[R-sig-ME] Interpretation of nonlinear mixed-effects modeling results

[Gang Chen]

I'm fitting a nonlinear mixed-effects model to some data with two
groups (controls and patients) with something like

fm <- nlme(response ~  myFunc(time, a, b), data=myData, fixed = a + b
~ group, start=...)

myFunc is a nonlinear function defined with two parameters a and b.
I'm very confused with the results between summary(fm) and anova(fm):

> summary(fm)

...
Fixed effects: a + b ~ group
                        Value       Std.Error   DF     t-value      p-value
a.(Intercept) 29.905889 10.532769 2196  2.839319  0.0046
a.groupPat     6.437218 16.045223 2196  0.401192  0.6883
b.(Intercept)  0.290943  0.072544 2196  4.010559  0.0001
b.groupPat    -0.138361  0.077339 2196 -1.789010  0.0738
...

> anova(fm)
                 numDF denDF  F-value p-value
a.(Intercept)     1  2196 497.8594  <.0001
a.group           1  2196  12.6109  0.0004
b.(Intercept)     1  2196  45.2787  <.0001
b.group           1  2196   3.2006  0.0738

If I understand it correctly, the last row in the fixed effects table
of summary(fm) is the difference in parameter b between the two
groups, and the t-statistic (and p-value) matches the F-statistic (and
p-value) from the last row of anova(fm): (-1.789010)^2 = 3.2006.
However, I'm totally at a loss for the other three rows in the two
tables? For example, I thought a.groupPat (2nd row) in the summary(fm)
table is the amount in parameter a in Patient group that is more than
parameter a in the Control group (1st row); but this interpretation is
not consistent with what is shown in the 2nd row of anova(fm) table.
What am I missing here?

Thanks,
Gang