Comparing perineuronal nets and parvalbumin development between blackbird species with differences in early developmental song exposure

Obligate brood parasitic species never build their own nests, incubate their own eggs or provision their own young. Instead, brood parasitic species lay eggs in the nest of other species (Payne, 1977; Winfree, 1999). Although this strategy allows parasitic species to evade the hard work of nest defense and finding food for hungry offspring, the strategy is loaded with a set of novel ontogenetic challenges for the young brood parasite that is left to be raised by an entirely foreign species. Young brood parasitic birds must avoid mis-imprinting on host parents and must eventually correctly identify conspecifics that they have never encountered while in the host nest (King and West, 1977; Sherman, 1999; Hauber et al., 2000, 2001; Slagsvold and Hansen, 2001; Lynch et al., 2017). Without the presence of conspecifics during this critical developmental period, learning the correct species-typical song is a greater challenge for the young brood parasite in comparison to birds raised by their own parents.

Some brood parasitic species, such as the brown-headed cowbird (Molothrus ater), are within the order Passeriformes, suborder Oscine. Brood parasitism within oscines (hereafter, referred to as songbirds) is particularly intriguing because these species learn their species-typical song from adult male conspecific tutors during discrete early developmental sensitive periods (Marler and Peters, 1987; Brainard and Doupe, 2002; Williams, 2004; Brenowitz and Beecher, 2005), a time frame when brood parasitic young have negligible exposure to conspecific songs. Early development of songbird vocalizations includes two sensitive periods: a perceptual phase, during which species-typical vocalizations are memorized, and a sensorimotor phase, when vocalizations are practiced using auditory feedback that helps refine song structure (Doupe and Kuhl, 1999; Konishi, 1965; Williams, 2004). Songbirds can either be ‘age-restricted’ or ‘close-ended’ learners, as opposed to ‘open-ended’ learners with a protracted time frame for song learning (Marler and Peters, 1982, 1987; Brainard and Doupe, 2002; Brenowitz and Beecher, 2005). Species in the former category, such as the zebra finch, exhibit a restricted time window for males to learn song from adult tutors (Böhner, 1990). Species in the latter category are capable of developing new songs after sexual maturity (Brenowitz and Beecher, 2005). Song learning systems, however, are not dichotomous. Rather, a spectrum of song learning exists ranging from closed- to open-ended learning (Brenowitz and Beecher, 2005). Because developing brood parasites are unable to acquire memories of conspecific songs during early phases of development but also do not acquire new songs throughout adulthood, these species present a unique opportunity to explore neural plasticity along this spectrum.

Young brood parasites spend sensitive developmental periods surrounded by heterospecifics or other juvenile conspecifics that have also recently fledged a heterospecific host nest (Friedmann, 1929; Friedmann and Kiff, 1985; O'Loghlen and Rothstein, 1993; Rothstein et al., 1980). Nevertheless, juvenile parasitic brown-headed cowbirds develop a species-typical song even in the absence of adult male tutors, indicating that song learning is tutor-independent and guided by social interactions (King and West, 1988). Juvenile males housed with only adult females and no adult males develop highly attractive courtship songs that elicit copulation solicitation display from females (White et al., 2002a,b). Consequently, young male brood parasites offer an opportunity to investigate neural plasticity associated with song learning that is independent of adult male tutors.

The brown-headed cowbird is a brood parasitic icterid species (i.e. blackbird; Fig. 1A) that is ubiquitous across North America. First-year juvenile cowbirds may join a flock with no adults and never interact with a single adult during their whole first year, whereas other juveniles join a flock with adults that are actively breeding and remain with these adults during their entire first year (Friedmann, 1929; O'Loghlen and Rothstein, 1993; Rothstein et al., 1980). Consequently, during a sensitive developmental time frame in other songbirds, young cowbirds experience unstable social environments exhibiting stark contrast from one population to another, which generates pronounced variability in the timing of song production and song dialects, which persists well into adulthood (O'Loghlen and Rothstein, 2010; White et al., 2002a,b). Brown-headed cowbirds can memorize songs from adult tutors in their first year, but can also overcome the lack of adult tutors by using social cues within the flock to develop new songs (O'Loghlen and Rothstein, 2002). Thus, cowbirds represent a natural system in which conspecific song isolation occurs during sensitive developmental periods, which requires this species to wait until their first year or later to learn conspecific songs. In contrast, a closely related non-parasitic blackbird species, the red-winged blackbird (Agelaius phoeniceus; Fig. 1A), has plenty of opportunities to memorize songs early in development because this species does raise its own young. The young red-winged blackbird is surrounded by conspecific song from the day it is hatched. Moreover, male red-winged blackbirds produce new song types beyond the second year, indicating that this species is a typical open-ended learner (Marler et al., 1972). However, there is no evidence that cowbirds continue to learn new songs well into adulthood, and therefore, they cannot be considered a typical open-ended learner (see Brenowitz and Beecher, 2005 for spectrum of closed to open-ended song learning). Clearly, their time frame for song learning is protracted, as is the case for an open-ended learner. Consequently, we predict that neural plasticity within the song learning brain regions will resemble that of an open-ended learner. Moreover, comparing the timing of neural plasticity patterns between these two closely related species with stark contrast in early conspecific song exposure provides a unique opportunity to examine the importance of early developmental song memorization in birds with protracted sensory learning.

To this end, in the present study, we investigated markers of brain plasticity associated with critical periods in song development in male brown-headed cowbirds and red-winged blackbirds by comparing the development of perineuronal nets (PNN) and neurons expressing the calcium-binding protein parvalbumin (PV; Fig. 2). PNN are aggregations of chondroitin sulfate proteoglycan chains (CSPG) with a number of proteins that form mainly around GABAergic interneurons containing PV (van't Spijker and Kwok, 2017). The appearance of PNN is assumed to control the closing of sensitive neural plasticity periods by acting namely as a physical barrier that prevents new synaptic contacts (Hensch, 2005; Karetko and Skangiel-Kramska, 2009). Although PV appearance generally corresponds to the onset of sensitive neural plasticity periods, the development of PNN around PV-positive neurons is supposed to be involved in their maturation (Beurdeley et al., 2012). This event often marks the closing of sensitive periods associated with sensory-based learning (Pizzorusso et al., 2006, 2002) and sensorimotor learning in mammals (Gogolla et al., 2009; Happel et al., 2014; Morikawa et al., 2017). It was recently found that PNN could also play an important role in the closing of the sensitive periods for vocal learning in songbirds (Balmer et al., 2009; Cornez et al., 2017b, 2018b). PV expression is more abundant in vocal learners than in non-learners in mammals and birds (Hara et al., 2012), and PNN expression during ontogeny correlates with song crystallization in the zebra finch (Cornez et al., 2018b) and canary (Cornez et al., 2018a).

Here, we compared the timing of PV and PNN appearance in brown-headed cowbirds and red-winged blackbirds across two age groups: hatch year (HY, i.e. juveniles) and after hatch year (AHY, i.e. adults). We examined PV+ and PNN+ cells in nuclei of the song control system (SCS) involved in song learning and production. These regions comprise the vocal motor pathway and the anterior forebrain pathway (Fig. 3). The nucleus HVC (acronym is now a proper name) connects to the premotor nucleus RA (robust nucleus of the arcopallium), and downstream to the neurons controlling the syrinx muscles. Together, they form the vocal motor pathway that functionally resembles the connection between Broca's area and the laryngeal motor cortex of humans (Pfenning et al., 2014). HVC also connects to Area X and LMAN (lateral magnocellular nucleus of the anterior nidopallium), which then projects to RA (Fig. 3). This forms the anterior forebrain pathway, which functionally resembles the corticostriatal motor loop of humans and is involved in the control of song learning and variability. We examined the interconnected nuclei that form the vocal motor and anterior forebrain pathways to compare PNN and PV patterns in male brown-headed cowbirds and red-winged blackbirds. We examined PV and PNN within these song control regions because our previous studies revealed that these markers are differentially regulated in these regions in zebra finches (Taeniopygia guttata) and starlings (Sturnus vulgaris) as compared with other song learning regions including the auditory forebrain (Cornez et al., 2015, 2017b, 2018b). We tested the hypothesis that brown-headed cowbirds would display similar patterns of PV and PNN in comparison to the closely related, open-ended song learner – the red-winged blackbird. We predicted these neural plasticity markers would be similar in these two species that have an open versus protracted song learning time frame. By co-opting a song learning system like a typical open-ended learner, brood parasites may have developed a delayed song learning time frame that allows them additional time to locate conspecifics with which to learn the song of their own species. Although it is also the case that brown-headed cowbirds may exhibit a great deal of variability in neural plasticity markers such as PNN and PV across development, the data presented here provide a foundation to understand whether species differences even exist between parasitic and non-parasitic species and where in the SCS heightened plasticity might be located in a brood parasitic species. This comparison will provide a novel perspective on the neural plasticity associated with delayed song learning in parasitic birds that helps them meet the challenges of their parasitic lifestyle.

Wild-caught brown-headed cowbirds (N=16 males and N=8 females) and red-winged blackbirds (N=14) were collected May through August 2018 in Travis County, TX, and Nassau County, NY, USA, using bait traps (see Fig. 1A for phylogenic relationships between these species). Tissue from all subjects was collected and fixed immediately upon capture in the field. All ages were estimated based on plumage in order to categorize subjects into hatch year (HY; N=6 for cowbirds, Fig. 1B; N=5 for redwings, Fig. 1D) and after hatch year (AHY=first breeding season and older; N=10 for cowbirds, Fig. 1C; N=9 for redwings Fig. 1E,F). HY birds were between 30 and 60 days old and are considered as juveniles here, whereas AHY birds included birds in their first year of breeding and beyond. These are considered as adults here. Additionally, adult female brown-headed cowbirds were used to explore the sex difference in adult cowbirds. The mean±s.d. testes size for adult male cowbirds and red-winged blackbirds was 5.9±0.78 and 6.9±1.5 mm, respectively. Although we were able to verify testes in juvenile male cowbirds, the testes were too small to measure in most subjects, whereas the average testes size in juvenile red-winged blackbirds was 2.0±0.48 mm. The largest follicle size was also measured in female cowbirds, which averaged 1.0±0.46 mm across subjects. All birds were perfused under isoflurane anesthesia with phosphate buffered saline (PBS) followed by 4% paraformaldehyde within 12 h after capture. Brain tissue was post-fixed in paraformaldehyde for 24 h before cryoprotection in 30% sucrose and flash-freezing in liquid nitrogen. All tissues were stored at −80°C until sectioning. Brains were sectioned at 30 µm into four consecutive coronal series of four wells and placed in a cryoprotectant solution for transport to the University of Liège. All procedures were approved by the Institutional Animal Care and Use Committee at Hofstra University (permit 16/17-11).

A single series of tissue was used to conduct free-floating double-label immunocytochemistry (ICC) for PV and PNN as described in Cornez et al. (2015, 2017b, 2018b), except for minor differences in the DAPI staining procedure and the mounting medium used for coverslipping. Briefly, sections were blocked in 5% normal goat serum (NGS) diluted in Tris-buffered saline (TBS) with 0.1% Triton-X-100 (TBST) for 30 min. They were incubated overnight at 4°C in a mixture of two primary antibodies diluted in TBST: a mouse monoclonal anti-chondroitin sulfate antibody (CS-56, 1:1000; C8035, Sigma Aldrich) specific for the glycosaminoglycan portion of the chondroitin sulfate proteoglycans that are the main components of the PNN, and a polyclonal rabbit anti-parvalbumin antibody (1:1000; ab11427, Abcam). Sections were then incubated at room temperature in a mixture of secondary antibodies diluted in TBST. A goat anti-mouse IgG coupled with Alexa 488 (green, 1:100, Invitrogen) was used to visualize PNN staining and a goat anti-rabbit IgG coupled with Alexa 546 (red, 1:200, Invitrogen) was used to visualize PV cells. Afterwards, sections were incubated in a bath of DAPI for 10 min (0.5 µg ml⁻¹ diluted in H₂O, D1306, Invitrogen) that was used to confirm that PNN that were not surrounding PV-positive cells were localized around a cell nucleus. Finally, sections were mounted on slides using TBS with gelatin and coverslipped with Fluoromount-G (00-4958-02, Invitrogen).

Quantification of (1) cells surrounded by PNN (PNN), (2) PV+ cells (PV) and (3) PV cells surrounded by PNN (PV+PNN) were made using the ImageJ software (https://imagej.nih/ij) from a total of four photomicrographs taken at 40× from both hemispheres in two sections. Photomicrographs of these regions were taken along the rostro-caudal axis for each brain region with a Leica fluorescence microscope (DMRB Fl 100). The targeted brain regions included HVC, RA, Area X and LMAN. Each photomicrograph was taken within only the nucleus under investigation, so that the same surface was counted each time. The results of the cell counting were transformed into densities (number of objects per mm²).

To quantify the volume of nuclei, photomicrographs of all sections containing the regions of interest were taken with the same microscope using a 2.5× objective. The area of each region of interest within each section was measured using ImageJ software based on the bright PNN staining highlighting these nuclei (see Fig. 4) and summed independently for each hemisphere, multiplied by the distance between sections (0.24 mm) and averaged between hemispheres to obtain the volume of each region as described previously (Cornez et al., 2017b).

The results of cell counting were then used to estimate the total numbers of PNN, PV and PV+PNN per nucleus using the following formula: (number of counted objects)×[nucleus volume/(counted area×section thickness)]. Cell counting and measurements were completed by an observer blind to the age, sex and species of the subjects.

In most birds, HVC, RA and Area X were identified based on their brighter chondroitin sulfate background staining, and PV staining (see Fig. 4 illustrating this staining in HVC). However, LMAN could not be delineated in any bird by this method because the borders of this nucleus are not clearly visible with these fluorescent stains. Therefore, counts in this region are only presented in densities and the total numbers per nucleus could not be estimated.

In female cowbirds, the faint staining of HVC did not allow a precise delineation based on PV or PNN staining, whereas the brighter nuclei of PV-positive cells allowed their identification for cell counting. It was also not possible to delineate Area X in females because the background staining was too faint. However, as this region is very large and present in multiple sections, 40× photomicrographs could be taken at the same location as in males based on the laminas that are clearly visible, as described for female zebra finches (Cornez et al., 2015). The probability that these counts were taken outside the female Area X is thus extremely low given the large size of this structure. We were only able to delineate RA accurately in females. Consequently, we analyzed the sex difference in adult cowbirds using only the densities of PNN, PV and PV+PNN for the four nuclei studied.

We were also not able to find HVC in one juvenile and one adult female cowbird and RA in one female cowbird. Additionally, we could not measure Area X volume in one juvenile brown-headed cowbird and in one juvenile red-winged blackbird. This led to a small reduction of the sample size for some or all measurements in the corresponding groups and nuclei.

Species and age differences in males were explored using two-way ANOVA for each region independently for measures of volume as well as for numbers and densities of PNN, PV and PV+PNN. Significant interactions were analyzed using Tukey post hoc tests. Using the adult male cowbirds, we compared PNN, PV and PV+PNN densities between male and female brown-headed cowbirds using independent t-tests to understand song-dependent PNN. The critical significance level was set at P<0.05 for all tests. Moreover, our overall conclusions remain identical when we used Benjamini–Hochberg false discovery rate (FDR) corrections to account for separate analyses for each nucleus, and therefore, P-values reported here do not include FDR corrections. All statistical analyses were performed with GraphPad Prism V6 Software.

The volumes of HVC, RA and Area X were all significantly different across species (HVC: F_1,25=34.0, P<0.001; RA: F_1,26=15.58, P<0.001; Area X: F_1,24=19.8, P<0.001; Fig. 5A–C). HVC and RA volumes were significantly different across ages and a marginal difference was detected in Area X volume (HVC: F_1,25=14.8, P<0.001; RA: F_1,26=8.8, P=0.006; Area X: F_1,24=3.3, P=0.082), with no significant interaction between these effects (HVC: F_1,25=3.3, P=0.081; RA: F_1,26=0.67, P=0.81; Area X: F_1,24=0.2, P=0.677). Overall, the volume was systematically larger in adults compared with juveniles and in red-winged blackbirds compared with brown-headed cowbirds.

We compared the number of PNN, PV and PV+PNN per nucleus across HVC in juvenile and adult male red-winged blackbirds and brown-headed cowbirds. The number of PNN per nucleus was not significantly different across species (F_1,25=1.7, P=0.199; Fig. 5D), but was significantly different between the two age groups (F_1,25=19.7, P<0.001), with no interaction between age and species (F_1,25=0.1, P=0.749). The number of PV per nucleus was not significantly different across species (F_1,25=3.1, P=0.089; Fig. 5G), but was significantly different between the two age groups (F_1,25=5.07, P=0.033). There was no significant interaction between age and species (F_1,25=0.1, P=0.808). The number of PV+PNN per nucleus was also not significantly different across species (F_1,25=1.3, P=0.263; Fig. 5J), but was significantly different between the two age groups (F_1,25=19.4, P<0.001), with no significant interaction between the two factors (F_1,25=0.4, P=0.529). The age differences were due to a higher number of PNN, PV and PV+PNN per nucleus in adults compared with juveniles.

A similar age difference was found for the density of PNN, PV and PV+PNN (no. per mm²), suggesting that the rise of PNN number with age exceeds the increase in HVC volume (see Table 1 for details). Additionally, there was a main effect of species as well as an interaction between species and age for the density of PV, which was significantly higher in juvenile compared with adult cowbirds, and also compared with juvenile red-winged blackbirds (Table 1). This higher PV density in juvenile cowbirds is probably due to the species difference in HVC volume accounting for the smaller HVC volume in juvenile cowbirds. But we were not able to verify that statistically because there was no interaction for this measure (see Fig. 5A).

In the RA, the number of PNN was significantly different across species (F_1,26=4.6, P=0.041; Fig. 5E) and ages (F_1,26=41.7, P<0.001) with a significant interaction between these two factors (F_1,26=5.8, P=0.024). Likewise, the number of PV+PNN per nucleus was significantly affected by the two factors and their interaction (age: F_1,26=7.3, P=0.012; species: F_1,26=24.9, P<0.001; interaction: F_1,26=6.2, P=0.019; Fig. 5K). The number of PV per nucleus was not different across species (F_1,26=0.05, P=0.825; Fig. 5H), but differed between ages (F_1,26=11.7, P=0.002), with no significant interaction between these factors (F_1,26=2.5, P=0.121).

Tukey post hoc tests revealed that the age difference in PNN numbers is significant in both species (cowbirds: P<0.001; redwings: P<0.05), even if there was a significant interaction between age and species. This was also the case for PV+PNN, but only in cowbirds (cowbirds: P<0.001; redwings: P>0.05). The significant interactions for PNN and PV+PNN were shown to result from species differences specifically affecting the adults (PNN: P<0.01; PV+PNN: P<0.01), but not the juveniles (PNN: P>0.05; PV+PNN: P>0.05).

Similar results were found for the densities of PNN and PV+PNN, suggesting that the age increase in PNN proportionally exceeds the increase in RA volume, as was the case in HVC. Nevertheless, there were no age differences for the density of PV, but a species difference instead, suggesting an overall higher PV density in cowbirds (Table 1).

In Area X, the number of PNN per nucleus was not significantly different across species (F_1,24=2.1, P=0.157; Fig. 5F), but was significantly different between ages (F_1,24=19.6, P<0.001), with no significant interaction between age and species (F_1,24=0.5, P=0.485). The number of PV per nucleus was not affected by species (F_1,24=3.8, P=0.064; Fig. 5I), age (F_1,24=2.7, P=0.111) or the interaction of species and age (F_1,24=0.1, P=0.758). Additionally, the number of PV+PNN per nucleus was not significantly different across species (Fig. 5L; F_1,24=0.4, P=0.536), but was significantly different between ages (F_1,24=13.2, P=0.001), with no significant interaction between the two factors (F_1,24=0.5, P=0.488). Again, the age difference reflected higher PNN and PV+PNN numbers in adults compared with juveniles.

Similarly, there was a main effect of age on the densities of PNN and PV+PNN without significant differences between species or interactions. Again, this suggests that the rise in PNN numbers exceeds the increase in volume of the nucleus. Additionally, there was a significant species difference for the density of PV cells that was higher in cowbirds (Table 1). This is possibly due to their lower nuclei volume (Fig. 5C).

In LMAN, there was no significant effect of age or species and no significant interaction between these factors in the densities of PNN, PV and PV+PNN. Nevertheless, there was a trend toward age exhibiting a meaningful difference affecting the densities of PNN (F_1,26=3.1, P=0.090) and PV+PNN (F_1,26=2.9, P=0.098), suggesting a possible increase with age (Table 1).

We also compared the densities of PNN, PV and PV+PNN between adult male and female brown-headed cowbirds because female brown-headed cowbirds do not sing and therefore serve as a reference point. In HVC, there was a trend for sex differences in the density of PNN (t₁₅=1.8, P=0.088; Fig. 6A) and PV+PNN (t₁₅=1.8, P=0.097; Fig. 6I) and a significantly higher PV density in females (t₁₅=2.5, P=0.024; Fig. 6E). In RA, the densities of PNN (t₁₅=7.0, P<0.001; Fig. 6B) and of PV+PNN (t₁₅=6.0, P<0.001; Fig. 6J) were both significantly lower in females. As in HVC, the density of PV cells in RA was significantly higher in females (t₁₅=4.3, P<0.001; Fig. 6F). The PV density in Area X was similar to that in HVC in that it was significantly higher in females (t₁₆=3.2, P=0.005; Fig. 6G), but there was no sex difference in PNN density (t₁₆=1.6, P=0.138; Fig. 6C) or PV+PNN density (t₁₆=1.0, P=0.315; Fig. 6K) in this nucleus. Finally, no sex differences in any measure were found in LMAN (PNN: t₁₆=0.6, P=0.538; PV: t₁₆=1.5, P=0.157; PV+PNN: t₁₆=1.4, P=0.174; Fig. 6D,H,K).

Finally, as we were unable to measure the volume of song control nuclei in females, except in RA, we could only compare the total numbers of PNN, PV and PV+PNN between male and female cowbirds in this nucleus. In RA, the volume was significantly larger in males compared with females (t₁₄=13.1, P<0.001; Fig. 7A) and there was significantly more PNN (t₁₄=6.4, P<0.001; Fig. 7B) and PV+PNN (t₁₄=5.4, P<0.001; Fig. 7D) in males than in females. Similarly, the numbers of PV cells counted were significantly higher in males than in females (t₁₄=5.8, P<0.001; Fig. 7C).

In the present study, we compared the development of two indicators of critical periods, PV and PNN, across two ages and two closely related icterid species: brown-headed cowbirds (i.e. brood parasite) and red-winged blackbirds (i.e. non-parasite). Both PV and PNN have been suggested as marking the onset (i.e. PV) and closure (i.e. PNN) of sensitive periods for vocal learning in songbirds (Balmer et al., 2009; Cornez et al., 2018b). PNN create a physical barrier that prevents new synaptic contacts and is assumed to control the closing of sensitive neural plasticity periods (Hensch, 2005; Karetko and Skangiel-Kramska, 2009). We used these neuroplasticity markers to understand the process of song learning in two species of song learners with striking contrast in early developmental experience. This approach provides insight into the timing of the sensitive song learning window in a brood parasitic species with exceptionally protracted delays in exposure to conspecific interactions and conspecific songs. Results revealed differences in neural plasticity patterns across two age categories and, to a lesser extent, between the two closely related species. Densities and total numbers of PNN were, in most cases, significantly greater in birds that were after their hatch year (i.e. adults) as compared with young birds in their hatch year (i.e. juveniles). Changes in PV with age were more variable. We discovered age-related differences in all three song control nuclei, HVC, RA and Area X. A similar trend occurred in the density of PNN and PV in LMAN, but there were no significant differences in age or species.

We identified an age difference in the number per nuclei for PV in HVC and RA, but not in Area X, reflecting an increase in PV expression in adults in these two nuclei. Additionally, we found a species difference reflecting higher PV density in HVC and RA of cowbirds compared with red-winged blackbirds. Moreover, in HVC, there was an interaction between age and species for this same measure. While developing cowbirds do not experience sensory deprivation sensu stricto, the age difference in PV expression is reminiscent of studies that examine the role of sensory deprivation on PV and PNN expression within the song system of zebra finches. Deprivation from a song tutor during development delays the closure of the sensitive period for vocal learning in zebra finches (Eales, 1987, 1985), which is associated with a significant decrease in the number of PV-expressing interneurons in HVC of young adults (Balmer et al., 2009). In contrast, in first-year zebra finch raised with tutors, PV expression in HVC reflects adult levels as early as day 20 post-hatching, which is the onset of sensory learning and just before the start of the sensorimotor learning phase (Cornez et al., 2018b). The PV density in HVC is high at 10 days and decreases as soon as sensory learning starts in these finches (Cornez et al., 2018b), but is significantly decreased at adulthood in acoustically isolated birds (Balmer et al., 2009), suggesting that HVC requires sensory input for normal PV levels to appear. Normal PV appearance in HVC may even require social interaction (Boseret et al., 2006), which is also lacking during cowbird development.

We also identified a greater overall PV density in HVC of cowbirds as compared with the red-winged blackbirds. Additionally, in HVC, an interaction between age and species indicates that juvenile cowbirds had a higher PV density than both juvenile red-winged blackbirds and adult cowbirds. This is similar to results seen in 10-day-old zebra finches, where PV density was also higher than in older birds (Cornez et al., 2018b), even though young birds in the present study were 30 to 60 days old. These results may also be related to species differences in early conspecific exposure. Thus, it is likely that the time of decreased PV density is related to the timing of sensory learning, as is the case in other measures of neural plasticity such as neurogenesis (Nordeen et al., 1989). Taken together, these results suggest that the high PV density observed in juvenile cowbirds may relate to a delayed sensory learning in a brood parasitic species. It is not simply a total sensory deprivation that modifies the timing of PV expression (as reported in zebra finch) but the relevance of that sensory experience (i.e. the tutor song; Balmer et al., 2009; Eales, 1987). It is the case that juvenile cowbirds do have early sensory experiences and exposure to song, but little to no exposure to conspecific songs. Moreover, there is substantial variability in the timing of juvenile cowbirds finally being exposed to conspecific songs. Such variability in exposure to conspecifics during early developmental periods may produce variability in SCS plasticity of cowbirds, depending on the social environment to which they are exposed.

Age-related differences in the total number of PNN per nucleus were also detected in both species in all three song control nuclei, and similar effects were observed in the number of PV surrounded by PNN (PV+PNN) per nucleus. In all three brain regions and in both species, PNN and PV+PNN per nucleus were significantly more numerous in adult males relative to first year males. Thus, regardless of differences in early conspecific song exposure between these species, both species exhibit an age-related increase in PNN and PV+PNN. It is not clear whether age-related increases in PNN mark the conclusion of the sensory or sensorimotor learning phase. Additional studies should isolate cowbirds from song at various periods during development and determine whether PNN marks the final stages of sensory or sensorimotor learning. Recent findings, however, indicate that full development of PNN corresponds to later stages of sensorimotor learning, well after the end of sensory learning in both zebra finches and canaries (Cornez et al., 2018a,b). A detailed time course comparison between brown-headed cowbirds and red-winged blackbirds of PNN development during the first 2 years of life would allow the determination of whether the delayed sensory learning in parasitic songbirds is also associated with a delayed PNN formation. This question cannot be answered based on the present study, which only included two time points.

Interestingly, species-specific differences in PNN and PV+PNN per nuclei specifically occurred in RA, a song control region that is implicated in the crystallization of song structure, controls the stereotypy of song syllables and directly provides motor commands to the syrinx for singing behavior (Nottebohm, 1999; Wild, 2004; Brenowitz and Beecher, 2005; Alward et al., 2017). A significant interaction between species and age was detected in RA for both PNN and PV+PNN. This interaction results from the fact that age differences in PNN and PV+PNN were markedly more pronounced in cowbirds than in red-winged blackbirds. Although the numbers of PNN in juveniles were similar in both species, the adult levels were much higher in cowbirds than in red-winged blackbirds. This somewhat surprising finding may be related to both the function of PNN and developmental differences between these species. PNNs are dependent on neural activity and they are thought to stabilize sensory circuits, affect neuronal excitability and mark the end of sensory sensitive periods (Balmer, 2016; Dityatev et al., 2007; Hensch, 2005). Although it is not clear whether PNN expression marks the end of sensory or sensorimotor periods, it is the case that PNNs stabilize neural circuits involved in song learning and species-typical song learning occurs differently in these two species. Specifically, male red-winged blackbirds continue to learn songs well into their second year of life (Marler et al., 1972) and add new song types well beyond their second year (Yasukawa et al., 1980), indicating that long-term plasticity in the neural control of song learning likely exists in this species. It was previously shown that typical open-ended learning species, such as European starlings, which also add new song elements to their song repertoire throughout their lives, also have a lower PNN density in HVC, RA, Area X and LMAN compared with zebra finches (Cornez et al., 2017a). This suggests that the lower PNN expression in RA of adult male red-winged blackbirds could be similarly related to their extensive vocal plasticity during adulthood. But why this pattern appeared only in RA, and not Area X, remains an open question, because Area X is specifically involved in song learning and manages variability during sensorimotor learning (Nottebohm, 2005; Scharff and Nottebohm, 1991). We were thus expecting also a lower PNN density in the Area X of adult red-winged blackbirds, but this is not the case. The results in RA, however, may be related to species differences in song production rather than in song learning. Because male cowbirds can develop a normal adult song in the absence of conspecific tutors and interaction with females plays an important role in the development of male songs (King and West, 1988), juveniles may develop a stereotyped song that solicits female copulation in the absence of adult male song exposure (White et al., 2002a,b). Thus, juvenile male cowbirds will attempt to sing even without a male tutor and this may be why RA, a song production-related structure, exhibits plenty of PNN in young male cowbirds.

In LMAN, there was no species difference in PNN and PV+PNN density and barely a statistical trend (P<0.10) for an age-related difference. This suggests that LMAN develops earlier than the other song nuclei and the adult level of PNN is probably almost, if not completely, attained at the age when the juvenile brains were collected. This may be similar to patterns demonstrated in zebra finches in which LMAN undergoes early changes in terms of nucleus volume, synaptic organization and PNN development that take place at the beginning of sensorimotor song learning, but not later (Cornez et al., 2018a; Miller-Sims and Bottjer, 2012; Nixdorf-Bergweiler, 1996). Together, these data indicate that delayed sensory learning in cowbirds, as compared with other icterid species such as the red-winged blackbird, is associated with a delay in the PV maturation and PNN development around PV interneurons. However, it is difficult to know whether these delays relate to sensory or sensorimotor learning, especially because cowbirds exhibit a delay in both processes underlying song learning (O'Loghlen and Rothstein, 2010, 2002, 1993).

Our results revealed species differences in PNN and PV+PNN in RA, a song production nucleus. Therefore, we explored song-dependent PNN by comparing between singing male cowbirds and non-singing female cowbirds. A robust sex difference in the density of PNN and PV+PNN was identified in RA, but not in the other SCS nuclei examined here. However, we were not able to measure the volume of HVC, Area X and LMAN in female cowbirds. We were therefore able to quantify the total numbers of PNN, PV and PV+PNN in RA only. RA was much larger in males than in females. Consequently, the sex difference in the total numbers of PNN and PV+PNN was even larger than the difference observed for the corresponding density (i.e. counts per volume). If we had been able to measure HVC volume in females, we might have also found a significant sex difference in favor of males in PNN and PV+PNN as found in RA. Indeed, it was previously shown that HVC volume measured using NeuN staining is larger in males than in females of this species (Guigueno et al., 2016), which implies that it is possible that PNN or PV+PNN expression may be different between the sexes once volume is incorporated into the measure. In contrast, no information is available concerning a possible sex difference in Area X volume.

Female cowbirds do produce vocalizations, specifically non-learned vocalizations called chatters, but they never produce species-typical songs. Because RA is involved in the motor control of species-typical song production, this might explain the large sex difference seen in RA, where almost no PNN were found in females. Because juvenile male cowbirds can learn species-typical songs in the absence of male tutors using female responses to their vocalizations (King and West, 1988; White et al., 2002a,b), it is possible that females have a representation of the male song in the song control nuclei, perhaps in the anterior forebrain pathway involved in song learning. This would allow them to tutor male song learning. In addition, PV density was larger in HVC, RA and Area X in females compared with males, but in RA, the total number of PV was larger in males. As female cowbirds do not sing, it is likely that PV expression in females relates more to song perception.

Powerful opportunities for investigating neural plasticity associated with song learning are rooted in the comparison of closely related species exhibiting striking differences in early exposure to song. These comparative studies are the first to use brood parasites as a natural system in which early song exposure is lacking. Our results reveal that plasticity-related patterns in the SCS are similar to those of closely related open-ended learners, but likely for an entirely different purpose. Non-parasitic red-winged blackbirds add songs to their repertoire even after their first year, whereas parasitic brown-headed cowbirds exhibit substantially delayed exposure to conspecific tutors and must practice their songs using social guidance from females (King and West, 1988). Thus, cowbirds do learn their song even in the absence of males, and the practice of song may explain the elevated PNN in RA, an SCS nucleus related to song production, in adult male cowbirds compared with red-winged blackbirds. Overall, these results may indicate that delays in song learning owing to a lack of early song tutoring can be co-opted from an open-ended learning system such that the delayed learner simply uses similar neural substrates with similar timing of neural plasticity.

The authors thank Melissa Malloy at New York USDA and Balcones Wildlife Refuge in Marble Falls Texas for help with bird collection. C.A.C. is a senior research associate of the Belgian Funds for Scientific Research (Fonds pour la Recherche Scientifique, FRS-FNRS).