Introduction

Over the past two centuries in the United States, rapid land conversion and human population growth have created novel urban landscapes (Gibson and Jung 2002; Leyk et al. 2020). Humans have changed urban ecosystems dramatically (Grimm et al. 2008), including alterations to physical habitat (Saunders et al. 1991; Arnold and Gibbons 1996; Fortuniak et al. 2006; Carey et al. 2013; Fletcher et al. 2013), trophic dynamics (Faeth et al. 2005; El-Sabaawi 2018; Sol et al. 1998; Shochat 2004), and community composition (Green and Baker 2003; Biamonte et al. 2011; Tomasevic and Marzluff 2017). Fischer et al. (2015) propose a classification system for how species respond to cities: urban dwellers are successful and form self-sustaining populations (McKinney 2006), urban utilizers can reproduce but do not form self-sustaining populations, and urban avoiders have low reproductive success (Lancaster and Rees 1979). A fundamental interaction which can synergistically affect and respond to host tolerance of the urban landscape is parasitism (Beldomenico and Begon 2010), potentially regulating host success and community structure.

There are two broad hypotheses for how parasitism changes in cities (Werner and Nunn 2020): (1) the ‘urban burden’, where transmission is exacerbated by dense populations (Bradley and Altizer 2007) and clumped resources (Wright and Gompper 2005; Giraudeau et al. 2014), or (2) the ‘urban refuge’, where abundant food and predation release create more immunocompetent populations which can expend more energy on defenses (Bradley and Altizer 2007). The two hypotheses may be linked to Fischer et al.’s (2015) classification of urban tolerance, such that urban avoiders are susceptible to parasites on the rare occasions they enter urban areas because they are less successful there, but urban dwellers are healthy in urban habitats and therefore resistant to parasites. A countervailing theory, the dilution effect, proposes that dispersed populations reduce transmission of diseases (Bradley and Altizer 2007; Civitello et al. 2015), which may apply to urban avoiders, whereas dense, homogeneous populations increase transmission, which may apply to urban dwellers. Which hypothesis dominates depends on a complex combination of host species, sex, age, and urban tolerance (Bentz et al. 2006; Delgado-V. & French 2012), as well as parasite life cycle complexity (Calegaro-Marques and Amato 2014; Bailly et al. 2016), transmission method (Evans et al. 2009; DeVore et al. 2020), endo- versus ectoparasitism (Gregoire et al. 2002; Arzua et al. 2003; Wright and Gompper 2005), and virulence (Bichet et al. 2020).

Birds have been widely studied for their urban responses, including higher boldness (Atwell et al. 2012; Minias 2015), a variety of stress responses (Bonier 2012), reduced body mass (Meillère et al. 2015), altered phenology (Tryjanowski et al. 2013; Hutton et al. 2021), and denser populations (Sorace 2002). In addition, nearly all bird species are associated with obligately ectoparasitic chewing lice in the clades Ischnocera and Amblycera (Psocodea: Phthiraptera) (Price et al. 2003; Clayton et al. 2008). In high numbers, chewing lice exact costs on their hosts by reducing feather mass (Booth et al. 1993; Clayton et al. 1999), increasing grooming time (Brown 1974; Redpath 1988; Price et al. 2003), making mates less competitive (Clayton 1990, 1991), and sometimes blood-feeding (Samuel et al. 1982; Price et al. 2003; Dik 2006). Chewing louse eggs, called nits, are cemented on feathers in hard-to-reach places (Eichler 1963; Foster 1969a, b; Price et al. 2003). Like many parasites, chewing lice follow an overdispersed count distribution, where a few hosts have high numbers of lice, and most hosts have few or none (Price et al. 2003). The two parvorders of avian chewing lice have important behavioral differences: Amblycera are more mobile both while a host is alive and in seeking a new host after death, and they often feed on blood as well as feathers, while Ischnocera are specialized to living and feeding on feathers, to the point of restricted mobility (Clayton et al. 2008; Kolenčík et al. 2024). For lice, low humidity is associated with lower prevalence and abundance, as well as lower diversity (Fabiyi 1996; Moyer et al. 2002; Bush et al. 2009, 2024), but Ischnocera mitigate this to some extent by absorbing atmospheric water vapor (Rudolph 1983; Bush et al. 2009). Although birds have often been studied in the context of urban environments or ectoparasitism, few studies have examined the intersection of these factors.

Urban development and ecosystems overall are temporally structured, but a lack of data on historical processes has hindered the study of past urban ecology (Ossola et al. 2021). Natural history collection specimens preserve chewing lice (Mey 2003), providing a unique opportunity to pinpoint when and how the development of urban spaces has affected parasitism (Winker 2004). This is especially true for the early 20th century, when high rates of urban growth presumably caused ecological transitions (Leyk et al. 2020; Ossola et al. 2021) and when American vertebrate collecting activity peaked (Joseph 2011). In the western United States, the House Finch, Haemorhous mexicanus (Statius Müller 1776), has been widely distributed across rural and urban habitats since at least the late 1800s (Bergtold 1913; Fig. 1a). The Hairy Woodpecker, Dryobates villosus (Linnaeus 1766), is more of an urban avoider (Ripper et al. 2007; Baz and Murphy 2023; Fig. 1b), but the two species overlap in distribution (Fink et al. 2023). The House Finch’s responses to urban environments include duller coloration (Giraudeau et al. 2018; Sykes et al. 2021), lower aggression (Hasegawa et al. 2014), lower sensitivity to artificial light (Hutton et al. 2024), and higher prevalence and severity of coccidiosis and poxvirus infections (Giraudeau et al. 2014; Sykes et al. 2021). The Hairy Woodpecker’s responses to cities have not been studied specifically, but as an urban avoider, it may be less healthy and less successful (Delgado-V. & French 2012). For these host species, chewing louse associations are documented (Table 1), but such interactions have received limited ecological attention, especially in urban spaces.

Therefore, to assess the major hypotheses of urban parasitism in a historical framework, we measured chewing louse abundance on House Finch and Hairy Woodpecker specimens collected from the western United States between 1905 and 1932 (Fig. 1c). As a predictor of louse abundance, we extracted values of historical urbanness from a time-series of land use from 1900 to 2010 (Leyk and Uhl 2018b). We also tested host species as a predictor, expecting different responses from urban dwellers and avoiders, and tested year, day of year, precipitation, host sex, and host age to control for plausible biological effects. We expected that urban environments in the western United States in the early 20th century would result in higher chewing louse loads, analogous to higher House Finch endoparasite loads in modern urban environments (Giraudeau et al. 2014). Alternatively, if urban habitat did not alter avian chewing louse abundance in this period, it would either mean that during this time urbanization was too limited to affect bird-lice interactions, or that, unlike endoparasites, chewing lice do not differ in abundance between rural and urban environments.

Fig. 1
figure 1

Annual mean relative abundance of (a) H. mexicanus and (b) D. villosus in 2022 in the western United States (Fink et al. 2023). Note that H. mexicanus is widely distributed, with dense populations around urban areas. In contrast, D. villosus is more patchily distributed, with a tendency to avoid cities. (c) Collection sites of the 104 H. mexicanus and 56 D. villosus UCLA Dickey Bird and Mammal Collection specimens used in this study, which were collected 1905–1932, and 2010 BUFA (Leyk and Uhl 2018b), a metric of how many m2 are covered by human structures per 250 × 250 m grid cell. Figure created in QGIS 3.36.1 (QGIS Development Team 2024), WGS 84 CRS

Full size image
Table 1 Documented chewing louse associations with H. mexicanus and D. villosus. Familial and ordinal placements of genera from Price et al. (2003)
Full size table

Methods

Bird specimens and louse abundance

To count lice, we searched 104 House Finch and 56 Hairy Woodpecker study skins (hosts) from the UCLA Dickey Bird and Mammal Collection. Specimens were collected between 1905 and 1932 across the states of Washington, Oregon, California, Idaho, Nevada, and Arizona (Fig. 1c). Specimens included information on collection date, geographic coordinates of collection, collector name, sex of individual specimen, and specimen age class (Juvenile, Immature, or Adult). Sex and age determinations were taken from specimen tags, while the rest of the information was derived from online catalog records. Geographic coordinates had precision of one arcsecond, giving a maximum uncertainty of 1861 m.

We applied standard search protocols to collect lice from each bird specimen (Clayton and Walther 1997; Mey 2003), with additional training and advice from Jessie Salter (pers. comm.). Under a magnifying circular LED light, we searched contour feathers around the ears, beak, neck, breast, and center of the dorsal side, but excluded the wings, lower belly, and thighs because neither host species is known to host wing lice and because it was difficult to search long, downy feathers consistently and thoroughly. We moved methodically across each section, lifting one row of feathers at a time, searching each dorsal and ventral surface, especially near the skin. Using forceps, we extracted (adult) lice and nits; in some cases, nits were counted but not removed because they were attached too strongly to remove without damaging the specimen. We anecdotally observed that most but not all nits were empty, though eggs may pop open as they dry, so empty nits were not necessarily hatched. While inspecting, we held each bird over glossy paper and examined fallen debris under a dissecting scope for additional lice. To ensure completeness, each specimen was searched separately by two individuals. All searchers (EXK, JMJ, AL, MCR, GAM) trained together for consistency. The median search time was 10 min, though times varied widely – search times were longer on birds with more lice because of the time required to remove ectoparasites.

We identified representative adult louse morphotypes by preserving them in ethanol, slide-mounting them in Hoyer’s medium, and imaging them with a Leica M205 FA Fluorescence Stereomicroscope and a compound light microscope. We identified lice with keys from taxonomic authorities (Table 1) and, in the case of Brueelia thorini, additional assistance from Dr. Daniel Gustafsson (pers. comm. 2023 Jul 30).

Historical urbanness index (BUFA)

Historical urbanness values were drawn from the Building Footprint Area (BUFA) raster dataset in the Historical Settlement Data Compilation for the United States (HISDAC-US) (Leyk and Uhl 2018b). Based on property records, BUFA calculates the land area covered by human structures in the contiguous United States at 250 m resolution, every decade 1900–2010 (Leyk and Uhl 2018a). Because we were interested in the environmental impacts of urbanization, the physical definition of urbanness measured by BUFA was preferable to a more sociological definition (e.g. Bretagnolle et al. 2015). Although 100 m2 covered by human structures may have quite different environmental effects in 1900 compared to 2010, this dataset is predictive of human population size at the county level (Leyk et al. 2020) and HISDAC-US provides by far the finest resolution metrics of historical urbanness currently available for the U.S. (Leyk and Uhl 2018a; Uhl et al. 2021).

With the terra package v.1.7–55, (Hijmans 2023) in R v.4.3.1 (R Core Team 2023), we extracted the average BUFA value in a 2000-m radius buffer around the collection site of each specimen to account for the 1861-m coordinate uncertainty, proportional to the exact fraction of each cell covered by the buffer. Values were extracted from the most recent layer before each collection date; for example, specimens collected in 1920 and 1929 both received BUFA values from the 1920 raster layer.

Precipitation

To account for effects of climatic variation, particularly humidity, we obtained bioclimatic variable BIO17, which is the total precipitation in the driest three months of the year in mm, in 2.5 arcminute resolution from WorldClim v.2.1, based on data from 1970 to 2000 (Fick and Hijmans 2017). We chose BIO17 because humidity is the main climatic variable thought to affect avian chewing lice (Bush et al. 2024). With the same terra package, we extracted precipitation values at the collection coordinates for each specimen. However, because of missing raster cell values, we extracted precipitation values at the exact coordinates using the package’s bilinear method, which interpolates the four nearest cells, without a buffer.

Statistical analysis

We conducted all statistical analyses in R v.4.3.1 (R Core Team 2023). We fit negative binomial generalized linear mixed models (GLMM) of louse abundance with the glmmTMB package (v.1.1.7, Brooks et al. 2017) using the ‘nbinom2’ family, which has a quadratic parameterization and a log link. Although our dataset potentially showed some zero-inflation, we found non-convergence of zero-inflation models during exploratory analyses, so we retained the simpler negative binomial GLMM for final analysis. Our model choice was supported by previous descriptions of louse abundance distributions (Price et al. 2003).

Because Amblycera and Ischnocera have important differences in their life histories, feeding patterns, and responses after host death, we separated them when modeling. However, we only encountered one amblyceran species, and it only occurred on a single Hairy Woodpecker. Therefore, there was an insufficient sample size for any amblyceran interaction, so we only modeled adult Ischnocera and total nits. The Ischnocera model included both host species, and its response variable was the summed abundances of Brueelia thorini, Philopterus sp., Brueelia straminea s. lat., Penenirmus auritus s. lat., and Picicola snodgrassi, with the first two occurring on House Finches and the latter three on Hairy Woodpeckers. We did not distinguish ischnoceran louse species because small sample sizes would have weakened our modelling power. We selected the best fitting model for each response variable based on the values of AICc calculated with the AICcmodavg package (v.2.3.3, Mazerolle 2023) and ΔAICc and Akaike weights calculated with the qpcR package (v.1.4-1, Spiess 2018; Anderson and Burnham 2002). As predictors, we tested BUFA, collection year, collection day of year (with linear and quadratic terms), precipitation, host sex, host age class, and host species. We z-transformed the continuous variables (BUFA, year, day of year, precipitation) before fitting. Because seasonal effects were unlikely to follow a linear pattern from January 1st to December 31st, day of year was always included as both a linear and a quadratic predictor, where the quadratic term was squared after scaling. We expected that the biological categories could alter the bird-louse interaction, and that there could be a seasonal effect. Collection year was mostly included as a nuisance variable, although it was possible that the interaction had changed linearly through time. Although collector name was available from the catalog, we excluded it because there were many collectors that only collected a few specimens, so there was a high risk of overfitting. We used BUFA as a predictor in every tested model because it was our variable of interest, and we used host species in every model except the BUFA-only model, because of the a priori assumption that two species of bird with two different louse communities would behave differently. We also tested the interactions of BUFA with collection year and BUFA with host species. These interactions were chosen because the same area covered by human structures could equate to very different impacts in 1900 versus 2010, and different bird species are known to respond differently to urban spaces, which might also affect their parasites. We would have tested host sex and age class with host species as the random intercept, but these models did not converge. Our full candidate model set can be found in SI Tables 1 and 2.

To evaluate the strength and effect of our variables, we used the ggeffects package (v.2.2.0, Lüdecke 2018) to predict counterfactual means, which reflect the distribution of the data by averaging the response of every observation as if they all belonged to each level or value of the predictor, for the two host species and across the ranges of BUFA and collection day of year. Estimates were back-transformed to the response scale (i.e., number of lice or nits).

Results

We documented two louse species parasitizing House Finches (host n = 104) and four parasitizing Hairy Woodpeckers (host n = 56), as well as abundant nits which were not identified to species (Figs. 2 and 3a). Of the 104 House Finches, 32 (31%) hosted Brueelia thorini (Fig. 2a) and 26 (25%) hosted Philopterus sp. (Fig. 2b), including 11 (11%) which hosted both species. Of the 56 Hairy Woodpeckers, just 1 (2%) hosted Menacanthus sp. (Fig. 2c), and this individual hosted 6 Menacanthus sp. and 27 total lice. Additionally, 7 (13%) woodpeckers hosted Brueelia straminea s. lat. (Fig. 2d), 4 (7%) hosted Picicola snodgrassi (Fig. 2e), and 11 hosted P. auritus s. lat. (Fig. 2f). Only 2 (4%) Hairy Woodpeckers hosted multiple louse species: one hosted both B. straminea s. lat. and P. auritus s. lat., while the other was an anomalously high-load individual which hosted all four species that we found on Hairy Woodpeckers.

Fig. 2
figure 2

The two louse species found in this study on House Finches were (a) Brueelia thorini and (b) Philopterus sp. The four louse species found in this study on Hairy Woodpeckers were (c) Menacanthus sp., (d) Brueelia straminea s. lat., (e) Picicola snodgrassi, and (f) Penenirmus auritus s. lat. Lice were slide-mounted in Hoyer’s medium and imaged with a Swiftcam SC1803R camera, Swift microscope SW380T, and Swift Imaging software v.3.0

Full size image

Ischnocera and nits were found on all host species (Fig. 3a), host sexes (Fig. 3b), and host age classes (Fig. 3c), with similar numbers between host sexes, though adult lice tended to be more abundant on immatures, and nits tended to be more abundant on House Finches. Specimens were collected between 1905 and 1932 (Fig. 4a), at sites with decadal BUFA values ranging from 0 to 2756 m2 built up area per 62,500 m2 grid cell (Fig. 4b). Specimens were collected across a gradient of 3–166 mm precipitation in the driest three months of the year (Fig. 4c) and throughout the year (Fig. 4d).

The top-ranked AICc model for nit abundance showed the same model structure as the second-highest ranked model for adult Ischnocera (SI Tables 1 and 2). Therefore, we selected this model structure for both response variables. Host species was a significant predictor in both models, though Hairy Woodpeckers were associated with higher Ischnocera and lower nit counts (Tables 2 and 3). Urbanness was not significant for either model (Tables 2 and 3), nor was precipitation (Tables 2 and 3), even though both variables were retained during model selection. Linear day of year had a significant negative effect only for nits, while day of year squared had a significant positive effect only for the adult Ischnocera model (Tables 2 and 3). These selected models did not include collection year, host sex, or host age as predictors.

Louse and nit abundance by species, host sex, and host age

On House Finches we found 0.86 ± 2.24 B. thorini and 0.38 ± 0.78 Philopterus sp. (mean ± s.d.) per specimen, with a total of 1.24 ± 2.35 lice per specimen (Fig. 3a). We also found 8.81 ± 19.69 nits per specimen (Fig. 3a), in two distinct types: larger, teardrop-shaped pairs on the underside of feathers and smaller, rounder ones in haphazard clumps, but we were unable to identify nits to species. On Hairy Woodpeckers, there were 0.11 ± 0.80 Menacanthus sp., 0.63 ± 2.35 B. straminea s. lat., 0.16 ± 0.68 P. mississippiensis, and 0.45 ± 1.16 P. auritus sp., with a total of 1.34 ± 3.87 lice per specimen, as well as 2.07 ± 5.90 nits (Fig. 3a). We again did not identify nits to species and did not notice obviously different types on Hairy Woodpeckers.

Host sex and age class were not predictors in our selected models. Excluding the six Menacanthus sp. on a single Hairy Woodpecker because they could not be modeled, on male hosts we found 1.22 ± 2.76 lice and 6.85 ± 19.00 nits, and on female hosts we found 1.27 ± 2.43 lice and 5.59 ± 9.60 nits (Fig. 3b). On juvenile hosts we found 0.36 ± 0.84 lice and 6.29 ± 13.70 nits, on immature hosts, we found 2.33 ± 3.75 lice and 4.60 ± 9.76 nits, and on adult hosts we found 1.21 ± 2.61 lice and 6.68 ± 17.50 nits (Fig. 3c).

Fig. 3
figure 3

Boxplots of Ischnocera and nit abundance across (a) host species, (b) host sex, and (c) host age class in the raw data. Note the logarithmic y-axes; louse and nit abundance values are plotted as (count + 1) to allow a log transformation for zero-counts. Amblycera are not included, because we only found six Menacanthus sp. on a single Hairy Woodpecker specimen

Full size image
Fig. 4
figure 4

Scatterplots of Ischnocera and nit abundance on each specimen against (a) collection year, (b) the BUFA value at the collection site, (c) precipitation in the driest quarter of the year at the collection site, and (d) collection day of year in the raw data. Note the logarithmic y-axes; louse and nit abundance values are plotted as (count + 1) to allow the log transformation for zero-counts. Amblycera are not included, because we only found six Menacanthus sp. on a single Hairy Woodpecker specimen

Full size image

GLMM selection and results

For predicting Ischnocera abundance, we found eleven models with ΔAICc < 2, of which all eleven included day of year, seven included year, four included precipitation, one included host sex, seven included host age, one included the interaction of BUFA with host species, and one included the interaction of BUFA with year (SI Table 1). For predicting nit abundance, there were eleven models with ΔAICc < 2, of which seven included day of year, two included year, six included precipitation, one included host sex, none included host age, four included the interaction of BUFA with host species, and none included the interaction of BUFA with year (SI Table 2). We selected the model parametrization that was highest weighted for nits (ΔAICc = 0), which was also second-highest weighted for Ischnocera (ΔAICc = 0.78). In contrast, the highest weighted parametrization for Ischnocera (ΔAICc = 0) was a poor model for nits (ΔAICc = 4.75). The selected model had BUFA, host species, day of year, and precipitation as its predictors, without interaction or random effects (Tables 2 and 3).

In the negative binomial GLMM of Ischnocera abundance (model I43 in SI Table 1), Hairy Woodpeckers had significantly higher louse loads than House Finches (Table 2; Fig. 5a). Urbanness was not a significant predictor (Table 2; Fig. 5b), nor were precipitation (Table 2; Fig. 5c) and linear day of year (Table 2; Fig. 5d), despite being retained in the model. However, day of year squared was a significant positive predictor (Table 2; Fig. 5d), indicating lower Ischnocera abundance in the summer than in the winter.

Table 2 Estimated regression parameters, standard errors, z-values, and p-values for the model of Ischnocera abundance. BUFA, day of year, and precipitation were scaled before running the model
Full size table

In the negative binomial GLMM of nit abundance (model N43 in SI Table 2), House Finches had significantly higher loads than Hairy Woodpeckers (Table 3; Fig. 5a), the inverse of the pattern for both adult louse models. Urbanness was not a significant predictor (Table 3; Fig. 5b), nor was precipitation (Table 3; Fig. 5c). However, linear day of year was a significant negative predictor (Table 3; Fig. 5d), whereas day of year squared was non-significant (Table 3; Fig. 5d), unlike the adult lice, indicating more abundant nits early in the year.

Table 3 Estimated regression parameters, standard errors, z-values, and p-values for the model of Nit abundance. BUFA, day of year, and precipitation were scaled before running the model
Full size table
Fig. 5
figure 5

GLMM-predicted adult louse and nit abundances with 95% confidence intervals in response to (a) host species, (b) the urbanness metric BUFA, (c) precipitation in the driest quarter of the year, and (c) collection day of year. The prediction method was counterfactual. Ischnocera and nits are displayed together, but their abundances were not compared to each other, because they were modeled separately

Full size image

Discussion

We found that urbanness did not have a significant effect on avian louse abundances, either demonstrating that unlike other parasites, chewing lice abundance is not affected by urban environments, or that ecological effects of urbanization in the western United States were much more limited a century ago compared with the present day. This held across both House Finches and Hairy Woodpeckers, even though we hypothesized that ectoparasitism on urban dwelling House Finches would be more strongly affected in urban environments, either suppressed by improved host health and increased preening or promoted by denser host populations. This null finding suggests that neither the urban tolerance/avoidance nor the dilution hypotheses can adequately explain avian chewing lice interactions, or that these effects only become large enough to be biologically meaningful in the western United States after the 1930s.

Our lack of evidence that urban environments change louse abundance must be understood in the context of different kinds of parasitic interaction and the source of our specimens. In contrast to our nonsignificant effect of urbanness, Giraudeau et al. (2014) documented higher severity and prevalence of coccidia and poxvirus infections for more urban House Finches. Potentially, lice are less responsive to urban environments than other infections because they are buffered from certain environmental factors, or because the changes that avian hosts undergo in urban environments have limited effects on anti-louse defenses. It should be noted that we only modeled Ischnoceran lice because we found very few Amblycera, and these parvorders may respond differently to the urban environment, as they do with humidity (Bush et al. 2009). Few other studies have examined the effect of urban environments on avian lice of any host species, except for some interactions of introduced species (Delgado-V. & French 2012), so future comparative research between bird species could elucidate if avian louse abundance is always unaffected by cities, or if abundance depends on the host species and the host’s tolerance for urban environments. Alternatively, because our specimens reflect urban and non-urban environments between 1905 and 1932, our findings may represent the state of urban ecosystems in the western United States a century ago and their differences from modern cities. More intensive sampling of specimens across a wider time series, and specifically comparison between historical and modern specimens, could elucidate whether the effects of cities on avian louse abundance have changed since 1905. If modern louse abundance is affected by cities, even though it was unaffected historically, this would be evidence that the intensification of urban environments through time has caused corresponding variation in parasitic interactions.

For House Finches and Hairy Woodpeckers collected in the western United States from 1905 to 1932, House Finches had fewer adult Ischnocera and more nits than Hairy Woodpeckers, precipitation did not have a significant effect, and linear day of year was a negative predictor for nits, while quadratic day of year was a positive predictor for both Amblycera and Ischnocera. It is not entirely clear why host species has opposite effects on adult lice and nits, but one potential factor is that they experience different infection dynamics. Adult lice are strongly controlled by host preening and other behaviors (Booth et al. 1993; Clayton et al. 1999; Clayton and Walther 2001), but nits are usually cemented in hard-to-reach places (Nelson and Murray 1971) and may escape preening pressure. Since one female louse can lay 12–20 nits every 4–10 days (Price et al. 2003), just one breeding pair of adult lice that escapes preening could cause high nit abundance. Therefore, adult louse abundance may represent the long-term infection success, modulated by energy the host allocates to anti-louse behaviors, whereas nit abundance may be a snapshot of a more variable metric that requires more frequent sampling to fully describe. Alternatively, the discrepancy for host species as a predictor may stem from differences in louse species life histories between hosts. For example, Brueelia thorini on House Finches may be more fecund than Brueelia straminea s. lat. on Hairy Woodpeckers. Such a pattern could potentially be due to evolutionary histories, host grooming pressure, or competition between louse species, but little concrete is known about variation among life histories of different louse species.

We additionally found a null ecological effect of precipitation, as a proxy for humidity, across the ranges of our louse species. The effect of humidity on the abundance of a single louse species which is adapted to an arid climate may be small (Carrillo et al. 2007), and the ecological scale of our study should be distinguished from the correlation of humidity with louse abundance and diversity at a larger biogeographic scale (Bush et al. 2024). The positive association of quadratic day of year with adult louse abundances indicates lower louse loads in the summer, but the nits instead showed high abundance early in the year. Foster (1969a) reported a peak in Ricinus nit loads corresponding with the host’s spring breeding season, and our data appear to show a similar pattern for nit abundance. There is wide variation in whether louse communities peak during breeding, during molt, or in response to unrelated seasonal patterns (Galloway and Lamb 2021), but since both host species that we studied have a single annual molt in late summer to fall (Pyle 1997), the decline in adult louse abundance in the middle of the year may be related to shedding and replacing feathers. It seems likely that the louse life stages were temporally decoupled in our study because adult lice and nits responded most strongly to different effects in their hosts’ life histories.

Notably, our study represents the first record of Philopterus sp. lice on House Finches (Table 1), with documented occurrences in our data originating from the U.S. states of Washington, California, and Nevada. We found Philopterus sp. so frequently on House Finches that they were almost certainly infesting those hosts in life, and Philopterus species have known associations with several other Fringillidae (Price et al. 2003). Brueelia thorini was already known from House Finches in Utah, California, and Mexico (Table 1). We did not find Myrsidea conspicua, Menacanthus alaudae, or Ricinus microcephalus on any specimens, despite previous reports from Hawaii for M. conspicua, Hawaii and Arizona for M. alaudae, and Hawaii and California for R. microcephalus (Table 1). The absence of these species may be partially because they are Amblycera, which are more mobile than Ischnocera (Price et al. 2003) and more likely to abandon a dead host (Mey 2003). For Hairy Woodpeckers, we found Menacanthus sp. on a single specimen from California, and this genus is widely reported, including on Hairy Woodpeckers in Manitoba (Table 1). Brueelia straminea s. lat., P. auritus s. lat., P. jungens, and Picicola snodgrassi have been reported across the U.S. (Table 1).

A complication for our findings is that we studied museum specimens collected in the early 1900s. We lack documentation on the exact preparation and handling methods used on our specimens over decades. However, chewing lice are often preserved with museum skins (Mey 2003), and the adult lice we found were still in identifiable condition. With the exception of Philopterus sp., we only found louse species that have documented associations with our host birds, so cross-contamination during specimen collection is of minimal concern. Mey (2003) notes that chewing lice will only leave a corpse if they encounter a new host, and Ischnocera are particularly reluctant to leave. Even assuming that a proportion of lice leave after death, high loads on specimens still must correlate with high loads in life. Moreover, since nits are not dislodged even after death, they are reliably correlated with nit abundance in life (Foster 1969b). Some modern collectors wash bird skins as they are prepared (Winker 2000; Szabo et al. 2013), which can remove ectoparasites (Szabo et al. 2014), but this is a modern practice that postdates most of our specimens (Mey 2003; e.g. Chapin 1923 does not mention washing), because it relies on modern dish detergents. We also note that nits are likely resistant to washing because they are cemented to the feathers, so they may be a reliable abundance metric even for washed specimens. These factors give us confidence that our ectoparasite counts are closely correlated with the true abundance of at time of collection. We urge future specimen preparators to keep records of preparation techniques and specifically to avoid washing skins when specimen condition and cleanliness permit.

Museum collections are a vital tool for understanding how urban ecology has changed through time. Because these museum specimens allowed us to study urban conditions from 1905 to 1932, we now have historical data that can track earlier phases of urbanization. The best interpretation of our findings is either that avian louse abundance on House Finches and Hairy Woodpeckers is not affected by urbanness, or that from 1905 to 1932 the degree of urbanization in the western United States was not yet severe enough to affect this louse abundance. By comparing louse abundance in present-day urban and rural habitats to our historical findings, future studies could provide insight into how urbanization and its effects on birds and their parasites have changed in the past century.