Influencias en la formación de la concepción del desarrollo

Summary

The annual scheduling of the replacement of flight feathers, and the length of time it takes, is generally well defined and predictable in resident, temperate bird species. In most cases, moult starts towards the end of the breeding season and must be completed before impending deterioration in local environmental conditions. This results in a moult duration (in smaller species) of <100 days and is usually finished in time to allow diversion of resources for improving body condition. In a tropical environment with reduced seasonality, these temporal constraints may be less severe allowing for an increase in the variation in order, timing and duration of moult.

I investigated the patterns of moult in a range of tropical African birds in order to make general comparisons with those of temperate species. All 44 of the species assessed followed the same basic order of moult through the flight feathers as seen for temperate species. The level of coincidental moult between wing feather tracts was much higher for tropical species, with up to five tracts moulting simultaneously in some species. This was considered to be a result of the length of time taken to complete moult, which averaged 131 ± 11 days across the 29 species assessed here and overall ranged from one to nine months.

Active remex moult was recorded in all months of the year highlighting the reduction in constraints for tropical species. Although the level of synchrony was high for certain groups (Ploceidae), timing of moult varied widely, both within and between species. In general, peak occurrence of moult occurred shortly after the peak time of breeding for many species (late rainy season). No broad differences in timing of moulting were identified across guilds

although insectivores moulted proportionally more in the wet season than granivores.

Overall this study has demonstrated the impact of reduced climate variability on the moult characteristics of birds in tropical areas. The results are discussed in the light of this difference from temperate species and the possible relationships between moult and other life history traits are examined.

Introduction

The replacement of feathers (moult) is a costly process with short-term thermoregulatory expense (Payne 1972), increased energy metabolism (King 1981; Lindström et al. 1993) and reduced flight performance (Swaddle & Witter 1997), balanced against future thermoregulation, signalling and flight performance gains. The timing of feather replacement and the length of time taken will also impact upon other essential processes (Langston & Rohwer 1996) unless it is efficiently incorporated within an annual schedule. This is especially important in temperate regions because of the strong seasonality of the climate (e.g. Cresswell & McCleery 2003; Crick 2004; Reed et al. 2006). In response, and because of the energetic costs associated with replacing feathers (Payne 1972; King 1981; Lindström et al. 1993), the replacement of feathers by individuals in temperate areas is generally scheduled to occur at the same time each year (Svensson & Hedenstrom 1999) such that there is minimal overlap with breeding (Payne 1969), migration (Holmgren & Hedenstrom 1995; Yuri & Rohwer 1997), autumnal territory defence and food storage (Jenni & Winkler 1994) or over-wintering (Barta et al. 2006). Although some annual variation will inevitably occur (due to differing weather conditions, timing of breeding and success), the time-constrained calendars of most temperate species will result in predictable and well-defined patterns of moult.

In contrast, moult strategies of tropical species are often considered more variable than those in temperate areas and less constrained by seasonal routine (Barta et al. 2006). As a result, tropical species have been found to have an array of supplemental moults not often found in temperate species (e.g. Miller 1961; Craig 1983; Brooke 1985; Herremans 2006). Further, and aside from the differences between species, there is also often assumed to be great variation and reduced synchrony in timing, duration and extent of moult the between individuals and populations in tropical areas (Jenni & Winkler 1994).

In this study I investigate the differences in moult characteristics between tropical and temperate species after first describing the characteristics of moult in a range of tropical African bird species. There have been comparatively few studies of moult in tropical birds in a life history context, despite it being an important component of avian life histories (Stiles & Wolf 1974 ; Tidemann & Woinarski 1994; Marini & Durães 2001; Pierce 2009), and even fewer for African resident birds (Payne 1969; Wilkinson 1983 ;Klaassen 1995; Oschadleus 2005; McGregor et al. 2007a). Therefore even basic descriptions of moult characteristics and phenology are needed to further our understanding of how seasonality and latitude constrains moult in birds. In this study I focussed on four specific hypotheses:

1. The sequence of moult in the wing and tail feathers will be similar in tropical and temperate species. Since the sequence of moult has

evolved in order to limit the impact of feather loss on aerial ability (Ginn & Melville 1983; Pennycuick 1975; Swaddle et al. 1996; Swaddle & Witter 1997), the requirement to maintain reasonable or good flight performance at all times of the year will result in the sequence of moult being the same for both temperate and tropical birds, i.e. for the remiges, Primaries → Secondaries → Tertials (= Inner Secondaries) and in the coverts, Primary Coverts → Greater Coverts → Carpal covert → Median Coverts → Alula.

2. Tropical species should take longer to complete moult than similar north temperate species. The degree of seasonality in north

temperate areas is more strongly defined than in most tropical areas (especially when considering the variation in daylight hours and temperature through the year). As a result, passerine species in north temperate areas are very strongly seasonal and predictable in their breeding and moulting periods. This seasonal constraint and the need to undertake a range of important behaviours (territory establishment and defence, breeding, moulting and preparation for harsh weather/migration) within an annual cycle is reflected in a comparatively short moult duration. Also, since birds induced to moult more rapidly produce feathers of reduced quality (Dawson 1998; Dawson et al. 2000) there must be an optimal duration of moult for each species in certain conditions to optimise the trade-off between feather quality, flight capability and clutch size/duration of breeding (e.g. Bensch et al. 1985; Jenni & Winkler 1994). If birds are adapted to the seasonal constraints operating in an environment then this should result in a reasonably stable moult strategy linked to that predictable seasonality. In comparison, tropical species generally have smaller clutch sizes and do not necessarily breed each year (Skutch 1949), which, because of the reduced definition in seasonality means that a greater period of time is available for the moult.

3. Pronounced seasonality should lead to a predictable pattern of timing and duration of moult for a species. Although seasonality is

present in West Africa it is not as clearly defined as in temperate areas. I therefore predicted that timing of moult in tropical species would vary in a manner similar to that for moult duration.

4. Moult speed will be determined by size and taxonomy. The feather

mass of larger and non-passerine species is, in most cases, much greater that of most passerines (Baker 1993; del Hoyo et al. 1994; Pyle 2006). Because the physiological processes necessary for feather production are energetically demanding (Walsberg 1983a; Walsberg

1983b) it follows that larger feathers take longer to produce than smaller feathers (Rohwer 1999). I therefore predicted that tropical species would show a similar relationship between size and moult duration as seen in temperate species (Ginn & Melville 1983; Jenni & Winkler 1994) and that moult duration in tropical non-passerines should be longer than passerines.

Methods Field Methods

This study was conducted primarily in the Amurum Community Forest Reserve, near Jos (09o 52’N, 08o 58’E), and Yankari State Park (09o45’N, 10o30’E), near Bauchi, both in central Nigeria, between 2001 and 2008. Additional, much smaller, datasets are also available and included from mist- netting operations in Gwafan, Jos and Pankshin (both, Plateau State). The habitat at all sites consists primarily of savannah with smaller areas of riverine woodland and gallery forest. Continuous operation of nets (i.e. nets being opened in order to catch birds) over the course of several hours during one day, are referred to here as ringing ‘sessions’ and, where several of these occurred over consecutive days, these groups of sessions are referred to as ‘periods’. Nets were operated at Amurum in all months between November 2001 and May 2008 except for June and December 2003, June-August 2004, and July 2007. One final capture period was also made here in October 2008. Catches were made at Yankari in September and April 2006, February, April and October in 2007 and January, February and March in 2008.

Capture methods

Two capture approaches were used during the course of the study, one (the CES method) following an existing protocol operating at the Amurum Community Forest Reserve since 2001 and aimed at standardising capture effort, and the second (the ad hoc approach) aimed at generating larger datasets for more widespread analyses of data for more species.

The CES method

This approach was performed at Amurum study site only. This method required the use of a standard number and length of mist netting with nets placed in specific set locations in one area of the Reserve for 14 consecutive days usually during two periods each year - during the ‘Spring’ (February- April) and ‘Autumn’ (October-December). Between 2001-2004 a third session was often included operating between June and September each year. Locations of mist nets remained the same for all capture sessions and periods. Initially, 14 four-panel mist nets totalling 187m were used, however this was increased to 20 nets (270m) in 2006 in an effort to both improve species diversity and numbers of recaptures. The nets were opened daily between dawn (usually around 06:00) and 10:30am each day, occasionally being closed earlier if weather conditions posed a risk to the welfare of birds in the net. Nets were checked every 20 minutes in order to reduce the chances of heat stress to individual birds. On occasions where rain or very strong winds prevented catching, the CES period was increased to account for lost days.

During the period of this study, the majority of the CES periods and associated data collection, were performed by two researchers (McGregor 2001-2004 and Stevens 2006-2009). During the intervening 18-month period, a further three persons operated the CES sessions.

Ad hoc method

Aside from the standardised protocol operating for the CES, mist nets were erected and operated at numerous other locations (outside of the CES capture area and capture periods) around the Amurum Reserve during the course of this study. These ringing periods usually involved the use of 100- 200m of mist net and lasted 5-10 days. Ad hoc ringing periods were also operated throughout the year across the other three study sites. Nets were opened mostly during the period between 06:00 – 10:30, however supplementary catches were also occasionally made between 16:00 –19:00 (dusk). Locations for these ad hoc sessions were rotated to improve coverage of the primary site. Data generated from these sessions is useful in

terms of increasing the number of recaptures and resightings and provides for an improved diversity of species included within the study.

The majority of the data generated under the ad hoc approach and included within this study was generated by two people (myself and McGregor) however, data from a further six people were included in the analyses. These additional data were assessed before inclusion in order to identify obvious errors or discrepancies.

Data collection

Ageing

Where possible, all birds caught were aged and sexed (according to plumage characteristics and biometrics), fitted with uniquely numbered metal rings, and had the moult status of remiges and rectrices assessed. Because the lifecycles of tropical birds rarely follow a standard annual calendar (i.e. breeding does not necessarily fall during the middle of the calendar year), it was not possible to reliably apply the methods used in temperate systems for coding age of individuals (Redfern & Clark 2001). Age was therefore assessed according to plumage characteristics Five age categories were identified: 1 – pullus; 2 – unknown or indeterminate; 3 – full juvenile plumage; 4 – full adult plumage; 5 – immature plumage. Juveniles were determined as those individuals carrying their first plumage, i.e. that grown whilst in the nest. Certain species have juvenile plumage that is markedly different from that of the adult (e.g. C.senegalensis, C.venustus and P.chrysoconus) and determination of age category for these was straightforward. For others however, assessment was based on a variety of other characteristics including general feather structure (juvenile body feathers tend to be looser textured than adult feathers and have fewer and more widely-spaced barbs (Göhringer 1951)), feather shape (Jenni & Winkler 1994; Svensson 1992), colour of bare parts (e.g. iris (e.g. in many Sturnidae and Ploceidae species (Bannerman 1948)), orbital ring (e.g. in some Sylviidae (Gargallo 1992)), wattles (e.g. in Platysteiridae (Brooke & Manson 1979; Harris & Arnott 1988)),

bill colour (e.g. in Prionopidae (Harris & Franklin 2000), the Estrildidae (Mackworth-Praed & Grant 1973) and Hirundinidae (Svensson 1992)) and tarsus and foot colour (e.g. in Sylviidae (Karlsson et al. 1988)), and presence of gape (Svensson 1992). Individuals were classified as immatures once any noticeable moult had occurred (even if just one feather), either in the body feathers or as a result of post-juvenile moult of remiges or rectrices. Conversely, adults were identified as those individuals having no trace of juvenile plumage.

Moult assessment

Moult scoring followed the system outlined by Ginn & Melville (1983) where an old feather was scored as a 0, growing and missing feathers scored between 1 and 4, depending upon length, and fully grown, new feathers, scored as 5. Moult score was therefore the sum of the scores for each tract. Occasionally some individuals lost remiges or rectrices as a result of handling. This was accounted for by recording all feathers found with the bird at the net (or in the bird bag) that could be attributed to the individual, and by identifying and noting all gaps in the plumage. Also, as each bird was fitted with a unique metal ring, it was possible to determine whether all unconventional or unusual moults were the result of feathers being lost during previous handling events. Moult of the major wing coverts (median, greater, carpal, alula and primary) was assessed using a simplified system in which feathers were scored either as old (0) or new (5). Missing or growing feathers were recorded as new. This method was used to provide a simple and rapid means of assessing the organisation and pattern of moult in the coverts. Moult of the contour feathers was assessed by scoring the number of feathers growing in the various pterylae across the body. For body moult, a score of 0 indicated that no growing feathers could be seen on quick examination of the ventral, spinal, or crural tracts, up to 10 growing feathers was represented by a score of 1, 11 to 20 feathers as a 2 and more than 20 as a 3. On the head, scores of 0 – 3 were given when 0, 1-5, 6-20 and more than 20 growing feathers were recorded in any area of the capital tract. In reality a more subjective assessment of body and head moult was performed with the

categories 0-3 referring to ‘zero’, ‘few’, ‘many’ and ‘lots/mostly’ of growing feathers, thus allowing for the differences in size between species.

Breeding assessment

Determination of breeding status was performed through examination of cloaca for seasonal enlargement (Svensson 1992) and checking for the presence of brood patches. Here tropical birds were assumed to show a progression of brood patch development similar to that in temperate species and, to prevent misidentification of breeding periods, only those individuals having brood patches defined by fully de-feathered breasts (i.e. stages 2-4 in Redfern (2008)) were considered to be actively engaged in breeding (Hinde 1962).

Other assessment

Other standard morphometrics were taken (wing length (maximum chord), mass, fat and muscle levels) according to British Trust for Ornithology protocols (Redfern & Clark 2001) to provide some measure of the variation in condition of the population at the sites and to further aid ageing and sexing.

Data analysis

Sequence of moult

Species were selected on the basis that they had a minimum of 15 individuals engaged in active moult. To ensure as many species as possible reached this criterion, data from multiple captures of the same individual were used. Although this introduces an element of pseudoreplication into the analyses it was felt that the information gained outweighed the negative statistical effects. 139 of these multiple captures were included, almost half (67) of which were for five species having some of the largest datasets (all >67 records). It is therefore unlikely that significant statistical biases were introduced for the species in the analyses. 44 Afrotropical bird species (40 of which were passerines) of 19 families were therefore assessed for moult sequence (mean = 49.9 individuals per species). The number of feathers moulting in each tract was assessed and the proportion of the tract in moult determined by dividing

the observed moult score by the maximum possible moult score for the relevant tract. Sequential pairwise correlations (two-tailed, Pearson) of the proportion of feathers in moult in each of the tracts was used to generate a correlation matrix. This allowed the identification of the general progress of moult through the wing and tail for all species.

Timing and duration of moult

Tropical and larger birds may show an increased tendency to show suspended, arrested moult or Staffelmauser in their primaries making determination of the relative ages of feathers more difficult (e.g. Pyle 2006; Symes & Wilson 2008). As interruption and resumption of moult will also result in lengthened estimates of moult duration, individuals that appeared to have more than two generations of flight feathers were omitted from the main analyses. All species for which there were sufficient data for actively moulting birds (i.e. more than 15 cases of Type 3 data in Underhill & Zucchini 1988) to plot moult score against time were included in the analyses. This further reduced the number of species assessed to 29 (4 of which were non- passerines).

Ordinary linear regression of moult score regressed on date, using only actively moulting birds, generally produces incorrect results (Pimm 1976) due to heteroscedasticity (i.e. at the start and end of moult there is less variability in moult score than during the middle period of moult). The regression line produced using this method encompasses the start and end dates of the entire population under study and not the dates for the average individual (Pimm 1976; Summers et al. 1983). Changing the dependent variable to date and regressing this against moult score appears to resolve this (Pimm 1976) however, this places constraints on the residuals in the regression model such that they are not identically distributed and do not have a mean of zero (Underhill & Zucchini 1988). This failure to meet the model assumptions will often result in bias in parameter estimates in particular the start date (underestimated) and the duration (overestimated) (Underhill & Zucchini 1988. To counter this, species were selected for inclusion in the analysis on the

basis of their having sufficient numbers of type 3 data (i.e. only birds actively moulting). The proportion of primaries in moult (i.e. observed moult score/50) was then plotted against date. In many cases the reference date (start date of the plot) was adjusted so that low moult scores (i.e. the start of moult) occurred at the start of the plot. Where possible this was identified from natural breaks in the data. This ensured that all plots showed a progressive increase in moult score through time and within the 12-month (or shorter) period shown by the plot. Extreme outliers that indicated likely mistakes in scoring were removed. The residuals of these new data were then assessed