Table 3. Analyses of densities (0.1 m²) of recruits at several distances from restored (R) or reference (Ref) patches of Phyllospora (a) six or (b) 12 months after the start of the second experiment.

	(a) 6 months					(b) 12 months
Source	df	MS	F		P	df	MS	F	P
Distance	2	3327	42		<0.01	2	12.74	56	<0.01
R vs Ref	1	7651	394		<0.01	1	10.97	13	0.07
Place(RvRef)	1	14	1		0.55	2	1.26	4	0.15
D x RvRef	2	2586	33		<0.01	2	3.38	15	<0.01
Patch(Pl(RvRef))	2	28	0		0.71	3	0.33	1	0.22
D x Pl(RvRef)	2	73	Pooled			4	0.27	Pooled
D x Pa(Pl(RvRef))	4	25	Pooled			6	0.21	Pooled
Residual	60	83				84	0.22
SNK		Inside: R>C					Inside: R>C
		Edge: R>C					Edge: R>C
		2.5 m: R = C					2.5 m: R = C
		R: In > Edge >2.5 m					R: Edge > In >2.5 m
		Ref: In  =  Edge  = 2.5 m					Ref: In  =  Edge >2.5 m

Distance was fixed with 3 levels (inside, edge, 2.5 m), R vs Ref was fixed with 2 levels, Place was random nested in RvRef (1 level for R; 2 [5 mo.] or 3 [12 mo.] levels for Ref), Patch was random nested in Place (1 level for R; 2 levels for Ref). The replicates were the quadrats (n = 5). Transformation of data in (a) failed to make variances homogeneous; data in (b) were ln(X+1) transformed. Cochran's test for homogeneity of variances: a: C = 0.57, P<0.01; b: C = 0.17 ns. Non-significant terms with P>0.25 were pooled.