatives establish an empirical evidence base to develop strategies specifically focused on identifying key hunt quality or opportunity factors. © 2020 The Wildlife Society. We comment on how waterfowl hunting opportunity and quality may influence waterfowl hunter recruitment, retention, and reactivation (R3). We suggest that hunt quality is an equal, if not more important facet of waterfowl hunter R3 as opportunity or access, but it is substantially overlooked and undervalued by current R3 initiatives.
... They identified numerous social factors associated with species distributions operating at scales ranging from the individual property to the community, such as family income, housing age, population density, and public versus private ownership (Decker et al. 2012;Hope et al. 2003). Their findings can inform conservation planning by identifying the types of landowners who or communities that are not achieving desired conservation outcomes; managers may then target outreach toward these specific groups (e.g., Metcalf et al. 2019). ...
Article
Full-text available
Conservation across human-dominated landscapes requires an understanding of the social and ecological factors driving outcomes. Studies that link conservation outcomes to social and ecological factors have examined temporally static patterns. However, there may be different social and ecological processes driving increases and decreases in conservation outcomes that can only be revealed through temporal analyses. Through a case study of the invasion of Falcataria moluccana in Hawaii, we examined the association of social factors with increases and decreases in invader distributions over time and space. Over 7 years, rates of invader decrease varied substantially (66-100%) relative to social factors, such as building value, whether land was privately or publically owned, and primary residence by a homeowner, whereas rates of increase varied only slightly (<0.1-3.6%) relative to such factors. These findings suggest that links between social factors and invasion in the study system may be driven more by landowners controlling existing invasive species, rather than by landowners preventing the spread of invasive species. We suggest that spatially explicit, time-dependent analyses provide a more nuanced understanding of the way social factors influence conservation outcomes. Such an understanding can help managers develop outreach programs and policies targeted at different types of landowners in human-dominated landscapes. © 2019 Society for Conservation Biology.
... They identified numerous social factors associated with species distributions operating at scales ranging from the individual property to the community, such as family income, housing age, population density, and public versus private ownership (Decker et al. 2012;Hope et al. 2003). Their findings can inform conservation planning by identifying the types of landowners who or communities that are not achieving desired conservation outcomes; managers may then target outreach toward these specific groups (e.g., Metcalf et al. 2019). ...
Preprint
Spatially explicit, time-dependent analysis reveals whether links between social factors and conservation outcomes are due to preservation or restoration. In Press at Conservation Biology.
Article
Declining participation in hunting, especially among young adult hunters, affects the ability of state and federal agencies to achieve goals for wildlife management and decreases revenue for conservation. For wildlife agencies hoping to engage diverse audiences in hunter recruitment, retention, and reactivation (R3) efforts, university settings provide unique advantages: they contain millions of young adults who are developmentally primed to explore new activities, and they cultivate a social atmosphere where new identities can flourish. From 2018 to 2020, we surveyed 17,203 undergraduate students at public universities across 22 states in the United States to explore R3 potential on college campuses and assess key demographic, social, and cognitive correlates of past and intended future hunting behavior. After weighting to account for demographic differences between our sample and the larger student population, we found 29% of students across all states had hunted in the past. Students with previous hunting experience were likely to be white, male, from rural areas or hunting families, and pursuing degrees related to natural resources. When we grouped students into 1 of 4 categories with respect to hunting (i.e., non‐hunters [50%], potential hunters [22%], active hunters [26%], and lapsed hunters [3%]), comparisons revealed differences based on demographic attributes, beliefs, attitudes, and behaviors. Compared to active hunters, potential hunters were more likely to be females or racial and ethnic minorities, and less likely to experience social support for hunting. Potential hunters valued game meat and altruistic reasons for hunting, but they faced unique constraints due to lack of hunting knowledge and skills. Findings provide insights for marketing and programming designed to achieve R3 objectives with a focus on university students.
Article
Despite the vital role of private lands as habitat for imperiled species and as important components of functioning protected area networks, incorporation of private lands into national and regional conservation planning has been challenging. Identifying locations where private landowners are likely to participate in conservation initiatives can help avoid costly conflict and clarify trade‐offs between ecological benefits and socio‐political costs. Empirical, spatially explicit assessment of the factors associated with conservation on private land is an emerging tool for identifying future conservation opportunities. However, most data on private land conservation are voluntarily reported and incomplete which complicates these assessments. We use a novel application of occupancy models to analyze the occurrence of conservation easements on private land. We illustrate the utility of the occupancy framework for modeling conservation on private land using a simulation study with a case study of easement data in Idaho and Montana (United States). We compared multiple formulations of occupancy models to a logistic regression model to predict the locations of conservation easements using a Spatially Explicit Social Ecological System framework. We found that occupancy models that explicitly account for variation in reporting produced substantially less biased estimates of predictors than logistic regression. Our results demonstrate that occupancy models produced substantially less biased estimates of the predictors of conservation than logistic regression. Results from our case study suggest that occupancy models also resulted in qualitatively different inferences regarding the effects of predictors on conservation easement occurrence than logistic regression. These results highlight the importance of integrating variable and incomplete reporting of participation into empirical analysis of conservation initiatives through an occupancy approach. Failure to do so can lead to emphasizing the wrong social, institutional, and environmental factors that enable conservation and underestimating conservation opportunities in landscapes where social norms or institutional constraints inhibit reporting. Article impact statement: Occupancy models improve identification of private land conservation opportunities even with incompletely reported data. This article is protected by copyright. All rights reserved
Article
Full-text available
The future of conservation and human-wildlife relationships in the American West is at a defining moment. The region consists of a mosaic of land-cover types, with large amounts of public land under varying degrees of protection, use, and ownership. This public land provides the foundation for high levels of connectivity and habitat for healthy populations of wildlife, including those with large resource requirements such as large and wide-ranging mammals (Barnes et al 2016). However, space for wildlife is under threat in the West. Energy development projects, urban and ex-urban sprawl, increasing road traffic and density, and amenity-driven human migration are dramatically changing the ecological landscape (Leu et al 2008). The social landscape is rapidly changing as well, with new residents bringing different worldviews, economic activities, and expectations regarding wildlife and their habitats (Teel and Manfredo 2010). Because maintaining and establishing landscape connectivity for wildlife in part depends on facilitating their movement across privately-owned lands that connect protected areas, balancing disparate human priorities with wildlife conservation across large landscapes in the American West requires novel approaches to conservation practice. Inclusion of multi-level drivers of social processes and human behavior in spatial analysis and conservation planning represents a tremendous opportunity to improve outcomes for both wildlife and humans in shared landscapes (Lischka et al. 2018). A growing body of work has demonstrated novel ways to spatially integrate social and ecological factors that can better inform decision making for human-wildlife coexistence under changing conditions (Bryan et al 2011, Behr et al 2017, Williamson et al 2018). Here, we build on that foundation to underscore the utility of integrating social factors into traditional spatial analysis to promote human-wildlife coexistence in the American West.
Article
Full-text available
Successful conservation in the United States relies on collective stewardship by millions of private landowners, challenging those agencies and nongovernment organizations tasked with engagement and outreach. Perennially limited resources compound this challenge, highlighting a deep need for efficient social marketing. In the following research, we test the efficacy of two social marketing strategies— microtargeting and normative appeals—through a randomized controlled trial of an integrated social marketing campaign targeting riparian landowners in the Pennsylvania portion of the Chesapeake Bay Watershed. We used a microtargeting algorithm to predict landowners’ likelihood of responding to a conservation outreach campaign to create treatment groups of high-likelihood prospects versus random prospects (i.e., no microtargeting). A normative appeal was also included as an experimental factor in the campaign communicating that forested riparian buffer investments were common among similar landowners. Among microtargeted landowners, we observed a 66% increase in response to a riparian restoration survey compared to the control group. Additionally, we found a significant influence of a normative message among random (nonmicrotargeted) prospects, increasing response by 23% over the control group.We conclude conservation outcomes may be more efficiently achieved by deploying these marketing techniques on a wider scale to a variety of conservation challenges.
Technical Report
Full-text available
This report extends and updates an ongoing program of research analyzing Americans’ interpretations of and responses to climate change. The research segments the American public into six audiences that range along a spectrum of concern and issue engagement from the Alarmed, who are convinced of the reality and danger of climate change, and who are highly supportive of personal and political actions to mitigate the threat, to the Dismissive, who are equally convinced that climate change is not occurring and that no response should be made. The Six Americas are not very different demographically, but are dramatically different in their beliefs and actions, as well as their basic values and political orientations. The groups were first identified in a nationally representative survey conducted in the fall of 2008, and were re-assessed in January and June of 2010. The current report is the fourth in the series; in it we provide new insights into the informational needs of the six groups, their understanding of the health impacts of global warming, beliefs about current environmental impacts of global warming in the U.S., and support for local adaptation and mitigation policies.
Technical Report
Full-text available
This report extends and updates an ongoing program of research analyzing Americans' interpretations of and responses to climate change. This research segments the American public into six audiences that range along a spectrum of concern and issue engagement from the Alarmed, who are convinced of the reality and danger of climate change, and who are highly supportive of personal and political actions to mitigate the threat, to the Dismissive, who are equally convinced that climate change is not occurring and that no response should be made.
Article
Full-text available
Current systems of conservation reserves may be insufficient to sustain biodiversity in the face of climate change and habitat losses. Faced with these pressures, calls have been made to protect the Earth's remaining wildlands and complete the system of protected areas by establishing conservation reserves that (i) better represent ecosystems; (ii) increase connectivity to facilitate biota movement in response to stressors including climate change; and (iii) promote species persistence within intact landscapes. Using geospatial data, we conducted an assessment for expanding protected areas within the contiguous U.S. to include the least human-modified wildlands, establish a connected network, and better represent ecosystem diversity and hotspots of biodiversity. Our composite map highlights areas of high value to achieve goals in the western U.S., where existing protected areas and lands with high ecological integrity are concentrated. We identified important areas in the East rich in species and contain ecosystems that are poorly represented in the existing protected area system. Expanding protection to these priority areas is ultimately expected to create a more resilient system for protecting the nation's biological heritage. This expectation should be subject to rigorous testing prior to implementation, and regional monitoring will ensure areas and actions are adjusted over time. This article is protected by copyright. All rights reserved.
Article
Full-text available
Context: Recent conceptual developments in ecosystem services research have revealed the need to elucidate the complex and unintended relationships between humans and the environment if we are to better understand and manage ecosystem services in practice. Objectives: This study aimed to develop a model that spatially represents a complex human–environment (H–E) system consisting of heterogeneous social–ecological components and feedback mechanisms at multiple scales, in order to assess multi-dimensional (spatial, temporal, and social) trade-offs in ecosystem services. Methods: We constructed an agent-based model and empirically calibrated it for a semi-arid region in Northeast China, and examined ecosystem service trade-offs derived from the Sloping Land Conversion Program (SLCP), which is based on payment for ecosystem services. This paper describes our model, named Inner Mongolia Land Use Dynamic Simulator (IM-LUDAS), using the overview, design concepts, and details + decision (ODD + D) protocol and demonstrates the capabilities of IM-LUDAS through simulations. Results: IM-LUDAS represented typical characteristics of complex H–E systems, such as secondary and cross-scale feedback loops, time lags, and threshold change, revealing the following results: tree plantations expanded by the SLCP facilitated vegetation and soil restoration and household change toward off-farm livelihoods, as expected by the government; conversely, the program caused further land degradation outside the implementation plots; moreover, the livelihood changes were not large enough to compensate for income deterioration by policy-induced reduction in cropland. Conclusions: IM-LUDAS proved itself to be an advanced empirical model that can recreate essential features of complex H–E systems and assess multi-dimensional trade-offs in ecosystem services.
Article
Full-text available
The participation of private landowners in conservation is crucial to efficient biodiversity conservation. This is especially the case in settings where the share of private ownership is large and the economic costs associated with the public land acquisition are high. We examine the revealed participation choice of Danish forest owners in a voluntary conservation program, and use the results to spatially predict the likelihood of participation for the full population of Danish forest owners. The outcome is included in a probability model for species survival. Uninformed and informed (includes land owner characteristics) models are then incorporated into a spatial prioritization analysis of setting aside unmanaged forest. The choice models are based on socio-demographic data on the entire population of Danish forest owners and historical data on their participation in conservation schemes. The results show a significant efficiency may be gained through including information on private landowners' willingness to supply land for conservation. The landowner choice model provides an example of moving towards more implementable conservation planning. This article is protected by copyright. All rights reserved.
Article
Leveraging cognitive biases and social influence can make conservation efforts more effective
Article
Satellite data, and the tools that ecologists use to analyse them, are more accessible and plentiful than ever.
Article
Conservation success is contingent on assessing social as well as environmental factors so that cost-effective implementation of strategies and actions can be placed in a broad social-ecological context. Until now, the focus has been on how to include spatially-explicit social data in conservation planning, whereas the value of different kinds of social data has received limited attention. In a regional systematic conservation planning case study in Australia, we examined the spatial concurrence of a range of spatially-explicit social values and preferences collected using public participation GIS (PPGIS) methods with biological data. We then integrated the social data with the biological data in a series of spatial prioritization scenarios using Zonation software to determine the effect of the different types of social data on spatial prioritization vis-à-vis biological data alone. We found that the type of social data included in the analysis significantly affected spatial prioritization outcomes. The integration of social values and land-use preferences under different scenarios was highly variable and generated spatial prioritizations that were 1.2% to 51% different from those based on biological data alone. The inclusion of conservation-compatible values and preferences added relatively little new area to conservation priorities while in contrast, including non-compatible economic values and development preferences as costs significantly changed conservation priority areas. The multi-faceted conservation prioritization approach presented herein that combines spatially-explicit social data with biological data can assist conservation planners in identifying the type of social data to collect for more effective and feasible conservation actions.