|
| |
|
| |
|
Extinction is one of the most fundamental processes in ecology. It shapes evolution, community
structure, and patterns of biodiversity. Yet studies of extinction are scarce. We have taken advantage
of an extraordinary database concerning occurrence records for native fishes in the southwestern US and
northwestern Mexico to investigate relationships between extirpation and patterns of rarity and life-
history traits. This unique database includes over 25,000 locality records spanning 160 years of field
research for all 50 known taxa of fishes from the Sonoran Desert region. Key to our project was the
development of new methods for quantifying patterns of rarity that are independent of the spatial scale
of analyses. We discussed how the patterns of spatial rarity these fish exhibit relate to their
extinction dynamics (e.g., frequency and pattern of extirpation events) and current levels of
endangerment. We also suggested ways in which patterns of rarity and extinction can be predicted by
life-history attributes.
|
 |
We created a map of the rivers in the Sonoran Desert region using the U.S. National Imagery and Mapping
Agency (NIMA) Digital Chart of the World (DCW), which is a comprehensive 1:1,000,000-scale vector
basemap of the world consisting of cartographic, attribute, and textual data. After cleaning and
splitting the lines representing these rivers into five-kilometer segments, we then created a buffer
extending one kilometer to either side in order to encompass the entire width. Sections of rivers
extending more than one kilometer in width, such as deltas, we covered with wider buffers as
appropriate. We gave each five-kilometer segment a unique identifier and then grouped these segments
into 25-kilometer, 100-kilometer, and 500-kilometer aggregates.
|
|
We brought the buffered river coverages into ESRI ArcView, a desktop GIS and mapping software package,
along with a point coverage representing occurrences of the native freshwater fish fauna of western
United States and Mexico (the SONFISHES database). Utilizing the spatial analysis functionality of
ArcView, we summarized these fish occurrences based upon their location within the buffered river
segments.
|
|
Linked to each of the occurrence records is a wide array of additional information, including the year
in which the field data was collected. We made use of this data to produce two different types of
analysis. Our first interest was in the presence and absence within river segments of each species
prior to and after 1980, in order to acquire a general concept of local species extirpations.
|
 |
|
We then employed a more in-depth calculation of a species’ likelihood of extirpation based upon the
earliest and most recent occurrence records, as well as the number of intermediate sightings within a
given section of the river -- relating the year of the most recent recorded sighting, the year of the
earliest recorded sighting, the year of the latest field data collection (whether this particular
species was sighted or not), and the number of times since the earliest sighting the species has been
recorded. We performed this calculation at each of the five-kilometer, 25-kilometer, 100-kilometer, and 500-kilometer scales.
|
| [top] |
| |
|
| |
|
The impacts of habitat destruction on the dynamics and extinction regimes of ecological communities
constitute an area of increasing concern to conservation biologists and theoretical ecologists alike.
In collaboration with a local academic institution, we studied the impacts of habitat destruction on
extinction dynamics in Lotka-Volterra competition communities. We treated space implicitly by adopting
the notion that habitat loss translates into reductions in species’ carrying capacities within the
region. We found a curvilinear decrease in species persistence with increasing habitat loss. The
severity of extinctions increased with increasing community size and with increasing intensity of
competition. By comparing the fraction of species ultimately going extinct with the fraction observed
going extinct during the course of habitat destruction, we found that an "extinction debt" exists in
these competitive communities. The magnitude of the "extinction debt" increases with increasing species
richness and with increasing intensity of competition. Among species that survived the loss of 95% of
their habitat, competitive hierarchies based on relative abundance were reshuffled in comparison to the
hierarchies in corresponding communities without habitat loss. The extent of this reshuffling
increased with the intensity of competition and the size of the competitive communities. Increases in
those variables also led to increased extinction of more highly ranked competitors, though no
consistent loss of competitive dominants occurred. In concert with other theoretical investigations of
habitat loss from a community perspective, this study suggests that some aspects of extinction dynamics
may be broadly predictable across modeling frameworks whereas others may depend sensitively on what one
assumes about the intensity and nature of interspecific competition.
|
 |
We generated competitive communities of a varying number of species using systems of coupled first-
order Lotka-Volterra differential equations. This model formulation implies spatial homogeneity ("well-
mixedness") in the communities. To generate a range of competing species, we drew intrinsic growth
rates and carrying capacities from uniform distributions of pre-specified ranges.
|
|
|
Intraspecific interaction strength terms (along the diagonal of the Jacobian matrix of each community)
were set to 1 throughout. Interspecific interaction strength terms were drawn from a log-normal
distribution whose median took a range of values in our principal simulations. We assumed this log-
normal distribution because an increasing number of field studies suggest that distributions of
interaction strengths in real communities are skewed such that most interspecific interactions are weak
with respect to intraspecific interactions. The values of median interaction strength we used result
in communities that are for the most part weakly competitive. Only in one case case were we able to
find any interspecific competitive relationship stronger than intraspecific competition. In all model
communities, initial population sizes were set at one half the carrying capacity for each species.
|
|
After an initial transient period, we simulated habitat loss by geometrically decreasing each species’
carrying capacity by 1% every year. This approach implied that the impacts of habitat loss were
proportionately equivalent among all species in the community. Habitat loss continued in this geometric
fashion until either 95% of the habitat was destroyed or all species went extinct. For comparison, we
also numerically solved each community’s system of equations under deterministic conditions without
habitat loss for the same time frame over which the habitat loss models ran.
|
 |
For the habitat loss model, we recorded for each community 1) the sequence and timing of species
extinctions relative to the percent of carrying capacity remaining at extinction and 2) the abundance
of each surviving species immediately after the initial transient period and every 20 years
thereafter. For the deterministic model, we recorded 1) the sequence and timing of species extinctions
and 2) the abundance of each species surviving at 0 and terminal (post transient) years. Species going
extinct in the deterministic model were factored out of the results for the corresponding habitat loss
model to yield a sequence of extinction events attributable solely to habitat destruction. Though the
potential for competitive loops among members of our model communities prohibits unambiguous
delineation of competitive hierarchies like those found in other models of habitat loss, we adopt
relative abundance after the terminal post-transient period as an index of competitive dominance.
To determine the extent of the extinction debt generated by habitat loss in these competitive
communities, we first determined the number of species extinct due to habitat loss after a given
percentage of the habitat had been destroyed. In the absence of any further habitat loss, we then
solved each model community to equilibrium to identify what additional species would ultimately go
extinct. We next solved the corresponding deterministic models to equilibrium to identify and factor
out those species that would ultimately go extinct even in the absence of habitat loss. This approach
allowed us to compare the severity of species extinctions immediately observed as being due to habitat
loss with the extinctions that would ultimately result from habitat loss.
|
 |
For each species going extinct in a habitat loss model that remained extant in the corresponding
deterministic model, we identified its rank abundance (competitive dominance) after the terminal post-
transient period in the deterministic model. To account for differences in the number of species
surviving among replicates of the deterministic models, we scaled these rank abundances to relative
rank abundance where the least abundant species had a rank of 0 and the most abundant species a rank of
1.
|
|
We performed arcsine square-root transformations on these relative rank abundances prior to
statistical analysis. We next used repeated measures analysis of variance on the transformed model
output (treating community size and median interaction strength as factors and communities as
replicates) to examine how community structure influenced species extinction due to habitat loss.
|
|
We used Kendall’s rank correlation analyses to compare the rank abundance of species (the structure of
competitive dominance hierarchies) in corresponding pairs of habitat loss and deterministic communities
after 95% destruction in the habitat loss model. Species extant in the deterministic (or habitat loss)
models that went extinct in the habitat loss (or deterministic) models were eliminated prior to
calculating competitive ranks. We then converted the correlation coefficient output to rank order
(across communities of different size and median interaction strengths) and conducted a Scheirer-Ray-
Hare test, a nonparametric form of two-way ANOVA, to assess whether the degree of agreement in rank
abundance varied according to community size or median interaction strength.
|
| [top] |
| |
|
| |
Metapopulation models offer a means of incorporating spatial structure into species interactions by
defining the local subpopulation or patch as the basic unit of description, thus forcing local dynamics
to be stochastic. This effectively divides available habitat into a grid or other structure through
which subpopulations colonize and become locally extinct over time, and eventually reach some
equilibrium based on population parameters and trophic structure. Habitat destruction, which in
essence removes available patches from the reach of subpopulation colonization, can destabilize these
coexistence equilibria and reorganize the trophic structure of communities, allowing previously
inferior species to dominate the landscape and drive their superior competitors to local extinction.
Our analysis focused on the five-species metapopulation model with two specialist predators (K1 and K2),
a competitively superior generalist predator (K3), and two prey species (P1 and P2),where Pi and Ki
represent the fraction of patches occupied by prey (P) or predator (K) species i, ci and Ci represent
colonization rates per occupied patch for prey and predator species, and ei and Ei represent rates at
hich patches go extinct. For this model prey species one is competitively superior to prey species two
(P1 can replace P2 in a given patch, but not vice-versa).
The coexistence equilibrium for this system is unstable, and collapses to four-species when initial
conditions are perturbed away from equilibrium values. The species which goes extinct depends upon
which initial conditions are changed, and in what direction, although the extinct species is always
either the generalist predator or one of the specialist predators. The generalist predator dies when
P1 or P2 is below equilibrium values, when K1 or K2 is above equilibrium values, and when K3
(generalist) is below equilibrium values. The specialist predator dies out when the above conditions
are reversed. Which of the specialist predators goes to extinction seems to depend most heavily on the
values of c1, the colonization rate for the competitively superior prey species P1.
|
| |
|
Habitat destruction (d) is incorporated by reducing the number of patches available for recolonization,
and is employed as a geometric decrease in the total number of patches. For the specific case of c2=
0.19, e=0.022, C1=0.77, C2=0.75, C3=0.095, E1=0.02, E2=0.02, E3=0.027, and d=0.08, if c1 is greater
than approximately 0.08, specialist predator two (K2) goes extinct; if c1 is less than approximately
0.07, specialist predator one (K1) goes extinct.The invasion criteria for this case (generalist
superior) are given at right.
|
 |
| |
Note that, since the coexistence equilibrium is unstable, even if the above criteria are satisfied and
K1 or K2 successfully invades, one of the five species will still eventually go to extinction. There
seems to be no relationship between the effects of c1 on switching of the specialist predator which
goes extinct and the invasion criteria.
If all other parameters are held constant and only habitat destruction (d) is changed, there is a
linear change in the equilibrium values for all five species (see figure). As d increases, prey
species P1 and P2 and generalist predator species K3 increase, and specialist predator species K1 and
K2 decrease. Equilibrium values for specialist predator K1 are always at lower levels, but for habitat
destruction levels below a certain threshold, the generalist predator K3 is at lower levels than the
specialist predator K2; beyond this threshold, equilibrium values for K3 surpass those for K2.
|
| |
 |
| |
|
| [top] |
| |
|
| |
|
The goal of the USDA - Farm Service Agency's Conservation Reserve Program is to establish long-term,
resource conserving covers on eligible farmland. In order to assess the value of land dedicated to CRP
we teamed up with a group of other small businesses to develop a method to evaluate the impact of CRP
lands on pheasants in North America. By combining regular bird censuses with known distribution of CRP
land in a GIS application, we have been able to develop a visual assessment of the change in pheasant
distributions over time and changes in land use. This project is ongoing, and statistical analyses are
currently being performed on the data produced from the GIS analysis.
|
 |
| |
 |
| [top] |
| |
| |
| |
| |
| |