Conserving Chile ’ s most critically endangered bird species : First data on foraging , feeding , and food items of the Másafuera Rayadito ( Aves : Furnariidae )

1 Institute of Landscape Ecology, University of Münster, Robert-Koch-Str. 28, 48149 Münster, Germany. 2 Departamento de Ecología, Pontificia Universidad Cat. de Chile, Av. Lib. B. O’Higgins 340, P.C.: 114-D, Santiago, Chile. ingo.hahn(at)uni-muenster.de 3 Institute of Biogeography, University of Trier, Am Wissenschaftspark 25-27, 54296 Trier, Germany. 4 Institut de Recherche pour le Développement (IRD), LMI-EDIA, 357, calle Teruel, Miraflores, Lima 18, Peru. 5 Department of Engineer Geography, Universidad de Santiago de Chile, Av. Lib. B. O’Higgins 3363, P.C.: 7254758, Santiago, Chile. 6 CEUS Llanquihue, Universidad de Santiago de Chile.


Introduction
The Másafuera Rayadito Aphrastura masafuerae (Aves: Furnariidae) is endemic to Isla Alejandro Sel-kirk, where it is restricted to the high altitudes dominated by fern stands and tree fern forest.Aphrastura masafuerae, described by Philippi & Landbeck in 1866, presently is Chile's most threatened bird species.It is listed as a "critically endangered species" in the Red List (BirdLife International, 2010).Having a total population of only about 140 individuals, it is the most prone to extinction amongst all bird species of Chile (Hahn et al., 2009).
Few fi eld observations of the Másafuera Rayadito have been made since its scientifi c description.The bird was not seen for fi fty years (comp.Millie according to Johnson & Goodall, 1967;Torres & Aguayo, 1971) and already categorised as probably extinct (Vaurie, 1980), until Bourne (1983) rediscovered it by seeing four individuals.Then Brooke (1988) and Hahn (1998) informed of a live population.On February 2 nd 1986Brooke (1988) ) saw an adult bird carrying unidentifi ed food, but it disappeared and was not followed.No other information of food sources and feeding behaviour existed.
Main reasons for the small population seem to be man-made habitat destruction and introduction of alien species.For a broader discussion of reasons for decrease refer to Bourne et al. (1992), Cuevas & Leersum (2001), and Hahn & Römer (2002).However, predation by native Másafuera Hawks Buteo polyosoma exsul was not seriously considered before our studies (Hahn et al., 2004a, b).Another not investigated and possibly limiting factor may be competition with a second insectivorous ovenbird in the same island, the Masafuera Cinclodes Cinclodes oustaleti baeckstroemii.
1) The reproductive period is of high critical importance to a species, in particular for island birds (e.g.Moors, 1985).As A. masafuerae is in danger of extinction, the fi rst purpose of this work is to make ecological information available on home range structure, foraging behaviour and food sources during the breeding season, as these will become important for future conservation activities.Therefore we investigate the orientation and distance of foraging fl ights from and to the nest by adult rayaditos.We additionally want to analyse the faeces transports away from the nest and their importance in keeping the nest locality secret from hunting hawks.
2) Islands are generally species poor ecosystems.This is especially obvious for the land birds of Alejandro Selkirk.We suggest a similar scarcity in the number of insect (prey) species, and, thus a strong reliance on certain types of prey.In this study we aim to present observations on the feeding behaviour and food items conducted at the only nest sites ever been found of the Masafuera Rayadito.Furthermore, we aim to discuss the potentially important effect of competition risk as a factor limiting its breeding performance by present-ing comparative data on the food items of Cinclodes oustaleti baeckstroemii.

Study area
The study area is Isla Alejandro Selkirk (formerly Másafuera), the westernmost island of the Juan Fernández Archipelago.The island is positioned in the southeast Pacifi c Ocean (33° 45' south and 80° 45' west of Greenwich) off the coast of Chile.It is 167 km west of Isla Robinson Crusoe, the other major island, and is about 769 km distant from the South American continent.The area of Alejandro Selkirk is about 44.6 km², some 10 km from north to south and six from east to west.Reaching 1320 m at the Pico del Inocentes, it is the highest island of the archipelago.Alejandro Selkirk is part of the Chilean national park "Islas Juan Fernández" that was founded in 1935.Because of its international value the archipelago was declared an UNESCO Biosphere Reserve in 1977.More detailed geographical descriptions may be taken from Castilla (1987), Skottsberg (1956), andStuessy &Ono (1998).

Field work and data collection
Field work on Juan Fernandez Archipelago was carried out during the breeding seasons of the years 1992/93, 1994/95, 2001/2002, and 2009. However, up . However, up to presence only four nest sites of this species have ever been found (Hahn et al., 2004a;in press).The data presented here comes from these four monitored nest sites (comp.Hahn et al., 2004a;in press).Optic identifi cation of birds based on the original species description (Philippi & Landbeck, 1866), which was supplemented later on (Hahn & Römer, 1996).As Másafuera Rayaditos generally rarely leave the dense vegetation, they are diffi cult to fi nd.Mostly they are only recognised by the slight moving of a fern frond.If seen directly (which is rare at all), they attend to rapidly consume their food, making it diffi cult to identify the specifi c items additionally.Thus the fi rst necessity was to detect a home range of a pair.This was possible after learning the vocalisations in the fi eld, especially common call and morning song (comp.Hahn & Mattes, 2000).Then it was necessary to fi nd a nest site where food items could be identifi ed in the bill of adult rayaditos.Finding a nest site and a suitable observation point was still among the most diffi cult requirements for the planned observations, as the nests are well hidden and the terrain is badly accessible.the geomorphology of the surrounding terrain (or factors resulting from it).The fi rst station after fl ying off the nest site was the basis of starting to forage.Seven distance categories were distinguished (Table 2, col.1).
Home ranges were foraged up to a maximum of 100 meters by adult rayaditos.However, already fl ights of more than 75 meters were rare compared to the other categories, especially at nest site 4. The low number of fl ights off nest site 2 into the 0 -10 m zone probably resulted from the nest site's downhill exposition (NE), by which the height above ground was greater than at nest site 4. Except for this, the distance categories of the home ranges 2 and 4 were frequented similarly and generally show the same tendency for both breeding pairs: the area up to 30 meters distance around the nest site was frequented most and the number of fl ights was decreasing along with larger distances.This pattern becomes even more obvious by calculating fl ight numbers per surface area.The area per distance category was calculated (Table 2, col.8), and based on this the probability for a visit in any 100 m category given (Table 2, col.9).The calculation results show a decrease in successive stages of fl ight numbers along with larger distance to the nest site.

Faeces transports
Principally the taking away of egg-shells and faeces over considerable distances is done by numerous passerine birds to not been discovered by predators.Hole-breeding birds additionally have to keep the nest cavity clean and accessible.Faeces were already registered in order to transports per time unit (Hahn et al., 2004b).Here spatial observations of faeces transports were done according to distance and drop-down point to obtain the information, if Másafuera Rayaditos try to keep their nest site location secret from predators by taking faecal matter far away and/or inside the fern stands.Five distance categories were distinguished at nest site 4, to which the nestling excrements were taken by adult Másafuera Rayaditos.Farther than 20 meters away from the nest site faeces transports could not always be followed.If the rayadito kept the faeces in the bill throughout the fl ight, the distance to its disappearance in the vegetation cover was noted.
The quantitative registrations (Table 3) show that most faeces (n = 16) were already dropped inside the 20 meter radius, close to the nest site.Considering the distances in the Tables 2 and 3, less than 30 % of the faeces were carried more than 20 meters from the nest site, although about 80 % of the fl ights led further anyway.Most faeces were dropped down during the fl ight off the nest hole.

Results
As Másafuera Rayaditos generally rarely leave the dense vegetation cover, they are seen seldom.Mostly they are recognised by the slight moving of a fern frond or a call only (Fig. 1).Thus, without telemetry it is diffi cult to study home range structure and home range size.However, during the nestling time the adults regularly leave the vegetation cover (Hahn et al., 2004b) (Figs. 2 & 7).Firstly, to enter the nest hole located in a rock upwards from the basing fern stand.Secondly, after feeding the nestlings and taking off, they fl y above the fern cover to their next site of food search.Time is probably scarce because of high food demands for own nutrition and juvenile feedings (ca.500 per pair and day).However, they fl y tight above the ferns (0.5-2 m) and by that quickly reach the point of fi rst food search.This point generally represents the most nest distant one during food search.From the point of landing they move inside the fern stand more or less directly towards the nest site, still searching for food.This behaviour allows studying the structure of the feeding range during the nestling time.

Orientation and distance of foraging flights
In regard to the topographical structure of the steep terrain, all detected home ranges were placed on slopes.200 fl ights of adult Másafuera Rayaditos leaving the hole of nest site 2 were optically followed, compass point and geomorphologic orientation registered.The home range, surrounding the central nest site, was subdivided into eight compass point classes, each covering an angle of 45°.Three of these compass point classes were facing downhill, three uphill, and two hill-parallel.
137 fl ights (68.5 %) followed hill-parallel directions (NW, SE), although these directions mathematically took a proportion of 90 degrees (25 %) only.Thus, hill-parallel fl ights were over 170 % more often than expected by assuming statistically equal distribution.Contrary, fl ights directed downhill (6.5 %) or uphill (25 %) were underrepresented: although these areas each covered a proportion of 135 degrees or 37.5 % of the studied home range.
An observation series on the fl ight distance during foraging was aimed to give further information of home range extension and size.The distance of 312 fl ights off the nest sites 2 and 4 to the destination location was registered.The two observed nest sites (in two different years; Table 2) were located at the same rock, just a few meters apart from each other.Thus, differences of home range size were not caused by  Five different types of prey were recorded (Table 4).Only arthropods were observed as food for nestlings, no vegetable one.More than two thirds of all recorded feedings contained butterfl y larvae.Catching larvae from the fern fronds with an insect net showed that all "small, green" specimens belonged to one single species (Fig. 3).The small whitish silver moth is very likely to be the imago of this larva, and to belong to the same species.Thus, about three quarters of the prey seems to come from only one species.The remaining quarter consisted of spiders (Fig. 4).Pill bugs and big brown lepidopteron larvae were fed by exception only, and were never present in samples of insect nets.As the small green lepidopteron larva was the most important food source, its numbers and aggregation with other prey items per feeding event were registered.
About half of the feeding events contained two small green lepidopteron larvae alone (Table 5).Mása fuera Rayaditos are also able to carry three of these larvae or one larva together with a different arthropod prey.The proportion of feeding events of one single larva is relatively high (over one third).Thus, this was obviously not caused by their transport ability, as clearly shown by the other feeding events and the small size of larvae (comp.Fig. 3).More probable is the low density of suitable arthropod prey throughout

Feeding behaviour and food items
Only at one of the four known nest sites of the Másafuera Rayaditos Aphrastura masafuerae (comp.Hahn et al., 2004a), a suitable and closely observation point was present.At nest site 4 prey items of the two adults were identifi ed in the bill, before they entered the nest hole for feeding the nestlings.It was qualitatively distinguished in prey types per feeding event.Mostly only one type of prey was carried during one feeding event.
However, sometimes two prey types were brought at once.ability and density of their prey seems small.Thus, it is probable that Másafuera Rayaditos depend on one specifi c prey species, which is small and often found single.
the home range.Adults were foraging in an area of up to 100 meters distance from the nest, feeding nestlings every ca. 3 ½ minutes on average (Hahn et al., 2004b).Although rayaditos are highly active (Vaurie (1980) describes them as "restless birds"), the vari-    Rayadito Aphrastura spinicauda, closely related and widely distributed on the mainland, is also known to mainly feed on small arthropods (Vuilleumier, 1967;Johnson & Goodall, 1967;own observations IH & PV 1993-2009).However, single observations re-

Discussion
The results of this study indicate that Aphrastura masafuerae is principally insectivorous.The Thorn-tailed     had to be done uphill, the most energy consuming way.Prey items are limited in number per area and may be found scattered in the dense vegetation, but not or only exceptionally clustered.Thus, it is of principal advantage for Másafuera Rayaditos to more or less regularly visit different areas around the nest site.However, energy and time costs signifi cantly increase with growing distances from the nest site (comp.Table 2), which might be critical during the nestling time.Adults have to balance between their own energy uptake and expenditures rates on the one hand, and the energetic investment into the needs of the brood on the other, combined with the necessity to control the nest site to shelter their nestlings.The staggered home range usage may represent an optimised economical strategy to invest energy and making sure that breeding efforts turn out to be successful.Predation on rayaditos by native Másafuera Hawks Buteo polyosoma exsul during the breeding season was shown to be a serious threat in previous studies (Hahn et al., 2004a, b).However, Másafuera Rayaditos do not take faeces far away for hiding their nest site, on vealed that A. spinicauda was pecking ripe berries from Berberis, Ribes (Humphrey et al., 1970), and Gunnera (R. Schlatter pers.comm.).Two Gunnera species occur as well on Alejandro Selkirk, but A. masafuerae yet has not been observed feeding on these.According to their reduced nutrition value these berries may not be used for raising nestlings.At present, resulting from the lack of more detailed data, we are not able to present a more reliable inter-specifi c comparison of feeding and food items of these (only) two Aphrastura species.
Our fi ndings on distribution of food sources, vegetation structure (Hahn et al., 2010), and geomorphologic features of the nest surrounding terrain did not give evidence for an unequal usage of the home range by Másafuera Rayaditos: the terrain structure and the vegetation cover were highly homogeneous.More likely the infl uence of energetic factors is reasonable for such a behaviour (comp.Table 1).This would also explain the fact that fl ights downhill were nearly four times less numerous than fl ights uphill: after fl ights directed downhill the search for food and the transport Tab. 4. Prey item types identifi ed in the bill of adults before entering the nest: Aphrastura masafuerae (nest site 4; January 9 th ) and Cinclodes oustaleti baeckstroemii (nest site 6; January 25 th ).Prey types were counted per nestling feeding event (qualitative), not the number of individual items of prey (quantitative).Finally, we conclude that more detailed studies on the breeding ecology and foraging of both co-existing ovenbirds on Alejandro Selkirk are required for setting up appropriate management plans and protection activities (including general evolutionary comparisons like in Remsen, 2003;Fjeldså et al., 2005).The Másafuera Rayaditos seem to have developed suitable adaptations for protecting their broods against Másafuera Hawks, but nothing is known about the interactions with introduced mammals like rats and cats.Maybe such investigations are the key for understanding the critical factors to save Chile's most fragile bird, and lead to successful species conservation.

Aphrastura masafuerae
average not even to the point of fi rst food search.Most faeces were even dropped down during the fl ight off the nest hole.Then they were clearly visible on top of the fern fronds, whitely refl ecting.These faeces must have been visible for Másafuera Hawks, too.Probably the locations of nest sites were all known to resident hawks, especially facing the high number of feeding visits per day and the open homogenous habitat.However, it seems that not the knowledge of the locations is the critical point but the narrowness of the nest entrance and the deepness of the nest itself (comp.Figs. 5 & 6).
Másafuera Rayaditos obviously prefer comparatively stationary food items like lepidopteron larvae (about two thirds of observed prey items; comp.Tables 4 & 5).They have been observed carrying relatively small insects or their larvae to the nest.These fi ndings may either be an artifi cial effect of the circumstance that the nestlings during our observations were still young, or of a simultaneous seasonal reproduction of the lepidopteron species on the island.The latter might explain the high amount of identical prey items as well as the comparably homogenous size of the larvae carried to the nest.Adult birds should optimise their feeding strategies concerning two major aspects: they fi rstly should forage as many biomass as fast as possible and secondly be able to calculate the optimal size of each portion to guarantee fastest growing of their offspring possible under the given circumstances.This is generally the case in situations in which the availability of sources is foreseeable depending from stable climate conditions.This should be the case here, as the climate on the Juan Fernandez Islands is fairly constant, and follows a regular seasonality.
Compared to the Másafuera Cinclodes, the difference in the food sources is striking.At present we do not have enough data for a signifi cant test of our hypotheses (see also Hahn et al., 2005), but we may be allowed to speculate, that two different reasons may be responsible for this effect (see .On the one hand the nestlings of the observed cinclodes may have already been older compared to that of the rayaditos.We know from many bird species, that adults feed mostly soft small larvae to young hatchlings.With growing offspring the hardness of food items also rises in most of these species and parents switch to feed more imagines of certain insect species during late nestling days.On the other hand the different choice of food may be the evolutionary result of inter-specifi c competition between the two endemic species (comp.also Blondel et al., 1999;Blondel, 2000).If this would be the case, the food ecology of these two island bird species would be a perfect example of nice building capacities of inter-specifi c competition preventing competitive interaction of both species in the most sensitive phase of their lifecycle, during reproduction.

Fig. 1 .
Fig. 1. Adult Másafuera Rayadito Aphrastura masafuerae forage in the alpine fern stands of Lophosoria quadripinnata for arthropod prey.They forage together in pairs, using vertical as well as horizontal structures, or stay in vocal contact in the shelter of a dense vegetation cover.Drawing by Erika Römer on base of photographs by IH.

Fig. 2 .
Fig. 2. Másafuera Rayadito Aphrastura masafuerae leave the fern stand cover only by exception, for example when disturbed during foraging or in the presence of a potential predator.The illustration shows the wren-like warning behaviour because of a nearby hawk close to the nest site.Drawing by Erika Römer on base of photographs by IH.

Fig. 4 .
Fig. 4. Brood-caring Másafuera Rayadito Aphrastura masafuerae carries a spider to the nest hole for feeding the nestlings.It landed just left of the nest entrance, controlling the surroundings before entering.

Fig. 5 .
Fig. 5. Small entry to the nest hole of the fi rst nest site discovered for the Másafuera Rayadito Aphrastura masafuerae.The entrance is just brought enough to allow the birds to enter, representing a fi rm natural protection from any larger predators.

Fig. 6 .
Fig. 6.A Másafuera HawkButeo polyosoma exsul has discovered the nest hole, sitting right in front of it, later trying to snap with its talons inside.An adult Másafuera Rayadito Aphrastura masafuerae is warning just two meters apart on a projecting rock.The hawk attack remained unsuccessful, as the nest entrance was small and the nest itself too deep.Photos: I. Hahn.

Fig. 7 .
Fig. 7. Foraging adult Másafuera Rayadito Aphrastura masafuerae stops shortly on top of a fern frond to search for arthropod prey items between the small leaves of a Lophosoria quadripinnata.

Fig. 8 .
Fig. 8. Másafuera Cinclodes Cinclodes oustaleti baeckstroemii mostly hunt on fl ying insects.They are acrobatic and fast fl yers, and do not depend on the shelter of a vegetation cover to escape hawks.Mostly frequenting open and rocky habitats, they are rarely seen on ferns like here.

Fig. 9 .
Fig. 9. Másafuera Cinclodes Cinclodes oustaleti baeckstroemii also nest in natural hole of rock walls.As their nest entrances are wider than those of Másafuera Rayaditos, the nest itself is located even deeper inside the rock for inhibit hawk predation.

Fig. 10 .
Fig. 10.A Másafuera CinclodesCinclodes oustaleti baeckstroemii carries several small whitish-silvery lepidopteron imagines inside the nest hole for feeding the nestlings near the top of Los Innocents Mountain.Photos: I. Hahn.
Tab. 1. Flight direction and geomorphologic orientation of brood caring Aphrastura masafuerae fl ying off nest hole 2 (January 6 th ).Foraging fl ights of brood caring Aphrastura masafuerae and their distance to the nest sites 2 (January 9 th ) and 4 (January 9 th ).