THE RING 40 (2018) 10.1515/ring-2018-0001
THE GENERAL PATTERN OF SEASONAL DYNAMICS
OF THE AUTUMN MIGRATION OF THE WOOD PIGEON
COLUMBA PALUMBUS IN ITALY
Enrico Cavina, Rinaldo Bucchi and Przemys³aw Busse
Cavina E., Bucchi R. and Busse P. 2018. The general pattern of seasonal dynamics of the
autumn migration of the Wood Pigeon Columba palumbus in Italy. Ring 40: 3-18.
Given the scarcity of studies on the migration of the Wood Pigeon through Italy, the first
systematic observations by a network of hunters, as citizen researchers, can be presented
as a starting point for more in-depth analyses. Observations from the years 1998-2006 are
analysed and presented in a generalized form. During this period more than 100 observation
sites, covering most of Italy, were active for about 40 days every autumn. Migration over
Italy was described in terms of the timing and intensity of migration. Special attention was
directed to the long-term number dynamics and seasonal dynamics of the passage. The
most intensive migration was observed within northern Italy, while lower intensity is
visible more to the south of the peninsula. Following tendencies in numbers of observed
migrants within the ten years of the study, we can find positive tendencies in most of the
northern provinces, while three negative trends are visible in central Italy. The study of the
seasonal pattern, in terms of the number dynamics of the passage and the frequency of
pronounced peak days, strongly suggests that there are five or six waves of pigeons passing
through Italy in different parts of the autumn that are quite stable between years. Every
year the time of the passage includes a few peak days of migration.
Keywords: seasonal migration pattern, autumn, migration peaks, migration waves, Wood
Italy is a peninsular bridge through the Mediterranean Sea, between the Euro-
pean continent and the North African and Spanish coasts (35-47°N, 6-18°E). It is di-
vided into Continental Italy (at 44°N), Peninsular Italy and Insular Italy. The Alps
(barrier at the Continental North) and Apennine Mountains and their foothills, rivers
E. Cavina, Club Italiano de Colombaccio, via Serraloggia 31, 60044 Fabiano (An), Italy;
R. Bucchi, Club Italiano de Colombaccio, Italy; P. Busse (corresponding author), Bird
Migration Research Foundation, Przebendowo 3, 84-210 Choczewo, Poland, e-mail:
and valleys are clear markers of the complex topography and hydrography of Italy
and determine its seasonal climatology and weather.
Autumn migration of wood pigeons takes place mainly in October, but also during
the first half of November. Wood pigeons arrive mostly from the eastern continental
parts of Europe and western parts of the Asiatic Palearctic (Hobson 2009). This
migratory route begins in Eastern Europe, Russia and Ukraine, but probably also in
Asiatic areas east of the Urals. The breeding origin of many populations of wood
pigeons arriving in Italy remains a problem to be better investigated.
The main route of migration from Eastern countries passes through Hungary,
Slovenia, Austria and Croatia, and enters Italy by what is known locally as ‘the East
Italian Gate’ in the region of Friuli Venezia Giulia, which includes the Carniche Alps
and the Tagliamento and Fella River valleys. This corridor gives flying migrants direct
access to continental Italy – the Po valley, south of the Alps up to the Maritime Alps,
and the coast on the French border in the west and the Ligurian Sea (Genoa) in the south.
It should be emphasized that several different populations migrate through Italy:
those that breed a great distance away, within eastern Russia, as well as others breed-
ing in the forests of the Carpathian Mountains and the Balkans. These can all have
different lengths of migration: they can be long-, medium-, or short-distance mi-
grants. Thus the seasonal pattern of migration could be varied in terms of timing and
The seasonal phenomenon of migration is an important part of the historical tra-
dition and culture of many thousands of Italian hunters. For this group of citizens,
knowledge of migration has been limited to practical, uncoordinated observations,
but for twenty years an idea has been developing among hunters to conduct system-
atic daily observations for potential scientific use. A better understanding of the prop-
erties of the migration may be useful in the practice of sustainable exploitation of
huntable bird resources.
The aim of this paper is to provide a general picture of the autumn migration of
this species in Italy and to suggest subsequent research for elucidating this phenome-
non. This could encourage the formation of a network of observation sites maintained
by birders and hunters.
MATERIAL AND METHODS
The source of the data is reports from Progretto Colombaccio (R. Bucchi) and
Progetto Colombaccio Italia (S. Giannerini). Examples of the data are presented in
Figures 1 and 2. The observation scheme was launched in 1997 and carried out until
2006, as a network of observation posts run by hunters during the hunting season. At
sites selected by active hunters, wood pigeons were hunted and birds passing the site
were counted simultaneously.
The counting routine consisted in recording flocks of wood pigeons using a few
standard parameters: the number of pigeons by flock (flock size), direction of flight
(according to the standard direction of the local migration stream and ‘reverse mi-
gration’, in the opposite direction), and the bag from the hunting. In this paper, flock
size and bag numbers are not taken into account, but they can be used in future work.
4 THE RING 40 (2018)
Table 1 presents the periods of work, numbers of sites, numbers of observation
days per season, and total numbers of pigeons observed. The observations were car-
ried out essentially day by day, but due to Italian hunting law there were breaks in ob-
servation activity two days a week (Tuesday and Friday), when hunting is prohibited
(Fig. 1). This meant that the ‘continuous day by day’ scheme was in fact a kind of sam-
pling method. This resulted in less accurate data for determining the peak days of mi-
gration in a season. However, as the total data used cover ten seasons 1998-2006 (the
1997 season is excluded due to incomplete data), at a general level the results should
be sufficiently accurate. Another parameter of the observation scheme that was insuf-
ficiently standardized was the number of observation sites operating each year. To
analyse compatible values, the numbers of birds observed were recalculated to
number/site/day and presented for the administrative regions (Tab. 2, Fig. 3). The re-
gions are grouped into three areas from the north to the south of Italy.
THE RING 40 (2018) 5
Fig. 1. Example of source data sheet for Italy in 1997 (total, limited data). Numbering of hunting
days is given in the first column, then dates (note dates of days without hunting/
observations), followed by numbers of pigeons flying in normal direction of the passage,
numbers for reverse movements and numbers of birds shot.
6 THE RING 40 (2018)
Fig. 2. Example of source data sheet for provinces and areas as given for the main data set (1998-2006). In these years
the basic data set is supplemented by data not used in the paper.
Data used in the analysis: number of observation sites, period of work, number of
observation days and number of pigeons observed following the migration direction
Year Nsites Period of work Ndays Nobserved
1997 – 1 Oct–31 Oct 22 27,661
1998 60 26 Sep–5 Nov 27 178,400
1999 100 25 Sep–18 Nov 40 349,587
2000 107 25 Sep–19 Nov 40 507,175
2001 113 25 Sep–18 Nov 40 588,385
2002 124 25 Sep–17 Nov 39 807,867
2003 138 24 Sep–16 Nov 39 703,758
2004 129 22 Sep–11 Nov 34 820,464
2005 143 21 Sep–17 Nov 41 489,878
2006 124 23 Sep–12 Nov 37 568,557
Total 359 5,041,732
THE RING 40 (2018) 7
Fig. 3. Average daily numbers of migrating pigeons in different provinces of
Basic data by provinces and areas. Yearly raw numbers of observed pigeons, numbers
of observation sites, numbers of observation days and standardized values: number
per site and number per site and day
8 THE RING 40 (2018)
The general presentation of the course of seasonal migration through Italy is
based on raw numbers of daily migrating birds, as well as an analysis of variation in
the course of the seasonal migration pattern. It is common knowledge among stu-
dents of the seasonal pattern of passage of birds, irrespective of the species, that day-
to-day numbers of migrants are typically extremely variable. Rushes of migration are
noted in which more than 25% of the yearly number of migrants of a given species
pass the observation site in one day, and fluctuations from one day to the next can
reach 1,000% or more. Various terms have been used for such rushes, usually peaks
or waves of migration, but frequently the precise meaning of these terms is not de-
fined. The terms are commonly explained as ‘a day or days with clearly higher fre-
quency in the course of the seasonal migration pattern’. This makes comparisons of
In this paper, we define the term peak day as a day within which the number of obser-
ved birds exceeds 5% of the individuals observed within the season (all birds observed
from the beginning to the end of the observation period = 100%). This means that if
during a period of two, three or more consecutive days the share of birds each day is
above 5% of the total number of observed birds, all these days will be called ‘peak
days’. For a more precise description, peak days with different values are designated
as ‘low peaks’ – 5.1-10.0% of the yearly total, ‘moderate peaks’ – 10.1-15.0% and
‘high peaks’ – >15%. Still the term ‘peak’ refers to one day. When we use the per-
centage value of the share of the day in the entire study, calculation of the Similarity
Index (SI – discussed below) is natural and easily understandable.
We use the term wave of migration to refer to a period of several days in sequence
in which the migration is more intensive than in periods with lower numbers
(shares). The wave can contain both peak days and days with very low numbers.
Within the entire period of seasonal migration, waves are usually smaller at the be-
ginning and at the end of migration period than in the middle period of migration,
and of course the probability that real peak days will occur then is lower.
To study whether two curves representing migration dynamics are similar or dis-
similar in terms of the course of migration, we can use statistical tools, such as chi-
square or similarity indices. The similarity index used in this paper is the Renkonen
SI =Sum min [n%
where: SI is the sum of minima in the daily pair of frequency values expressed as
a percentage of the total samples compared (n%d1,n%
d2). The Renkonen coefficient is
commonly used to compare species population structure in botanical and zoological
ecology. The logical and statistical structure of our problem exactly fits the assump-
tions of the Renkonen coefficient. Since in this paper the data for establishing the
dates of peaks of migration were recalculated to daily percentage values, the results
were ready to use for calculation of SI values. In this study, SI values were calculated
to define the similarity of the migration dynamics in each year to the average dynam-
ics for all ten years together. The level of statistical significance of the results was esti-
mated by comparing the SI values with those obtained in a detailed study of the prob-
lem by Nowakowski et al. (2005), where the same method was used. The same proce-
THE RING 40 (2018) 9
dure could be used to compare the synchrony of migration through different areas in
Italy, but available data are still too limited.
Intensity of migration and multi-year number trends
The basic results for intensity of migration at sites situated in different areas and
provinces of Italy are presented in Table 2 and Figure 3. The most intensive migration
is observed within Area I in the northern part of Italy, where pigeons move over fairly
open and flat landscapes, while lower intensity is visible more to the south of the Ital-
ian peninsula, where birds must travel along several valleys to cross the Apennine
Examination of the tendencies in numbers of observed migrants within the ten-
year study period reveals positive tendencies in most of the northern provinces, while
three negative trends are visible in the central part of Italy (Fig. 4). Thus trends for
10 THE RING 40 (2018)
Fig. 4. Trends in numbers of pigeons as observed in different provinces
Areas I and II are positive (Fig. 5). The regression coefficient for Area II is statistically
highly significant (p< 0.01, F
= 18.25), while for Area I it is not (p> 0.05, F
Note, however, that the slope is similar to that of Area II, and the lack of formal sig-
nificance is due to the high variation within the data for that area (especially in 2004,
when the number of birds observed far surpassed the numbers in other years). The
trend for Area III is slightly, insignificantly negative (p> 0.05, F
= 0.29). At this time
the number of birds observed in Area I is one order higher than in Areas II and III.
Traditionally, when the sampling method is used, records are grouped to obtain
a more general picture. In this way, Figure 6 shows the general pattern of the autumn
migration of wood pigeons as one maximum, but asymmetric curve of the number of
observed individuals. The shape of the distribution pattern of migration peaks is very
THE RING 40 (2018) 11
Fig. 5. Total number trends in 1998-2006 by areas. Dots – yearly values (number per day), lines –
regression equation lines
similar. Both show clearly that the distribution is not normal, but probably composed
of a few basic normal or quasi-normal distributions caused by internal group differ-
entiation (migration in several migratory waves).
A more detailed – day-by-day – analysis of the distribution of numbers and peaks
confirms this supposition (Fig. 7). Both distributions show a clear multi-wave pattern.
Comparison of these distributions using the SI (Similarity Index) gives an extremely
high SI-value of 85.6, which is significant at the level of at least p< 0.001 (Nowakow-
ski et al. 2005). Thus only day-by-day analysis is suitable for studying the problem of
peaks and waves of migration. Distributions of peak days in different years are pre-
sented in Figure 8, where yearly patterns are shown against the background of the to-
tal pattern of migration.
The next step of the analysis is to answer the question, ‘are the yearly patterns
based on a common, repeatable general migration pattern?’ The first verification that
the distribution of peaks corresponds to the basic structure of migration has already
been presented in Figure 7, where the curves for the number distribution as well as
the peak distribution clearly show at least five waves (a sixth wave is probably hidden
in an asymmetric first wave at the beginning of the distribution). The second argu-
ment is even stronger – the analysis of the similarity of the yearly patterns with a com-
mon template of migration shows that all yearly courses of migration dynamics are in
highly significant agreement (pat least at the 0.001 level) with the average course of
migration (Table 3). This table also shows that the variance in SI-values for different
years is quite low (Avg = 67.7, SD = 6.01, coefficient of variation V= 8.87). Compari-
son of the data in this table also shows that the number of birds observed at the aver-
age site influences the SI-value less (Pearson’s r= 0.61, p> 0.05) than the number of
sites (r= 0.75, p< 0.05). This underscores the importance of the number of observa-
tion sites when studying details of the migration dynamics pattern.
12 THE RING 40 (2018)
Fig. 6. Seasonal pattern of Wood Pigeon migration in Italy presented traditionally as numbers
(Individuals - per station and day) grouped into pentades [5-day periods] (line with black
circles) and as number of peaks in these pentades (line with empty circles)
Similarity Indices for yearly seasonal dynamics of the pigeons’ passage. Numbers of
observation sites and average numbers of birds observed per site in year are given.
Year Sites Pbserved Similarity
1998 60 2 973 58.4
1999 100 3 496 59.3
2000 107 4 718 65.6
2001 113 5 207 69.3
2002 124 6 517 64.6
2003 138 5 074 72.8
2004 129 6 360 78.6
2005 143 3 407 69.4
2006 124 4 585 69.7
Avg 115 4 704 67.5
SD 23.5 1 182 6.01
THE RING 40 (2018) 13
Fig. 7. Left panel: day-by-day total seasonal pattern of migration – numbers: bars – average daily
numbers of pigeons (per station and day) for the whole 1998-2006 period, expressed for year
as percentage share of the yearly total, line – daily data smoothed by 5-day moving average.
DISCUSSION AND CONCLUSIONS
In-depth studies on seasonal differentiation in the intensity of migration are rela-
tively scarce. Frequently the problem of a very high level of fluctuations in numbers
of observed/caught individuals of migrating species is treated as a stochastic variation
caused by the unpredictable influences of various physiological properties of birds,
sensitivity to weather, habitat, and climate. This kind of explanation of the peaks and
breaks observed in seasonal dynamics patterns is convincing mainly when the data
14 THE RING 40 (2018)
come from single years or a short time series. If the time series is long enough, the
stochastic variation should lead to a chaotic distribution of peaks in the summarized
seasonal patterns, most likely in a form close to the Gaussian, normal distribution (if
only one bird population is passing the observation site). In the most common case,
when we have different populations (in terms of source areas, destination winter
quarters and/or direction of migration), a general normal distribution is very unlikely.
This is very well confirmed by the long-term data (since 1961) of Operation Baltic in
Poland, e.g. ten years of data for many species ringed/observed – Busse and Halastra
1981; thirty years of data for the Willow Warbler Pylloscopus trochilus – Piotrkowska
1995 and for the Blackcap Sylvia atricapilla – Busse 1996, Kopiec 1997, Kopiec-Mokwa
1999; and fourteen years of data for the Robin Erithacus rubecula – Nowakowski et
al. 2005, as well as data from other areas, e.g. a nineteen-year study on many species
from Hungary – Gyuracz et al. 2017, or a ten-year study of many species from Pales-
tine – Awad et al. 2017. There are many, many examples of multi-wave distributions.
Thus the results obtained in the current study suggesting several waves in the sea-
sonal dynamics of the Wood Pigeon are very well grounded.
Here we have five or six groups of pigeons passing through Italy in different
parts of the autumn, and the time of the passage, including a few peak days, is quite
THE RING 40 (2018) 15
Fig. 8. Time distribution of migratory peaks (bars – left scale) within certain years shown against
a background of the general migration pattern (line – right scale – corresponds to
smoothed percentage pattern in Figure 7 – left panel)
stable between years. Yearly peaks in different waves, as well as the waves them-
selves, are not regularly of the same relative number sizes, but it is quite normal that
different groups of migrants have their own number size and migration variation.
The general problem here that should be solved in the future is what the waves
found for pigeons really mean. If, as is common in many bird species within the Bal-
tic area, directions of migration are differentiated (e.g. Goldcrest – Busse 1981, Re-
misiewicz and Baumanis 1996), the waves could contain different migratory popula-
tions migrating one by one or they may be composed of a mix of individuals belong-
ing to different migratory populations (Busse and Maksalon 1978). Birds migrating
the same way, but with various areas of origin, can form more or less uniform waves
migrating sequentially – as was supposed for the Coal Tit migrating in the 1974 irrup-
tion (Busse 1978). This was suggested by some signs of separation mechanisms dur-
ing observations of movement through a few bird stations along the Baltic coast.
A similar case was found for the Goldcrest on the central part of the Polish Baltic
coast (Busse 1981). Finally, Blyumenthal (1971) explained the typical occurrence of
waves in migration as the result of bioenergetic mechanisms. From the general popu-
lation pattern of migration of Wood Pigeon in central and southern Europe (Fig. 9)
estimated on the basis of analysis of radioisotopes (Hobson 2009) and bird recovery
atlases for Italy (Spina and Volponi 2008), the Czech Republic and Slovakia (Cepák
et. al. 2008), and Germany (Bairlein et al. 2014), we can rule out the formation of sub-
sequent waves by different migratory populations (birds from the southern flow vs.
16 THE RING 40 (2018)
Fig. 9. General spatial pattern of Wood Pigeon migration in Europe as a rough estimation from
stable isotopes analysis and bird recovery atlases (cited in the text).
those belonging to the northern flow). Therefore, what we have in the area studied is
a wave structure caused by sequential starts of groups originating in more or less dis-
tant territories along the same flyway; or, any mechanism that sorts birds according
to inherited migration distance may play the main role in the creating the wave. In-
herited tendencies may, however, be modified by climate change and/or selection
pressure, as shown for partial migrants. Which explanation better fits wood pigeon
migration is to be discovered in the future, as it could be important for sustainable ex-
ploitation of the pigeon as a hunting resource.
Because our data for Italy are very general, it is not yet possible to discuss
whether these populations travel over the entire country or pass over different re-
gions. We still we do not know whether we have been studying different subpopula-
tions or the entire stream is differentiated only by the time of migration of subgroups
of inhabitants of the same area. Another problem with the seasonal dynamics await-
ing more in-depth study is the occurrence of peak days of migration within waves of
migration. Valuable attempts to study the causes of peaks observed in migration are
presented online by Cavina (2015) and seem to be worthy of further work. Therefore
we need more data, from more years and more sites in various regions, to be able to
draw a detailed picture of the wave and population structure of pigeon migration.
We would like to express our gratitude to all members/observers from Club Itali-
ano Colombaccio. The data collection would not have been possible without their co-
Awad S.I., Farhoud M. H. and Saada Abu R. K. 2017. Long-term bird ringing in Palestine. Ring
Bairlein F., Dierschke J., Dierschke V., Salewski V., Geiter O., Hüppop K., Köppen U. and Fiedler W.
2014. Atlas des Vogelzugs. Aula-Verlag Wiebelsheim.
Blyumenthal T. I. 1971. The development of the autumnal migratory state in some wild Passerine
birds: bioenergetic aspect. In: Ekologiceskye i fizyologiceskyie aspekty piereletow ptic: 111-182.
Busse P. 1978. Wave and population structure during Coal Tit autumn migration in 1974. Not.
Orn. 19, 1-4: 15-36.
Busse P. 1981. Finding of local passage direction as the result of an analysis of retraps and short
distance direct-recoveries. Not. Orn. 22: 3-4.
Busse P. 1996. Modelling of the seasonal dynamics of bird migration. Ring 18, 1-2: 97-119.
Busse P. and Halastra G. 1981. The autumn migration of birds on the Polish Baltic sea coast. Acta
orn. 18, 3: 167-290.
Busse P. and Maksalon L. 1978. Some aspects of Song Thrush migration at Polish Baltic coast.
Not. Orn. 19, 1-4: 1-14.
Cavina E. 2015. Decision making of autumn migrations of woodpigeons (Columba palumbus) in
Europe: analysis of the abiotic factors and atmospheric pressure changes. www. scienceher-
Cepák J., Klvaòa P., Škopek J., Schöpfer L., Jelínek M., Hoøák D., Formánek J. and Zárybnický J.
2008. Czech and Slovak Bird Migration Atlas. Aventinum, Praha.
THE RING 40 (2018) 17
Gyurácz J., Bánhidi P., Góczán J., Illés P., Kalmár S., Koszorús P., Lukács Z., Németh C. and
Varga L. 2017. Bird number dynamics during the post-breeding period at the Tömörd Bird
Ringing Station, western Hungary. Ring 39: 23-82.
Hobson, K. A., H. Lormée, S. L. Van Wilgenburg, L. I. Wassenaar, and J. M. Boutin. 2009. Stable
isotopes (?D) delineate the origins and migratory connectivity of harvested animals: The case of
European woodpigeons. Journal of Applied Ecology 46: 572–581.
Kopiec K. 1997. Seasonal pattern of the Blackcap (Sylvia atricapilla) autumn migration at the Pol-
ish Baltic coast. Ring 19, 1-2:41-58.
Kopiec-Mokwa K. 1999. Dates of migration waves – a coincidence or an effect of biologically
based mechanism? Improvement of the method of analysing the seasonal migration dynamics.
Ring 21, 2: 131-144.
Piotrkowska L. 1995. [Analysis and comparison of the dynamics of autumn migration of Willow
Warbler (Phylloscopus trochilus) at Bukowo, Hel and Vistula Spit.] Diploma work at Univer-
sity of Gdañsk, Poland. (in Polish)
Remisiewicz M., Baumanis J. 1996. Autumn migration of Goldcrest (Regulus regulus) at the east-
ern and southern Baltic coast. Ring 18, 1-2: 3-36
Nowakowski J.K., Remisiewicz M., Keller M., Busse P. and Rowiñski P. 2005. Synchronisation of
the autumn mass migration of passerines: a cse of Robins Erithacus rubecula. Acta orn. 40, 2:
Spina F, Volponi S (2008) Atlante della Migrazione degli Ucelli in Italia. 1. non-Passeriformi.
Ministero dell’Ambiente e della Tutela del Territorio e del Mare, Istituto Superiore per la Protezi-
one e la Ricerca Ambientale (ISPRA). Tipografia SCR-Roma
18 THE RING 40 (2018)