Page 3 - Eagles_Circaetus_gallicus_2016

Migrating eagles and water barrier 545

2011). Wind data at 1500 m. above sea level were Table II. Reverse migration (flock and solitary bird sampling
obtained from the National Centers of units) reported in the study area.
Environmental Prediction/National Center for Species Inland zone Coastal zone
Atmospheric Research (NCEP/NCAR) reanalysis
project (NOAA/OAR/ESRL PSD, Boulder, CO, Accipiter nisus 2 1
Buteo buteo 1 0
USA, http://www.esrl.noaa.gov/psd/; National
Circus aeruginosus 15 13
Oceanic and Atmospheric Administration/Oceanic Circus pygargus 1 0
and Atmospheric Research/Earth System Research Circaetus gallicus 4 15
Laboratory, Physical Sciences Division) and down- Falco tinnunculus 1 0
loaded using RNCEP package in R software (Kemp Falco tinnunculus/naumanni 3 0
Hieraaetus pennatus 2 2
et al. 2012). Variables in the model were chosen
Milvus migrans 11 0
following a backward stepwise approach using an
Pandion haliaetus 1 0
information theoretic approach (Burnham & Pernis apivorus 85 1
Anderson 2002), selecting the variables by the AIC
(Akaike 1973).
Table III. Results of binary logistic regression analysis; asterisk
flags significance.
Results
Explanatory terms Wald df P
A total of 12,147 directions of migration were
recorded, involving 67,235 raptors belonging to 23 Watch site 10.5 1 0.001*
species; of these only 158 (1.3%) resulted in reverse Wind speed 0.03 2 0.9
migration (Tables I, II). The short-toed snake eagle Wind direction 1.9 3 0.8
Wind direction*Wind speed 0.01 3 1
Circaetus gallicus (Gmelin, 1788) was the only species
showing a significant proportion of reverse migration
(19 vs. 64, 23%; Table II). As a result, we focused
observations involved migrating birds. The median
the analysis on this species. Since this eagle does not
date of the passage of this species was on 26
breed in the study area (Panuccio et al. 2015), our
September, and among aged eagles (N = 40) 33
(82.5%) were first calendar year (cy) birds. Finally,
Table I. Numbers of observed directions of migration (flock and we reported 69 solitary individuals and 14 flocks
solitary bird sampling units) reported during autumn migration containing on average 2.7 ± 0.7 (standard error,
2011, 2012 and 2013 in southern continental Italy. SE) birds. Among birds and flocks showing reverse
migration (N = 83), we directly observed the change
Species Inland zone Coastal zone
in the direction (from expected to reversed) in six
Accipiter nisus 42 26 cases (two flocks and four solitary birds), and always
Aquila clanga 1 0 in the coastal zone; all birds of the two flocks
Aquila pomarina 1 0 behaved in the same way, showing a strong social
Buteo buteo 139 47
Buteo rufinus 2 0 attraction. As regards statistical analysis, variables
Circus pygargus/macrourus 39 9 were not auto correlated (P > 0.05). The
Circus aeruginosus 3329 1085 ANCOVA run on the most parsimonious BLRA
Circus cyaneus 2 1 model (Table S1) shows that only the observation
Circaetus gallicus 47 36 zone significantly affected the behavior of short-toed
Circus macrourus 19 2
Circus pygargus 62 17 snake eagles (Table III). In particular along the
Falco subbuteo/eleonorae 24 5 mountain chain eagles were more likely to migrate
Falco tinnunculus/naumanni 1075 138 in the expected direction of migration (heading
Falco eleonorae 13 5
south) than along the coastal zone. The AUC of
Falco naumanni 106 4
this model was 0.96 (P < 0.001), so the accuracy of
Falco peregrinus 14 9
the model was good.
Falco subbuteo 116 31
Falco tinnunculus 25 5
Falco vespertinus 14 2
Discussion
Hieraaetus pennatus 138 73
Milvus migrans 427 137
The low proportion of raptors showing reverse
Milvus milvus 10 1
migration was probably caused by the location of
Neophron percnopterus 4 0
Pandion haliaetus 99 29 watch points, far from the shoreline, but also because
Pernis apivorus 4081 656 the Strait of Messina is not an insurmountable water