Compensatory Mitigation

humidity, temperature, and wind could account for variation in the spectral characteristics of Dupont’s lark song (as larks are aerial singers). However, no prior hypothesis has suggested that these could cause shifts in song-/calltype abundance, the focus of our paper, and such abiotic factors are even less likely to covary with habitat fragmentation. We are aware of the studies on natural vocal variation within bird species and we cited those that treated birdsong repertoires from a qualitative point of view (occurrence of song/syllable types rather than quantitative switches in acoustics), as these better matched the scope of our paper. We also cited studies on bird communication systems in fragmented landscapes (Laiolo and Tella 2005). Watson and Tack also addressed the possibility of founder effect, one of the hypotheses we considered. This hypothesis was discarded because there is published evidence of historically large population sizes and steppeland cover in the study area (Aragües 1992), in addition to the personal observations of one of us (JLT), who born there. For readers interested in the fate of Iberian steppe-lands, information on past and present changes can be found in Tella et al. (2005), which notes recent Dupont’s lark range restriction and local extinctions, and in Laiolo and Tella (2006a), which documents the loss of the species’ habitat throughout the Iberian peninsula. We stress that two populations went extinct during our study, and one uttered song types that were not recorded at any other Spanish site; these unique tunes were lost with the last male in the area. We did not consider the hypothesis of pre-existing historic variation because we found no evidence of dialect boundaries in a large and less fragmented geographic zone comparable in size to our study area (see Laiolo and Tella 2006b for a comparison of the Iberian Mountains and Ebro Valley). Notably, Westcott and Kroon’s (2002) study cannot be cited as an example of song dialects pre-dating isolation, as the authors themselves hypothesize a link between Prionodura newtonia song variation and recent forest clearance. Finally, Watson and Tack question the relationship between call diversity and isolation due to the influence of a supposedly outlying point (see figure). However, the relationship remains significant after this point is removed (r = –0.17, F1,21 = 4.33; P = 0.049; which, in the interest of clarity, is not a statistical outlier. We also stress that the key role of isolation was previously addressed in detail in Laiolo and Tella (2006b), where we analyzed call diversification with respect to the composition of the habitat matrix and alternative isolation indices. Finally, the relationship between repertoire and fitness was not within the scope of this paper. Paola Laiolo and José L Tella Estación Biológica de Doñana (CSIC), Sevilla, Spain ([email protected])