‘Small’ sample size scenarios in animal behaviour: alternatives to the use of null hypothesis significance testing