Skip to main content
Dryad logo

Marabou Stork Stable Isotope and Trace Metal Data

Citation

Francis, Roxane; Kingsford, Richard; Brandis, Kate; Murray-Hudson, Michael (2021), Marabou Stork Stable Isotope and Trace Metal Data, Dryad, Dataset, https://doi.org/10.5061/dryad.pg4f4qrm2

Abstract

We compared diets of marabou storks Leptoptilos crumenifer foraging from urban landfills and natural areas in northern Botswana using stable isotope analyses and inductively coupled plasma mass spectrometry (ICP-MS) on moulted feathers. 

Methods

Stable Isotope and Trace Metal ICP-MS Analyses

We removed any remaining surface dirt on feathers with distilled water, followed by vigorous washing in deionized water (RO) and a chloroform methanol solution wash (see methods in Paritte and Kelly (2009)) to remove any surface oils. Feathers were then left to air dry for 24-48 hours. Feathers were identified as either body or flight feathers based on their size, colouration and structure. For stable isotope analyses, feather barbs from the tip of each feather were clipped, placed in tin capsules and weighed (~500 ug). Standards of glutamic acid 40 and glutamic acid 41 were analysed at the beginning, middle and end of each run through the mass spectrometer (Seminoff et al. 2007), with their accuracy measured as continuous flow isotope ratio mass spectrometry for δ15N and δ13C values (Brenna et al. 1997). For ICP-MS analyses of trace metals (Aluminium, Cadmium, Chromium, Copper, Iron, Lead, Manganese, Nickel, Potassium and Zinc), a piece ~4x2cm of feather was removed from the top of the vane of each feather, avoiding the centre rachis, sampling about 0.2g (DeNiro and Epstein 1981). Samples were digested with HNO3 (open), and then analysed using an inductively coupled plasma mass spectrometer (ICPMS, Perkin Elmer, NexION 300D with universal cell technology). Calibration standards were prepared from commercial stock standard solutions, referenced to certified bovine liver (Altmeyer et al. 1991, Kim et al. 1998, Cardiel et al. 2011) (Supplementary Table 1).

Statistical Analyses

We separately modelled for differences in 𝛿13C and 𝛿15N ratios across sites, varying in distance from the closest landfill (Fig. 1). We used a linear modelling approach, with the glmmTMB package (Brooks et al. 2017), with fixed predictor variables including collection site (converted to a numerical variable based on the distance from the closest landfill); collection region (Chobe in the East vs the Okavango in the West) and feather type (body vs flight). We included region in the model to account for geological differences which may alter trace metal concentrations naturally present in the environment (Huntsman-Mapila et al. 2005, Kelepile et al. 2020). We performed a power analysis on the 𝛿13C and 𝛿15N model results using the pwr package (Champely et al. 2018), with both falling above the standard 0.8 threshold indicating our significance testing was valid (Cohen 1965).

To determine differences in the isotopic niches of the marabou population, we then divided the feathers into three distance divisions. This separated marabous feeding at the landfills in Maun and Kasane (<10 km); in Chobe National Park, where they potentially still visit the Kasane landfill (10-55km) and; those unlikely to be frequently visiting landfills in the Okavango Delta (>55km). We used the R package SIBER (Jackson et al. 2011) to visualise the standard ellipse areas (SEA) representing niche widths and fitted Bayesian models to the data, using the rjags package (Plummer 2013). We also calculated the overlap of the niche area occupied by each group using the package nicheROVER (Lysy et al. 2014), with the Monte Carlo method to bootstrap the same number of samples for each group (500), repeated 500 times (α=0.05), averaged to provide final percentages.

To model trace metal concentrations in feathers, we used separate glmmTMB models (with a gaussian family), including collection region, distance from landfill sites and feather type as predictor variables. No trace metals fell below the measurement detection limit, and so all were included in the modelling (Supplementary Table 1). We used the DHARMa package (Florian Hartig 2019) to visualise the QQplot and the residual vs predicted values of all glmmTMB models, checking the data satisfied the assumptions of normality and homogeneity of variance. We log transformed Al, Cr, Cu, Fe, K, Ni and Pb concentrations because data were skewed. Finally, we compared trace metal concentrations to suggested avian healthy limits (Malik and Zeb 2009, Ullah et al. 2014, Abdullah et al. 2015).

Funding

Taronga Conservation Society*, Award: NA

Taronga Conservation Society