- Top of page
In recent decades, numerous studies have examined factors affecting risk of host nest parasitism in well-known avian host–parasite systems; however, little attention has been paid to the role of host nest availability. In accordance with other studies, we found that nest visibility, reed density and timing of breeding predicted brood parasitism of Great Reed Warblers Acrocephalus arundinaceus by the Common Cuckoo Cuculus canorus. More interestingly, hosts had a greater chance of escaping brood parasitism if nesting was synchronized. Cuckoo nest searching was governed primarily by nest visibility at high host-nest density. However, even well-concealed nests were likely to be parasitized during periods when just a few hosts were laying eggs, suggesting that Cuckoos adjust their nest-searching strategy in relation to the availability of host nests. Our results demonstrate that host vulnerability to brood parasitism varies temporally and that Cuckoo females are able to optimize their nest-searching strategy. Moreover, our study indicated that Cuckoos always manage to find at least some nests to parasitize. Thus, in this case, the co-evolutionary arms race should take place mainly in the form of parasitic egg rejection rather than via frontline pre-parasitism defence.
Brood parasitism can result in a considerable decrease in the breeding success of host populations (Ortega & Ortega 2003, Barabás et al. 2004, Jewell & Arcese 2008, but see Brooker & Brooker 1996), or at least of individual hosts (Davies 2000). The impact of brood parasitism on host fitness, however, differs significantly between host–parasite systems. The most detrimental effect occurs in hosts of evicting cuckoos (Cuculidae) or honeyguides (Indicatoridae) where brood parasitism has an even greater impact than nest predation, since predation at least allows immediate re-nesting. Successful brood parasitism, on the other hand, precludes re-nesting and confers lower, or even no host breeding success (Payne & Payne 1998, Davies 2000). Thus, hosts have evolved numerous strategies to defeat their parasitic enemies. Two lines of defence can be distinguished: (1) defence preventing brood parasitism; and (2) defence after brood parasitism has occurred, such as nest desertion or egg ejection. The first includes nest guarding, aggressive host behaviour against adult parasites, breeding at sites safe from brood parasites or inconspicuous behaviour near the nest (Robertson & Norman 1977, Patten et al. 2011, Feeney et al. 2012).
Several hypotheses have been put forward to explain nest-site safety in terms of brood parasitism, including distance to the nearest perch site (Alvarez 1993, Hauber & Russo 2000), nest concealment or cover (Burhans 1997, Clarke et al. 2001), nest height above the ground (Clotfelter 1998, Banks & Martin 2001) and distance to habitat edge (Moskát & Honza 2000, Patten et al. 2006) or other active host nests (Spautz 1999). Other hypotheses emphasize host behaviour, such as physical (Sharp & Kus 2006) or acoustic (Gochfeld 1979) activity near the nest and aggression towards a brood parasite. The role of host aggression is equivocal as it may serve both as a nest-searching cue for brood parasites – ‘nesting cue hypothesis’ (Fiorini et al. 2009) – or as a means of driving them off – ‘nest defence hypothesis’ (Robertson & Norman 1977). In addition, the timing of breeding (Patten et al. 2011), nest size (McLaren & Sealy 2003) or height of the nearest perch site (Antonov et al. 2006) have been shown to affect the probability of parasitism. All these factors can be influenced by individual hosts, i.e. hosts can build nests far from trees, in denser vegetation or can be more aggressive in order to expel brood parasites. There are other factors, however, that hosts cannot influence or, at best, influence very little. For example, it is reasonable to expect that the probability of brood parasitism increases with increasing numbers of brood parasites present in the locality, or with decreasing numbers of breeding host pairs. Brood parasites lay a finite number of eggs per season and make use of a limited time-window successfully to parasitize individual host nests (Davies 2000). These constraints should favour host pairs breeding simultaneously, providing them with a better chance of avoiding parasitism through, for example, better nest concealment.
This scenario would represent an important density-dependent mechanism with a substantial role in brood parasitism, influencing the probability of parasitism at the population level much more than nest-site characteristics, for example. To our knowledge, however, only two studies have tested this hypothesis directly. Martínez et al. (1996) found that Eurasian Magpie Pica pica pairs that nested synchronously were parasitized by the Great Spotted Cuckoo Clamator glandarius less than non-synchronous pairs, and Clark and Robertson (1979) observed a similar relationship in Mangrove Warblers Setophaga petechia parasitized by Brown-headed Cowbirds Molothrus ater. It is quite surprising that there has been no evidence provided to date that such a density-dependent mechanism influences the likelihood of parasitism in hosts of the Common Cuckoo Cuculus canorus (hereafter Cuckoo), despite female Cuckoos parasitism being host-specific and the probable impact of the relationship being likely to be more important than in cowbird hosts.
Of importance in this context is that the numbers of brood parasites and hosts fluctuates between years, and hence the proposed density-dependent relationship could be influenced by their phenology. Cuckoos arrive at our study site well before their major host, the Great Reed Warbler Acrocephalus arundinaceus, starts to breed (Jelinek V, Prochazka P, Pozgayova M, Honza M, pers. obs.), enabling them to synchronize egg-laying with the host species (Moskát et al. 2006). Moreover, host breeding starts gradually, with older birds breeding ahead of younger birds, reflecting a protracted arrival that can extend over 3–4 weeks (Jelínek V, Procházka P, Požgayová M, Honza M unpubl. data). This means that Cuckoos have an advantage over their hosts at the beginning of the nesting season. A similar pattern may also occur at the end of the breeding season, when the majority of Great Reed Warbler pairs have either already bred or are feeding nestlings; very few are initiating late replacement or second clutches, whereas Cuckoos are still present in similar numbers as at the beginning of the season. This uneven pattern of breeding (Fig. 1) provides us with the opportunity to assess whether a density-dependent relationship influences the probability of brood parasitism.
Figure 1. Number of nests which started (first egg laid) in 10-day periods over the five study years (first 10-day period begins on 1 May). Grey solid line – parasitized nests, white dashed line – non-parasitized nests, black solid line – all nests together).
Download figure to PowerPoint
In light of the above-mentioned studies we predicted that the probability of parasitism would depend on the number of suitable host nests, with more nests available, resulting in a lower proportion being parasitized. We additionally predicted that a Cuckoo's nest-searching strategy would depend on host nest density. More specifically, predictors of brood parasitism (e.g. nest concealment or perch proximity) would be more important the greater the number of host nests available. To test these hypotheses, we extended a previous study by Moskát and Honza (2000) of a different heavily parasitized Great Reed Warbler population, with two main modifications. First, we used data from five consecutive years, and, secondly, we quantified host aggression to test for nest defence and the nesting cue hypothesis.
- Top of page
This study was conducted between 20 April and 20 July 2008–2012 in two adjacent fishpond areas between Hodonín (48°51′N, 17°07′E) and Mutěnice (48°54′N, 17°02′E) in South Moravia, Czech Republic. The studied Great Reed Warbler population numbered 80–100 pairs and all individuals were marked with a unique combination of a standard aluminium ring and up to three coloured plastic rings. We systematically searched for nests in the littoral vegetation, which was dominated by the Common Reed Phragmites australis, with a smaller proportion of Narrow-leaved Cattails Typha angustifolia (hereafter referred to as reeds for both types of vegetation). We were able to find almost all nests each year thanks to regular mapping of male territories and checking for male mating status (Bensch 1996). Most nests were found during the building stage or at the beginning of egg-laying and were checked daily until clutch completion. Thereafter, the nests were checked less often (typically every 4 days), except for parasitized nests, which were checked daily to determine host response towards parasitic eggs. Each nest was tagged with a small piece of coloured tape and its location recorded with GPS. Eggs were numbered using a felt-tip pen according to laying order.
Nest and nest-site characteristics
Nest and nest-site characteristics were measured for almost all host pairs during the egg-laying period or incubation. Some nests were found late and had large nestlings or had been destroyed by predators before measurements could be taken (from 3% to 9% per year). In these cases, we measured only those nest and nest-site characteristics that remained intact. The following variables were recorded: laying date of the first egg (1 May = day 1), distance to the nearest tree serving as a potential Cuckoo perch (nearest shrub, tree or electric wire more than 5 m high), distance to nearest water's edge for the reed bed, distance to nearest active conspecific nest (measured from GPS positions), reed height above the water's surface, height of upper nest rim above water or ground level, height of vegetation above the nest, reed density (estimated as sparse, intermediate or dense), nest volume and Cuckoo nest view. The last variable was scored following the protocol of Øien et al. (1996), i.e. direct – a Cuckoo sitting in a nearby tree can see the nest; indirect – the Cuckoo cannot see the nest directly but can locate it by activity of nesting birds; or no nest view – the nest is well hidden and the Cuckoo cannot see or locate it by activity of the nesting birds. Nest volume (V) was calculated as half an ellipsoid according to the formula: where a = nest radius and b = nest height (Palomino et al. 1998). Both dimensions were obtained as the mean of two measurements of nest width and height. The underside of the nest usually terminated in loose tags of nesting material. As these contributed to the nest silhouette, nest height was measured from the nest rim to the end of these tags. Nest width was measured using callipers and all other measurements were obtained using a folding ruler. Distances to nearest potential perch site and the water's edge were assessed on the basis of previous training and checked with the assistance of aerial photos (http://www.mapy.cz). All measurements were collected by one person (V.J.), including scoring of Cuckoo nest view and reed density. We used the number of host nests available for Cuckoos (host nest density) as a measure of host breeding synchrony, considering a given host nest as available from the first day of egg-laying to the first day after the last egg was laid. As a typical Great Reed Warbler clutch in the study population was known to consist of five eggs (median = 5, mean ± sd = 4.65 ± 0.73, n = 333; unparasitized clutches from 2008–2012), this interval was set to 6 days. For each day within a season, we computed the number of host nests available for Cuckoos using the egg-laying dates of all nests found each year (i.e. not parasitized and unparasitized nests only as used in the following analyses). Host nest density for each nest was then expressed as the sum of these ‘day nest values’ over the 6-day interval when the nest was suitable for parasitism. This interval was shorter in nests depredated during the egg-laying phase.
In 2009 and 2010, we presented a stuffed Cuckoo dummy at host nests (both parasitized and unparasitized) and recorded aggressive behaviour. The experiment took place around the turn of the egg-laying and incubation stage, on average 3.6 days after clutch initiation (sd = 1.0, n = 78). The dummy was attached to a pole < 1 m from a focal nest and placed level with the nest rim. The behaviour of each member of the host pair was observed for 5 min from its first arrival from a distance of 10–15 m, after it had appeared within a 5-m diameter around the dummy. If there was no reaction and no bird was seen in the vicinity of the nest for 20 min from dummy exposure, the experiment was stopped and the dummy removed. Such nests were excluded from further analysis (seven nests from both years). The experiments were carried out between 08:00 and 20:00 h CET using one of three different Cuckoo dummies presented randomly. All experiments were carried out by M.P.
As a measure of host aggression, we used the sum of contact attacks on the dummy of both pair members, choosing this behavioural characteristic as it showed the highest level of individual variation in comparison with other traits (see also Trnka & Prokop 2012). Moreover, we considered this trait the most risky (in general) and, at the same time, the most effective behaviour for expelling Cuckoo females and preventing them from laying an egg.
Nest parasitism rates
Nests were considered as parasitized when they contained a Cuckoo egg or chick. Nests with a known uninterrupted egg-laying sequence, or those found later but with a complete clutch of four or five eggs, were classified as non-parasitized. All other nests were not considered for further analysis, except when they were used for calculation of host nest density (see above). We also included replacement clutches by the same females into our analyses, as Cuckoo females are able to choose between all active nests within their territories.
Parasitism rates differed markedly among the study years. In 2009–2012, more than 30% of nests were parasitized (2009 – 32%, 2010 – 47%, 2011 – 38%, 2012 – 50%), but only six of 100 nests (6%) were parasitized in 2008. This extremely low parasitism rate prevented us from using these data in the same way as data from the other 4 years; hence they were used only as a supplement for comparison with predicted patterns of brood parasitism. Data from 2009 to 2012 were pooled and analysed together (n = 404: 109, 93, 102, 100 nests in respective years).
All statistical analyses were performed with logistic regression within generalized linear models (GLM) with a binomial error distribution and logit link function in R 2.15 (R Development Core Team 2012). Model simplification was performed through backward stepwise elimination of insignificant terms from the initial model based on change in deviance between the full and reduced models tested using a chi-square test (Faraway 2006, Crawley 2007). Comparisons between categorical predictor levels or their interactions in the minimum adequate model were performed using ‘treatment’ contrasts (Crawley 2007). All continuous predictors were centred (Schielzeth 2010) due to the inclusion of interaction terms. The initial model included nest parasitism (0 = nest unparasitized, 1 = nest parasitized) as a response variable, and host nest density, Cuckoo nest view, reed density, distance to nearest tree, distance to nearest water's edge, distance to nearest active conspecific nest, height of nest above water level, height of reed above the nest, nest volume and laying date as predictors. To test the hypothesis that Cuckoos change their nest-searching strategy in response to host nest availability, we included interactions between host nest density and all predictors, apart from laying date, into the model. The height of the reed was excluded from the initial model due to a strong correlation with the height of reed above the nest (rS = 0.78).
The impact of host aggression on brood parasitism was tested for 71 nests over 2 years (2009 and 2010), 18 of which nests belonged to the same males. For this reason, we used generalized estimating equations (GEE; R package geepack; Yan 2002, Yan & Fine 2004, Hojsgaard et al. 2006) with a binomial error structure and independent correlation structure. Model selection was performed in the same manner as for GLM. The initial model included nest parasitism as a response variable, host nest density, number of attacks, reed density, nest status (monogamous, primary, secondary), laying date and three interactions as predictors, and male identity as a grouping variable. The first interaction was between reed density and number of attacks, as hosts may adjust aggressive behaviour based on the level of nest concealment. As Požgayová et al. (2013) found that polygynous males defend their nests less intensively compared with monogamous males, we also included interaction between nest status and number of attacks. Finally, to test whether the Cuckoo's nest-searching strategy depends on availability of host nests, we included the interaction between host nest density and number of attacks.
- Top of page
Seasonal host and parasite egg-laying patterns differed over the 5 years (Fig. 1). In 2008 and 2010, the number of nesting Great Reed Warbler pairs increased gradually, with a clear egg-laying peak in the third 10-day period of May followed by a similarly gradual decrease. In 2009, 2011 and 2012, the numbers of breeding pairs reached their maximum faster over at least two 10-day periods in May, followed by a rapid (2009), slow (2011) or moderate (2012) decrease.
The probability of brood parasitism increased with higher nest visibility and lower reed density, and decreased over the course of the breeding season (Table 1). Moreover, the interaction of Cuckoo nest view and host nest density was also found to be significant (Table 1). More specifically, the best-concealed nests (with no nest view) were likely to be parasitized at lower host densities, whereas they were quite safe at higher host densities. No such relationship was found for nests with indirect or direct Cuckoo nest view (Fig. 2).
Table 1. The effect of nest and nest-site characteristics on the probability of brood parasitism in Great Reed Warbler nests in 2009–2012. Only the minimum adequate model from the logistic regression model analysis is presented. P-values of particular model terms are based on Type III sum of squares. P-values of differences between the levels of categorical predictors are in parentheses. Values for host nest density are not given because the presence of this term in the significant interaction prevents the reliable interpretation of this term.
|Variable||χ2||df|| P ||Estimate ± se|
|Intercept|| || || ||–1.595 ± 0.626|
|Host nest density||–||–||–||—|
|Cuckoo nest viewa*||24.04||2||< 0.001||1.673 ± 0.597 (indirect: P = 0.005)|
|2.488 ± 0.638 (direct: P < 0.0001)|
|Reed densityb*||6.67||2||0.036||–0.435 ± 0.274 (middle: P = 0.112)|
|–0.884 ± 0.345 (dense: P = 0.011)|
|Laying date||7.70||1||0.006||–0.026 ± 0.009|
|Cuckoo nest viewa* × Host nest density||7.84||2||0.019||0.169 ± 0.069 (indirect: P = 0.014)|
|0.157 ± 0.073 (direct: P = 0.031)|
Figure 2. Relationship between the probability of brood parasitism and host nest density in Great Reed Warbler nests with different Cuckoo nest view in 2009–2012 (grey solid line and grey-filled circles – direct nest view, grey dashed line and grey open circles – indirect nest view, black solid line and black filled circles – no nest view). Dotted lines are 95% CIs. Predicted probabilities are based on simple logistic regression models. Points below zero represent unparasitized nests, points above one, parasitized nests; the symbols are depicted in separate lines to distinguish nests with different Cuckoo nest view.
Download figure to PowerPoint
During the nest defence experiments, 59 pairs attacked the Cuckoo dummy between two and 180 times, whereas 12 arriving pairs did not directly attack the dummy (mean number of attacks ± sd = 38 ± 38.2). Despite this variation, neither host aggression towards the Cuckoo nor any other variable except host nest density (β ± se = −0.069 ± 0.03, = 4.52, P = 0.033) had a significant effect on the probability of brood parasitism.
- Top of page
Here, we demonstrate for the first time that the number of simultaneously breeding host pairs significantly influences the probability of brood parasitism by a Cuckoo, and that the number of available host nests may modulate the Cuckoo's nest-searching strategy. Host pairs nesting during periods of high nest density were less parasitized than those nesting at lower densities. These results are similar to those of Martínez et al. (1996) and Clark and Robertson (1979) for other host–parasite systems.
In accordance with many other studies (e.g. Øien et al. 1996, Moskát & Honza 2000, Clarke et al. 2001, Avilés et al. 2009), we found that the best-concealed host nests (built in dense reed or those that were poorly visible with respect to Cuckoos) were at the lowest risk of brood parasitism. Accordingly, the optimal strategy to escape brood parasitism should be to build nests in such places and, in general, we confirmed this assumption. We also demonstrated that such a tactic proved useless when only a few host pairs were breeding simultaneously. More specifically, well-concealed nests were also parasitized at low host nest densities (Fig. 2) as Cuckoos could devote more time to searching for such nests, while ignoring them during periods of nest surplus. It appears therefore that Cuckoos adopt a strategy of finding as many host nests as possible. There are two possibilities why they should do so: first, finding enough host nests is so time-consuming that the Cuckoo must concentrate on locating only poorly hidden nests, and second, parasitic females pursue a strategy of locating as many host nests as possible to be able to select the most suitable nests, e.g. on the basis of egg colour matching (M. Honza et al. unpubl. data) or other characteristics (reviewed by Parejo & Avilés 2007). This latter possibility is in accordance with the finding of Nakamura et al. (2005), who showed that one Cuckoo female visited at least 16 host nests in her territory but parasitized only nine of them.
Consequently, the longer the hosts breed simultaneously in ample numbers, the higher the number of pairs that have a chance of escaping parasitism by building their nests in less visible places, e.g. in denser vegetation and far from trees that offer the Cuckoo perches. In doing so, they gain an advantage over other pairs and increase their reproductive success despite the high overall parasitism rate in the population. Briskie et al. (1990) suggested a similar relationship at the interspecific level for American Yellow Warblers Setophaga aestiva, which were preferentially parasitized by Brown-headed Cowbirds over Least Flycatchers Empidonax minimus, whose nests were only used when nests of the primary host species became rarer.
In addition, our results showed a temporal pattern in parasitism pressure, with earlier nests parasitized more often than later nests. For example, when only six of 100 nests were parasitized in 2008, three of those were from the six earliest nests. Similarly, Welbergen and Davies (2009) found that the earliest nests of Eurasian Reed Warblers Acrocephalus scirpaceus suffered higher Cuckoo parasitism than later nests, a pattern also suggested by the data of Øien et al. (1998) for the same species. Nevertheless, it is hard to say what is behind this date-related effect. As stated above, in part it could be that Cuckoos have more time for nest searching at the beginning of the season; alternatively, as Eurasian Reed Warblers begin breeding on average 2 weeks later than the Great Reed Warbler, some Cuckoo females may switch host species later in the season. A further possibility is that Cuckoos preferentially parasitize earlier nests, as offspring in such nests have greater reproductive value than offspring that hatch later (e.g. Verhulst & Nilsson 2008, McKim-Louder et al. 2013).
Although distance to nearest perch site is generally regarded as an important predictor of brood parasitism in both cuckoo (Alvarez 1993, Moskát & Honza 2000, Antonov et al. 2007, Welbergen & Davies 2009) and cowbird hosts (e.g. Clotfelter 1998, Hauber & Russo 2000, Patten et al. 2006), we did not confirm such a relationship. In agreement with Moskát and Honza (2000) and Antonov et al. (2007), we also found no effect of habitat edge on brood parasitism, despite such a relationship being substantially supported in the cowbird–host system (Patten et al. 2011). Finally, similarly to Moskát & Honza (2000) and Avilés et al. (2009), we were also able to show that nest size did not predict probability of brood parasitism (but see McLaren & Sealy 2003).
It is surprising that the influence of host aggression on probability of brood parasitism has been tested relatively rarely compared with the effect of nest-site characteristics, despite aggressiveness of host species generally increasing with vulnerability to brood parasitism (Moksnes et al. 1991, Røskaft et al. 2002). Furthermore, populations of some host species living in sympatry with brood parasites have been shown to display higher aggression to dummies than populations living in allopatry (Burhans et al. 2001, Røskaft et al. 2002). Despite results indicating that host aggressiveness towards brood parasites is associated with nest parasitism, we know almost nothing about its effectiveness. We found no relationship between host aggression towards the Cuckoo dummy and probability of brood parasitism. It would appear therefore that although our host species is very aggressive and vigorously attacks enemies near its nest (Molnár 1944, Janisch 1948–51, Bártol et al. 2002, Požgayová et al. 2009), it is unable to prevent Cuckoo females from laying eggs in the nest. Nakamura et al. (2005) reported that a female Cuckoo laid an egg into the nest of an Oriental Reed Warbler Acrocephalus orientalis at the ninth attempt, despite always being attacked by up to four Warblers. Similarly, host aggressiveness does not prevent brood parasitism in several cowbird host species (Gill et al. 1997, Olendorf & Robinson 2000). However, Welbergen and Davies (2009) have shown a strong correlation between local parasitism risk and mobbing propensity in the Eurasian Reed Warbler, with mobbers in areas of high parasitism risk suffering a more than 20% lower parasitism rate. This is surprising as mobbing in this small-bodied host is based mainly on alarm calls (Welbergen & Davies 2008, Čapek et al. 2010) and heavier and much more physically aggressive Great Reed Warblers were unable effectively to protect their clutches. We believe that Eurasian Reed Warblers may be able to avoid Cuckoo parasitism using another method. Previous studies have shown that some Cuckoo hosts, including the Eurasian Reed Warbler, tend to reject model eggs if a Cuckoo dummy is presented before experimental parasitism (Davies & Brooke 1988, Moksnes & Røskaft 1989) and it may be this that helps explain the Cuckoo's unwillingness to parasitize the better guarded nests of Eurasian Reed Warblers.