Durbin-Watson test vs Breusch-Godfrey test

I performed DW test (DW=4) and Godfrey test (Godfrey=4) on a small sample size (60 observations). The DW test results accept the null hypothesis (4 Pr>W value are all >0.05). However, for Godfrey test, only AR(1) accept the null hypothesis (Pr>LM value >0.05). AR(2)-AR(4) all have very small P value. What does this mean? Will the small sample size impact the Godfrey test results? Can anyone help me out here? Very appreciated!