Limits...
Cross-Scale Variation in Biodiversity-Environment Links Illustrated by Coastal Sandflat Communities.

Kraan C, Dormann CF, Greenfield BL, Thrush SF - PLoS ONE (2015)

Bottom Line: Across all scales, less than 30% of the variation in spatial structure was captured by our analysis.The large number of species (48) making up the 10 highest species scores based on redundancy analyses illustrate the variability of species-scale associations.Our results emphasise that abiotic variables and biodiversity are related at all scales investigated and stress the importance of assessing the relationship between environmental variables and the abundance and distribution of biological assemblages across a range of different scales.

View Article: PubMed Central - PubMed

Affiliation: National Institute of Water and Atmospheric Research, Hamilton, New Zealand.

ABSTRACT
Spatial variation in the composition of communities is the product of many biotic and environmental interactions. A neglected factor in the analysis of community distribution patterns is the multi-scale nature of the data, which has implications for understanding ecological processes and the development of conservation and environmental management practice. Drawing on recently established multivariate spatial analyses, we investigate whether including relationships between spatial structure and abiotic variables enable us to better discern patterns of species and communities across scales. Data comprised 1200 macrozoobenthic samples collected over an array of distances (30 cm to 1 km) in three New Zealand harbours, as well as commonly used abiotic variables, such as sediment characteristics and chlorophyll a concentrations, measured at the same scales. Moran's eigenvector mapping was used to extract spatial scales at which communities were structured. Benthic communities, representing primarily bivalves, polychaetes and crustaceans, were spatially structured at four spatial scales, i.e. >100 m, 50-100 m, 50-15 m, and < 15 m. A broad selection of abiotic variables contributed to the large-scale variation, whereas a more limited set explained part of the fine-scale community structure. Across all scales, less than 30% of the variation in spatial structure was captured by our analysis. The large number of species (48) making up the 10 highest species scores based on redundancy analyses illustrate the variability of species-scale associations. Our results emphasise that abiotic variables and biodiversity are related at all scales investigated and stress the importance of assessing the relationship between environmental variables and the abundance and distribution of biological assemblages across a range of different scales.

No MeSH data available.


Range of spatial autocorrelation of each significant positive MEM variable.Broad: MEM variables with a range > 100 m; Meso: MEM variables with a range < 100 m and > 50 m; Small: MEM variables with a range < 50 m and > 15 m; Fine: MEM variables with a range < 15 m. Delineation into 4 distinct spatial scales is based on visual appraisal (see Materials and Methods).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4640831&req=5

pone.0142411.g002: Range of spatial autocorrelation of each significant positive MEM variable.Broad: MEM variables with a range > 100 m; Meso: MEM variables with a range < 100 m and > 50 m; Small: MEM variables with a range < 50 m and > 15 m; Fine: MEM variables with a range < 15 m. Delineation into 4 distinct spatial scales is based on visual appraisal (see Materials and Methods).

Mentions: Next, (3) we constructed a spatial weighting matrix (SWM) to define linkages between sampling points, used for the decomposition in orthogonal spatial variables. We trialled connectivity based on Delaunay triangulation, minimum spanning tree, relative neighbourhood, nearest neighbours, Gabriel neighbourhood, and distance thresholds (see [26]), selecting a distance-based SWM (Table 1). This particular matrix optimised performance, as determined by the AICc (Table 2), and reflects a data-driven approach [15,24,26]. Subsequently, (4) this SWM was used in eigen decomposition of community data, providing spatial eigen functions (“MEM-variables”) that can be used as spatial predictors in ordination approaches (see, e.g. [2]). Significant positive MEM-variables, representing positive spatial autocorrelation (p ≤ 0.05, 9999 permutations), were grouped (Fig 2) based on a visual comparison of similarities in their range of spatial autocorrelation. This represents a routine method of clustering as single MEM-variables harbour little significance [28,29]. This grouping in MEM-subsets was constrained by our sampling design, such that “spatial scales” were limited between the smallest (30 cm) and largest (1 km) inter-sample distance.


Cross-Scale Variation in Biodiversity-Environment Links Illustrated by Coastal Sandflat Communities.

Kraan C, Dormann CF, Greenfield BL, Thrush SF - PLoS ONE (2015)

Range of spatial autocorrelation of each significant positive MEM variable.Broad: MEM variables with a range > 100 m; Meso: MEM variables with a range < 100 m and > 50 m; Small: MEM variables with a range < 50 m and > 15 m; Fine: MEM variables with a range < 15 m. Delineation into 4 distinct spatial scales is based on visual appraisal (see Materials and Methods).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4640831&req=5

pone.0142411.g002: Range of spatial autocorrelation of each significant positive MEM variable.Broad: MEM variables with a range > 100 m; Meso: MEM variables with a range < 100 m and > 50 m; Small: MEM variables with a range < 50 m and > 15 m; Fine: MEM variables with a range < 15 m. Delineation into 4 distinct spatial scales is based on visual appraisal (see Materials and Methods).
Mentions: Next, (3) we constructed a spatial weighting matrix (SWM) to define linkages between sampling points, used for the decomposition in orthogonal spatial variables. We trialled connectivity based on Delaunay triangulation, minimum spanning tree, relative neighbourhood, nearest neighbours, Gabriel neighbourhood, and distance thresholds (see [26]), selecting a distance-based SWM (Table 1). This particular matrix optimised performance, as determined by the AICc (Table 2), and reflects a data-driven approach [15,24,26]. Subsequently, (4) this SWM was used in eigen decomposition of community data, providing spatial eigen functions (“MEM-variables”) that can be used as spatial predictors in ordination approaches (see, e.g. [2]). Significant positive MEM-variables, representing positive spatial autocorrelation (p ≤ 0.05, 9999 permutations), were grouped (Fig 2) based on a visual comparison of similarities in their range of spatial autocorrelation. This represents a routine method of clustering as single MEM-variables harbour little significance [28,29]. This grouping in MEM-subsets was constrained by our sampling design, such that “spatial scales” were limited between the smallest (30 cm) and largest (1 km) inter-sample distance.

Bottom Line: Across all scales, less than 30% of the variation in spatial structure was captured by our analysis.The large number of species (48) making up the 10 highest species scores based on redundancy analyses illustrate the variability of species-scale associations.Our results emphasise that abiotic variables and biodiversity are related at all scales investigated and stress the importance of assessing the relationship between environmental variables and the abundance and distribution of biological assemblages across a range of different scales.

View Article: PubMed Central - PubMed

Affiliation: National Institute of Water and Atmospheric Research, Hamilton, New Zealand.

ABSTRACT
Spatial variation in the composition of communities is the product of many biotic and environmental interactions. A neglected factor in the analysis of community distribution patterns is the multi-scale nature of the data, which has implications for understanding ecological processes and the development of conservation and environmental management practice. Drawing on recently established multivariate spatial analyses, we investigate whether including relationships between spatial structure and abiotic variables enable us to better discern patterns of species and communities across scales. Data comprised 1200 macrozoobenthic samples collected over an array of distances (30 cm to 1 km) in three New Zealand harbours, as well as commonly used abiotic variables, such as sediment characteristics and chlorophyll a concentrations, measured at the same scales. Moran's eigenvector mapping was used to extract spatial scales at which communities were structured. Benthic communities, representing primarily bivalves, polychaetes and crustaceans, were spatially structured at four spatial scales, i.e. >100 m, 50-100 m, 50-15 m, and < 15 m. A broad selection of abiotic variables contributed to the large-scale variation, whereas a more limited set explained part of the fine-scale community structure. Across all scales, less than 30% of the variation in spatial structure was captured by our analysis. The large number of species (48) making up the 10 highest species scores based on redundancy analyses illustrate the variability of species-scale associations. Our results emphasise that abiotic variables and biodiversity are related at all scales investigated and stress the importance of assessing the relationship between environmental variables and the abundance and distribution of biological assemblages across a range of different scales.

No MeSH data available.