Limits...
The shape of terrestrial abundance distributions.

Alroy J - Sci Adv (2015)

Bottom Line: However, previous studies have focused on either a few theoretical distributions or a few empirical distributions.However, unlike any other model, it assumes both that richness is finite and that species compete unequally for resources in a two-dimensional niche landscape, which implies that niche breadths are variable and that trait distributions are neither arrayed along a single dimension nor randomly associated.The hypothesis that niche space is multidimensional helps to explain how numerous species can coexist despite interacting strongly.

View Article: PubMed Central - PubMed

Affiliation: Department of Biological Sciences, Macquarie University, New South Wales 2109, Australia. E-mail: john.alroy@mq.edu.au.

ABSTRACT
Ecologists widely accept that the distribution of abundances in most communities is fairly flat but heavily dominated by a few species. The reason for this is that species abundances are thought to follow certain theoretical distributions that predict such a pattern. However, previous studies have focused on either a few theoretical distributions or a few empirical distributions. I illustrate abundance patterns in 1055 samples of trees, bats, small terrestrial mammals, birds, lizards, frogs, ants, dung beetles, butterflies, and odonates. Five existing theoretical distributions make inaccurate predictions about the frequencies of the most common species and of the average species, and most of them fit the overall patterns poorly, according to the maximum likelihood-related Kullback-Leibler divergence statistic. Instead, the data support a low-dominance distribution here called the "double geometric." Depending on the value of its two governing parameters, it may resemble either the geometric series distribution or the lognormal series distribution. However, unlike any other model, it assumes both that richness is finite and that species compete unequally for resources in a two-dimensional niche landscape, which implies that niche breadths are variable and that trait distributions are neither arrayed along a single dimension nor randomly associated. The hypothesis that niche space is multidimensional helps to explain how numerous species can coexist despite interacting strongly.

No MeSH data available.


Predicted and actual dominance (frequency of the most common species) in 1055 ecological samples, including trees (97), bats (159), small terrestrial mammals (161), birds (119), lizards (77), frogs (110), ants (77), dung beetles (115), butterflies (83), and odonates (57).(A) Double geometric distribution. (B) Lognormal distribution. (C) Geometric series distribution. (D) Log series distribution.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4643760&req=5

Figure 3: Predicted and actual dominance (frequency of the most common species) in 1055 ecological samples, including trees (97), bats (159), small terrestrial mammals (161), birds (119), lizards (77), frogs (110), ants (77), dung beetles (115), butterflies (83), and odonates (57).(A) Double geometric distribution. (B) Lognormal distribution. (C) Geometric series distribution. (D) Log series distribution.

Mentions: The double geometric model consistently yields accurate estimates of dominance (Fig. 3), as indicated by several statistics (Table 1). First, the estimates and observed values are highly correlated (as with the lognormal and log series but not the geometric series). Second, regression of actual values on double geometric–predicted values produces a slope very close to 1 and an intercept very close to 0. By contrast, the relationship is steep for the lognormal and shallow for the geometric series, whereas the intercepts are much farther from 0 for these two distributions. Although the log series performs better, its intercept is nonetheless substantially farther from 0. The lognormal might appear to perform well upon casual inspection of the data (Fig. 3), but its estimates are consistently too high (Table 1). Finally, the double geometric produces a median offset between predicted and observed values that is considerably smaller than the offsets generated by all other models.


The shape of terrestrial abundance distributions.

Alroy J - Sci Adv (2015)

Predicted and actual dominance (frequency of the most common species) in 1055 ecological samples, including trees (97), bats (159), small terrestrial mammals (161), birds (119), lizards (77), frogs (110), ants (77), dung beetles (115), butterflies (83), and odonates (57).(A) Double geometric distribution. (B) Lognormal distribution. (C) Geometric series distribution. (D) Log series distribution.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4643760&req=5

Figure 3: Predicted and actual dominance (frequency of the most common species) in 1055 ecological samples, including trees (97), bats (159), small terrestrial mammals (161), birds (119), lizards (77), frogs (110), ants (77), dung beetles (115), butterflies (83), and odonates (57).(A) Double geometric distribution. (B) Lognormal distribution. (C) Geometric series distribution. (D) Log series distribution.
Mentions: The double geometric model consistently yields accurate estimates of dominance (Fig. 3), as indicated by several statistics (Table 1). First, the estimates and observed values are highly correlated (as with the lognormal and log series but not the geometric series). Second, regression of actual values on double geometric–predicted values produces a slope very close to 1 and an intercept very close to 0. By contrast, the relationship is steep for the lognormal and shallow for the geometric series, whereas the intercepts are much farther from 0 for these two distributions. Although the log series performs better, its intercept is nonetheless substantially farther from 0. The lognormal might appear to perform well upon casual inspection of the data (Fig. 3), but its estimates are consistently too high (Table 1). Finally, the double geometric produces a median offset between predicted and observed values that is considerably smaller than the offsets generated by all other models.

Bottom Line: However, previous studies have focused on either a few theoretical distributions or a few empirical distributions.However, unlike any other model, it assumes both that richness is finite and that species compete unequally for resources in a two-dimensional niche landscape, which implies that niche breadths are variable and that trait distributions are neither arrayed along a single dimension nor randomly associated.The hypothesis that niche space is multidimensional helps to explain how numerous species can coexist despite interacting strongly.

View Article: PubMed Central - PubMed

Affiliation: Department of Biological Sciences, Macquarie University, New South Wales 2109, Australia. E-mail: john.alroy@mq.edu.au.

ABSTRACT
Ecologists widely accept that the distribution of abundances in most communities is fairly flat but heavily dominated by a few species. The reason for this is that species abundances are thought to follow certain theoretical distributions that predict such a pattern. However, previous studies have focused on either a few theoretical distributions or a few empirical distributions. I illustrate abundance patterns in 1055 samples of trees, bats, small terrestrial mammals, birds, lizards, frogs, ants, dung beetles, butterflies, and odonates. Five existing theoretical distributions make inaccurate predictions about the frequencies of the most common species and of the average species, and most of them fit the overall patterns poorly, according to the maximum likelihood-related Kullback-Leibler divergence statistic. Instead, the data support a low-dominance distribution here called the "double geometric." Depending on the value of its two governing parameters, it may resemble either the geometric series distribution or the lognormal series distribution. However, unlike any other model, it assumes both that richness is finite and that species compete unequally for resources in a two-dimensional niche landscape, which implies that niche breadths are variable and that trait distributions are neither arrayed along a single dimension nor randomly associated. The hypothesis that niche space is multidimensional helps to explain how numerous species can coexist despite interacting strongly.

No MeSH data available.