Sexual habitat segregation in migrant warblers along a shade gradient of Jamaican coffee farms

Shade coffee has been shown to provide wintering habitat for migratory bird species. Canopy cover has been cited as a good indicator of habitat quality on coffee farms. Many species of sexually dimorphic migrant warblers show sexual hab - itat segregation on their wintering grounds, often related to habitat quality. To test if degree of habitat segregation followed a gradient of habitat quality, we sampled canopy cover and avian communities at nine different sites in coffee farms ranging from full sun to highly shaded. Proportion of all species of wintering warblers that were male was positively correlated with per - cent canopy cover. Species-specific male proportions of American Redstarts ( Setophaga ruticilla ), Black-throated Blue Warblers ( S. caerulescens ), and Prairie Warblers ( S. discolor ) showed positive correlations with canopy cover, though small sample sizes limited statistical power.

ever, numerous species appear amenable to habitat modification in tropical agricultural landscapes if farms contain adequate tree cover and are embedded in areas with patches of residual forests (Wunderle andWaide 1993, Petit et al. 1995).This is especially apparent in shade coffee farms in the Neotropics, which provide refugia for local biodiversity (Perfecto et al. 1996, Jha et al. 2014) and habitat for wintering warblers in the family Parulidae (Greenberg et al. 1997, Sherry 2000).
Warblers are largely insectivorous (Lack 1976) and can provide a valuable pest-control service for farmers (Greenberg et al. 2000, Kellermann et al. 2008, Karp et al. 2013, Railsback and Johnson 2014).On Jamaican coffee farms, birds have been shown to provide an economic benefit of up to US$310/ha (Johnson et al. 2010).Higher quality habitats have been shown to support more birds (Johnson et al. 2006), thus potentially providing a greater service to the farmer.Habitat quality in coffee plantations has typically been associated with the extent of canopy cover and other vegetation complexity (Greenberg et al. 1997, Tejeda-Cruz andSutherland 2004), but avian-based characteristics of good habitat in coffee are not well defined (Johnson et al. 2006).
Many species of sexually dimorphic migratory parulid warblers display habitat segregation on wintering grounds, with males and older individuals typically occupying higher quality habitats (Lynch et al. 1985, Ornat and Greenberg 1990, Marra et al. 1993, Wunderle 1995, Marra and Holmes 2001).According to this theory of habitat-mediated behavioral dominance, the age and sex ratios of warblers should vary along a gradient of habitat quality.Coffee farms fall on a gradient from full sun to highly shaded (Moguel and Toledo 1999).Since highly shaded farms have been suggested to contain higher quality habitat, there should be an older or more male-biased population, or both, on these farms than on sun coffee farms.In this study, we tested for this pattern by examining age and sex ratios of migratory warblers along a gradient of canopy cover in Jamaican coffee farms.

Study Area
Nine study sites were selected from eight different coffee farms in Jamaica, West Indies (Appendix 1).Farms were located in major coffee growing regions across the island.Shade regimes varied from full sun to highly shaded.Two sites were selected on the Baronhall Estate, which is one of the largest plantations on the island and has markedly different shade regimes in different portions of the farm.The two sites on this farm were more than 1 km apart and differed strongly in shade cover.All sites were within 1 km of patches of disturbed second growth forest.
Study sites varied in the presence and abundance of shade trees.Common species of hardwood shade trees included Inga vera, Cedrela odorata, Spathodea campanulata, Mangifera indica, Gliricidia sepium, and Leucaena leucocephala.Most farms also had banana trees (Musa sp.) planted sporadically between rows.Coffee trees were generally pruned to stand between 1.8 and 3.0 m tall, with about 1.5 m between rows.Ground cover varied from bare soil to grass to a duff layer.All farms applied the insecticide Suldan (fenitrothion) locally to combat pests, especially the coffee berry borer (Hypothenemus hampei).Applications were made during the fruit ripening stages, and no insecticides were applied during our study.Several farms at higher elevations also used copper-based fungicides.

Vegetation and Bird Surveys
Data were collected in late December and January in 2010-2011 and 2011-2012.Canopy cover was quantified using a spherical densiometer (Lemmon 1956), which involves tallying cover from an image of the canopy projected onto a concave mirror.For the 2010-2011 field season, vegetation data were collected at five random locations within each site.For the 2011-2012 season, we sampled vegetation at each net (n = 12-15).
We sampled birds using mist nets (9 or 12 m, 30 mm or 36 mm mesh; n = 12-15) deployed to thoroughly sample 3-5 ha of each coffee farm.Nets were at least 10 m apart and precise locations were chosen to optimize bird capture success.Nets were oriented either parallel or perpendicular to coffee rows.We opened nets shortly after sunrise for at least 5 hr.Nets were operated for three consecutive days at each site, except at Forres Park (two days) and Abbey Greene (three non-consecutive days due to rain).Each bird was batch-marked by clipping about 0.5 cm off of a tail feather.A different tail feather was used for each day, allowing identification of within-and between-day recaptures.We determined age class and sex of migrant warblers based on plumage characteristics (Pyle 1997).We used the term "adult" to describe individuals with plumages indicative of being in Jamaica for their second or later winter (designated as afterhatch-year, AHY, before 1 January or after-second-year, ASY, after 1 January), and we used the term "young" to describe individuals with plumages indicative of being in their first winter in Jamaica (designated as hatch-year, HY, before 1 January or second-year, SY, after 1 January; Pyle 1997).

Analyses
We quantified migrant population structure by calculating the proportion of captured individuals that were males and those that were adult birds.Young male American Redstarts (Setophaga ruticilla) exhibit delayed plumage maturation, and like females they are behaviorally subordinate to adult males, so they were grouped with females for analyses (Marra and Holmes 2001).The overall proportion of males for migrant species was obtained at the site level by dividing the total number of males by the total number of sexed birds.For three of the most common wintering migrants on Jamaica (American Redstart, Black-throated Blue Warbler [S.caerulescens], and Prairie Warbler [S.discolor]), species-specific male proportions were calculated in the same manner.Only farms that had at least three known-sex individuals (of the species in question) were included in the species-specific analysis.We calculated site-level age proportions by dividing the number of adult birds by the total number of birds assigned an age class.We then ran Spearman's rank-order correlations on each of these sex and age proportions with percent canopy cover.We set alpha at 0.10 due to a low sample size (nine sites), which increased chances of making a type 1 (false positive) error, but it decreased the likelihood of making a type 2 (false negative) error and is appropriate given the importance of understanding variation in habitat quality in human disturbed habitat for declining species (Askins et al. 1990).One farm, Wallenford, had very few shade trees and scored 0.0% canopy cover with our methods (see Results).This outlier could have a large effect on statistical significance, so we ran the correlation tests with and without this farm included.

Results
Nets were run for 792 net-hours in the 2010-2011 field season and 586 net-hours in 2011-2012, with 298 total captures of 14 species of migratory warblers, of which 216 were assigned a sex.The most common migrant captures were Black-throated Blue Warbler, American Redstart, and Prairie Warbler (Table 1).There were 82 between-day recaptures in 2010-2011 and 33 in 2011-2012; these were excluded from analyses.
Canopy cover varied from 0.0% to 89.0% with a mean of 56.7% (Table 1).Site-level overall proportions of males and adults varied from 0.21 to 0.67 and 0.27 to 0.79, with means of 0.52 and 0.50, respectively (Table 2).The average proportions of Black-throated Blue and Prairie Warblers that were male were 0.54 and 0.86, respectively.Average proportion of adult male American Redstarts was 0.39 (Table 2).
There were no significant correlations between proportion adult migrants and percent canopy cover for any species, nor all species combined (all r s < 0.5, all p > 0.28).Across all species, there was a significant correlation between proportion male and percent canopy cover (r s = 0.61, p = 0.08, n = 9; Fig. 1a).Similarly, the proportion male increased significantly with canopy cover for Black-throated Blue Warbler (r s = 0.60, p = 0.08, n = 9; Fig. 1b).Correlations between the proportion male and canopy cover were also positive for American Redstart (r s = 0.80, p = 0.20, n = 4; Fig. 1c) and Prairie Warbler (r s = 0.67, p = 0.22, n = 5; Fig. 1d), but these relationships were not statistically significant.When  the Wallenford farm was removed from analyses, all regression coefficients remained positive but were lower in magnitude (r s = 0.49, 0.50, 0.43, and 0.26 for all species, American Redstart, Black-throated Blue Warbler, and Prairie Warbler, respectively), and none of the relationships were statistically significant (all p > 0.21).

Discussion
Research on habitat segregation of wintering warblers has focused on highly contrasting habitats (Marra et al. 1993, Wunderle 1995, Marra and Holmes 2001, Studds and Marra 2005, Reudink et al. 2009).Results presented here are not strong; our small number of study sites limited statistical power, and the small number of individual birds of some species at some sites reduced precision in estimating the proportion male.Nonetheless, the proportions of males of all three species examined (American Redstart, Black-throated Blue Warbler, and Prairie Warbler) appeared to increase with increasing canopy cover, suggesting that segregation may also operate within a single habitat (coffee farms) along a gradient of habitat quality.This may merit further study.The trends we observed were driven especially by the low canopy cover and male proportions at the Wallenford farm, which was nearly a pure sun coffee farm.Future work will benefit from the inclusion of more intermediate canopy cover values, such as between 0 and 40% cover.
Low proportions of males in farms with little shade are consistent with the notion that shade coffee provides better habitat for wintering parulid warblers than does sun coffee (Greenberg et al. 1997, Sherry 2000, Tejeda-Cruz and Sutherland 2004).This trend was most conspicuous in American Redstarts and Blackthroated Blue Warblers.Behaviorally-mediated sexual habitat segregation can be signaled by plumage variation among dominance classes (Marra and Holmes 2001), so it is interesting to note that the relationship between canopy cover and proportion male was weakest with Prairie Warblers, which are not as drastically dimorphic as either American Redstarts or Black-throated Blue Warblers.
Sexual segregation in wintering warblers has been explained as either an innate difference in habitat preference (Hooded Warblers [Setophaga citrina]; Morton 1990) or as the result of competitive exclusion by males for sites of higher quality, such as those with more food (American Redstarts; Marra et al. 1993).In coffee farms, food availability for insectivores is significantly higher in the canopy of shade trees than in the coffee shrub understory (Greenberg et al. 1997, Smith et al. 2012).Therefore, the patterns of male-biased populations in farms with higher canopy cover is consistent with the hypothesis of behavioral dominance for food-rich sites.Additional behavioral research on whether warblers exhibit vertical stratification within a habitat (canopy versus understory) may further reveal the extent of intersexual aggression.Regardless of the mechanism, our results suggest that the degree of sex class segregation of migrant warblers can be indicative of habitat quality on Jamaican coffee farms.However, there were myriad other environmental variables we did not include in our analyses that could also affect habitat quality and influence the proportion of males in a migrant population, such as distance to nearby forests, variation in insect abundance owing to differences in coffee shrub or ground cover layer, variation in shade tree species, and so on.Our results are generally consistent with the hypothesis that the proportion male could provide an index of habitat quality for Jamaican coffee farms, but confirmation awaits more controlled studies or experimentation.
Two migrant species (Palm Warbler [Setophaga palmarum] and Prairie Warbler) were most abundant at Wallenford, the site with the least shade.This is not surprising as both species are at least moderately specialized for disturbed or early succession scrub-type habitats (Wunderle and Waide 1993), which a sun coffee farm resembles.The second highest number of captures of American Redstart was also at Wallenford; these individuals were overwhelmingly female-type (Table 2).This magnitude of captures was slightly unexpected but other studies have also found large numbers of females in similar habitats (Ornat andGreenberg 1990, Marra andHolmes 2001), which may reflect an alternative strategy for obtaining food, especially for subordinate individuals.

Fig. 1 .
Fig. 1.Associations between shade and sex ratios of (a) all species of migrant warblers, (b) Black-throated Blue Warblers, (c) American Redstarts, (d) and Prairie Warblers on coffee farms in Jamaica, West Indies.Lines indicate linear correlations.

Table 1 .
Number of migrants captured during 2010Number of migrants captured during  -2011Number of migrants captured during   and 2011Number of migrants captured during  -2012winter field seasons at nine coffee farm sites in Jamaica, West Indies.Canopy cover from spherical densiometer also presented.

Table 2 .
Population structure of wintering warblers at nine coffee farm sites in Jamaica, West Indies.
a Proportion of adult males.b Excluded from further analyses due to insufficient sample size.