User manual | EO Specs: Separation Distance for Animals

EO Specs: Separation Distance for Animals
Version 1.0 (29 October 2004)
1. Movement Data and Determination of Separation Distance
Ideally, separation distances should yield occurrences that are, for all practical purposes,
demographically independent. Thus separation distances should be based on information on the
dispersion patterns and dispersal characteristics of a species or group of related species (“EO
Specs Group”). Unfortunately, adequate information on these characteristics does not exist for
most species. However, even when data are available, there is no single, strictly correct measure
for identifying “independent” populations or metapopulations. Demographic and genetic
connections among populations form a continuum rather than discrete categories. Additionally,
due to often substantial ecogeographic and temporal variation in average or modal movement
characteristics of a species, and variations resulting from differences in study methods, any
particular data set is not necessarily the best reflection of a species’ or group’s movement
characteristics or an ideal basis for separation distance determination. So separation distance
generally must reflect dispersion patterns and movement characteristics in a general way rather
than as a precise mathematical outcome of specific data. Consequently, in element occurrence
specifications, certain somewhat standardized separation distances (e.g., 5, 10, 15, 20 km) have
been adopted. Although separation distances are necessarily fairly arbitrary, they do generally
attempt to reflect mobility similarities and differences among species or groups of species,
except in the case of birds and other highly mobile species for which separation distances must
be much smaller than movements would indicate such that occurrences are of practical size for
conservation purposes (such occurrences do not attempt to represent populations or
metapopulations).
2. Minimum Suitable-Habitat Separation Distance
For most species, occupied locations separated by short distances of suitable habitat not known
to be occupied generally do not represent discrete populations; usually the intervening area is
occupied (if not in the short-term then usually over a period of several years), or at least it is not
a significant discontinuity between populations. Thus the minimum suitable-habitat separation
distance should not be too small. Accordingly, the minimum suitable-habitat separation distance
is 3-5 kilometers for most species and groups, and 1 or 2 kilometers for only the most sedentary
ones. (The standardized minimum separation distance for suitable and unsuitable habitat is 1
kilometer; see EO Data Standard.) These values, though necessarily arbitrary, are intended to
reduce the incidence of inadvertent splitting of populations and metapopulations into multiple
occurrences.
1
3. Habitat Suitability and Occurrence Separation
The following definitions have been adopted for distinguishing suitable and unsuitable habitat:
Suitable: habitat capable of supporting reproduction or used regularly for feeding or other
essential life history functions; a habitat in which you would expect to find the species (assuming
appropriate season and conditions); includes marginally suitable habitat that is contiguous with
clearly suitable habitat.
Unsuitable: habitat through which the species may successfully disperse but that cannot support
reproduction or long-term survival.
Use the following guidelines to determine whether to use the separation distance for suitable
habitat or unsuitable habitat when circumscribing occurrences:
Habitat thoroughly surveyed
Habitat inadequately surveyed
Apparently suitable
Suitable
Suitable
Suitability uncertain
Unsuitable
Suitable
Apparently unsuitable
Unsuitable
Unsuitable
Note that treatment of inadequately surveyed habitat as suitable or unsuitable depends on a
subjective assessment of whether or not the habitat is likely to be suitable. If the habitat seems to
be more likely to be suitable than unsuitable, or if suitability is completely unknown, treat it as
suitable.1
4. Overriding the Separation Distance Criteria
Circumstances occasionally indicate that patches of occupied habitat (i.e., source features) that
are farther apart than the nominal separation distance should be joined as parts of a single
occurrence. For example, if the separation distance for a riverine species is 10 km and patches of
occupied habitat along a river are known to be sequentially separated by distances of 5, 8, 7, 9, 2,
4, 12, 6, 3, 5, and 8 km, and there are no barriers present and no reason to believe that more than
one population is involved, then it makes biological sense to include all of the patches along the
river in a single occurrence despite the existence of a gap that is in excess of the separation
distance. Similarly, with arid country taxa especially, but also others, it is reasonable to cluster
all populations in an obvious landscape feature such as a canyon or along a stream as one
occurrence even if the separation distances are somewhat violated. Joining as a single occurrence
patches that are beyond the separation distance is particularly appropriate when the apparent gap
that exceeds the separation distance has not been adequately surveyed.
1
In contrast, botanists in the heritage network treat habitat of unknown suitability as unsuitable [explanation to be
added]
2
Separation distances should be overridden only when professional consensus indicates that it is
more biologically appropriate to do so than to establish multiple occurrences. Additionally, as a
general rule, patches of occupied suitable habitat that are more than approximately 1.5 times the
separation distance from the nearest occupied patch should not be recorded as parts of the same
occurrence. However, in vast, relatively uniform landscape units for which survey data are
sparse, the nominal separation distance may be increased even further, if it is more biologically
reasonable to do so than to establish multiple occurrences.
If it appears that overriding the separation distance seems to be appropriate on a regular basis,
please contact the responsible NatureServe staff in order to discuss the possibility of increasing
the separation distance. Changing a separation distance may affect occurrences and data
management in multiple states/provinces, so changes are made only if deemed appropriate by
consensus of responsible NatureServe staff and affected subnational jurisdictions.
5. Alternative Separation Procedures
For some species (or EO specs groups), occurrence separations may be based on an alternative
procedure rather than the two usual distance values (for suitable and unsuitable habitat). For
example, occurrence specifications may provide separation distance values for more than two
categories of habitat, or they may employ a qualitative method (e.g., based on hydrographic units
or population migration patterns) rather than numerical distances for distinguishing occurrences.
For some highly mobile, occurrence-tracked species for which locational data are typically
recorded as points or tiny polygons (e.g., bald eagle nests), EO specifications may indicate that
every location or territory qualifies as an occurrence. This simplified procedure may be
particularly appropriate for species characterized by a large number of locations over a wide
area. For these, appropriate numerical separation distances would be very large and occurrences
would be impracticably large for all but the most esoteric uses. Using a small separation distance
would yield smaller but completely arbitrary occurrences, and ranking these would be
meaningless.
6. Separation Distance and Occurrence Mapping
Separation distance does not affect how occupied patches of habitat (source features) are
mapped, but it does alter how these patches are allocated among occurrences. For example, two
distinct occupied patches that are separated by a gap of 4 kilometers are mapped as two polygons
regardless of whether the separation distance is 3 kilometers or 5 kilometers. With a 3-kilometer
separation distance, they would represent two occurrences, whereas a 5-kilometer separation
distance would yield one occurrence with multiple polygons. Occurrence boundary delineation is
a process that is independent of separation distance; see “Mapping Guidelines.”
3
EO Specs: Mapping Guidelines for Animals
Version 1.0 (29 October 2004)
1. Mapping Occurrences
Occurrence mapping usually involves the translation of point locations (e.g., collection sites,
locations of brief observations) into one or more polygonal source features. A basic guideline for
occurrence mapping is that mapped features should be based on what we know rather than what
we suspect about the distribution of a species in an area. Documented observation locations are
the primary basis for mapping an occurrence. Interpolation (mapping of habitat between adjacent
known points or locations) usually is appropriate, whereas extrapolation (mapping of areas
outside the known locations) should be limited. Thus mapped occurrences generally are
conservative representations of occupied (or formerly occupied) habitat (contrast with Inferred
Extent; see section 2).
Occurrence mapping often is not as straightforward as it may sound. Over the years, biologists in
the heritage network have adopted a wide range of protocols to address ambiguous mapping
situations. Some highly conservative approaches map only points that represent known collection
or observation locations. Others regard one or a few observations as an appropriate basis for
mapping a large patch of suitable habitat as the occurrence. Given identical information,
different biologists may produce dramatically different mapped occurrences. Such
inconsistencies confound multijurisdictional status assessments and inhibit effective range-wide
conservation planning. To increase mapping consistency across the network, we here
recommend some standardized mapping policies. These recommendations cover some frequently
encountered mapping situations, but the wide range of potential circumstances makes it
impossible to cover all mapping situations. We anticipate that ad hoc mapping decisions will
remain a common practice, but we encourage those who have adopted policies for situations not
covered by this document to share their experiences with other programs (e.g., via the zoology
listserve).
A. Map the whole patch or not?
A single location in a small patch of suitable habitat may be sufficient to justify mapping the
entire patch as the occurrence. For any patch that is not larger than a typical or average home
range size or (for species with limited mobility) not longer than an ordinary dispersal distance,
the entire patch should be mapped as the occurrence. Information on home range size or
dispersal characteristics (if available) can be found in the EO Specifications or Characterization
files for each species or EO Specs group.
The appropriate size of the mapped occurrence becomes ambiguous in situations characterized
by just one or a few locations in a small portion of a large patch of contiguous suitable habitat
(larger than the typical or average home range size). In such cases, we recommend that the
occurrence be mapped as a polygon that minimally includes all locations that are not separated
by a gap larger than the separation distance. In addition, depending on locational uncertainty, this
polygon may also include a variable amount of suitable habitat outside a minimum convex
4
polygon defined by available observation locations. However, if the area of the mapped feature
includes more than 70 percent of the total contiguous area of suitable habitat, then include the
entire patch of suitable habitat in the occurrence, using unsuitable habitat to define the boundary.
Note: The value of 70 percent serves as an arbitrary threshold deemed appropriate as an indicator
that, for all practical purposes, the entire patch likely is occupied.
The boundaries of patches of occupied suitable habitat sometimes can be readily identified by
their abrupt transitions into adjacent areas of unsuitable habitat (e.g., wetland/upland edge, or a
forest/grassland edge). In other situations, suitable habitat may grade gradually into unsuitable
habitat. In these latter circumstances, occurrences should include only suitable habitat and
contiguous marginally suitable habitat.
Suitable habitat may be interrupted by narrow or wide areas of unsuitable habitat that are smaller
than the separation distance. How do these areas affect what is mapped as part of an occurrence?
Basically, this is a subjective determination that should take into consideration the relative size
and nature of the suitable and unsuitable habitats and the behavioral characteristics of the
species. Narrow habitat interruptions should be ignored if those areas are relatively
inconsequential discontinuities in an otherwise continuous tract of suitable habitat.
B. Map it as part of the occurrence, or treat it as inferred extent?
Frequently there is an substantial gray area between what we know for certain and what may be
inferred about the extent of occupied habitat. For example, we often have data on the breeding
locations of pool-breeding but otherwise terrestrial amphibians, or the lek sites of grouse, but
usually we lack adequate site-specific information on the occupied upland habitat of the
amphibians or the nesting and feeding sites of the grouse. Nevertheless, these other habitats
adjacent to breeding pools and lek sites are absolutely required and necessarily occupied, so we
“know” that they are part of the occurrence even if we lack specific information on the full
extent and precise occupancy of that adjacent occupied habitat. Therefore, based on available
data for these examples, suitable habitat extending several kilometers around a sage grouse lek
site, or at least a few hundred meters around a wood frog breeding site, should be included in the
mapped representation of the occurrence. This area, regarded as a biologically appropriate
extrapolation, may be defined, and its biological basis is described, in the mapping guidance
section of the occurrence specifications for the species or group. Beyond this area, additional
habitat that is thought to be part of the occurrence but for which there is greater uncertainty, may
be represented in an Inferred Extent polygon or in a predicted distribution layer for the species.
2. Inferred Extent
An inferred extent (IE) polygon and a mapped occurrence (represented by a point, line, or
polygon) are independent but related spatial representations. An occurrence represents known
occupied habitat. An inferred extent polygon is a representation of the known occupied habitat
plus additional habitat surrounding that area. Conceptually, inferred extent for an occurrence
consists of one or more polygons that represent the area likely encompassed by the occurrence,
as inferred from mapped source features (known occupied habitat) and knowledge of the biology
of the species. For example, the IE polygon for an occurrence based on a single bog turtle
5
observation in a fen could encompass the entire fen and any closely adjacent fens, as well as
presumed migration/dispersal corridors. The outer extent of the IE polygon should be kept within
a reasonable distance of the known occupied habitat (i.e., the occurrence itself), generally not
more than the maximum known single-year migration distance for the species (assumes species
is nonvolant), and certainly not more than the separation distance. It should exclude areas not
likely to be regularly occupied. The extent of the IE polygon should be determined by a biologist
experienced with species and its habitat. As a projection of an occurrence's probable extent, the
IE polygon is useful for environmental review, especially when data limitations result in
excessively conservative approximations of occurrence dimensions.
6

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project