Ecological Archives E095-165-A2

Irit Altman, James E. Byers. 2014. Large-scale spatial variation in parasite communities influenced by anthropogenic factors. Ecology 95:1876–1887. http://dx.doi.org/10.1890/13-0509.1

Appendix B. A full description of methods to reduce metal data prior to analytical modeling of trematode prevalence and diversity. In addition, we present results of a principal compnents analysis applied to the subset of metals included in trematode models (Table B1) and comparisons of metal concentrations at sites to the informal toxicity benchmark: Effect Range Low (Table B2).

Data reduction associated with heavy metal results

We examined pairwise relationships between concentrations of metals from estuarine sediments (n = 13) and the five trematode prevalence and diversity metrics described in the main text. Results indicate only a small handful are the best predictors of trematodes. Using an R² > 0.15 (absolute value) as a minimum threshold, we identified a total of four metals that exhibited the strongest relationships with any trematode response: copper, lead, zinc, and arsenic. Considering only these four, we then examined the correlation coefficient (r) across all pairs to determine the extent of collinearity within this set. Results reveal extremely high correlations among copper, zinc, and lead (avg r = 0.97 across these pairs) with arsenic exhibiting only moderate collinearity with the others (avg r = 0.62 for all pairwise relationships that include arsenic). Based on these findings and to reduce the presence of the most strongly collinear metals, we applied principle component analysis to copper, lead, and zinc measurements. In contrast arsenic, which was only moderate collinear with the others, was retained in its measured form for use in later analyses.

Table B1. Variable loadings from principal component analysis performed on concentrations of copper, lead, and zinc which are three of the four metals found to be strongest predictors of trematodes. The first principle component explains 97.7% of the variance in these data. Principal component scores associated with this first axis (referred to as PC-Cu,Pb,Zn) were used as independent variables in multiple regression analyses.

Metal

Loadings

PC 1

PC 2

Zinc

0.98

0.19

Lead

0.99

-0.08

Copper

0.99

-0.11

 

 

 

Eigenvalue

2.93

0.05

Variance explained (%)

97.69

1.76

 

Table B2. Concentration (μg/g) of 13 heavy metals measured from estuarine sediments across salt marsh sites where Ilyanassa obsoleta trematode parasites, hosts, and other abiotic attributes were assessed. Measured concentrations of metals (μg/g) are compared to the informal benchmark, Effect Range Low (ERL), of marine sediment quality developed by Long et al. (1998). Values below ERL are rarely associated with adverse effects in marine organisms. Highlighted values are those found to exceed ERL. "na" indicates that ERL is not described for an analyte.

Site

Aluminum

Arsenic

Cadmium

Chromium

Copper

Iron

Lead

Manganese

Mercury

Nickel

Silver

Tin

Zinc

Bellamy

23000

7.6

0.3

85

17

23000

37

180

0.11

14

0.4

5

91

Bunker

23000

7.9

0.4

89

20

23000

42

200

0.13

16

0.5

6

100

Coombs

23000

4.4

0.2

37

15

22000

20

210

0.04

19

0.3

2

76

Duxbury

17000

3.8

0.4

41

15

20000

29

200

0.07

13

0.5

4

58

Freeport

21000

6.3

0.2

35

25

24000

51

220

0.03

16

0.4

3

100

Harpswell

27000

4.2

0.2

42

14

24000

19

240

0.03

22

0.3

2

76

Middle Bay

28000

5.5

0.3

45

16

26000

22

270

0.04

24

0.4

3

90

Pickering

24000

11

0.8

88

21

25000

48

230

0.25

19

0.5

5

130

Plum Island 1

14000

3.6

0.6

51

17

15000

27

150

0.08

11

0.4

5

58

Plum Island 2

79000

1.9

0.3

22

7

8700

12

99

0.03

7

0.3

3

33

Rumney 1

21000

10

2

140

86

25000

160

220

0.28

24

1.6

19

260

Rumney 2

18000

6

0.8

92

32

19000

54

190

0.08

16

0.4

5

110

Scarborough

11000

2.2

0.2

18

6

10000

10

130

0.03

8

0.3

2

32

Thompson

18000

5.1

0.3

28

22

18000

47

170

0.14

14

0.3

5

95

Vols

28000

11

0.5

110

20

29000

51

220

0.11

19

0.7

6

95

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Effect Range Low (ERL)

na

8.2

1.2

81

34

na

47

na

0.15

21

1

na

150

Literature Cited

Long, E. R., L. J. Field, and D. D. MacDonald. 1998. Predicting toxicity in marine sediments with numerical sediment quality guidelines. Environmental Toxicology and Chemistry 17:714–727.


[Back to E095-165]