Slow explorers take less risk: a problem of sampling bias in ecological studies