Rank-abundance Geometric series: found in very communities such as the Log series: group of species that occur _ time are the most frequent. Useful for calculating a diversity metric (Fisher s alpha) Most abundant Least abundant Log normal: More even distribution of abundance but still skewed to rare species. Can be normalized by taking of abundance classes (octaves)
Log-normal Species Abundance Distributions (SAD) Number of spp # individs log 2 # individs log 3 Octaves -3-2 -1 0 1 2 3 4 5
Back to the Rothamstead moths: Preston showed that fits to log series represented inadequate sampling of a log normal distribution Number of species Moth abundance
Why do species abundances commonly follow a lognormal distribution? No clear answer. May (1975) and others argued that results as consequence of the Central Limit Theorem (product of interacting effects of many random processes (e.g., competition, predation, etc.) J. H. Brown (1995, Macroecology, pg. 79): just as normal distributions are produced by additive combinations of random variables, lognormal distributions are produced by multiplicative combinations of random variables (May 1975) http://www.stat.sc.edu/~west/javahtml/clt.html
Rank-abundance plots Rank abundance plots of four sppabundance models x axis = rank order of abundance (most abundant least abundant) y axis = log species abundance Most abundant Least abundant
Broken Stick model: Proposed by MacArthur (1957) Imagine trophically similar species dividing up a common pool of resources, so that relative abundance is proportional to the fraction of total resources each species uses (rem: geometric series) Broken stick because imagine placing S-1 points at random along a resource axis (stick) and then breaking it into stick sections according to the position of the points result not a log-normal (abundances are too even )
Broken Stick: The sub-division of niche space among species is analogous to randomly breaking a stick into S pieces (MacArthur 1957) Results in a somewhat more even distribution of abundances among species than the other models, which suggests that it should occur when an important resource is shared more or less equitably among species Sp. 1 Sp. 2 Sp. 3 Sp. 4 Sp. 5 Total Available Resources
Hubbell (2001) Problems with the development of theories of species relative abundances 1. Approach is either inductive or statistical - fit a model to the data without reference to an underlying mechanism 2. When more deductive (e.g. McArthur s broken stick) the particular mechanism partitioning resources is unclear (what is the stick??!!) 3. The expected relative species abundances are only determined once the number of species partitioning the resources has been specified
Community phylogenetics the rebirth of assembly rules Last 10 years push to combine phylogenetic analysis of species relationships with community assembly and structure 3 perspectives on how communities assemble: (1) Niche-assembly rules dictated by local environmental filters and the principle of competitive exclusion (Tilman, Diamond) (2) Neutral assembly (the null model approach) where species are assumed to be ecologically equivalent (Hubbell, Simberloff) (3) History-based assembly. Starting conditions and historical patterns of speciation matter more than local processes (Ricklefs)
(1) Niche assembly view Tilman: competitive interactions determine which species can coexist locally Diamond: resource availability determines which species can coexist locally
(2) Neutral assembly view Hubbell (1979) Development of SAD from neutral processes (also considered a non-equilibrium theory for the maintenance of diversity) Non-equilibruim view: Competition is minimal. Disturbances occur at sufficiently frequent intervals to prevent competitive exclusion Furthermore in species rich communities, selection for specialized niches might be weak, and most species are ecologically equivalent generalists and few selective forces can drive their elimination from a community (rem: character displacement?)
Hubbell (1979) Community drift model: Imagine a forest saturated with K trees (all species). Each tree controls one unit of canopy space and resists invasion by other trees until killed.
Hubbell (1979) Community drift model: Imagine a forest saturated with K trees (all species). Each tree controls one unit of canopy space and resists invasion by other trees until killed - Suppose a windstorm or landslide kills D trees with mortality randomly distributed across species (loss of each species will be proportional to its current relative abundance) - Let D new trees replace the vacancies, with the proportion of replacement trees contributed by each species given by the proportional abundance of the species in the community after the disturbance. This is the only assembly rule. No species is inherently better than any other in occupying a site. - Run the model of simulated tree death and replacement over time, what would the outcome be??
4 species: Simulated forest stand K=20 Each model iteration: one tree dies. Probability of replacement is proportional to its relative abundance in the community Probability of replacement by green is 8/20, red is 7/20... Could apply to any system where dynamics is a zero sum game
Hubbell (1979) Model predictions Species abundance patterns will take a random walk. Over the long term, with no immigration or recolonization of the local site, all but one species will be lost by extinction. Start with even distribution. Over the short term the model leads to lognormal relative abundances, and over the mid-term to a geometric distribution (assuming no immigration or recolonization) How quickly would a species be lost by extinction? Depends on magnitude of D relative to K For D=8, K=512, will take 90,000 disturbance events to remove or fix a species with starting population of 256 individuals
For realistic population sizes and mortality rates Hubbell argues that species can be viewed as essentially immune to extinction over geologically significant time spans - long enough for speciation to become an important process
Phylogenetic approach: identification of historical processes that underlie community assembly - Emphasis on competitive exclusion/limiting similarity led to convenient assumption that evolutionary processes are not relevant on the time scale of ecological processes. Cavender-Bares et al. (2009) Ecol. Lett. 12:693-715
The paradox of phenotypic similarity species of the same genus have usually, though by no means invariably, some similarity in habits and constitution Darwin (1859) So closely related species should also experience strong competitive interactions due to their ecological similarity. One the one hand environmental filtering will select for species with similar traits in the same environment. On the other hand ecological similarity may prevent closely related species from sharing environments. Community phylogenetics explores the relative importance of competitive exclusion and ecological character displacement in community assembly.
What might community phylogenetic structure look like? Scenario #1 Strong phylogenetic signal in community assembly Clustering is a consequence of trait conservatism closely related species have similar ecologies Phenotypic clustering in turn results from environmental filtering
What might community phylogenetic structure look like? Scenario #1 Strong phylogenetic signal in community assembly Clustering is a consequence of trait conservatism closely related species have similar ecologies Phenotypic clustering in turn results from environmental filtering
What might community phylogenetic structure look like? Communities composed of species from different branches of phylogeny. Why? Species on different branches converge on similar traits Environmental filtering controls what traits can occur in a niche/ community
What might community phylogenetic structure look like? Communities composed of species from different branches of phylogeny. Why? Species on different branches converge on similar traits Environmental filtering controls what traits can occur in a niche/ community
Explaining phylogenetic structure (Webb 2002) If environmental filtering dominates, co-occurring species sharing the same abiotic environment should have more similar traits (phenotypically more similar) than expected (trait clustering)
Explaining phylogenetic structure (Webb 2002) If competitive interactions dominate, co-occurring species sharing the same abiotic environment should be phenotypically less similar than expected (trait overdispersion)
Cavender-Bares et al. (2004) Phylogenetic overdispersion of oak communities 17 oak species occur in North Central Florida in sites that range in moisture availability. Environmental filtering oaks that live in similar environments should show similar phenotypic traits. But species that are too similar are unlikely to co-occur because of competitive exclusion Explored correlations between phylogenetic relatedness of oaks, degree of co-occurrence, and similarity in physiological traits.
Floridian oak phylogeny and a mapped on trait (soil moisture preference) Any evidence for phylogenetic clustering??
Found: Significant negative correlation between species differences in soil moisture preference and phylogenetic distance so, distantly related species converge on the same habitat conditions. Despite significant phylogenetic overdispersion there is evidence for environmental filtering in this study: Bark thickness, radial growth, rhizome resprouting potential, seedling growth rate all show phenotypic clustering, indicating that co-occurring species across a soil moisture gradient were phenotypically similar.
Graham et al. (2009) Phylogenetic structure of 189 hummingbird communities in Ecuador Hummingbirds: Compete strongly for nectar, and have striking phenotypes that influence foraging capacity and diet choice across different environments
Of 189 hummingbird communities, 134 (71%) had positive net relatedness index (NRI) indicating phylogenetic clustering NRI is a measure of how closely related the hummers are in a single community. It is calculated using a null model that includes information on the relatedness of all hummers in the study. Positive NRI hummers are more closely related than expected. Blue and red are significant departures from null expectation Clear elevational gradient in NRI
Break down pattern in NRI by clade Clade: bee, brilliant, coquette, emerald, hermit, mangoe etc
benign environments wet lowlands E and W of Andes. y axis is proportion of communities where the clade is represented. Numbers on bar are number of taxa/clade Overdispersion here (potentially) represents interspecific competition for shared nectar resources.
Clustering occurs in challenging environments high elevation (C) or in arid habitats (D) Results consistent with other work that suggests that harsh environments are a stronger filter on species traits. If traits are phylogenetically conserved then communities will show clustering.
Conclusions: Habitat filtering and biotic interactions can act together to assemble communities. Phylogenetic approaches can help to partition these two effects Evidence for interspecific competition as a driver of overdispersion however is mixed (see Wednesday discussion papers). Phylogenetic patterns also tend to be scale dependent Regional scale: environmental filtering leads to clustering Local scale: competitive exclusion/limiting similarity may lead to overdispersion.