Fault bars timing and duration: the power of studying feather fault bars and growth bands together Roger Jovani and Javier Diaz-Real R. Jovani (jovani@ebd.csic.es) and J. Diaz-Real, Estación Biológica de Doñana, Consejo Superior de Investigaciones Científicas (CSIC), Américo Vespucio s/n, ES-41092 Sevilla, Spain. Growth bands and fault bars, widespread features of feathers that form during regeneration, have largely been studied independently. Growth bands result from normal regeneration: each pair of dark/light bands forms every 24 h. Fault bars are a response to stress during regeneration, creating a translucent line that can break the feather. We studied the relative position and width of these two structures in feathers of nestling and adult white storks Ciconia ciconia. We first confirmed that one growth band represents 24 h of feather regeneration. Fault bars did not occur at random within growth bands: 65.7% (in nestlings) and 45.6% (in adults) of them occurred in one out of six defined segments within a growth band, namely that segment corresponding to the first one-third of night time hours. The width of fault bars relative to growth bands suggested that fault bars were produced during a median (range) of 7.0 h (2.7–27.0) in nestlings and 3.7 h (1.8–7.9) in adults. Fault bars were concentrated at feather tips in nestlings, but at central locations in adult feathers. Our results suggest that, in general, fault bars are a discrete event of a finite duration occurring mainly during the night (particularly in nestlings). This, along with current knowledge, suggests that acute stressors, rather than chronic ones, are responsible for fault bar formation. Thus, such acute punctual stressors (a matter of minutes) can have long-lasting (months–years) physiological effects due to the wing load increase from feather breakage caused by fault bars. Growth bands are pairs of subtle, alternating, dark/light bands perpendicular to the feather axis along the entire length of a feather with an appearance similar to watermarks (Riddle 1907, Michener and Michener 1938, Grubb 2006, Jovani et al. 2011). A growth band forms every 24 h (less in some cases; Results) during normal feather regeneration. Despite a lack of a 1:1 correspondence between photoperiod and the light/dark width ratio, the light band is produced during the night, and the dark band during the day (Jovani et al. 2011 and references therein). This circadian origin of growth bands suggests their link to bird body condition: the better the bird’s nutritional status, the wider the growth bands (Riddle 1908, Wood 1950, Grubb 1989, 2006). Fault bars are also widespread in feathers, but indicate abnormal regeneration (dysmorphogenesis). Like growth bands, they lie perpendicularly to the feather axis. They can be as minor as a slight notch on the feather surface (‘light’ intensity) or so severe as an absence of barbules allowing one to see through the feather (‘medium’ and ‘strong’ intensities, following Sarasola and Jovani 2006). All hypotheses attempting to explain fault bar origin consider some stressor as the cause. However, they differ on the nature of the stressor. Malnutrition was independently proposed by Riddle (1908) and Michener and Michener (1938) as the cause of fault bars. This idea held until Murphy and King (1982) did detailed experiments with synthetic food, showing that malnutrition, per se, did not produce fault bars, but that bird handling during the experiments caused the fault bars (King and Murphy 1984, Murphy et al. 1988). More recent studies also have excluded malnutrition as a cause of fault bars. An experiment with American kestrel Falco sparverius nestlings showed that food deprivation produced no increase in fault bar number, but nestling handling did (Negro et al. 1994). Providing captive European starlings Sturnus vulgaris with food and water ad libitum, but changing the availability of dense shrubs for hiding produced an increased abundance of fault bars in aviaries without shrubs, also challenging the malnutrition hypothesis (Witter and Lee 1995). However, despite human handling and failed predation attempts occur in the wild, it seems highly unlikely that they may account for the widespread occurrence of fault bars in wild birds. Therefore, a general explanation of the causes producing fault bars remains elusive. Whatever the cause of fault bars what we do know is that fault bars often result in partial feather damage or even complete breakage (Sarasola and Jovani 2006). Broken feathers are not replaced until the next moult (as opposed to plucked feathers, the replacement of which is initiated immediately), and experimental reduction of wing area is known to increase energetic demands resulting in direct fitness costs as reflected in lower reproductive success 97 (Mauck and Grubb 1995, Velando 2002). Therefore, understanding when and for how long are fault bars produced is a missing piece of knowledge with implications for our understanding of bird biology and on the potential implications of fault bars in bird plumage evolution. In summary, growth bands and fault bars in feathers reflects normal and abnormal feather regeneration, respectively. Until now, the two have been studied as independent sources of information: growth bands as proxies of feather growth rate (Grubb 2006), and fault bars about stressors (Bortolotti et al. 2009). Also, studies trying to understand the very nature of these structures have focused on one or the other structure (Murphy and King 1991, Jovani et al. 2011), but rarely on both (Riddle 1907, 1908). We propose that studying growth bands and fault bars together could provide a new way to study the potential intimate relationship between the two (Riddle 1907, 1908), answer basic questions on fault bar formation, and provide new information on avian biology. Specifically, our questions were 1) do fault bars occur randomly within growth bands, and if not, when are they most abundant within the circadian formation of growth bands? 2) When are fault bars produced during feather regeneration? 3) How wide are fault bars relative to growth bands? 4) If phenomena in these contexts differ between nestling and adult birds. When are fault bars formed? We cut off the third scapular feather from a sample of 100 nestlings. Also, during our incursions into the colony in June 2003, we collected from the ground freshly moulted primary, secondary and scapular wing feathers. These were considered to be adult feathers (see detailed justification in Jovani and Blas 2004). In the laboratory, we selected feathers having at least one fault bar within a conspicuous growth band (i.e. sometimes growth bands were difficult to detect in some feather portions). To quantify this, we partitioned the growth band into six imaginary, equallysized, arbitrary segments perpendicular to the longitudinal feather axis (Fig. 1). The first feather portion grown is the tip. Within growth bands, the dark section is produced during the day and the light section during the night (Introduction). Therefore category 4 (Fig. 1) roughly corresponds to the hours after sunset (Riddle 1907, 1908). (a) Growth band 1 2 3 4 5 6 Methods Does one growth band equal 24 h? We tested whether white storks conform to the general rule (Jovani et al. 2011) that one growth band is produced every 24 h during feather regeneration. This was an important prerequisite to the study as a few exceptions to the rule have been found (Langston and Rohwer 1996, Kern and Cowie 2002). On 3 June 2006 we measured the third scapular (see a detailed schematic of feather position in Jovani and Blas 2004) of nine nestlings with a ruler (nearest 1 mm). We repeated the measurement at the same time eight days later and then cut off the feather at its base. In this way, we calculated the average daily growth rate of the feather as (second measure – first measure)/8. Later in the laboratory, we measured a median (range) of 4 (2–6) growth bands with a caliper (nearest 0.01 mm) directly from the feather as closely as possible to the cut base to compare the mean width of the growth bands with the daily feather elongation. Our expectation was that the width of the growth bands was equal to the daily growth rate of feathers. 98 Fault bar Feather growth direction (b) sunset 75 % fault bars We employed feathers from nestling and adult white storks Ciconia ciconia, an especially suitable species by virtue of its long feathers with abundant fault bars (Jovani and Blas 2004) and conspicuous growth bands of considerable width. Field work was performed during the 2003 breeding season at the Dehesa de Abajo site (Puebla del Río, Sevilla, SW Spain; 37°13′N, 6°09′W), where ca 250 pairs of white storks nest colonially on wild olive trees (see further details in Jovani and Blas 2004, Jovani and Tella 2004). 50 25 0 1 2 3 4 5 6 Position within growth band Figure 1. (a) Diagram showing the six arbitrary segments designated to record the position of fault bars within growth bands. A pair of dark-light bands comprises a growth band. Because feathers grow from the tip to the proximal feather section, and the light sections are produced during the night, a fault bar in growth band positions 1–3 is produced during the day, and one occurring in position 4 would be produced in the first hours of the night. In (b) it is shown the proportion of fault bars occurring in each of the six arbitrary segments within growth bands of nestlings (white bars) and adults (black bars). Since we have previously shown (Jovani et al. 2011) that the ratio of dark/light width is independent of photoperiod, the transition between the light and the dark portion is more difficult to allocate to a particular moment of the day. When a fault bar occupied more than one category we recorded the category best describing the midpoint of the fault bar. We inspected feathers by holding them by the calamus with the dorsal side inclined at ca 20–45° towards natural indirect sunlight. Fault bars were identified and classified as light, medium or strong following Sarasola and Jovani (2006). Independent of fault bar strength, we selected examples for which we could record fault bar position within the growth band with confidence. We found at least one such fault bar in 55 of the 100 nestling feathers. In nestlings, 83.3% of fault bars were scored medium or strong (n 114 fault bars, n 55 feathers), but only 29.3% were so in adults (n 58 fault bars, n 26 feathers). In nestlings, fault bar intensity was not related to position within the growth band (PROC FREQ, c2 10 11.00, p 0.357, n 105 fault bars); there was not a large enough sample size to test this in adults. Also, the proximity of the fault bar to the feather tip was not related to its position within the growth band (PROC CORR, Spearman r 20.151, p 0.125; n 105 fault bars). Therefore, we pooled fault bars of different strengths when analysing their position. Overall, we located the position within growth bands of 105 fault bars from nestlings (n 55 feathers) and 57 fault bars from adults (n 26 feathers). Moreover, from the 55 nestling feathers we calculated feather length and the distance of each fault bar to the tip (whether or not the fault bar was used to study its position within the growth band). The same analysis was done for 36 adult scapular feathers. We divided each feather into 25 mm sections starting from the tip (i.e. 0–25, 26–50, etc.). For each section, we calculated the proportion of feathers with at least one fault bar in the section. This variable was compared between ages along the feather with a generalized linear model with binomial probability distribution (PROC GENMOD). We also tested whether the number of fault bars differed between ages and feather sections with a generalized linear model with Poisson probability distribution (PROC GENMOD). Only feathers long enough to be contributing to a given feather section were used in the calculations; i.e. a 0 meant that the feather had no fault bar in the section, given that the feather was long enough to potentially have a fault bar in the section. For how long do fault bars form? From these same feathers (cut scapulars from 100 nestlings and adult feathers collected from the ground) we selected mediumstrong fault bars (following Sarasola and Jovani 2006) to measure the time elapsed during fault bar formation because they reflect a clear period of dysmorphogenesis. Because light fault bars are merely a simple pinch on the feather surface, comparing their width with that of medium-strong fault bars would be misleading. Moreover, we only studied fault bars that had at least one growth band nearby easy to measure and representative of the feather portion in which the chosen fault bar occurred. Since we confirmed that white storks follow the general rule that one growth band corresponds to 24 h of feather regeneration (Results) we estimated the number of hours elapsing during fault bar formation as (maximum fault bar width/mean growth band width) 24 h. To do so, we measured the maximum width of these fault bars along the long feather axis as well as the mean width of adjacent growth bands with a calliper (nearest 0.01 mm). Overall, for 50 nestling feathers we measured widths of 95 fault bars and a median (range) of 3 (1–8) growth bands around them. The same was done for 17 fault bars from 14 adult feathers; median (range) growth bands measured 3 (2–6). SAS 9.2 (SAS Inst.) was used for all analyses (used SAS procedures are indicated at first appearance). We used unequal variance t-tests following Ruxton (2006) when appropriate. Results Does one growth band equal 24 h? For the third scapular of nine nestlings the mean (SD, range) growth rate was 5.6 (1.2, 4.0–7.1) mm d21, and the mean width of growth bands was 6.0 (1.1, 3.7–7.6) mm d21. That is, mean growth band width was only 0.4 mm wider than the growth rate of the feather, a non-significant difference (PROC TTEST, two independent samples Satterthwaite t-test, t15.97 20.89, p 0.386), and there was no systematic difference between growth band width and daily feather growth rate within feathers (paired t-test, t8 0.90, p 0.393). The alternative hypothesis would be that two growth bands (two pairs of light-dark growth bands) are produced every day, according to previous findings in laysan albatross Phoebastria immutabilis (Langston and Rohwer 1996) and nestling pied flycatchers Ficedula hypoleuca (Kern and Cowie 2002). In our specific case, the average growth band would be 5.6/2 2.8 mm wide, but growth bands clearly departed from this ‘one growth band 12 h’ hypothesis (one-sample t-test, t8 8.75, p 0.0001). When are fault bars formed? Fault bars were found in any position within growth bands, but the first third of the light portion accounted for 65.7% of the fault bars in nestlings (c2 5 185.46, p 0.0001) and 45.6% in adults (c2 5 38.05, p 0.0001; Fig. 1); note that under a perfect even distribution only 16.6% of the fault bars would occur in this position. More generally, fault bars were highly concentrated in the light portion of the growth bands of nestling feathers (86.7%; c 2 1 56.47, p 0.0001), while a slighter bias occurred in adults (64.9%, c2 1 5.07, p 0.0243; Fig. 1); note that an even distribution would yield 50%. Fault bars were more concentrated towards the feather tip in nestlings, while the probability of harbouring a fault bar was higher at the central sections of the adult feathers (PROC GENMOD, feather section age: c29 46.68, p 0.0001; Fig. 2a). Very similar patterns were found for the number of fault bars per feather section (feather section age: c2 8 85.33, p 0.0001; Fig. 2b). For how long do fault bars form? Nestling fault bars were estimated to be growing during a median (range) of 7.0 h (2.7–27.0), slightly departing from 99 25 0.8 20 0.6 15 0.4 10 0.2 0.0 (b) 2.5 Mean number of fault bars 2.0 5 0 10 1.5 8 1.0 6 0.5 Adults 4 226–250 201–225 176–200 151–175 126–150 101–125 51–75 76–100 0–25 26–50 0.0 Feather section (mm; 0=tip) Figure 2. Occurrence of fault bars at different times during feather regeneration (the tip is the first part of the feather to regenerate). Proportion (SE) of feathers harbouring at least one fault bar in each feather section (a). Mean (SE) number of fault bars in each feather section (b). In both panels, only feathers contributing to each feather section (not all feathers were of the same length) were used for the calculations. The number of nestling feathers (white dots; from left to right) was: 55, 55, 55, 52, 50, 46, 40, 22, 5, and 2; adults (black dots): 36, 36, 36, 36, 36, 35, 22, 8, 1, and 1. a normal frequency distribution (PROC UNIVARIATE, Kolmogorov–Smirnov D 0.103, p 0.014; skewness 2.36; Fig. 3). Excluding the largest value, the distribution did not depart from normal (Kolmogorov–Smirnov D 0.080, p 0.148; skewness 0.66). Adult fault bars showed a much reduced range of values, with a median (range) of 3.7 h (1.8–7.9) and did not depart from a normal distribution (Kolmogorov–Smirnov D 0.185, p 0.122; skewness 0.673; Fig. 3). Nestling fault bars were wider than those of adults (Satterthwaite t-test, t37.123 5.65, p 0.0001). The same conclusion was obtained when only scapulars were compared (Satterthwaite t-test, t27.098 4.74, p 0.0001). Discussion Fault bars did not occur randomly within growth bands, but were highly concentrated at the beginning of the formation of the light growth band (during the first hours of the night according to current evidence, see Introduction). This could be a simple by-product of the two (the formation of the fault bar and the beginning of the 100 Nestlings 30 Number of fault bars Proportion of feathers with at least one fault bar (a) 1.0 2 0 0 4 8 12 16 20 24 28 Growth time (hours) Figure 3. Fault bar duration (hours) in nestlings and adults. formation of the light growth band) being produced simultaneously because of a third causative factor, or it could denote an intimate developmental connection between them. Interestingly, one hypothesis posed by Riddle (1907, 1908) suggests that fault bars (particularly his ‘types one and three’ defects) are the extreme manifestation of the same underlying phenomenon that also creates growth bands (his ‘fundamental bars’), namely, the circadian fluctuations of blood pressure. As the lowest pressure in birds was said to occur in the first hours of the night, if it was especially low on a particular night, this would produce follicular collapse and thus a fault bar as the result of defective keratinization. Our results partially support this hypothesis suggesting a developmental connection between growth bands and fault bars. However, frequent fault bar occurrence in locations other than those predicted by Riddle’s hypothesis suggests that other processes could be involved. In any case, our approach comparing growth bands and fault bars promises to be a suitable way to better understand the potential connection between the two, and suggests that a good starting point would be to investigate why (developmentally, physiologically, behaviourally, ecologically) the first hours of the night are so prone to the creation of fault bars. The relatively good fit of time taken to form a fault bar to a normal frequency distribution (if we exclude the single exceptionally wide fault bar; Fig. 2) suggests that it is a discrete phenomenon. This is interesting because stressors for birds can span from seconds (escaping from a predator) or minutes (a severe territorial fight) to hours (an intense storm), days (the acute stage of parasite infection), or more. However, our study shows that fault bars either occur or do not occur, and when they occur they are of a specific size range, showing a definite mean (Fig. 2). Captive starlings exposed to 30 min of chronic stress (including restraint in cloth bag, loud music, human voice, cage moving) four times a day for 20 d, or the same chronic stress coupled with food deprivation, showed fewer fault bars than birds exposed to an acute punctual stress consisting of 30 min of restraint in a cloth bag every other day (Strochlic and Romero 2008). Another study showed that corticosterone levels were higher in feather locations containing a fault bar than in contiguous locations lacking such (Bortolotti et al. 2009). Overall, the evidence suggests that fault bars are not the result of a chronic stress, but that of an acute stressor. It could thus be speculated that fault bar strength (reflected in their width) is not related to the duration of the stress, but rather, to the intensity of the stress (or the intensity of the response of the bird to the stress). Broken feathers by fault bars are not replaced until the next moult (Introduction). Therefore, an acute stress lasting minutes, or even seconds, could translate into long-lasting physiological costs over many months or years if this leads to the formation of a fault bar. This shows the relevance of further investigating the relationship between stressors and fault bar formation, and of using the position of fault bars within growth bands to understand which are these punctual stressors producing long-lasting flight costs in wild bird. Fault bars were concentrated in the first third of the light growth band (Fig. 1). Moreover, we have confirmed here the ‘one growth band 24 h’ hypothesis, and current knowledge strongly suggests that the light portion is produced during the night (see details in Jovani et al. 2011). Therefore, fault bar formation was likely concentrated during the first hours of the night both in nestlings and adults, but particularly so in nestlings. More generally, fault bars accumulated more often during the night than during the day (particularly in nestlings). Moreover, nestling fault bars were wider than in adults. This extends previous studies of the same population where we found that a) fault bars were more abundant in nestlings, and that b) nestling mortality was especially high during the first days of life, sharply declining at older ages when nestlings are able to thermoregulate (Jovani and Blas 2004, Jovani and Tella 2004). Interestingly, fault bars did not concentrate at the tips of adult feathers. This shows that accumulation of fault bars on the tips of nestling feathers (when they are youngest) is not an artefact implying feather weakness in this feather section and/ or feathers being more prone to dysmorphogenesis. Rather, this suggests a real connection between stress and fault bar origin. However, the peak of fault bars in the mid-sections of the adult feathers is an unexpected result demanding further study. In conclusion, we have shown that studying the relative position and width of fault bars in the context of growth bands is valuable in understanding the fundamental nature of both and can provide new interesting information about avian biology. Acknowledgements – We are grateful to José Luis Tella for his support. Julio Blas, Joaquín Lamas, Judit Smits, Gary Bortolotti, José Luis Tella, Martina Carrete, Raquel Baos, Isa Luque, Mariola Martín, MaCarmen Roque, Fatima León, Ernesto Fedriani and especially Juan Manuel Terrero were of great help in the field. Paul F. A. Maderson made interesting suggestion to an earlier draft of the manuscript. RJ is supported by a Ramón y Cajal research contract (RYC-2009-03967) from the Ministerio de Ciencia e Innovación. RJ wants to dedicate this paper to the memory of Gary Bortolotti. JD-R did this work during a stay at the Estación Biológica de Doñana (CSIC) under an internship program agreement with the Univ. de Salamanca. References Bortolotti, G. R., Marchant, T., Blas, J. and Cabezas, S. 2009. Tracking stress: localization, deposition and stability of corticosterone in feathers. – J. Exp. Biol. 212: 1477–1482. Grubb Jr, T. C. 1989. Ptilochronology: feather growth bars as indicators of nutritional status. – Auk 106: 314–320. Grubb Jr, T. C. 2006. Ptilochronology: feather time and the biology of birds. – Oxford Univ. Press. Jovani, R. and Blas, J. 2004. Adaptive allocation of stress-induced deformities on bird feathers. – J. Evol. Biol. 17: 294–301. Jovani, R. and Tella, J. L. 2004. Age-related environmental sensitivity and weather mediated nestling mortality in white storks (Ciconia ciconia). – Ecography 27: 611–618. Jovani, R., Blas, J., Navarro, C. and Mougeot, F. 2011. Feather growth bands and photoperiod. – J. Avian Biol. 42: 1–4. Kern, M. D. and Cowie, R. J. 2002. Ptilochronology proves unreliable in studies of nestling pied flycatchers. – Ibis 144: 23–29. King, J. R. and Murphy, M. E. 1984. Fault bars in the feathers of white-crowned sparrows: dietary deficiency or stress of captivity and handling? – Auk 101: 168–169. Langston, N. E. and Rohwer, S. 1996. Molt-breeding tradeoffs in albatrosses: life history implications for big birds. – Oikos 76: 498–510. Mauck, R. A. and Grubb, J. 1995. Petrel parents shunt all experimentally increased reproductive costs to their offspring. – Anim. Behav. 49: 999–1008. Michener, H. and Michener, J. R. 1938. Bars in flight feathers. – Condor 40: 149–160. Murphy, M. E. and King, J. R. 1982. Semi-synthetic diets as a tool for nutritional ecology. – Auk 99: 165–167. Murphy, M. E. and King, J. R. 1991. Ptilochronology: a critical evaluation of assumptions and utility. – Auk 108: 695–704. Murphy, M. E., King, J. R. and Lu, J. 1988. Malnutrition during the postnuptial molt of white-crowned sparrows: feather growth and quality. – Can. J. Zool. 66: 1403–1413. Negro, J. J., Bildstein, K. L. and Bird, D. M. 1994. Effects of food deprivation and handling stress on fault-bar formation in nestling American kestrels. – Ardea 82: 263–267. Riddle, O. 1907. A study of fundamental bars in feathers. – Biol. Bull. 14: 165–174. Riddle, O. 1908. The genesis of fault-bars in feathers and the cause of alternation of light and dark fundamental bars. – Biol. Bull. 14: 328–370. Ruxton, G. D. 2006. The unequal variance t-test is an underused alternative to Student’s t-test and the Mann–Whitney U test. – Behav. Ecol. 17: 688–690. Sarasola, J. H. and Jovani, R. 2006. Risk of feather damage explains fault bar occurrence in a migrant hawk, the Swainson’s hawk Buteo swainsoni. – J. Avian Biol. 37: 29–35. Strochlic, D. and Romero, L. 2008. The effects of chronic psychological and physical stress on feather replacement in European starlings (Sturnus vulgaris). – Comp. Biochem. Physiol. A 149: 68–79. Velando, A. 2002. Experimental manipulation of maternal effort produces differential effects in sons and daughters: implications for adaptive sex ratios in the blue-footed booby. – Behav. Ecol. 13: 443–449. Witter, M. S. and Lee, S. J. 1995. Habitat structure, stress and plumage development. – Proc. R. Soc. B 261: 303–308. Wood, H. B. 1950. Growth bars in feathers. – Auk 67: 486–491. 101