Fig 1.
Map of study areas including the number of samples within each location in parentheses.
Table 1.
Data inventory table with references.
RPD and LOI refer to the apparent redox potential discontinuity depth and sediment organic matter measured by loss on ignition, respectively.
Fig 2.
Example of categorical acoustic provinces (right) created from backscatter data (left) from the Haverstraw Bay study site.
Points indicate the location of sampling stations. Basemap from https://nationalmap.gov.
Table 2.
Variogram nomenclature [14,47,48].
Table 3.
Summary of faunal data including mean abundance of all taxa per m2, mean species density per sample, and overall species richness at each study location.
Table 4.
Summary of environmental data.
Mean values (and range) of the grab-scale environmental variables acquired via in situ sampling. RPD and LOI refer to the apparent redox potential discontinuity depth and sediment organic matter measured by loss on ignition, respectively.
Table 5.
Percent surficial cover from maximum likelihood analysis of images extracted from underwater video surveys at the Tappan Zee, Robins Island, and Shelter Island locations.
Fig 3.
Empirical multivariate variogram of community data (points) and the percentage of sample pair comparisons between acoustic provinces (bars).
Distance intervals ranged from 0.25 to 0.75 km among study locations. Arrows on the vertical and horizontal axes indicate nugget and range estimates, respectively.
Table 6.
Fitted variogram models, total variance estimates (s2), variogram parameter estimates (c0, c1, and a), and other variogram relationships for each study location.
c0 is an estimate of the nugget effect and c0/s2 is an estimate of the fraction of the total variance represented by small-scale variability. The sum c0 + c1 is the sill, and a defines the rate and, in the case of the spherical and piecewise linear models, the range at which the sill is reached. For the exponential model, the range was estimated as the distance where the variogram model reached 95% of the sill. The maximum extent is the largest geographic distance between two sampling locations in a study.
Fig 4.
Ordination triplots illustrating the nonspatial RDA results.
Scores for individual samples plotted as points. Sample points close to one another tend to have similar faunal structure than those further apart. Different point shapes and color represent samples collected from different acoustic provinces. Polygons enclose samples from each province. Continuous explanatory variables and individual taxa are plotted as vectors. The vector arrowheads represent high, the origin averages, and the tail (when extended through the origin) low values of the selected environmental variables. Projections of sample points onto an individual taxa vector approximate the Hellinger transformed abundances for that taxon. For clarity, only species with the highest amounts of explained variance (typically those with > 10%) are plotted. Acca = Acteocina canaliculata, Antr = Anadara transversa, Baca = Batea catharinensis, Baim = Balanus imporvisus, Balamp = Balanus amphitrite, Boclig = Boccardia ligerica, Cdeco = Chironomus decorus, Chispp = Chironomidae sp., Crfo = Crepidula fornicata, Crpl = Crepidula plana, Crychi = Cryptochironomus sp., Cscap = Coelotanypus scapularis, Cpoli/Cypo = Cyathura polita, Dstag = Dorylaimus cf. stagnalis, Elle = Elasmopus levis, Endi = Ensis directus, Eusa = Eumida sanguinea, Geudem = Geukensia demissa, Gtigr = Gammarus tigrinus, Hefi = Heteromastus filiformis, Hofl = Heteromysis formosa, Hypgra = Hypaniola grayi, Ifrey = Isochaetides freyi, imTubwo = immature Tubificidae without hair setae, Lepplu/Lepl = Leptocheirus plumulosus, Lesm = Lembos smithi, Lhoff = Limnodrilus hoffmeisteri, Lyhy = Lyonsia hyalina, Mazo = Macroclymene zonalis, Melnit/Meni = Melita nitida, Meme = Mercenaria mercenaria, Mula = Mulinia lateralis, Mytleu/Myle = Mytilopsis leucophaeata, Neasuc/Nesu = Neanthes succinea, Nema = Nematode, Olig = Oligochaeta, Ophi = Ophiuroidea, Pago = Pandora gouldiana, Pahe = Panopeus herbstii, Phalt = Polypedilum halterale, Pygo = Polygordius sp., Racu/Racun = Rangia cuneata, Scfr = Scoloplos fragilis, Stbe = Streblospio benedicti, Teag = Tellina agilis, and Thsp = Tharyx sp.
Table 7.
RDA forward selection results, eigenvalues, and AICc for each study location.
Minimum AICc values are indicated in bold. Capital letters listed in the “Variable” column are provinces, and the variables are listed in order of selection.
Fig 5.
MSO plots for each study area.
Crosshairs are the sum of fitted and residual variograms, solid lines are Bonferroni-corrected point confidence envelope of the empirical variogram, open diamonds represent the fitted variogram, and open squares are the residual variogram. Numbers above the distance axis are the number of sample pairs in each distance class. See the explanation of Eq 7 definitions of these quantities. Solid squares, although absent in all cases, would indicate the presence significant spatial autocorrelation in the residuals (detected via Mantel tests). Distances are in kilometers.
Fig 6.
Community variance explained by RDA models (fit) compared to nugget corrected residual variance (res − c0).
For each data set, comparisons are made for the entire nonspatial RDA and for results broken down into components representing variation within and between acoustic provinces.