The resistor from the + input to ground will act to equalize the errors due to bias current (Ib). Bias current flows through both the feedback resistor and the resistor from the - input to ground (through the impedance of the input source), and generates a small voltage. If there is no resistor present at the + input, that voltage due to the bias current produces an error at the output.
By placing a resistor from the + input to ground, the bias current flowing from the + input generates a voltage across the resistor equal to that in the negative side and the error voltage cancels out at the output.
The ideal value for the resistor is: the parallel combination of the feedback resistor, and the sum of the input resistor with the source input impedance. Any value other than zero-ohms will reduce the bias current error, though (as is the case in your example).