Ecological Archives E084-072-A1

K. M. Jenkins and A. J. Boulton. 2003. Connectivity in a dryland river: short-term aquatic microinvertebrate recruitment following floodplain inundation. Ecology 84:2708–2723.

Appendix A. The statistical design for ANOVA showing factors, degrees of freedom, multipliers, and formulas to determine estimated mean squares, variance components, and F-ratio tests. A pdf version is also available.

Factor R has i =…a levels; factor L(R) has j =…b levels; factor S(L( R)) has k =…c levels; factor FT has l =…d levels and FD has m =…e levels.  All combinations of between sample effects are replicated n times (r =…n).  All combinations of within sample effects are replicated N times (s =…N).  There are three orthogonal factors and 2 nested factors.  Factors Reach (R), Flood Type (FT) and Flood Day (FD) are fixed and Lakes within Reaches (L(R)) and Sites within (L(R)) are random.  Multipliers and formula’s to determine estimated mean squares (E(MS)) and variance components, denominators for F-ratio tests (den.) and degrees of freedom (df) are shown.

Xijkl= m + Ri + L(R)j(i) + S(L(R))k(j(i))  + FTl + Ri x FTl + FTl x L(R)j(i) + FTl x S(L(R))k(j(i))  + E r[j(i)kl]

Xijklm= m + FDm + FDm x  Ri + FDm x  L(R)j(i) + FDm x  S(L(R))k(j(i))  + FDm x  FTl + FDm x  Ri x FTl + FDm x  FTl x L(R)j(i) + FDm x  FTl x S(L(R))k(j(i))  + E s[mj(i)kl]

Source of variation

Multipliers

E(MS)

Den.

df

Variance component

Between samples

i

j

k

l

r

-

-

 
 

1. Ri

1

b

c

d

n

-

-

2e+ dn2S(L(R))+ cdn2(L(R))+bcdn2R

2

1,2

2R = [MSR- MSL(R)  ] / bcdn

2. L(R)j(i)

1

1

c

d

n

-

-

2e + dn2S(L(R))+ cdn2(L(R))+

3

2,8

2L(R) = [MSL(FH)- MSS(L(R)) ] / cdn

3. S(L(R))k(j(i))

1

1

1

d

n

-

-

2e + dn2S(L(R))

8

8,96

2S(L(R)) = [MSS(L(R))- MSe ] / dn

4. FTl

a

b

c

0

n

-

-

2e + n2FTS(L(R))+ cns2FTL(R)+ bcn2FTR+ abcns2FT

5

1,1

2FT = [MSFT- MSFTR ] / abcn

5. FT x Rli

1

b

c

0

n

-

-

2e + n2FTS(L(R))+ cns2FTL(R)+ bcn2FTR

6

1,2

2FTR = [MSFTR  - MSFTL(FH)] / bcn

6. FT x L(R) lj(i)

1

1

c

0

n

-

-

2e + n2FTS(L(R))+ cns2FTL(R)

7

2,8

2FTL(R) = [MSFTL(R) - MSFTS(L(R))  ] / cn

7. FT x S(L(R)) lk(j(i))

1

1

1

0

n

-

-

2e + n2FTS(L(R))

8

8,96

2FTS(L(R)) = [MSFTS(L(R)) – MSe] / n

8. e r[ijkl]

1

1

1

1

1

-

-

2e

2e =  MSe

Within samples

i

j

k

l

-

m

 
 

9. FDm

a

b

c

d

-

1

N

2E  + dN2FDS(L(R))+ cdNs2FDL(R)+bcdN2FDR+abcdN2FD

10

3,3

2FD = [MSFD - MS FDR ] / abcdN

10. FDxRi

0

b

c

d

-

1

N

2E  + dN2FDS(L(R))+ cdNs2FDL(R)+bcdN2FDR

11

3,6

2FDR = [MSFDR - MS FDL(R) ] / bcdN

11. FDxL(R) mj(i)

1

1

c

d

-

1

N

2E  + dN2FDS(L(R))+ cdNs2FDL(R)

12

6,16

2FDL(R) = [MS FDL(R) - MS FDS(L(R)) ] / cdN

12. FDxS(L(R)) mk(j(i))

a

b

0

d

-

1

N

2E  + dN2FDS(L(R))

17

24,288

2FDS(L(R)) = [MSFD S(L(R)) - MS E ] / dN

13. FDxFTml

a

b

c

0

-

1

N

2E  + N2FDFTS(L(R))+bN2FDFTL(R)+bcN2FDFTR+abcN2FDFT

14

3,3

2FDFT = [MSFDFT - MS FDFTR ] / abcN

14. FDxFTxRmli

0

b

0

d

-

1

N

2E  + N2FDFTS(L(R))+ bN2FDFTL(R)+ bcN2FDFTR

15

3,6

2FDFTR = [MSFDFTR- MS FDFTL(R)] / bcN

15. FDxFTxL(R) m lj(i)

0

b

c

0

-

1

N

2E  + N2FDFTS(L(R))+ bN2FDFTL(R)

16

6,24

2FDFTL(R) = [MSFDFTL(R) -MSFDFTS(L(R))] / bN

16. FDxFTxS(L(R)) m lk(j(i))

a

b

0

0

-

1

N

2E  + N2FDFTS(L(R))

17

24,288

2FDFTS(L(R)) = [MSFDFTS(L(R)) - MS E ] / N

17. E s[mijkl]

1

1

1

1

-

1

1

2E

2FD = MSE

 



[Back to E084-072]