Study Area.

The island of Madeira (32°44′N, 17°00′W; 737 km²) is approximately 630 km off the western coast of Morocco. This mountainous island is dominated by a volcanic landscape, with deep valleys, steep slopes, and scarps. The watercourses are enclosed within cliffs which run roughly perpendicular to the coastline. The maximum altitude is achieved at the top of the eastern mountain range (1861 m at Pico Ruivo). Climatic conditions vary across the island, with the northern part being colder and wetter, and the southern part warmer and drier. The highest annual precipitation (2966.5 mm) occurs at the highest altitudes (>1000 masl) and lower rainfall amounts (about 550 mm) in lowland areas (<700 masl). The average annual temperature also depends on altitude, with values of 17°C at the southern coast and 19°C at the northern coast, falling to 9°C in the highlands.

The Laurel forest occupies 20% of the total area of Madeira, mainly on the northern slopes between 300 and 1300 masl in altitude. Several invasive plant species currently threaten the Madeiran Laurel forest, primarily striated broom (Cytisus striatus) and dwarf gorse (Ulex minor). Species that are harvested (several pine species [mainly P. pinaster], eucalyptus, acacia [Acacia spp.], and chestnut [Castanea sativa]) prevail mainly in the south of the island between 300 and 800 masl. There are no indigenous pine species in Madeira (Vieira 2002). The reforestation of some areas between 700 and 1200 masl also included species such as Douglas-fir (Pseudotsuga menziesii), oak (Quercus spp.), and Japanese cedar (Cryptomeria japonica). Above 1400 masl, the forest was severely damaged by overgrazing and fire and only a few small woodlots still stand (Capelo et al. 2004).

Monitoring Nests of Macaronesian Sparrowhawk.

This subspecies begins mating in February, with egg-laying occurring in April. Young hatch in May, and juveniles eventually leave the nest in July (Delgado et al. 1987). We carried out fieldwork during two successive breeding seasons in 2014 and 2015. In 2014, we surveyed preliminary transects around the island between 24 May and 28 August, using accessible paths and “levadas” (walking trails that follow the network of water channels) to search for sparrowhawk nests. However, we detected all nests when the chicks were near fledging age (July and August). During transect surveys, we recorded all Macaronesian sparrowhawks observed and their behavior; we also searched for droppings, plucked prey, and molted feathers as signs of sparrowhawk presence. Breeding females start molting while incubating, so molted primary and secondary feathers can be found near the nest tree (Newton and Marquiss 1982).

In 2015, we conducted 84 1-hr observations with scope and binoculars at high visibility points near forest patches, from 15 February to 31 March (mating and nest-building phases). For some patches, we observed at more than one point in order to cover the whole forest patch area. From the point, we observed courtship flights and then recorded the direction where the pair entered the forest, so that we could later search for their nest. In addition, we performed transect walks to search for nests and other signs of sparrowhawk presence from 1 April to 31 August (incubation, nestling, and fledgling phases). Specifically, we first visited forest patches where courtship flight had been observed in the mating phase; second, we visited forest patches with signs of presence from the 2014 breeding season, and finally, we visited the remaining forest patches around the island. We searched each forest patch from 1–6 d (8hr/d) until the nest was found. We classified nests as either “in use” or old. From the ground, we classified nests as in use if they had freshly broken twigs or flecks of white excreta (during incubation). We classified nests as old if they contained fallen leaves and spider webs (Newton 1976).

During the 2015 breeding season, all nests in use found prior to egg-laying were visited every 4 d in the incubation, hatching, and fledging periods, in order to estimate the incubation, hatching, and fledging dates of each nest, and number of fledglings. We observed nests from the ground using binoculars and a spotting scope, but did not climb to them.

Variables Measured at Nests, Nest Trees, Nest Sites, and Nesting Territories.

For each nest, we measured height aboveground using scaled photographs, and determined nest orientation with respect to the trunk and the orientation of access to the nest used by the birds. We classified the position of the nest in the tree as either near the trunk or on a side branch.

We defined nest tree as the tree in which the nest was placed. We classified the altitude of each nest tree as one of three categories: <700 masl, 700–1000 masl and >1000 masl and measured the distance to the nearest accessible trail or path using a GPS. We also recorded tree species, tree height, and diameter at breast height (dbh); the radius of the top of the tree was measured using scaled photographs.

We defined a nest site as the area within a 10-m radius around each nest tree. Due to the complex and steep topography of Madeira Island, this plot size was the largest accessible area for which nest-site variables could be accurately measured. Within each nest site, we classified the interior watershed (northern or southern). Moreover, we measured slope and assessed the proportion of ground vegetation cover and canopy closure.

We grouped all nests (in use and old) within the same forest patch into the same nesting territory. We identified all dominant tree species, and classified each nesting territory as one of three types of habitat: Laurel, mixed (indigenous and introduced tree species), or exotic forest. We recorded the presence or absence of valleys with watercourses, as well as the presence or absence of open areas (Patrimonio 1984). We also recorded whether the forest patch was in a regeneration process or not, and the causes of this (fire, lumber harvesting, etc.).

Statistical Analyses.

We examined the relationship between breeding phenology and altitude of the nest sites using a Pearson's correlation. For this analysis, we used hatching dates because during the visits we could visually confirm the age of the young, whereas the egg-laying dates had to be estimated for some nests. We tested for normality of data using a Shapiro-Wilk test.

We defined breeding success as the percentage of young fledged per egg-laying pair, and reproductive rate as the number of young fledged per egg-laying pair. To evaluate whether pairs breeding earlier had greater reproductive rate than those breeding later, we compared the number of fledglings per nest with eggs with hatching dates using one-way ANOVA. We used the difference (in d) between the mean hatching date and the hatching date of each pair as a categorical variable and transformed the values using natural log. We used chi-square tests to analyze whether the number of fledglings per nest with eggs varied with nest orientation and position, watershed of the nest site, dominant habitat of the nesting territories, and between territories in regeneration or not. We used one-way ANOVAs to test whether the number of fledglings per nest with eggs was influenced by nest height, distance to path, height of nest tree, tree diameter, tree top radius, the slope, vegetation cover, canopy closure of the nest sites, and the number of dominant tree species of nesting territories and the number of territories in regeneration.

Nests, Phenology, and Reproductive Rates of Macaronesian Sparrowhawks.

We found 54 different nests on 35 nesting territories between 4–22 August 2014 and from 31 March to 29 August 2015 (Fig. 1; Table 1); 25 of those nests were in use and 29 were classified as old nests. Within the same territory, the average distance between nests (in use and old) was 129.84 m (±26.8 SE, range = 12–230 m) and the mean number of nests was 1.74 (±0.2, 1–7). One nest tree contained two old nests; all others had only one nest. Considering the nest in use as the center of the nesting territory and measuring the distance between nests, we found that the average distance between nests/nesting territories was 4764 m (±238 m, 782–7534 m). In six of the nine nesting territories found in 2014, we confirmed sparrowhawk presence in 2015. In five of these territories, the birds built a new nest near the previous one; in one territory, the breeding pair reused the same nest.

Figure 1. 

Spatial distribution of the Macaronesian sparrowhawk nests found on Madeira Island from 2014 to 2015.

Table 1. 

Number of nesting territories, nests in use, and old nests found on Madeira Island from 2014 to 2015.

The earliest dates of incubation (7 May; mean = 19 May ±1.5 SE; range = 7 May–13 June), hatching (14 June; 27 June ±1.4; 14 June–22 July), and fledging (16 July; 29 July ±1.6; 16 July–25 August) occurred at a lowland nest (234 masl). The latest dates were observed at an upland nest (1538 masl) more than 1 mo later.

Mean breeding success of the Macaronesian sparrowhawk in 2015 was 73.2% ±0.1 SE (n = 18). Of five recorded nest failures, four (22.2%) failed due to forest exploitation, by direct cutting or subsequent collapse of the nest due to direct exposure to the wind. At the remaining nest, the cause of failure was unknown. The mean number of fledglings/nest with eggs was 2.27 (± 0.04, 0–3). In 2014, a dead nestling found headless under its nest had an incisor-mark on its right wing. There are two potential predators of Macaronesian sparrowhawk on Madeira, cats (Felis silvestris catus) and rats (Rattus rattus), and the incisor-mark found on this particular nestling matched the incisor size of cats. Moreover, in three different nests, we recorded one dead nestling each, but the causes of death were unknown.

Variables Measured at Nests, Nest Trees, Nest Sites and Territories.

Nest heights ranged from 4.5–17 m (Table 2). The most frequent position of the nest was on a side branch of the tree. Macaronesian sparrowhawks nested in 11 tree species (Table 3). Although all nests were located <165 m from the nearest accessible path, the number of fledglings/nest with eggs was not influenced by the distance from nest tree to path (F = 0.250, df = 14, P = 0.935). The number of fledglings/nest with eggs was higher for nest sites on the northern side than on the southern side of the island (χ² = 5.000, df = 1, P = 0.025; n = five nests on the northern side and n = 15 on the southern). Most territories (88.6%) were located in forest patches where valleys with watercourses were present, and all of them were in close proximity (average 25 m) to open areas. At eleven of the 35 nesting territories (31.4%), the forest was in regeneration, mainly due to fires.

Table 2. 

Mean values (SE and ranges) of environmental characteristics of Macaronesian sparrowhawk nesting territories at three different altitudes monitored on Madeira Island from 2014 to 2015. Nest orientation and access orientation are the most common cardinal directions.

Table 3. 

The number and percentage of tree species used for nesting by the Macaronesian sparrowhawk, and the territories where these tree species were found, on Madeira Island from 2014 to 2015.

Altitude of the nest tree was positively associated with hatching dates (Fig. 2): Macaronesian sparrowhawk pairs breeding in lowland nest trees initiated breeding earlier than those in upland nest trees (r = 0.977, P < 0.0001). However, pairs breeding earlier did not have higher reproductive rates than those breeding later (F = 0.186, df = 2, P = 0.832). Mixed forests were the dominant habitat of nesting territories, except in upland territories where exotic forest was prevalent. The number of fledglings/nest with eggs was higher in mixed habitats compared to exotic or Laurel forest (χ² = 7.929, df = 4, P = 0.040). None of the remaining variables influenced the number of fledglings/nest with eggs: nest-level variables (nest height F = 0.511, df = 9, P = 0.832; nest orientation χ² = 3.333, df = 7, P = 0.853; access orientation χ² = 4.222, df = 7, P = 0.754; position χ² = 3.556, df = 1, P = 0.059); nest tree-level variables (height of nest tree F = 0.634, df = 11, P = 0.755; tree diameter F = 0.390, df = 9, P = 0.905; tree-top radius F = 2.996, df = 7, P = 0.064); nest-site-level variables (slope F = 0.560, df = 7, P = 0.774; vegetation cover F = 0.876, df = 10, P = 0.586; canopy closure F = 1.582, df = 6, P = 0.235); nesting territory-level variables (number of dominant tree species F = 0.620, df = 4, P = 0.656; territories in regeneration χ² = 0.059, df = 1, P = 0.808).

Figure 2. 

Correlation between altitude (m) of the nest trees and the hatching dates of Macaronesian sparrowhawks for 2015 on Madeira Island (r = 0.977, P <0.000).

Reproductive Ecology of the Macaronesian Sparrowhawk.

Macaronesian sparrowhawk nests were distributed around the entire island of Madeira (Fig. 1). In this study we identified two nesting territory-level variables that seem to be important in the breeding habitat of this subspecies, and these variables were also identified for A. nisus wolterstorffi on Corsica Island (Patrimonio 1984). All nesting territories were near open areas; these were mainly agricultural lands (abandoned or in use), and probably used as hunting areas as sparrowhawks hunt over a wider area than the forest patch containing the nest (Newton et al. 1977). Moreover, most of the territories were in deep valleys where watercourses are numerous. A few territories were found in tiny forest patches (less than 40 m²), although canopy cover inside these patches was dense; this finding coincided with that observed for A. nisus wolterstorffi on Corsica (Patrimonio 1984).

Distances between old nests within a nesting territory in our study area (130 ±27 SE, range = 12–230 m) were shorter than those observed in Britain and Corsica (50–400 m, Newton et al. 1977; >400 m, Patrimonio 1984). This was probably because forest patches on Madeira are frequently interrupted by open meadows and rock ledges. We found fewer old nests per nesting territory (1.74 ±0.2, range = 1–7), compared to those in Britain (15 nests, Newton et al. 1977) and Tenerife (52 nests in 17 nesting territories; Delgado et al. 1987). In the Canaries, the Macaronesian sparrowhawk also shows high fidelity to its nesting territory, and usually builds a new nest each year close to the previous one (Barone and Atienza 2004). The small number of old nests found in the present study on Madeira may indicate that some nesting territories were not consistently occupied for many years, although it is also possible we missed some old nests located farther from the nest in use, due to the steep terrain. On Madeira, the destruction of forest patches by fires and the timber industry may be an important factor in influencing changes in nest site, but other factors such as adverse climatic conditions could also be significant (Martínez et al. 2013). Accurate indices of fidelity to nesting territories would help us understand temporal variation in occupancy.

The two closest nesting territories were 782 m apart; these were located in the same valley, but in different forest patches in the lowlands (<700 masl). However, we did not find a consistent pattern of increasing distance between nesting territories with higher altitude, as reported by Newton et al. (1986).

We found the same number of nesting territories in exotic and Laurel forest (25.7%), and nearly twice as many in mixed habitat (48.6%), composed of indigenous species of the Laurel forest and introduced species (mainly acacia and eucalyptus). In the Canaries, Macaronesian sparrowhawk nests are abundant in Laurel forest, and we believe that more nesting territories than we found in this study should exist in this habitat. However, due to the complex topography of Madeira and the inaccessibility of the Laurel forest, additional research effort is needed.

Nest heights of the Macaronesian sparrowhawk on Madeira (4.5–17 m) were similar to those on Tenerife (5–16 m, n = 13 nests; Delgado et al. 1987). Seven of the eleven tree species used for nesting on Madeira (Table 3) are newly reported for this subspecies, whereas the remaining species were reported previously in the Canaries (Delgado 1985, Delgado et al. 1987, Martín and Lorenzo 2001). The three most frequently used tree species were Douglas-fir (very common in exotic habitats), followed by two indigenous species: firetree (Morella faya) and laurel (Laurus novocanariensis), which are found in the Laurel forest and also in mixed habitats. Although some nests were found in Pinus spp., pines do not seem as important in Madeira as in the Canaries (Martín and Lorenzo 2001). This finding may be explained by the decrease of pine populations due to the pinewood nematode (Bursaphelenchus xylophilus) introduced on Madeira (Fonseca et al. 2012). Macaronesian sparrowhawks probably adapted to using the large pine plantations for nesting, as they do in the Canaries. However, the lack of previous knowledge on the historical breeding habitat of the Macaronesian sparrowhawk does not allow evaluation of the ecological consequences of the decreasing pine forest on Madeira.

Which Factors Influence Breeding Success and Phenology?

The breeding success of the Macaronesian sparrowhawk in Madeira (73.2%) was lower than the value found in the Canaries (81.2%; Delgado et al. 1987). The mean number of fledglings per egg-laying pair (2.3) was also lower than that found in Tenerife (mean = three young, n = 15 nests; Delgado et al. 1987). For long-lived raptor species, such as the Macaronesian sparrowhawk, an increase in reproductive success could benefit its population. Because the population of Madeira Island is small (Romano et al. 2010), conservation efforts should be focused on protecting extant nesting territories and improving reproductive rate.

The influence of altitude on breeding phenology was first reported here for this subspecies, although this pattern has been observed in A. melanoleucus of Africa (Sebele 2012) and is not unexpected. Although birds aim to breed when environmental conditions favor breeding success (Newton 1979), ideal breeding conditions in Madeira seem to occur at different times according to altitude. The influence of this factor on breeding phenology may be related to an increase of rainfall and decrease in temperature with altitude. Moreover, the Macaronesian sparrowhawk, in order to successfully breed, may have to shift the timing of breeding to match reproduction with peak food availability, which likely also varies with altitude (Wright et al. 2009). Future studies on the variables influencing breeding phenology are needed to evaluate other potential factors, such as breeders' experience and mate choice (see Newton and Marquiss 1982), and to consider their overall effect.

Threats Potentially Limiting the Population of Macaronesian Sparrowhawk on Madeira.

Our study revealed that the principal cause of breeding failure of Macaronesian sparrowhawk was forest cutting. Currently, forest exploitation is one of the most important economic activities on the island; however, this activity seems to be destroying recent breeding areas of the Macaronesian sparrowhawk. We found evidence of the effect of the forestry industry on Macaronesian sparrowhawk breeding success. To avoid a population decline of this short-range subspecies, forest management practices should consider species' requirements by: (1) determining the location of nests before harvesting; (2) conserving a minimum number of suitable nesting and hunting areas in regions subject to major logging; (3) postponing logging during the breeding season; and (4) purchasing mature plots of suitable nesting habitat for the species.

We found 31.4% of nesting territories in forest patches that had been burned. In recent years (especially 2010 and 2012), fires had devastating effects on Madeira Island, and they have been responsible for the destruction of many hectares of forest (Fontinha et al. 2014). In some forest patches, fires have created a heterogeneous mosaic of areas with varying degrees of damage, and areas with unburnt trees that were occupied by the Macaronesian sparrowhawk as nest sites. Once an area has been destroyed by fire, it loses potential not only as a nesting site, but also as a hunting site, and it takes many years for the habitat to recover (Haslem et al. 2011). The restoration of burned areas on Madeira is urgently needed to prevent soil erosion, expansion of invasive species, and loss of indigenous species (Fontinha et al. 2014). Fires have also devastated large forest areas used by the Macaronesian sparrowhawk on some of the Canary Islands (Barone and Atienza 2004). Therefore, accurate knowledge about the Macaronesian sparrowhawk's response to fire is needed in order to meet the goals of the Birds Directive for this short-range taxa.