I'd always thought you couldn't say that something was "close to significant" like this. And the results mean it is insignificant so to be highlighting hairdressers regardless has me a little concerned.After excluding correlated predictors, we ran generalized linear models and applied a stepwise selection procedure of variables based on variance inflexion factors (see Material and methods for details). We identified noise pollution and air pollution as the main predictors of mutilation occurrence (Table1). No other predictors were significant, though the quadratic term of hairdresser density was close to significance (P=0.075, Fig. 2c). [my emphasis, p4]
The authors also point out that,
It's important research and I like the point that "There is no ethical reason for accepting that pigeons should suffer from mutilation due to human development without trying to reduce their pain" though I could have done without the highlighting of hairdressers without more robust evidence.there are also more hairdressers where human population density is higher (r=0.57 see Fig.S2 in Supplementary material), leading to a potential indirect effect of human population density [p4]
P.S. The paper is behind a paywall but if anyone wants a copy DM me.