E CO L O G I CA L IN F O R MA TI CS 3 ( 2 0 08 ) 87–9 6

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / e c o l i n f

Raccoon ecology database: A resource for population

dynamics modelling and meta-analysis

Erin E. Reesa,⁎, Bruce A. Pondb , Julia R. Phillipsa , Dennis Murraya

a
Trent University, DNA Building, 2140 East Bank, Peterborough Ontario, Canada K9J 7B8
b
Ontario Ministry of Natural Resources, Peterborough, Canada

AR TIC LE I N FO ABS TR ACT

Article history: The value of scientific studies increases and is extended when their data are stored in a
Received 9 February 2007 manageable and accessible format. This is demonstrated through development of a raccoon
Received in revised form ecology database (REDB) to store, manage and disseminate available peer-reviewed and
22 January 2008 unpublished data on raccoon (Procyon lotor) biology, ecology and raccoon rabies, including
Accepted 23 January 2008 citations for data sources. Over 800 documents were identified and citations for them
entered into the database as literature references. Approximately 1000 trait values were
Keywords: entered from almost 200 of these sources. These data included estimates of population
Data warehouse density, survival rates, rabies incubation period, litter size, body weight, dispersal distance
Ecological database and home-range size, often by age or sex class. Each datum is linked to a citation for its
Relational data management system source, and to information about location and land use in the study area, time of year the
Meta-data study was undertaken, sample size, and variance. The relational database design enables
Meta-analysis querying and easy updating and manipulation of data.
The relational data model is presented, as is its application in further developing an
individual-based, spatially-explicit population model of raccoon rabies. Using information
queried from the REDB benefits model development by: i) assessing the appropriateness of
input parameter values, ii) providing sources for citing input values, iii) parameterising the
model to different geographic regions, iv) enabling meta-analyses for evaluating model
structure, as well as further contributing to parameterisation at specific locations, and v)
providing biologically appropriate parameter input values for model sensitivity testing. The
REDB is a useful research resource that will increase in value with ongoing inclusion of data
from future raccoon and raccoon rabies studies and serves as a model for database design
and research applications to other species. The database and an empty database for use
with other species are available online (http://redb.nrdpfc.ca).
© 2008 Elsevier B.V. All rights reserved.

1. Introduction studies. Traditionally, ecological reviews were qualitative and

The synthesis of data from many independent studies into data
warehouses encourages a more comprehensive analysis of
ecological systems (Jones et al., 2006). This is useful for advancing
ecological theory, for developing appropriate nature conserva-
tion strategies and extracting additional value from past research
used a standard “vote counting” approach. However, progress in
the methodology of meta-analysis is increasing the ability of
review studies to analyse and present their results more
objectively and quantitatively (Arnqvist and Wooster, 1995).
There are now well-established methodologies for properly
undertaking such studies (Roberts et al., 2006; Gates, 2002), and

⁎ Corresponding author.
E-mail address: erin.rees@nrdpfc.ca (E.E. Rees).

1574-9541/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecoinf.2008.01.002
88 E CO L O G I CA L IN F O RM A TI CS 3 ( 2 0 08 ) 87–9 6

Table 1 – Description of traits contained in the REDB, and the number of studies contributing data to the traits
Trait Description Count

age_of_first_movement_with_mother
age_of_independence_from_mother
age_of_weaning
age_ratio
body_condition
body_neck_circumference
body_tail_length
body_total_length
body_weight
body_weight_change
capture_rate
contact_rate
dens
density
density_dependence
disease_cycling_period
disease_enzootic_duration
disease_epizootic_period
disease_first_epizootic_duration
disease_first_epizootic_maximum
disease_first_epizootic_period
disease_incubation_period
disease_infectious_period
disease_natural_immunity
disease_positive_cases_time_to_peak
disease_prevalence
dispersal_distance
dispersal_period
habitat_selection
home_range
home_range_core
home_range_core_overlapping
home_range_overlapping
life_span
mortality_causes
movement_rate
population_size
reproduction_birth_period
reproduction_breeding_period
reproduction_chance_of_giving_birth
reproduction_consortship_duration
reproduction_consortship_partners
reproduction_consortship_success
reproduction_fecundity
reproduction_gestation
reproduction_litter_size
reproduction_mating_system
reproduction_oestrus_duration
reproduction_oestrus_period
reproduction_parous
reproduction_placental_scars
roadkill_index
sex_ratio
site_fidelity
species_range
survival
Age at which an offspring explores freely, but is still dependent on its mother 2
Age at which an offspring is no longer dependent on its mother for survival 4
Age at which an offspring is being weaned by its mother 1
Ratio of juveniles to adults 2
Body condition 15
Circumference of an animal's neck 9
Tail length of an animal 9
Total length of an animal 9
Body weight 39
Change in body weight over time 1
Rate at which animals are capture in traps 21
A measure of animals interacting among each other per unit time 6
Number of dens used by a animal for a defined period of time (e.g. season, year) 23
Density of animals: density = population size / area 117
Documented factors regulating population levels of animals 1
Duration of a period in a cycling pattern of disease incidence in a population 1
Length of time an established disease exists in an animal population 11
Duration of time between cycling peaks of disease incidence this is epizootic 0
Length of time of the first epizootic (e.g. number of consecutive months 11
when disease incidence is greater than median
Number of rabid raccoons in the first epizootic 11
Length of time of the first epizootic AND the interepizootic period 17
before the start of the second epizootic
Length of time an animal carriers the disease from infection until 1
they are symptomatic
Length of time an animal is symptomatic and can infect another individual 1
Measure of animals natural immunity to a disease 30
Length of time from initial outbreak to the time of maximum disease incidence 63
Prevalence of disease among a group of animals (e.g. percentage, count) 1
Distance that an animal disperses 11
Time of year an animal disperses from its mother 4
Proportion and/or type of habitat which the animal chooses to occupy 12
Area of the activity space of an animal 106
The area of the most highly used portion of an animal's home range 13
(e.g. daily activity space)
Area that animals' core home ranges overlap 8
Area that animals' home ranges overlap 43
Length of time an animal lives 2
Amount of mortality (e.g. count, probability) attributable to documented 11
factors, as reported in the “comments” field
Rate of movement of animals 15
Number of animals within a population 15
Time of year when mothers are giving birth 6
Time of year when animals are mating 2
Chance of a raccoon giving birth 12
Length of time spent courting and mating with another animal 4
Number of partners an individual has mated with during a breeding season 6
The percentage of consortships that resulted in the production of offspring 15
Offspring/individual productivity of an animal 5
Length of the gestation period in days 1
Number of living offspring in a litter 41
A numerical value indicating the type of mating system: 1 = polygynous 2
(single male mates with multiple females), 2 = promiscuous
(both males and females have multiple mates)
Length of time a female is in oestrus 3
Time of year a female is in oestrus 1
Percentage of females that have given birth more than once relative to 24
all the sampled females
Number of placental scars 12
Road kills per kilometre of road (can be used as an index for density) 23
Ratio of males to females (or indicate if otherwise) 3
Measure of site fidelity 9
Description of species range recorded in “comments” field 1
Survival of animals (e.g. proportion) for a defined period of time 177
 E CO L O G I CA L IN F O R MA TI CS 3 ( 2 0 08 ) 87–9 6
consequently the need for ecologists to systematically preserve
and make their data available has become increasingly impor-
tant and feasible (Michener, 2006; Michener et al., 1997).
The multifaceted value of ecological data warehouses is
demonstrated through the creation of a raccoon (Procyon lotor)
ecology database (REDB). The REDB was designed to store and
manage traits about raccoon biology, ecology and rabies
collected from available peer-reviewed published literature
and unpublished “grey literature”, and includes citations
referencing the data sources. Traits were collected primarily
from ecological and biological field and laboratory studies
undertaken in North America. Examples of the collected data
include: mortality rates, litter size, home range size and
raccoon rabies incubation period. Meta-data, such as sample
size, a measure of the variance for each demographic and
disease trait, sex and age class, time of year and geographic
coordinates of the study area, also were collected.
The motivation for creating the REDB was to further develop
the Ontario Rabies Model (ORM). The ORM is an individual-based
spatially-explicit disease-host simulation model currently con-
figured for raccoons (Tinline et al., 2007). It simulates raccoon
population dynamics, raccoon rabies viral transmission, and
rabies control strategies. The ORM was built using the expertise
of field biologists who have studied Ontario raccoons. To
increase credibility in model design and results beyond the
builders it is necessary to assess and document sources of model
design as extensively as possible (Bart, 1995; Conroy et al., 1995).
Creation of the REDB was largely motivated to provide an
efficient source of information for checking model parameter-
isation (the appropriateness of ORM parameter input values and
model structure), and for determining biologically reasonable
ranges of input values for model sensitivity testing. Assessing
parameterisation and sensitivity testing are critical to proper
model development (Jackson et al., 2000). A meta-analysis using
Fig. 1 – Relational database design between the five tables of the REDB.
89
REDB data is also demonstrated as an additional contribution to
model parameterisation. REDB data obtained from studies
spanning across the North American species range enables an
exploration of parameterisation in different geographic regions.
Furthermore, specific aspects of model structure can be
assessed, such as, males and females having equal home
range sizes and males dispersing greater distances than females.
In this paper the REDB data model is presented and the utility
of the REDB for model development is revealed. Beyond ORM
development, the REDB is of value to wildlife biologists,
ecologists and infectious disease researchers whose studies
pertain to raccoons and raccoon rabies. Also, managers of
raccoon populations and/or raccoon rabies disease control
programs would benefit by having access to a wide range of
data from many North American locations spanning decades to
enable more informed decision-making. Similar studies have
created databases for other organisms that are also available for
research (e.g. Onstad, 2007; Gachet et al., 2005). The REDB is a
useful research tool that will increase in value with on-going
inclusion of data from future raccoon and raccoon rabies studies.
2. Methods
ISI Web of Science® and Google Scholar™ search engines were
used to identify literature containing information about
raccoons and raccoon rabies. Examples of key words used in
searches are: a) {raccoon OR P. lotor} AND density, b) {raccoon
OR P. lotor} AND {mortality OR survival}. Bibliographies and
reference lists from books and review articles, respectively,
provided an additional method for identifying primary
literature. The literature search included unpublished mate-
rial (grey literature). Full citations of all literature sources were
managed in ProCite version 5.0 (www.procite.com) and also
90 E CO L O G I CA L IN F O RM A TI CS 3 ( 2 0 08 ) 87–9 6

stored for the convenience of data users in the REDB. In the tion (pdf). Pdf's defining model behaviours (e.g. mortality
REDB a study attribute is used to flag grey literature so that it is rates, dispersal distances, density, litter size, rabies incubation
distinguishable from published material. As many publica- period) were derived from field, laboratory and modelling
tions as were accessible from university and Ontario Ministry studies undertaken in Ontario by the Ontario Ministry of
of Natural Resources (OMNR) provincial government libraries Natural Resources and their collaborators (e.g. Totton et al.,
were reviewed. Statistics were collected from primary litera- 2004; Tinline et al., 2002; Totton et al., 2002). In the ORM,
ture for 56 different traits (Table 1). Traits were chosen based
on their importance in defining raccoon biology, ecology and
raccoon rabies disease dynamics. Traits commonly reported Table 2 – Names and descriptions of the lookup (LUP) and
in the reviewed studies were also included. For every trait data tables found in the REDB
value entered into the database, ancillary data were collected Table name Description
pertaining to the value (e.g. sample size, variance, sex class)
LUP_age_class Age class categories
and study site (e.g. location coordinates, season; See Fig. 1.
LUP_captivity_status Captivity status categories
tables [T3_Trait_Values] and [T2_Study_Area] for a complete
LUP_country Countries where studies were undertaken
list). A “comments” field in table [T3_Trait_Values] was used to LUP_estimate_source Type of study undertaken (e.g. field, lab)
record information specific to each trait value. For example, LUP_landscape Types of landscape where studies were
for the “movement_rate” trait, the comments field was used to undertaken
record if telemetry data were acquired during the hunting LUP_latitude “North” and “south” hemispheres for
season, because hunting may affect raccoon movement. The study location
LUP_longitude “East” and “west” hemispheres for study
“comments” field was also used to store verbal definitions of
location
trait values, (e.g. providing the following detail about the
LUP_method Methods used to collect data (e.g. radio
“species range” trait: “Assiniboine and Red River valleys limit collars, mark recapture)
the northern range of raccoons in Manitoba”), though, the LUP_month_season Month or season when the data were
search was focused on obtaining quantitative data. collected
Five people reviewed and entered information from docu- LUP_organism Type of organism to which the trait
ments into the REDB. To check for bias and consistency in pertains
LUP_place Types of administrative units
reader interpretations, all readers initially reviewed and
LUP_season Periods of time as the seasons and an
entered data from the same 10 documents and the results overall annual period
were compared. Readers also entered study site and trait LUP_sex Gender codes (e.g. male, female, both,
information using data input forms built into the REDB to unknown, N/A)
ensure consistency and minimise data entry errors. Com- LUP_state Provinces in Canada and states in the
monly entered information was selected from predefined lists USA
LUP_trait_name Trait names for data values (See Table 1
stored in the database in lookup tables. For example,
for list)
“hectares”, “kilometres^2”, or “miles^2” could be selected
LUP_units Common units pertaining to
from the units field. measurement data (e.g. “hectares” for
area of home range)
2.1. REDB data model LUP_USA_counties All counties by state in the USA
LUP_year Years from 1930 to 2010 from which
The data were organized and stored in a relational database studies have been (and will be)
undertaken
using Microsoft Access 2000®. Five tables store data and are
REDB_VERSION
related to each other using one-to-many relationships (Fig. 1). REFERENCE_Julian_dates User reference table containing calendar
“T” prefixes the names of these tables. Lookup tables are days of the year and the Julian date
prefixed by “LUP”, and are linked to commonly entered fields equivalent
in the data tables. One remaining table serves as a reference T0_Reader_contact_info Data table containing contact
for users needing to convert calendar dates into Julian days information for people reading literature
[REFERENCE_Julian_dates], as was required for some of the and entering data into the REDB
T1_Authorship Data table containing full citation
temporal traits (e.g. birth period, dispersal period; Table 2).
information for every study found
during the literature search, and for
2.2. Ontario rabies model (ORM) those subsequent studies that have trait
values stored in the REDB. This table
The ORM is an individual-based spatially-explicit disease-host also stores the names of the “readers”
simulation model (For details see: Tinline et al., 2007). In this who reviewed the studies and entered
the data into the REDB
model, raccoon rabies dynamics operate on a lattice of
T2_Study_Area Data table containing information about
hexagonal cells, each with a radius of 2 km (from cell centre
the study area (e.g. geographic, time at
to each vertex of the hexagon), the approximate activity range which the study was undertaken)
of raccoons in Ontario (Rosatte, 2000). Model processes operate T2b_Place Administrative unit and name where
at one-week intervals and are stochastically determined. For study occurred
example, each year in week 18, juvenile (52–74 weeks old) and T3_Trait_Values The main data table of the database that
adult females (≥75 weeks old) have a chance of producing a contains the biological and/or ecological
information published in the study
number of offspring as determined by a probability distribu-
 E CO L O G I CA L IN F O R MA TI CS 3 ( 2 0 08 ) 87–9 6 91

default parameter values are chosen to simulate disease
spread in rural raccoons.
2.3. ORM parameterisation
To assess whether ORM default parameter input values fell within
the known range of raccoon and rabies ecology, rural studies were
queried from the REDB, which were then further defined to
retrieve trait information equating to ORM parameters. For ex-
ample, queries were defined to retrieve data for sex-specific home
ranges and annual mortality rates (as opposed to seasonal esti-
mates); ranges of the trait values were then compared with ORM
input values.
2.4. Meta-analysis
Meta-analyses were performed on data for home range size,
density, and litter size. A trait value from a study was included
in the analysis if the study provided a mean, measure of
variance and sample size for the required trait. This rule was
not followed for the meta-analysis of litter size, as explained
below. In general, studies for home range meta-analysis
yielded two estimates, one for each of the sexes, while studies
for density and litter size meta-analyses yielded one estimate
of the required trait. Data for trait values were converted to
common units (e.g. raccoons/km2 for density, and hectares for
home range).
A meta-analysis, analogous to a one-way analysis of
variance (ANOVA; Lipsey and Wilson, 2001), was used to
assess the effect of sex on home range size. QB was used as the
test statistic since it accounts for the portion of variation in the
dependent variable, home range size, which is explained by
the categorical variable, sex. A weighted least squares regres-
sion model was used to test for the relationship between
latitude and density. Both analyses used a mixed-effects
model. This accounted for variation resulting from known
factors within studies (e.g. sex, latitude, season), while still
accounting for the existence of factors causing variation
among studies that are not being explicitly modelled (e.g.
different sampling protocols, unique study area characteris-
tics, unknown random factors; Lipsey and Wilson, 2001).
Mixed-effect models are also among the most “appropriate,
powerful, and informative for the analysis of weighted meta-
analysis data” (Gurevitch and Hedges, 1999).

Table 3 – ORM trait default values from Ontario field data and range of values found for rural raccoons in the REDB
Parameter ORM default REDB range
2
Annual rural raccoon density 5 raccoons/km 4.6–13.6 raccoons/km2
Age of independence 20 weeks 20–36 weeks
Year 0 male (female) mortality rate 0.5 Sub-adult: 0.4–0.51
Year 1 male (female) mortality rate 0.4 Adult: 0.3–0.9
Year 2 male (female) mortality rate 0.3
Year 3 male (female) mortality rate 0.3
Year 4 male (female) mortality rate 0.3
Year 5 male (female) mortality rate 0.6
Year 6 male (female) mortality rate 0.6
Year 7 male (female) mortality rate 0.6

Birth week Calendar week 18 (end of Mid-end April to end of May

April/beginning of May)
Juvenile birth rate 60% 29–66%
Adult birth rate 95% 73–100%
Average litter size 4 2–5
Litter size variance 1 1
Sex ratio at birth 50:50 male:female 50:50 male:female
Male juvenile/adult movement weeks Calendar weeks: 8–43 No movement during severe winters
(spring, summer, autumn)
Female juvenile/adult movement weeks Calendar weeks: 12–17, 38–43 No movement during severe winters
(spring, autumn)
Male young-of-year movement weeks calendar weeks: 38–43 (autumn) No movement during severe winters;
disperse July to November
Female young-of-year movement weeks Calendar weeks: 38–43 (autumn) No movement during severe winters;
disperse July to November
Juvenile/adult male dispersal distance a 1.36 km 1.2–10.2 km
Juvenile/adult female dispersal distance a 0.68 km 0.3–8.5 km
Young-of-year male dispersal distance a 2.12 km 0.2–16.1 km
Young-of-year female dispersal distance a 1.02 km 0.4–6.3 km
Incubation period a 5.47 weeks 1–6 weeks
Infectious period a 1.00 week 1 week
Chance of spread 1.5% –
Contact within cell 77.8% –

Unlisted REDB range values were not definable using data from the REDB.
a
Mean of ORM values derived from a probability distribution function.
92 E CO L O G I CA L IN F O RM A TI CS 3 ( 2 0 08 ) 87–9 6

Effect size (ES) and inverse variance weight (w) variables meterised models (McKay et al., 1979). LHS draws parameter
were calculated to standardise the results among studies so values without replacement and with equal probability over
that they could objectively be compared. These variables were the entire parameter space of its probability distribution
calculated as outlined by Lipsey and Wilson (2001) for function (pdf). The pdf defines a distribution of mean trait
arithmetic means: values. By using an empirical pdf, the shape of the distribution
P is reflective of the data and not constrained to specific
P xi distributions (e.g. Gaussian), which enables more realistic
ES ¼ X ¼ ð1Þ
n parameterisation. Some ORM parameter input values are
1 defined by the mean and variance of a pdf (e.g. mean litter
w¼ ð2Þ size and litter size variance). For other ORM parameters, the
v
user must convert pdf data into the probability of obtaining
where xi is an individual observation within a study, from i = 1 to discrete values. More specifically, input values for these
n, n is the sample size of the observations and v is the variance of parameters constitute a probability mass function (pmf; e.g.
the observations. The weighted least squares mixed-effects the incubation period parameter being defined as the prob-
regression models were performed in SPSS for Win/Version 11.0 ability of incubating rabies 1, 2 or more weeks).
as specified from macros provided by Lipsey and Wilson (2001) Thus, it was necessary to obtain as much information as
and available at http://mason.gmu.edu/∼dwilsonb/ma.html. possible for each parameter to define their pdf, and the REDB
A formal meta-analysis was not possible with the litter size was an efficient means of storing, managing and querying
data because too few studies reported information on varia- data for deriving pdf's for sensitivity analysis. In this paper,
tion of litter size; thus, w could not be calculated. Insufficient example pdf's are given for home range size by sex and litter
data is common in meta-analysis, but does not always size. Parameter pdf's were fitted using Palisade Corporation's
preclude analysis from the raw data, since it still can be @RISK software, version 4.5.4 (www.palisade.com).
informative (Gurevitch and Hedges, 1999). Therefore, correla-
tion analyses were used to test the relationship between
degrees of latitude and litter size. 3. Results

2.5. Sensitivity analysis
A sensitivity analysis procedure was developed to test model
behaviour for demographic and disease parameters charac-
terising raccoon rabies dynamics in the ORM (Rees, 2008).
Parameter input values were defined using Latin hypercube
sampling (LHS) because it is substantially more efficient than
using a fully factorial design when assessing highly para-
The literature search of raccoon biology, ecology and raccoon
rabies produced a list of 838 documents. From these, over 200
articles and books were obtained. Most documents were
primary literature. Readers reviewing literature were 100%
consistent and accurate in their input of data from an initial
review of the same 10 studies. Overall, 160 studies contained
quantitative trait data that were entered into the REDB, and
another 5 studies contributed qualitative data. A total of 993

Fig. 2 – Scatterplot showing the correlation between the mean number of juveniles per litter and degrees latitude, and the
associated 95% confidence bands.
 E CO L O G I CA L IN F O R MA TI CS 3 ( 2 0 08 ) 87–9 6 93

Fig. 3 – Probability distribution function of male and female home range sizes.

traits were entered into the REDB. The majority of data
pertained to density, litter size, survival, home range, dis-
persal period and incubation period (Table 1).
Most ORM parameters had equivalent traits in the REDB. Of
these parameters, the default values were found to lie within
the variation for rural raccoons across North America (Table 3).
Meta-analysis indicated that males had significantly larger
home ranges (HR) than females (HRmale = 352.1 ha ± SE 52.5,
HRfemale =128.1 ha±SE 52.6; QB (one tailed, α = 0.05) =9.1, p-value=
0.003). There was a significant negative relationship of latitude
and raccoon densities (density = 584.2–12.3latitude, R2 = 0.11,
p=0.04). Meta-analysis of litter size revealed a significant positive
correlation with latitude (r=0.38, p=0.032, n=32). The correlation
was strongest when studies were restricted to those occurring in
rural landscapes and pertaining to adult reproduction, omitting
juvenile mothers (r=0.74, pb 0.001, n=18); a second order poly-
nomial model fits most closely to the data compared to linear,
power or exponential models (Fig. 2).
The pdf's for male and female home range sizes illustrate
that males have significantly larger home ranges than
females, as determined by meta-analysis of raccoon home-
range size (Fig. 3). The pdf of litter size indicates a range

Fig. 4 – Probability distribution function of litter size reported in 32 studies.

94 E CO L O G I CA L IN F O RM A TI CS 3 ( 2 0 08 ) 87–9 6
between 1 and 5 juveniles per litter, with a mean of 3.5
animals, for the reviewed North American studies (Fig. 4).
4. Discussion
The relational design of the REDB follows the well-established
relational data model to store traits from available literature,
and enabled subsequent flexible management and manipula-
tion of these data. The REDB provided an efficient source of data
for assessing the appropriateness of ORM parameter values and
model structure. For example, it was easy to query information
about raccoon densities in rural areas to compare with the ORM
rural density parameter value; and, obtain annual mortality
rates as opposed to season-specific values to assess the annual
mortality parameter values in the model. At the conclusion of
the exercise, all ORM parameter default values were found to fall
within the range of data queried from the REDB that were
equivalent to ORM parameters. A further advantage of using the
REDB is that queried information provides additional sources to
support the choice of ORM parameter values.
Meta-analysis is also demonstrated as another method
contributing to ORM parameterisation. For instance, model
structure defines male and female home range size to be equal,
though, meta-analysis results indicate male home ranges are
larger. This is not unexpected since raccoons are mostly
solitary carnivores (Sanderson, 1987; Kaufmann, 1982) that
demonstrate sexual differences in behavioural strategies for
habitat use and breeding success (Kamler and Gipson, 2003;
Gehrt and Fritzell, 1997; Fritzell, 1978). Male raccoons tend to be
larger than female raccoons (Zeveloff, 2002), and a positive
correlation between body size and home range size has been
reported in other species (Anderson et al., 2005; Grigione et al.,
2002), which is likely due to larger energetic requirements
needed to support a greater body mass (White et al., 2007). The
ORM has the capability of modelling sex-specific home range
sizes; however, a sensitivity analysis is necessary to determine
if sex-specific home ranges produce more accurate model
outcomes beyond the cost of increasing outcome uncertainty
resulting from using two estimates of home range size instead
of one.
The meta-analysis demonstration also highlights the
benefit of the REDB as a data source for parameterising the
ORM to geographic regions beyond southern Ontario. For
instance, meta-analysis of litter size against latitude clearly
demonstrates a trend of larger litter sizes at higher latitudes
(Fig. 2). Positive relationships between litter size and latitude
have been reported in many species (e.g. Bilenca et al., 1994;
Cockburn et al., 1983; Innes, 1978; Conaway et al., 1974). A
possible casual mechanism is attributed to higher latitudes
having more extreme environments, causing increased mor-
tality risk due to seasonal variation in food, hence, a shorter
breeding period. To increase fitness, animals invest more per
reproductive event by having one large litter, as opposed to two
or three smaller litters produced over a longer period of time in
more favourably climates of lower latitudes (Rademaker and
Cerqueira, 2006; Zeveloff, 2002).
Latitude had a significant negative correlation with density.
This result may be a consequence of the geographic sampling
distribution of the studies. Species are expected to have higher
densities at the core of their range (Williams et al., 2003;
Hengeveld and Haeck, 1982), because of habitat limitation at
range peripheries (Erb and Boyce, 1999). The raccoon species
range extends as far south as Central America (Zeveloff,
2002). Meta-analysis for this study did not include studies
further south than the United States. Thus, it is possible that
the negative correlation is caused by sampling raccoons near
the core of their range in the United States, and then north-
wards to studies sampled at the periphery of their range in
Canada.
An important benefit of the REDB is providing data for
meta-analyses. Science is advanced by comparing new ideas
to old ideas (Gates, 2002; Arnqvist and Wooster, 1995), and in
the case of the REDB, the analysis of raccoon biology, ecology
and raccoon rabies is advanced by enabling comparisons
among different landscapes, seasons, age classes, sexes and
geographic locations. Results from a single study may seem
unimportant, but may appear more meaningful when as-
sessed in context of other studies.
The REDB was also a valuable data source to define bio-
logically appropriate parameter values for sensitivity testing
the ORM. Biologically-defined values are necessary for deter-
mining processes that most significantly affect model out-
comes (Ginot et al., 2006). This also means that the effect of
traits on model outcomes can be interpreted in a more
biologically meaningful context. This is more appropriate for
a primarily mechanistic biological model, such as the ORM,
than halving or doubling trait inputs as described in Voigt et al.
(1985), because even a trait of minor importance can impact
model results given a large enough variation in input values.
Doubling or halving input values has the potential to create an
unrealistic scenario for the natural system being modelled,
and might yield results that are less informative in terms
of understanding the true biological effect of the trait being
examined.
There are several challenges to creating and using the REDB
and ecological data warehouses in general. The first is obtaining
studies to contribute to the database. Over 800 raccoon and
raccoon rabies documents were identified, but approximately
only 200 documents were obtained. ISI Web of Science® and
Google Scholar™ search engines were effective methods for
identifying literature. It was easy to run many different queries
using various key words. Furthermore, bibliographies and
reference lists from review books and articles, respectively,
provided a means of identifying primary literature not de-
tected using search engines. Unfortunately, journal articles
dating before 1995 are often unavailable online from publishers.
University, government and private libraries must then be used
to obtain hard copies of older literature, however, this requires
considerably more time and effort.
Publication bias and “the file drawer problem” (Bauchau,
1997) are also problematic for using data from ecological data-
bases that warehouse data from multiple studies (Arnqvist and
Wooster, 1995). These phenomena refer to the greater likelihood
of publishing studies reporting significant results (Kotiaho, 2002;
Gurevitch and Hedges, 1999). The strategy for countering
publication bias is to include grey literature in analysis (Roberts
et al., 2006; Burdett et al., 2003; McAuley et al., 2000). Unfor-
tunately, grey literature is difficult to identify and obtain
because, being unpublished, it is less likely to be referenced. In
 E CO L O G I CA L IN F O R MA TI CS 3 ( 2 0 08 ) 87–9 6
the health care industry, efforts are being made to make grey
literature more accessible (e.g. publishing abstracts of non-
significant results, grey literature citation databases; Benzies
et al., 2006; Conn et al., 2003); however, without these resources,
obtaining grey literature is dependent on receiving permission
from researchers to use their unpublished data. Furthermore,
quality of data from grey literature is uncertain if the study has
not undergone or has not passed a scientific peer review (Conn
et al., 2003).
Locating grey literature for the REDB was an onerous task.
Of the 165 studies entered into the database, only 11 were from
unpublished studies. However, data quality of the REDB likely
benefits from not including grey literature. This is because the
REDB is designed to contain descriptive data and not results
from studies testing specific hypotheses. Hence, a peer-re-
viewed process of descriptive studies more likely excludes
lower quality studies than biasing REDB information. Data
warehouse users should still be diligent since poor quality
studies degrade the value of these data. These studies should
be identified through objective measures and eliminated from
potential analyses. Roberts et al. (2006) recommend “assessing
the experimental design, implementation and analysis” of
each study being included in the meta-analysis. It was difficult
to rigorously assess every study, given the high number
included in the database; however, a field was included in the
database to “comment” on any cautions or peculiarities of a
study.
The research question or interpretation of data can
also bias selection of data for populating the REDB. The
REDB is largely free from this bias because all available data
were collected without filtering to predetermined ecological
questions. Several readers collected and entered data,
further reducing the potential bias of reader-misinterpreta-
tion. Additionally, an initial review of the same 10 studies
by the data entry readers was important to ensure high
quality and consistent data collection and to enhance the
database design. This exercise demonstrated that readers
were similarly thorough and accurate in their data input.
The collective review of sample articles also precipitated
discussion to adjust the design of the database to be more
efficient and helped formulate more useful data entry
forms.
5. Conclusions
The availability of the REDB and other similar databases
contribute greatly to the exhaustive effort required to locate,
store, manage and query data needed for model development
and meta-analyses. Results from ecological models and meta-
analyses are frequently used by conservation and wildlife
managers, biologists and ecologists (Sterner and Smith, 2006;
Pullin et al., 2004; Oreskes, 2003). Ecology benefits from
databases like REDB by increasing the amount of data
accessible for comparative and broad-scale analyses and
thus, increasing the value of science's investment in past
studies. The REDB itself offers a robust and simple data model
for compilation, maintenance and analysis of ecological traits
from published studies. The database, and an empty database
for use with other species, comprised of empty data tables,
95
lookup tables and their relationships, are available online
(http://redb.nrdpfc.ca).
Acknowledgements
We are grateful to Rick Rosatte from the Ontario Ministry of
Natural Resources for his knowledge of raccoon ecology and
access to his extensive reprint library. We also greatly appre-
ciated an insightful review from Rowland Tinline, professor
emeritus, Queen's University, Canada. The database greatly
benefited from the careful and meticulous data entry by Denis
Carr, to whom we are indebted. This research has been sup-
ported by a Strategic Grant from the Natural Sciences and
Engineering Research Council of Canada to Bradley White
of the Natural Resources DNA Profiling and Forensics Centre,
Trent University and by a cooperative research agreement be-
tween the Ontario Ministry of Natural Resources and Queen's
University.
REFERENCES
Anderson, D.P., Forester, J.D., Turner, M.G., Frair, J.L., Merrill, E.H.,
Fortin, D., Mao, J.S., Boyce, M.S., 2005. Factors influencing
female home range sizes in elk (Cervus elaphus) in North
American landscapes. Landscape Ecology 20, 257–271.
Arnqvist, G., Wooster, D., 1995. Meta-analysis: synthesizing
research findings in ecology and evolution. Trends in Ecology
and Evolution 10, 236–240.
Bart, J., 1995. Acceptance criteria for using individual-based
models to make management decisions. Ecological
Applications 5, 411–420.
Bauchau, V., 1997. Is there a “file drawer problem” in biological
research? Oikos 79, 407–409.
Benzies, K.M., Premji, S., Hayden, K.A., Serrett, K., 2006.
State-of-the-evidence reviews: advantages and challenges of
including grey literature. Worldviews on Evidence-Based
Nursing, Second Quarter, pp. 55–61.
Bilenca, D.N., Zuleta, G.A., Kravetz, F.O., 1994. Latitudinal
variation in litter size and length of the breeding season in
populations of pampas mice, Akodon azarae. Mammalia 58,
563–568.
Burdett, S., Stewart, L.A., Tierney, J.F., 2003. Publication bias and
meta-analyses. International Journal of Technology
Assessment in Health Care 19, 129–134.
Cockburn, A., Lee, A.K., Martin, R.W., 1983. Macrogeographic
variation in litter size in Antechinus (Marsupalia: Dasyuridae).
Evolution 37, 86–95.
Conaway, C.H., Sadler, K.C., Hazelwood, D.H., 1974. Geographic
variation in litter size and onset of breeding in cottontails.
Journal of Wildlife Management 38, 473–481.
Conn, V.S., Valentine, J.C., Cooper, H.M., Rantz, M.J., 2003. Grey
literature in meta-analyses. Nursing Research 52, 256–261.
Conroy, M.J., Cohen, Y., James, F.C., Matsinos, Y.G., Maurer, B.A.,
1995. Parameter estimation, reliability, and model
improvement for spatially explicit models of animal
populations. Ecological Applications 5, 17–19.
Erb, J.D., Boyce, M.S., 1999. Distribution of population declines in
large mammals. Conservation Biology 13, 199–201.
Fritzell, E.K., 1978. Aspects of raccoon (Procyon lotor) social
organization. Canadian Journal of Zoology 56, 260–271.
Gachet, S., Vela, E., Tatoni, T., 2005. BASECO: a floristic and
ecological database of Mediterranean French flora. Biodiversity
and Conservation 14, 1023–1034.
96 E CO L O G I CA L IN F O RM A TI CS 3 ( 2 0 08 ) 87–9 6
Gates, S., 2002. Review methodology of quantitative reviews using
meta-analysis in ecology. Journal of Animal Ecology 71,
547–557.
Gehrt, S.D., Fritzell, E.K., 1997. Sexual differences in home ranges
of raccoons. Journal of Mammalogy 78, 921–931.
Ginot, V., Gaba, S., Beaudouin, R., Aries, F., Monod, H., 2006.
Combined use of local and ANOVA-based global sensitivity
analyses for the investigation of a stochastic dynamic model:
application to the case study of an individual-based model of a
fish population. Ecological Modelling 193, 479–491.
Grigione, M.M., Beier, P., Hopkins, R.A., Neal, D., Padley, W.D.,
Schonewald, C.M., Johnson, M.L., 2002. Ecological and
allometric determinants of home-range size for mountain
lions (Puma concolor). Animal Conservation 5, 317–324.
Gurevitch, J., Hedges, L.V., 1999. Statistical issues in ecological
meta-analyses. Ecology 80, 1142–1149.
Hengeveld, R., Haeck, J., 1982. The distribution of abundance. I.
Measurements. Journal of Biogeography 9, 303–316.
Innes, D.G.L., 1978. A reexamination of litter size in some North
American microtines. Canadian Journal of Zoology 56,
1488–1496.
Jackson, L.J., Trebitz, A.S., Cottingham, K.L., 2000. An introduction
to the practice of ecological modelling. Bioscience 50, 694–706.
Jones, M.B., Schildhauer, M.P., Reichmann, O.J., Bowers, S., 2006. The
new bioinformatics: integrating ecological data from the gene to
the biosphere. Annual Reviews of Ecological Systems 37, 519–544.
Kamler, J.F., Gipson, P.S., 2003. Space and habitat use by male
and female raccoons, Procyon lotor, in Kansas. Canadian
Field-Naturalist 117, 218–223.
Kaufmann, J.H., 1982. Raccoon and allies. In: Chapman, J.A.,
Felhamer, G.A. (Eds.), Wild mammals of North America:
Biology, management, and economics. John Hopkins
University Press, Baltimore, pp. 567–585.
Kotiaho, J.S., 2002. Meta-analysis, can it ever fail? Oikos 96,
551–553.
Lipsey, M.W., Wilson, D.B., 2001. Practical Meta-Analysis. Sage,
Thousand Oaks, California. 247 pp.
McAuley, L., Pham, B., Tugwell, P., Moher, D., 2000. Does the inclusion
of grey literature influence estimates of intervention effectiveness
reported in meta-analysis? The Lancet 356, 1228–1231.
McKay, M.D., Conover, W.J., Beckman, R.J., 1979. The epidemiology and
transmission dynamics of HIV/AIDS. Technometrics 21, 239–245.
Michener, W.K., 2006. Meta-information concepts for ecological
data management. Ecological Informatics 1, 3–7.
Michener, W.K., Brunt, J.S., Helly, J., Kirchner, T.B., Stafford, S.,
1997. Non-geospatial metadata for the ecological sciences.
Ecological Applications 7, 330–342.
Onstad, D.W., 2007. EDWIP: Ecological Database of the World's
Insect Pathogens. Illinois Natural History Survey, Champaign,
Illinois. http://insectweb.inhs.uiuc.edu/Pathogens/EDWIP.
Oreskes, N., 2003. The role of quantitative models in science. In:
Canham, C.D., Cole, J.J., Laurenroth, W.K. (Eds.), Models of
Ecosystems. Princeton Press, Princeton N.J., pp. 13–31.
Pullin, A.S., Knight, T.M., Stone, D.A., Charman, K., 2004. Do
conservation managers use scientific evidence to support
their decision-making? Biological Conservation 119,
245–252.
Rademaker, V., Cerqueira, R., 2006. Variation in the latitudinal
reproductive patterns of the genus Didelphis (Didelphimorphia:
Didelphidae). Austral Ecology 31, 337–342.
Rees, E.E., 2008. Approaches to modelling raccoon rabies.
Peterborough, Canada: Trent University (PhD thesis).
Roberts, P.D., Stewart, G.B., Pullin, A.S., 2006. Are review articles a
reliable source of evidence to support conservation and
environmental management? A comparison with medicine.
Biological Conservation 132, 406–423.
Rosatte, R.C., 2000. Management of raccoons (Procyon lotor) in
Ontario, Canada: do human intervention and disease have
significant impact on raccoon populations? Mammalia 64,
369–390.
Sanderson, G.C., 1987. Raccoon. Ministry of Natural Resources,
Ontario, Canada.
Sterner, R.T., Smith, G.C., 2006. Modelling wildlife rabies:
transmission, economics, and conservation. Biological
Conservation 131, 163–179.
Tinline, R.R., Rosatte, R.C., MacInnes, C.D., 2002. Estimating the
incubation period of raccoon rabies: a time-space clustering
approach. Preventative Veterinary Medicine 56, 89–103.
Tinline, R.R., Ball, D., Broadfoot, J., 2007. The Ontario rabies model.
Wildlife Research and Development Section. Ontario Ministry
of Natural Resources, Peterborough, Canada.
Totton, S.C., Tinline, R.R., Rosatte, R.C., Bigler, L.L., 2002. Contact
rates of raccoons (Procyon lotor) at a communal feeding site in
rural Eastern Ontario. Journal of Wildlife Diseases 38, 313–319.
Totton, S.C., Rosatte, R.C., Tinline, R.R., Bigler, L.L., 2004. Seasonal
home ranges of raccoons, Procyon lotor, using a common
feeding site in rural Eastern Ontario: rabies management
implications. Canadian Field-Naturalist 118, 65–71.
Voigt, D.R., Tinline, R.R., Broekhoven, L.H., 1985. A spatial
simulation model for rabies control. In: Bacon, P.J. (Ed.),
Population Dynamics of Rabies Wildlife. Academic Press Inc.
Ltd., London, pp. 311–349.
White, E.P., Ernest, S.K.M., Kerkhoff, A.J., Enquist, B.J., 2007.
Relationships between body size and abundance in ecology.
TRENDS in Ecology and Evolution 22, 323–330.
Williams, C.K., Ives, A.R., Applegate, R.D., 2003. Population
dynamics across geographical ranges: time-series analysis of
three small game species. Ecology 84, 2654–2667.
Zeveloff, S.I., 2002. Raccoons: A Natural History. UBC Press,
Vancouver, Canada.