The authors have declared that no competing interests exist.
Conceived and designed the experiments: JFL JF MF. Performed the experiments: JFL AR TG MF. Analyzed the data: JFL JF MF. Wrote the manuscript: JFL JF TG AM MF.
When species competing for the same resources coexist, some segregation in the way they utilize those resources is expected. However, little is known about how closely related sympatric breeding species segregate outside the breeding season. We investigated the annual segregation of three closely related seabirds (razorbill
Precisely how ecological communities consisting of competing species are able to maintain long-term biodiversity has been debated for decades [
Seabirds provide an illustrative case, as they often breed highly concentrated in large multi-species colonies, and spend the rest of the year dispersed over large areas at sea. Whilst breeding, seabirds are central place foragers and their foraging behaviour is strongly affected by the need to return to land to reproduce [
In this study we used geolocation (GLS, Global Location Sensing) and GPS (Global Positioning System) tracking technologies in combination with stable isotope analyses and dive depth analyses to examine the year-round resource partitioning among three closely related seabird species breeding sympatrically in Southwest Greenland. In the Sub-Arctic Atlantic Ocean, three auk species (razorbill (
We hypothesized that as a consequence of sympatric colonial breeding followed by migration away from the colony, different spatial constraints might operate inside and outside the breeding season. Hence, we predicted that dietary segregation would be strongest during the breeding season, but spatial segregation more likely outside the breeding season when birds are away from the colony.
Fieldwork was carried out in July and August 2009, 2010 and 2011 at Kitsissut Avalliit (Southwest Greenland, 60°46’ N, 48°28’ W), the most southern major seabird colony in Greenland [
The study was performed in accordance with the ethical guidelines promoted by the Association for the Study of Animal Behaviour. All handling of birds in this study was performed in the shade to avoid heat stress and the head of the bird was covered throughout in order to minimise stress. Birds were released after less than 15 minutes of handling. All necessary permits to conduct fieldwork at Kitsissut Avalliit as well as handling, ringing, deploying loggers and taking samples from the different species of birds were obtained from the Ministry of Fisheries, Hunting and Agriculture (APNN) in Greenland (Jr. Nr. 66.01.13 and Dok. Nr. 657095). No approval from an Ethics committee is required to conduct fieldwork and handling birds (ringing, sampling and deployment of loggers) in Greenland. However, as part of the approval process the Ministry of Fisheries, Agriculture and Hunting do take animal welfare into account. Our application thus contained information on fieldwork protocols (including blood sampling), and the permit was given conditional on us following these protocols.
To map the spatial distribution of foraging during the three breeding seasons, we deployed GPS loggers on a total of seven razorbills, five common guillemots and six Brünnich’s guillemots. Three types of GPS loggers were used: i-gotU GT120 (Mobile Action Technology, Inc; removed from their housing; sampling rate = 10 min), TM-TAG (e-Shepherd Solutions, Guardbridge, UK; sampling rate = 2 min) and EP-3.1 (Ecotone, Poland; sampling rate = 2 min), recording date, time, longitude and latitude for approximately two days. GPS loggers were water-proofed using heat shrink tubing (Finishrink) and attached to the back feathers using Tesa® cloth tape, allowing loss of the logger through feather moult, or decay of the tape’s adhesive, in case of failed recapture. Total mass of GPS loggers was approx. 16 g (including attachment material), and corresponded to < 2.4% of adult body mass of the lightest bird. GPS data were processed in ArcGIS 10. Foraging areas were defined as areas where birds were diving and/or resting on water, therefore all points where birds were recorded in flight or at the colony were excluded.
GLS loggers were used to map the overwintering areas of razorbills, common guillemots and Brünnich’s guillemots. GLS loggers use ambient light to estimate the timings of sunset and sunrise. The latitudinal position can then be estimated from day length, and longitudinal position from the time of local midday [
Two types of GLS loggers were used in this study: GLS-MK9 (British Antarctic Survey, U.K), and Lotek LAT2500 (Lotek Marine Technology, St. John’s, Newfoundland, Canada), both of which were attached to metal rings and deployed on the tarsus of the birds. The masses of GLS-MK9 and Lotek LAT2500 in air were approximately 5.3g and 6.2g respectively (including attachments), corresponding to < 1% of adult body mass of the lightest bird.
All birds were caught on the nest, weighed, and a total of 51 loggers were deployed on razorbills, common guillemots and Brünnich’s guillemots (18, 16 and 17 respectively). In total 33 loggers were retrieved, of which 23 provided data. Results were obtained from nine razorbills, eight common guillemots and six Brünnich’s guillemots. Seven birds were tracked for two years (two razorbills, three common guillemots and two Brünnich’s guillemots), thus our data represent 30 bird-years.
Upon recapture, birds were re-weighed and a blood sample of 50µl was taken from either the tarsus or brachial vein and stored in heparin coated vials for use in sex determination by DNA analysis.
Geolocator light data from BAS loggers were processed using the BASTrak software package. A light threshold of 10 and a sun angle of -3° for razorbills and -3.5° for common and Brünnich’s guillemots were used. Sun angles were selected by visually inspecting positions derived using a range of angles during periods when birds were presumed to be near the colony (May–June). Light data from Lotek loggers were processed with LAT Viewer Studio Software (Lotek Wireless, Newmarket, Ontario) using the template fitting method [
Kernel densities for GLS and GPS data were analysed in ArcView GIS 3.2 (ESRI) using the Animal movement extension. We used 50% and 95% kernel density contours to represent the core area of foraging activity and the area of active use, respectively [
Vertical habitat use of birds was determined from diving information. In addition to geolocation, Lotek loggers recorded pressure throughout the year every sixth or tenth minute depending on the programming of the individual logger. None of the three species have been found to have dive durations of more than 5 min (summarized in Table 2 in [
To obtain information on trophic status across the annual cycle, stable isotope analyses were performed on three different tissues (blood, back feathers and throat feathers) collected from breeding adult birds during July 2011. The two batches of feathers were sampled according to the moulting sequence of the three auk species. Soon after breeding (in August–September), all species undergo a complete moult when cover and flight feathers are replaced within a few weeks, leaving the birds flightless. In the spring, they have a partial pre-breeding moult, when only face and throat feathers are replaced [
Blood samples were taken from 14 razorbills, 13 common guillemots and 10 Brünnich’s guillemots and stored in 70% ethanol before preparation for stable isotope analysis. Prior to analysis, samples were dried for 48 h at 60°C and then homogenized. Feathers were collected from 15 razorbills, 13 common guillemots and 11 Brünnich’s guillemots, and stored in zip lock bags before analysis. Feathers were rinsed in a 2:1 chloroform: methanol solution, rinsed 2× in a methanol solution, dried for 48 h at 60°C and homogenized with scissors. Analyses were performed at the Biosystems Department (BIO-309) at Risø National Laboratory, Denmark on 1 mg subsamples of dried and homogenized tissues loaded into tin cups, using a CE 1110 elemental analyser (ThermoFinnigan, Milan, Italy) coupled in continuous flow mode to a Finnigan MAT, Delta PLUS isotope ratio mass spectrometer. Stable isotope abundances were expressed in δ notation as the deviation from standards in parts per thousand (‰) according to the following equation: δX = [(Rsample/Rstandard)-1] × 1000, where X is 13C or 15N and R is the corresponding ratio 13C/12C or 15N/14N. The Rstandard values were based on Vienna-PeeDee Belemnite (VPDB) for 13C and atmospheric N2 (air) for 15N. The δ15N isotopic values reflect the relative trophic position of birds, while the δ13C values reflect the source of carbon (allowing distinction between nearshore and pelagic foraging habitats [
We applied multivariate analysis of variance (MANOVA Pillai’s trace) to examine the isotopic data for differences between species. Each tissue type was analysed separately. The dependent variables were δ15N and δ13C with species and sex as fixed factor. If a significant result was found in the dependent variables we applied one-way analysis of variance (ANOVA) followed by post-hoc Tukey tests. We used linear mixed effect models (LME) with species as a fixed factor to examine species variation in dive depth. Non-independence of data within individuals was accounted for by including individual random effects. All data were transformed when necessary, to meet the requirements of normality. All analyses were carried out in R 2.15.0 (R Development Core Team 2012), and p < 0.05 was regarded as statistically significant.
During the breeding season razorbills, common and Brünnich’s guillemots overlapped in their core foraging areas (50% kernel distribution,
Razorbills (n = 7) shown in blue, common guillemots (n = 5) in red and Brünnich’s guillemots (n = 6) in black. 50 and 95% kernel contours are represented by solid and dashed lines, respectively. The breeding colony is marked with a yellow star and differences in depth between isobaths are 100 m.
During the first half of September, all three species had overlapping activity areas (95% kernel distribution), but where common guillemots had overlapping core foraging areas with both razorbills and Brünnich’s guillemots, razorbills and Brünnich’s guillemots did not overlap in their core area (50% kernels) (
50% (solid) and 95% (dashed) kernel distribution of razorbills (n = 9, blue), common guillemots (n = 8, red) and Brünnich’s guillemots (n = 6, black) during the first half of September (left panel) and April (right panel). Razorbills left the east coast of North America at the end of April, thus the activity area north of 50° N represents the period from 24th of April and onwards. The breeding colony is marked with a yellow star.
All three species left the breeding colony in mid-August and most individuals moved north along the West Greenland coast (
During the pre-breeding period (April), razorbills were spatially segregated from the two guillemot species until the 24th of April and onwards (
During the breeding season, we found a clear difference in dive depth between razorbills (mean = 12.1 m) and the two guillemot species (
Overlapping core and activity area | Overlapping core and activity area | Overlapping core and activity area | ||
p = 0.0031 | p = 0.0008 | p = 0.2812 | ||
δ13C: p = 0.73 | δ13C: p < 0.0001 | δ13C: p < 0.0001 | ||
δ15N: p = 0.32 | δ15N: p = 0.081 | δ15N: p = 0.003 | ||
Overlapping core and activity area | Non-overlapping core area. Overlapping activity area | Overlapping core and activity area | ||
p < 0.0001 | p < 0.0001 | p = 0.7909 | ||
δ13C: p = 0.56 | δ13C: p = 0.004 | δ13C: p = 0.05 | ||
δ15N: p = 0.83 | δ15N: p = 0.63 | δ15N: p = 0.93 | ||
Non-overlapping core area. Overlapping activity area from 24th of April and onwards. | Non-overlapping core area. Overlapping activity area from 24th of April and onwards. | Overlapping core and activity area | ||
p = 0.004 | p = 0.004 | p = 0.92 | ||
δ13C: p = 0.42 | δ13C: p = 0.74 | δ13C: p = 0.89 | ||
δ15N: p = 0.028 | δ15N: p = 0.16 | δ15N: p = 0.78 |
P-values for vertical and trophic partitioning are derived from linear mixed effect models (LME) and post-hoc Tukey tests, respectively. p < 0.05 is regarded as statistically significant.
Median maximum dive depth of Brünnich’s guillemot (BC), common guillemots (CG) and razorbills (RB) during the breeding season (A), post-breeding period (September, B) and pre-breeding period (April, C). The bottom and top of the box show the 25th and 75th percentiles, respectively around the median (solid line). The vertical dashed lines (whiskers) show either the maximum value or 1.5 times the difference in the response variable between its first and third quartiles, which ever is the smallest. Outliers are shown with circles. When there are no outliers the whiskers show the maximum and minimum values.
There was a clear difference in dive depth during the post-breeding season (September) between razorbills (mean = 8.6 m) and both Brünnich’s guillemots and common guillemots (
During the pre-breeding season (April), there was again a clear difference in dive depth between razorbills (mean = 12.4 m) and the two guillemot species (
We did not detect any sex difference in the overall isotopic signature during the breeding, post-breeding or pre-breeding season for common guillemot (MANOVA, F4,20 = 1.454, p = 0.253, F4,20 = 1.429, p = 0.200 and F4,20 = 1.282, p = 0.310 respectively) and Brünnich’s guillemot (F2,7 = 1.550, p = 0.277, F2,7 = 0.2409, p = 0.791 and F2,8 = 0.300, p = 0.748 respectively). For razorbills, we did not detect any sex difference in the overall isotopic signatures during the breeding and pre-breeding season (MANOVA, F4,22 = 0.309, p = 0.868 and F4,24 = 0.612, p = 0.658 respectively), but detected a significant difference in the overall signature during the post-breeding season (F4,22 = 8.459, p = 0.000). We therefore tested for between-sex differences separately for δ13C and δ15N, and found a difference in δ15N between the sexes (ANOVA, p = 0.104 and p = 0.012 respectively). Despite the difference in δ15N for razorbills during the post-breeding season, we decided to pool the sexes in the comparisons of diet of all three species due to the small sample sizes in each season.
During the breeding season, no difference was found between razorbills and common guillemots in δ13C or δ15N (
δ13C and δ15N values (mean ± SE) of blood (breeding), back-feathers (post-breeding) and throat feathers (pre-breeding), of adult razorbills (blue), common guillemots (red) and Brünnich’s guillemots (black).
No difference was found during the post-breeding season between razorbills and common guillemots in δ13C (
During the pre-breeding season, no difference in δ13C was found between the three species (
By combining GPS and GLS spatial data with dive depth and stable isotope analysis, this study presents novel information on the dietary and spatial partitioning during the entire annual cycle of three closely related sympatric breeding auks. In accordance with our predictions, we demonstrate that during the breeding season, razorbills, common guillemots and Brünnich’s guillemots partitioned in either dive depth, diet or both, presumably due to the presence of potential competitors. During both the post-breeding and pre-breeding periods, the three species had an increased overlap in their isotopic niche, and presumably their diet. In addition, razorbill distribution during post-breeding and pre-breeding did not overlap with either of the guillemot species (
During the breeding season, we found a strong overlap in the core foraging areas (50% kernel contours) used by razorbills, common guillemots and Brünnich’s guillemots, although razorbills and common guillemots were found to forage closer to shore than Brünnich’s guillemots. This contrasts with previous studies which showed a segregation in foraging areas of sympatric breeding auks [
Support for prey partitioning during the breeding season comes from the isotopic signatures, where differences in δ15N and δ13C indicated a strong segregation in diet between Brünnich’s guillemots and both razorbills and common guillemots (
An increase in the abundance of resources would be expected to result in a reduction in competition and a subsequent increase in resource overlap, since food partitioning is considered as a mechanism preventing food shortage [
Outside the breeding season, as expected, we found the three species to be distributed on a larger spatial scale and show a reduced segregation in trophic level. During the post-breeding period, Brünnich’s guillemots had slightly lower δ13C values indicating a more pelagic distribution. The core distribution area for razorbills and partly for common guillemots as well as the 95% kernel distribution suggests an earlier departure from the breeding area compared to Brünnich’s guillemots. Desynchronized departure from the breeding colony by different species has also been found elsewhere [
During the period November through most of April, razorbills and the two guillemot species were completely segregated spatially, whereas common guillemots and Brünnich’s guillemots overlapped in their spatial distribution (
Previous studies have shown that Brünnich’s guillemots shift their diet to a lower trophic level during late winter compared to immediately after breeding [
Tracking razorbills, common and Brünnich’s guillemots breeding in southern Greenland with GLS loggers provided new insight into the ecology and non-breeding movements of these sympatric breeding auks. Immediately after the breeding season, birds migrated north as far as Disko Bay (approx. 69° N), following the West Greenland Current (
We found two distinct overwintering areas. One area, exclusively used by razorbills, was situated off the eastern coast of North America. This area is also extensively used by many other seabird species during the non-breeding season, not only birds breeding in North America but also from Europe [
It is very interesting that of the three species, the razorbill was the only one to leave Greenlandic waters, although in Europe razorbills are also known to migrate further south than common guillemots [
To our knowledge, this study is the first to reveal patterns of resource partitioning among three closely related sympatric breeding auks during an entire annual cycle. This study gives insight into how closely-related sympatric breeding seabirds partition resources during the breeding season when they are constrained by the need to return to the breeding site to feed their chicks; and how this resource partition is reduced in the non-breeding season due to the ability to move more freely and thereby segregate spatially. We found dive depth to be very similar across the annual cycle, indicating considerable conservatism, probably due to morphological adaptations to deep diving. Prey choice on the other hand seemed much more flexible and thus is more likely to be affected by the presence of potential competitors. In general, morphological and physiological traits are less plastic and evolve much more slowly than behavioural traits [
This study was conducted at a small colony in Southwest Greenland, where breeding season competition probably is not very intense; we suggest that additional studies are conducted at larger colonies where competition is more pronounced.
(TIF)
We are very grateful to Lou Maurice, Lars Renvald, Charlotte Main, Bob Pratt, Nicholas Per Huffeldt and Knud Falk for help in the field, and especially Per N. Hansen for help with logistics. We thank Anja Nielsen and Per Ambus at the Biosystems Department (BIO-309) at Risø National Laboratory, Denmark for conducting the stable isotope analyses, Kasper L. Johansen for help with spatial modeling, Jens Mogens Olesen for useful discussions and Kelly Edmunds, David Grémillet, Steve Votier and two anonymous reviewers for comments on an earlier draft.