In 1963, in one of his insightful papers on birds in Africa, R.E. Moreau made a shrewd observation about mangroves: “The importance of the mangroves in the bird ecology of the territory seems unknown, but it may be considerable, especially as these trees remain evergreen. Limited inspection of the small Avicennia mangroves, where they were accessible on the edge of dried mud-flats, revealed several species of both African and Palaearctic birds. The big Rhizophora, which line so many miles of river banks, serve as perches for many species, especially herons, raptors and rollers, but nothing seems to be known of their use by small birds. In fact, the bird population of mangroves generally in The Gambia, or anywhere else in Africa, remains to be carefully investigated. Since the habitat is so exceptionally simple, with extensive pure stands, it would be interesting to work out the seasonal fluctuation of food resources and utilization of the mangroves and it would not be difficult once the physical obstacles are overcome” (Cawkell & Moreau 1963). Half a century later, Moreau's questions are still unresolved. This is remarkable, especially in the light of detailed studies on the dynamics of bird communities and food resources in Central and South American mangroves (Lefebvre et al. 1994, Lefebvre & Poulin 1996), in Malaysia (Noske 1995) and Australia (Mohd-Azlan et al. 2014). The single exception is the pioneering study by Altenburg & van Spanje (1989) in Guinea-Bissau. In the winter of 1986/87, they found a low diversity of Palearctic insectivorous species in the mangroves of Guinea-Bissau, but densities of Willow Warbler Phylloscopus trochilus, Melodious Warbler Hippolais polyglotta and especially European Reed Warbler Acrocephalus scirpaceus, were high. In fact, such densities would imply the entire mangrove area in West Africa (8400 km²; Bos et al. 2006, Giri et al. 2010, Zwarts 2014) to be the winter home of many millions of Palearctic migrants. Such an extrapolation is tentative, however. According to Altenburg & van Spanje (1989), their method of calculating bird densities would likely have resulted in conservative estimates.

The West African mangroves, found between 20°N (Banc d'Arguin in Mauritania) and 7.5°N (Sherbro Island in Sierra Leone), but mostly south of 14°N (Saloum estuary in Senegal), are situated in a rapidly changing world. Consequently, many Afro-Palearctic birds are in double jeopardy: with adverse changes occurring both on the breeding grounds and in West Africa, resulting in steep declines of migratory birds (Zwarts et al. 2009). The declines were particularly large between 1950 and 1993, when the rainfall in the Sahel declined from, on average, +12.7% relative to the long-term average in 1949–1968 to -13.5% in 1969–1993; the latter period is known in West Africa as the Great Drought. Did mangroves, a wintering area hitherto neglected in Afro-Palearctic studies, suffer less from the Great Drought in West Africa than the adjacent Sahelian drylands? And if so, did mangrove forests act as a refugium for Palearctic warblers affected by drought-related habitat loss in the Sahel? In northern Australia, for example, mangroves may provide temporary refugia from fire in adjacent savanna woodland (Mohd-Azlan et al. 2014).

Mangroves, however, are also under pressure, worldwide (Valiela et al. 2001) as well as in West Africa. The Great Drought especially impacted mangrove forests in Senegal and northern Guinea-Bissau, where estuaries became hyper-saline following reduced river discharges (Savenije & Pagès 1992) and mangroves massively died off. The Saloum estuary, for instance, lost 200 km² (40%) of its mangrove vegetation between 1972 and 1986 (Diève et al. 2013) and the Somone estuary even more than 90% (Sakho et al. 2011). After rainfall improved (since 1994), the surface area of mangroves increased, partly due to local replanting of Rhizophora (Conchedda et al. 2008, Sakho et al. 2011, Diève et al. 2013). At the same time, many rice farmers in Guinea-Bissau abandoned their rice fields after 1970, which were subsequently recolonised by mangroves (Lourenço et al. 2009, Zwarts 2014).

Based on fieldwork in early 2014, we present bird densities in West African mangroves along a latitudinal gradient between 16ºN and 11ºN, separately for tree species and tree height. These data are used to estimate the importance of West African mangroves for wintering Palearctic warblers in the Sahel region at large.

METHODS

The fieldwork took place between 8 January and 8 February 2014 in Senegal (Senegal Delta, Somone estuary, Saloum estuary) and Guinea-Bissau (Rio Cacheu, Rio Mansoa, Rio Geba, Rio Grande de Buba, Rio Cacine) and between 9 and 15 March 2014 in the Casamance (Senegal; Figure 1). The bird counts were done during the entire daylight period. We recorded all bird species in 380 plots (Table 1), but this paper exclusively deals with insectivorous birds feeding in mangrove vegetation. We did not distinguish between Iberian and Common Chiffchaff Phylloscopus ibericus and P. collybita because most were silent.

We aimed to obtain bird densities as close to real densities as possible. To that end, we used a habitatspecific and time-consuming method. In areas with scattered Avicennia scrub (trees <3 m) 3–4 persons slowly walked in broad front along a transect of 10–15 m wide, recording all birds present. In Avicennia forests with large trees (12–18 m high), we intermittently made stops to intensively scan the canopy for birds. Fragmented mangrove stands not wider than 15–30 m were simultaneously watched by 1–3 persons from the outside, to account for birds leaving the transects. With this method, we covered on average 22 m²/min, varying between 4 and 80 m²/min, depending on tree density and tree height. Dense or impassable forests (mostly Rhizophora) were entered along small creeks, sometimes by boat. The surrounding woody vegetation was then penetrated on foot to perform a half-circle focal watch of 5–20 m during 10 minutes. We covered 8 m²/min with this sit-and-wait strategy, varying between 3 and 15 m²/min.

We took great pains to record all birds by sight and ear, until we were convinced that no more birds were present within the transects or plots. Upon detection, birds were scored as either silent or vocal and whether they started calling or remained silent within the observation period. Activity and position in the tree were scored for birds within visual range. After we finished the density count, we used 5 min of playback song of European Reed Warbler, Melodious Warbler and Subalpine Warbler Sylvia cantillans to elicit vocalisations of silent (and possibly undetected) birds, then continued our observations for another 5 to 10 min. Using playback song in mangroves, at least in January, was not successful in terms of detecting hitherto undetected birds. Palearctic passerines were either calling or not, irrespective of playback sounds. Yellow-Crowned Gonolek Laniarius barbarus was the only species that - when present - always showed a reaction. As we were unable to determine whether playback attracted birds from outside the plots/transects, which would have jeopardized the reliability of our census method, we stopped using playback sound halfway through the fieldwork period. The additional data obtained via playback are not used here.

Our method to determine bird densities covered, depending on the method, 8 to 22 m²/min, and was a slow procedure compared to other census methods. A faster method would have been to base the inventory on bird sounds only (with occasional sight observations), as did Altenburg & van Spanje (1989); they covered 28.4 ha in about 15 hours, equivalent to 315 m²/min, and concluded that the resulting density should be considered as a minimum. Our observations give an idea of the order of error. Of 13 Reed Warblers detected by eye and 15 by ear, 7 and 6, respectively, remained silent during the rest of the observation period. Hence, if the inventory would have been based on vocalisations only, we would have missed a quarter (7/28) of the Reed Warblers. This overlooked fraction would have increased if less time had been spent in the sampling plot, a condition met in the study of Altenburg & van Spanje (1989).

Photo1.

JvdK doing a plot count in a dense Rhizophora mangrove. The ‘physical obstacles’ mentioned by Cawkell & Moreau (1963), especially the impenetrability and the deep and sticky mud, are still very much in attendance when entering most mangrove forests (photo Leo Zwarts).

Figure 1.

The nine study areas along the coast of West Africa (names given left of the coast line; red dots show the visited sites); mangroves are indicated in black. The average annual rainfall (right y-axis) varies from less than 300 mm/year in the Senegal Delta to 2400 mm at the border between Guinea-Bissau and Guinea-Conakry.

Photo 2.

Extensive Avicennia mangrove with Rhizophora (being higher and dark green compared to Avicennia) along the creeks; Casamance River, 12.875°N and 16.716°W; 26 August 2008 (photo Leo Zwarts).

Wilson & Cresswell (2010) found a decline in detectability with time after sunrise, but such a temporal effect was not found in our data, probably due to our slow and stationary sampling methods. Moreover, Reed Warblers in particular were quite vocal throughout the day: 6 out of 7 birds were vocal during plot counts before 9:00 h local time, compared to 14 out of 16 between 9:00 and 16:00 h, and 1 out of 3 after 17:00 h.

We also tried to validate the accuracy of our method by performing a repeat census of birds in the same area, each time with three observers. On 8 January 2014, during one hour in the late afternoon, we saw and heard 2 Subalpine Warblers and 2 Reed Warblers on 1141 m² of mangroves (Avicennia, 1–5 m high) in the Senegal Delta. The next early morning we arrived at the same numbers after an inventory of again one hour. The second check was based on a survey of three hours in the afternoon of 4 February 2014 along the Rio Mansoa, where 5 Reed Warblers and 2 Beautiful Sunbirds Cinnyris pulchellus in 1486 m² of mangroves (mainly Avicennia, 2–3 m high) were recorded. Between 7:15 and 7:20 h the next morning, 5 Reed Warblers were heard in about the same sites as located the day before. In the next two hours, we recorded 5 Reed Warblers and no other birds. These few data suggest that our intensive method of quantifying bird densities is rather accurate.

Table 1.

Average bird density (n/ha canopy) of insectivorous bird species in West African Avicennia germinans mangroves, for five regions.

In 380 plots, we measured height and crown width of the seven tree species (Avicennia germinans, Rhizophora racemosa, R. mangle, R. harrisonii, Laguncularia racemosa, Conocarpus erectus, Drepanocarpus lunatus) that constituted the mangrove habitat. The height of trees was estimated by eye or measured (trees >3 m) with a laser rangefinder. Height and crown width were noted for each tree when individual trees could be discerned. Crown width was used to calculate tree cover, assuming trees are circular. When tree species formed a closed canopy, average height and cover were estimated for the entire plot rather than for the individual trees. Most study plots lacked an understorey except young mangrove shoots, but Avicennia habitat in the Senegal Delta was less homogeneous with a scattering of tamarisk Tamarix senegalensis and cattail Typha australis nearby.

In the field, we used a GPS linked to a laptop with geo-referenced, high-resolution imagery (resolution 0.6–1.5 m) of our study plots. The tracks were used to validate our field estimates of plot size and tree cover. Tree cover in the plots varied between 5 and 100%, but usually exceeded 80%. Bird densities are expressed as numbers per ha canopy (thus 100% tree cover).

RESULTS

In several types of mangrove forest, insectivorous birds were absent, i.e. in Laguncularia racemosa (1272 m² surveyed) and in (the rare) Drepanocarpus lunatus (110 m²); in Conocarpus erectus (3698 m²) a sunbird was recorded only once. In contrast, mangrove forests consisting of Avicennia and Rhizophora (for the latter, three species combined), held high bird densities, especially Avicennia (Table 1 and 2).

In West African mangroves, Palearctic birds outnumber African birds in the insectivorous guild during the northern winter, but the fraction of African birds gradually increases south of 14°N (Figure 2). Latitudinal gradients were also recorded for Palearctic passerines. Subalpine Warbler was common in the Senegal Delta (16.5°N), less common in the Saloum (13.5°N) and absent in plots further south. European Reed Warbler was the most common bird species along the Rio Geba and further north (>11.5°N), but absent further south. In contrast, Willow Warblers were observed only at 11.0–11.5°N, i.e. in our most southerly plots.

Figure 2.

The percentage Palearctic birds among foliage-foraging insectivorous species between 11°N and 17.5°N in West African coastal mangrove areas and in upland habitats in Mauritania, Senegal and Mali (Zwarts et al. in prep.; data from December–February 2007–2012. The square refers to data collected by Altenburg & van Spanje (1989) in mangroves in Guinea-Bissau in the winter of 1986/87. The third-degree polynomial relationship refers to both data sets combined.

Table 2.

Average bird density (n/ha canopy) of insectivorous bird species in West African Rhizophora mangroves, for four regions (see Table 1).

Figure 3.

Average position within the canopy of mangrove trees where birds were observed feeding. The average height (m above ground) is plotted against the average position of the bird relative to the height of the tree (%). The trend is significant, but based on a small sample of 30 birds.

Except the (mostly) nectivorous sunbirds, all bird species encountered during the surveys were foliageand branch-foraging insectivores. The bird density is given as numbers per surface area, but one would expect that bird density is higher in large trees than in low scrubs, because there are more leaves, and thus likely more insects, per surface area in taller vegetation. Contrary to expectation, bird density per surface area decreased with tree height (Table 3). Avicennia scrubs were more attractive than large trees. A similar trend was found in Rhizophora, with 16.1 birds/ha canopy in trees 1–3 m high (3095 m² surveyed) against only 4.7 birds/ha canopy in higher trees (4–6 m; 4231 m²).

Palearctic insectivorous species, and Northern Crombec Sylvietta brachyura, were most frequently recorded in the lower half of the canopy. In contrast, African species like Common Wattle-Eye Platysteira cyanea, Brown Sunbird Anthreptes gabonicus and Beautiful Sunbird Cinnyris pulchellus were mostly found in larger trees where they foraged high in the trees (Figure 3).

DISCUSSION

Bird density

The species-specific latitudinal variations in density (Tables 1 and 2) correspond with existing knowledge of the distribution of the same species in the West African Sahel, e.g. Subalpine Warbler residing north of 13°N and Willow Warbler south of 12°N (Zwarts et al. in prep.). The similar latitudinal distribution of Palearctic birds within the mangrove belt, therefore, is unlikely to be typical of mangrove habitats but rather reflects habitat preferences associated with latitude and/or rainfall (see Figure 1).

Within the mangrove belt, European Reed Warblers were found across a wide latitudinal range (12–16.5°N). In West Africa, this species is common in mangrove habitat and more sparsely observed in other habitats. Its density was low (0.07 birds/ha) in coastal rice fields adjacent to mangroves in Guinea-Bissau (Bos et al. 1986), and it has been captured in small numbers in cattail stands and reed beds in Senegal (Sauvage et al. 1998, Flade 2008), a rare habitat in West Africa. During the northern winter, Reed Warblers are almost completely absent from the same latitudinal band in the rest of West Africa, in wetlands as well as in drylands (Zwarts et al. 2009), although recorded in dry habitats further south where they are associated with low trees and rank grass (Dowsett-Lemaire & Dowsett 1987, Procházka et al. 2008).

Table 3.

Bird density per ha canopy in Avicennia germinans as a function of tree height. To rule out latitudinal variation (Table 1), the Senegal Delta in the north and the Rio Grande de Buba and Cacine in the south were left out.

The density counts by Altenburg & van Spanje (1989) in mangroves in the northwestern part of Guinea-Bissau covered the same region as our counts along the Rio Geba and Mansoa (Table 1). Their density of European Reed Warblers (9.3/ha in Avicennia and 3.5/ha in Rhizophora and mixed mangrove vegetation) was lower than our estimate, for which several explanations can be put forward. First, their densities are mostly based on vocalising birds, a method leading to underestimates. Secondly, our estimates refer to ha canopy, while their estimate refers to total surface, including open spaces between the trees, which makes for a difference in canopy cover of about 10%, on average. Thirdly, in the 1980s Reed Warblers were at a low ebb. In Great Britain and Ireland, for example, one of the presumed major origins of mangrove wintering Reed Warblers in West Africa (Procházka et al. 2008, Zwarts et al. 2009), the distribution of Reed Warblers in 2008–11 increased by 44% as compared to the distribution in 1968–72, and numbers more than doubled, with a 36% increase during 1995–2010 (Balmer et al. 2013).

Of Palearctic migrants, Reed Warblers were the only species present in large numbers during our surveys. In contrast, Altenburg & van Spanje (1989) recorded a density of 4.2 Melodious Warblers/ha, a species we only recorded frequently in adjacent upland south of 15°N, mainly in Faidherbia albida. Apparently, their presence in mangroves varies from year to year, as exemplified for the mangroves of the Saloum where it was the most common bird species in February 2011 (1.3 ha investigated; van der Kamp, Sikkema & Zwarts, unpubl.).

Bird densities in mangroves are known to fluctuate. The abundance of insectivorous migratory birds in mangroves in Venezuela showed seasonal variations, reaching a maximum in the late wet and early dry season when their food supply (arthropods) is most abundant (Lefebvre et al. 1994). The same dynamics have been described for Panamanian (Lefebvre & Poulin 1996) and north Australian mangroves (Mohd-Azlan et al. 2012, 2014). The period of flowering of Avicennia and Rhizophora is subject to large seasonal, latitudinal, annual and local variations (Duke 1990, Clarke & Myerscough 1991), although generally peaking in the late wet and early dry season (October–November). Flowering of Avicennia, and of mangroves in general, has a major impact on insectivorous birds (Lefebvre & Poulin 1996). When flowering phenology differs between species of mangroves (Mohd-Azlan et al. 2014), successive waves of flowering (and hence, insect abundance) may offer insectivorous birds a longer period of food abundance.

In our study, the average density of insectivorous birds amounted to 11 birds/ha canopy in Rhizophora and 22 birds/ha canopy in Avicennia. Altenburg & van Spanje (1989) also recorded a lower density of insectivorous birds in Rhizophora (14/ha) than in Avicennia (19/ha). An explanation might be that the leathery leaves of Rhizophora attract fewer herbivorous insects than the succulent leaves of Avicennia (Robertson & Duke 1987). Why migrant warblers in West Africa prefer low mangroves and the lower part of the canopy (Figure 3) is still unknown. Elsewhere in the world, bird species differ regarding the foraging height in mangroves (Lefebvre et al. 1992, Noske 1995), but in general are less specialized in their use of foraging heights than in foraging behaviours (Mohd-Azlan et al. 2014).

The guild of insectivorous birds in West African mangroves reaches a density similar to that in mangroves in Mexico (Hutto 1980), Malaysia (Noske 1995) and Australia (Noske 1996, Mohd-Azlan et al. 2012). The density is high compared to scrubs and trees in dryland West Africa: the average density of insectivorous birds in 164 tree species in Sahelian West Africa varied between 0 and 102 birds/ha canopy and was higher than in Avicennia in only 15 (i.e. 9%), of the tree and scrub species (Zwarts et al. in prep).

Estimate of bird populations in West African Mangroves

The densities of birds in mangroves can be used to calculate total wintering numbers in this habitat type, taking into account latitudinal variations in density. Reed Warblers occur in the mangroves north of the Rio Grande de Buba in Guinea-Bissau, with a total surface of 3500 km² canopy (Zwarts 2014). We have no quantitative data on the fraction of the vegetation consisting of Avicennia and Rhizophora, but we estimate from our field work that Avicennia is twice as common as Rhizophora. Nearly all mangroves are smaller than 6 m, hence we use an average density of 19 Reed Warblers /ha canopy for Avicennia (Table 3) and of 2.4 for Rhizophora, from which we arrive at a wintering population of 5 ± 1 million Reed Warblers in the West African mangroves. This estimate would be 10% lower or higher when respectively 60% or 75% of the vegetation had consisted of Avicennia (instead of 67%); we doubt that the proportion of Avicennia exceeds either estimate.

Figure 4.

The % ringed birds found dead in the Sahara in January–June relative to the numbers found dead between 4°N and 36°N between 1 July and the preceding 1 July; total n is given. Calculated for the 9 (or 20%) most wet (preceding) years in the Sahel between 1961 and 2005 (1962–1966, 1968, 1995, 2000, 2004; rainfall, on average, +8.8% relative to average of the 20th century) and the 9 driest years (1973–1974, 1983–1985, 1987–1988, 1991, 2003; rainfall, on average, -24.8% relative to average of the 20th century). Based on EURING data and a reanalyse of data given by Zwarts et al. (2009). χ²-test: *P < 0.02, **P < 0.01, ***P < 0.001.

The breeding population of the Reed Warbler in Europe has been estimated at 2.6–5.0 million pairs (BirdLife International 2004). Given that 42% of the wintering population in Africa consists of first-year birds (EURING data, analysis in Zwarts et al. 2009), the winter population can be estimated at 9 to 17 million birds. If correct, 30–50% of the total European population would then be concentrated in the small strip of West African mangroves during the northern winter. This proportion is even higher for European Reed Warblers from West, North and East Europe. These birds migrate through Iberia and spend the winter in West Africa, while the birds breeding east and south of Austria migrate through northeastern Africa and winter in central and eastern Africa (Dowsett-Lemaire & Dowsett 1987, Procházka et al. 2008, Zwarts et al. 2009, Andueza et al. 2013, Procházka et al. 2013). Without the Reed Warblers from SE Europe (wintering population: 3–6 million birds; BirdLife International 2004), 50–80% of the population from West, North and East Europe (6–11 million birds) would spend the northern winter in the mangroves.

However, only 10% of the ring recoveries from the winter period refer to Reed Warblers from coastal West Africa and 90% from the Sudan-Guinean zone within West Africa between 6 and 15°N (Procázka et al. 2008, Zwarts et al. 2009). The reporting rate of ringed birds in Africa is extremely low, but must be close to zero in the inaccessible mangrove habitat. Hence it would be incorrect to assume that Reed Warblers mostly spend the northern winter in the Sudan-Guinean zone, even considering the huge range from which field observations and ringing recoveries are available. Although much is still unclear, it is obvious that a substantial part of the West European Reed Warbler population spends the winter in West African mangroves. This is particularly true for the populations from Britain, Ireland and Spain, for which ring recoveries suggest a coastal winter distribution (Procházka et al. 2008). Cresswell (2014) predicts that many juveniles will end up in unsuitable habitat, but due to lack of data on age composition per latitude and habitat type we can only speculate whether adult Reed Warblers dominate in the (northern) mangroves.

A similar exercise for the Subalpine Warbler indicates a wintering population of 0.9 million birds in the West African mangroves, given an average density of 2.8 and 1.9 birds/ha canopy in respectively Avicennia and Rhizophora north of the Rio Grande de Buba. BirdLife International (2004) estimated the European breeding population of the Subalpine Warbler at 1.4–3.2 million pairs, equivalent to 5–11 million birds in winter, to which must be added an unknown number originating from breeding grounds in northern Africa. From this, one may derive that about 10% of the estimated world population of this species winters in the West African mangroves.