Seasonal effects on foraging behaviour of two sympatric species of couas in the western dry forest of Madagascar

Authors


*Insitute of Zoology, Chinese Academy of Sciences, Datun Road, Beijing  1001010, China. E-mail: chouteau@ioz.ac.cn

Abstract

Coquerel’s Coua (Coua coquereli) and Red-capped Coua (Coua ruficeps) occur in the western dry forest of Ankarafantsika in northwest Madagascar. This kind of forest is characterized by an alternating of a dry and a rainy season. Although they belong to the same genus, the two species differ in appearance, with Red-capped Coua being slender than Coquerel’s Coua. We analyse their respective foraging strategies, according to the seasonal variation. The foraging behaviour of both species was different and was also influenced by the seasonal variations. They tended to use the same main substrates but differed in the proportion of foraging techniques according to the season. Seasonal variations probably have an important effect on the prey availability (estimated by the rate of prey capture), the nature of prey and also alternative foraging substrates used, forcing the two species to use different techniques and probably to capture different prey. These different foraging strategies could maintain the coexistence between these two species.

Résumé

Le coua de Coquerel (Coua coquereli) et le coua à tête rousse (Coua ruficeps) se rencontrent dans la forêt sèche d’Ankarafantsika, au nord-ouest de Madagascar. Ce genre de forêt est caractérisé par une alternance de saisons sèches et de saisons des pluies. Bien qu’elles appartiennent au même genre, les deux espèces sont d’apparence différente, le coua à tête rousse étant plus fin que le coua de Coquerel. Nous analysons leurs stratégies alimentaires respectives en fonction du changement de saison. Le comportement alimentaire des deux espèces était différent et était aussi influencé par les variations saisonnières. Ils avaient tendance à fréquenter les mêmes substrats principaux mais la proportion des différentes techniques alimentaires variait selon la saison. Les variations saisonnières ont probablement un effet important sur la disponibilité des proies (estimée par le taux de capture des proies), la nature des proies et aussi les substrats alimentaires alternatifs fréquentés, forçant les deux espèces à recourir à des techniques différentes et probablement à capturer des proies différentes. Ce sont ces différentes stratégies alimentaires qui pourraient permettre la cohabitation de ces deux espèces.

Introduction

Bird distribution is usually correlated with habitat characteristics, particularly vegetation structure (Cody, 1985). Foraging behaviour for the insectivorous forest birds is affected by abundance, distribution and detectability of prey, and the efficiency of different capture techniques (Robinson & Holmes, 1982, 1984; Vanderwerf, 1994). Other determinants of the foraging behaviour are seasonal variations (Hejl & Verner, 1990; Miles, 1990), which can have important consequences on food availability (Block & Brennan, 1993). In some habitats, such as the tropical dry forests, arthropods, especially the litter arthropods, are influenced by the seasonality (Janzen & Schoener, 1968). In a Neotropical dry forest, the availability of ground arthropods varied from one to 50 between dry and wet season (Cuevas, 1995). This can be very important to take into consideration because the abundance of arthropods and diversity affect foraging behaviour of the insectivorous birds (Holmes & Schultz, 1988). Such changes may result in foraging traits: morphology of a bird constrains the type of foraging manoeuvres that can be used efficiently. But morphology does not change between season, and the birds do not modify their foraging manoeuvres as easily as other foraging characteristics, e.g. substrate choice (Martin & Karr, 1990).

We show here the results of a 2-year study of two sympatric terrestrial couas species in a western dry forest in Madagascar. The aim of the study was (i) to compare the morphology of both species and (ii) to compare the foraging strategies of two sympatric and congeneric couas species in the dry forest, and to assess whether, to track food availability, the two species seasonally change their foraging strategies.

Study area and methods

Study site

The study area was the Station forestière d’Ampijoroa, in the National Park of Ankarafantsika, 120 km south of Mahajanga in the dry forest zone of north-western Madagascar (Fig. 1).

Figure 1.

 Location of the study area in the Ankarafantsika region. The bold lines indicate the limits of the burnt forest plots (once-burnt and twice-burnt forests). The other lines indicate the trails around the Jardin Botanique A. The study was performed in the undisturbed area

Rainfall is between 1000 and 1500 mm per year and January is the wettest month. Daily temperatures range from 17 to 35°C with a daily mean of 22°C in the dry season and 29°C in the rainy season (Nicoll & Langrand, 1989). More information about Ankarafantsika can be gained in Alonso et al. (2002).

For the purpose of this study, we considered the rainy season, defined by the period of heaviest rainfall, (December–March), while the dry season was from April to November. All the observations were performed in the unburned part of the Jardin Botanique A (Fig. 1).

Studied species

Couas are a monophyletic group of endemic birds of Madagascar and belong to the Cuculidae (Langrand, 1990; Payne, 1997, 2005). Two terrestrial couas were encountered in Ampijoroa: Coquerel’s Coua and Red-capped Coua. There is no external morphological sexual dimorphism in these species. Previous works were conducted on some aspects of the ecology of Coquerel’s Coua and Red-capped Coua (Appert, 1966, 1970, 1980; Hawkins, 1994; Urano et al., 1994; Masuda & Ramanampamonjy, 1996). They are typically large insectivorous birds, and as such, are likely to be among the most sensitive bird species to deforestation (Thiollay, 1992).

Although these two species belong to the same genus, we were interested to know if they presented some differences in their morphology. We first analysed some morphological variables and then assessed if these possible differences could be linked to different foraging behaviour.

To study morphology of the two species, we analysed 21 specimens of Coquerel’s Coua and 12 specimens of Red-capped Coua from the Museum National d’Histoire Naturelle in Paris. We measured three variables on the bill (length, width and depth) and three variables on the legs (length of tarsus, length of the medium toe and length of the claw of this toe). We measured the total length of the specimens too. We divided each measure by the total length of the individual. This ratio was used to compare the relative proportions of each species. We used a nonparametric test (Kruskal–Wallis test, Sas, 1989) to compare both species.

Foraging data collection

Foraging data were recorded from 1997 to 1998 during the two different seasons as defined above. Observations of foraging events were opportunistic, but we attempted to observe both species at various times of the day, although couas were difficult to locate. We obtained several foraging sequences during at least 1 min but no more than 5 min, always leaving a period of 30 min between two successive observations. Although some investigators recommend taking only the first foraging event for analysis, we retained all to ensure recording inconspicuous foraging events and to reduce biases towards most common foraging techniques and substrates used. Some birds were ringed, and we assumed that the observations were taken from different individuals. For the unmarked birds, we obtained several observations, temporally and spatially separated, so that these observations were likely taken from different individuals.

In all, we obtained for Coquerel’s coua: 66 foraging sequences during the rainy season and 60 during the dry season. For Red-capped Coua, we obtained 44 sequences during both seasons.

Six foraging variables were considered: ability to climb in the understorey vegetation, mean height reached by climbing in the vegetation, techniques used; substrates used, prey size and rate of capture.

First, although couas were considered terrestrial birds foraging on or near the ground level (up to 30 cm above the ground), we observed them foraging sometimes in the upper levels of vegetation. We were interested to know the proportion of prey captured higher in the vegetation, compared with the proportion of prey captured on or near the ground. To capture these prey, the birds had to leave the ground and jump or climb in the vegetation. We hypothesized that, if there was a possibility to climb and search for prey in the upper levels of vegetation, they did it efficiently. The heights of the places where the birds foraged and captured prey were recorded to the nearest 10 cm, by comparing with the height of the observer.

Techniques of capture were recorded as follows (modified from Remsen & Robinson, 1990): GLEAN: prey captured on the substrate, without manipulation of the substrates. Gleaned prey were usually spotted nearby (<0.3 m) and the attack did not involve a flight component. LUNGE: prey captured on the ground by running, (running prey) or in the air (flying prey). LEAP: prey captured on a substrate above the ground by a jump from the ground (without using their wings), SALLY: prey captured on a substrate above the ground, by a powerful jump from the ground and with utilization of the wings. Prey captured by sallying was always higher than prey captured by leaping. PROBE: prey captured after a manipulation of the substrate (e.g. by using the bill to push the dead leaves on the litter or to chase the prey into the dead curled leaves) or to extract a prey from a hollow stem by breaking off some parts of the stem, until the prey was apparent enough to be captured (usually a wingless grasshopper). OTHER included techniques not listed above.

Categories of substrates included: GROUND (on and into litter), LEAVES, TRUNK, AIR (for flying prey) and OTHER. We suspected taht these substrates harboured different kinds of prey, and some differences in their utilization might be useful to segregate the foraging of the two coua species.

Prey size was estimated from the size of the bird’s bill. Three size classes for prey were defined: smaller than 0.5 cm (noted A); 0.5–1.5 cm (B) and longer than 1.5 cm (C). The diets of both species will not be studied here. However, seeds were identified because they were grouped on the ground under a particular tree, and usually we were able to identify the remaining seeds. But because all seeds taken as food were not identified, we presumed that the quantity of seeds eaten was underestimated.

The rate of capture was used as an indirect measure of the prey availability. To calculate it, we selected 20 foraging sequences for each species and within each season. These periods were chosen spatially and temporally independent, with different individuals, to assure the independence of the data. These periods were longer than 30 min to reduce the bias introduced by inactive behaviour (preening, basking, singing and resting). To assess the rate of capture, we used the total number of attempts of capture, observed during each period, divided by the total duration (in minutes) for this period. Comparison was made by using each period as an independent data point (number of capture/minute).

We usually could not determine if an attack was successful, so attack rate refers only to the rate at which prey was attacked, not captured.

Data analysis

For techniques and substrates variables, expressed as proportions, we calculated the average value from all sequences pooled by species and by season. We used multivariate analyses of variance (MANOVA) to compare the intraspecific variations (one coua species compared between two seasons) and the interspecific variations (the two coua species compared into each seasons). Statistical analysis was performed using the procedure GLM (Sas, 1989). We did not include the ‘others’ categories in the MANOVA to avoid nonindependance of proportions (Aebischer, Robertson & Kenward, 1993).

Height reached by climbing to search for prey in upper levels of vegetation, for each species and during the different seasons, was calculated using the procedure TTEST (Sas, 1989). The rate of climbing was analysed by chi-squared tests (procedure FREQ, Sas, 1989). The rate of capture was calculated by a Kruskal–Wallis test (procedure NPAR1WAY, Sas, 1989).

Results

Size of the two coua species

Coquerel’s Coua and Red-capped Coua are similar in size, using the total length as the measure (Table 1), but not in morphology. We obtained significant differences in the variety of morphological measurements for the bill and the tarsus dimensions (Table 1). Differences were also significant for the size of the head, for the Red-capped Coua, it was larger compared with the size of the bird (Table 1).

Table 1.   Morphological measurements of couas
Morphological variablesNo. measurementsMean (variable/total length)χ²
Coua coquereliCoua ruficepsCoua coquereliCoua ruficeps
  1. The second column indicates the length of the different variables in centimetre (±SD) for each species.

  2. χ2 = Kruskal–Wallis test.

  3. *P < 0.05; **P < 0.01; others not significant (P > 0.05); ***P < 0.001; –not calculated.

Upper mandible length1.95 (±0.13)2.15 (±0.09)0.0510.0510.001
Lower mandible length2.55 (±0.11)2.96 (±0.15)0.0670.0702.23
Bill depth0.95 (±0.07)0.95 (±0.04)0.0250.0229.06**
Bill width0.92 (±0.06)1.08 (±0.04)0.0240.0263.95*
Tarsus length4.11 (±0.34)5.10 (±0.20)0.1080.12113.09***
Middle toe length2.58 (±0.16)2.79 (±0.18)0.0680.0661.29
Middle claw length1.07 (±0.09)1.22 (±0.08)0.0280.0290.57
Head length5.45 (±0.14)5.70 (±0.09)0.1350.1438.78**
Total length38.2 (±1.91)42.3 (±1.69) 

Foraging behaviour of Coquerel’s Coua between seasons

We recorded no significant difference in the rate of capture between seasons for Coquerel’s Coua (Table 2). This species captured more prey by climbing during the rainy season, compared with the dry season.

Table 2.   Analysis of the foraging variables used by each species, between the two seasons
Foraging variablesCoua coquereliCoua ruficeps
Rainy seasonDry season Rainy seasonDry season 
  1. The rate of capture is calculated using the Kruskall–Wallis test. The height of capture is calculated using t-test. The % of climbing and the proportion of prey captured are calculated by a qui-square. A MANOVA is used to analyse the proportions of techniques and substrates used.

  2. *P < 0.05; **P < 0.01; ***P < 0.001; others not significant (P > 0.05).

Rate of capture (prey per minute)0.25 ± 0.110.38 ± 0.11H = 2.650.24 ± 0.030.41 ± 0.08H = 3.09*
Height of capture (m)2.89 ± 2.202.13 ± 2.51= 1.170.72 ± 0.481.17 ± 0.75= 2.30
% of capture by climbing7.651.60χ2 = 28.26**6.522.33χ2 = 6.16*
Techniques used
 Glean0.823 ± 0.1060.803 ± 0.121F = 0.940.706 ± 0.1380.682 ± 0.124F = 0.75
 Leap0.074 ± 0.0710.022 ± 0.024F = 28.72***0.095 ± 0.0990.051 ± 0.04F = 7.54***
 Lunge0.038 ± 0.0540.040 ± 0.033F = 0.050.108 ± 0.1520.058 ± 0.073F = 3.81
 Probe0.056 ± 0.0640.119 ± 0.121F = 13.88***0.015 ± 0.0390.142 ± 0.095F = 67.83***
 Sally0.001 ± 0.0030.002 ± 0.005F = 2.450.027 ± 0.0560.006 ± 0.01F = 5.82*
Substrates used
 Air0.003 ± 0.0170.0002F(1,124) = 1.180.062 ± 0.0820.011 ± 0.036F(1,86) = 13.71***
 Ground0.702 ± 0.1270.884 ± 0.088F(1,124) = 84.98***0.642 ± 0.1430.778 ± 0.058F(1,86) = 34.32***
 Leaf0.232 ± 0.1090.089 ± 0.075F(1,124) = 71.82***0.226 ± 0.1350.145 ± 0.055F(1,86) = 13.32***
 Trunk0.061 ± 0.10.026 ± 0.029F(1,124) = 6.85**0.058 ± 0.1180.044 ± 0.031F(1,86) = 0.54
 Prey size  χ2 = 65.59**  χ2 = 38.94**

Difference in techniques used were significant between the two seasons (MANOVA, n = 126, F5,120 = 9.30, P < 0.005). There was also a significant variation for the substrates used (MANOVA, n = 126, F4,121 = 22.67, P < 0.005). Ground was used more frequently during the dry season by the two species (Table 2). However, leaves were more often used as substrates during the rainy season compared with the dry season (Table 2). Coquerel’s Coua also captured more prey in ‘trunk’ during the rainy season (Table 2). Air was a minor substrat for this species.

Coquerel’s Coua always captured prey of different sizes according to the season. More prey of size C was captured during the rainy season; and more prey of size B captured during the dry season (Fig. 2).

Figure 2.

 Prey size captured in the undisturbed forest by each Coua species by seasons

Foraging behaviour of Red-capped Coua between seasons

We recorded a significant difference in the rate of capture between seasons for Red-capped Coua, which captured more prey during the dry season than during the rainy season (Table 2). This species also captured more prey by climbing during the rainy season in comparison with the dry season.

Differences in techniques used were significant between the two seasons (MANOVA, n = 88, F5,82 = 17.34, P < 0.005). There was also a significant variation for the substrates used (MANOVA, n = 37, F4,83 = 16.31; P < 0.005). Ground was more often used during the dry season by this species (Table 3), whereas leaves were more often used as substrates during the rainy season (Table 2). ‘Air’ was more frequently used by Red-capped Coua during the dry season compared with the rainy season.

Table 3.   Analysis of foraging variables used by the two species during each season
 Rainy seasonDry season
  1. Statistical analysis is the same as in Table 2.

  2. *P < 0.05; **P < 0.01; ***P < 0.001; others not significant (P > 0.05).

Foraging variables
 Rate of capture (prey per minute)H = 0.23H = 0.40
 Height of capture (m)t = 12.58***t = 1.22
 % of capture by climbingχ2 = 0.27χ2 = 0.72
Techniques used
 GleanF = 25.16***F = 25.11***
 LeapF = 1.74F = 20.76***
 LungeF = 11.79***F = 3.02
 ProbeF = 14.06***F = 1.13
 SallyF = 14.02***F = 7.55**
Substrates used
 AirF(1,108) = 32.13***F(1,102) = 5.64*
 GroundF(1,108) = 5.34*F(1,102) = 47.35***
 LeafF(1,108) = 0.076F(1,102) = 17.73***
 TrunkF(1,108) = 0.018F(1,102) = 7.12**
 Prey sizeχ2 = 52.59***χ2 = 23.3**

Red-capped Coua captured more prey of size A during the dry season in comparison with the rainy season (Fig. 2).

Interspecific difference between seasons

There was no significant difference between the two species for the rate of capture, whatever the season (Table 3). No difference occurred between Coquerel’s Coua and Red-capped Coua for the proportion of prey captured by climbing (Table 3). Differences in the height reached by the two species appeared during the rainy season: Coquerel’s Coua climbed higher than Red-capped Coua.

There was a significant difference in the use of the techniques during the rainy season (MANOVA, n = 110, F5,104 = 18.34, P = 0.0001) and during the dry season (MANOVA, n = 104, F5,98 = 35.47, P = 0.0001). Differences occurred for ‘glean’, always more frequently used by Coquerel’s Coua in both seasons. Likewise, this species also used `probe’ more often during the rainy season. Red-capped Coua used ‘lunge’ more often during the rainy season and ‘sally’ during the two seasons (Table 3).

The two species differed also in the substrates used during the two seasons (Table 2). This difference was significant during the rainy season (MANOVA, n = 110, F4,105 = 14.32, P = 0.001), with ‘air’ more used by Red-capped Coua and ‘ground’ more used by Coquerel’s Coua. The difference was also significant during the dry season (MANOVA, n = 88, F4,99 = 16.10, P = 0.001). During this period, there was a significant difference between the two species for ‘ground’, being more frequently used by Coquerel’s Coua, and ‘leaves’, more frequently used by red-capped Coua (Table 3).

There was always a significant difference between the two species for the prey size captured, whatever the season: Red-capped Coua always captures bigger prey than Coquerel’s Coua (Table 3 and Fig. 2).

Kleptoparasitism of Red-capped Coua.  We saw Red-capped Coua practising kleptoparasitism, even if this phenomenon was marginal (1 % of eaten prey). It was clear that this bird watched the other species when foraging, and tried to take their food. Red-capped Coua was attracted by at least four species of birds: three cases were with Coua cristata (Cuculidae), one case with Coquerel’s Coua (Cuculidae); five cases with Shetba rufa (Vangidae) and two cases with Bernieria madagascariensis (Bernieridae). This behaviour was not recorded during the rainy season, and was never seen in Coquerel’s Coua. It was also clear that Red-capped Coua was only attracted by the insectivorous birds. Bernieria madagascariensis and S. rufa occurred often in the lowet layers at the beginning of the dry season, and thus, became more visible for Red-capped Coua, especially if the trees were leafless. This behaviour of stealing prey from other birds was recorded among all Red-capped Coua studied during this season.

Diet.  The prey taken by the two species were mainly arthropods. We obtained the following proportions of arthropods among the identified prey: 99.7 % and 98.9 % for Coquerel’s Coua (respectively for the rainy season and the dry season), and 98.3 % and only 91.6 % for Red-capped Coua. Arthropods eaten were mainly caterpillars and orthopteras. Ants were not recorded in the diet, although they were abundant in Ampijoroa. Red-capped Coua captured a snail at least once. Seeds and other vegetal food were also eaten by Red-capped Coua, especially during the dry season. This species ate a lot of seeds of Margaritaria rhomboidalis (Euphorbiaceae) that were very abundant at Ampijoroa, especially at the beginning of the dry season. Seeds involved 35.5 % of the food items taken from the ground during this period. But these results are probably underestimated because of the difficulty of identifying all the seeds eaten on the ground. Because seeds stay on the ground during all the dry season, they can be a very important source of food for this bird. Other vegetal food for Red-capped Coua included resin of a tree Rhopalocarpus sp. (Sphaerosepalaceae) and some flowers of an unidentified Asclepiadaceae, which blossoms during the beginning of the dry season. Coquerel’s Coua was not observed at Ampijoroa eating seeds, but it is known to do so at other places in Madagascar such as Kirindy, in the west of Madagascar. We observed this species eating some lizard eggs and some flowers in Ampijoroa.

Discussion

Differences observed between Coquerel’s coua and Red-Capped coua indicated a difference in their foraging strategy during each season. Both species used different proportions of the techniques and substrates, whatever the season. In addition, seasonal variations modified the foraging strategies used by each species.

Martin & Karr (1990) suggest that techniques used by a bird can be constrained by their morphology, and birds could change other aspects of its foraging, such as substrate or nature of the prey captured.

Our study suggests that there is a correlation between the techniques used and the morphology of the two couas studied here. Although their sizes were similar, Red-capped Coua was slender and has proportionally longer legs than Coquerel’s Coua. This may help the former to jump higher than the latter species, and its more powerful bill could be useful to capture large and fleeing prey, as it has been recorded in some other sympatric species (e.g. Strong, 2000). Rabenold (1978) and Martin & Karr (1990) showed that birds must use different techniques to capture different types of prey. During the dry season and the rainy season, Coquerel’s Coua captured prey more often by gleaning, especially on the ground, whereas Red-capped Coua used leap, lunge or sally more frequently on the other substrates; suggesting that the first species captured prey easily on the surface of the substrate, and that Red-capped Coua captured prey that were generally not accessible to Coquerel’s Coua by using these energetically costly techniques. However, Red-capped Coua cannot climb easily in the vegetation, whereas Coquerel’s Coua is more agile for climbing and therefore efficiently exploiting the prey in the foliage. This prey was usually caterpillars, which live on leaves during all the dry season (Griveaud, 1977). Red-Capped Coua was not able to exploit this prey. This latter species has some difficulties seeing prey in foliage and/or climbing easily and quickly into the foliage. Red-capped Coua captured prey more often while flying, in places where this species can run efficiently to catch them. Morphological differences separate the two species as follows: Coquerel’s Coua is a ‘walker and climber’ and Red-capped is ‘a jumper and runner’.

In addition, Red-capped Coua captured larger prey than those of Coquerel’s Coua, and ate more vegetal matter, especially during the dry season, when animal prey were supposedly rare. The consumption of seeds by this species during the dry season could also explain the greater rate of capture observed during the dry season: Red-capped Coua tended to revisit the same place several times (P.C. pers. obs.). The dry forest may be lower-quality habitat for this species, because it provides a more difficult foraging environment during the dry season. Seeds may be suboptimal food items for predominantly insectivorous species like couas.

Other studies of foraging strategies in birds showed similar patterns. Recher (1987) studied two congeneric insectivorous forest birds, the Brown and the Striated thornbills, similar in appearance and in their habitat in a forest of south-eastern Australia. This forest was seasonal too, with an alternating winter and summer. He found that the two species differed at each season, in the techniques and the substrates used. Seasonal transitions could implicate a difference in food availability, with the dry season corresponding to the expected period of lowest invertebrate availability. Martin & Karr (1990) suggested that the changes in the pattern towards more costly capture techniques reflected the periods of demanding environmental conditions and food limitation. Increased use of energetically costly techniques as probe, during the dry season, especially by Red-capped Coua, may indicate a greater food limitation, and the need to capture some prey, which would have been neglected during the rainy season. Data from Hawkins (1994) indicate that arthropods were more abundant during the rainy season in Ampijoroa. But the rate of capture indicates that more prey was captured during the dry season. According to Hawkins’s (1994), we can explain this paradox by a seasonal variation in the composition of arthropods incorporated into the diet. Spiders and orthopteras were more abundant during the rainy season, but cockroaches, caterpillars, coleopteras and dermapteras were more abundant during the dry season (Hawkins, 1994). All these arthropods are eaten by couas. Ecology of these arthropods can explain the shifts in techniques used, because spiders and cockroaches live under the dead leaves on litter, while orthopteras and caterpillars live on the surface of the litter or the leaves, suggesting the use of different techniques to catch them. However, we have no data about the energy value these prey can provide.

Differences observed in the foraging variables for each coua species, during the seasonal transition, indicate that these transitions play an important role on foraging behaviour, by modifying the strategy used among each species, and probably the nature of the prey captured. Season has to be considered an important factor for food availability in the foraging studies, and it is necessary not to pool all observations, if the aim of the study is to analyse the foraging behaviour. In this way, we considered Urano et al.’s (1994), obtained in the same station, as biased, because these authors did not take into consideration the effects of the seasonal variations.

Difference in behaviour can be easily seen between the two species. Red-capped Coua was more frequently seen in open areas. Usually, Red-capped Coua was not shy, but moved quickly when alarmed by jumping, instead of running. Coquerel’s Coua appeared shier, and tended to run instead. Coquerel’s Coua always appeared reluctant to cross the wide trails, and only did it by running. They never stayed a long time on the trails, even for basking. On the contrary, Red-capped Coua stay a long time on the trails, for basking, singing and feeding.

These differences can have an evolutionary significance, because Red-capped Coua would be more adapted to open environments, and Coquerel’s Coua to a forested environment, as already suggested by Urano et al. (1994). Our own results suggest that the Red-Capped Coua is encountered in open habitats, which in Ampijoroa are also their most usual habitats (Chouteau, Fenosoa & Rakotoarimanana, 2004).

Masuda & Ramanampamonjy (1996), by mapping the territories of the two species, suggested a competitive exclusion between the two species because their territories appeared clearly separated. But our field observations, with marked birds, showed that their territories overlapped (Chouteau, 2003) and that Red-capped Coua, the dominant species, frequently chased Coquerel’s Coua when they came into contact. Recher (1987) observed no interspecific aggression between the Brown and the Striated thornbills, because one species was generalist, and the other one was more specialized. The pattern observed for the couas is different. Both species can be called generalists. We suggest that all the differences between their foraging could play a role in maintaining the coexistence of the two sympatric and congeneric species in a fluctuating environment.

Habitat alteration, by burning or logging, could modify this optimal exploitation, and threaten these species. Comparison of these species, in a primary habitat and a disturbed habitat (by fire or wood exploitation), can contribute to a better understanding of the effects of the vegetation structure on the foraging habits of these birds.

Acknowledgements

We thank the ‘Commission Tripartite’ of the Malagasy Government, the Ministère pour la Production Animale et des Eaux et Forêts for the permission to work in Madagascar. The staff of WWF Madagascar and Steven M. Goodman provided logistic support and very pleasant hospitality in Antananarivo. O. Langrand initiated the study and advised on the methods. J. Gignoux and G. Lacroix helped on statistical analyses. Jean-Marc Thiollay and Steven M. Goodman gave some useful comments on the manuscript. Dorothy Hunt helped to edit the English language. We thank Conservation International for permission to work in Ampijoroa and for accommodation in the station Forestière; Durrell Wildlife Conservation Trust, especially Don Reid, for logistic support in Ampijoroa, Mister Zozime, Doyen de la Faculté des sciences de Mahajanga for his collaboration in its study; U. Thalmann, A. Muller, H. Randriamahazo, T. Mizuta, Charles, Jacky and Rabemazava for their reception in Ampijoroa. Chris Birkinshaw (Missouri Botanical Garden) helped in plant identification.

Ancillary