Effect of nest age and habitat variables on nest survival in Marsh Harrier (Circus aeruginosus) in a fishpond habitat

Study area

The study was conducted over five breeding seasons (2008, 2009, 2011, 2018 and 2019) in eastern Poland on four fishpond complexes: Siedlce, Rudka, Szostek, Mościbrody (52°05′–52°11′N, 21°58′–22°18′E); all are mainly used for the commercial breeding of Common Carp Cyprinus carpio. Pond areas varied from 65 to 203 ha and were surrounded by a mixture of arable fields (50–60%) where mostly cereals were cultivated, meadows (25–30%) and forested areas (5–25%). All four fishpond complexes were located within 20 km of each other. Most of the ponds were partially covered by tall marsh vegetation consisting of Bulrush (Common Reedmace) Typha latifolia, Common Reed Phragmites australis and Sedges Carex spp. These plants tend to proliferate rapidly, quickly covering the surface of a pond or wetland, thus creating a suitable breeding habitat for Marsh Harrier. The ponds were similar in water depth but water levels in the emergent vegetation varied from 7 to 92 cm in spring, falling as the breeding season progressed.

During the fieldwork, the presence of several possible opportunistic predators of aquatic birds’ nests were recorded: the invasive American Mink Neovison vison and Raccoon dog Nyctereutes procyonoides, and the native Wild Boar Sus scrofa, Red Fox Vulpes vulpes, European Otter Lutra lutra, European Badger Meles meles, Magpie Pica pica, Raven Corvus corax and White-tailed Eagle Haliaeetus albicilla. With its rapid colonisation of Poland in recent years, American mink is currently considered the biggest threat to the native waterbird fauna (Brzeziński et al., 2019a, 2019b).

Field procedures

At the beginning of each breeding season, each study pond was visited at 1–3-day intervals between mid-April and mid-May to locate breeding pairs and nests. The observations were made with 8 × 42 binoculars from the fishpond dyke. The birds were observed carrying nest material to the emergent vegetation belt and during aerial food-passes near their potential nest site. After selecting a potentially favourable site, the observers inspected the vegetation belt on foot along fixed line transects. When located, the nests were numbered and their positions recorded with a hand-held GPS unit. A total of 82 nests were discovered in the study areas—27 during the egg-laying phase, 53 during egg incubation and 2 during the early nestling period. To minimise disturbance, each nest was visited at 5–7 day intervals to determine clutch size, hatching date, nest fate and the number of live chicks. The first-egg laying date was calculated on the assumption that eggs are laid at 2-day intervals, and that incubation starts after the laying of the first egg and lasts for an average of 33 days (Witkowski, 1989). A nest was considered to have been depredated when clear signs (egg shells or nestling feathers) were found or the nest was found empty before the predicted date of fledging. A successful nest was defined as one where at least one young bird survived up to 35 days old (Witkowski, 1989). To be sure of this success, additional inspections of such a nest were made 5 days before and 3 days after this period. Six habitat variables were obtained for each active nest along with the fate of the nearest neighbour nest (Table 1). All the vegetation measurements (height, diameter and density) were made in 100 × 100 cm quadrats placed around the nest during the first visit. The distances of a nest to the fishpond dyke and to open water was measured using GPS equipment. All operations were conducted as part of the ‘Research of birds in the disclimax ecosystems’ programme, approved by the Institute of Biological Sciences, Siedlce University of Natural Sciences and Humanities (number of approval: IB.5030.8.2018). The study took place in compliance with current Polish Law and was approved by Ministry of the Environment (permit number: 425/2019) and also by the Regional Directorate for Environmental Protection in Warsaw (permit number: WSTS.6401.34.2018.MO).

Statistical analyses

Daily survival rate, the probability that a nest will survive a single day, was calculated using known-fate models with the RMark package (Laake, 2019). RMark is an R package (R Core Team, 2019) that provides a formula-based interface for the MARK programme (White & Burnham, 1999). The analysis covered only nests which succeeded or were depredated (N = 82). The dates were scaled such that day 1 was the day when the first nest was found and day 84 was the day the last nest was checked. Hence, the 84-day nesting season was defined as beginning on 30th April and ending on 22nd July. The season thus consisted of 83 intervals, which represent an 83-day nesting cycle with each interval equivalent to 1 day. We therefore modelled DSR as a function of temporal (nest age), the fate of the nearest neighbour nest and habitat variables (water depth under the nest, density of reed stems, height and diameter of vegetation, distance to open water and distance to the dyke). The multi-co-linearity of habitat variables was checked—the correlations between them were <0.4. We constructed models of nest survival that incorporated combinations of individual covariates, and compared them to the null model of constant survival rate S(.). The set of competing models was based on a combination of factors assumed a priori to affect DSR.

The better concealed nests were expected to have a greater chance of survival. We used an information-theoretic approach (AIC) to compare the competing models (Burnham & Anderson, 2002) and analysed model support using the Akaike’s information criterion with small-sample bias adjustment (AICc) value, which corrects for small sample sizes and evaluates the strength of evidence for each model using normalised weights (w_i). We applied selection approach suggested by Richards (2008), in which more complex variants of any model with a lower AICc value are excluded from the candidate set. The models selected with the smallest AICc as being the best of all the models compared, where the models were within a ∆i AIC of 2.00, were considered to be equally supported (Burnham & Anderson, 2002).

Nest age

Our results showed that Marsh Harrier DSR was not constant from the egg phase to fledging, decreasing gradually with nest advancement. It is likely that multiple factors influence age-specific patterns of nest survival. One potential explanation of this pattern is that after hatching, parents and young provide more behavioural cues at the nest, which increase the possibility of its being detected by predators. Increased parental visits to the nest may raise the risk of nest predation (Martin, Scott & Menge, 2000). Indeed, in the late nesting period, Marsh Harrier parents make more food deliveries daily (Kitowski, 2006). However, as nest age increases, adults invest more in the nest and typically intensify defensive behaviour (Caro, 2005); this pattern has also been confirmed in other birds of prey (Carrillo & Aparicio, 2001; Sergio & Bogliani, 2001). Despite the fact that Marsh Harrier actively defend its nests (Kitowski, 2006), this does not compensate for increased predation. One possible explanation is that with nest advancement, parents need to make more frequent foraging flights to provide for their offspring and spend less their time defending the nest. The times when the parents are absent are when the nest is more vulnerable to predation. Decreasing DSR may also reflect a cumulative risk: the longer a nest is active, the more likely it will lose eggs to predation. In Marsh Harrier, the period from the start of incubation to the fledging of the young birds is relatively long in comparison with other species nesting in the same fishpond habitat. For example chicks of Eurasian Bittern (Botaurus stellaris) leave the nest at the age of just 2 weeks post hatching, that is before reaching full independence (Kasprzykowski & Polak, 2012). The female continues to care for the young, which hide in vegetation near the nest until fully fledged. This could be an adaptive strategy diluting the risk of detection by predators and preventing DSR from decreasing with nest age in this species. In contrast, Marsh Harrier chicks cannot leave the nest until they are capable of flight: this species is therefore especially vulnerable to detection by predators with increasing nest advancement.

Nest concealment

In our study, the diameter of reed stems but not the density or height of vegetation influenced Marsh Harrier nest DSR. Nests built over deeper water, where reed stems were thicker, were less vulnerable to predation. Nests located in sites with thicker vegetation may be better protected from predators because the habitat structure at these sites may improve visual concealment of nest thus reducing probability of nest detection. Vegetation thickness can also improve protection from wind thus reducing olfactory cues to potential mammal predators (Guyn & Clark, 1997). The height of the vegetation was not a variable significantly improving the chances of brood success in Marsh Harrier; this factor has been proven significant, though mostly for populations of birds depredated by avian predators (Jedlikowski, Brzeziński & Chibowski, 2015). The distances from nests to open water or the fishpond dyke were not significant. This finding is consistent with previous studies on this species (Stanevičius, 2004). It is well-known that ground-nesting birds, including Marsh Harrier, are particularly vulnerable to predation. Thus, obligate ground-nesters have evolved a method of placing their nests in well-concealed, evenly-spaced sites to reduce the likelihood of detection (Redmond, Keppie & Herzog, 1982). We expected the density of vegetation to have an impact on nest DSR of Marsh Harrier, as was found to be the case in other marsh nesting birds (Kristiansen, 1998; Polak, 2016). The reason for the lack of such a relationship is that parental behaviour (nest defence) may compensate for any effects of insufficient nest cover (Lima & Dill, 1990). Marsh Harrier is considered a top avian predator of wetland habitats and actively defends its own nests with alarm calls and physical attacks (Witkowski, 1989).

Water depth

A strong positive relationship between water depth and nest survival has been observed in marsh-nesting birds, for example in Eurasian Bittern (Polak, 2016) and Common Pochard Aythya ferina (Albrecht et al., 2006). In the present study, Marsh Harrier nests located at sites with deeper water exhibited the same trait. Such nests were particularly successful, because water presents a barrier to many mammalian predators (Koons & Rotella, 2003). It is worth noting that the water level is not constant throughout the breeding season. Previous studies have shown that predation rates for wetland nesters decreased with increasing water depth (Purger & Mészáros, 2006). On the other hand, water depth was not important for the DSR of either Little Crake Porzana parva or Water Rail Rallus aquaticus, as their main predators are mostly avian (Jedlikowski, Brzeziński & Chibowski, 2015). This pattern suggests that Marsh Harrier builds its nests over deeper water because they are more susceptible to predation by mammals than by birds. The relationship between a low water level and mammalian predation was demonstrated in the Netherlands, where Red Fox was a frequent predator of Marsh Harrier nests in dry reedbeds (Dijkstra & Zijlstra, 1997). In his study of a breeding population of Marsh Harrier in the Barycz Valley, Witkowski (1989) also noted that dried out ponds were avoided, although the same ponds containing water could have one of the highest densities of nesting pairs. This may suggest that locating nests at deep water sites is an important antipredator strategy in this species. During studies in eastern Poland 30 years ago, Buczek & Keller (1994) highlighted corvids and mustelids as being the main predators of Marsh Harrier nests on retention reservoirs (mostly resembling neglected fishponds) and bogs, respectively. This was further explained by the low water level in bogs, decreasing during the course of the season, which enabled mammalian predators to penetrate reedbeds. Since that study, predator and prey interactions could well have changed significantly, following the spread of non-native invasive predators such as American Mink. This has been confirmed by recent research, which links the declines of several waterbirds and semi-aquatic mammals with the colonisation of Poland by American Mink (Brzeziński et al., 2019b). Even though water is not a barrier for American Mink, the occurrence of this invasive species may also be having an impact on Marsh Harrier, making it even more vulnerable to mammalian predation. But to explain this possible shift, further studies will be needed to evaluate the causes of nest loss in Marsh Harrier in greater detail.

Neighbour fate

The influence of nearest neighbour fate on nest success has been addressed in observational and experimental studies, but their results are inconsistent (Ackerman, Blackmer & Eadie, 2004; Ringelman, Eadie & Ackerman, 2012). In our study, the chances of nest survival were lower if a neighbouring nest was depredated. A similar pattern is observed in other marsh nesting birds like Common Pochard Aythia ferina (Folliot et al., 2017). In other wetland bird species, nearest-neighbour nests tend to share the same fate but only when they form clusters, which survive or are depredated as a group (Ringelman, Eadie & Ackerman, 2012; Kasprzykowski & Polak, 2014). The reason why neighbours’ fates influence a nest’s fate is that predators use area-restricted searching and employ search images to find similar nests nearby. However, other studies on waterfowl have not detected any influence of nearest neighbour fate on the success of artificial or natural nests (Ackerman, Blackmer & Eadie, 2004). The reason for this difference could be the predator community and the spatial scale at which potential predator species operate that is large vs. small home ranges (Schmidt & Whelan, 1999).