1. Ecology
Download icon

Wind prevents cliff-breeding birds from accessing nests through loss of flight control

  1. Emily Shepard  Is a corresponding author
  2. Emma-Louise Cole
  3. Andrew Neate
  4. Emmanouil Lempidakis
  5. Andrew Ross
  1. Swansea University, United Kingdom
  2. Max Planck Institute for Animal Behaviour, Germany
  3. University of Leeds, United Kingdom
Research Article
Cite this article as: eLife 2019;8:e43842 doi: 10.7554/eLife.43842
4 figures, 1 table and 4 additional files


Figure 1 with 1 supplement
Airflows around Skomer Island (51.73611°N 5.29628°W), modelled with a SW wind.

(A) A horizontal cross section of wind speed (m s−1) at 10 m above sea level (which intersects the island, given in grey) shows the reduction in wind strength near the cliffs (see supplement 1). (B) The horizontal wind vectors within the inset in A, modelled at 2 m normal to the surface, and coloured according to the total wind speed (m s−1). Wind is funnelled into the canyon on the left of the image (the Wick colony is located along the South side of this canyon), forcing birds to enter this area with a tailwind. (C) The turbulence intensity, TI, (a dimensionless ratio of the RMS of the turbulent wind fluctuations to the mean wind) at a distance of 2 m normal to the surface, within the inset shown in A. Typical values are ~0.1, so values of ~1 (red areas), indicate highly variable winds. Note these high values occur in areas with low mean winds (blue colours in B), so actual gust strength is low.

Figure 1—figure supplement 1
The mean wind speed at each colony is shown relative to the value at over the sea, where values < 1 indicate a reduction in wind speed close to the cliffs.

Within each colony, relative wind speeds are shown at different heights (observer locations are marked as top, the vertical mid-point of the cliff is given as middle, and bottom gives conditions at the base of the cliff). At most cliff sites the speed is less than upwind, a pattern that is particularly evident at the base of the cliffs. The only exception to this is at the Mew Stone where flow is channelled between the rock and the main island in westerly/easterly wind directions, leading to higher speeds.

Landing success decreases with increasing wind speed for both guillemots (solid line) and razorbills (dashed line).

(A) The probability of landing (derived from the statistical model), declines to ~0.1 in winds of 10 m s−1. Binned raw data are shown for both species (guillemots as filled circles, razorbills as open circles; data are grouped with n ≥ 30 observations per bin). As success also varies with ledge type, both the model output and raw data refer to birds landing on long narrow ledges. (B) The probability of landing according to the mean, at-sea wind speed, as derived from the probabilistic model (also for long narrow ledges and a TI value of 0.2). The difference between the x-axes indicates the increase in wind speed over open water, compared to near the cliffs, as estimated using airflow model outputs averaged across all wind directions. (C) The distribution of at-sea wind speeds across the breeding season (for 2005–2018, recorded at the M5 wave buoy and reduced to 2 m ASL).

Figure 2—source data 2

The raw data on landing observations, along with associated data on species, ledge type and wind speed (see also ‘parameter definitions’).

Figure 3 with 2 supplements
The cumulative probability of landing according to wind speed, ledge type and species.

Seasonal wind speeds near the breeding cliffs will vary with wind direction and colony location (Figure 1—figure supplement 1 ). For the prevailing SW wind direction, median wind speeds across the breeding season are predicted to be ≤7 m s−1 (first two columns). Upper quartile speeds are predicted to be ≤ 9 m s−1 (third column), and reasonable maxima ≤ 16 m s−1.

Figure 3—figure supplement 1
Observations of landings grouped according to ledge size for each species.
Figure 3—figure supplement 2
The largest guillemot colony on Skomer; the Wick.

A long ledge runs roughly half way down the cliff. This has sections where it expands into large platforms and others, such as that to the left of the picture, where it forms a long, narrow ledge.

The distribution and orientation of guillemot breeding colonies on Skomer Island (digitised from the 2015 breeding bird survey [Stubbings et al., 2015]).

(A) Breeding colonies (grey and black regions) appear to be distributed all around Skomer, however, bearings of the 11 densest colonies (marked in black), given in (B), show that while colonies appear uniformly distributed overall (Z = 0.314, p=0.346, df = 10, Rayleigh test), none are oriented towards the prevailing south-westerly wind direction, despite the availability of cliff habitat (unoccupied sections are indicated with open circles). Study colonies are indicated with stars.



Table 1
The output of the best performing model.

High frequency wind speed measurements were obtained for 6140 observations and all models of landing success were run using this dataset. Height and ledge were included as factors.

ParameterDfEstimate ± SEp-valueF valueDeviance explained
wind1−0.62 ± 0.22<0.001223.2225.04
    medium ledge−1.21 ± 0.12
    small ledge−2.48 ± 0.15
turbulence1−0.81 ± 0.430.4670.530.05
    razorbill1.10 ± 0.11
    lowest height−0.58 ± 0.19
wind * ledge2<0.00114.213.19
    wind*medium ledge0.03 ± 0.04
    wind*small ledge0.22 ± 0.05
turbulence * ledge20.0154.230.95
    turb*medium ledge0.95 ± 0.38
    turb*small ledge0.07 ± 0.50

Data availability

Landing observations are included in the supporting files. The raw data for the GLMM (Source data 1) and the GLMM code (Source code 1) have been uploaded as additional data files (csv and R files respectively). The parameters listed in the raw data file are defined in an accompanying txt file (Supplementary file 2).

Additional files

Source code 1

Models of landing success.

Supplementary file 1

Summary information for the breeding cliffs where landing data were collected.

Numbers of guillemots (GM) are taken from the 2015 Skomer Island breeding bird survey (Stubbings et al., 2015).

Supplementary file 2

Definitions of parameters within "Skomer Landings" data.

Transparent reporting form

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)