Predators and nest success of Sky Larks Alauda arvensis in large arable fields in the Czech Republic

Abstract

Capsule Nest failure owing to a range of predators was high, but the level and specificity of nest depredation cannot be generalised.

Aims To determine fates and predators of Sky Lark nests in conventionally managed arable fields in the Czech Republic.

Methods Sky Lark nests in large fields (mainly Maize, Sugar Beet and Opium Poppy) were monitored by means of continuous video surveillance.

Results Primary nest fates of 42 active nests were fledging (13), depredation (22), desertion (5), nestling death (1), and flooding (1). The overall nest success (Mayfield estimate) was 17% (all mortality factors considered) or 27% (only depredation). Depredation events were caused by Marsh Harrier Circus aeruginosus (11), Hooded Crow Corvus cornix (4), Stone Marten Martes foina (3), Montagu’s Harrier Circus pygargus (2), Red Fox Vulpes vulpes (2), Hedgehog Erinaceus sp. (2) and Eurasian Jackdaw Corvus monedula (1). Successful nests were only slightly more away from field edge than depredated nests; nests taken by birds tended to be closer to field edge than those depredated by mammals. The possible reasons for the absence of a clear edge effect include comparatively large field parcels (about 50 ha) and location of nests far from field edge (median = 195 m).

Conclusion Nest survival and composition of nest predators are site‐specific and contingent upon the study method and may not be simply generalised.

INTRODUCTION

Sky Larks Alauda arvensis are one of the most abundant and widespread of ground‐nesting farmland birds (Donald 2004). No other passerine in Europe shows such a strong association with arable land (Reif et al. 2008). Sky Lark populations have declined over past decades across much of western Europe and these declines have coincided temporally and spatially with a rapid intensification in farming (Erhard & Wink 1987, Chamberlain & Crick 1999). Therefore, Sky Larks are commonly included among top indicators of loss of biodiversity in agriculture landscapes (Gregory et al. 2005). Nest depredation, the principal cause of low nesting success in most passerines, usually accounts for over 70% of all failures of Sky Lark nests (Weibel 1999, Donald et al. 2002). Yet, the question of whether nest depredation may have increased as a result of farmland habitat changes, has rarely been addressed (Evans 2004) and no studies have reported depredation as a major driver of population decline for Sky Larks (Donald 2004). Nevertheless, nest success has been shown to vary among crop types (Weibel 1999, Donald et al. 2002, Eraud & Boutin 2002, Kragten et al. 2008) as well as within individual fields with a proximity to tramlines (Donald et al. 2002) or field edges (Weibel 1999, Morris & Gilroy 2008).

Although previous studies often implied predator‐specific effects on nest survival, there is little definitive evidence of nest predator identity in farmland birds (MacDonald & Bolton 2008, Teunissen et al. 2008). We are aware of only one video surveillance study of any ground‐nesting passerine in crop fields (Morris & Gilroy 2008). Because predator–prey systems are locally specific, conclusions from one area may not be directly applicable to other regions (Thompson 2007). Agricultural landscapes have developed differently in the western and eastern parts of Europe over the past 50 years. Temporary decline in the intensity of agriculture around 1990 is the likely reason why populations of farmland birds, including Sky Larks, have declined at a slower rate in central and eastern Europe compared with western Europe (Donald et al. 2001; see Reif et al. [2008] for details on the Czech Republic). Contrary to the well‐documented population trends, data on breeding performance and nest predators in the former socialist eastern and central European countries are lacking (Shurulinkov 2005, Erdös et al. 2009).

Here we report on breeding success and nest predators of Sky Larks in intensively managed arable fields in the Czech Republic. The results should help to asses whether predictors of Sky Lark nest success identified in western European farmlands can be generalised to other regions.

METHODS

Study area and field methods

The study was conducted in the northeast part of Chrudim District, the Czech Republic (49° 54′ N, 15° 59′ E, altitude 270–460 m) in 2009. The study area of 55 km² was characterised by predominance of conventionally managed arable land (>80% of total area, mean parcel size about 50 ha), interspersed with remnants of deciduous woodland (<5% of total area) and villages. Main crop types were: Winter Wheat Triticum aestivum (50%); Maize Zea mays (25%); Sugar Beet Beta vulgaris, Lucerne Medicago sativa and Spring Barley Hordeum vulgare (each 5%); Opium Poppy Papaver somniferum, Oilseed Winter Rape Brassica napus subsp. napus, Caraway Carum carvi, Pea Pisum sativum and Sunflower Helianthus annuus (each ≤2%).

Fieldwork was carried out from early April to late July; roughly 100 days were devoted to nest searching in the whole study area. This period covers the breeding season of Sky Larks in the Czech Republic (from end of March to end of July [Hudec 1983]). Sky Larks avoid fields with dense coverage and tall vegetation (Toepfer & Stubbe 2001); therefore, we located nests by systematically searching only the ground of sparse and low‐vegetated fields (≤60% coverage and ≤60 cm vegetation height). Searches for nests were conducted mainly in Winter Wheat (during April), Maize, Opium Poppy and Sugar Beet (all from May to July). We did not keep systematic observations of adult birds, but we opportunistically searched in places where birds were flushed. Age of nestlings was estimated from their development; stage of incubation was estimated by egg floatation. First‐egg laying date was back‐calculated from brood size and hatch date or clutch size and commencement of incubation. Position of each discovered nest was determined with a global positioning system; nearest distance to field edge, woodland and settlement were measured from orthophotomaps. Height of vegetation at the nest site (m) and the degree of nest cover (scored as: 1, well visible from above; 2, intermediate; 3, completely covered from above) were measured and the nest was photographed at each nest visit.

During field work we kept notes on occurrence of potential nest predators, including direct observation (corvids, raptors), occupation of potential breeding sites (completely known only for Marsh Harrier Circus aeruginosus) and records of trails and feaces (carnivorous mammals).

Video monitoring

Video monitoring systems consisted of a video camera (40 × 35 mm) with nine infrared‐emitting diodes, a portable security digital video recorder (DVR) (Yoko RYK‐9107), and a 12V/65Ah deep cycle battery. The DVR was housed in a weatherproof plastic box (125 × 95 × 50 mm) and connected to the camera by a 5‐m cable and to the battery by a 1‐m cable. All outer parts of the system were camouflaged by brown‐green spotted painting. The camera was mounted on a piece of wire that was inserted in the ground and allowed the adjustment of camera position. Cameras were placed 0.7–2.0 m from nests (depending on nest visibility) and 2–20 cm above the ground, never exceeding the height of surrounding vegetation. We used local natural material (dry vegetation, stones) to mask the camera; all other parts (box, battery, cables) were buried under ground. We set the DVR to record continually with a frequency of 10 frames s⁻¹ at 640 × 480 pixel resolution and a medium quality. These settings allowed for 4.5 days of recording on a 16 GB memory card. Cameras were deployed either immediately at the time of nest discovery (22 nests), on the same day after a period of 2–5 hours (11 nests) or within the next 4 days after discovery (10 nests). Deployment of the video system by two people took about 20 minutes. We avoided disturbing vegetation or leaving dead‐end tracks (to/from the nest). Also, we postponed deployment of the camera (or subsequent nest visits) in the case of rain or when the nest was exposed to another human disturbance (agricultural operations). We visited the nests usually (70% cases) every fourth day to check nest content and to change the battery and memory card; the rest of the nest visits were at 1‐ to 3‐day intervals. To keep the time spent at nests short (≤15 min), we offloaded all data in the lab. About half of our visits took place in the morning (08:30–10:30 Central European Time) and half in the afternoon (14:30–17:30). When a nest was found to be empty, we recorded and photographed its state and searched in the immediate vicinity for signs of nest failure (eggshells, feathers, dead nestlings) or success (alive fledglings, droppings, juvenile feathers). We continued video monitoring of deserted or partially depredated nests with intact egg(s) for at least two 4‐day periods, depending on availability of active nests at that time.

Data analysis

When the nest checked was empty or when some eggs/nestlings were missing we viewed the recordings backward to find the event and to determine the nest fate, timing of the event and the species of nest predator. Although the exact survival times and nest fates were available, the limited sample size precluded sensible application of the methods suitable for continuous survival data (e.g. Weidinger 2008). Instead, we measured the exposure period in nest‐days where each nest‐day was treated as an independent binary observation (survived or failed). We estimated the daily survival rate (DSR) as a simple ratio of survived to exposed nest‐days. With the present type of data this method is equivalent to the conventional logistic (Aebischer 1999) or logistic‐exposure (Shaffer 2004) models without covariates. Contrary to these methods applied to ordinary nest‐visitation data (i.e. interval survival data) the survival time in this study was known and did not need to be estimated from the length of nest visit intervals. In this way we estimated DSR separately for the egg (laying and incubation) and nestling stage of active nests, and for deserted (= inactive) nests with eggs. For the active nests we considered either nest losses owing to predation, or combined mortality.

Given the limited sample size we performed only simple univariate analyses; we focused on indicating possible patterns in the data rather than on formal statistical testing. We examined an edge effect on nest survival by means of logistic regression (sensu Aebischer 1999; with scale adjustment for over‐dispersion) with distance to field edge entered as a single continuous nest‐level covariate. We checked for potential observer effects on nest survival by analysing the relationship between timing of observer visits and timing of depredation events. We compared DSR among the four days that followed after a nest visits by means of logistic regression with day entered as either a categorical (day 1 vs. subsequent days) or continuous time‐dependent covariate. Lowered DSR on the day just after nest visit would indicate attraction of predators by observer presence at the nest, while DSR gradually increasing from the first to the forth day after observer visit would indicate a vanishing effect of tracks left by observer at the nest (for detailed reasoning see Weidinger [2008]).

RESULTS

In all, 44 nests (42 active nests and two abandoned nests with fresh intact eggs) were found in 16 different fields representing seven crop types: Maize (19 nests, five fields), Sugar Beet (12, 4), Opium Poppy (7, 2), Lucerne (2, 2), Caraway (2, 1), Spring Barley (1) and Pea (1). Median age of nests at discovery was 4 days (day 1 = first‐egg laying date); the nests were found before commencement of egg laying (5), during laying (8), incubation (28) or nestling stage (1). Nests found by accidental flushing of incubating birds (n = 15) were similar in vegetation characteristics (difference ± se; vegetation cover score: 0.01 ± 0.26; vegetation height: −0.07 ± 0.12 cm) to nests found directly by visual search (n = 27). The electronic supplementary data file details all nest depredation events.

The median laying date was 3 June (24 April–5 July, n = 42), size of completed clutches was 3.86 ± 0.12 se (2–5, n = 35), number of hatchlings was 3.53 ± 0.19 se (2–4, n = 15) and number of fledglings was 3.31 ± 0.33 se (1–4, n = 13). Primary nest fates of the 42 active nests were: fledging (13); complete depredation (17); partial depredation (5); desertion (5); nestling death (1); and flooding (1). Almost all (26/29) nest losses occurred during the egg stage. Overall nest success was 17% (all mortality factors considered) or 27% (only predation considered; Table 1).

Table 1. Survival rates of active (primary depredation, all mortality) and abandoned (secondary depredation) Sky Lark nests in crop fields in the Czech Republic.

Mortality factor	Total nests	Failed nests	Exposure nest‐days	DSR^a	95% CI	Nest survival^b	95% CI
Nesting stage
Primary predation
Eggs	42	21	245	0.914	0.872	0.946	0.29	0.15	0.46
Nestling	16	1	120	0.992	0.954	1.000	0.94	0.69	1.00
Total	43	22	365	0.940	0.910	0.962	0.25	0.13	0.42
Total^c	–	–	–	–	–	–	0.27	0.10	0.46
Secondary predation
Eggs (abandoned)	13	7	118	0.941	0.882	0.976	–	–	–
All mortality factors
Eggs	42	26	245	0.894	0.848	0.930	0.21	0.10	0.36
Nestling	16	3	120	0.975	0.929	0.995	0.82	0.55	0.96
Total	43	29	365	0.921	0.888	0.946	0.16	0.07	0.30
Total^c	–	–	–	–	–	–	0.17	0.06	0.34
^aDaily survival rate; ^bnest survival = DSR^t, where t = 14 (eggs), 8 (nestlings) or 22 (total) days; ^ccalculated as the product of nest survival for the egg and nestling stage, respectively.

Video monitoring of 43 nests (342 active nest‐days and 106 abandoned nest‐days) documented 38 primary nest fates and seven secondary depredations on nests previously deserted or partially depredated, yielding a total of 25 documented depredation events at 23 nests (see supplementary data which is available on the supplementary content tab of the article’s Informaworld page at http://dx.doi.org/10.1080/00063657.2010.506208). In all, seven species (four birds, three mammals) were recorded depredating Sky Lark nests: Marsh Harrier (11 events, 11 nests), Hooded Crow Corvus cornix (4, 3), Stone Marten Martes foina (3, 3), Montagu’s Harrier Circus pygargus (2, 2), Hedgehog Erinaceus sp. (2, 2), Red Fox Vulpes vulpes (2, 2) and Eurasian Jackdaw Corvus monedula (1, 1). Appearance of the depredated nests did not provide any useful cues to determine the species or class of the predator. Predation by birds and mammals was restricted to day (05:34–20:34) and night (21:24–03:43), respectively. The median distance from Sky Lark nests depredated by Marsh Harrier to the nearest occupied harrier nest was 3.5 km (0.3–4.5 km, n = 11).

Daily nest survival rate (considering predation losses) did not change appreciably with distance to field edge: DSR (logit scale) = 2.3647 (± 0.6710 se) + 0.0024 (± 0.0039 se) × distance (m). Correspondingly, successful nests (n = 13) were, on average, only slightly further away from field edge (difference ± se: 31 ± 25 m) than depredated nests (primary predation, n = 22). Nests depredated by birds (n = 17) differed little from those depredated by mammals (n = 7) in vegetation cover score (difference ± se: 0.34 ± 0.41), but tended to be closer to field edge (difference ± se: −76 ± 34 m).

Daily survival rate (considering depredation losses during the egg stage) was non‐significantly higher on the first day after observer visit compared with subsequent days (difference ± se, logit scale: 0.3855 ± 0.4268) and did not show a trend across the four days in between nest visits (regression slope ± se, logit scale: −0.0412 ± 0.1425).

DISCUSSION

The overall nest success (Mayfield estimate) in this study (17%) was lower than the average values found in the UK (23–40% [Wilson et al. 1997, Chamberlain & Crick 1999, Donald et al. 2002]), Germany (22% [Jeromin 2002]) and Switzerland (22% [Weibel 1999]). However, our estimate is based on single‐year data and it falls at the lower end of annual variation reported from elsewhere (e.g. 18–38% [Weibel 1999], 16–30% [Donald et al. 2002]); more data are needed to assess an ‘average’ nest success in our study area.

At least 76% (22/29) of primary nest losses were caused by depredation, which is in agreement with other Sky Lark studies (Weibel 1999, Donald et al. 2002). About 72% (18/25) of all depredation events were attributable to birds, which is in stark contrast with a study from the UK (Morris & Gilroy 2008), where all depredation was caused by mammals. While corvids were traditionally suspected as primary nest predators of ground‐nesting farmland birds across Europe, the recent video evidence from western Europe revealed carnivorous mammals as the principal nest predators of waders (MacDonald & Bolton 2008, Teunissen et al. 2008) as well as passerines (Morris & Gilroy 2008). Yet, we showed that raptors (Harriers) might account for as much as 52% (13/25) of total depredation events within our study area. This is unlikely to be a site‐specific effect as the local abundance of Marsh Harriers (two breeding pairs in a 55 km� area) was representative of the large‐scale density for the whole Czech Republic (2.0–3.6 pairs/100 km²; based on data from Št’astný et al. 2006). Corvids and carnivores each accounted for about 20% (5/25) of total nest depredation events. Predation by foxes was low, despite their local abundance (inferred from hunting statistics) and frequently reported effects on farmland birds (Tryjanowski et al. 2002, MacDonald & Bolton 2008). Unlike foxes, predators like martens (López‐Martín et al. 1992, Virgós & García 2002) and hedgehogs (Micol et al. 1994) are thought to require more complex habitat structure and their activity is supposed to be concentrated along field edges and linear habitats (Šálek et al. 2009). Surprisingly, we documented their nest depredations and occurrence (based on trails and faeces) even in central part of large crop fields (>300 m from edge).

Contrary to some previous studies (Weibel 1999, Donald 2004, Morris & Gilroy 2008) we did not find depredation rates to vary with distance to field edge. The possible reasons include comparatively large size of field parcels (about 50 ha in this study versus e.g. 12 ha in a UK study [Vickery et al. 2002]), small number of monitored nests and their location rather far from field edge (median = 195 m, minimum = 30 m). The edge effect on nest success in crop fields is presumably a result of increased mammalian nest depredation near edges, but the principal predators in our study were birds. Moreover, nests depredated by birds tended to be closer to field edge than those depredated by mammals.

Breeding performance of Sky Larks varies markedly among crop types (Wilson et al. 1997, Weibel 1999, Donald et al. 2002), but the number of Sky Lark nests found in each crop type was not entirely proportional to the distribution of crops in the study area. Despite the fact that our results might not be representative of the whole landscape, we believe they are valid for the crops well represented in our sample (Maize, Sugar Beet). Because we searched for nests visually, our sample could be potentially biased towards poorly concealed nests that in turn might be more vulnerable to visually guided predators. Nevertheless, 36% (15/42) of active nests in our sample were found by accidental flushing of incubating birds and these nests did not differ in vegetation characteristics from nests found directly by visual search. Moreover, nests depredated by birds were not less concealed than those depredated by mammals.

Although the use of cameras may bias nest depredation rates and species composition of predators (Richardson et al. 2009), this effect is unlikely to be serious in this study, for several reasons. We minimised the visible parts of video equipment (see Methods), while various human artefacts not associated with research activity (plastic garbage, cans, etc.) were common on all fields under study. Most importantly, the observed composition of predators is conservative with regard to the direction of potential bias. Predators potentially under‐represented because of neophobia to human artefacts (cameras) – raptors – dominated in our sample of video records, while predators potentially over‐represented through attraction to the monitored nests (human trails, infrared illumination) – carnivores – were less frequent. The absence of a temporal relationship between timing of our nest visits and timing of depredation events indicates that overall nest survival in this study was not seriously influenced through changed depredation risk (Weidinger 2008). All cases of ‘spontaneous’ (unexplained) nest desertion occurred more than 12 hours (i.e. night hours on the next day or later) after our last nest visit. Contrary to potential bias (see earlier) video monitoring allowed us to eliminate the problem of uncertain nest fates in estimating nest survival (Manolis et al. 2000, Weidinger 2007).

To conclude, we showed that video surveillance represents an efficient tool to study nest success and nest predators of a farmland breeding passerine. This study of Sky Larks revealed a distinctly different composition of nest predators than studies of farmland birds in other regions or an earlier study of woodland passerines in the same area (Weidinger 2009). Identification of principal predators is vital for causal interpretation of observed nest depredation rates as well as for efficient conservation of populations threatened by nest depredation (Gibbons et al. 2007). Yet, we caution against generalisation of site‐specific results and unsupported assumptions on predator identity.

ACKNOWLEDGEMENTS

This work was supported by the Ministry of Education of the Czech Republic (grant no: MSM 6198959212). We would like to thank Peter Adamík, James Gilroy and Simon Gillings for comments on the manuscript.