Research Article |
Corresponding author: Marie-Sophie Garcia-Heras ( ms.garciaheras@gmail.com ) Academic editor: Klaus Henle
© 2016 Marie-Sophie Garcia-Heras, Beatriz Arroyo, François Mougeot, Arjun Amar, Robert E. Simmons.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Garcia-Heras M, Arroyo B, Mougeot F, Amar A, Simmons R (2016) Does timing of breeding matter less where the grass is greener? Seasonal declines in breeding performance differ between regions in an endangered endemic raptor . Nature Conservation 15: 23-45. https://doi.org/10.3897/natureconservation.15.9800
|
The timing of breeding can strongly influence individual breeding performance and fitness. Seasonal declines in breeding parameters have been often documented in birds, particularly in the Northern Hemisphere. Fewer studies have investigated whether seasonal declines in productivity vary in space, which would have implications for a species’ population dynamics across its distributional range. We report here on variation in the timing of breeding in the Black Harrier (Circus maurus), an endangered and endemic raptor to Southern Africa. We investigated how key breeding parameters (clutch size, nesting success and productivity) varied with the timing of breeding, weather conditions (rainfall and temperature) and between contrasted regions (coastal vs. interior-mountain). Black Harrier onset of breeding extended over an 8-month period, with a peak of laying between mid-August and end of September. We show a marked seasonal decline in all breeding parameters. Importantly, for clutch size and productivity these seasonal declines differed regionally, being more pronounced in interior-mountain than in coastal regions, where the breeding season was overall shorter. Timing of breeding, clutch size and productivity were also partly explained by weather conditions. In coastal regions, where environmental conditions, in particular rainfall, appear to be less variable, the timing of breeding matters less for breeding output than in interior-mountain regions, and breeding attempts thus occurred over a longer period. The former areas may act as population sources and be key in protecting the long-term population viability of this threatened endemic raptor. This study provides unique evidence for a regionally variable seasonal decline in breeding performance with implications for population biology and conservation.
Black Harrier, Circus maurus , Conservation, Breeding success, Productivity, Fynbos, Karoo, South Africa
Understanding spatial-temporal variations in breeding parameters is an essential component of population ecology, and is particularly important for species that are of conservation concern, as this may help identify reasons for population decline or scarcity (
The timing of breeding is a key determinant of breeding success and productivity (e.g.
Recent research has also indicated that seasonal declines in breeding performance may vary in strength depending on habitat type or location. For example,
Research on the relationship between timing of breeding (i.e. lay date) and breeding output (e.g. clutch size, success or productivity) in birds, up until now, has been mainly conducted in temperate and boreal regions (
The Black Harrier (Circus maurus) is a ground-nesting medium-sized bird of prey, endemic to southern Africa. The species is very scarce with an estimated total world population of less than 1000 mature breeding birds, a distribution range of approximately 500,000 km2 and a far more restricted breeding range of approximately 170,000 km2 (
In this study, we use a large data set of nearly 400 breeding events of this scarce endemic species collected over 15 years (2000–2014) in South Africa to investigate spatial-temporal variations in breeding performance. We first report on regional variation in the timing of breeding, and its association with weather conditions (i.e. rainfall and temperature). We then investigate whether key breeding parameters (clutch size, nesting success and productivity) vary depending on the timing of breeding, geographical location (coastal vs. interior-mountain regions) and weather conditions. Lastly we evaluate whether seasonal declines in breeding performance differ in strength between regions, and the potential implications this might have for the conservation of this species.
Breeding data were collected opportunistically over a large area (ca. 170,000 km2) of temperate southwestern South Africa (29°-34°S; 17°-27°E) from 2000 to 2011. More focused studies took place along the west coast of the Western Cape Province and inland in the Northern Cape Province around Nieuwoudtville (31°19’S; 19°05’E) first from 2000 to 2002, and then from 2012 to 2014. Nests were located in and around national parks (i.e. South African National Parks – SANParks), provincial protected reserves (i.e. Cape Nature), or on private lands. They were spread across a mosaic of different biomes with diverse habitats and vegetation types, many of which are nationally and internationally protected and considered of high biological and ecological values (see e.g.
Black Harriers are ground-nesting birds and, unlike other raptor species, breeders rarely re-use the same nest over the years (
After discovery, nests were visited regularly (usually 2–3 times per breeding event) where possible to assess nesting success and productivity. However, because of the extensive nature of the study area, not all breeding areas and nest sites were monitored consistently each year, and for some remote areas, nests sites were only visited once, or were last visited prior to fledging. During each nest visit, we noted the nest contents (i.e. number of eggs or nestlings) and, if the nests contained nestlings, a visual estimate of age was taken. In a subsample of nests, wing, tail and tarsus length (mm), and mass (g) of chicks were measured. Nest visits were kept as brief as possible (< 20 min) and an effort was made to leave the vegetation around the nest undisturbed. The location of nests was recorded using a global positioning system (GPS). A total of 490 nests were located between 2000 and 2014, although not all variables examined in this study were available for each breeding attempt, so sample size varies among analyses.
Lay dates were estimated by subtracting 31 days (
Clutch size was defined as the maximum number of eggs laid. When possible, nests were visited twice during the incubation period with the second visit timed to coincide with the estimated date of hatch. This ensured that we recorded the exact number of eggs laid per breeding event. Nests that were visited before the clutch was finished and that subsequently failed, or only during the nestling period were excluded from clutch size analyses. Clutch size was known for 191 breeding attempts.
Breeding output was measured in two ways, nesting success (known for n = 263 breeding attempts) and productivity (n = 261). Nesting success was classified as 1 for those nests where at least one young was raised to 35 days old, or 0 otherwise. Productivity was defined as the number of young reaching 35 days of age (range 0–4) for pairs that laid a clutch. Black Harriers fledge at approximately 40 days old (
Nest coordinates were incorporated in a geographical information system (QGIS Valmiera 2.2.0), projected on WGS84-UTM-34S as the coordinate reference system. Using this GIS, we calculated and identified the following variables for each nest: i) Altitude, from the Shuttle Radar Topography Mission (SRTM) 90 m Digital Elevation Database v4.1 (Srtm90m). ii) Region (coastal and interior-mountain) was defined using a combination of nest altitude (from SRTM) and distance to the coast. Coastal nests were defined as those located within 15 km from the coast and with a maximal altitude of 100 mASL (n = 328). Nests located further than 15 km from the coast and with an altitude higher than 100 mASL were considered as interior-mountain (n = 146). However, this classification excluded nine nests that were located higher than 100 mASL (average of 118 m), but within 15 km from the coast and for the purpose of our analysis these were classified as coastal. Another eight nests were located at an altitude lower than 100 m, but 45 km from the coast, and these were classified as interior-mountain. In both cases, we believe our classification to more accurately describe conditions for those 17 nests. This regional classification was initiated by
Weather data were obtained for the period 2000–2014 from 17 weather stations distributed throughout the study area (source: South African Weather Services: http://www.weathersa.co.za) (Figure
All statistical analyses were conducted using R 3.2.3 (the
To reduce the number of weather variables and to account for potential collinearity among them, we conducted a Principal Component Analysis (PCA) on monthly rainfall and temperature data for each station and study year. We selected the first four weather Principal Components (PCs) for subsequent analyses (a scree plot showed a marked drop in explained variance between the fourth and the fifth PC). PCs were chosen for analyses on the effect of weather on breeding parameters, rather than using raw weather data, because we did not have a strong a priori hypothesis of the time period over which weather may be more influential. Therefore, using raw data would have meant exploring the effect of a high number of potential explanatory variables (weather over different time periods). Furthermore, our PCs had clear biological meanings (see results), which helped in interpreting the relationships found. However, because PCs include information about weather in all months, in our discussion we placed most emphasis on the meaning of each PC for the months prior to the variable in question (e.g., for the relationship between lay date and weather, we focus on the meaning of each PC for the months prior to laying, not subsequently).
We investigated regional differences in the weather PCs using General Linear Mixed Models (GLMMs, statistical package lme4,
To analyse factors affecting variation in breeding phenology, we used GLMMs that included year as a random effect, so that we could identify patterns that would describe what happens in an average year. The “lay date” of each nest (response variable) was fitted with a Gaussian distribution and an identity link function. The initial model included the explanatory variables of region and weather variables (the first four weather PCs). These models were conducted on a subsample of 393 nests for which both lay date and weather data were available.
GLMMs with year as a random effect were also used to explore clutch size, nesting success and productivity (response variables) in relation to region, lay date, and weather (explanatory variables). Initial models also included the interaction between region and lay date to look for regional differences in seasonal variations in breeding performance. For models where this interaction was significant, we re-ran the same model but without the interaction to test for differences between regions. Nesting success was fitted with a binomial distribution, and clutch size and productivity were fitted with a Gaussian distribution. Even though the latter may not be ideal for productivity data, using a Poisson distribution produced models with large dispersion parameters, whereas Gaussian models performed well and model residuals were normally distributed. Analyses of clutch size were conducted on a subsample of 183 breeding events for which clutch size, lay date and weather data were available. Analyses of variation in nesting success and productivity were conducted on a subsample of 223 and 222 breeding events, respectively, for which lay date and weather data were also available.
A stepwise backward procedure was performed for model selection (with the function drop1), and likelihood ratio chi2 tests (LRT) on AIC differences were used to select the best models.
Samples sizes differed between regions and our slope estimates for the relationships between lay date and breeding parameters could be influenced by this or hinge on data from a few very early or very late nests (see Figure
Study regions were characterized by different weather conditions (Figure
Monthly average temperature (a) and rainfall (b), according to region (coastal, white bars; interior-mountain, dark grey bars). Also presented are Coefficient of Variation (100× SD/Mean) for both climatic variables (dashed line for coastal, solid line for interior-mountain), as well as frequency distribution of breeding initiation (n = 402) (c) during the study period (2000–2014).
Black Harrier breeding performance (a clutch size b nesting success c productivity) variation according to lay date and region (coastal nests: white circles/dashed line; interior-mountain nests: grey dark circles/ solid line). Lines represent modelled data from the GLMM results (Table
The PCA analysis on monthly rainfall and temperature data rendered four PCs explaining approximately 60% of the variance (Table
Results of the Principal Component Analysis conducted on weather data (monthly averages of daily rainfall and daily temperatures) collected in 2000–2014 at 17 weather stations (see Figure
PC1 | PC2 | PC3 | PC4 | |
---|---|---|---|---|
Rain. JAN | 0.014443 | 0.129292 | -0.01123 | 0.353204 |
Rain. FEB | -0.02877 | 0.171029 | 0.098396 | 0.416038 |
Rain. MAR | -0.01663 | 0.196125 | 0.032239 | 0.45707 |
Rain. APR | -0.06009 | 0.303045 | 0.03444 | 0.067087 |
Rain. MAY | -0.11894 | 0.281114 | 0.093557 | -0.23835 |
Rain. JUN | -0.12415 | 0.308008 | 0.131795 | -0.26365 |
Rain. JUL | -0.124 | 0.302469 | 0.084539 | -0.23318 |
Rain. AUG | -0.13295 | 0.325751 | 0.102831 | -0.27611 |
Rain. SEP | -0.09693 | 0.350088 | -0.01698 | -0.04674 |
Rain. OCT | -0.0634 | 0.28149 | 0.001479 | 0.105491 |
Rain. NOV | -0.07844 | 0.32201 | 0.029495 | 0.174901 |
Rain. DEC | 0.006074 | 0.168862 | 0.060451 | 0.290475 |
Temp. JAN | 0.292895 | 0.057873 | 0.310684 | 0.011043 |
Temp. FEB | 0.315231 | 0.065366 | 0.257761 | -0.01914 |
Temp. MAR | 0.303027 | 0.066636 | 0.231843 | -0.11742 |
Temp. APR | 0.335735 | 0.082603 | -0.02066 | -0.20133 |
Temp. MAY | 0.230807 | 0.16073 | -0.29901 | -0.07898 |
Temp. JUN | 0.127701 | 0.149213 | -0.44499 | -0.0962 |
Temp. JUL | 0.127055 | 0.120859 | -0.43149 | -0.10167 |
Temp. AUG | 0.190731 | 0.153715 | -0.36607 | 0.128207 |
Temp. SEP | 0.272202 | 0.082607 | -0.18223 | 0.075789 |
Temp. OCT | 0.352615 | 0.052975 | 0.017441 | 0.024666 |
Temp. NOV | 0.344244 | 0.035273 | 0.119129 | -0.00367 |
Temp. DEC | 0.290036 | 0.046253 | 0.258563 | 0.009631 |
Variance explained | ||||
Proportion | 0.254 | 0.1554 | 0.1361 | 0.08704 |
Cumulative | 0.254 | 0.4084 | 0.5454 | 0.63249 |
All weather PCs varied significantly among years, but only PC1 and PC3 were significantly different between regions (Table
Results of the General Linear Mixed Models (GLMMs) testing for differences between years and regions (coastal vs. interior-mountain) in weather variables (PC1, PC2, PC3, and PC4; see Table
Dependent variables | Explanatory variables | DF | LRT | P |
---|---|---|---|---|
PC1 | Year | 15 | 82.15 | <0.0001 |
Region | 1 | 4.67 | 0.031 | |
PC2 | Year | 15 | 32.11 | 0.006 |
Region | 1 | 0.02 | 0.88 | |
PC3 | Year | 15 | 214.52 | <0.0001 |
Region | 1 | 21.11 | <0.0001 | |
PC4 | Year | 15 | 95.19 | <0.0001 |
Region | 1 | 0.04 | 0.83 |
Lay date (n = 393 nests) was remarkably well spread through the year, spanning 8 months, from mid-May to mid-December, and followed a unimodal distribution in each region (Shapiro normality test, w = 0.98, p = <0.0001, n = 287 for coastal region; w = 0.95, p = 0.0009, n = 106 for interior-mountain region) with a peak during mid-August to end of September (Figure
Lay date was negatively associated with weather PC2 (slope = -0.26 ± 0.07) and PC4 (slope = -0.27 ± 0.10; Table
Results of the Generalized Linear Mixed Models (GLMMs) testing for variations in lay date (15-day periods), clutch size, nesting success and productivity. “Year” was included as a random effect in all models. Initial models included region (coastal vs. interior-mountain), weather variables (PCs) and lay date (for clutch size, nesting success and productivity), as well as interactions between region and lay date. Stepwise backward model selection was performed based on AIC values. We present the results of final models.
Dependent variables | Explanatory variables | DF | LRT | P |
---|---|---|---|---|
Lay date | PC2 | 1 | 15.00 | 0.0001 |
PC4 | 1 | 7.28 | 0.007 | |
Region | 1 | 16.14 | <0.0001 | |
Clutch size | PC2 | 1 | 4.23 | 0.039 |
Region×Lay date | 1 | 7.45 | 0.006 | |
Nesting success | Lay Date | 1 | 17.59 | <0.0001 |
Productivity | PC2 | 1 | 5.08 | 0.024 |
Region×Lay Date | 1 | 2.84 | 0.092 |
Clutch size averaged 3.58 ± 0.64 eggs (range: 2–5; n = 183 nests). Clutch size varied with rainfall (PC2, Table
In total, 31% of nests (n = 223) monitored during the study period failed to produce fledglings. Nesting success declined significantly with lay date (Table
Productivity among monitored nests averaged 1.66 ± 1.30 fledglings (range 0 - 4 fledglings, n = 222 nests). Productivity was positively associated with weather PC2 (Table
This study revealed an extended breeding period for the Black Harrier and profound consequences of the timing of breeding on breeding performance. Moreover, it is one of the few studies that document a seasonal decline in breeding performance in a southern African species (
Most strikingly, we found that seasonal declines varied among regions for clutch size and also (less markedly) for productivity. The seasonal decline in these parameters was progressive and moderate in coastal regions but much more abrupt in interior-mountain regions. Thus, clutch size and productivity were overall higher in interior-mountain than in coastal regions early in the season (until September), but differences were not found or values were higher in coastal regions for nests initiated from October onwards (Figure 3ac). Interestingly, we did not find a significant difference between regions for nesting success, suggesting that regional differences in declines in productivity may simply result from differences in clutch size patterns. Additionally, this suggests that differences between regions are more influential early in the breeding cycle. Ultimately, neither clutch size nor productivity were, on average, significantly different between regions, indicating that differences between regions early and late in the season balanced each other out.
Seasonal declines in breeding performance can be explained by differences in the quality of individuals breeding early or late and/or by a worsening of environmental conditions as the breeding season progresses (
Temperature was overall higher in coastal regions until August, when clutch sizes were smaller there, but the opposite pattern was found from October onwards, when clutch sizes were greater in coastal regions. Temperature variation could thus be an indicator of the temporal variation in quality of environmental conditions among regions. However, temperature (PC1) did not significantly influence clutch size (or any other breeding performance parameter), so differences are likely to be related to other factors, such as food availability or habitat quality. Black Harriers mostly feed on small mammals (ca. 65% of the diet), particularly on Four-Striped Mouse (Rhabdomys pumilio) and African Vlei Rats (Otomys sp.) (
Black Harriers showed a remarkably extended breeding period, with the onset of laying spread over 8 months (mid-May to mid-December). A wide spread in timing of breeding has been reported in other raptors from the Southern Hemisphere [e.g., 8 months for the Black Sparrowhawk (Accipiter melanoleucus),
Nevertheless, we found a clear seasonal peak, with most laying (ca. 50% of clutches) occurring between mid-August and the end of September. This, together with the strong seasonal decline in breeding performance observed, indicates that optimal timing for breeding is limited for this species, despite the overall large extended breeding period. This peak coincides with a sharp drop in rainfall levels and an increase in temperature (Figure
The strong associations between timing of breeding, temperature and rainfall also indicate that climate change may further influence shifts in breeding phenology of southern African birds (
We also found differences in lay date between regions: Black Harriers breeding in coastal regions started laying on average about 15 days earlier, and clutches occurred over a more extended period than those breeding in interior-mountain (Figure
Black Harriers have been described as Fynbos specialists (
Recent changes in climate conditions within Africa during the last decades (
This study provides unique evidence for spatial variation in the strength of seasonal declines in breeding performance. This main finding has broad implications for population biology and conservation. Environmental heterogeneity needs to be accounted for when considering overall population viability, and our findings suggest that where environmental conditions are less variable and more predictable, the timing of breeding may have less importance for the production of young. Relative differences in individual quality between early and late breeders, which can explain the breeding seasonal declines (
This study was funded by the NRF (National Research Foundation, South Africa; Grant no. 90582 to R.E.Simmons), by the DST-NRF Center of Excellence at the Percy FitzPatrick Institute, University of Cape Town, by CSIC (Consejo Superior de Investigaciones Cientificas- PIE 201330E106) and by private landowners and organizations. Particular thanks for economic support are due to BirdLife South Africa, Inkwazi and Wits Bird Club, “Golden Fleece Merino”, URC (University Research Council), Jakkalsfontein Private Nature Reserve, TOSS (Two Oceans Slope Soarers), Natural Research UK, Hawk Mountain (USA), the Peregrine Fund, Sven Carlsson-Smith, Nial Perrins, Chris Cory, Gisela Ortner and James Smith. We are also grateful to Graham Cumming to help M-S Garcia-Heras with financial support in 2014. We also thank SANParks and Cape Nature to access to study sites, and the South African Weather Services for the acquisition of weather data. We also thank Andrew Jenkins, Odette Curtis, Anne Williams, Bettie Bester, Binks Mackenzie, Txuso Garcia, Juan Jose Luque-Larena and all the students from the Percy FitzPatrick Institute of African Ornithology who helped with fieldwork and data collection, and an anonymous reviewer for comments on the manuscript.