Every year more than 2 billion passerines move between Africa and Europe (Hahn et al. 2009). Several of these species, mainly warblers, are occasional nectarivores, which land at stopover sites with pollen grains stuck on their body (Cecere et al 2010). Nectar feeding by warblers is well described at a stopover site located in the Sahara desert (Salewski et al. 2006) but also at Ventotene, an Italian island reached by birds after the crossing of the Mediterranean Sea during spring migration (Schwilch et al. 2001). Nine species of passerines are recorded as nectar feeders on Ventotene Island, six of which are trans-Saharan migrants. These birds use at least six plants belonging to different families to forage on nectar: Ferula communis, Brassica sp., Lavatera arborea, Prunus avium and Pittosphorum tobira and probably Malva sylvestris (Cecere et al. 2011). As a consequence of nectar feeding, pollen grains can remain stuck to the bill and the forehead of the birds (Ash 1959, Ash et al. 1961, Laursen et al. 1997, Schwilch et al. 2001, Ortega-Olivencia et al. 2005, Salewski et al. 2006). In Denmark, Laursen et al. (1997) found pollen grains belonging to at least nine plant taxa on the bill of some migrating warblers. In the same study, the authors found pollen belonging to plants that are abundant in the Mediterranean region and absent in Denmark, and they suggested that the pollen found on birds can be used as a marker for migratory routes.

Similarly to exotic birds, European songbirds seem to be able to pollinate plants by carrying pollen grains on their plumage and bill. This has been shown for Blue Tits Cyanistes caeruleus, which can pollinate an exotic plant, Fritillaria imperialis, in the United Kingdom (Búrquez 1989), but also for Sardinian Warbler Sylvia melanocephala, Eurasian Blackcap S. atricapilla and Chiffchaff Phylloscopus collybita, which regularly pollinate the European native plant Anagyris foetida in Southern Spain (Ortega-Olivencia et al. 2005).

Despite several reports about pollen transport by Old World songbirds, an in-depth analysis of pollen species carried by migrants landing in Europe and coming directly from Africa is missing. This kind of analysis is necessary to assess which plants are used for nectar consumption at stopover sites by songbirds crossing the Mediterranean Basin. Furthermore, it can show specific relationships between bird and plant species. Overall, this information can add to the knowledge of the ecology of European migrants and inform habitat management at stopover sites. Lastly, these studies could be able to assess the role of European songbirds in plant pollination.

In this paper we analyze pollen grains found on 147 birds belonging to eight species landing at a stopover site located in the Central Mediterranean during spring migration, and discuss the possible role of migrants in the dispersion of pollen. Moreover, we analyze the flowering patterns of the two plants most frequently used by songbirds on Ventotene Island and the staging patterns of occasional nectarivore migrants at the stopover sites (Cecere et al. 2010), in order to assess whether there is a relationship between flower availability and pollen grains found on bird species, thus between bird species and the plants they feed on.

METHODS

The study was carried out in spring at Ventotene Island during the years 2006–2009. The island (1.3 km²) is 50 km off the Tyrrhenian coast of Italy; it is inhabited by a small community (450 people) and covered by several small fields characterized by crop rotation (cultivated and set aside). Since 1998, the Italian Ringing Centre (CNI-ISPRA) has carried out the “Progetto Piccole Isole — Small Islands Project” at a large network of sites, including Ventotene, in order to investigate stopover strategies adopted by songbirds during spring migration (Spina et al. 1993). In this study, continuous and standardised ringing activity was carried out from 1 April to 31 May. A total length of 282 m of mist-net was checked for the presence of birds every hour and every day, from one hour before sunrise till one hour after sunset; nets were not operated in bad weather conditions (strong wind or heavy rain).

Seasonal flowering patterns and presence of migrants

Seasonal flowering patterns were obtained for the two plants most used by passerines on Ventotene, following Cecere et al. (2010): F. communis (Apiaceae) and Brassica sp. (Brassicaceae). Anzalone & Caputo (1976) referred to the presence of Brassica oleracea L. ssp. robertiana (Gay) Rouy et Fouc. and B. fruticulosa Cyr. ssp. fruticulosa on the island. However, Brassica is an important genus in agriculture, with a wide range of crops and it is likely that hybridisation may have occurred in the past few decades. Taking into account this possibility, and the well known taxonomic uncertainty of this genus (Gladis & Hammer 2001), we refer to this species as Brassica sp.

The census of blooming Brassica, sp. and F. communis was carried out within 10 plot areas in 2006 and 2007. All flowers on plants inside 5 plots were counted every five days, from 1 April to 31 May. Plots had an area of 2 m², were placed at least 200 m from each other on land with the following exposures: SE, E, NE, N, and W. In the second year, each plot was moved a few meters in order to sample different plants. For Brassica sp. all open flowers were directly counted. In the case of F. communis, the percentage of open flowers per each inflorescence was estimated. Generally, the branches of F. communis have inflorescences of two sizes, which were categorised as big or small. After each count, we computed the flowering index of F. communis as follows:

where FoP = open flowers; Ftb = total flowers of big inflorescence; Fts = total flowers of small inflorescence; N = total number of big inflorescences; M = total number of small inflorescences. This formula allowed us to obtain an index of abundance of blooming F. communis by the estimated percentage of open flowers for each inflorescence present within the plot areas. The index was used because a count of each single little flower was not possible without dissecting the inflorescence.

The migrating patterns of the four most common migrants landing on Ventotene, which regularly feed on nectar, viz. Eurasian Blackcap, Subalpine Warbler S. cantillans, Garden Warbler S. borin and Common Whitethroat S. communis, (Schwilch et al. 2001, Cecere et al. 2010) were estimated by summing up all captures made every five days in 2006 and 2007. Because data were not always normally distributed (Shapiro—Wilk test), trends in 2006 and 2007 were compared by Kolmogorov—Smirnov test, both for the four migrating species and for blooming Brassica sp. and F. communis.

Figure 1.

A) Garden Warbler feeding on the nectar of Ferula communis. B) Garden Warbler on F. communis with breast dirty with nectar and pollen (photos by Marzia Mirabile). C) Garden Warbler showing fresh pollen on the bill and forehead. D) Garden Warbler with dried pollen on the bill and forehead (photos by Jacopo G. Cecere).

RESULTS

Pollen collection

A total of 147 pollen samples (95 fresh and 52 dried) were collected on as many birds. With regards to fresh pollen, an average number of 235.3 ± 2.3 (n = 82) pollen grains/sample were recorded on a fraction of 86.3% of samples, 50–200 pollen grains/sample (100 ± 16.6, n = 8) on 8.4% of samples, while less than 50 pollen grains/sample (8.4 ± 4.1, n = 5) were recorded on 5.26% of samples. As for dry samples, over 200 pollen grains/sample (245.1 ± 7.8, n = 17) were recorded on a fraction of 32.7% of samples, 50–200 pollen grains/sample (99.4 ± 8.6, n = 18) on 34.6% of samples, and less than 50 (16.3 ± 3.01, n = 17) on 32.7% of samples. Frequency classes were not defined for samples with less than 50 pollen grains. As a consequence, the total number of classified samples was 125 (90 fresh and 35 dried). On the whole, pollen grains from 38 plant taxa were recorded in all samples, of which 7 taxa are certainly not present on Ventotene, while 3 taxa are very rare (Anzalone & Caputo 1976; pers. obs.). The lists of plant species recorded in fresh and dried samples are reported in Table 1 A and B, respectively.

Figure 2.

Flowering of Brassica sp. and Ferula communis recorded from 1 April to 30 May, in 2006 and in 2007. For Brassica, the trend was obtained by counting flowers in 10 plot areas, while for Ferula communis, the trend was obtained by a flowering index calculated after each control (for details see Methods).

A total of 26 plant taxa were recorded in fresh samples (Table 1A), among which Brassica sp. (Fig. 4A) was the most frequent. This taxon was found in 72.2% of fresh samples and, among these, it was the predominant pollen species in 75.4% of cases. F. communis (Fig. 4B) was present in 51.58% of fresh samples and, among these samples it was the predominant pollen species in 81.63% of cases. Pollen grains from another 24 plant taxa were recorded with lower frequencies in fresh samples, and their presence was generally classified as ‘important minor pollen’ or ‘minor pollen’, except for Malvaceae (Fig. 4C), which represented the ‘predominant pollen’ in one fresh sample (Table 1A).

Figure 3.

Estimated trends of landings of four migrants on Ventotene island, derived from the ringing activity carried out from 1 April to 30 May, in 2006 and in 2007.

Plant species that contributed to the predominant pollen in dry samples differed from those in fresh samples (χ² = 73.35, df = 3, P < 0.0001); Malvaceae, Cyperaceae and Fraxinus sp. were excluded from the analysis since each of them was present as predominant pollen in only one sample. A total of 32 plant taxa were recorded in dry pollen samples (Table 1B) and among these, F. communis was the most frequent species. It was present in 23.2% of dry samples, and among these, it was the predominant pollen in 40.9% of cases. Citrus sp. (Fig. 4D) was the second most frequent pollen taxon, present in 21.05% of dry samples and, among these, it was the predominant pollen in 65% of cases. Brassica sp. was the third most frequent pollen taxon, present in 17.9% of dry samples and, among these it was the predominant pollen in 11.7% of cases. The fourth most abundant pollen taxon was Eucalyptus sp. (Fig. 4E), present in 11.6% of dry samples, and among these, it was the predominant pollen in 72.7% of cases. The other 28 plant species showed lower frequencies, being classified in most cases as important minor pollen or minor pollen (Table 1B).

Plant species recorded as predominant among fresh samples differed between short and long-distance migrants (χ² = 77.35, df = 1, P < 0.0001). All Eurasian Blackcaps sampled with fresh pollen (n = 7) and 97% of Subalpine Warblers (n = 37) had Brassica sp. as predominant pollen. On the contrary, the two longdistance migrants carried mainly pollen of F. communis: 94% of Garden Warblers (n = 35) and 100% of Common Whitethroats (n = 7). Malvaceae was excluded from the analysis since it was present only in one sample.

Unfortunately, data and sample size did not allow us to perform a Chi-square test for dry samples. Nevertheless, differences between short and long-distance migrants in recorded predominant plants appeared to be present among dry samples too. Citrus sp. was the most common taxa as the predominant plant among dry samples collected on the two short-distance migrants: 60% of Subalpine Warblers (n = 10) and 50% of Eurasian Blackcaps (n = 10), while it was absent among dry samples collected on the plumages of long-distance migrants. On the contrary, F. communis was recorded only on Garden Warbler, (60%; n = 10) and on the two sampled Common Whitethroats. Pollen grains of Eucalyptus sp. were found both on short and distance migrants (40% of Subalpines and 20% of Eurasian Blackcaps and Garden Warblers).

Table 1.

Occurrence of plant species per abundance class recorded among samples of pollen collected on the plumage and bill of migrating songbirds staging at Ventotene Island (Italy) during spring 2008. A) In fresh samples, and B) in dried samples. Pollination mechanisms: A = anemophilous; Z = zoophilous; M = mixed anemophilou- zoophilous (Oberdorfer 1949, Dommée et al. 1999, Horlock et al. 2000, Smeins 2003, Escudero et al. 2008).

Figure 4.

Pollen grains of A) Brassica sp., 63×, B) Ferula communis, equatorial view, 63×, C) Cirrus sp., 63×. D) Aggregate of pollens stuck to a feather, 10×.

Pollen percentages were not calculated for Common Whitethroat with dry pollen or for the other species, Sardinian Warbler, Chiffchaff, Wood Warbler P. sibilatrix and Lesser Whitethroat S. curruca, due to small sample sizes (sample size of each species < 4).

DISCUSSION

In this study we show that pollen grains of a large array of plant species were carried every spring across the Mediterranean Region by migrating warblers. On the plumages of 147 birds staging at the stopover site of Ventotene Island, we found pollen grains belonging to 38 different plant taxa, seven of which are absent from the island (Table 1). This can only be explained by nectar feeding of warblers at previous African stopover sites (e.g. see Salewski et al. 2006) and on Mediterranean islands (Cecere et al. 2010). In addition, many other pollen species stick accidentally onto the plumage of birds, such as pollen from anemophilous and nectar-less plants. The nine plant taxa recorded on the plumage of migrants trapped in Denmark had also been found on migrants landing on Ventotene, with the exception of Cannabis sp. (Laursen et al. 1997).

In our study, Brassica sp. and F. communis were by far the most common species in the fresh pollen samples, which is consistent with the importance of these plant species for nectar-feeding songbirds on Ventotene (Schwilch et al. 2001, Cecere et al. 2010). However, these two plant species showed a different distribution among bird species. Almost all fresh pollen samples collected on Subalpine Warblers and Eurasian Blackcaps contained Brassica sp. as the predominant pollen, while almost all Garden Warblers and Common Whitethroats carried F. communis. The first two bird species are short-distance migrants staging at Ventotene earlier than Sub-saharan migrants, coinciding with the flowering peak of Brassica sp. (Fig. 2 and 3). In contrast, Garden Warbler and Common Whitethroat are long-distance migrants passing later in the season, when the number of flowers of Brassica sp. starts decreasing, while the flowers of F. communis are abundant (Fig. 2 and 3). This suggests that the pollen of fresh samples may have originated from nectar feeding in situ.

The remaining 24 plant taxa found in the fresh samples were only recorded with low frequencies, and a good proportion (38.5%) of the total taxa found in fresh samples belong to anemophilous plants (Table 1A). In these cases, the presence of pollen on plumage does not originate from nectar feeding, but rather from the wind or through simple contamination when the birds perch on such plants.

The pollen found in dry and agglutinated samples was likely to derive mainly from sites visited by migrants before their staging at Ventotene (Cecere et al. 2010). Most spring migrants spend less than two days on Ventotene island (Goymann et al. 2010; Tenan and Spina, unpubl. data) and this period is likely too short for pollen to dry. The pollen composition of dry and fresh samples were quite different: Eucalyptus sp., Citrus sp. and F. communis, were mainly present in dried samples as predominant pollen, while Brassica sp. was also common, but mainly present as secondary pollen. The other 28 plant taxa recorded in the dry samples showed very low frequencies and, like in fresh samples, a good proportion (46.87%) of the total taxa found in dry samples belonged to anemophilous plants (Table 1B).

F. communis and Brassica sp. are abundant on Ventotene but also at many other sites within the Mediterranean; Citrus sp. and Eucalyptus sp. are cultivated in the whole Region, but are scarce and absent, respectively, on Ventotene. Pollen grains of Citrus sp. were recorded on migrating Sylvia warblers trapped in the United Kingdom and in France (Ash 1959, Ash et al. 1961). Eucalyptus and Citrus were the genera most abundant among the pollen samples collected on warblers trapped in Denmark by Laursen et al. (1997). Based on our results, the pollen grains of Citrus sp. and Eucalyptus sp. were mainly carried by the shortdistance migrants Eurasian Blackcap and Subalpine Warbler, whereas Garden Warbler, which is a longdistance migrant wintering in tropical and equatorial areas, was mainly found carrying F. communis.

The difference in species composition between the pollen carried by long- vs. short-distance migrants may be due to the different migration phenologies across the Mediterranean Region, which coincide with different flower availability, as occurs at Ventotene for F. communis and Brassica sp. In addition, different migration routes and patterns may have a role. Short and long-distance migrants could use different stopover sites, characterized by dissimilar vegetations; thus they may come into contact with different nectar resources, carrying, consequently, different pollen species on their plumages.

We hypothesize here that migrating songbirds, and in particular short-distance migrants, can play a role as pollen vectors of Citrus and Eucalyptus species through the Mediterranean region, especially when considering that some migrants land on Ventotene coming directly from North Africa (Pilastro et al. 1995). But in the case of dried samples, we have also to consider that pollen might not be viable for pollination.

Migrants could also have a role as pollen vectors in the interbreeding of Brassica sp. growing at Ventotene, where the plant is largely exploited by birds, or across the Mediterranean. The pollen grains of Brassica sp. were common on bird plumages, principally on the bill, as for Citrus sp. and Eucalyptus sp., and the plant is characterized by large flowers. On the contrary, it is harder to hypothesize a role of songbirds as pollen vectors for F. communis, since it is characterized by small flowers and its pollen is mainly found stuck on breast feathers. We cannot exclude that when a bird is nectar-feeding at a flower, the breast feathers can accidentally pollinate other flowers of the same inflorescence. Contrary to Citrus and Eucalyptus species, pollen of Brassica sp. and F. communis were not found on the body of migrants trapped at other Mediterranean sites. However, this may depend on a lack of data due to the very few studies on this topic in Europe, and future studies are necessary to assess this idea.

For other plant taxa, migrating warblers seem not to play a relevant role in pollen dispersal. The pollen of these plants was scarce on bird plumages and mostly derived from anemophilous plants. However, as suggested by Laursen et al. (1997), less frequent pollen taxa can be used as markers for the distribution of stopover areas visited before landing at the trapping site.

A recent study shows that warblers gain a physiological benefit from nectar consumption: nectar uptake, in fact, led to an increase in the plasma blood glucose concentrations of migrants (Cecere et al. 2011). With the present study we showed that as a consequence of nectar feeding, European warblers carried large amounts of pollen grains across the Mediterranean, suggesting a possible role of songbirds as vectors for pollination. A possible mutualistic relationship might thus occur between Sylvia warblers and some plant species, as for Brassica sp., Citrus sp. and Eucalyptus sp. Moreover, the large number of plants recorded in pollen samples marks the tendency of migrating songbirds to use nectar as a resource during spring migration. In particular, in urban and semi-urban habitats, generally less favourable to migrants, the presence of plants that flower between March and May can be an aid for birds to restore themselves during migration.