Leiothlypis peregrina, (A. Wilson, 1811)) (A. Wilson, 1811) Resume

Poirier, Vanessa, Frei, Barbara, Lefvert, Mathilde, Morales, Ana & Elliott, Kyle H., 2024, Moult migrant Tennessee Warblers undergo extensive stopover in peri-urban forests of southern Quebec, Canadian Journal of Zoology 102 (3), pp. 272-285 : 276-278

Moult migrating Tennessee Warblers at the MBO

From 2013 to 2022 (excluding 2014 and 2020), the MBO captured 107 actively moulting Tennessee Warblers during their fall migration. The years 2019 and 2022 recorded the lowest numbers of moult migrants (only seven and eight individuals, respectively), while 2018 was the highest, having captured 27 moult migrants ( Table 1 ). Note that we found two radio-tagged moult migrants deceased in the wild. The average proportion of AHY individuals that were moulting upon capture was 73% ± 19% (n = 145, Table 1 ). The number of HY individuals caught per year (as a proxy for productivity at the breeding grounds) was not significantly correlated with the number of moult migrants (Pearson’s r = 0.277, p = 0.506) but was positively correlated with moult migrants’ median arrival date (Pearson’s r = 0.844, p = 0.008). Finally, years with high numbers of AHY individuals had higher numbers of moult migrants as they were strongly correlated (Pearson’s r = 0.831, p = 0.010). Note that a similar table including data from 2013 to 2022 (excluding 2014 and 2020) can be found in the Supplementary material (see Table S1 ).

Moult rate and stopover duration

Across all years (2013–2022), moult migrants were most frequently captured in early August (mode: 10 August; median: 17 August, n = 107). Upon capture, the average individual had completed 39% ± 38% of their moult. The average moult rate for Tennessee Warblers was 2.5% ± 0.6% per day (n = 72, Fig. 1 ). Based on this estimation, the median arrival date for moult migrants across all sampling years (from 2013 to 2022) was 6 August (mode: 4 August, n = 107). The median arrival date for moult migrants did not differ significantly between years (Kruskal– Wallis test, including data from 2013 to 2022; p = 0.068, Χ 2 [7] = 13.2, n = 107) nor did it differ for post-moult migrants (p = 0.076, Χ 2 [6] = 11.4, n = 38). Median arrival date between moult and post-moult migrants, however, was significantly different (p <0.001, W [1] = 5980, n = 145). While moult migrants arrived in early August, post-moult migrants arrived more than a month later in September (median: 14 September, n = 38; see Table 1 ).

Departure dates were statistically similar between moult and post-moult migrants (p = 0.89, W [1] = 36, n = 22). Note that departure date calculations are biased as moult migrants were much more numerous (18 compared to 4 post-moult migrants) and most were tagged in 2018 (n = 14), whereas post-moult migrants were more evenly distributed between years (see Fig. 2 ). The median and mode departure date for moult migrants was 16 September (n = 18), while the median departure date for post-moult migrants was 20 September (n = 4, with no mode; Fig. 2 ). Among both moult and post-moult migrants, year had no effect on departure dates (Kruskal– Wallis test: p = 0.158, Χ 2 [3] = 5.19, n = 22).

Assuming individuals arrived at the start of their moult and maintained a constant 2.5% moult rate throughout their stay, moult migrants spent 46 ± 5 days (n = 18) at their stopover, while post-moult migrants stayed for 8 ± 6 days (n = 4, see Fig. 3 ). The minimum stay was 39 days (in 2018), while the maximum was 56 days (in 2022). According to the estimated moult rate (2.5%/day), it would take a bird 40 days to moult their feathers, which was near the average stay of moult migrants.

Stopover home ranges

The one post-moult migrant had an estimated 95% stopover home range of 5.79 ha (with 95% confidence intervals extending from 1.58 to 12.70 ha), while the moult migrants (n = 16) had an average 95% stopover home range size of 14.6 ha (with standard error of ± 4.4 ha). The largest stopover home range across all tagged Tennessee Warblers was 58.8 ha and the smallest was 1.33 ha, both obtained from moult migrants in 2022 and 2021, respectively (see Table S 2 in the Supplemental material). Figure. 4B shows a map of the stopover home ranges.

Almost half (48.83% ± 21.43%) of the Tennessee Warblers’ stopover home ranges were composed of forests ( Fig. 4B ). More than a third (33.03% ± 25.25%, n = 17) of the area spanned agricultural land. Water and wetland types covered 10.12% ± 11.20% and 7.31% ± 5.46% of the total area of the stopover home ranges, respectively. The least prevalent land cover type over which the stopover home ranges were found was anthropogenically modified habitat (excluding agriculture), which represented 0.71% ± 1.99% of the total area. See Table S 3 in the Supplementary material for a summary of the land types present in each bird’s stopover home range. We found no correlation between the proportion of forested stopover home range and stopover duration (p = 1, W [16] = 32, R 2 = 0.164, n = 8) nor between stopover duration and stopover home range size (p = 0.959, W [16] = 31, R 2 = 0.095, n = 8).

Use–availability analysis: migrants’ resource selection

A comparison of the global models at different buffer sizes (50, 100, and 200 m) revealed that the 100 m radius buffer was the best (>2.0 ΔAICc between all other models). The models discussed in this section therefore include landscape measurements obtained from a 100 m radius buffer around the used and available points.

The forest model (including the proportion of forest and length of forest edge (m)) and the global model (including the same variables plus percent anthropogenic land) were the two best models (>2.0 ΔAICc between all other models) to predict resource selection of migrating Tennessee Warblers ( Table 2 ). In both models, the proportion of forest and length of forest edge had a significantly positive influence on Tennessee Warbler presence (p <0.001, see Table 3 and Fig. 5 ). The conditional version of the forest model including bird ID as a random effect had a higher theoretical R 2 (=0.82) than the marginal model excluding bird ID (R 2 = 0.69). These two best models also had high scores following the AUC goodness of fit test for discrimination (AUC = 0.942). Both the forest and global models, however, failed the Hosmer–Lemeshow goodness of fit test for calibration (Χ 2 [8] = 36500 and 37100, n = 702 GPS points from 18 individuals, and p <0.001, respectively).

Departure decisions

The null model, which included only calendar date, year, and moult status as random effects, was the highest ranked model for predicting a migrant’s departure ( Table 4). Although the exogenous model was ranked secondbest (within <2.0 AICc of the null model), it explained little variability in the data. None of the predictor variables in any of the models was significant (p> 0.05), and the goodness of fit measurement of concordance for all models was similar to the null model (C = 0.5).

Figures

Fig. 1. Tennessee Warblers’ stage of moult (i.e., percent of moult completed) over time during their fall migration stopover at the McGill Bird Observatory (MBO). (A) Visual representation of Tennessee Warblers’ flight feather moult stages (photos taken during 2021 and 2022 fall migration). (B) Graph showing the moult stage of individuals upon capture. Each point represents a date when an individual was captured at the MBO and their moult stage was recorded. This figure represents 72 individuals across 8 years (from 2013 to 2022).

Fig. 2. Barplot showing the number of radio-tagged Tennessee Warblers making migratory departures from the McGill Bird Observatory stopover on any given night during the fall season. Data are color-coded by year, and moult versus post-moult migrants are differentiated by pattern as seen in the legend.

Fig. 3. A boxplot illustrating the difference in stopover duration between moult and post-moult migrating Tennessee Warblers at the McGill Bird Observatory in southern Quebec. Stopover duration for post-moult migrants is defined as the number of days between their first capture and departure dates, while for moult migrants, it is their estimated arrival (based on their moult stage and an estimated moult rate (2.5%/day), assuming migrants arrived at the start of their moult) and their departure dates.

Fig. 4. (A) Map of the West Island of Montreal showcasing the McGill Bird Observatory (MBO) and Motus tower locations. In the bottom left corner shows where the West Island of Montreal is located in northeastern North America. The blue circle indicates the area represented in map “B”. (B) Map of the 95% stopover home ranges (and their associated GPS points) of moult and post-moult migrating Tennessee Warblers at the MBO. Individuals were radio-tagged and tracked in 2018, 2019, 2021, and 2022. Both maps use a 30 m resolution landscape cover layer from 2019 (obtained from Partenariat Données Québec). Map projection is WGS84, and the coordinate system used is UTM zone 11. Inset map data provided by Google © 2023, INEGI.

Fig. 5. Probability of presence of migrating Tennessee Warblers within a 100 m radius area as predicted by a generalized linear mixed effects model. The model’s predictors include the proportion of forest and wetland, and the length of forest edge (in meters), and individual (bird ID) as a random effect. (A) shows the positive relationship between migrants’ presence and the amount of forest edge (coefficient = 3.27, p <0.001) and (B) shows the positive relationship between migrants’ presence and the proportion of forest (coefficient = 0.55, p <0.001).