^{1}

^{*}

^{2}

^{3}

^{4}

^{5}

^{4}

^{5}

The authors have declared that no competing interests exist.

Conceived and designed the experiments: AM AB. Performed the experiments: AM AB. Analyzed the data: AM A. Kane. Contributed reagents/materials/analysis tools: AM AB DD A. Kotze. Wrote the paper: AM A. Kane.

Investigating the ecology of long lived birds is particularly challenging owing to the time scales involved. Here an analysis is presented of a long term study of the survival and population dynamics of the marabou stork (

Survival estimates are critical parameters for various ecological models, particularly population dynamics, which in turn inform conservation

The marabou stork is a widespread scavenging bird occurring in savanna habitats throughout sub-Saharan Africa

The inset shows the location of Hlane National Park within Swaziland.

The nesting colonies are represented by black circles and are as follows: (A) Hlane National Park, Swaziland

One way to infer a metapopulation structure for this species would be through a population dynamics model. Such models require a number of parameters for them to have any predictive power, notably probability of survival and fecundity rates

The aim of this paper is to elucidate the population ecology of the marabou stork in southern Africa. It is hypothesized that survival is age dependent and that this species has a positive population growth rate. The objectives of this paper are to:

Relate fecundity of marabou storks breeding in Swaziland over a nine year period with climatic conditions,

Estimate survival rates based on resightings of tagged nestlings and free-flying marabou storks in southern Africa,

Use the breeding data and survival estimates of the breeding population in Swaziland to model the population dynamics of this species.

The collection of breeding data and the tagging of nestlings was conducted in the 16 000 ha Hlane National Park (31°53′S, 26°18′E), Swaziland. The vegetation is dominated by knobthorn

Free-flying marabou storks were captured and tagged at the Moholoholo Wildlife Rehabilitation Centre (24°30′S; 30°54′E, 600 m above sea level), north-eastern South Africa. The site is situated in low-lying savanna, to the west of the Kruger National Park.

There is no Ethics Committee at the University of Swaziland to oversee the compliance of biological studies. However, the current study was conducted under a permit from the Swaziland National Trust Commission to Ara Monadjem. As outlined in our methods, the only contact we had with the study animals was tagging 210 birds based on a standard technique applicable throughout the World. Furthermore, we did not harm or compromise the health of any species during this study.

Patagial tagging was implemented as the preferred method of colour-marking for large raptors, vultures and storks by the Birds of Prey Programme of the Endangered Wildlife Trust in 2006. An extensive review and assessment process of a range of methods was conducted over a period of 18 months before the members and associates of the Programme agreed on and approved the tagging protocol for the use of this method at its Annual Conference

Breeding marabou storks were monitored regularly between 2003 and 2011 at Hlane National Park. Originally nests were located on foot, but from 2008 we used a microlight to search for nests from the air once per year in July. Active nests (a nest on which eggs were found or adult activity was observed) were visited twice weekly until the chicks hatched or the nest failed. Fledging date was taken as the date of the last visit on which the chick was still on the nest. The mass and wing length was measured during each visit.

Free-flying birds were captured in a specially designed walk-in trap for vultures ^{st} year bird), subadult (2–4 years old) and five years old or older. Each bird was fitted with a metal ring issued by AFRING (Animal Demography Unit, University of Cape Town) and a patagial tag. Patagial tags were fitted according to the standard protocol adopted for this practice in southern Africa

A dedicated resightings programme was established using radio and television broadcasts, newspaper and magazine articles, and posters in Kruger National Park rest camps. A significant proportion of resightings was submitted by the staff at Moholoholo who kept a daily watch at the vulture restaurant, at which the marabou storks were frequently observed. Resightings were also reported inter alia by managers of other vulture restaurants, game ranchers, farmers and tourists.

Only nestlings were sexed. DNA extraction was conducted using the QiagenDNeasy® Blood and Tissue Kit. The extraction protocol as outlined in the manufacturer protocol was followed. CHD1 gene amplification was conducted using the 2550F/2718R

Fecundity is a measure of reproductive success and was defined as the number of fledglings successfully raised per pair per annum

The program MARK was used to estimate survival and recapture of marabou storks using the standard Cormack-Jolly-Seber model _{c}) _{c} was deemed the best model; where ΔAIC_{c} for any two (or more) models was <2.0, they were both deemed to be equally good.

Survival was estimated separately for the birds tagged as nestlings from those tagged as free-flying. A subset (n = 100, tagged between 2008 and 2011) of those tagged as nestlings were sexed, and were used to test for the role of sex in survival of marabou storks. Models which included sex performed poorly compared with those that did not (ΔAIC_{c}>2.9) and hence sex was removed as a factor. Subsequently, the data for the sexes of the nestlings were pooled and only age and time dependence were included.

To test for violations of the assumptions of homogeneity of survival and recaptures, GOF (goodness of fit) tests were conducted in the program Release

Leslie Matrices are used to chart the development of a population over time by separating the given group into distinct age classes, with matrix elements representing probability of survival and fecundity of these classes _{t} such that the number determined at time t+1 is given by _{t+1} = _{t}, where L represents the Leslie Matrix. This process is repeated for each time step. In this case t represents one year. This gives the following general equation:

A variation of the Leslie Matrix known as the Lefkovich Matrix was used in order to separate the storks into distinct life stages

Gx (on the subdiagonal) is the probability of surviving for a year and moving into the next stage; Px (on the diagonal) is the probability of surviving for a year and remaining in the same stage; Fx represents fecundity. The model did not take into account density dependent effects. The matrix was created in MS Excel using a variation of a template developed by Spangenberg and Jungck

The probability of remaining in an age class at next year is given by p_{i} and q_{i} is the probability of moving up an age class at next year. S_{i} is the survivor rate for i^{th} year and d_{i} is the length of time spent in this i^{th} stage (taken from

The below matrix was developed using the values from our results which corrected the original parameters for stage duration and fecundity,

This matrix has the three defined age classes i.e. the columns (there is a small non-zero probability that the birds will live past the third stage and fall into a fourth category but any individuals here have zero probability of surviving to the next time step). Subadult birds spend years two to four in their stage and adults spend from age five to 25 in their stage. This assumes a life expectancy of 25 years for wild marabou storks

Between 19 and 31 pairs of marabou storks bred annually at Hlane National Park that fledged between 11 and 43 chicks. The fecundity of marabou storks breeding at Hlane National Park differed greatly between years (

A total of 193 nestlings and 17 free-flying marabou storks have been tagged since 2005 and which have been resighted 811 and 834 times, respectively. On fledging, male marabou storks had significantly larger mean (± SE) wing lengths (663.1±4.63 mm vs 596.1±5.85 mm; t = 8.98, P<0.0001, DF = 98) and heavier mean (± SE) masses (7437±110 g vs 6155±114 g; t = 8.09, P<0.0001, DF = 98) than females. Free-flying birds had mean (± SE) wing length of 700.7±7.88 mm (n = 17), and mean (± SE) mass of 5038±291 mm (n = 13).

In the nestling analysis covering the seven years between 2005 and 2011, survival of nestlings post-fledging was age-dependent. The best model for survival had three age classes, 1^{st} year birds, 2^{nd}–4^{th} year birds and ≥5^{th}year birds (_{c}<2 compared with the best model, and had survival separated into five age classes. For both of these models recapture rates were age and time independent (_{c}>5 (

Model | AICc | Delta AICc | AICc Weights | N |

phi(age1, 2–4, ≥5) p(.) | 489.751 | 0 | 0.13917 | 4 |

phi(age1, 2, 3, 4, ≥5) p(.) | 490.1439 | 0.3928 | 0.11435 | 5 |

phi(age1, 2, 3, 4, ≥5) p(age1, ≥2) | 492.1082 | 2.3572 | 0.04282 | 6 |

phi(age1, 2, 3, 4, ≥5) p(t) | 494.1198 | 4.3688 | 0.01566 | 10 |

phi(age1, ≥2) p(.) | 494.3246 | 4.5736 | 0.01414 | 3 |

phi(t) p(.) | 495.4601 | 5.7091 | 0.00801 | 7 |

phi(age1, 2, ≥3) p(.) | 496.0157 | 6.2647 | 0.00607 | 4 |

phi(t) p(t) | 496.0794 | 6.3284 | 0.00588 | 10 |

phi(age1, ≥2) p(age1, ≥2) | 496.2424 | 6.4914 | 0.00542 | 4 |

phi(age1, 2, 3, 4, ≥5) p(.) | 497.811 | 8.0600 | 0.00247 | 5 |

phi(age1, ≥2+t) p(.) | 500.4836 | 10.7326 | 0.00065 | 11 |

phi(age1–2, ≥3+t) p(.) | 502.8863 | 13.1353 | 0.0002 | 11 |

phi(age1, 2, ≥3+t) p(.) | 505.8063 | 16.0553 | 0.00005 | 15 |

Estimates of survival (phi) and recapture (p) were modelled with time (t), and/or age class of the birds (age). Age1–5 refers to age classes of 1^{st} year birds through to 5^{th} year birds. The number of parameters is indicated by “n”. The models are arranged from best (top of table) to worst (bottom).

In the analysis of the free-flying birds, the best model had survival separated into two age classes (^{st} year birds, and subadult/adult birds (i.e. no difference in survival of subadults and adults). The next three best models all had ΔAIC_{c}<2, suggesting that they were not statistically distinguishable from the best model. In all three of these models survival was separated into three age classes. In the top three models, recapture rates were age dependent, whereas in the fourth best model recaptures were independent of both age and time (

Model | AICc | Delta AICc | AICc Weights | n |

phi(age1, ≥2) p(age1, ≥2) | 52.0191 | 0 | 0.29804 | 3 |

phi(age1, 2, ≥3) p(age1, 2, ≥3) | 52.554 | 0.5349 | 0.22810 | 4 |

phi(age1, 2, ≥3) p(age1, ≥2) | 52.5604 | 0.5413 | 0.22737 | 4 |

phi(age1, 2, ≥3) p(.) | 53.2647 | 1.2456 | 0.15988 | 4 |

phi(age1, 2, ≥3) p(t) | 59.7627 | 7.7436 | 0.00621 | 6 |

phi(age1, 2, ≥3+t) p(age1, 2, ≥3+t) | 66.2284 | 14.2093 | 0.00024 | 8 |

phi(age1, 2, ≥3 *t) p(age1, 2, ≥3 *t) | 79.696 | 27.6769 | 0 | 11 |

Estimates of survival (phi) and recapture (p) were modeled with time (t), and/or age class of the birds (age). Age1 refers to age classes of 1^{st} year birds, age2 to subadults (2^{nd} to 4^{th} year birds) and age3 to adults (≥5^{th} year birds). The inclusion or exclusion of interactions in the models are symbolized by (*) or (+), respectively. The number of parameters is indicated by “n”. The models are arranged from best (top of table) to worst (bottom).

The survival rates of marabou storks varied considerably between age classes and the datasets analysed (^{st} year birds was only 0.2500. Survival of older age classes (subadults and adults) was generally high ranging from 0.7917 to 0.8727 (^{th} year birds tagged as nestlings where survival dropped to 0.2193, suggesting loss or fading of tags.

Analysis | Estimated parameter | rate | SD |

Nestlings | Survival of 1^{st} year birds |
0.6440 | 0.0765 |

Nestlings | Survival of 2^{nd}–4^{th} year birds |
0.7917 | 0.0597 |

Nestlings | Survival of ≥5^{th} year birds |
0.2193 | 0.1459 |

Nestlings | Recapture rate | 0.4226 | 0.0523 |

Free-flying | Survival of 1^{st} year birds |
0.2500 | 0.2165 |

Free-flying | Survival of ≥ subadult birds | 0.8727 | 0.2483 |

Free-flying | Recapture of 1^{st} year birds |
0.9999 | 0.0004 |

Free-flying | Recapture of ≥ subadult birds | 0.2545 | 0.1559 |

The analysis refers to the specific dataset used for the estimates: nestlings = all nestlings tagged and resighted between 2005 and 2011; free-flying = all free-flying birds tagged and resighted between 2007 and 2011. See Methods for further details.

Stage | Fx | Px | Gx |

Juveniles | 0 | 0 | 0.644 |

Subadults | 0 | 0.587 | 0.205 |

Adults | 0.525 | 0.865 | 0.008 |

Fx = Fecundity; Px = Probability of remaining in age class at next year; Gx = Probability of moving up an age class at next year.

The resulting λ value was 1.0212 indicating a population increase over time (see

The initial population was set to 10 females in this case. λ = 1.0212.

Stage | Stable Age Distribution | Reproductive Value |

Juveniles | 0.2251 | 1.00 |

Subadults | 0.3337 | 1.59 |

Adults | 0.4378 | 3.36 |

The final population in the model is clearly dependent on the initial number of founding individuals and the fecundity of the storks. Starting with three females a total population of 26 birds is produced over the 50 year time frame; 10 females resulted in 86 birds (

This study presents the first estimates of survival for the marabou stork based on resightings of tagged individuals that also accounts for recapture (resightings) probability. A previous study estimated the survival of marabou storks based on the proportion of immature birds to adult birds, yielding survival rates of 28% for first year birds, 72% for subadults and 92% for adults ^{st} year mean survival is 64%. This is higher than survival rates for 1^{st} year white storks estimated at 47% ^{st} year wood storks at 44% ^{st} year survival based on the tagging of free-flying birds was exceptionally low (25%), lower than that reported for the white or wood storks. Undoubtedly, this estimate suffers from small sample sizes as only four free-flying 1^{st} year birds were captured and tagged compared with 193 nestlings.

Adult survival was higher, but the specific value depended on the dataset that was used. Based on the free-flying birds the survival rate was 87% and apparently did not differ between subadults and adults. Subadult (2^{nd}–4^{th} year birds) mean survival rates based on resightings of tagged nestlings was 79%, with adult (≥5^{th} year birds) survival rate dropping to 22%. This clearly is not an accurate reflection of adult survival, where captive marabou storks may live to 31 years

Survival rates of the marabou stork were not time-dependent. In contrast, the variation in survival rates of the white stork has been linked to rainfall in the Sahel where this species migrates to in the northern winter

There are a number of hypotheses that have been put forward to explain the relationship between rainfall and breeding success. One suggestion is that high rainfall impedes the formation of thermals on which the Marabous soar while they forage

There was no apparent difference in the survival of male and female marabou storks. A similar result has been reported for the white stork

Tag loss or fading is a serious violation of the assumptions of capture-mark-recapture studies ^{th} year birds). A similar resightings study of African white-backed vultures fitted with the same tags showed that they were fading and become illegible from about 5–6 years of age and older ^{th} year of a bird carrying the tags. We suggest that future studies investigate the possibility of using more durable tags for tagging of marabou storks.

Pastor

The lambda value from the matrix produced a population that either underestimated the modern day population completely or hit the lower range of the known breeding population encountered today of 19 pairs. This is with three and 10 breeding females respectively.

In general, this may be indicative of a metapopulation structure for marabou storks whereby more birds must be recruited to the local population in order to produce the numbers seen today

Specifically the lambda values reported here show that the Swaziland population of marabou storks is increasing. The metapopulation structure of marabou storks in southern Africa could explain the presence of a large population (300–400 individuals) of non-breeding birds in the Kruger National Park, South Africa

The high reproductive value of adult marabou storks is worth highlighting. This value represents the contribution that birds at any stage make to population growth. Our analysis showed that adults contributed twice as much as subadults and over three times as much as juveniles to the growth of the population. The population growth of wood storks was similarly shown to be highly sensitive to adult survivorship

We thank Mickey Reilly for permission to study marabou storks at Hlane National Park and for logistical support. Mduduzi Ngwenya, Martha Surridge, Elaine Franklin, Marie Dahl Gydesen and various All Out Africa volunteers assisted with the finding of nests and processing of the chicks. Dave Ducasse has flown various assistants annually over the study area in a microlight in search of marabou stork nests since 2008.