Limits...
Large ‐ scale dark diversity estimates: new perspectives with combined methods

View Article: PubMed Central - PubMed

ABSTRACT

Large‐scale biodiversity studies can be more informative if observed diversity in a study site is accompanied by dark diversity, the set of absent although ecologically suitable species. Dark diversity methodology is still being developed and a comparison of different approaches is needed. We used plant data at two different scales (European and seven large regions) and compared dark diversity estimates from two mathematical methods: species co‐occurrence (SCO) and species distribution modeling (SDM). We used plant distribution data from the Atlas Florae Europaeae (50 × 50 km grid cells) and seven different European regions (10 × 10 km grid cells). Dark diversity was estimated by SCO and SDM for both datasets. We examined the relationship between the dark diversity sizes (type II regression) and the overlap in species composition (overlap coefficient). We tested the overlap probability according to the hypergeometric distribution. We combined the estimates of the two methods to determine consensus dark diversity and composite dark diversity. We tested whether dark diversity and completeness of site diversity (log ratio of observed and dark diversity) are related to various natural and anthropogenic factors differently than simple observed diversity. Both methods provided similar dark diversity sizes and distribution patterns; dark diversity is greater in southern Europe. The regression line, however, deviated from a 1:1 relationship. The species composition overlap of two methods was about 75%, which is much greater than expected by chance. Both consensus and composite dark diversity estimates showed similar distribution patterns. Both dark diversity and completeness measures exhibit relationships to natural and anthropogenic factors different than those exhibited by observed richness. In summary, dark diversity revealed new biodiversity patterns which were not evident when only observed diversity was examined. A new perspective in dark diversity studies can incorporate a combination of methods.

No MeSH data available.


Consensus and composite dark diversity estimates (species predicted by both or one method) at the European scale (A) consensus dark diversity, (B) composite dark diversity, and at the regional scale (C) consensus dark diversity and (D) composite dark diversity. Note that each map has own scale. Projection: Lambert azimuthal equal area.
© Copyright Policy - creativeCommonsBy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5016647&req=5

ece32371-fig-0003: Consensus and composite dark diversity estimates (species predicted by both or one method) at the European scale (A) consensus dark diversity, (B) composite dark diversity, and at the regional scale (C) consensus dark diversity and (D) composite dark diversity. Note that each map has own scale. Projection: Lambert azimuthal equal area.

Mentions: At the European scale, SCO and SDM methods both showed that dark diversity is low in north Europe and high in south Europe (Fig. 2), although in some areas, for example, in the Iberian Peninsula, SCO assigned less dark diversity than SDM. Consensus and composite dark diversity showed relatively similar patterns (Fig. 3). Consensus dark diversity by both methods estimated dark diversity to be greatest in southern Europe. Compared with consensus dark diversity, the composite dark diversity showed rather greater dark diversity in central Europe and in the Iberian Peninsula. Heterogeneity (PCA1) showed differences between observed species richness and dark diversity estimates: All dark diversity estimates were significantly weaker than for observed species richness. Dark diversity calculated by SDM method and composite dark diversity both showed positive relationships with latitude (PCA2), and these relationships differed significantly from that of observed species richness (no relationship in spatially informed model). Seasonality (PCA3) showed the greatest differences between observed species richness and dark diversity estimates: Observed species richness was related positively whereas dark diversity estimates related negatively (Table 1). The regional scale shows also a similar pattern with some differences, for example, dark diversity by SCO is less than by SDM in north Germany. Dark diversity estimates showed similar negative relationships with latitude (PCA1) and heterogeneity (PCA2, Table 2), and all dark diversity estimates had significantly different relationships than patterns from observed species richness.


Large ‐ scale dark diversity estimates: new perspectives with combined methods
Consensus and composite dark diversity estimates (species predicted by both or one method) at the European scale (A) consensus dark diversity, (B) composite dark diversity, and at the regional scale (C) consensus dark diversity and (D) composite dark diversity. Note that each map has own scale. Projection: Lambert azimuthal equal area.
© Copyright Policy - creativeCommonsBy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5016647&req=5

ece32371-fig-0003: Consensus and composite dark diversity estimates (species predicted by both or one method) at the European scale (A) consensus dark diversity, (B) composite dark diversity, and at the regional scale (C) consensus dark diversity and (D) composite dark diversity. Note that each map has own scale. Projection: Lambert azimuthal equal area.
Mentions: At the European scale, SCO and SDM methods both showed that dark diversity is low in north Europe and high in south Europe (Fig. 2), although in some areas, for example, in the Iberian Peninsula, SCO assigned less dark diversity than SDM. Consensus and composite dark diversity showed relatively similar patterns (Fig. 3). Consensus dark diversity by both methods estimated dark diversity to be greatest in southern Europe. Compared with consensus dark diversity, the composite dark diversity showed rather greater dark diversity in central Europe and in the Iberian Peninsula. Heterogeneity (PCA1) showed differences between observed species richness and dark diversity estimates: All dark diversity estimates were significantly weaker than for observed species richness. Dark diversity calculated by SDM method and composite dark diversity both showed positive relationships with latitude (PCA2), and these relationships differed significantly from that of observed species richness (no relationship in spatially informed model). Seasonality (PCA3) showed the greatest differences between observed species richness and dark diversity estimates: Observed species richness was related positively whereas dark diversity estimates related negatively (Table 1). The regional scale shows also a similar pattern with some differences, for example, dark diversity by SCO is less than by SDM in north Germany. Dark diversity estimates showed similar negative relationships with latitude (PCA1) and heterogeneity (PCA2, Table 2), and all dark diversity estimates had significantly different relationships than patterns from observed species richness.

View Article: PubMed Central - PubMed

ABSTRACT

Large‐scale biodiversity studies can be more informative if observed diversity in a study site is accompanied by dark diversity, the set of absent although ecologically suitable species. Dark diversity methodology is still being developed and a comparison of different approaches is needed. We used plant data at two different scales (European and seven large regions) and compared dark diversity estimates from two mathematical methods: species co‐occurrence (SCO) and species distribution modeling (SDM). We used plant distribution data from the Atlas Florae Europaeae (50 × 50 km grid cells) and seven different European regions (10 × 10 km grid cells). Dark diversity was estimated by SCO and SDM for both datasets. We examined the relationship between the dark diversity sizes (type II regression) and the overlap in species composition (overlap coefficient). We tested the overlap probability according to the hypergeometric distribution. We combined the estimates of the two methods to determine consensus dark diversity and composite dark diversity. We tested whether dark diversity and completeness of site diversity (log ratio of observed and dark diversity) are related to various natural and anthropogenic factors differently than simple observed diversity. Both methods provided similar dark diversity sizes and distribution patterns; dark diversity is greater in southern Europe. The regression line, however, deviated from a 1:1 relationship. The species composition overlap of two methods was about 75%, which is much greater than expected by chance. Both consensus and composite dark diversity estimates showed similar distribution patterns. Both dark diversity and completeness measures exhibit relationships to natural and anthropogenic factors different than those exhibited by observed richness. In summary, dark diversity revealed new biodiversity patterns which were not evident when only observed diversity was examined. A new perspective in dark diversity studies can incorporate a combination of methods.

No MeSH data available.