Urban areas can be attractive to certain species because of increased food abundance and nesting availability which in turn may increase productivity or breeding rates. However, there are also potential costs associated with urban living such as higher nest failure, poorer body condition or increased prevalence of disease. These costs may result in species trading off the number of young produced against the condition of their young. African Crowned Eagles (Stephanoaetus coronatus) are a rare example of large, powerful apex predators that breed in some urban areas in Africa. In this study, we explored the breeding performance of these eagles across an urbanization gradient in KwaZulu-Natal Province, South Africa, over seven breeding seasons. We predicted that living in an urban environment would increase productivity through an increase in breeding rate (shifting from typically biennial breeding to annual breeding). We then explored if there were any hidden costs associated with such a change in breeding strategy by examining the body condition of chicks from pairs which had successfully bred in the previous year. We found that pairs in more urban areas were more likely to breed annually, resulting in higher breeding rates, but were also less likely to successfully fledge a chick (i.e., lower breeding success). These two contrasting responses counteracted each other and resulted in similar productivity across the urbanization gradient. For those eagles that bred in consecutive years, annual breeding did not appear to have a negative cost on chick condition. The switch to annual breeding is thought to be a response to improved or more constant food sources in urban areas, while higher failure rates might be because of increased nest disturbances from anthropogenic sources (e.g., vegetation clearing, development of industrial areas, human and car traffic). However, although urbanization negatively affected the breeding success of African Crowned Eagles, they are able to persist and thrive in this highly transformed environment likely through an increased breeding rate.

Study Area and Data Collection

The study area covered ~20,000 km² in southern KwaZulu-Natal Province, South Africa, centered on the metropoles of Durban and Pietermaritzburg, and extended to several coastal towns both north and south of Durban (Figure 1; McPherson et al. 2016a, b, 2019). Crowned Eagle nesting sites were initially found by networking with interested individuals/groups (local birding experts, Birdlife and Falconry club members, and online community groups), and accessing unpublished databases from Durban Natural History Museum, eThekwini Municipality, and SABAP2, as well as by direct searching in suitable habitat or where territorial displays were observed.

Crowned Eagle nest monitoring was conducted from August through January the following year for the years 2011 to 2017 (i.e., 7 breeding seasons), which covers their peak annual breeding period in the area (McPherson et al. 2016a, b). Territories were visited regularly, at least twice in the first month, in order to assess occupation (e.g., nest building, incubation or brooding behavior). A nest was classified as active if nest building or fresh green leaves were seen on the nest or if the adults were present in either of these first two nest visits. A nest was classified as having a breeding attempt if incubation or brooding behavior was seen. Nests with a breeding attempt were then monitored during 2-3 nest site visits until conclusion of the breeding event (i.e., until the chicks were around 70±5 days old). Breeding success was defined as having a chick survive until banding age (70 ± 5 days old). After this age, failures in this species and most other large raptors are relatively low (Brown, 1976). Nests were observed from vantage points generally 50 – 200 m away from the nest (see details in McPherson et al. 2016a).

            Crowned Eagle chicks were banded when their estimated age was 65 - 75 days, a time window recommended by experts (S. Thomsett and B. Hoffman pers. comm.). The age estimates used in this study were ascertained by photo reference material of pulli of known age (McPherson et al. 2017)  and were based on size and plumage development. During banding, chicks were weighed (with an electronic hanging scale to the nearest 5 g) and the total length and unfurled length of the 8^th primary feather was taken (with a straight ruler to the nearest 1 mm). All measurements were done in accordance with the SAFRING user manual (de Beer et al. 2001).

Urbanization Score

In order to establish the percentage of urbanization around each nest site, we used the LandCover 2014 raster (GEOTERRAIMAGE, 2015), which classifies land use into 72 different categories. We chose a circular buffer area of 10 km² (radius = 1784.1 m) based on the mean home range size of the species during the breeding season from four telemetered adults in the study area (McPherson et al. 2019). Once the percentage of each land class around each nest site had been calculated, the values for all land classes containing sealed surface (see Rose et al. 2017) were used to calculate an urban score (%) for each nest. Examples of what land classes constituted sealed surface are urban residential, industrial, townships, and mines. In territories where there was more than one nest, the mean urban score was taken to represent the territory.

Statistical Analyses

All analyses were conducted in R version 3.5.1 (R Core Team 2018) with the packages ‘lme4’ (Bates et al., 2015), ‘car’ (Fox and Weisberg, 2018), and ‘effects’ (Fox, 2003). All means are presented with standard deviations. Generalized Linear Models (GLMs) or Linear Mixed Models (LMMs) were used to analyze the data. An initial model selection for the GLMs considered both the linear or quadratic relationship between urbanization and our response variables, as a quadratic relationship could reveal changed breeding demography at intermediate levels of urbanization. In all cases, the linear relationship had the best model fit (lowest AIC) and thus only linear relationships were considered in the final analyses.

            We explored how urbanization affected several Crowned Eagle breeding parameters over seven breeding seasons. GLMs with a binomial distribution were used to investigate the effect of urbanization on three key breeding parameters using the cbind function.  These three variables were i) breeding rates: modelled as the total number of attempts and number of non-attempts (i.e., no nest building activity (nest lining, mating behaviour, incubation etc.) at a previously occupied nest) across the years a territory was monitored; ii) breeding success: modelled as the total number of successful breeding attempts and number of failures across the years in which a territory was active; and iii) breeding continuity: modelled as the number of continuous breeding attempts (i.e., no gap between breeding attempts) and the number of non-continuous breeding attempts (i.e., with at least 1 year gap between breeding attempts) for the total number of years monitored. This binomial approach also accounted for differences in the number of years of data for each territory, by effectively weighting each sample according to the total number of years monitored (models i and iii) or total number of active years (model ii). Additionally, a different GLM was used to investigate Crowned Eagle productivity in relation to urbanization. Here the response variable was the total number of young fledged across all the years each territory was monitored. Models were fitted with a Poisson distribution, with an offset specified as the log of the number of years monitored.

            An LMM was used to explore whether a Crowned Eagle breeding attempt or, more importantly, a breeding success in the previous year, had an influence on the body condition of chicks. For this LMM, the response variable was the condition of each chick (n = 72), where chick condition was the residual from a linear regression of weight against the length of the 8^th primary feather. The explanatory variable was either attempt (t-1), where 0 = no attempt previous year, and 1 = attempt the previous year; we also ran the same model but specifying success (t-1), where 0 = no successful chick produced in the previous year, and 1 = chick successfully produced in the previous year. ‘Year’ and ‘Territory Identity’ were included as random terms to account for the repeated measures from the same territory and from different territories in the same year. As Crowned Eagles only fledge 1 chick per breeding attempt we did not need to control for the number of chicks in a nest.

Legend for data file

The headings for each data sheet included in the data file are described below. Units are included where required.

Nest productivity data sheet:

Chick body condition data sheet: