Figure 1.
Study area, The Mariana Archipelago showing the location of all islands and banks.
Table 1.
Environmental predictor variables included in the boosted regression tree analysis.
Table 2.
Boosted regression tree (BRT) analysis: Optimal parameter settings, predictive performance, and relative influence of environmental variables on total large-bodied reef fish biomass and presence/absence of key species.
Table 3.
List of taxonomic groups and reasons for inclusion in the analysis.
Figure 2.
Biomass of large-bodied reef fish (all species and years pooled) in the Mariana Archipelago.
Error bars are 1 standard error.
Figure 3.
Two-dimensional nMDS ordination plots of islands in the Mariana Archipelago based on multivariate species-level dispersion-weighted biomass of large-bodied (>50 cm TL) reef fishes and Bray-Curtis similarities.
Cluster contours represent significant SIMPROF groups (∼30% similarity within groups). In the first panel, symbols reflect level of human population. Aguijan is categorized as “Mid” human population due to the influence of nearby Tinian and Saipan as well as reported visitation by fishers from Guam. Subsequent panels are the same nMDS as panel 1, with symbols representing relative biomass of each species.
Table 4.
SIMPER analysis results showing species contribution to the dissimilarity between significant SIMPROF groupings based on species-level biomass (dispersion-weighted) and human population.
Figure 4.
Relative importance plots for boosted regression tree analysis of total large-bodied reef fish biomass (all species pooled) and presence/absence of key species.
Species selection was based on maximum data density, and management and ecological importance. CAAB = Carcharhinus amblyrhynchos, LUBO = Lutjanus bohar, MASP = Macolor spp., TROB = Triaenodon obesus, NAHE = Naso hexacanthus, SCRU = Scarus rubroviolaceus, CAME = Caranx melampygus, CHUD = Cheilinus undulatus.
Figure 5.
Partial dependency plots for boosted regression tree analysis of total large-bodied reef fish biomass (all species pooled) and presence/absence of key species.
Species selection was based on maximum data density, and management and ecological importance. CAAB = Carcharhinus amblyrhynchos, LUBO = Lutjanus bohar, MASP = Macolor spp., TROB = Triaenodon obesus, NAHE = Naso hexacanthus, SCRU = Scarus rubroviolaceus, CAME = Caranx melampygus, CHUD = Cheilinus undulatus.
Figure 6.
Pairwise interaction between depth and temperature with respect to total biomass of large-bodied reef fish.
The effect of temperature on total large-bodied reef fish biomass is magnified in areas of warm water and the effect of temperature is magnified at deeper depths.
Table 5.
Pairwise interactions between predictor variables used to relate total biomass and occurrence of each taxa to environment.
Figure 7.
Pairwise interaction between human population density and depth with respect to occurrence of Carcharhinus amblyrhynchos.
Depth is only influential in areas of low human population density and the effect of human population density is reduced in shallow waters.
Figure 8.
Pairwise interaction between wave energy and temperature with respect to occurrence of Lutjanus bohar.
The effect of temperature is magnified by wave energy while wave energy is primarily influential at temperatures below 28°C.
Figure 9.
Pairwise interaction between human population density and depth with respect to occurrence of Triaenodon obesus.
The effect of depth is influential only at the lowest human population densities and the effect of human population density is reduced in shallow areas.
Figure 10.
Pairwise interaction between human population density and wave energy with respect to occurrence of Triaenodon obesus.
The effect of wave energy is influential only at the lowest human population densities and the effect of human population density is magnified in higher wave energy areas.
Figure 11.
Pairwise interaction between depth and wave energy with respect to occurrence of Scarus rubroviolaceus.
The effect of wave energy is influential only in depths shallower than 15 m. The effect of depth is somewhat magnified in lower wave energy areas.
Figure 12.
Pairwise interaction between wave energy and sand cover with respect to occurrence of Scarus rubroviolaceus.
The effect of wave energy was highest in areas with low sand cover and the effect of sand cover was greatest in areas were wave energy was between 100 and 200 kW/m.
Figure 13.
Pairwise interaction between depth and percent cover of crustose coralline algae (CCA) with respect to occurrence of Caranx melampygus.
The effect of CCA was greatest at depths greater than 15 m while the effect of depth was reduced in areas with lower CCA cover.