Journal of Ornithology

, Volume 153, Issue 3, pp 735–746

Patterns of nest placement in a population of Marsh Tits Poecile palustris

Authors

    • Centre for Ecology and Hydrology
  • Ross A. Hill
    • School of Applied SciencesBournemouth University
  • Lindsay J. Henderson
    • Faculty of Biomedical and Life Sciences, Ecology and Evolutionary BiologyUniversity of Glasgow
  • Paul E. Bellamy
    • Royal Society for the Protection of BirdsThe Lodge
  • Shelley A. Hinsley
    • Centre for Ecology and Hydrology
Original Article

DOI: 10.1007/s10336-011-0790-2

Cite this article as:
Broughton, R.K., Hill, R.A., Henderson, L.J. et al. J Ornithol (2012) 153: 735. doi:10.1007/s10336-011-0790-2

Abstract

The factors influencing nest placement by territorial birds are not fully understood, including the roles played by habitat, conspecific attraction and female experience of a previous nesting location. We used 7 years of Marsh Tit (Poecile palustris) nest-site and territory data, and high-resolution vegetation models derived from remote sensing, to investigate spatial patterns of nest placement with regard to previous female experience and age, conspecific attraction, and habitat in a woodland environment. We found no evidence for an effect of conspecific attraction or previous nest location on nest placement within the territory. However, first-year (FY) females placed nests in a random spatial pattern within their territories, and after first-year (AFY) females predominantly placed nests within the central parts of their territories, away from conspecifics. The core area of each breeding territory was centred on a region of comparatively taller overstorey and less understorey than other parts of the territory. Nest-sites were situated in localised areas of a similar structure, although absolute differences between selected and non-selected areas of the territory were not substantial. Both female age groups nested in areas of the territory where the overstorey contained relatively more Common Ash (Fraxinus excelsior) and Field Maple (Acer campestre), which may have been related to tree height, but there was no selection for English Oak (Quercus robur). We found no significant habitat differences between the territories of FY and AFY females that explained their differing patterns of nest placement.

Keywords

Conspecific attractionLidarMarsh TitNest placementRemote sensing

Zusammenfassung

Nistplatzwahl in einer Population von SumpfmeisenPoecile palustris

Die Faktoren, die die Nistplatzwahl territorialer Vogelarten beeinflussen, sind nicht gänzlich verstanden. Dies beinhaltet die Funktion des Habitats, innerartliche Attraktion, aber auch die Erfahrung der Weibchen mit früheren Neststandorten. Mit Hilfe eines Datensatzes aus sieben Jahren zu Nistplätzen und Revieren von Sumpfmeisen (Poecile palustris) sowie hoch aufgelösten Vegetationsmodellen (aus Fernerkundungsdaten hergeleitet) untersuchten wir räumliche Muster der Nistplatzwahl im Hinblick auf Alter und Erfahrung der Weibchen, innerartliche Attraktion und Habitat in einem Waldgebiet. Wir fanden keine Beweise für einen Effekt der innerartlichen Attraktion oder früherer Neststandorte auf die Nistplatzwahl innerhalb der Reviere. Allerdings legten vorjährige Weibchen, räumlich gesehen, ihre Nester innerhalb ihrer Reviere in einem zufälligen Muster an. Ältere Weibchen legten dagegen ihre Nester überwiegend in den zentralen Bereichen ihrer Reviere an, abseits von Artgenossen. Das Kerngebiet der einzelnen Brutreviere lag in Bereichen mit vergleichsweise höherem Oberholz und weniger Unterholz als in anderen Bereichen der Reviere. Die Neststandorte befanden sich in Bereichen mit ähnlicher Struktur, obwohl grundsätzliche Unterschiede zwischen gewählten und nicht gewählten Bereichen innerhalb der Reviere nicht substantiell waren. Beide Altersgruppen der Weibchen nisteten in Bereichen der Brutreviere, in denen die höherwüchsige Struktur relativ mehr aus Gewöhnlicher Esche (Fraxinus excelsior) und Feldahorn (Acer campestre) bestand, was mit der Baumhöhe zusammenhängen könnte. Allerdings gab es keine Selektion für Stieleichen (Quercus robur). Wir fanden keine signifikanten Habitatunterschiede zwischen den Revieren der beiden Altersgruppen der Weibchen, die die Verteilung der Neststandorte erklären.

Introduction

The causes and consequences of nest-site selection are recurrent themes in ornithological research. Studies have commonly investigated the influence of habitat and/or previous experience on nest placement, often within the context of resource availability, reproductive success and nest predation (Nilsson 1984; Marzluff 1988; Martin 1993; Muller et al. 1997; Sockman 2000). There is also growing interest in nest placement with regard to the spatial arrangement of conspecifics, with attempts to understand the relative importance of social and physical determinants of nest-site selection (e.g. Ramsay et al. 1999; Melles et al. 2009; Bayard and Elphick 2010), including their impact on the incidence of extra-pair paternity (Mennill et al. 2004; Westneat and Mays 2005).

For territorial species, nest placement decisions can be constrained by the distribution of specific resources within the territory, such as the availability of snags for primary cavity-nesters like the Black-backed Woodpecker (Picoides arcticus) (Bonnot et al. 2009). The limited availability of nest resources may even suppress the breeding density of secondary cavity-nesters, as in the Great Tit (Parus major) (Mänd et al. 2009), while nest placement by Crested Tits (Lophophanes cristatus) appears tied to a particular habitat structure (Summers et al. 1992; Atiénzar et al. 2009). Male Cerulean Warblers (Dendroica cerulea) can alleviate such constraints on female nest placement by defending territories containing multiple patches of potential nesting habitat (Jones and Robertson 2001). However, nest placement by female Hooded Warblers (Wilsonia citrina) was random with regard to habitat variables and was instead clustered among conspecifics (Melles et al. 2009). Similarly, Ramsay et al. (1999) found that Black-capped Chickadees (Poecile atricapillus) placed nests close to their territory boundary, discounting habitat, food availability and female previous experience as significant factors in nest-site selection at the territory scale. Mennill et al. (2004) concluded that this pattern of nest placement was driven by conspecific attraction among Black-capped Chickadees, with females that were paired to low-ranking males placing their nests close to the territories of high-ranking neighbouring males. This behaviour did not facilitate extra-pair copulations, however, and the advantage was unclear. Similar work on Red-winged Blackbirds (Agelaius phoeniceus) has also failed to show a significant relationship between the location of a female’s nest within the territory and the incidence of extra-pair paternity by neighbouring males (Weatherhead et al. 1994; Westneat and Mays 2005). As such, the relative importance of social and habitat cues to nest placement within the territory, and the function of observed patterns of conspecific attraction, remain unclear (Mennill et al. 2004; Melles et al. 2009).

Characterising complex vegetation at the territory level can be problematic, however, as localised ground-based sample plots may not adequately describe available habitat at an appropriate spatial extent or resolution (e.g. Jones and Robertson 2001). The understanding of the influence of vegetation structure and composition on patterns of bird distribution is being advanced by the increasing use of remote sensing methods (Bradbury et al. 2005; Gottschalk et al. 2005). For woodland birds in particular, lidar (light detection and ranging) technology can describe habitat structure at a high spatial resolution and at a landscape-scale (Broughton et al. 2006; Martinuzzi et al. 2009; Goetz et al. 2010), while optical imagery can provide information on the vegetation species composition (Goetz et al. 2010; Hill et al. 2010). The combination of habitat datasets derived from remote sensing, and bird distribution data derived from field surveys, can permit powerful analyses of bird–habitat interactions (Fuller et al. 2005; Goetz et al. 2010).

The Marsh Tit (Poecile palustris) is a Eurasian 10–12 g sedentary woodland species which adopts a similar breeding strategy to the closely-related Black-capped Chickadee, whereby socially monogamous pairs occupy exclusive spring territories and nest-sites are selected by the female (Morley 1953; Broughton et al. 2006). Marsh Tits are obligate users of secondary cavities, primarily derived from natural decay processes in the stems of trees and shrubs, but rarely use the abandoned cavities of primary cavity-nesting birds (Wesołowski 1996; Broughton et al. 2011). There have been no investigations into the social and habitat factors that may govern nest placement within the territories of this species.

In this study, we aimed to investigate the roles of habitat, female experience of previous nest locations, and conspecific attraction on Marsh Tit nest placement, using 7 years of territory, bird and nest-site data. The application of high-resolution models of canopy and understorey vegetation, derived from airborne remote sensing, enabled the spatial analyses of nest-sites, territory composition and habitat structure in exceptional detail. Studies of nest placement by cavity-nesting birds within the context of territory configuration and social organisation remain rare. This work on the Marsh Tit provides the first investigation into such patterns for an additional species, and permits comparison with results from the congeneric Black-capped Chickadee (Ramsay et al. 1999; Mennill et al. 2004).

Methods

Territory and nest data

The study was conducted in 153 ha of woodland in the Monks Wood National Nature Reserve in Cambridgeshire, UK (52°24′N, 0°14′W; hereafter Monks Wood), and six neighbouring patches of woodland that lie within 4 km to the east of that site (hereafter the Eastern Woods), four of which are 4–7 ha in size and the other two 27 and 70 ha. Monks Wood and the Eastern Woods are fragments of lowland, semi-natural ancient woodland, dominated by Common Ash (Fraxinus excelsior), English Oak (Quercus robur) and Field Maple (Acer campestre) in the tree canopy, with variable amounts of Silver Birch (Betula pendula), European Aspen (Populus tremula) and elm (Ulmus spp.). The understorey layers are dominated by hawthorn (Crataegus spp.), Blackthorn (Prunus spinosa) and Common Hazel (Corylus avellana) (Broughton et al. 2006). The study sites are mature woodland and broadly correspond to the ‘old-growth’ stage of development (Quine et al. 2007), with little visual change in vegetation structure between years. The four smallest Eastern Woods each held a single Marsh Tit territory during the study, the two larger woods held 6 and 10 territories, respectively, and Monks Wood held 21–23 territories each year.

Almost the entire breeding population of Marsh Tits in Monks Wood (96%) were individually marked with colour-rings between 2003 and 2010. Birds were sexed and aged as first-years (FY) or older (after first-years: AFY) according to Broughton et al. (2008). Six female spring immigrants that could not be aged before the post-breeding moult were assumed to be FY, because this age group formed 92% of the female immigrants of known age (Broughton et al. 2010). Each spring, the boundaries of all territories in Monks Wood were identified during a minimum of six visits to each territory between February and May. Territory occupants were followed for 1–4 h on each visit, and movements were plotted on 1:10,000-scale geo-referenced maps (Broughton et al. 2006, 2010). Particular attention was paid to territorial behaviour such as boundary disputes and singing, in order to identify territory boundaries, and playback of recorded song was also employed to elicit behavioural responses. The mapped bird registrations were digitised in a Geographical Information System (GIS), and territories were delineated as non-overlapping maximum defended areas based on standard territory-mapping techniques (Bibby et al. 2000), using the behavioural attribute data of bird registrations for interpretation (see Broughton et al. 2006 for further detail).

A total of 117 nest locations (112 in natural cavities, 5 in nest-boxes) were identified in Monks Wood during survey visits to each territory in April and May of 2004–2010 (14–21 nests per year, with 36–72% belonging to FY females). Nest locations were digitised using coordinates from a hand-held GPS receiver (Garmin eTrex H model), with further correction based on reference to features in digital habitat models (see below), giving an estimated accuracy of 3–12 m. Only nests from a pair’s initial breeding attempt each year were included in analyses, as second attempts may be constrained by the availability of time (Wesołowski 2000). Although Marsh Tits are secondary cavity-nesters, and thus reliant on pre-existing tree holes, we have previously shown that there is a very low rate of cavity re-use which indicates that nest-sites were abundant throughout Monks Wood (Broughton et al. 2011). On this basis, we were able to exclude the possibility that female choice of nest placement was significantly constrained by nest-site availability.

The Eastern Woods were used to test the wider applicability of nest–habitat interactions observed in Monks Wood (see below), and were each surveyed 4–8 times during April–May in 2008 or 2010 in order to determine the approximate territory positions of their un-ringed Marsh Tit populations, using territory-mapping techniques as per Monks Wood. The boundaries of 15 territories were estimated and digitised, and were termed ‘approximate territories’ to acknowledge the lower precision than that achieved for the digitised Monks Wood territories due to the birds not being marked. The locations of nests within these territories were digitised as per Monks Wood, with 14 nests located in 2010 (6 and 4 nests in the two larger woods, respectively, 1 nest each in three of the four smaller woods) and 1 in 2008 (in the fourth small wood), all belonging to different females of unknown age.

Nest placement in relation to patterns of social hierarchy among males was assessed in Monks Wood using male age as a proxy for dominance, as age-related prior residency correlates strongly with social status in the Poecile genus (Nilsson and Smith 1988; Koivula et al. 1993; Schubert et al. 2007). For each nest, the age disparity between the resident male and the male in the nearest territory to the nest was calculated. A positive value indicated that the neighbouring male was older than (and socially dominant to) the resident male, and a negative value indicated that the neighbouring male was younger than (and sub-dominant to) the resident. Values of zero indicated males of the same age where dominance could not be inferred, and were excluded from analyses.

To investigate whether females were placing nests closer to the borders than to the centres of their territories, we defined the territory core as the area containing all parts of the territory that were closer to the geographical centre (centroid) than to the territory border. The remainder of the territory was defined as the margin, containing the area that was closer to the border than to the centroid. If females were placing nests with regard to the social hierarchy of neighbouring males, we predicted a greater proportion of nest placement within the territory margins than by random chance, and these locations would be associated with a positive age disparity between the nearest neighbouring male to the nest and the resident male.

In order to test for an effect of previous female experience of nest locations on placement of nests in Monks Wood, we measured the distance between nests of the same female in subsequent years, where the territory centroid fell within the previous year’s territory and where the previous year’s nest was successful. This distance was then compared with that between the nests of new FY females nesting within an area for the first time, where the new territory centroid fell within the previous territory, and the successful nest of the previous female in the preceding year. As such, the distribution of the nests of FY females was essentially acting as a ‘random’ comparison with that of AFY females. We were unable to test for any effect of unsuccessful nesting on subsequent nest placement by the same female, as the nest failure rate was low (16.4%) (Broughton et al. 2011) and only three females survived to breed in the study area in the year following nest loss.

Habitat models of woodland canopy overstorey and understorey structure

Digital habitat models of Monks Wood and the Eastern Woods were used to investigate the role of habitat selection in relation to nest placement within the territory. For each territory and approximate territory, 25-m buffers were defined around nest sites to encompass the mean maximum distance used by foraging adults when provisioning nests at Monks Wood (Carpenter 2008), and were then used to compare habitat within the remainder of the territory. To assess woodland structure, a 0.5-m resolution raster canopy-height model (CHM) of all woods was generated during leaf-on conditions from airborne lidar data acquired in June 2005 (data for all woods processed as described in Hill and Broughton 2009). Lidar is an active remote-sensing technique whereby a short pulse of near-infrared light is fired at the ground by an aircraft-mounted laser scanner. The timing and intensity of the reflected return signals from the surfaces below are then used to calculate a ranging measurement (Lefsky et al. 2002), with the first-return signal measuring the range to the first object encountered (e.g. a tree top) and the last-return signal measuring the range to the last object encountered at or above the ground at the same location (e.g. sub-canopy shrubs or the ground). The CHM of the woods thus described the structure of the woodland canopy surface, containing height information for the tallest vegetation structure present in each 0.5-m grid cell using the first-return data (Hill and Broughton 2009). Lidar data has been shown to be capable of representing the height of deciduous woodland vegetation with a high degree of accuracy (Lefsky et al. 2002; Lim et al. 2003), including the Monks Wood study site (Hill et al. 2002), so we did not attempt to validate data in the field. However, field observations in Monks Wood were used to ascribe height values of >8 m to the overstorey layer of mature tree crowns (Hill et al. 2010), which were extracted to create a separate overstorey height model. This model was used to determine mean values of overstorey height for the nest buffers and the corresponding remaining parts of the territories for comparison, and also to compare overstorey height between the territory cores and margins. Data acquisition and processing for the Eastern Woods followed the methodology used for Monks Wood, which we justified based on the similar woody species present, age-structure and management. The results from the limited single-year dataset from the Eastern Woods were used to assess the wider applicability of the comprehensive multiple-year territory data from Monks Woods. As the Eastern Woods data involved unique females and territory locations for each nest, issues of pseudo-replication and bias were minimised, and the data were considered to be a representative snap-shot of the woods in 1 year.

The CHM height values of 1–8 m corresponded to the understorey vegetation (Hill and Broughton 2009). The CHM from first-return data during leaf-on conditions contained only limited information on the understorey layer, as much of it was hidden beneath the overstorey in the airborne lidar data. Therefore, a leaf-off CHM (from April 2003) was also acquired using last-return data in order to derive additional information on the understorey coverage that was exposed beneath the dormant overstorey (Hill and Broughton 2009). Height values in the range 1–8 m in either the leaf-on or leaf-off CHMs were extracted as the total understorey data. The mean height and spatial coverage of this understorey vegetation was generated for each nest buffer and corresponding territory, and each territory core and margin. As with the overstorey, the Eastern Woods were used to assess the wider applicability of the results derived from Monks Wood.

Overstorey tree species composition model

A 1-m resolution raster map of the six tree species comprising the Monks Wood overstorey was produced from a supervised classification of time-series Airborne Thematic Mapper (ATM) data acquired in 2003. The distribution of overstorey in this map was determined using a corresponding 1-m resolution lidar-derived CHM acquired in June 2000 (Hill and Thomson 2005), applying a vegetation height threshold of >8 m, as above. Each 1-m grid cell, where the canopy was 8 m or taller, was thus assigned to a tree species, and the remaining grid cells (below 8 m in height) were assigned to an unclassified category. The resulting canopy tree species model had a surveyed overall accuracy of 88% (Hill et al. 2010), and was used to compare the proportions of tree species within the overstorey of each nest buffer and corresponding territory, and the territory cores and margins.

Statistical analysis

To investigate patterns of nest placement, we carried out randomisation tests by generating single random points within each territory over 999 iterations, determining the frequency with which points fell within the territory cores. The position of the observed frequency of nests within cores on the frequency distribution was used to calculate statistical significance. Nest placement within the territory was compared between FY and AFY females, and examined for age groups combined.

Mean overstorey height, understorey height and understorey coverage were treated as paired data for comparison of nest buffers and their corresponding territories, and for territory cores and their corresponding margins. We applied angular transformation to proportional understorey coverage data, with non-parametric tests being employed where normalisation of data could not be achieved by transformation, and Spearman’s rank order correlation was used to test for inter-variable correlations. All tests were two-tailed. As the vegetation structure in the study sites was predominantly mature, we expected little dynamic change over the study period (2004–2010). As we were comparing relative values of vegetation height and coverage between the nest buffers and remainder of the territories, and the territory cores and margins, absolute changes in the vegetation structure after the 2003 and 2005 lidar data acquisition would only be problematic if they occurred unevenly within territories. If such uneven change had occurred over the timeframe of the study, we reasoned that any significant patterns of vegetation structure relating to nest placement that were evident in the early years of the study would become less distinct or unapparent in the later years, as the time between acquisition of the lidar and bird data increased. We tested for this effect by comparing the pattern of results of the nest buffer and territory analyses of structure (overstorey and understorey height, and understorey coverage) for the early years of study (2004–2006 combined) and the later years (2008–2010).

To test whether the proportion of any tree species in the overstorey differed between the nest buffers and territories, the proportional values required transformation to address the problem of non-independence when some or all were considered together, i.e. the mutually exclusive relationship between each tree species in a limited area whereby an increase in the proportion of one species must be at the expense of one or more others. To overcome this, the proportions of each tree species within the overstorey of each territory area (nest buffer, territory remainder, core or margin) were divided by that of the unclassified vegetation category in the same defined area. This reduced the dimensionality of the data by one, in the manner of a compositional analysis (Elston et al. 1996). Taking the natural logarithm of each ratio left the transformed values unbounded above or below and closer to a normal distribution, allowing each to be treated as an independent variable in comparative analyses.

Results

Nest placement within the territory

Of the 117 nests located in Monks Wood, 95 nest cavities were unique, and no territorial pairs that were present during a breeding period failed to acquire a nest cavity. Monks Wood territory sizes ranged from 1.45 to 14.12 ha (mean = 5.55, SD = 2.03, n = 153) and, for those territories used in nest placement analyses, geometry determined that the area of the territory core was substantially smaller than that of the territory margin (cores: mean = 1.42 ha, SD = 0.48; margins: mean = 4.09 ha, SD = 1.42; n = 117). For Monks Wood nests, all AFY females had nested in the area as FYs, and 42.4% of birds in the AFY age category nested in more than 1 year. Breeding attempts by the same female in different years were treated as independent, however, as each combination of territory boundary, nest-site and the arrangement of neighbours between and within years was unique. In addition, we found no evidence of an effect of previous female site-selection on nest placement, as AFY females did not place nests closer to successful sites in the previous year (n = 45), when compared to naïve FY females nesting in the same area for the first time (n = 26) (Mann–Whitney U test: U = 1,058.0, P = 0.15). For all females combined, nests were located in the territory core more frequently than random points (Table 1). This was also true of AFY females when the age groups were considered separately, but there was no statistical difference between the distribution of the nests of FY females among the territory cores and margins when compared to that of the random points (Table 1).
Table 1

Observed nest placement by Marsh Tits (Poecile palustris) within the territory cores (areas closer to the territory centroid than to the boundary) compared to results of a randomisation test using 999 iterations to generate a random point within each territory

 

Observed number of nests in territory core (%)

Median (range) of random points in territory cores

P

All females (n = 117)

44 (37.6)

29 (27)

<0.01

FY females (n = 58)

14 (24.6)

14 (20)

0.13

AFY females (n = 59)

30 (50.9)

15 (24)

<0.01

FY first-year females, AFY after first-year females

The median age of males paired to FY females was 1 year old (range = 5 years), this being significantly younger than the 3 years of age (range = 7 years) of males paired to AFY females (Mann–Whitney U test: U = 2,666.0, P < 0.01). For females of all ages, there was no relationship between the nests placed in the territory core or margin and the age disparity (younger or older) between the resident male and the nearest neighbouring male to the nest (χ2 = 0.18, df = 1, P = 0.67). This was also true when considering FY females in isolation (Fisher’s exact test, P = 0.46), despite their greater tendency to locate their nests within the territory margin (Table 1). Furthermore, for FY females, nest placement within the margin was not associated with the age of their breeding partner (χ2 = 0.15, df = 1, P = 0.68). These results indicated that the age difference between neighbouring males and the resident males was no greater for nests placed within the territory margins compared to those placed within the territory cores, and that the pattern of nest location of females paired to young males did not differ from that of females paired to older males.

Habitat in the nest buffers compared with the territories

For females of all ages in Monks Wood, overstorey height in the nest buffers was significantly taller than that of the corresponding territories (Table 2), but only by 3.8%. For all females in the Eastern Woods, mean overstorey height within the nest buffers was 6.8% taller than in the remainder of the approximate territories, and this result approached statistical significance (Table 2). This pattern was also observed when considering female age-groups separately in Monks Wood (Table 2), and there was no difference in the overstorey heights of nest buffers between FY and AFY females (two-sample t test: t = 0.40, P = 0.69).
Table 2

Mean habitat variables in the nest buffer and remainder of the territory for each breeding territory in Monks Wood (MW, n = 117) and the Eastern Woods (EW, n = 15) for all females

 

Nest buffers

Territories

Test statistic

P

Overstorey mean height (m), mean (SD)

 MW: all females

15.99 (2.14)

15.38 (1.33)

t = 3.99

<0.01

 MW: FY females

16.07 (1.94)

15.32 (1.25)

t = 3.24

<0.01

 MW: AFY females

15.91 (2.33)

14.44 (1.41)

t = 2.36

0.02

 EW: all females

17.92 (3.24)

16.70 (3.02)

t = 2.07

0.06

Understorey mean height (m), median (range)

 MW: all females

2.87 (4.48)

3.28 (1.75)

W = 855.00

<0.01

 MW: FY females

2.97 (2.13)

3.34 (1.69)

W = 149.00

<0.01

 MW: AFY

2.78 (4.40)

3.26 (1.67)

W = 305.00

<0.01

 EW: all females

2.95 (1.41)

3.34 (1.33)

W = 168.00

<0.01

Understorey mean coverage (proportion), mean (SD)

 MW: all females

0.39 (0.09)

0.43 (0.06)

t = 4.83

<0.01

 MW: FY females

0.40 (0.09)

0.44 (0.05)

t = 3.59

<0.01

 MW: AFY females

0.39 (0.09)

0.42 (0.06)

t = 3.22

<0.01

 EW: all females

0.44 (0.11)

0.50 (0.10)

t = 2.36

0.03

Comparisons performed using paired t tests and Wilcoxon Signed Rank W tests, with angular transformation being applied to proportion data before testing

FY first-year females, AFY after first-year females

For all females combined, understorey height in the nest buffers was significantly less than in the corresponding territories in Monks Wood and the Eastern Woods, and this effect persisted when female age groups were considered separately (Table 2). Understorey height was greater in the nest buffers of FY than AFY females (Mann–Whitney U test: U = 3,840.0, P < 0.02), though only by 6.4%, but values for the remainder of the territories indicated that understorey was comparatively taller across the entire territory of FY females. The proportion of understorey coverage was significantly lower in the nest buffers than the remainder of the territories for all females in Monks Wood and the Eastern Woods, and for the FY and AFY female age groups in Monks Wood. There was no difference in understorey coverage between the nest buffers of FY and AFY females (two-sample t test: t = 0.62, P = 0.54). When comparing the observed pattern of the results for understorey height, coverage and overstorey height for nest buffers and corresponding territories during the early years of study (2004–2006) and later years (2008–2010) there was little difference in the pattern between time periods (Table 3), or between each period and the full duration of study (Tables 2 and 3). The only discrepancy was the near-significant result for overstorey mean height in 2004–2006, but this was considered a minor inconsistency. The results in Table 3 indicated that the lidar data was equally valid in assessing relative within-territory variables throughout the study period.
Table 3

Mean habitat variables in the nest buffer and remainder of the territory for breeding territories in Monks Wood during the early years of study (2004–2006, n = 52) and the later years of study (2008–2010, n = 44) for females of all ages

 

Nest buffers

Territories

Test statistic

P

Overstorey mean height (m), mean (SD)

 2004–2006

15.87 (2.11)

15.44 (1.34)

t = 2.01

0.07

 2008–2010

16.20 (2.31)

15.31 (1.27)

t = 2.02

<0.01

Understorey mean height (m), median (range)

 2004–2006

2.95 (1.94)

3.30 (1.75)

W = 1,378.00

<0.01

 2008–2010

2.91 (4.47)

3.44 (1.63)

W = 990.00

<0.01

Understorey mean coverage (proportion), mean (SD)

 2004–2006

0.38 (0.06)

0.43 (0.08)

t = 4.67

<0.01

 2008–2010

0.40 (0.09)

0.43 (0.05)

t = 2.09

0.04

Comparisons performed using paired t tests and Wilcoxon signed rank W tests, with angular transformation being applied to proportion data before testing

Table 4 shows that Silver Birch, European Aspen and elm accounted for a negligible proportion of the overstorey in the Monks Wood nest buffers and territories, and these species were excluded from analyses. Of the remaining tree species, a significantly greater coverage of Common Ash and Field Maple were present in the nest buffers compared to the territories, although the absolute difference for Field Maple was only 3%. There was no difference between FY and AFY females in the proportions of Common Ash in the nest buffers (two-sample t test: t = 0.93, P = 0.36), nor of Field Maple (two-sample t test: t = 1.13, P = 0.26). There was less coverage of English Oak in the nest buffers than the territories of both female groups, but the difference was not statistically significant (Table 4).
Table 4

Mean proportional composition of overstorey tree species in 25 m radius nest buffers (n = 117), the corresponding remainder of the Marsh Tit territories, and the whole of Monks Wood (MW)

 

Mean (SD) proportion in nest buffers

Mean (SD) proportion in territories

Wilcoxon signed rank W statistic

P

Proportion of MW

Mean (SD) height in MW (m)

Common Ash

0.54 (0.22)

0.45 (0.16)

5,760.0

<0.01

0.48

16.22 (3.05)

European Aspen

0.05 (0.08)

0.05 (0.04)

0.06

14.17 (3.11)

Silver Birch

0.01 (0.01)

0.02 (0.02)

0.02

12.15 (2.54)

Elm

<0.01 (0.01)

<0.01 (0.01)

0.01

17.50 (3.98)

Field Maple

0.15 (0.12)

0.12 (0.07)

5,302.0

<0.01

0.12

15.93 (3.30)

English Oak

0.11 (0.12)

0.15 (0.08)

3,892.0

0.23

0.18

14.88 (2.87)

Unclassified vegetation

0.13 (0.15)

0.20 (0.10)

0.15

11.95 (2.52)

Mean heights (and SD) of each tree species in the overstorey of the whole of Monks Wood are also given, derived from height values of the 1 m grid cells assigned to each species. Wilcoxon signed rank W tests were performed on transformed data (see text) for common species only

Significant negative correlations were apparent between the ranked values of overstorey and understorey mean heights in the nest buffers and corresponding territories of FY and AFY females in Monks Wood, these being moderate relationships in the nest buffers but strong relationships in the territories (Table 5). This indicated that the vertical structure of the overstorey and understorey was more diverse in the nest buffers compared to the territory remainders, which was also indicated by the larger standard deviations of the means and ranges around the medians for nest buffer overstorey and understorey mean height in Table 2. There were only weak or insignificant relationships between understorey height and coverage, but overstorey mean height was positively and significantly correlated with the coverage of Common Ash for both female age classes. A positive relationship between overstorey mean height and Field Maple was also apparent, being stronger for AFY females.
Table 5

Correlations between habitat variables in Marsh Tit nest buffers (B) and their corresponding territory remainders (T), and territory cores (C) and their corresponding margins (M), for first-year (FY, n = 58) and after first-year (AFY, n = 59) females in Monks Wood

 

FY females

AFY females

Overstorey mean height

Understorey mean height

Overstorey mean height

Understorey mean height

B

T

B

T

B

T

B

T

Understorey mean height

−0.42**

−0.82**

−0.43**

−0.87**

Understorey coverage

−0.06

0.08

0.29**

0.03

−0.43

0.17

0.24

−0.04

Common Ash coverage

0.37**

0.37**

0.44**

0.35**

Field Maple coverage

0.27

0.26

0.44**

0.47**

English Oak coverage

−0.06

−0.10

−0.01

−0.06

 

C

M

C

M

C

M

C

M

Understorey mean height

−0.61**

−0.79**

−0.67**

−0.87**

Understorey coverage

0.34**

0.00

0.09

0.29**

0.20

−0.28**

0.23

0.49**

Common Ash coverage

0.59**

0.65**

0.62**

0.71**

Field Maple coverage

0.24

0.43**

0.33**

0.62**

English Oak coverage

−0.36**

−0.16

−0.11

−0.09

Values are Spearman’s rank order correlation statistic rs

P < 0.05, ** P < 0.01

Habitat in the territory cores and territory margins

Although AFY females nested disproportionately within the territory cores, compared to random nest placement by FY females, the original analyses of social and habitat factors did not account for this difference. We therefore performed additional analyses on habitat within the territory cores and margins of Monks Wood in an attempt to detect habitat differences between the territories of FY and AFY females to explain the difference in the patterns of nest placement. For females of both age groups, territory cores contained significantly taller overstorey and shorter understorey than the corresponding margins, with greater coverage of all three tree species and less coverage by understorey (Table 6). There was no statistically significant difference for any variable between the territory cores of FY and AFY females (results of two-sample t tests not presented).
Table 6

Mean habitat variables for the territory cores and margins of each breeding territory in Monks Wood for first-year (FY, n = 58) after first-year (AFY, n = 59) females

 

Territory cores

Territory margins

Test statistic

P

Overstorey mean height (m), mean (SD)

 FY females

15.85 (1.66)

15.17 (1.24)

t = 3.24

<0.01

 AFY females

15.84 (1.94)

15.31 (1.31)

t = 2.36

0.02

Understorey mean height (m), median (range)

 FY females

2.98 (1.97)

3.43 (1.79)

W = 224.00

<0.01

 AFY females

3.07 (2.60)

3.31 (1.75)

W = 232.00

<0.01

Understorey mean coverage (proportion), mean (SD)

 FY females

0.42 (0.06)

0.46 (0.06)

t = 3.59

<0.01

 AFY females

0.42 (0.07)

0.43 (0.06)

t = 3.22

<0.01

Common Ash coverage (proportion), mean (SD)

 FY females

0.45 (0.21)

0.43 (0.17)

t = 3.58

<0.01

 AFY females

0.53 (0.20)

0.50 (0.16)

t = 3.03

<0.01

Field Maple coverage (proportion), mean (SD)

 FY females

0.15 (0.11)

0.14 (0.07)

t = 5.34

<0.01

 AFY females

0.12 (0.09)

0.11 (0.07)

t = 5.11

<0.01

English Oak coverage (proportion), mean (SD)

 FY females

0.20 (0.15)

0.17 (0.09)

t = 5.49

<0.01

 AFY females

0.14 (0.10)

0.13 (0.07)

t = 4.96

<0.01

Comparisons performed using paired t tests and Wilcoxon signed rank W tests, with proportions being transformed before testing (see text for details)

Moderate or strong negative correlations were found between overstorey and understorey mean heights in the territory cores and margins of both female age groups, with positive associations between overstorey height and Common Ash coverage (Table 5). There was a weak yet significant correlation between understorey coverage and understorey mean height in the margins of FY females, with a stronger relationship for AFY females. Positive correlations were also apparent between overstorey height and Field Maple coverage, again being stronger for AFY than FY females. There were weak or insignificant negative correlations between overstorey mean height and the coverage of English Oak for both female age groups.

Discussion

The results confirmed that nest-site availability was unlikely to have influenced nest placement by Marsh Tits in Monks Wood, as the frequency of cavity re-use was lower than that recorded for this species in the primeval conditions of Białowieża forest, Poland, where holes were super-abundant (Wesołowski 2006; Broughton et al. 2011). We also found no evidence that the pattern of nest placement by female Marsh Tits was influenced by conspecific attraction. Indeed, placement by FY females was essentially random within the territory, while AFY females appeared to site their nests preferentially within the territory core, away from conspecifics. This differed from the work by Ramsay et al. (1999) and Mennill et al. (2004), which found that females of the closely-related Black-capped Chickadee placed their nests close to their territory border. They suggested that this may have been due to conspecific attraction, although geometry determines that the part of a convex polygonal territory which is closer to the boundary is always much larger in area than that which is closer to the centre. Thus, nests have a higher probability of being located nearer the territory boundary by chance and not necessarily by selection. By dividing each territory into a core and margin and using random points to derive expected patterns of nest placement, we were able to account for this effect. However, this discrepancy does not explain the finding of Mennill et al. (2004) that nests of females with low-ranking partners were close to neighbouring high-ranking males, a result that we could not replicate using age as a proxy for male rank in the Marsh Tit. This may be because age was an insufficient surrogate for social rank in our analyses, or that Marsh Tits differ from Black-capped Chickadees in this aspect of their social behaviour, but the purpose of any such behaviour remains elusive. We did not investigate extra-pair paternity in this study, but evidence from the Black-capped Chickadee (Mennill et al. 2004), and also the Red-winged Blackbird (Weatherhead et al. 1994; Westneat and Mays 2005), suggested that access to extra-pair copulations had little bearing on nest placement within the territory.

Our results further disagree with those for the Black-capped Chickadee (Ramsay et al. 1999; Mennill et al. 2004), and also the Hooded Warbler (Melles et al. 2009), in finding that nest placement by Marsh Tits was related to habitat variables. We identified statistically significant differences in habitat structure between the nest buffers and the remainder of the Marsh Tit territories, and the results supported the previous finding that Marsh Tit occupation was associated with taller canopy height at our Monks Wood study site (Broughton et al. 2006). The nest buffers of both female age groups in Monks Wood had a significantly taller overstorey than other areas of the Marsh Tit territories, and the trend was replicated in the nest buffers of the smaller single-year dataset for the Eastern Woods, and also the territory cores of Monks Wood. Additionally, understorey height and coverage were significantly lower in the nest buffers compared to the remainder of the territories, and also in the territory cores compared to the margins. The mean height of the overstorey was negatively correlated with that of the understorey, but there appeared to be no strong or consistent relationship between understorey coverage and other structural variables, and no selection for greater understorey height or coverage by Marsh Tits.

The results for understorey conflicted with Hinsley et al. (2007) and Carpenter et al. (2010), who concluded that Marsh Tit occupation and abundance was positively associated with the density and coverage of this vegetation layer. The failure to detect this trend around nest sites in our study suggests that selection for understorey might not be directly related to nesting behaviour or nest-site selection; it is possible that greater visibility around nest-sites confers some advantage in relation to predator detection and avoidance. As there was no substantial change in the relationships between nest placement and vegetation structure between the early and late years of the study period (Table 3), we were confident that these results were not compromised by the lidar data becoming invalid over the time span of the study due to changes in vegetation structure. Any such changes were likely to have been relatively homogeneous or insignificant by the latter years of the bird data collection.

Carpenter (2008) found that Common Ash and Field Maple were the tree species most frequently used for foraging in Monks Wood when Marsh Tits were provisioning young, and we detected a significantly higher proportion of these species in the overstorey of the nest buffers compared to the rest of the territories, and in the territory cores compared to the margins. However, as Common Ash was the tallest tree species in Monks Wood, followed by Field Maple, it was unclear whether selection by Marsh Tits was operating on the basis of tree species or tree height. There were significant positive correlations between overstorey height and the coverage of Common Ash and Field Maple, and these relationships were apparent in almost all subdivisions of the territories. In contrast, English Oak was the shortest tree species and showed a weak or very weak negative correlation with overstorey height in all territory areas. Yet English Oak was also more common in the territory cores than the margins, particularly for FY females, but not in the nest buffers compared to the rest of the territories. The territories of both female age groups were, therefore, centred on core areas containing relatively less understorey and a taller overstorey containing greater amounts of all three tree species, compared to the margins. Nest-sites, meanwhile, were centred on more localised areas of the territory that were likewise dominated by a modest understorey and relatively tall overstorey, but the trees were composed of Common Ash and Field Maple in preference to English Oak. This suggested a possible selection process driven by tree height at the scale of the territory core, but with additional selection based on tree species at the smaller scale of the nest buffer. Such a selection is supported by Carpenter’s (2008) finding that Common Ash and Field Maple were preferred over English Oak by foraging Marsh Tits when breeding.

Why Common Ash and Field Maple may be selected over English Oak around nest-sites is unclear. We have no information on food availability on the various tree species within the study site, although Carpenter (2008) suggested that Marsh Tits may have been competitively excluded from foraging areas by other tit species. As such, Marsh Tits may not select for Common Ash and Field Maple per se, but instead be avoiding competition with more dominant species in areas rich in English Oak, which offers high quality habitat for Great Tits (Wilkin et al. 2009) and Blue Tits (Cyanistes caeruleus) (Stenning 2008). The potential for inter-specific avoidance or attraction is rarely considered in nest placement studies (though see Burger 1987), but is worthy of further investigation for the Marsh Tit due to the potential contribution of competitor species to its negative conservation status (Carpenter et al. 2010; Broughton et al. 2011).

It was logical to surmise that AFY females preferentially nested in the territory cores because these areas contained a concentration of the same habitat structural characteristics as those found in the nest buffers, namely a tall overstorey with a comparatively modest understorey. However, the territory cores of FY females contained a similar concentration of these attributes, so it was unclear why the nests of FY females were randomly placed throughout the territory. Some aspects of habitat selection may have been based on variables not examined in our analyses, perhaps involving other attributes of tree quality, which may account for this finding and also some of the subtle or non-significant differences in our results. As FY females tended to be paired to younger, and presumably lower-ranking, males than the breeding partners of AFY females, these sub-dominant males may have been less able to acquire and centre their territories upon large areas of high-quality habitat, which we may have been unable to detect. Alternatively, although FY females appeared capable of assessing habitat quality at the scale of the nest buffer, they may have been less able to assess habitat quality at the larger scale of the whole territory, perhaps due to less familiarity with their environment than older females, or lack of previous breeding experience.

While we were unable to identify the underlying basis for the random pattern of nest location observed among FY females, we found no evidence to support the hypothesis that Marsh Tit nest placement was significantly influenced by conspecific attraction or a female’s previous breeding location. Instead, nest placement appeared to be related to habitat variables within the foraging distance around the nest site and, for AFY females at least, habitat characteristics in the territory core compared to the margin. Although many studies have reported a relationship between localised habitat characteristics and nest placement in other territorial species (Jones and Robertson 2001; Oppel 2004; Atiénzar et al. 2009), few have addressed this within the context of explicit territory boundaries and the arrangement of neighbouring conspecifics. Where this has been undertaken, conspecific attraction has appeared to be more influential than within-territory habitat variation in determining nest placement (Ramsay et al. 1999). To date, however, the resolution or spatial extent of habitat data in such studies has been much lower than that of the bird data, which may obscure fine-scale patterns of habitat selection. Our study appears to be the first to combine data for nest-sites, territory boundaries and population structure with high-resolution lidar data and optical imagery in the spatial analysis of nest placement. Due to the complex nature of vegetation in the spatially extensive territories of many bird species, remote-sensing may offer the best opportunity of obtaining habitat data of sufficient detail with which to investigate nest placement at the territory scale.

Acknowledgments

The authors thank Natural England for access to Monks Wood National Nature Reserve and private landowners for access to the Eastern Woods, Dr. Stephen N. Freeman for statistical support, Dr. Daria Dadam and Dr. Jane Carpenter for additional fieldwork, and the three anonymous reviewers of the manuscript. This work was funded by the Natural Environment Research Council (NERC). The remote sensing data were acquired by the NERC Airborne Research and Survey facility (ARSF) in conjunction with the Unit for Landscape Modelling (ULM) at the University of Cambridge. All ringing activities were licensed by the British Trust for Ornithology, and complied with UK law. The authors declare that they have no conflict of interest. Richard K Broughton is a Visiting Research Fellow at Bournemouth University, UK.

Copyright information

© Dt. Ornithologen-Gesellschaft e.V. 2011