Bird diversity indicates ecological value in urban
Michael C. Farmer &Mark C. Wallace &Michael Shiroya
#Springer Science+Business Media, LLC 2011
Abstract It is known that public greenspaces contribute positively to urban home prices;
yet urban ecologists also have known that not all greenspaces are equally valuable. Also
some ecologically valuable space appears on private residences, not only public spaces.
This work examines directly whether using a variable derived from bird species richness
and relative abundance adds new information regarding ecological value and if high values
of that variable significantly improve urban housing prices. We collected information on
approximately 368 home sales in Lubbock, TX from 2008 to 2009 from the Multiple
Listing Service: Sale Price, Square Footage, Lot Size and Age in 17 neighborhoods
identified by the Lubbock Realtor Association. We conducted bird counts in the vicinity of
each home sale and recorded both the total numbers of birds and the number of bird species
identified in a particular class—less ubiquitous bird species. Finally, we used GIS to record
the percentage of tree cover in the immediate area surrounding each sale. We constructed a
predictive model for a bird relative abundance and species richness variable (Bird) from
statistics. Home price for each sale then was regressed against the predicted value of
‘Bird’from the selected model and regressed against home price along with other attributes
from the Multiple Listing Service. The predicted value for Bird finds that the addition of
another desirable, less ubiquitous bird species improves mean home price by $32,028,
likely due to the human created landscapes on private properties immediately surrounding a
home sale. Curiously, the presence of a nearby park neither explained variation in the
ecological indicator nor contributed to home price elevation. This deliberately simple and
inexpensive indicator helped to direct attention to the composition of local landscapes in
specific areas to assess joint ecological and economic gains rather than presume a priori
that open greenspace jointly satisfies these dual objectives.
Keywords Hedonic price analysis .Greenscape .Urban wildlife
M. C. Farmer (*):M. Shiroya
Department of Agricultural and Applied Economics, Texas Tech University, Agricultural Science Room
204C, Lubbock, TX 79409, USA
M. C. Wallace
Department of Natural Resources Management, Texas Tech University, Lubbock, TX 79409, USA
A problem urban ecologists and ‘new urbanists’often face is the difficult task of
completing ecological-environmental evaluations in a timely and cost effective fashion
when land use decisions operate under short timelines and modest budgets. If we add a
need by decision-makers for follow-up ecological information with local zoning and
planning boards, this places the urban ecologist and ecological economist with a
conundrum. Communities slated for new development or redevelopment seek permits and
zoning rules in a relatively rapid policy design, decision and development timetable. The
time to conduct a full wildlife population assessment or even a landscape inventory is often
Currently in these environments certain defaults frequently prevail. For example, it is
common to designate any greenspace as ecologically beneficial, environmentally low foot-
print and beneficial to economic development. The mantra can be invoked to exculp policy-
makers from closer ecological examinations. Perhaps this is because those same defaults
have been proven to benefit housing values with little need for nuanced ecological
assessment (Cavailhés et al. 2007; Jim and Chen 2006; Wolters 2001). Recently studies are
beginning to show that some people pay more to reside near amenities that show more
nuanced urban landscapes or greenscapes (Cho, et al. 2008); for example, homes with
highly educated owners may prefer quality habitat directly (Bark et al. 2009).
From an ecological perspective, the type and the nature of urban greenspaces yield
very different outcomes (Pickett et al. 2001; Mansfield et al. 2005; Sandström et al.
2006). Greenspace can be simple open space such as a greenway, or it can be a rich
ecological space populated with urban wildlife and forests (Oleyar, et al. 2008).
Nonetheless studies regularly report that undifferentiated open space is an unambiguous
public good yielding multiple environmental benefits (Bolund and Hunhammar 1999). So
as the urban ecologist aspires to add context to an assertion that all greenspace is
necessarily ‘green,’we propose to consider methods wherein a little extra field work by
the urban ecologist working with economists might identify even stronger gains to
housing values, environmental footprint and urban wildlife that might otherwise go
unnoticed in a quick-paced development cycle.
This work is motivated by a concern that the form of urban landscapes, of which
greenspace is part, needs to be assessed more explicitly. Economic and ecological-
environmental effects are more subtle; yet that is often difficult to inject into public
discourse in time limited policy settings. One method to motivate rethinking is to
design filters—not full ecological evaluations—that can be completed quickly and
cheaply but nonetheless compelling to local planners and zoning boards. Filters likely
to be especially effective in this policy setting would indicate four things that we
consider in this work:
&Does the filter indicate a potential to increase economic value of a development?
&Does the filter indicate a potential to improve ecological and other environmental
&Does the filter require few resources and little time to complete?
&Does the filter help to identify locations where follow-up work is likely to be most
This work presents a case study for the first three, pointing to the last. Given prior
literature, a measure that emphasizes the number of bird species present in the study area,
but that tend to be observed in areas with a variable vegetative height density, is argued to
be an indicator of an array of desirable ecological outcomes. We measure if high values for
the indicator map to higher home prices in an immediate area about the size of a city block.
In this case, the filter was relatively easy to complete in terms of time to conduct the
repeated observations and of cost to conduct the study.
For the last concern, ecologists may use a filter to point to quite specific locations in the
urban space that represent specific opportunities to increase economic development values
while improving ecological-environmental outcomes. The filter not only may provide
breathing space as a precautionary practice during the planning and development approval
process, but it may identify where higher resolution follow-up field work is likely to be
beneficial, but also cheaper, faster and reasonably robust.
A seminal work by MacArthur and MacArthur (1961) offers one opportunity to construct
such a filter that could be useful in numerous regions. This work drew a strong relation
between landscape structure and avian wildlife diversity: the more complete the vertical
landscape composition of vegetation across heights ranging from low shrubs to higher tree
canopy, the richer the avian species diversity. A fuller vertical composition attracts bird
species beyond the ubiquitous species. For examples of ubiquitous species in our study
area, these included the house sparrow (Passer domesticus), European starling (Sturnus
vulgaris), great-tailed grackle (Quiscalus mexicanus), and Eurasian collared dove
(Streptopelia decaocto). Some members of these species are observed at nearly every site
in our study area and often in great numbers, notably the great-tailed grackle.
Subsequent studies validate the MacArthur and MacArthur (1961) assertion that the
indicator works both ways: diverse landscape compositions attract diverse avian species;
but, also, if the researcher observes more species diversity, or presence of several less
ubiquitous species, that is a powerful indicator of a diverse vertical landscape composition
(Sandström et al. 2006). That is important. As ecologists have noted that urban ecosystems
are quite diverse and scale dependent (Pickett et al. 2001), a common urban ecosystem
indicator adopted by ecologists has been tree presence, density and location. DeGraaf and
Wentworth (1986) noted that bird diversity varied with, amongst other dimensions, tree
density, or tree density predicted bird diversity. Oleyar, et al. (2008) noted the isomorphic
relationship: urban forest functionality was an integration of many factors; and bird
diversity was a good predictor of tree density. Fernández-Juricic (2000) reached a similar
finding. They noted that trees located at the street level had a positive effect on bird
diversity and population abundance. Fernández-Juricic (2000) also noted that bird diversity
varied with vegetation in urban settings. Crooks et al. (2004) finally equate the presence of
bird diversity with urban ecosystem diversity broadly. Temporal variation in canopy and
vegetation would be indicated by an observation of neighborhood bird diversity and
population. This implies that a more height variable canopy and vegetation, which has
known positive environmental externalities, likely supports a richer population of birds.
Some of the associated benefits of a rich greenscape with varied vegetative height range
from reduced noise and air pollution, climate regulation, watershed management, rainwater
capture, drainage and sewage treatment (Bolund and Hunhammar 1999). Even health (Maas
et al. 2006) and physical activity benefits to avert chronic disease (Farmer and Lipscomb
2006) are associated with a rich urban landscape pattern. Other ecological-environmental
benefits also flow from simple tree canopy by itself: reductions in air conditioning costs,
improvements in air quality, sequestration of carbon, assistance in recharge of water tables
and provision of habitat for animals in urban settings (Breuste et al. 1998).
We accept the received literature that bird diversity among less ubiquitous bird species
responds to a more progressed and height varied landscape structure; as such, a diverse,
progressed vegetation structure delivers a raft of environmental benefits over a high input
monoculture lawn (Pickett et al. 1997).
As environmental and ecological studies have shown robust benefits to greenspace
generally, economists have identified robust responses to greenspace and open-space,
generally, as improving home values (Wolters 2001; Morancho 2003; Bell and Irwin 2002;
Lipscomb and Farmer 2005; Mansfield et al. 2005; Cavailhés et al. 2007). Some have gone
on to examine gross number of trees to value rental property (Thompson et al. 1999; Luttik
2000). As ecologists began to differentiate among greenspace benefits from studies
conducted at higher levels of resolution with respect to the composition of greenspace
(DeGraaf et al. 1985; Pickett et al. 2001; Fernández-Juricic 2000; Sandström et al. 2006),
data began to permit economists to examine if specific ecological characteristics also
benefited home values (Farmer and Lipscomb 2006; Bark et al. 2009; Cho, et al. 2008).
This work developed an ecological indicator to facilitate economic study and to improve
the precision of ecological-economic linkages while saving study costs and time. Since little
has been done in residential areas on the non-public spaces where most developed green
landscapes often are located, urban ecology benefits include private greenscapes in the
yards of individual homes, alleys and in the collective canopy of, say, a street.
This level of resolution facilitates sharper economic evaluations. This work adapted a
higher resolution ecological indicator that operates at the scale of a city block; and then
estimated if home prices close to bird observation sites responded positively. The reduced
cost and time is informative.
We anticipate that a full landscape inventory that evaluates landscape composition
according to a consistent and meaningful coding protocol will be difficult and expensive.
We sought an indicator easy to measure yet likely to be associated with the array of benefits
identified by bird species diversity. The indicator is not a stand alone ecological assessment
of bird diversity, but a broad robust indicator of environmental and urban ecological
benefits consistent with the literature, that is also easy to collect.
We adapted an indicator to serve as a filter that more quickly highlights potential urban
ecological and economic gains to zoning and development practices. The indicator is the
simple product of the number of less ubiquitous bird species observed at a site and the
number of birds observed. The indicator weights highly repeated observations of less
ubiquitous species. If only ubiquitous species were observed, the value for the indicator was
zero. This is because our goal was not directly to complete a bird population study but to
infer diversity of vegetative cover and broad ecological diversity. To predict values for the
indicator variable (Bird), tree canopy cover was observed over an area the size of a city
block near a bird observation site. Home sales were observed within the same range. So the
indicator, while neither a comprehensive landscape inventory nor a wildlife population
study nonetheless was observed at a relatively high spatial scale to indicate higher
resolution environmental and ecological benefits. These allow higher resolution economic
response estimates of ecological-environmental services. In our study area the less
ubiquitous species included the American robin (Turdus migratorius), blue jay (Cyanocitta
cristata), mourning dove (Zenaida macroura), northern mockingbird (Mimus polyglottos)
and western kingbird (Tyrannus verticalis).
In evaluating areas of rich vegetation composition where the built environment was
largely private residences, bird diversity and abundance is often easier to capture than direct
completion of a comprehensive inventory of vegetative structure and composition. The
question is whether this local vegetation diversity supporting more diverse bird life is also
more valuable to home owners, especially in comparison to other forms of public
greenspace such as a local park. Rather than construct a continuous or categorical metric of
landscape quality, we measured the closely associated urban wildlife variable directly to
assess the wildlife variable’s effect on housing prices across a 40 sq-mi urban space.
We collected a data set of recent house sales, tree canopy density and bird species richness
in 17 neighborhoods of interest. The area spans approximately 40 square miles in southwest
Lubbock, TX (Fig. 1.). We chose 17 neighborhoods in Lubbock, TX suggested by realtors
as diverse for the area and also thought to have a high number of recent sales.
Neighborhoods are of similar size and most had comparable numbers of sales. Two
neighborhoods (Southhaven and Melonie Park South) had only four or five sales, but were
somewhat unique neighborhoods. They were retained in the housing price model; but
eliminated in the model to estimate significant effects on the Bird index.
Fig. 1 Map of neighborhoods for study. Shaded neighborhoods were used in final model for Predicted Bird
(Model 3). Each grid square is approximately one square mile
Data sources and collection
The first set of data collected was housing residential sales data from June 2006 to
December 2008 within the neighborhoods of interest by extracting information from the
Lubbock area Multiple Listing Service (MLS). We drew from MLS documents physical
attributes of each house and the selling price.
Percent tree canopy cover was estimated from Google Earth photos from 2008. Grid size
adopted used a 192 point dot-grid overlay on Google Earth images scaled at an elevation of
3,200 ft above the surface, with a resolution of 0.56 by 0.70 miles, and recorded percent of
grid dots overlapping trees.
Relative abundance of birds was collected using point count surveys (Bibby et al.; 1992)
recording detections <or> 50 m. Birds detected were identified to species and were
recorded at n≥8 sites in each of n=17 neighborhoods over ≥2 different mornings during
summer 2009. We categorized bird species as synanthropic species and suburban species.
Species A (synanthropic species) birds were ubiquitous native and invasive species for
example, house sparrow (Passer domesticus), European starling (Sturnus vulgaris), great-
tailed grackle (Quiscalus mexicanus), Eurasian collared dove (Streptopelia decaocto); and
Species B birds (suburban species) were desirable urban birds like: American robin (Turdus
migratorius), blue jay (Cyanocitta cristata), mourning dove (Zenaida macroura), northern
mockingbird (Mimus polyglottos), western kingbird (Tyrannus verticalis). The bird variable
used for analyses was calculated by multiplying the total number of birds at an observation
site by the number of Category B bird species, for an indexed variable labeled Bird. All
data, such as tree canopy or the measure of the bird variable were referenced to each
residential unit sold in the survey; so the bird variable, for example, realized a different
value for each housing unit sale that we examined.
Model selection methods for bird richness and abundance
Our approach, based on bird richness and relative abundance, resembles other models that
use birds as a barometer of ecosystem complexity (MacArthur and MacArthur 1961), urban
ecosystem biodiversity (Crooks et al. 2004) or the quality of greenspace (Sandström et al.
2006; and Fernández-Juricic 2001), or urban forests (Oleyar, et al. 2008). These studies
build a theory associating urban avian wildlife abundance as well as diversity with a sizable
number of favorable environmental outcomes.
First, we sought to explain variation in the bird variable. Bird is defined as an inter-
action variable between the total numbers of birds multiplied by the number of different
kinds of Species B birds. We regressed Bird against the percent tree cover, 16 dummy
variables to capture variation due to neighborhood effects across the 17 neighborhoods, and
a variable ‘near park,’defined by the presence (1) or absence (0) of a park fitting two
criteria: the park was located within the neighborhood and within a half mile of the
residential unit of interest. Model 1 predicts Bird using all 18 independent variables. Using
Model 1 results, we eliminated three insignificant variables. The public greenspace, ‘near
park’was eliminated at this stage. We then regressed Bird against the significant variables
to derive Model 2. Following this, we derived Model 3 from the significant variables in
Model 2, eliminating two more variables. The process eliminated neighborhoods that had
less than 6 sales in 2008 and 2009 to form Model 3. We regressed the remaining 11
neighborhood dummy variables and tree cover on Bird.
We compared the three Models for Bird using AIC
values for the three regressions.
Having chosen the Model with the lowest AIC
score (results below) and the highest weight
to predict Bird, we extracted the predicted value of Bird from that model, which we labeled
We then tested regressed each observed home sale price against household structural
variables and Predicted Bird to examine how well this urban wildlife measure correlated
with home prices.
Two models of home price compared
We estimated several factors expected to affect housing price. The housing characteristics
used were Square Footage, Age of House, Garage (0/1) and Lot Size. We compared two
housing price models to predict home price for sales in 2008 and 2009. The first housing
price model did not include Predicted Bird (Generic Model); the other model included
Predicted Bird (Modified Model). The two models are:
Generic Housing Price Hedonic Model:
House value ¼b0þb1
»Square Footage þb3
Modified Housing Price Hedonic Model:
House value ¼b0þb1
»Square Footage þb2
»Lot Size þb5
Model selection for bird We conducted 296 point counts and used total number of birds and
total number of species B detected on 2 point counts (randomly selected if there were >2
conducted) for each neighborhood to attach a Bird value to each sale. The results of model
comparisons for Bird are presented in Table 1.
Model 3, in which Bird was regressed against tree cover and remaining significant
neighborhood dummy variables was the best model from our tests. Model 3 showed the
lowest overall AIC
value; was assigned the entirety of the explanatory weight; and
generated the highest R
(=0.4965) value. Therefore we used Model 3 to generate
‘Predicted Bird,’which was then used as an independent variable to explain home sale
prices for which we had MLS data in 2008–2009 across 17 neighborhoods.
Model selection method for home price We collected MLS data for 368 home sales in
Lubbock, TX from 2008 to 2009. Three hundred twenty-three had no missing data and
were used for analyses.
Table 2reports our hedonic price model for the Modified Housing Price Model. Home
Price was regressed against physical house characteristic variables (house square footage,
garage, lot size, and age) and one ecological variable (Predicted Bird). Predicted Bird,
square footage, age and lot size of each house were all significant (P< 0.001). R
improved for Modified Model that included Predicted Bird over the Generic Model,
from 0.7348 to 0.7792 using Predicted Bird.
Not shown are several alternatives to the Generic Housing Price Model and Modified
Housing Price Model. ‘Near park’was added to both models; and in a third case, ‘near
park’directly replaced Predicted Bird in the Modified Model. In all three cases, ‘near park’
added virtually no explanatory value (t-stat <0.18 in all cases) and, in each case, overall
model performances fell (see below).
Table 3summarizes key information. Table 3records neighborhood name (referenced on
Fig. 1) and reports the mean values for Bird, Tree Cover, and Home Price. Implications are
Model for predicted bird Overall, Model 3 in displayed the lowest overall AIC
predicting Bird variation; was assigned the entire explanatory weight (100%); and showed
the highest R
. Because Model 3 was the most efficient estimate of bird species richness
and relative abundance, this model was used to predict Bird (i.e. Predicted Bird). Predicted
Bird from this model then was used in the modified hedonic regression (Modified Model)
to predict of house value. The final model of Bird (Model3) showed a strong positive
relationship between bird species richness and relative abundance with tree cover. Tree
cover producing a t stat=4.13. Yet paradoxically, proximity to a park was not a significant
predictor of bird richness and relative abundance. In fact, the ‘near park’variable produced
a t stat=0.15 in Model 1. When we added ‘near park’to Model 3 as a test, ‘near park’again
realized a virtually identical t stat (0.148); and when we replaced ‘near park’for Predicted
Bird the t-stat was low (0.185).
This seemingly counterintuitive result for parks is explained by the composition of tree
cover in most Parks in Lubbock, TX, which is not particularly dense, nor a multi-layered
canopy. Tree cover is composed of punctuated trees heavily pruned below 8′to 10′. Also
many city parks are converted (filled) playa lakes, managed partly to serve as flood control
drainage as well as public space. So parks on average were not valuable in predicting higher
Bird outcomes. A development model premised on clearing space for a park but not
Table 1 Akaiki (AIC
) comparison of regression models to derive Predicted Bird variable
Statistics for model comparison
Model SSE −2log(L) K n AIC
1 242279 −4299.63 17 319 −4263.60 0
2 247221 −4292.51 12 319 −4267.49 0
3 288441 −4386.90 11 319 −4364.04 1
Model 1 Bird regressed against near park, tree cover and neighborhood dummy variables
Model 2 Bird regressed against tree cover and significant neighborhood dummy variables from Instrument 1
Model 3 Bird regressed against tree cover, significant neighborhood dummy variables with >6 sales from
Variables: bird, was derived using total birds times number of suburban bird species; near park, was distance
to nearest open space/city park; and, neighborhood dummy variables. Data are from summer 2009 in
Table 2 Modified hedonic price model
Hedonic Price Model With Predicted Bird
Dependent Variable: Price
Number of Observations Read 368
Number of Observations Used 323
Number of Observations with Missing Values 45
Analysis of Variance
Source DF Sum of Squares Mean Square F Value Pr > F
Model 5 3.746561E12 7.493121E11 228.27 <.0001
Error 317 1.040565E12 3282540056
Corrected Total 322 4.787126E12
Root MSE 57293 R-Square 0.7826
Dependent Mean 234809 Adj R-Sq 0.7792
Coeff Var 24.40002
Variable Label DF Parameter Estimate Standard Error t Value Pr >−t|
Intercept Intercept 1 –45434 16924 –2.68 0.0076
Square Footage 1 94.25576 3.95636 23.82 <.0001
Age of House 1 –2926.60790 286.06107 −10.23 <.0001
Garage 1 8655.86801 6582.18243 1.32 0.1894
Lot Size 1 1.55797 0.34163 4.56 <0001
Predicted Bird 1 1243.93485 172.42051 7.21 <.0001
Mean total number of bird= 24.14. Mean value of B Species=2.08; for $32,028 gain in expected mean home value from a gain of one more Type B species observed
Regressing MLS characteristics (house square footage, garage, lot size, and age) and ecological variables (Predicted Bird) to prices of most recent home sale in Lubbock, TX.
improving zoning and building requirements for an improved landscape on residential
property would remove an ecological benefit and substitute for it an expensive public
commitment to upkeep that cleared area. The same decision also would lower home
Home price effects of predicted bird
The Modified hedonic housing price regression included Predicted Bird; again the expected
value of Bird associated with each home sale from Model 3 discussed above. The
regressors for the Generic housing price model are identical, except they do not include
Predicted Bird had a significant effect on housing prices. First, the significance level of
Predicted Bird is well above 99%, with a t-stat=7.21. This stands in contrast to results for
‘near park.’When entered into either the Generic Home Price Model or the new Modified
Home Price Model, ‘near park’had an insignificant effect on home price. The Modified
Model, for example, generated a t stat=0.49 for near park. Further, in both models, the
addition of ‘near park’as a regressor resulted in a decline in adjusted R
. So greenspace
alone was not significantly related to improvements in either the urban wildlife variable,
Bird, nor in average home prices.
In contrast, the inclusion of Predicted Bird increased the value of the adjusted R
72.6% to 77.9%, meaning the variable added to real explanatory power for the regression
adjusting for the decline in the degrees of freedom. Again, Adjusted R
fell when ‘near
park’was added. The inclusion of Predicted Bird into the regression model for home price
was not only significant, but passed several internal consistency checks when the Modified
Model was compared to the Generic Model.
Table 3 Mean values for key variables by neighborhood
Mean bird Mean home price Mean tree cover
Tech terrace 91 178,285 0.33824
Whisperwood 51 167,881 0.37414
Rushland 65 386,933 0.37905
Brentwood 56 145,500 0.3
Tanglewood 123 470,838 0.385
Greenlawn 37 98,124 0.28
Melonie Park South 67 174,806 0.29
Melonie Park 53 177,742 0.24167
Melonie Gard 27 318,358 0.11739
Ravenwood 10 280,960 0.05
Southhaven 30 307,437 0.29426
Lakeridge 26 148,243 0.28556
Farrar 10 222,712 0.04107
Regal Park 34 155,861 0.05071
Pleasant Run 46 340,056 0.23444
Regency Park 9 162,448 0.05
For Neighborhoods Used in Bird Model, See Fig. 1
Model consistency checks
Compare hedonic regressions with and without Predicted Bird (the Generic and the
Modified Model above). Theory suggests some estimated parameters that explain home
prices should change when Predicted Bird is added to the Modified Model. Theory also
suggests others should not change.
Square footage is one example. The parameter value for square footage, an interior home
attribute, should not shift between the Generic and the Modified models. Adding Predicted
Bird into the Modified Model is an exterior attribute, which we expect to be independent
from values of interior home attributes. Comparing the Generic to the Modified model, the
parameter estimate for square footage shifted only from $94.25 to $94.48 per square foot—
almost no change. Both results were well in line with the average $92 to $98 per square
foot value provided to us by local realtors (Linda Gaither, Westmark Realty, personal
Lot Size and AgeofHouse parameter estimates, however, are predicted by theory to
change between the Generic Model and the Modified Model, which includes Predicted Bird
as a regressor.
The regression parameter for Lot Size (in square yards) fell when Predicted Bird was
entered into the Modified regression. Some lots will be improved with added tree cover and
others will not. So we expect that when Predicted Bird is added to the regression, the value
of lot size should fall. By entering Predicted Bird into the Modified Home Price Model, the
model ought to differentiate better between the values for lot size on unimproved space
versus private lots improved by mature trees. Indeed, the value for lot size fell from $1.97
per square yard in the Generic Model to $1.56 per square yard in the Modified Model.
What is noteworthy is that the ecological indicator helped to tease out a qualitative
difference among lots with and without mature trees in the economic model.
We expect a similar type of outcome for the coefficient values for house AGE. Older
houses are less valuable on average. Yet older homes can accommodate older, more mature
trees. That feature should increase home values. Again, some older houses have mature
trees and some do not. So with the addition of Predicted Bird into the Modified Model,
house age by itself should exert a more powerful negative effect. In the Generic Model, it
was estimated that age diminished home price by an expected $1516.84 per year old (e.g. a
10 year old home would fall by $15,168). Adding Predicted Bird to the Modified Model
increased the predicted negative impact of home age, reducing housing price by $2926.61
per year. Differentiating between older homes with mature trees and older home without
mature trees, the stronger depreciating value of home age was more properly identified.
These shifts in parameter values from the Generic Housing Price Model and the
Modified Housing Price Model appeared as expected. Further, no shifts in parameters arose
between where they were not expected, such as square footage. This internal consistency
would explain the improvement in adjusted R
from 0.7348 to 0.7792 by adding Predicted
Estimated mean home price premium for bird richness and abundance
There is a strong average premium for features that attract more desirable bird species, Bird.
For the mean home sale in this market, a premium of $32,028 was correlated with the
presence of just one more Type B, or suburban, species observed near the home site. The
number, while high, makes sense when we note the average number of type B species
observed was only 2.08. So the value on average from, say, 2 to 3 of such species may
suggest a strong improvement in the aesthetics, neighborhood walkability and the wildlife
diversity that collectively helps to form a ‘streetscape.’
The average number of Type B bird species observed for all positions was 2.08. The
mean number of total birds observed was 24.14. On Table 2, as the total value of Predicted
Bird increased by one unit, home price, on average, improved by $1243.93. That regression
coefficient is equivalent to observing another bird at a site where at least one type B species
was observed. The critical contribution to home value, however, arose likely from a
landscape that on average improved the number of Type B species by one. Such a change,
on more observed type B species, realized a gain of $32,028.47 in home price; or average
home price changed by 24.14(mean of total birds observed)*$1243.93 (expected home
price change as Predicted Bird value increases by one unit).
Not to be taken too literally, the result does divide the observations into those with
landscapes nearby that favor a vegetative cover that would improve diversity in Bird
species in terms of including non-ubiquitous bird species from those that do not. The home
price contribution of the suite of services correlated with a higher Bird value is not only
significant, but it is a substantive contribution to overall home price. Whatever are the
component details of the landscape differential or, perhaps, ecological function, the sharp
impact on home prices broadly across this study area is one that we submit would draw
attention from planners and developers.
Efficient targeting of subsequent field work
Our final comment is that our empirical results direct attention to higher resolution studies at
specific locations. This is item four regarding the benefits of a good filter in the Introduction. A
way to target locations for further study is to examine those sites that have high recorded Bird
values yet were predicted to have much lower Bird values. We hypothesize that unobserved
landscape characteristics, such as vertical height cover, would be responsible for higher
observed than expected Bird values. This would isolate quite specific spots for closer
inspection. Landscape architects or forest ecologists would need to conduct site studies, but
some features stand out to suggest those studies would be constructive. We start with cross
neighborhood comparisons and then drill down to specific blocks.
Starting with neighborhood by neighborhood comparisons, when we look a little closer
at the neighborhoods in the Bird prediction model (Model 3) that explain very high Bird
values that are statistically significant and compare them to neighborhoods with
insignificant or negative effects on mean Bird values, the vegetative canopy has a visibly
more complete vertical composition. On Table 3, consider two neighborhood comparisons.
Comparing Tanglewood to Rushland, they neighborhoods have similar tree canopy
percentages (0.385 versus 0.379) and other home characteristics, such as size, were similar.
Tangelwood realized much higher values for Bird (123 versus 65) and for home value
($471 K versus $387 K). As the neighborhoods are adjacent (Fig. 1), it is clear that older
trees with high canopies with lawns below dominate the landscape in Rushland compared
to a more height varied and denser tree composition for the same overall canopy coverage.
Three other neighborhoods with reasonably similar characteristics, Tech Terrace, Wisperwood,
and Brentwood, have similar values for tree cover (0.38, 0.37, 0.30, respectively); yet Tech
Terrace realizes higher home prices ($178 K, $167 K, $146 K, respectively) and Bird values
(91, 51, 56), and by casual inspection has more landscape diversity.
The is good reason to suspect the area ‘beneath the canopy’explains much of the
variance and serves as a guide to a higher resolution enunciation of ecological services.
Oleyar, et al. (2008), evaluate urban forest functionality, especially in light of the
connection between avian assembleges in a fragmented urban landscape and vegetative
structures (Crooks, et al. 2004).
Drilling down just a little further to street block by street block, there is variation in
vertical composition of the vegetative landscape at this resolution. We compare six bird
sites in Wisperwood and three in Tech Terrace with identical tree cover percentages,
evidence we believe useful to policy.
In Wisperwood areas with equivalent low tree canopy sites, those that predicted Bird
accurately were largely not vegetated, marked by younger ornamental fruit trees and
impervious surfaces. This contrasts with sites yielding higher than expected Bird values,
which had, in one case, clearly more undergrowth (still modest) and another site near an
alley that possessed both more overgrowth and more vegetative layers than dominant lawn
landscapes. In Tech terrace we identified local blocks with higher than, lower than and
expected bird values for canopy cover values that all realized the identical cover percentage
measure. The highest Bird valued site had a dense tree cover, as Oleyar, et al. (2008) found
elsewhere, a high canopy cover, yet a filled understory, including some fruit and nut bearing
plants. The site that predicted Bird higher than actually observed was not a layered
landscape and the site was adjacent to park open space inhabited by ubiquitous birds
(grackles) with no barrier. The site that predicted well also had open space nearby; but it
was more protected. Also a strong canopy existed with limited undergrowth and somewhat
larger lawns. These all conform to the developed theory.
These results, using a very accessible ecological indicator for the region as a filter, focus
attention for comparison to a very few select neighborhoods and even fewer selected blocks
for quite close, high resolution inspection and study. The higher resolution results will not
surprise ecologists who have found repeatedly that a progressed or layered vegetative
structure tends to encourage diversity of avian and other urban wildlife (Sandström, et al.
2006; Fernández-Juricic 2000), as well as many other environmental benefits (Crooks, et al.
2004); yet there was some corroboration in this exercise. What is perhaps valuable is the
speed and relative ease to obtain these comparisons from this one indicator that served as a
filter to identify particular sites to conduct more close study. Fortunately, those more
nuanced assessments correlated with direct economic benefits. Persons appear to be willing
to pay for some elements of the suite of services delivered, demonstrable by filtered
illustration to a few points.
These clearly casual inspections of neighborhoods or blocks with similar tree canopy but
higher Bird measures corresponded to landscapes with intermediate zone vegetation at the
one meter and two meter levels. So, a planner may wish to know what shapes contribute to
ecological value per se; and, then, with even greater specification from very local fieldwork
filtered through our ecological indicator coupled with an economic assessment, a planner
may be able to isolate which improvements lead to economic improvements and at what
point ecological enhancement starts to improve at the expense of further economic
outcomes. If the ecologist, economist and planner seek ecological economic outcomes, the
analysis here points to subsequent fieldwork where landscape design contributes to
ecological outcomes and to economic development. The sort of analysis conducted here
opens the pathway for defining where to sharpen field work to study specific ecological and
social conditions that affect ecological and economic outcomes, separately and together.
Acknowledgements Linda Gaither and Moses Russell, Lubbock Realtors; The 2008 real estate economics
class in Agr and Applied Economics, TTU; L. Navarette, J. Strova, and K. McCabe, field technicians in
Natural Resources Management, TTU; and C. Casanova, Landscape Architecture, TTU.
Bark RH, Osgood DE, Colby BG, Katz G, Stromberg J (2009) Habitat preservation and restoration: do
homebuyers have preferences for quality habitat? Ecol Econ 68:1465–1475
Bell K, Irwin E (2002) Spatially explicit micro-level modeling of land-use change at the rural–urban
interface. Ag Econ 27(3):217–232
Bibby CJ, Burgess ND, Hill DA (1992) Bird census techniques. Academic, New York
Bolund O, Hunhammar S (1999) Ecosystem services in urban areas. Ecol Econ 29(2):293–301
Breuste J, Feldmann H, Uhlmann O (1998) Urban ecology. Springer, Berlin
Cavailhés J, Brossard T, Foltête J-C, Hilal M, Joly D, Tourneux F-P, Tritz C, Wavresky P (2007) Seeing and being
seen: a GIS-based hedonic price valuation of landscape, working paper. INRA-Cesaer and CNRS-Thma,
Cho S-H, Poudyal NC, Roberts RK (2008) Spatial analysis of the amenity value of green open space. Ecol
Crooks KR, Suarez AV, Bolger DT (2004) Avian assemblages along a gradient of urbanization in a highly
fragmented landscape. Biol Cons 115(3):451–462
DeGraaf RM, Wentworth JM (1986) Avian guild structure and habitat associations in suburban bird
communities. Urban Ecol 9:399–412
DeGraaf RM, Tilghman NG, Anderson SH (1985) Foraging guilds of North American birds. Environ Manag
Farmer MC, Lipscomb C (2006) The role of economic analysis in the design and evaluation of healthy
communities and regions. In: Shevliakova E (ed) Regional economics: social and economic processes.
Toglatti University Press. 326–340
Fernández-Juricic E (2000) Bird community composition patterns in urban parks of Madrid: the role of age,
size, and isolation. Ecol Res 15(4):373–383
Fernández-Juricic E (2001) Avian spatial segregation at edges and interiors of urban parks in Madrid, Spain.
Biodivers Conserv 10(8):1303–1316
Jim CY, Chen WY (2006) Impacts of urban environmental elements on residential housing prices in
Guangzhou (China). Landsc Urban Plan 78(4):422–434
Lipscomb CA, Farmer M (2005) Household diversity and market segmentation within a single
neighborhood. Ann Reg Sci 39(4):791–810
Luttik J (2000) The value of trees, water and open space as reflected by house prices in The Netherlands.
Landsc Urban Plan 48(3/4):161
Maas J, Verheij RA, Groenewegen PP, De Vries S, Spreeuwenberg P (2006) Green Space, urbanity, and
health; how strong is the relation? J Epidemiol Commun H 60(7):587–592
MacArthur RH, MacArthur JW (1961) On bird species diversity. Ecol 42:594–598
Mansfield C, Pattanayak SK, McDow W, McDonald R, Halpin P (2005) Shades of green: measuring the
value of urban forests in the housing market. J For Econ 11:177–199
Morancho AM (2003) A hedonic valuation of urban green areas. Landsc Urban Plan 66(1):35–41
Oleyar D, Greve A, Withey JC, Bjorn A (2008) An integrated approach to evaluating urban forest
functionality. Urban Ecosys 11:289–308
Pickett STA, Burch JWR, Dalton SE, Foresman TW, Grove JM, Rowntree R (1997) A conceptual framework
for the study of human ecosystems in urban areas. Urban Ecosys 1:185–199
Pickett STA, Cadenasso ML, Grove JM, Nilon CH, Pouyat RV, Zipperer WC, Costanza R (2001) Urban
ecological systems: linking terrestrial ecological, physical, and socioeconomic components of
metropolitan areas. Annu Rev Ecol Syst 32:127
Sandström UG, Angelstam P, Mikusiński G (2006) Ecological diversity of birds in relation to the structure of
urban green space. Landsc Urban Plan 77(1/2):39–53
Thompson R, Hanna R, Noel J, Piirto D (1999) Valuation of tree aesthetics on small urban interface
properties. J Arboric 25(5):225–234
Wolters MJJ (2001). The business of modularity of business.PhD Thesis, Rotterdam, Erasmus University