Content uploaded by Mason Maron
Author content
All content in this area was uploaded by Mason Maron on Sep 29, 2024
Content may be subject to copyright.
1
Range Expansion Dynamics of the Lesser Goldfinch (Spinus psaltria) in the Pacific
Northwest
Mason W. Maron1,*, Emma I. Greig2, Jordan Boersma2
1Independent Researcher, PO Box 3546 Bay City, OR
2Cornell Lab of Ornithology, 159 Sapsucker Woods Rd. Ithaca, NY, USA
*Corresponding author: Mason Maron, mmaron101@gmail.com
ABSTRACT
The Lesser Goldfinch (Spinus psaltria) has undergone significant range expansion throughout
the Pacific Northwest in recent decades, expanding further northward through Oregon, Idaho,
and Washington. We used data from Project FeederWatch and eBird Status and Trends to reveal
a 9.3% northward shift in occupancy since 2001, while breeding season populations increased
significantly in Oregon (16.9%), Idaho (66.3%), and Washington (110.5%) from 2012 to 2022.
We found that colonization in the expanded range was characterized by higher mean and
maximum temperatures and associated with proximity to major rivers, particularly in areas of
increased urban development. In contrast, our historical range model indicated that occupancy
was predicted by minimum temperatures and precipitation, though the model's relatively weak
predictive power suggested additional factors influencing site selection. Bird feeders and
invasive plant species likely played a key role in facilitating this expansion, especially in urban
areas along riverine corridors. Additionally, our models showed sites typically remained
occupied once colonized, indicating permanent range shifts rather than temporary nomadic
movements. Our study highlights how S. psaltria's range dynamics are shaped by both
anthropogenic and natural landscape features and reflect broad ecological responses to climate
change and habitat modification, suggesting continued expansion as these factors persist.
Keywords: Lesser Goldfinch, range expansion, Pacific Northwest, colonization, climate change,
occupancy modeling, riverine corridors, urbanization
2
Lay Summary
• The Lesser Goldfinch (Spinus psaltria) is a seed-eating songbird which traditionally lives
throughout the Southwestern U.S. but has expanded its range into the Pacific Northwest
in recent decades.
• We used participatory science data from Project FeederWatch and eBird Status and
Trends to understand what factors are driving this expansion.
• We found that warmer temperatures, especially near major rivers in developed areas, are
critical in supporting the species' expansion.
• Once colonized, areas tend to remain occupied, indicating that the range shift is likely
permanent as groups of birds spread and settle in new locations.
• We identified increased access to food through bird feeders and the increased growth of
weedy plants, which these finches prefer as a food source, as likely facilitators of this
range expansion.
• Our findings show how human activities, such as urbanization and habitat modification,
along with climate change, are reshaping the distribution of bird species in North
America.
3
INTRODUCTION
Bird species globally are facing the consequences of landscape development, introduction of
invasive species, and climate change (Diamond 1989). In the United States, many species have
gone extinct or shifted ranges because of these anthropogenic forces (Auer and King 2014). For
instance, Conuropsis carolinensis (Carolina Parakeet) went extinct by the early 1900s due to
habitat loss, while many species of Hawaiian honeycreeper (subfamily Carduelinae) were pushed
to extinction by habitat degradation and the introduction, both intentionally and unintentionally,
of invasive species to the Hawaiian Islands (Saikku 1990, Atkinson and LaPointe 2009).
Additionally, numerous species across the country have responded to anthropogenic impacts by
shifting their ranges to higher latitudes and elevations (Hitch and Leberg 2007, Sekercioglu et al.
2008). These factors continue to push many species historically found in the Southern and
Eastern U.S. into the Pacific Northwest region (Battey 2019, Hampton 2020, Maron and Borre
2024).
In recent decades, large-scale participatory (citizen) science initiatives have proven
invaluable in documenting these avian range shifts in response to landscape modification and
climate change (Cooper et al. 2014). Datasets such as eBird and Project FeederWatch offer the
geographic, temporal, and taxonomic scale needed to analyze trends across multiple species, a
feat made possible through contributions of citizen scientists (Princé and Zuckerberg 2015, La
Sorte et al. 2017). These datasets can be paired with public data on factors such as climate,
temperature, and land use to gain insight into the specific forces behind range shifts in birds
(Princé and Zuckerberg 2015). Understanding the factors driving these avian range shifts is
crucial, as species moving into new areas can disrupt existing ecosystems, while the departure
from their historical range may indicate significant habitat loss (Princé and Zuckerberg 2015,
4
Holm et al. 2016). Identifying these factors also allows us to develop conservation strategies that
anticipate and mitigate the ongoing human impacts on these ecosystems (La Sorte et al. 2017).
One species undergoing this northern range expansion is Spinus psaltria (Lesser
Goldfinch), a finch species found across the Western United States (Watt and Willoughby 2020).
S. psaltria has been historically found year-round in California, Nevada, and parts of Oregon and
as a summer resident in much of Utah (Peterson 1984, Robbins et al. 1996). The species was first
documented in Oregon as resident in the southern Willamette Valley in the late 1800s, with data
collected throughout the early 20th century showing its presence throughout the valley, though
rare in Oregon’s coast and eastern deserts (Noel and Jewett 1940).
Since then, S. psaltria has spread throughout the Pacific Northwest region of the United
States, defined in this paper as Oregon, Washington, and Idaho (Schwantes 1996). Idaho
documented its first record in 1967 (Stephens et al. 1990). Before 1988, the state had only eight
records, with its first breeding pair that year (Stephens et al. 1990). Similarly, S. psaltria was not
documented in Washington until 1951 and was determined to be a breeding species in 1957 only
as far north as Clark County, just across the Columbia River from Portland (Mattocks et al.
1976).
In recent decades, the range of S. psaltria has expanded further, with the species now
breeding throughout much of Eastern Oregon and maintaining year-round residency in several
large population clusters around Bend, OR, Portland, OR, Boise, ID, Pullman, WA, Moscow, ID,
Lewiston, ID, and Clarkston, WA (eBird Status and Trends 2022). S. psaltria is also found
across much of the Columbia and Snake Rivers and throughout the entire Willamette Valley in
Oregon and is becoming an increasingly common rarity in the Puget Sound region and in parts of
British Columbia (eBird Basic Dataset 2024). This study aims to examine the factors driving the
5
northward range expansion of S. psaltria using data from the participatory science database
Project FeederWatch and pre-compiled Status and Trends data from eBird.
METHODS
Study Area and Data Sources
For this study, we focused on observations from Oregon, Washington, Idaho, California, Nevada,
and Utah. We identified much of Nevada, California, Utah, and parts of Oregon as the
“historical” range of S. psaltria as detailed in the field guide “A Guide to Field Identification:
Birds of North America” (Robbins et al. 1966, Figure 1). We therefore defined the “expanded
range” as all detections of S. psaltria occurring north of the historical range within our selected
states.
Figure 1. Map depicting the historical (yellow) and expanded (green) ranges of the Lesser Goldfinch (Spinus psaltria) in the
study area. The expanded range was estimated using kriging of Project FeederWatch sites with detections of more than one
6
individual (Yalcin & Leroux 2017). Unoccupied areas (gray) represent regions with either no detections or detections consisting
of only single individuals as to exclude vagrancy from the expanded range. S. psaltria illustration provided by Aly Stuart. River
maps sourced from NOAA Fisheries, USGS, the Lincoln Institute of Land Policy, Deschutes County, In Time & Place, and
ESRI. State boundaries map sourced from the U.S. Census Bureau.
We utilized data collected through Project FeederWatch (“PFW”), a long-term
participatory science project run by the Cornell Lab of Ornithology and Birds Canada (Bonter
and Greig 2021). PFW uses volunteer-collected data from the United States and Canada,
recording standardized counts of species observed at count sites from December through March.
By concentrating observations at consistent sites during a designated time of year, it ensures data
is recorded across multiple years at each location, offering a distinct advantage for occupancy
modeling over less consistent datasets. This method of data collection strengthens the modeling
approach compared to the more sporadic and uneven distribution of data often encountered with
eBird across different months and years. Additionally, PFW data is based on standardized
protocols designed to provide consistent data collection and avoid repeated records of the same
individuals within a given count period. The selection of data for this study was processed
following the methodology used in Greig et al. (2017) to ensure consistency with previous range
expansion studies using PFW.
We selected all PFW surveys within the study range from 2001 through the end of 2023.
Since all species detected during each count are reported by the volunteers, we derived S.
psaltria presence and absence from data series through zero-filling. All PFW entries are also
accompanied by a record of time spent surveying (effort) and the date the survey occurred. To
ensure effort could be accounted for, any entries lacking effort values were discarded.
To identify climate characteristics associated with S. psaltria range, we extracted the
following values from the PRISM Climate Group gridded dataset (2024): annual mean
temperature (tmean), annual maximum temperature (tmax), annual minimum temperature (tmin),
7
and total annual rainfall (ppt). We chose these annual traits to characterize the impact of year-
round climate on the landscape, habitat, and availability of resources being selected by S.
psaltria.
We used data from the 2013 National Land Cover Database (NLCD) to assess
urbanization and land use (Multi-Resolution Land Characteristics Consortium 2013). The 2013
dataset was selected because it represents the temporal median of our study period. For each
PFW site, we extracted the proportion of urban land cover within a 1 km circular buffer
surrounding each site in ArcGIS Pro 3.3.0. The urban land cover categories included developed
open space, low intensity, medium intensity, and high intensity (NLCD classes 21, 22, 23, and
24). We also used the 2020 U.S. Census dataset (U.S. Census Bureau 2020) to determine housing
density, defined as the total number of housing units per square kilometer. This measure was
calculated within the same 1 km circular buffer around each observation site.
In addition to these factors, we considered proximity to major rivers as a potential impact
on S. psaltria range. Rivers in the Pacific Northwest often provide more stable habitats with
warmer temperatures, higher urban development density, and greater vegetation diversity
compared to the surrounding landscape (Detling 1958, Graf 1992, Tennant et al. 2015). We
selected the Columbia, Snake, Willamette, Deschutes, Bear, Sacramento, Humboldt, and
Colorado Rivers for this analysis, based on their size, length, and ecological significance in the
region. Distance was calculated using the “Near” tool in ArcGIS Pro 3.3.0.
We used eBird Status and Trends data to illustrate broad breeding season trends of S.
psaltria. This program provides pre-compiled and pre-processed data from eBird on population
trends in range and abundance across different regions and time periods (Fink et al. 2023). eBird
Status and Trends offers additional seasonal coverage and a broader geographical scope, which
8
can supplement and support trends observed through PFW. Additionally, since the data is pre-
analyzed by eBird, statistics can be directly extracted from the Status and Trends dataset and
utilized without requiring further processing.
Data Analysis
To investigate the range expansion of S. psaltria, we used the R packaged “unmarked”, a species
occupancy-modeling framework (Fiske and Chandler 2011). This method is preferred for PFW
data, as it can account for imperfect detection in estimating site occupancy probabilities (Greig et
al. 2017). Following the protocols established by Greig et al. (2017), we applied both single-
season and multi-season occupancy models to explore how environmental factors, such as
urbanization and climate, influence the colonization and extinction dynamics of S. psaltria
populations.
We began by fitting single-season occupancy models to assess how S. psaltria occupancy
over time was influenced by latitude and climate variables. Following the single-season
framework described by Greig et al. (2017), we established three single-season models for the
years 2002 (n = 1201 sites), 2012 (n = 1305 sites), and 2022 (n = 1868 sites) during the winter
months of December through March, when S. psaltria is expected to be sedentary. In these
months, emigration should be minimal (Greig et al 2017). Despite variation in sample sizes
across years as PFW participation has changed, we have no reason to suspect that the sites do not
represent S. psaltria distribution over time. Occupancy modeling effectively handles such
differences in sample size, provided the sample is representative of the region and study period
(Greig et al. 2017). Given the geographic spread of sites included, we consider the sampled
locations to be an accurate reflection of the study area across all years.
9
For each year, we modeled occupancy as a function of latitude, mean annual temperature,
and total annual precipitation. We incorporated survey effort and observation date as detection
covariates to account for variability in observer effort and temporal differences in detectability
across the winter season. For each climate variable, we created separate models for each year,
along with a null model that included only the detection covariates. Precipitation data were log-
transformed to normalize distribution, while temperature variables were left untransformed due
to their normal distribution. Latitude was incorporated as a predictor to assess potential range
expansion of S. psaltria. We used Akaike Information Criterion (AIC) to evaluate the models,
with lower AIC values indicating a better fit. Models with ΔAIC less than 2 were considered to
have similar support. This approach allowed us to determine which environmental factors best
explained occupancy patterns in 2002, 2012, and 2022, and whether climatic conditions
associated with S. psaltria occupancy had shifted over time. To further explore the relationship
between temperature (tmean, tmin, tmax) and S. psaltria occupancy within the expanded range,
we compared the mean values for each variable across all years in our dataset. To assess any
trends in these variables, we performed Welch two-sample t-tests between sites with and without
detections.
To investigate whether the range expansion of S. psaltria was driven by climate,
urbanization, or proximity to rivers as hypothesized, we employed a multi-season occupancy
model to analyze colonization and extinction dynamics between 2001 and 2023. Multi-season
models are well-suited for this analysis as they allow for the estimation of colonization and
extinction rates between seasons, making them ideal for examining changes in occupancy over
an extended period (Greig et al. 2017). We omitted years prior to 2001 due to low sample sizes
and grouped remaining sites into two categories: those within the historical range (n = 5023) and
10
those in the expanded range (n = 3950). We modeled colonization by S. psaltria for each defined
range as a function of the selected site covariates housing density, proportion of urban land
cover, proximity to rivers, annual temperature variables (tmin, tmax, tmean), and annual
precipitation. Covariates that were static (housing density, urban cover, and proximity to rivers)
were not averaged across years, while dynamic variables (temperature and precipitation) were
averaged to account for temporal variation. Pearson correlation tests were conducted to ensure no
multicollinearity between covariates. While most included variables had weak to moderate
correlations (r < 0.5), the selected temperature variables were highly correlated with each other (r
> 0.9). To prevent multicollinearity, we tested models excluding one or more of these
temperature variables. Otherwise, no pair of covariates had a strong enough correlation (r > 0.7)
to warrant exclusion based on collinearity alone (Dormann et al. 2013).
Detection probability was modeled as a function of effort and observation date to account
for variation in observer effort across time, and latitude was included as a covariate influencing
overall occupancy. We compared the explanatory power of each covariate using a backward
stepwise model-selection approach, iteratively removing the least significant variables until all
remaining covariates had p-values below 0.2 (Wang et al. 2007). For each range, we compared
global models with all covariates, univariate models with a single covariate, and models
excluding one covariate at a time. Additionally, we tested models that included interaction terms
between covariates to assess whether combined effects influenced occupancy. A null model,
which included only detection covariates and no site predictors, was used as a baseline
comparison. All models converged and produced reasonable parameter estimates.
To evaluate the predictive power of the models, we calculated the Area Under the
Curve (AUC) statistic for the global models in both the historical and expanded ranges following
11
Greig et al. (2017). AUC values above 0.80 were considered to indicate excellent discriminatory
power, values between 0.70 and 0.80 were deemed acceptable, and values below 0.70 were
interpreted as poor (Zuckerberg et al. 2010).
RESULTS
Single-Season Models
The single-season models of occupancy in 2002, 2012, and 2022 indicated a 9.3% increase in the
latitude of predicted occupancy between 2002 (50% predicted occupancy = 42.23°) and 2022
(50% predicted occupancy = 46.18°) (Figure 2a). In all modeled years, latitude was a strong
negative predictor of occupancy, with steeper declines in occupancy probability at higher
latitudes over time (Table 1). In all years, mean temperature was positively associated with
occupancy probability (Table 1), indicating that warmer temperatures facilitated colonization
across the species' range. Precipitation consistently showed a negative effect on occupancy, with
higher rainfall reducing the probability of colonization in all years (Table 1). In contrast to
latitude, the influence of mean annual temperature and total annual precipitation on occupancy
probability remained relatively consistent over time (Figures 2b and 2c). In the expanded range,
sites with S. psaltria detections had significantly higher mean, minimum, and maximum
temperatures than those without detections (tmean: 11.22°C vs 10.45°C, t = 61.53, p < 0.001;
tmin: 5.21°C vs. 5.15°C, t = 2.61, p = 0.009; tmax: 17.24°C vs 15.74°C, t = 155.56, p < 0.001).
12
Figure 2. Predicted occupancy probability of the Lesser Goldfinch (Spinus psaltria) as a function of (a) latitude (degrees), (b)
mean temperature (°C), and (c) total precipitation (log mm) for 2002 (n = 1201 sites, blue), 2012 (n = 1305 sites, red), and 2022
(n = 1868 sites, yellow). Shaded regions represent 95% confidence intervals for each modeled year.
Table 1. Estimates from single-season models relating occupancy of Lesser Goldfinch (Spinus psaltria) to latitude, mean
temperature, and total precipitation across 2002, 2012, and 2022. The results indicate a northward shift in occupancy over th e 20-
year span, with minimal changes in occupancy by mean temperature and total precipitation. All predictors are considered
significant towards occupancy.
Predictor
Year
Estimate
SE
z
p
Latitude
2002
-0.66
0.05
-14.11
< 0.001
2012
-0.66
0.06
-12.07
< 0.001
2022
-2.88
0.15
-19.03
< 0.001
Mean
Temperature
2002
0.78
0.05
17.48
< 0.001
2012
0.60
0.03
20.41
< 0.001
2022
0.54
0.03
21.29
< 0.001
Total
Precipitation
2002
-4.06
0.36
-11.19
< 0.001
2012
-5.64
0.45
-12.62
< 0.001
2022
-4.22
0.28
-14.86
< 0.001
13
Multi-Season Models
The multi-season models revealed additional drivers of occupancy within the historical and
expanded ranges. In the historical range, the best-supported model included precipitation and an
interaction between precipitation and minimum temperature as factors impacting occupancy.
Precipitation was a strong predictor of colonization (Table 2), as was its interaction with
minimum temperature (Table 2), suggesting that colonization is more likely in areas with a
suitable combination of moderate rainfall and favorable temperature conditions. Minimum
temperature alone was not a significant predictor (Table 2). The Area Under the Curve (AUC)
statistic for the global model in the historical range indicated poor discriminatory power (AUC
mean ± SD = 0.62 ± 0.002), consistent with the limited predictive strength of the model.
In the expanded range, the best model included maximum temperature, proximity to
rivers, and an interaction term between river proximity and urban proportion. While urban
proportion by itself was not a significant predictor (p > 0.05), its interaction with river proximity
(Table 2) indicated that S. psaltria colonization was more likely in urban areas specifically when
they were closer to major rivers. Maximum temperature, river proximity, and their interaction
were also strong predictors of colonization (Table 2), indicating a preference for warmer areas in
closer proximity to major rivers. The global model for the expanded range exhibited excellent
discriminatory power (AUC mean ± SD = 0.84 ± 0.001), suggesting strong model performance.
Detection probability differed between the historical and expanded ranges. In the
historical range, detection probability was significantly influenced by both survey effort and date
(Table 2). Detection probability in the expanded range was significantly influenced by survey
effort but not by date (Table 2), suggesting that the timing of surveys had less of an effect on
detectability in the expanded areas compared to the historical range. Extinction probability in
14
both the historical range (Estimate = -4.18, SE = 0.09, z = -46, p < 0.001) and the expanded
range (Estimate = -3.39, SE = 0.13, z = -25.7, p < 0.001) were low, suggesting that colonized
sites were relatively stable once occupied.
Table 2. Multi-season best model estimates for Lesser Goldfinch (Spinus psaltria) from 2001-2023 in the expanded and historical
ranges. Colonization is provided for all site covariates (“predictors”). Effort and date are included as predictors of detection
probability, and latitude as a predictor of occupancy. Results show the strength and directionality of each predictor; negative
estimates suggest the parameter decreases as the predictor increases, while positive estimates suggest the parameter increases as
the predictor increases. Interaction terms identify compounding predictor effects.
Range
Predictor
Parameter
Estimate
SE
z
p
Expanded
Range
Latitude
occupancy
-1.17
0.09
-12.9
< 0.001
Maximum
Temperature
colonization
1.52
0.18
8.43
< 0.001
River Distance
colonization
-1.22
0.17
-7.16
< 0.001
Maximum
Temperature:River
Proximity (interaction
term)
colonization
1.38
0.19
7.15
< 0.001
Urban
Proportion:River
Distance (interaction
term)
colonization
-0.66
0.12
-5.54
< 0.001
Effort
detection
0.04
0.02
2.32
0.02
Date
detection
0.02
0.02
0.86
0.39
Historical
Range
Latitude
occupancy
-0.53
0.06
-9.04
< 0.001
Precipitation
colonization
-0.42
0.09
-4.49
< 0.001
Minimum Temperature
colonization
0.12
0.1
1.23
0.22
Precipitation:Minimum
Temperature
(interaction term)
colonization
0.41
0.12
3.44
< 0.001
Effort
detection
0.17
0.01
13.34
< 0.001
Date
detection
-0.1
0.01
-8.51
< 0.001
15
eBird Status and Trends Data
The eBird Status and Trends data indicated significant population increases in the breeding
season across several regions in the Pacific Northwest (Fink et al. 2023). In Washington, the
median breeding season population trend was 110.5% (80% CI: 78.1% - 161.9%) from 2012 to
2022. In Oregon, the median trend was 16.9% (80% CI: 10.7% - 22.3%), and in Idaho, it was
66.3% (80% CI: 41.3% - 89.0%). These positive trends contrast with more stable or declining
trends in parts of the species' historical range; in California, the median trend was 0.2% (80% CI:
-3.5% - 2.8%). In Utah, the trend was -6% (80% CI: -8.9% - 1.1%), and in Nevada, the trend was
-2.3% (80% CI: -8.5% - 2.7%) (Fink et al. 2023).
DISCUSSION
Our study sought to document the northward range expansion of S. psaltria into the Pacific
Northwest and investigate which environmental factors are associated with colonization. Our
analyses indicate a significant expansion of S. psaltria range into the Pacific Northwest region in
recent decades, with increasing likelihood of occupancy at higher latitudes over time. We
identified this expansion as being primarily driven by shifting landscape conditions resulting
from a combination of anthropogenic factors and climate suitability. While our results do not
reflect the full extent of warming reported by large-scale climate studies (Heeter et al. 2023), S.
psaltria colonization in its expanded range appears to be facilitated by warmer microclimates,
particularly at sites near major rivers and urban development. This interaction likely reflects a
broader trend of habitat modification, including urbanization and changes in land use, which
reshape the surrounding environment and create favorable environments for S. psaltria, even at
sites which have not experienced significant climatic change in recent decades.
16
In the historical range of S. psaltria, we found that site selection was most significantly
shaped by minimum annual temperatures and total annual precipitation, specifically for warmer,
drier sites. These relationships likely align with the dietary and habitat preferences of the species,
as precipitation levels typically control plant community composition (Silvertown et al. 1994).
The impact of minimum temperature on site occupancy is likely a consequence of migration, as
S. psaltria is known to retreat from the coldest parts of its historical range despite being a
relatively sedentary species otherwise (Watt and Willoughby 2020). The lower AUC score for
our historical range model also suggests that additional factors in this range may be impacting
occupancy in ways not captured by our selected covariates. One possible factor is habitat
degradation, as S. psaltria relies heavily on riparian habitats in arid regions and native
grasslands, which are experiencing ongoing loss due to ranching and development across its
historical range (Watt and Willoughby 2020).
In the expanded range, we found higher S. psaltria occupancy near major rivers,
particularly in areas with urban development. This relationship is likely due to a combination of
thermal stability and increased food availability. Occupancy may be controlled by supplementary
and year-round food provided by bird feeders, as previous studies have suggested that bird
feeders have contributed significantly to the range expansions of several bird species in the
United States (Versaw 2000, Greig et al. 2017, Watt and Willoughby 2020). Numerous
extralimital records of S. psaltria have been associated with feeders, supporting the idea that
feeders have driven northward expansion of this species (Versaw 2000, Johnston 2009, Toochin
2024). The differing impact of survey date on detection probability between the historical and
expanded ranges may suggest differences in local foraging behavior of S. psaltria. In the
historical range, small-scale migration patterns and nomadic foraging behavior may lead S.
17
psaltria to rely more on natural food sources (Greig et al. 2017, Watt and Willoughby 2020). In
contrast, in the expanded range, individuals might rely more consistently on feeders throughout
the winter, when natural food sources are scarcer (Watt and Willoughby 2020). This behavioral
difference could lead to more stable detection probabilities across survey dates despite lower
overall detection rates, likely due to smaller population sizes compared to the historical range.
In our expanded range models, the strength of both the interaction between river
proximity and urban proportion and the strength of river proximity alone align with S. psaltria’s
dietary preference for typically weedy, invasive plants, including thistle, groundsel, mullein, and
dandelions (Linsdale 1957, Watt and Willoughby 2020). These plants, along with other weedy
species S. psaltria feeds on, are prevalent across the Pacific Northwest, especially in disturbed
habitats such as roadsides, waste lots, and other unmanaged open areas (Giblin and Legler 2003,
Watt and Willoughby 2020). Proximity to rivers is a significant factor in the presence of these
food sources, both as a vector for seed dispersal and as a source of soil moisture (Hoffman et al.
2008, Rood et al. 2010). While these sites are often already warmer than surrounding, higher
elevation areas (Tennant et al. 2015), the presence of these weedy plants may be further
benefitted by the urban heat island effect (Zipper et al. 2016, Woudstra et al. 2024). The more
developed sites along major rivers may experience warmer, more stable temperatures throughout
the year, increasing the availability of these food resources, particularly during winter (Taha
1997, Melaas et al. 2016).
Data from eBird Status and Trends (Fink et al. 2023) provides a broader context for
understanding S. psaltria's range expansion. The positive breeding season trends observed in
Washington, Oregon, and Idaho suggest that the species has established increased residency in
these regions, supporting the low extinction probability found in our analysis. Alongside the
18
areas where S. psaltria has become resident throughout the Pacific Northwest, there are several
active frontiers of its ongoing range expansion. Our analysis suggests that the expansion of S.
psaltria into Washington may have occurred through two separate pathways: one following the
Willamette River north through Oregon, and another from Utah through Idaho along the Snake
River. While the populations in northern Oregon have been established and growing for several
decades, Idaho’s stable populations have only formed and established over the past two decades
(Jewett and Gabrielson 1929, Strope et al. 2020). These populations have increased significantly
in density at certain locations along their route and are nearly converging along the Columbia
River in central Washington, where they will likely establish a population soon (Figure 3).
Additionally, vagrants across the region likely disperse directly from these established
populations (Figure 3). Despite the geographically separate origin of these groups, they are
unlikely to be significantly separate on a genetic level (Willoughby 2007). Though not yet
considered established, the species has also become increasingly common in British Columbia
throughout the Vancouver area and the Okanogan and Fraiser Valleys and has begun to breed in
the region (Toochin 2024). Additionally, S. psaltria has been increasing in population throughout
much of Washington’s Puget Sound region (Hampton 2021, eBird Basic Dataset 2024). S.
psaltria has been identified as one of the North American bird species best adapting to climate
change (Neate-Clegg et al. 2024), so it is highly likely its population will continue to grow and
advance throughout suitable habitats across the Pacific Northwest region.
19
Figure 3. The distribution of the Lesser Goldfinch (Spinus psaltria) during the winter (December to March) at Project
FeederWatch count sites (small black dots) over three time periods: 2001–2008 (n = 3141 sites), 2009–2015 (n = 3178 sites), and
2016–2023 (n = 4994 sites). The mean number of birds per count is depicted in shades of gray when greater than 0: light gray
(≤1), medium gray (1–3), and dark gray (>3). Maps were generated using the kriging interpolation tool in ArcGIS 3.3.0.
Ongoing monitoring of S. psaltria will be important to determine the extent of northward
range expansion and whether this species remains closely tied to habitats along major rivers and
human built environments as it extends or shifts northward. Charting where S. psaltria
populations are expanding or decreasing can help determine which habitat characteristics are
most critical for colonization. In addition, learning how S. psaltria adapts to its new
environments will offer critical insights in the broader context of wildlife adaptations to human-
induced climate and ecosystem change. Future research on dietary preferences in the Pacific
Northwest, particularly on the balance between natural food sources and bird feeders, could be
especially important for understanding S. psaltria colonization. As climate and landscape
changes alter species distributions, it is essential to track range shifts in real-time and determine
which factors drive extinction and expansion to inform effective, species-specific conservation
strategies.
20
ACKNOWLEDGEMENTS
We thank Aly Stuart for contributing the S. psaltria illustration used in Figure 1 of this
publication and Garrett Hughes for assistance in developing the lay summary. Additionally, we
thank all Project FeederWatch and eBird participants who provided data that allowed this study
and others like it to be performed.
Funding Statement: No funding was sought for this study.
Ethics Statement: This study used publicly available participatory science data. No fieldwork or
interaction with animals occurred, and therefore no specific ethics approvals were required.
Conflict of Interest Statement: We declare no conflict of interest.
Author Contributions: M.W.M conceived the idea and design; M.W.M performed the
experiments; M.W.M, E.I.G, and J.B. wrote or substantially edited the paper; M.W.M and E.I.G
developed methods; and M.W.M, E.I.G, and J.B. analyzed data.
Data Depository: All data are accessible through Project FeederWatch (Cornell Laboratory of
Ornithology and Birds Canada), eBird Status and Trends (Cornell Lab of Ornithology), or
publicly available sources described in the methods. Processing of PFW data not described in the
methods can be reproduced using the code at https://github.com/Visorbearer/PFWDataPrep. This
material uses data from the eBird Status and Trends Project at the Cornell Lab of Ornithology,
eBird.org. Any opinions, findings, and conclusions or recommendations expressed in this
material are those of the author(s) and do not necessarily reflect the views of the Cornell Lab of
Ornithology.
21
LITERATURE CITED
Atkinson, C. T., and D. A. LaPointe (2009). Introduced avian diseases, climate change, and the
future of Hawaiian honeycreepers. Journal of Avian Medicine and Surgery 23:53–63.
Auer, S. K., and D. I. King (2014). Ecological and life‐history traits explain recent boundary
shifts in elevation and latitude of western North American songbirds. Global Ecology and
Biogeography 23:867–875.
Battey, C. J. (2019). Ecological release of the Anna’s Hummingbird during a northern range
expansion. The American Naturalist 194:306–315.
Bonter, D. N., and E. I. Greig (2021). Over 30 years of standardized bird counts at
supplementary feeding stations in North America: A citizen science data report for
Project FeederWatch. Frontiers in Ecology and Evolution 9.
Cooper, C. B., J. Shirk, and B. Zuckerberg (2014). The invisible prevalence of citizen science in
global research: migratory birds and climate change. PLoS ONE 9:e106508.
Detling, L. E. (1958). Peculiarities of the Columbia River gorge flora. Madroño 14:160–172.
Diamond, J. (1989). The present, past and future of human-caused extinctions. Philosophical
Transactions of the Royal Society of London. B, Biological Sciences 325:469–477.
Dormann, C. F., J. Elith, S. Bacher, C. Buchmann, G. Carl, G. Carré, J. R. G. Marquéz, B.
Gruber, B. Lafourcade, P. J. Leitão, T. Münkemüller, et al. (2013). Collinearity: a review
of methods to deal with it and a simulation study evaluating their performance.
Ecography 36:27–46.
eBird Basic Dataset (2024). Version: EBD_relApr-2024. Cornell Lab of Ornithology, Ithaca,
NY, USA.
22
Fink, D., T. Auer, A. Johnston, M. Strimas-Mackey, S. Ligocki, O. Robinson, W. Hochachka, L.
Jaromczyk, C. Crowley, K. Dunham, A. Stillman, et al. (2023). eBird Status and Trends.
Cornell Lab of Ornithology, Ithaca, NY, USA. https://ebird.org/science/status-and-
trends/data-access.
Fiske, I., and R. Chandler (2011). unmarked: An R package for fitting hierarchical models of
wildlife occurrence and abundance. Journal of Statistical Software 43.
Giblin, D. E., and B. S. Legler (2003). WTU image collection web site: vascular plants,
macrofungi, and lichenized fungi of Washington state. University of Washington
Herbarium. http://biology.burke.washington.edu/herbarium/imagecollection.php.
Graf, W. L. (1992). Science, public policy, and western American rivers. Transactions of the
Institute of British Geographers 17:5.
Greig, E. I., E. M. Wood, and D. N. Bonter (2017). Winter range expansion of a hummingbird is
associated with urbanization and supplementary feeding. Proceedings of the Royal
Society B: Biological Sciences 284:20170256.
Hampton, S. C. (2020). The invasion of the Pacific Northwest: California’s birds expand north
with warmer winters. The Cottonwood Post.
https://thecottonwoodpost.net/2020/03/09/the-invasion-of-the-pacific-northwest-
californias-birds-expand-north-with-warmer-winters/.
Hampton, S. C. (2021). The song of the Lesser Goldfinch: Another harbinger of a warming
climate. The Cottonwood Post. https://thecottonwoodpost.net/2021/03/15/the-song-of-
the-lesser-goldfinch-another-harbinger-of-a-warming-climate/.
23
Heeter, K. J., G. L. Harley, J. T. Abatzoglou, K. J. Anchukaitis, E. R. Cook, B. L. Coulthard, L.
A. Dye, and I. K. Homfeld (2023). Unprecedented 21st century heat across the Pacific
Northwest of North America. Climate and Atmospheric Science 6:5.
Hitch, A. T., and P. L. Leberg (2007). Breeding distributions of North American bird species
moving north as a result of climate change. Conservation Biology 21:534–539.
Hoffman, J. D., S. Narumalani, D. R. Mishra, P. Merani, and R. G. Wilson (2008). Predicting
potential occurrence and spread of invasive plant species along the North Platte River,
Nebraska. Invasive Plant Science and Management 1:359–367.
Holm, S. R., B. R. Noon, J. D. Wiens, and W. J. Ripple (2016). Potential trophic cascades
triggered by the barred owl range expansion. Wildlife Society Bulletin 40:615–624.
Jewett, S. G., and I. N. Gabrielson (1929). Birds of the Portland area, Oregon. Cooper
Ornithological Club. University of California, Berkeley, CA, USA.
Johnston, M. (2009). Record of a Lesser Goldfinch in the south-central interior of British
Columbia. Wildlife Afield 6:20–22.
La Sorte, F. A., D. Fink, P. J. Blancher, A. D. Rodewald, V. Ruiz‐Gutierrez, K. V. Rosenberg,
W. M. Hochachka, P. H. Verburg, and S. Kelling (2017). Global change and the
distributional dynamics of migratory bird populations wintering in Central America.
Global Change Biology 23:5284–5296.
Linsdale, J. M. (1957). Goldfinches on the Hastings Natural History Reservation. American
Midland Naturalist 57:1.
Maron, M. W., and C. Borre (2024). First record of breeding behavior by the Chestnut-sided
Warbler in Washington. Western Birds 55:137–142.
24
Mattocks, P. W., E. S. Hunn, and T. R. Wahl (1976). A checklist of the birds of Washington
state, with recent changes annotated. Western Birds 7:18.
Melaas, E. K., J. A. Wang, D. L. Miller, and M. A. Friedl (2016). Interactions between urban
vegetation and surface urban heat islands: a case study in the Boston metropolitan region.
Environmental Research Letters 11:054020.
Multi-Resolution Land Characteristics Consortium (2013). National Land Cover Database 2013:
Land Cover. https://www.mrlc.gov/data
Neate-Clegg, M. H. C., B. A. Tonelli, and M. W. Tingley (2024). Advances in breeding
phenology outpace latitudinal and elevational shifts for North American birds tracking
temperature. Nature Ecology and Evolution. https://doi.org/10.1038/s41559-024-02536-z
Peterson, R. T. (1984). Birds of the American West: (A field guide to western birds). Roger Tory
Peterson field guides. Easton Press, Norwalk, CT, USA.
Princé, K., and B. Zuckerberg (2015). Climate change in our backyards: the reshuffling of North
America’s winter bird communities. Global Change Biology 21:572–585.
PRISM Climate Group (2024). Oregon State University. https://prism.oregonstate.edu. Data
created 13 June 2024, accessed 13 June 2024.
Robbins, C. S., B. Bruun, H. S. Zim, and A. Singer (1979). A guide to field identification: birds
of North America. Western Publishing Company, Inc. Racine, WI, USA.
Rood, S. B., J. H. Braatne, and L. A. Goater (2010). Favorable fragmentation: river reservoirs
can impede downstream expansion of riparian weeds. Ecological Applications 20:1664–
1677.
Saikku, M. (1990). The extinction of the Carolina Parakeet. Environmental History Review 14:1–
18.
25
Schwantes, C. A. (1996). The Pacific Northwest: an interpretive history. University of Nebraska
Press. Lincoln, NE, USA.
Sekercioglu, C. H., S. H. Schneider, J. P. Fay, and S. R. Loarie (2008). Climate change,
elevational range shifts, and bird extinctions. Conservation Biology 22:140–150.
Silvertown, J., M. E. Dodd, K. McConway, J. Potts, and M. Crawley (1994). Rainfall, biomass
variation, and community composition in the Park Grass Experiment. Ecology 75:2430–
2437.
Stephens, D. A., C. Webb, and C. H. Trost (1990). First report of nesting Lesser Goldfinch in
Idaho. Western Birds 21:33–34.
Strope, C., B. Robinson, and H. W. Carlisle (2020). Using citizen science to examine the range
expansion of Bewick’s Wren and Lesser Goldfinch into Idaho.
https://scholarworks.boisestate.edu/under_showcase_2020/183.
Taha, H. (1997). Urban climates and heat islands: albedo, evapotranspiration, and anthropogenic
heat. Energy and Buildings 25:99–103.
Tennant, C. J., B. T. Crosby, and S. E. Godsey (2015). Elevation‐dependent responses of
streamflow to climate warming. Hydrological Processes 29:991–1001.
Toochin, R. (2024). The status and occurrence of the Lesser Goldfinch (Spinus psaltria) in
British Columbia. BC Rare Bird Records. https://bcrarebirdrecords.ca/species/lesser-
goldfinch#1.
U.S. Census Bureau (2020). 2020 Census Demographic and Housing Characteristics File
(DHC). https://www.census.gov/data/tables/2023/dec/2020-census-dhc.html
Versaw, A. (2000). Winter range expansion of Leser Goldfinches. Journal of the Colorado Field
Ornithologists 34:91–97.
26
Wang, Q., J. J. Koval, C. A. Mills, and K.-I. D. Lee (2007). Determination of the selection
statistics and best significance level in backward stepwise logistic regression.
Communications in Statistics - Simulation and Computation 37:62–72.
Watt, D. J., and E. J. Willoughby (2020). Lesser Goldfinch (Spinus psaltria). In Birds of the
World (S. M. Billerman, B. K. Keeney, P. G. Rodewald and T. S. Schulenberg, Editors).
Cornell Lab of Ornithology. Cornell Lab of Ornithology, Ithaca, NY, USA.
https://doi.org/10.2173/bow.lesgol.01
Willoughby, E. J. (2007). Geographic variation in color, measurements, and molt of the Lesser
Goldfinch in North America does not support subspecific designation. The Condor
109:419–436.
Woudstra, Y., R. Kraaiveld, A. Jorritsma, K. Vijverberg, S. Ivanovic, R. Erkens, H. Huber, B.
Gravendeel, and K. J. F. Verhoeven (2024). Some like it hot: adaptation to the urban heat
island in common dandelion. Evolution Letters qrae040.
Yalcin, S., and S. J. Leroux (2017). Diversity and suitability of existing methods and metrics for
quantifying species range shifts. Global Ecology and Biogeography 26:609–624.
Zipper, S. C., J. Schatz, A. Singh, C. J. Kucharik, P. A. Townsend, and S. P. Loheide (2016).
Urban heat island impacts on plant phenology: intra-urban variability and response to
land cover. Environmental Research Letters 11:054023.
Zuckerberg, B., D. N. Bonter, W. M. Hochachka, W. D. Koenig, A. T. DeGaetano, and J. L.
Dickinson (2011). Climatic constraints on wintering bird distributions are modified by
urbanization and weather: Wintering birds, weather, food, and climate. Journal of Animal
Ecology 80:403–413.