Figure 2. Putative scenarios of the causation of feather holes. (A) ‘Cause–effect’ scenario: the conventional wisdom that lice chew the holes; (B) ‘mediator scenario’: parasites affect the ‘state’ of hosts, which in turn influences feather quality (cf. arrowheads with C); (C) ‘false co-variation scenario’: a third-party variable is related to both lice infestation and feather damages, but through independent routes (e.g. ‘good-quality’ birds better withstand a parasite attack and grow higher quality feathers that are e.g. less holey); (D) ‘current state-of-the-art’: compression of previous scenarios. This scenario represents our current knowledge and it is so intricate because a) the cause is unknown and might be multiple (1, 2, 3) and b) causes 1 and 2 might be related directly (black numbers) or through co-variation (type B or C, grey numbers). Note that cause 3 affects directly the damage of feathers.
Download figure to PowerPoint
First, there is a positive correlation between the number of holes and the abundance of Amblyceran lice (Møller 1991). Second, the frequency distributions of lice and holes among host individuals have a similar shape (Møller 1991) and, in populations in which hosts breed colonially, the frequency distribution of holes is less skewed (Pap et al. 2005, Vas et al. 2008). Third, lice prefer staying on white rectrix spots, which are claimed to be more easily damaged by them, and holes are more frequent on these feather parts (Kose et al. 1999). Fourth, holes are associated with the infestation by Brueelia spp. lice (Ischnocera) in comparisons of sister avian taxa (Vas et al. 2008).
These associations, however, do not substantiate the hypothesis clearly, as they are all correlative. The major concerns identified can be structured according to the following subheadings.
Identity of damaging lice taxa. Recently, Ischnoceran lice were proposed in general and the Brueelia genus in particular as hole-makers (Vas et al. 2008). The Ischnoceran fraction of the lice removed by Møller (1991) with chloroform vapour is unknown, and the denominated Amblyceran species was later discarded as a potential delinquent (Vas et al. 2008). As suggested by Clayton et al. (2010), it would be informative to repeat the studies describing the lice–hole abundance correlation found by Møller (1991) and feather preference of lice showed by Kose et al. (1999) with the new candidate genus.
Sample size, host populations and species. First, the hole–lice association was based on a small sample size (DF = 18, p < 0.01) and assessed for a single barn swallow population (Møller 1991). Given the aggregated frequency distribution of parasites, small sample sizes can be problematic and increase the probability of overlooking indirect relationships (Booth et al. 1993, Jovani and Tella 2006). Second, hole count as substitute for lice abundance was later adopted for many other swallow populations and other avian hosts (Supplementary material Appendix 1, Table A1) without verifying the hole–lice correlation. However, differences among host population and species might well exist at the level of both parties (parasite identity, infestation level and host defence, condition). It is also possible that the causative agent differs in space or among hosts (hypotheses 2 and 3).
Lack of experimental support. Correlations cannot distinguish between cause and effect. A third-party trait (e.g. condition) can mediate the co-variation between the two parameters under investigation (Vágási et al. 2011; Fig. 2B) or even co-vary simultaneously with both leading to a false correlation between the two (Fig. 2C). These cases imply the decoupling of the cause (lice) from effect (holes). If mediators such as condition play a role, the direction of correlation between two traits tells little owing to a ‘big house, big car’ paradox (van Noordwijk and de Jong 1986). This uncertainty is inherent in the discussions of several papers in which condition is identified as a potential confounder.
Free-living barn swallows treated with a general insecticide (pyrethrin) did not differ from controls in terms of changes in hole load in one month (P. L. Pap pers. comm., n = 24, p = 0.93). However, the absolute breakthrough would be to test experimentally whether feather lice puncture holes directly or their effect is manifested indirectly (Fig. 2A–C). If lice chew the holes, loss of feather mass and temporal increase in hole counts is predicted for intact feathers (in vitro) or birds (in vivo) experimentally infested by lice. Such experimentation might gain insight also into the pace at which holes are created. Sometimes the abundance of holes is relatively stable over time (Møller 1991, Vágási et al. 2011). The apparent temporal pauses in hole count accrual is surprising because Ischnoceran lice cannot cease feeding on feathers, their sole carbon source.
How to eat flight feathers? Ischnoceran lice damage only the thin barbule filaments of the plumulaceous part of down feathers, but never those at distal pennaceous part (Clayton 1990) due to the physical constraint imposed by mandible size (Bush et al. 2006, Bush and Malenke 2008). Given that the best current candidate lice Brueelia spp. are slender-bodied Ischnocerans, it seems unlikely that they have mandibles robust enough to masticate the barbules (and sometimes the barbs) of flight feathers (Fig. 1). Finally, chewing lice do not hesitate to eat almost entirely the downy barbules that they are capable to graze (Clayton 1990); hence, small feather punctures demand an explanation.
Host defence mechanisms. It has been suggested that white feathers are more susceptible to lice chewing (Kose and Møller 1999, Kose et al. 1999) because they are devoid of melanin pigments that endow feathers with strength (Bonser 1995). Although lice prefer to stay on white feather spots in vitro, lice-related damage was not assessed (Kose et al. 1999). Moreover, lice do not harm white feathers any more severely than black feathers (Bush et al. 2006), and plumage colouration does not explain the load of holes or lice (Moreno-Rueda 2005, Bush et al. 2006). Uropygial gland secretions were supposed to serve lice-deterrent function because birds with larger glands harbour fewer holes (Moreno-Rueda 2010). However, the accessibility of gland oils did not protect hosts from lice in vivo (Moyer et al. 2003) and the size of the uropygial gland was not associated with the number of Ischnoceran genera (Møller et al. 2010).