Ecological Archives E095-170-A7
Chelsea L. Wood, Stuart A. Sandin, Brian Zgliczynski, Ana Sofía Guerra, Fiorenza Micheli. 2014. Fishing drives declines in fish parasite diversity and has variable effects on parasite abundance. Ecology 95:1929–1946. http://dx.doi.org/10.1890/13-1270.1
Appendix G. Multivariate analysis.
Methods
To confirm the findings of the abundance and diversity analyses described in the main text, we also analyzed the data with a multivariate approach, using non-parametric statistics based on the Bray–Curtis dissimilarity index. Because many host species did not share any parasite species in common (and were therefore infinitely different in parasite species composition), each host species was analyzed separately. Host individuals carrying no parasites were excluded, since a fish with no parasites is undefined in parasite-species space. We used both 2- and 3-dimensional configurations to find an optimal ordination for each host species, as assessed by stress value. To equalize the weight of common and rare parasite species in the index, we performed a fourth-root transformation of parasite species abundances (Field, Clarke et al. 1982). We then developed a non-metric multidimensional scaling (MDS) plot to visualize ranked within-group and between-group Bray–Curtis dissimilarities. Bray–Curtis distances between parasite infracommunities at fished and unfished atolls were tested using PERMANOVA, in a model that included island fishing status (fished vs. unfished) as a main effect, and also tested for the additional explanatory value of the covariates island (Jarvis, Kingman, Palmyra, Teraina, Tabuaeran, Kiritimati), host body size (TL), and depth and latitude of host collection. P values were obtained with 1000 permutations of the fourth-root transformed data. A SIMPER analysis was performed on fourth-root transformed data to identify the parasite species driving the differences between fished and unfished atolls identified by the PERMANOVA analysis. All analyses were carried out with functions in the R package vegan.
Results
The multivariate approach confirmed that the composition of parasite assemblages differed between fished and unfished atolls in the directions revealed by the GLMM/meta-analytic approach. This strong agreement between the two approaches suggests that our results are robust, and comes despite the fact that (1) the multivariate approach does not include any fish that carried no parasites and (2) the individual GLMMs do not assess differences in abundance between fished and unfished atolls for any parasite that was found at extremely low abundance. In PERMANOVA analyses, fishing status was a significant driver of variation in the composition of parasite infracommunities for all hosts (Appendix Table A1): Cephalopholis urodeta (F1,153 = 27.68, p = 0.001), Paracirrhites arcatus (F1,69 = 13.39, p = 0.001), Acanthurus nigricans (F1,120 = 2.66, p = 0.043), Ctenochaetus marginatus (F1,128 = 30.34, p = 0.001), Chromis margaritifer (F1,49 = 10.33, p = 0.001), Pseudanthias bartlettorum (F1,11 = 5.84, p = 0.004), and Stegastes aureus (F1,86 = 23.84, p = 0.001). According to the SIMPER analysis, these differences were overwhelmingly driven by the parasite taxa that displayed significant responses to the effect of fishing status in the individual GLMMs summarized in Fig. 3 (Appendix Table A2). Most parasite taxa that displayed a significant response to fishing pressure in individual GLMMs (Fig. 3, main text) were identified by SIMPER as contributing >10% of overall dissimilarity, with the exception of larval nematodes in Cephalopholis urodeta and Ctenochaetus marginatus. Some parasite taxa that did not have a significant response to fishing pressure in individual GLMMs were found to contribute >10% of overall dissimilarity in the SIMPER analysis (Appendix Table A2). For this subset of parasites, responses generally corresponded with the directional patterns uncovered in the meta-analysis described above: that is, directly transmitted crustacean parasites were more abundant on fished than unfished atolls (Grandiunguid sp. 1 in Chromis margaritifer and Grandiunguid sp. 1 and 2 in Pseudanthias bartlettorum), trophically transmitted cestode parasites were more abundant on unfished atolls than fished atolls (Tetraphyllidean sp. in Acanthurus nigricans), and the response of trematode parasites was mixed (more abundant on unfished than fished atolls: fin metacercariae and Bucephalid sp. in Paracirrhites arcatus, Microscaphiid sp. in Pseudanthias bartlettorum, fin metacercariae in Stegastes aureus; more abundant on fished than unfished atolls: gill metacercariae in Chromis margaritifer; Appendix Table A2).
Table G1. PERMANOVA of fourth-root transformed Bray–Curtis dissimiliarities of parasite abundance.
Host
Source
df
SS
MS
Pseudo-F
P
Cephalopholis urodeta
fishing status
1
1.7562
1.75620
27.6781
0.000999
island
4
3.5898
0.89744
14.1439
0.000999
latitude
1
0.7910
0.79101
12.4664
0.000999
TL
1
0.5791
0.57914
9.1273
0.000999
depth
1
0.3290
0.32897
5.1846
0.001998
Paracirrhites arcatus
fishing status
1
2.9245
2.92447
13.3953
0.000999
island
3
3.5897
1.19657
5.4808
0.000999
latitude
1
0.3310
0.33104
1.5163
0.233766
TL
1
0.1374
0.13741
0.6294
0.633367
depth
1
0.0592
0.05924
0.2713
0.887113
Acanthurus nigricans
fishing status
1
0.1725
0.17253
2.6606
0.042957
island
3
4.4135
1.47116
22.6872
0.000999
latitude
1
0.1784
0.17844
2.7518
0.031968
TL
1
0.2589
0.25890
3.9926
0.007992
depth
1
0.3140
0.31403
4.8427
0.002997
Ctenochaetus marginatus
fishing status
1
4.2829
4.2829
30.3351
0.000999
island
4
4.1829
1.0457
7.4066
0.000999
latitude
1
0.4277
0.4277
3.0291
0.018981
TL
1
0.5542
0.5542
3.9254
0.001998
depth
1
0.2281
0.2281
1.6157
0.174825
Chromis margaritifer
fishing status
1
2.1597
2.15974
10.3261
0.000999
island
3
5.0721
1.69069
8.0835
0.000999
latitude
1
0.6082
0.60822
2.9080
0.025974
TL
1
0.3564
0.35638
1.7039
0.157842
depth
1
0.3858
0.38577
1.8444
0.136863
Pseudanthias bartlettorum
fishing status
1
1.6513
1.65133
5.8355
0.003996
island
1
0.6124
0.61239
2.1641
0.106893
latitude
1
0.0736
0.07363
0.2602
0.832168
TL
1
0.0453
0.04526
0.1599
0.942058
depth
1
0.1239
0.12389
0.4378
0.765235
Stegastes aureus
fishing status
1
2.9703
2.97035
23.8400
0.000999
island
4
5.5276
1.38190
11.0911
0.000999
latitude
1
0.1246
0.12455
0.9996
0.325674
TL
1
0.4864
0.48640
3.9038
0.043956
depth
1
0.2968
0.29677
2.3819
0.123876
Table G2. Results of SIMPER analysis. # gives the number denoting each parasite taxon in Fig. 7. Contribution % gives the percent of overall dissimilarity explained by parasite taxon. Cumulative % gives the cumulative percent, starting with the parasite taxon contributing the most to the dissimilarity. Only parasite taxa contributing more than 10% of overall dissimilarity are reported below. Displayed under the host species name are p values indicating the significance of the difference in composition of parasite infracommunities between fished and unfished atolls, from PERMANOVA of fourth-root transformed Bray–Curtis dissimiliarities of parasite abundance (Appendix Table G1). Most parasite taxa that displayed a significant response to fishing pressure in individual GLMMs (Fig. 3) were identified by SIMPER as contributing >10% of overall dissimilarity, with the exception of larval nematodes in Cephalopholis urodeta and encysted larval nematodes in Ctenochaetus marginatus. Those parasite taxa that did not have a significant response to fishing pressure in individual GLMMs but did contribute >10% of overall dissimilarilty in the SIMPER analysis (i.e., those parasites where SIMPER was more sensitive to differences between fished and unfished atolls than GLMM) are denoted with an asterisk (*). Images of fishes courtesy of Cynthia Clark.
Host species
#
Parasite taxon
Relative abundance
Contribution %
Cumulative %
Cephalopholis urodeta
(p = 0.001)1
fin metacercariae
unfished > fished
0.2142529
0.2142529
2
Grandiunguid sp.
fished > unfished
0.1452153
0.3594682
3
dead nematodes
unfished > fished
0.1260119
0.4854801
4
gill metacercariae
unfished > fished
0.124869
0.6103491
5
Stephanostomum sp.
fished > unfished
0.1095607
0.7199098
Paracirrhites arcatus
(p = 0.001)1
Stephanostomum sp.
fished > unfished
0.2400277
0.2400277
2
fin metacercariae *
unfished > fished
0.2021064
0.4421341
3
larval nematodes
unfished > fished
0.172104
0.6142381
4
Bucephalid sp. *
unfished > fished
0.132215
0.7464531
Acanthurus nigricans
(p = 0.043)1
fin metacercariae
fished > unfished
0.2546453
0.2546453
2
Microscaphiid sp.
fished > unfished
0.2347066
0.4893519
3
gill metacercariae sp. 1
unfished > fished
0.1697191
0.6590710
4
Tetraphyllidean sp. *
unfished > fished
0.1194311
0.7785021
Ctenochaetus marginatus
(p = 0.001)1
Grandiunguid sp. 1
fished > unfished
0.2046055
0.2046055
2
Ancyrocephalidsp.
fished > unfished
0.1670085
0.3716140
3
Grandiunguid sp. 2
fished > unfished
0.1592844
0.5308984
4
Tetraphyllidean sp.
unfished > fished
0.1388083
0.6697067
5
gill metacercariae
unfished > fished
0.1098921
0.7795988
Chromis margaritifer
(p = 0.001)1
Stephanostomum sp.
fished > unfished
0.3318341
0.3318341
2
Grandiunguid sp. 1 *
fished > unfished
0.329962
0.6617961
3
gill metacercariae *
fished > unfished
0.152218
0.8140141
Pseudanthias bartlettorum
(p = 0.004)1
Microscaphiid sp. *
unfished > fished
0.4145455
0.4145455
2
Grandiunguid sp. 2 *
fished > unfished
0.2145454
0.6290909
3
Grandiunguid sp. 1 *
fished > unfished
0.2018182
0.8309091
Stegastes aureus
(p = 0.001)1
fin metacercariae *
unfished > fished
0.5105877
0.5105877
2
Stephanostomum sp.
fished > unfished
0.4113535
0.9219412
Fig. G1. Non-metric multi-dimensional scaling (MDS) plot on the first two axes of three- or two-dimensional solutions of fourth-root transformed Bray–Curtis dissimilarities between parasite infracommunities for each host species. The stress level and number of dimensions (k) of the ordination, as well as the significance level (p value) of the contrast between hosts from fished and unfished atolls (results of PERMANOVA from Table S7) is shown in the upper corner of each plot. Species that contribute more than 10% are indicated on the plot in order of declining contribution (see Table S7): (a) in Cephalopholis urodeta, 1 = fin metacercariae, 2 = Grandiunguid sp., 3 = dead nematodes, 4 = gill metacercariae, 5 = Stephanostomum sp., (b) in Paracirrhites arcatus, 1 = Stephanostomum sp., 2 = fin metacercariae, 3 = larval nematodes, 4 = Bucephalid sp., (c) in Acanthurus nigricans, 1 = fin metacercariae, 2 = Microscaphiid sp., 3 = gill metacercariae sp. 1, 4 = Tetraphyllidean sp., (d) in Ctenochaetus marginatus, 1 = Grandiunguid sp. 1, 2 = Ancyrocephalidsp., 3 = Grandiunguid sp. 2, 4 = Tetraphyllidean sp., 5 = gill metacercariae, (e) in Chromis margaritifer, 1 = Stephanostomum sp., 2 = Grandiunguid sp. 1, 3 = gill metacercariae, (f) in Pseudanthias bartlettorum, 1 = Microscaphiid sp., 2 = Grandiunguid sp. 2, 3 = Grandiunguid sp. 1, (g) in Stegastes aureus, 1 = fin metacercariae, 2 = Stephanostomum sp. Images of fishes courtesy of Cynthia Clark.
Literature cited
Field, J. G., K. R. Clarke, et al. 1982. A practical strategy for analyzing multispecies distribution patterns. Marine Ecology Progress Series 8:37–52.