Introduction
Many of the world's nocturnal birds are in the family Caprimulgidae (Sibley and Monroe Reference Sibley and Monroe1990). Species of the genus Caprimulgus are generally ground nesters associated with forested habitats. However, several African species in this genus (e.g., Caprimulgus aegyptius, C. eximius) are known to nest in open and forest-savanna habitats (Fry et al. Reference Fry, Keith and Urban1988). Caprimulgids lay their eggs directly onto a particular substrate (i.e., sand, rock, tree branch) without building a nest. Chicks of caprimulgids are semi-precocial and move repeatedly from the nest site shortly after hatching (Holyoak Reference Holyoak2001, Aragonés Reference Aragonés2003). The nocturnal habits and cryptic plumages of caprimulgids make them difficult to study. Most reports on the nesting habits of Neotropical caprimulgids have consisted of descriptive accounts of nests and breeding behaviour, some of which were accidentally discovered (Tate Reference Tate1994, Forcey Reference Forcey2002, Delannoy Reference Delannoy2005). Conversely, a number of studies on nesting behaviour, habitat use and management of Nearctic caprimulgids have been published (Mills Reference Mills1986, Grand and Cushman Reference Grand and Cushman2003).
The Puerto Rican Nightjar Caprimulgus noctitherus, hereafter termed nightjar, is endemic to coastal dry and lower montane forests of southwest Puerto Rico. This single-island endemic is presently listed by the Puerto Rico Department of Natural and Environmental Resources (DNER), the U.S. Fish and Wildlife Service (FWS), and the International Union for the Conservation of Nature (IUCN) as ‘Critically Endangered’ throughout its range (Diaz Reference Diaz1983, BirdLife International 2004). The available information on the nesting biology of the nightjar is limited (Kepler and Kepler Reference Kepler and Kepler1973, Vilella Reference Vilella1995, Delannoy Reference Delannoy2005). Moreover, published information on nest habitat use patterns of the nightjar is not available. Here I report on nest habitat use of the nightjar at Guánica Biosphere Reserve, Puerto Rico, and discuss implications for habitat management and conservation for this critically endangered species.
Study Site
Data on the reproductive ecology of the nightjar were collected on the section of Guánica Forest east of Guánica Bay (17°57’56” N, 66°52’44” W). This forest reserve was first established in 1917; however, timber extraction, charcoal production, grazing, and subsistence farming persisted in some areas of the current reserve through the 1940s (Molina-Colón and Lugo Reference Molina-Colón and Lugo2006). In 1982, protection and management efforts were increased when Guánica Forest became a part of the UNESCO Man and the Biosphere programme (Canals Reference Canals1990).
The forest is underlain by limestone that surfaces in some areas, and is virtually the only substrate for plant growth at elevations below 80 m (Roberts Reference Roberts1942). The dominant soil type is of calcareous origin with drainage areas dominated by clay soils. Soils are rocky and have low moisture retention. The climate of Guánica Forest is dry, with short periods of water recharge. Rainfall is seasonally variable (Murphy and Lugo Reference Murphy and Lugo1986). More than half of the precipitation falls between August and November, with a small secondary peak occurring in May. The dry season extends from January to May; precipitation does not exceed 30 mm during this period.
The Guánica Forest is classified in the Subtropical Dry Forest Life Zone (Ewel and Whitmore Reference Ewel and Whitmore1973). The vegetation of this region has been described by a number of authors (Gleason and Cooke Reference Gleason and Cooke1927, Wadsworth Reference Wadsworth1950, Little and Wadsworth Reference Little and Wadsworth1964, Kepler and Kepler Reference Kepler and Kepler1973, Little et al. Reference Little, Woodbury and Wadsworth1974, Gonzalez-Liboy et al. Reference González-Liboy, Cintrón and Lugo1976, Lugo et al. Reference Lugo, Ramos, Molina, Scatena and Vélez-Rodríguez1996). For this study, habitat descriptions followed Lugo et al. (Reference Lugo, González-Liboy, Cintrón and Dugger1978) and plant taxonomy followed Liogier and Martorell (Reference Liogier and Martorell1982). Approximately 246 tree species have been recorded at Guánica Forest; 35% of which are deciduous during the dry season. The dominant plant families are Fabaceae, Euphorbiaceae and Myrtaceae (Silander et al. Reference Silander, Gil de Rubio, Miranda and Vázquez1986).
The boundary of the deciduous forest was characterized by shallower slopes where the tree species Bursera simaruba and Bucida buceras constituted the emergent overstorey (Figure 1). This association was also characterized by the layering of the forest, with a shorter layer of shrubs and trees found under the emerging canopy. The most common plant species of the midstorey included Coccoloba microstachya, C. krugii, Colubrina elliptica, Plumeria alba, Capparis spp., and Pisonia albida (González-Liboy et al. Reference González-Liboy, Cintrón and Lugo1976).
Figure 1. Map of the Guánica Biosphere Reserve indicating major habitat types. Vegetation classification follows Lugo et al. (Reference Lugo, González-Liboy, Cintrón and Dugger1978).
The evergreen forest association occurred in areas with higher soil moisture. This habitat type was dominated by some of the same species found in the deciduous forest plus evergreen species such as Krugiodendron ferreum, Amyris elemifera, Guaiacum officinale, G. sanctum, Coccoloba diversifolia, and several species of Eugenia. In some areas presence of the grasses Panicum maximum, Uniola virgata, and Aristida adcensionis, the shrubs Lantana involucrata, Croton spp., and the trees Leucaena leucocephala and Prosopis juliflora are indicative of fire and grazing disturbance (Canals Reference Canals1990). During the 1930s, higher elevation areas of evergreen forest (Figure 1) were planted with Dominican Mahogany (Sweetenia mahogany) and Logwood (Haematoxyumn campechianum). Over time these have developed into areas of mature evergreen forest dominated by a mahogany and logwood overstorey.
Methods
I collected data during 1985–1987 on nightjar nest habitat within Guánica Forest along all existing footpaths, hiking trails and vehicle trails. See Vilella (Reference Vilella1995) for a detailed description of nightjar nest searching procedures. Vegetation data were collected at each nest site and an equivalent number of randomly selected sites in order to characterise nightjar nesting habitat. Random sites were selected using a stratified random sampling scheme (Scheiner and Gurevitch Reference Scheiner and Gurevitch2001). Plots were chosen from the two dominant upland habitat types where nightjars nest (deciduous forest and evergreen forest) by overlaying a vegetation map with a numbered grid, and selecting sites using a random numbers table. Nest sites were selected by designating the location of the nest as the centre of the plot.
Due to the paucity of information on nightjar nest sites I decided to record as much information as possible. Elevation at each vegetation plot was recorded using a Thommen® altimeter to the nearest metre. Vegetation data were collected for overstorey, midstorey and understorey using a modified nested circular plot (Litvaitis et al. Reference Litvaitis, Titus, Anderson and Bookhout1994). The overstorey was recorded as all vegetation over 2 m tall with a 20 cm or greater diameter at breast height (dbh) and was sampled on a 25 m diameter plot (0.05 ha); taxonomic composition, stem density, dbh (cm) classes (20–39.9, 40–59.9, 60–100, > 100) and height of the tallest tree were recorded. Canopy closure was visually estimated from plot centre.
Midstorey vegetation (vegetation 1–2 m tall) was sampled on a 2.5 m diameter plot (0.005 ha). Taxonomic composition and stem density were recorded for the midstorey. Understorey vegetation (plants < 1 m tall) was sampled on a 0.25 m diameter plot (0.0005 ha). Taxonomic composition, cover, distance to nearest tree (metres), and soil condition were recorded in the understorey. Several measurements were taken to quantify microhabitat: cover found within the first 50 cm above plot center, the amount of limestone cover, and leaf litter from a 30 cm diameter circle at the plot centre. Dry weight and composition were obtained for each leaf litter sample.
I tested habitat variables for normality using a Shapiro-Wilk's test. I then used a two-step approach to model nightjar nest habitat. To reduce the variable set I used a one-way ANOVA or Kruskal-Wallis test (as appropriate) to univariately compare nightjar nest site variables with corresponding variables for random sites. For these analyses, each habitat variable was the dependent variable and variable type (random versus nest) was the main effect. Variables that did not differ significantly were not further considered. To avoid excluding biologically important variables, I set a significance level for univariate comparisons at 0.10 (Pielou Reference Pielou1977).
The second step was to develop a nightjar nest habitat model using logistic regression by comparing nest sites to random sites (Brennan et al. Reference Brennan, Block, Gutierrez, Verner, Morrison and Ralph1986, Morrison et al. Reference Morrison, Marcot and Mannan2006). My intention was to assess which variables best described a nest (albeit the small sample) within the context of the surrounding habitat. The reduced variable set was used for the logistic regression as multivariate analyses are sensitive to ratios of variable to sample size, which can lead to unstable classification rates (Brennan et al. Reference Brennan, Block, Gutierrez, Verner, Morrison and Ralph1986, Williams et al. Reference Williams, Titus and Hines1990, Anderson et al. Reference Anderson, Burnham, Gould and Cherry2001). I used logistic regression because I was interested in a binary response variable (used vs. random), and because logistic regression is not constrained by the assumptions of multivariate normality of independent variables nor equality of the covariance matrices among each grouping variable (Agresti Reference Agresti1996, Hosmer and Lemeshow Reference Hosmer and Lemeshow2000). I paired each nest site observation with a random observation so that prior probability of group membership was 0.5 (Hosmer and Lemeshow Reference Hosmer and Lemeshow2000). I used a stepwise variable selection procedure to reduce model overspecification and multicolinearity (Myers Reference Myers1990). Probability to enter to model was 0.25 and probability to stay was 0.05. Means are reported as ± standard deviation. All data were analysed using the SAS system (SAS Institute 2004).
Results
A total of 23 nightjar nests were located during the study (Vilella Reference Vilella1995). Vegetation was quantified at only 22 nest plots as nightjars nested in the same exact location in two successive years. However, I could not determine whether these were the same individual nightjars as birds were not marked. Nightjars nested at elevations ranging from 55 m at the edge of the evergreen and limestone scrub associations to 220 m in deciduous forest near the highest elevations of Guánica Forest. Nightjar nests were located from 2–125 m (
= 32.6 ± 28 m) into the forest from the nearest road or trail. Overall habitat characteristics of nightjar nests and random sites are presented in Table 1. Ten nests were located in evergreen forest, six of these were found in relict mahogany plantations. The remaining 13 nests were found in deciduous forest (Table 2).
Table 1. Nest habitat characteristics (mean ± SD, range) measured within nightjar nests and random sites in Guánica Forest, Puerto Rico.
a Significant One-way ANOVA (P < 0.05)
b Significant Kruskal-Wallis ANOVA (P < 0.05)
Table 2. Habitat characteristics of nightjar nests in the evergreen forest association at Guánica Biosphere Reserve, Puerto Rico.
All nightjar nests were located within a metre of the base of a small tree. Nearest tree species in evergreen forest included Sweetenia mahogany, Albizia lebbeck, Bursera simaruba, Eugenia rhombea, Haematoxylum campechianum, and Euphorbia petiolaris. Leaf litter biomass of nightjar nests in evergreen forest ranged from 29.2 g to 72.6 g (
= 50.7 ± 13.8 g) and was higher than nests in deciduous forest (Table 2). The canopy of nest sites located in evergreen forest where mahogany was not present was dominated by Guaiacum officinale and Bourreria suculenta. Overstorey trees ranged from 20 cm to 120 cm dbh (
= 44.2 ± 21.7 cm). Canopy height ranged from 6–12 m (
= 8.1 ± 1.8 m) and canopy closure ranged from 40% to 90% (
= 68.5%).
Nearest trees in deciduous forest included Exostema caribaeum, Thouinia portoricensis, Capparis cynophallophora, Leptocereus quadricostatus, Leucaena glabra, Reynosa uncinata, Eugenia rhombea, and Pisonia albida (Table 3). Leaf litter biomass of nests in deciduous forest ranged from 19.7 g to 51 g (
= 31.4 ± 9.2 g/sample). The overstorey of nightjar nests in deciduous forest was dominated by the trees Bucida buceras and Bursera simaruba (Table 3). However, the canopy of deciduous forest at Guánica was not always well defined and consisted in some places of a few emergent trees protruding over the midstorey. Emergent trees ranged in height from 6–11 m (
= 7.9 ± 1.4 m).
Table 3. Habitat characteristics of nightjar nests in the deciduous forest association at Guánica Biosphere Reserve, Puerto Rico.
a Same nest site location used by nesting nightjars in successive years.
Of the 13 structural habitat variables examined at nightjar nests and random sites; leaf litter biomass, height of the nearest tree, and midstorey species richness were normally distributed (P > 0.20); the remaining variables failed the normality test. Univariate analysis of variance indicated four of 13 habitat variables differed between nests and random sites (Table 1). These included elevation (H 1 = 4.09; P = 0.043), leaf litter biomass (F 1,42 = 27.9; P < 0.00001), midstorey number of stems (F 1,42 = 3.17; P = 0.042), and canopy closure (H 1 = 7.02; P = 0.008). Thus, a 4-variable logistic regression model was utilised to assess which habitat variables best described a nightjar nest. Logistic regression produced a best nest site model containing 3 variables; leaf litter biomass (parameter = 0.364, SE = 0.143, P = 0.011), midstorey stem density (parameter = −0.104, SE = 0.0592, P = 0.05), and canopy closure (parameter = 0.084, SE = 0.048, P = 0.042). Overall classification rate was 79.5%; the model variable combination correctly classified 77.3% of nightjar nests and 81.8% of random sites.
Discussion
No nightjar nests have ever been reported from the lower elevation coastal scrub forest of Guánica Biosphere Reserve (Kepler and Kepler Reference Kepler and Kepler1973, Vilella Reference Vilella1995). At lower elevations (< 50 m) the substrate was exposed limestone with occasional solution holes and shallow soil pockets. As a result, these areas had few widely spaced, large deciduous trees (e.g., Bursera simaruba) interspersed among which were many shrubs, grasses, cacti, and areas of exposed limestone, providing virtually no usable nesting habitat for nightjars. Above 75 m elevation as the soil became deeper, deciduous tree species and evergreen species were located in deeper red soils (Murphy and Lugo Reference Murphy and Lugo1986).
Leaf litter biomass, midstorey stem density, and canopy closure best predicted the occurrence of nightjar nests. This suggests sites in Guánica Forest characterised by deep leaf litter, an open midstorey, and a closed canopy constitute optimal nightjar nesting habitat (Table 1). Moreover, breeding nightjars selected structural habitat features associated with the first 2 m above the forest floor. Habitat attributes such as structure of the overstorey contributed less to the variation between nest and random sites. At Guánica Forest the availability of leaf litter and structural complexity of the vegetation increased with elevation and soil development. Lugo et al. (Reference Lugo, González-Liboy, Cintrón and Dugger1978) reported the highest amounts of total litter at Guánica Forest were found in mahogany plantations (1.48 g m−2/day) followed by deciduous forest (0.79 g m−2/day). Nightjars frequently nest in the relict mahogany plantations of Guánica Biosphere Reserve. The openness of the lower layers of the forest constitutes a major requirement of suitable nightjar nesting habitat and adult nightjars foraged primarily in these layers. Areas of open vegetation near the ground also offered suitable foraging habitat to fledged young, who remain within the parent's territory for up to a month after fledging (Vilella Reference Vilella1995). A number of locations searched at Guánica Forest provided no evidence of nightjar nesting activity. Many of these areas were characterised by a dense, tangled understorey and midstorey dominated by Lantana involucrata and Croton rigidus, typical of disturbed areas. At Guánica Forest, the structure of the lower layers of the forest is believed to be directly related to the intensity of past disturbance and the amount of time a given area has been protected (Canals Reference Canals1990, Molina-Colón and Lugo Reference Molina-Colón and Lugo2006).
Resident nightjar congeners of the Caribbean include Cuban Nightjar Caprimulgus cubanensis, endemic to Cuba, Hispaniolan Nightjar Caprimulgus ekmani endemic to Hispaniola, Caprimulgus rufus otiosus limited to the island of St. Lucia, and White-tailed Nightjar Caprimulgus cayennensis restricted to the island of Martinique. With the exception of the White-tailed Nightjar, which nests in open grassy fields in Martinique, all other Caribbean caprimulgids are associated with forest habitats, particularly semi-arid forests (Garrido Reference Garrido1983, Raffaele et al. Reference Raffaele, Wiley, Garrido, Keith and Raffaele1998). In St. Lucia, C. rufus otiosus is most abundant in the coastal dry forests around Grand Anse. The vegetation in this region is very similar in structure and composition to Guánica Forest (Vilella Reference Vilella1993). Unfortunately, quantitative information on nest habitat use of other Caribbean forest caprimulgids is lacking.
At the time of its discovery, the island of Puerto Rico was practically totally forested (Wadsworth Reference Wadsworth1950). However, by the early decades of the 20th century deforestation had peaked and less than 10% of the island remained forested. Guánica Forest was heavily cut during this period and available nightjar nesting habitat must have been greatly reduced (Figure 2). It is possible that the relatively small spatial requirements (Vilella Reference Vilella1995) of nightjar breeding pairs may have helped the species survive periods of severe habitat destruction. Moreover, tree plantations established during the 1930s may have provided patches of closed canopy forest allowing nightjar breeding populations to persist (Figure 2). These plantation forest stands included the most important requirements for adequate nightjar breeding habitat; abundant leaf litter, little or no vegetation near the ground, and a closed canopy. At present, suitable nesting habitat exists at higher elevations in naturally regenerated areas of evergreen and deciduous forest without relict plantations (Molina-Colón and Lugo Reference Molina-Colón and Lugo2006).
Figure 2. Past and present at the Guánica Biosphere Reserve. A 1931 photograph (top) shows a mahogany plantation amid extensive deforestation near the present reserve headquarters (Photo courtesy Puerto Rico Department of Agriculture). Bottom photograph of the same area under present conditions showing the abandoned mahogany plantation (centre left) now embedded within mature evergreen forest (Photo by Francisco J. Vilella).
If the objectives of the recovery plan approved for the nightjar (Diaz Reference Diaz1983) are to be met, several habitat management alternatives should be pursued. Habitat management options available for the nightjar will depend primarily on whether the birds are found in private or public lands. This is due to access limitations and control over land use practices that are encountered within privately owned lands. Presently, the most urgent conservation need for nightjars within privately owned land is habitat acquisition. At the moment, the amount of terrestrial protected areas in Puerto Rico represents about 7.6% of the total land area (Puerto Rico and satellite islands) which is below the regional average of 8.6% for the Caribbean and Central America (World Resources Institute 2003).
Nightjars occur at moderate to high densities on a number of privately owned areas of dry limestone and lower cordillera forests in southwestern Puerto Rico (Vilella and Zwank Reference Vilella and Zwank1993). Therefore, measures should be taken to mitigate the impacts of changes on areas that presently possess nightjars and will likely remain under private ownership. Information on the status of the nightjar and the desirability of conserving the region's coastal dry forests should be made available to the relevant landowners. Additionally, reforestation using a mixture of mahogany and native tree species (e.g., Bucida buceras, Bursera simaruba, Pisonia albida, Exostema caribaeum) should be strongly encouraged. Landowners surrounding Guánica Forest should be encouraged to pursue silvicultural practices that promote nightjar nesting habitat and informed of conservation programs available for private lands (CEDES 2007). Land uses such as industrial and residential development that promote forest clearing should not be allowed in the periphery of Guánica Forest as Biosphere Reserves are mandated to strongly protect their peripheral buffer zones and promote the establishment of corridors along these buffers (Dyer and Holland Reference Dyer and Holland1991).
Nevertheless, the most immediate and effective measures to ensure the continuing existence of the nightjar can be applied on presently occupied habitat under public ownership. Studies of the European Nightjar (Caprimulgus europaeus) on heathlands of the United Kingdom provide evidence of disturbance from walkers and pets moving in the vicinity of nests (Langston et al. Reference Langston, Liley, Murison, Woodfield and Clarke2007). In Guánica, the unlimited access by birders and amateur nature photographers to forest stands during the peak months of nightjar nesting activity (April–June) should be evaluated. Moreover, my results indicate the main limiting factors associated with use of habitat for nesting by nightjars were the presence of dense, tangled vegetation near the ground. These results illustrate nightjar-habitat ecological relationships that should serve as the basis for future experimental work. Further research should concentrate on experimentally testing the hypotheses raised by my results. For example, the effect on nightjar habitat use of experimentally manipulating the vegetation in the understorey and midstorey layers of the forest should be investigated and habitat management alternatives explored. Management of nest habitat in protected areas and private lands will ensure the long-term persistence of nightjar populations across southwestern Puerto Rico.
Acknowledgements
This paper is dedicated to the memory of Phillip J. Zwank, mentor and friend; and Baudilio Hernández, former manager of Guánica Forest who guided biologists to the rediscovery of the Puerto Rican Nightjar in 1961. Funding was provided by the U. S. Fish and Wildlife Service, Caribbean Field Office. I am grateful to the following individuals for assistance with nest searches and collection of vegetation information during the course of this study; M. M. Canals, G. Padilla, S. Silander, E. Silva, and A. Román.
