2459
CONCEPTS & SYNTHESIS
EMPHASIZING NEW IDEAS TO STIMULATE RESEARCH IN ECOLOGY
Ecology, 80(8), 1999, pp. 2459–2473
᭧ 1999 by the Ecological Society of America
PALEOREFUGIA AND NEOREFUGIA: THE INFLUENCE OF
COLONIZATION HISTORY ON COMMUNITY PATTERN AND PROCESS
JEFFREY C. NEKOLA1
Curriculum in Ecology, 229 Wilson Hall, CB 3275, University of North Carolina,
Chapel Hill, North Carolina 27514 USA
Abstract. Two types of biological refugia (habitats that support populations not able
to live elsewhere in a landscape) can be defined from relative refugium age as compared
to surrounding matrix age; paleorefugia are now-fragmented relicts of a formerly widespread
matrix community, whereas neorefugia have formed more recently than the matrix. This
difference should make extinction a relatively more important process in determining species
occurrence in paleorefugia, whereas immigration should be relatively more important
in neorefugia. Based on these differences, a series of eight a priori predictions relating to
the diversity and distribution patterns for the biota of such sites can be generated: (1) the
slope of the species–area relationship, and amount of variance explained by it, should be
greater in paleorefugia as compared to neorefugia; (2) the negative relationship between
habitat isolation and species richness should be stronger in neorefugia as compared to
paleorefugia; (3) species richness should be expected to decrease over time in paleorefugia,
but to increase over time in neorefugia; (4) the inverse correlation between site distance
and community similarity (distance decay) should be stronger in neorefugia as compared
to paleorefugia; (5) neorefugia should be enriched in highly vagile species relative to
paleorefugia, whereas paleorefugia should be rich in less vagile species relative to neorefugia;
(6) geographic factors should be more important predictors of species occurrence
for neorefugia than for paleorefugia; (7) paleorefuge sites should possess more and stronger
correlations between community composition and environmental covariables (such as soil
chemistry, climate, etc.) as compared to neorefuge sites; and (8) the number of competitive
co-equivalents held within a system of neorefugia should be greater then the number held
within a series of paleorefugia.
The most readily testable predictions (numbers 1, 2, 4, and 5) were evaluated by comparing
species-richness and community-composition patterns within two northeastern Iowa
refugia: algific talus slopes (paleorefugia) and fens (neorefugia). Results from these tests
were consistent with predictions. These results illustrate that colonization history may
influence contemporaneous species diversity and community-composition patterns. They
also suggest that (1) equilibrium has yet to be achieved in the example systems after 5000–
10 000 yr, (2) the ecological-biogeographic debate centered around the mutual exclusivity
of vicariance and dispersal is intrinsically flawed, and (3) optimum reserve-design strategies
for biodiversity protection within paleorefuge and neorefuge systems will differ.
Key words: biodiversity; biogeography, historical; colonization history, influence on community
pattern; conservation biology; dispersal; fens; Iowa (USA), northeastern; island biogeography; refugia,
paleorefugia vs. neorefugia; species richness; talus slopes, algific; vicariance.
INTRODUCTION
Past environmental changes have caused repeated
fluctuations in species and habitat distributions that can
Manuscript received 6 February 1998; revised 10 November
1998; accepted 27 November 1998; final version received
21 December 1998.
1 Present address: Department of Natural Applied Sciences,
University of Wisconsin, 2420 Nicolet Drive, Green Bay, Wisconsin
54311 USA. E-mail: nekolaj@uwgb.edu
lead to formation of island-like habitats and isolated
populations through two contrasting pathways. They
may represent remnants of once-more-widespread distributions
that have become fragmented, or they may
represent the de novo development of habitats or populations
in a landscape where they were previously
absent. Borrowing terminology used for endemic species
(Stebbins and Major 1965, Krukeberg and Rabinowitz
1985), I refer to habitats that have been colonized
through these processes, respectively, as ‘‘paleo-
2460 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
FIG. 1. Diagrammatic representation of the development of paleorefugia and neorefugia. At time period A, environmental
conditions are conducive to the existence of a particular habitat (shaded). After critical environmental change (time period
B), this habitat has retreated completely from the focal landscape (inner box) in the case of neorefugia but persists as isolated
patches in the case of paleorefugia. By time period C, the environment has changed again, allowing the habitat to exist again
within portions of its original range. At time period C, paleorefugia are found within essentially the same microenvironmentally
buffered patches present at time period B. Neorefugia, in contrast, have formed as appropriate microhabitats develop in
isolation from the main habitat area.
refugia’’ and ‘‘neorefugia.’’ More specifically, paleorefugia
represent habitats that are older than the
surrounding biological matrix, whereas neorefugia represent
habitats that are younger than the surrounding
biological matrix. Such habitats are considered refugia
as they support communities or populations unable to
survive elsewhere in the landscape (i.e., Pielou 1979).
As these terms reflect the relative age of a refugium to
its matrix, a paleorefugium in one landscape may be
of more recent origin than a neorefugium in another.
It is also possible that a particular habitat type that
developed as a paleorefugium in one landscape may
be a neorefugium elsewhere.
In this paper, I ask how such paleorefugia and neorefugia
would be expected to differ in their contemporaneous
species richness, species distribution, and
community-composition patterns. I develop eight predictions
about these differences and then investigate
the four most easily testable ones with data collected
from two types of insular habitats that co-occur in
northeastern Iowa.
THE INFLUENCE OF EXTINCTION AND IMMIGRATION
The relative importance of extinction and immigration
processes varies greatly between paleorefugia and
neorefugia. First, consider the development of paleorefugia
sites. Before environmental change, a given
community is continuous over a wide geographic extent,
harboring a diverse and well-mixed biota potentially
near equilibrium with the environment. Following
environmental change, the habitat becomes restricted
to a few discrete patches in the landscape (Fig.
1). As the community becomes fragmented, habitat size
decreases and isolation between fragments increases,
leading to increased extinction rates on individual fragments
(MacArthur and Wilson 1967, Diamond 1975,
Bierrgaard et al. 1992). The richness of individual sites
falls, with larger fragments retaining more taxa than
smaller ones (Diamond 1975). Although some recolonization
from nearby sites undoubtedly occurs (the rescue
effect of Brown and Kodric-Brown [1977]), this
should not greatly influence large-scale compositional
pattern as nearby sites should have been originally colonized
from similar species pools.
In contrast, neorefugia sites are created through the
development of novel environmental conditions in a
landscape (compare time periods B to C in Fig. 1).
Consequently, the composition of neorefugia must be
a product of immigration. Factors that effect immigration,
such as habitat isolation or age, should therefore
exert influence on community composition and species
richness (Carlquist 1974, Diamond 1975, Bush and
Whittaker 1993). Although extinction certainly occurs
on these sites as newly recruited populations disappear,
this process should be of limited importance as populations
can only become extinct if a successful immigration
has previously occurred, and as incomplete
dispersal between sites will limit rates of competitive
exclusion (Shmida and Ellner 1984).
The differential contributions of extinction and immigration
have been empirically documented for various
insular habitats. Fernald (1925) hypothesized (incorrectly)
about the role played by unglaciated nuntacks
(reputed paleorefugia) in the development of eastern
North American species ranges. Diamond (1973,
1975) noted that the species richness of land-bridge
islands (paleorefugia) is primarily determined by extinction
rates. Brown (1971) also discussed the role of
extinction in shaping rodent faunas of Great Basin
montane and alpine communities, which became frag-
December 1999 2461NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
mented following the close of Wisconsinan glaciation
(paleorefugia). The importance of immigration in the
development of the biota of young volcanic islands
(neorefugia) has been shown on Krakatau, Long, and
Ritter islands in the eastern Pacific (Diamond 1975,
Bush and Whittaker 1993). Similar patterns were demonstrated
by Simberloff and Wilson (1969) in their manipulative
experiments of mangrove islands. Research
on the importance of immigration in shaping community
composition and structure has also been referred
to as ‘‘supply-side ecology’’ (Roughgarden et
al. 1987). However, none of these previous works have
attempted to compare the contemporaneous ecological
patterns of isolated communities that possess different
colonization histories and that have, therefore, been
exposed to different levels of extinction and immigra-
tion.
PREDICTIONS ABOUT SPECIES DISTRIBUTION, SPECIES
RICHNESS, AND COMMUNITY COMPOSITION
My primary assumption is that across a large range
of taxonomic groups, extinction will most influence the
species composition of paleorefugia, whereas immigration
will most influence the species composition of
neorefugia. From this assumption, I have developed
eight a priori expectations regarding differential species
richness, distribution, and community-composition
patterns:
(1) All else being equal, the slope of the species–
area relationship, and the amount of variance explained
by it, should be greater in paleorefugia than in neorefugia.
Paleorefugia can be expected to have steeper
species–area curves than neorefugia, because for a given-size
island the paleorefuge will be approaching
equilibrium number through species loss, whereas the
neorefuge will be approaching this number through
species accretion. Paleorefugia were originally more
connected, and presumably once harbored similar numbers
of species per unit area. Consequently, differences
in species richness should be largely a function of postfragmentation
extinction rates. Such rates are believed
to be strongly correlated with habitat size (e.g., Diamond
1975, Bierrgaard et al. 1992). The species richness
of neorefugia, however, should also be a function
of immigration processes, which are related less to habitat
size than to other factors like habitat age and isolation.
Consequently, paleorefugia can be expected to
have less variance around the species–area curve than
neorefugia. This prediction may be violated for neorefugia
that have remained isolated over evolutionary
time scales, as strong species–area relationships may
be generated on island habitats through speciation
(Carlquist 1974).
(2) All else being equal, the negative relationship
between habitat isolation and species richness should
be stronger in neorefugia than in paleorefugia. I expect
this as propagule movement (and hence immigration
rate) is largely a function of barrier width (Okubo and
Levin 1989). Thus, less-isolated neorefugia should
have access to a larger propagule pool than more-isolated
ones. This result would not be anticipated from
paleorefugia as, over limited geographic extents, sites
were likely colonized from similar species pools. As
such, contemporaneous site isolation should not strongly
affect richness. Isolation could influence paleorefugia
richness when recolonization from nearby sites
(Brown and Kodric-Brown 1977) is important.
(3) Species richness should decrease over ecological
time scales in paleorefugia, while increasing over similar
time scales in neorefugia. Extinction, or ‘‘relaxation’’
(sensu Diamond 1975), should eliminate taxa
from paleorefugia, as is typical for land-bridge islands.
Such taxa may include those that are not well adapted
to local site conditions, that are competitively inferior
(e.g., Tilman 1988), or that exhibit area sensitivity
(McDonald and Brown 1992). However, as neorefugia
form de novo out of a hostile landscape, they will initially
be void of species. Over time their richness
should increase as taxa from the surrounding species
pools immigrate onto sites (Zobel 1997).
It is also possible that over time the species richness
of neorefugia and paleorefugia will converge. As time
increases, immigration between neorefugia should become
more and more complete, thereby increasing the
importance of extinction in the structuring of site biotas
and strengthening the species–area relationship. On paleorefugia,
the stochastic loss of species may eventually
lower site richness to levels anticipated for neorefugia.
It is not clear what factors will most strongly
influence this rate of convergence.
(4) The inverse correlation between site distance and
community similarity (distance decay) should be stronger
for neorefugia than for paleorefugia. A fundamental
principle of geography is the inverse correlation between
similarity and intersample distance (Tobler
1970). In the geographical literature this decrease in
similarity has been termed ‘‘distance decay’’ and been
applied to phenomena as diverse as human communication
networks, migration patterns (Bennett and
Gade 1979, Fotherinham 1981), and spatial interpolation
(Burrough 1986). The rate of distance decay
among natural communities has been shown to vary
with the dispersal strategy of organisms and with the
degree of landscape fragmentation (Nekola and White,
in press).
I expect that the rate of compositional distance decay
for paleorefugia should primarily reflect pre-fragmentation
rates as long as no strong environmental gradients
are present in the current landscape. Consequently,
in most cases the rates of compositional distance decay
for paleorefugia should be low within regions. However,
because immigration is required for development
of neorefugia biotas, sites across a region may be colonized
through different species pools when source areas
or migration corridors differ. As a result, similarity
2462 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
of neorefugia communities should be more strongly
correlated with intersample distance.
(5) Neorefugia should be enriched in highly vagile
species relative to paleorefugia, whereas paleorefugia
should be enriched in less vagile species relative to
neorefugia. I expect that species with good dispersal
abilities will be more frequently represented in neorefugia,
as colonization on such sites is only possible
through the crossing of dispersal barriers–effectively
eliminating poor dispersers while increasing the relative
importance of good dispersers in the neorefugia
species pool. Poor dispersers not only can be a component
of paleorefugia biotas (as sites were colonized
prior to fragmentation), but may be favored over time
on these sites if they lose a smaller percentage of propagules
beyond site boundaries (Carlquist 1974), and
have larger, more competitively fit progeny (Harper
1977). These mechanisms have been suggested by Carlquist
(1974) to explain the higher frequency of poorly
dispersing plant and animal species on remote oceanic
islands.
(6) Geographic factors (such as distance to nearest
population) should be more important predictors of
species occurrence on neorefugia over ecological time
scales. All paleorefuge species (theoretically) had similar
access to sites prior to fragmentation. For this reason,
contemporaneous geographic factors should be of
relatively limited importance in predicting current species
distribution. Those environmental factors that directly
affect the survival of populations should be of
pronounced importance in predicting species occurrence
on these sites due to the prolonged competitive
sorting that has occurred there. However, the distance
to contemporaneous propagule sources should be an
important predictor of species occurrence on neorefugia,
as the probability of colonization will increase
with decreasing dispersal distance. Additionally, direct
environmental factors may be less important in predicting
species occurrence in neorefugia due to increased
habitat breadths that result from limited competitive
sorting on these sites and incomplete immigration
of potential competitors (Cox and Ricklefs
1977).
(7) The higher levels of competitive sorting in paleorefugia
should lead to a stronger tie between environmental
and compositional gradients (Peet and Christensen
1988) than in neorefugia. Although community
composition within paleorefugia should initially contain
more and stronger correlations with environmental
covariables than neorefugia, convergence can be expected
over ecological time.
(8) The number of competitive co-equivalents held
within an archipelago of neorefugia should be greater
than that held within a similar set of paleorefugia. The
more intense competitive sorting within and the lower
levels of pre-fragmentation isolation between paleorefugia
should have resulted in similar initial compositions
and a strongly competitive environment. As a
consequence, there should have been little opportunity
for establishment of equivalents with varying fitnesses
(Shmida and Ellner 1984). However, the always-isolated
nature of neorefugia should allow for less complete
intersite dispersal, making it easier for less fit coequivalents
to persist in the landscape.
As is true for many ecological processes that operate
over large spatial and temporal extents, collection of
data to test these hypotheses may be difficult, or in
some cases impossible. And, even if appropriate data
exist, it may be equally difficult to control for confounding
factors (Brown 1995). This should not imply,
however, that hypotheses regarding differential processes
between paleorefugia and neorefugia are inherently
unfalsifiable. By comparing current ecological
patterns between sites within the same geographic region,
confounding effects of the local environment, climatic
history, and migration history can be at least
partially controlled.
The most easily tested hypotheses are those that concern
directly quantifiable variables such as current species
richness, community composition, habitat size,
habitat isolation, and habitat location. Consistency with
Hypothesis 1 can be easily tested as it is based on a
comparison of species richness to habitat area. Hypothesis
2, which is based on species richness and siteisolation
data, can also be readily assessed. Hypothesis
4 can also be easily tested, as it only requires community-composition
and geographic-location data.
Lastly, Hypothesis 5 can be easily assessed by investigation
of the dispersal strategies exhibited for paleorefuge
and neorefuge species.
Hypothesis 7 can also be relatively easily tested, as
it is based on analysis of how quickly and predictably
composition changes over a given amount of environmental-gradient
space. Testing of this question is
slightly more complex, however, as the documentation
of these patterns requires use of multivariate data-reduction
procedures such as ordination. It is possible
that use of different data-reduction techniques could
produce different outcomes. Previous knowledge of
which environmental variables are critical in the system
may also be necessary.
Hypotheses 6 will be more difficult to test. To assess
the effect of geography on species occurrence, it is first
necessary to control for variation in distribution related
to environmental gradients. Although in some cases it
may be relatively easy to quantify this through use of
parametric or nonparametric regression techniques
(Austin et al. 1990), such analyses will be problematic
when environmental variables demonstrate strong spatial
autocorrelation or when habitats are clustered in a
landscape. In such situations it is difficult to know what
the null expectation will be for the effect of geography
on species occurrence, independent of environmental
variation. While these problems are likely tractable
through use of Monte Carlo or other randomization
December 1999 2463NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
FIG. 2. Cross-section view of a Blanding
Formation algific talus slope showing major geomorphological
features (adapted from Frest
[1991: Fig. 2]). A complete system has small
sinks in the forested upland; vertical mechanical
karst fissures in the Hopkington Formation; ice
reservoirs in mostly horizontal mechanical karst
in the thin-bedded Blanding Formation; underlying
relatively impervious Tete des Mortes,
Mosalem, and/or Maquoketa formations.
techniques (Hilborn and Mangel 1997), the exact form
of these tests have yet to be formalized.
Hypotheses 3 and 8 will both be hard to test. Determination
of absolute habitat age will be difficult, if
not impossible, to measure in most cases. Even when
it is possible to age habitats through 14C dating of organically
rich basal sediments (peatlands, for example),
it would be difficult or impossible to document total
species richness over this time period. Hypothesis 8 is
difficult to address as it requires determination of which
species represent co-competitors. It is not clear how
one could adequately assess this without knowing, for
a given set of species, all interactions between all potential
limiting environmental variables.
Based on these factors, a priori Hypotheses 1, 2, 4,
and 5 appear to be the most easily testable. To provide
a preliminary assessment for these hypothesized differences,
I have used a paleorefugia system (algific
talus slopes; Frest 1982) and neorefugia system (fens;
Nekola 1994) that co-occur in northeastern Iowa to
examine four specific predictions: (1) the species–area
relationship is steeper and stronger in algific slopes as
compared to fens; (2) the species–isolation relationship
is stronger in fens as compared to algific talus slopes;
(3) distance-decay patterns are more important in fens
as compared to algific talus slopes; and (4) the occurrence
frequency of good and poor dispersers varies
between algific talus slopes and fens.
METHODS
Study sites
Algific talus slopes.—Algific talus slopes (‘‘algific’’
coined from the Greek root ‘‘algos’’ and translates to
‘‘cold producing’’) occur within a deciduous-forest matrix
on steep, usually north-facing, carbonate talus
slopes where cold air flows out of ice-filled caves (Frest
1981; Fig. 2). These caves developed through ice wedging
associated with periglacial activity (Hedges 1972),
and have maintained a below-freezing mean annual
temperature through thermal buffering provided by
overlying cliff-forming carbonates and subtending talus
slopes (Frest 1981, 1982). Algific talus slopes are
characterized by an unique buffered microclimate
where soil temperatures rarely exceed 15Њ C in the summer
(Frest 1982). Approximately 300 algific talus slope
sites have been located across a 10-county region of
northeastern Iowa (USA) in exposures of Prairie-duChien,
Galena, Blanding, or Hopkington Formation
carbonates (Fig. 3; Frest 1981, 1982, 1983, 1984,
1986a, b, 1987). A limited number of sites are also
known from adjacent portions of Minnesota, Wisconsin,
and Illinois, USA (Frest 1991).
Algific talus slope communities range from densely
forested to moderately open. Characteristic woody species
include Acer saccharum, A. spicatum, Abies balsamea,
Betula lutea, B. papyrifera, Fraxinus americana,
Juglans cinerea, Pinus strobus, Salix bebbiana,
Sambucus pubens, Taxus canadensis, and Tilia americana.
The ground layer is typically dominated by a
dense bryophyte cover that supports large populations
of pteridophytes such as Cystopteris fragilis, Gymnocarpium
robertianum, and Equisetum scirpoides.
Woody plant cover decreases and bryophyte/pteridophyte
cover increases in general with increasing intensity
of cold-air seepage. Algific talus slopes harbor populations
of over 60 vascular plant species that are disjunct
in Iowa from wet woods and bogs in northern or
western boreal forests, including Adoxa moschatellina,
Carex media, Cornus canadensis, Linnaea borealis,
Mertensia paniculata, Poa paludigena, Pyrola asarifolia,
Rhamnus alnifolia, Ribes hudsonianum, Streptopus
roseus, and Viola renifolia (Thorne 1964, Frest
1982). At least three near-endemic vascular plant taxa
(Aconitum noveboracense, Chrysosplenium iowense,
and Sullivantia renifolia) occur on these sites. Algific
talus slopes also harbor populations of at least eight
2464 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
FIG. 3. Distribution of algific talus slopes and major bedrock units in northeastern Iowa, USA.
land snail taxa once thought to have become extinct at
the end of the Wisconsinan (Frest 1991). Relict mites
in the family Rhagidiidae have also been discovered
on these sites (V. Ruzicka, personal communication).
These taxa were components of a full-glacial community
associated with quasi-maritime climates found
south of the ice margin (Frest and Dickson 1986). Populations
became restricted to algific talus slopes when
this climatic regime disappeared following the retreat
of glacial ice from the continental interior (Frest 1981).
Due to the microclimatic buffering provided by coldair
seepage, which mimics the full-glacial regional climate,
populations were able to persist on algific talus
slopes through the Hypsithermal to the cooler and wetter
conditions of modern times. Algific talus slopes thus
represent paleorefugia whose biota predates the surrounding
deciduous forest matrix.
Fens.—In northeastern Iowa, fens occur within a
tallgrass-prairie matrix at sites of local groundwater
discharge from pre-Illinoian tills, bedrock, aeolian
sands, fluvial sands, or oxbow-meander cutoffs (Nekola
1994, Fig. 4). Sites typically occur as hillside terraces,
or more rarely as flat expanses in low swales or
mounds. The constant issue of groundwater at fen sites
creates a buffered habitat somewhat similar to algific
talus slopes, with soil temperatures being cooler in the
summer, warmer in the winter, and with more constant
soil moisture than is otherwise found within the surrounding
landscape. Fens are scattered across a 30county
region where at least 2333 sites occurred as
recently as 50 yr ago and where 160 sites remain extant
(Fig. 5). The bulk of presettlement and extant fens are
found on the Iowan Erosional Surface, which was
formed during the Wisconsinan through intense periglacial
erosion that removed older tills, created a
stepped landscape surface, and prevented the accumulation
of deep loess deposits (Hallberg et al. 1978,
Prior 1991). Similar fen habitats are known from northwestern
Iowa (Anderson 1943), Illinois (Moran 1981),
Minnesota (Coffin and Pfannmuller 1988), South Dakota
(Ode 1985), and Wisconsin (Curtis 1959).
In northeastern Iowa, fen habitats are dominated by
a dense sedge turf, which includes Carex buxbaumii,
C. interior, C. lanuginosa, C. lasiocarpa, C. prairea,
C. stricta, C. suberecta, and C. tetanica. In discharge
zones Scirpus validus, Typha angustifolia, and T. la-
December 1999 2465NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
FIG. 4. Cross section of the five geologic features that
give rise to fen formation in northeastern Iowa: (A) pre-Illinoian
till sites, with groundwater seepage emanating from
sand and gravel lenses interbedded between till units; (B)
bedrock sites, with groundwater seepage emanating from
fractured carbonate, sandstone, or shale units; (C) aeolian
sand dune sites, with groundwater seepage emanating from
the contact between sand and an impervious till or paleosol;
(D) fluvial sand sites, with groundwater seepage emanating
from the base of fluvial or outwash terraces; and (E) oxbow
sites, which form through hydric succession of cutoff oxbow
meanders.
tifolia are often dominant. Scattered clumps of low
shrubs are also frequent, and include Cornus stolonifera,
Salix bebbiana, S. discolor, S. petiolaris, and S.
rigida. In areas of highest soil saturation, vascular-plant
growth often becomes stunted and sparse, with bryophytes
becoming dominant. Approximately 50% of the
vascular-plant species of these sites are regionally rare
(Nekola 1994). Over 80 of these species are disjunct
from open peatlands in boreal or northeastern North
American, including Aster junciformis, Betula pumila,
Carex sterilis, Epilobium strictum, Eriophorum angustifolium,
Galium labradoricum, Gentiana procera, Lobelia
kalmii, Menyanthes trifoliata, Mimulus glabratus
var. fremontii, Parnassia glauca, Rhynchospora capillacea,
Salix candida, S. pedicellaris, Solidago uliginosa,
Triglochin maritimum, and T. palustris. Northeastern
Iowa fens also harbor populations of 19 rare
Iowa butterfly and skipper species (Nekola 1994), and
seven rare Iowa land snail species (Frest 1990).
14
C dates of basal peat layers indicate that northeastern
Iowa fens initiated peat development following
the end of Hypsithermal warming, less than 6000 yr
ago (Thompson 1992). Although a few sites possess
basal peat dates extending to 10 000 yr ago (Thompson
and Bettis 1994), the presence of oxidized layers within
these beds (Hall 1971, Van Zant and Hallberg 1976)
indicate that deposition was not continuous, and that
during the Hypsithermal even these sites dried out.
Thus, although fens may have occurred at modern locations
in the late-glacial landscape, the geological record
indicates that they did not maintain environmental
buffering during the Hypsithermal. Because of this,
isolated populations of boreal taxa occurring within
these sites cannot represent late Pleistocene or early
Holocene relicts, but are instead relatively recent immigrants.
In northeastern Iowa, therefore, fens represent
neorefugia that are younger than their surrounding
tallgrass-prairie matrix.
Data sets
Species richness.—The ecological distance between
fen and algific talus slope habitats and the surrounding
matrix habitats is not so great as to preclude occurrence
of matrix taxa. Such species can colonize sites via
short-range dispersal or mass effect (Shmida and Ellner
1984), independent of the colonization history of a given
site. As such, inclusion of these ubiquitous or waif
species could seriously swamp any potential ecological
differences caused by differential colonization histories.
By limiting analysis to only those species restricted
to fens or algific talus slopes, this source of noise
can be filtered out. For this reason, observations of
species composition and richness for fen and algific
talus slope sites was limited to the universe of vascularplant
taxa restricted (Ͼ80% of known northeastern
Iowa occurrences) to either of these habitats. In total,
80 restricted vascular-plant species were identified
from fens and 51 from algific slopes, based on occurrence
records reported in Frest (1982), Howe et al.
(1984), and Nekola (1990, 1994), as well as field data
collected during this study. The presence or absence of
each of these taxa was subsequently noted for all 160
extant northeastern Iowa fens and all 76 extant algific
talus slopes in the Blanding Formation outcrop region.
Habitat size.—Sizes of fen sites were estimated
through digitization of soil pedons (named variously
2466 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
FIG. 5. Presettlement and 1991 distribution of northeastern Iowa fens within major landform boundaries, based on Prior
(1991).
December 1999 2467NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
as Palms Muck, Houghton Muck, Muck, or Peat soils)
associated with fen communities on USDA Soil Conservation
Service soil maps. Sizes of algific talus slope
sites were determined through digitization of site
boundaries recorded from field observations on 7.5minute
U.S. Geological Survey topographic maps.
Isolation.—To quantify the isolation of sites, a
weighted summation was calculated for the distance
between each site and all others, assuming that the
influence of surrounding sites exponentially decreases
as intersite distance increases:
n
Ϫcd
I ϭ e͸iϭ1
where I ϭ isolation index, n ϭ number of other sites
in a landscape, d ϭ distance between the reference site
and site i, and c ϭ a constant modifying the rate of
exponential decay
The smaller the value of this index for a site, the
less it is expected to interact with surrounding sites.
This index differs from a summed addition of site distances
as it down-weights the importance of distant
areas to the isolation of a given site. Justification of
this form of weighting is based upon theoretical (Okubo
and Levin 1989) and quantitative (Preston 1962, Nekola
and White, in press) studies that have shown that
dispersal ability and community similarity tend to exhibit
an exponential decay with increasing intersample
distance.
No a priori rules exist to suggest the appropriate
value of c, the constant that modifies the level of downweighting
of distant sites. To allow a robust test of the
effect of isolation on species richness, isolation-index
scores were calculated over five different values of c
(1.0, 0.5, 0.25, 0.1, and 0.05), corresponding to five
different neighborhood radii (roughly equivalent to 3,
6, 12, 30, and 60 km) that demarcate those sites that
most strongly contribute to isolation values. Sites that
fall outside these neighborhoods contribute very little
to the overall isolation-index values for a given loca-
tion.
Statistical analysis
Prediction 1: The species–area relationship will be
steeper and stronger in algific slopes as compared to
fens.—The slope of the vascular plant species–area relationship
was estimated by multiple least-squares linear
regression after both the species-richness and habitat-area
values had been log transformed. A double
log transform was used as these residuals were more
homoscedastic than those generated from untransformed
richness vs. log-transformed area. To allow inclusion
of sites with zero species, a one was added to
the species richness of each site prior to log transformation.
Tests for differences between the slopes and
intercepts of the two species–area relationships were
obtained by analysis of a binary variable representing
habitat type that was added into the model following
methods outlined in Kleinbaum et al. (1988). The form
of this regression model was
ln(species richness ϩ 1)
ϭ ␤0 ϩ ␤1(ln(habitat size)) ϩ ␤2(habitat type)
ϩ ␤3(habitat type ϫ ln(habitat size)) ϩ ␧ (1)
where ␤0 ϭ intercept for algific slopes, ␤1 ϭ slope for
algific slopes, ␤2 ϭ difference in intercept between algific
slopes and fens, and ␤3 ϭ difference in slope
between algific slopes and fens.
As fens can be significantly larger than algific talus
slopes, comparison of species–area slopes and intercepts
calculated from the entire data set may lead to
inaccurate conclusions (Conner and McCoy 1979). To
compensate, the regression analysis was repeated using
only those sites falling within the habitat-size overlap
between algific talus slopes and fens.
The relative strength of the species–area relationship
was determined through comparison of the correlation
coefficients generated through separate regressions of
log-transformed richness vs. log-transformed area for
each habitat. The significance of observed differences
between these r values was quantified using the correlation-coefficient
homogeneity test (Sokal and Rohlf
1981).
Prediction 2: Species–isolation relationships will be
stronger in fens as compared to algific talus slopes.—
The significance and amount of extra variance in vascular-plant
species richness explained by site isolation
was estimated by separately adding each of the five
isolation-index values into the previously defined species–area
model. The form of this regression model,
for each of the isolation-index values, was
ln(species richness ϩ 1) ϭ ␤0 ϩ ␤1(ln(habitat size))
ϩ ␤2(isolation index) ϩ ␧ (2)
where ␤0 ϭ intercept, ␤1 ϭ slope of habitat-size relationship,
and ␤2 ϭ slope of isolation relationship. The
amount of additional variance each of the isolationindex
values explained was determined through subtraction
of r2
values generated from these models from
the r2
values observed from the species-richness vs.
habitat-size regression.
Prediction 3: Distance-decay patterns will be more
pronounced in fens as compared to algific talus
slopes.—Distance-decay rates in the restricted vascular-plant
floras were estimated in both habitats using
the procedures outlined in Nekola and White (in press).
These analyses were conducted at five different sample
grain sizes: single sites, plus grid cells of 10, 20, 40,
and 80 km on a side laid over the study region. However,
80 ϫ 80 km grid cells were not analyzed for algific
talus slopes because of their more limited extent. The
flora of a given cell was determined by summing together
the floras of all sites whose centroids fell within
the boundaries of that cell. The centroid location of
each cell was set as the average location of all sites
2468 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
FIG. 6. Species–area curves (Eq. 1). Log-transformed restricted-species richness as a function of log-transformed habitat
size for northeastern Iowa algific talus slope and fen habitats. Each group of three curves shows the position of best-fit
regression line and 95% confidence interval around the mean of each.
found within that quadrat. Floristic similarity between
samples was calculated using Jaccard’s index, and distance
was calculated as the aerial separation between
cell centroids. The amount of variance in similarity
explained by distance was estimated through a linear
regression of natural-log-transformed similarity vs. untransformed
distance (Nekola and White, in press). The
significance of the similarity–distance relationship was
estimated by Mantel matrix randomization tests (RT
software package, Manley 1992). All Mantel P values
were generated using 10 000 replications, as suggested
by Jackson and Somer (1989) for biological-data sets.
Prediction 4: The proportional frequency of species
with good or poor dispersal abilities will vary between
the algific talus slope and fen flora.—Each restricted
vascular-plant species found in either of the two habitat
types was assigned to one of four dispersal classes
expected to differ in their dispersal abilities: (1) macroscopic
(Ͼ0.1 mm), non-plumose seeds in dry fruits
or capsules; (2) fleshy berries, fruits, or nuts; (3) microscopic
(Ͻ0.01 mm) seeds or spores; and (4) plumose
seeds. The frequency of these four categories in the
flora of each habitat was determined, and differences
tested using the Pearson chi-square statistic.
RESULTS
Prediction 1: The species–area relationship will be
steeper and stronger in algific slopes as compared
to fens
Both habitats demonstrate a significant positive relationship
between species richness and habitat size
(Fig. 6, Table 1), with both slope and intercept for
algific talus slopes being significantly (P Ͻ 0.0005)
larger than those reported for fens. When this analysis
was repeated for only those sites within the range of
habitat-size overlap, the slope and intercept were again
found to be significantly larger in algific slopes (P Ͻ
0.017 and P Ͻ 0.0005, respectively). In both analyses,
small algific talus slopes had fewer vascular-plant species
than fens of identical size, whereas large algific
talus slopes had greater richness than fens of identical
size.
The amount of variance explained in these regressions
also differed greatly (Table 2). Log-transformed
habitat size accounted for 32.6% of observed variance
in algific talus slope species richness but accounted for
only 6.3% of the variance in fen species richness. This
December 1999 2469NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
TABLE 1. Results of multiple-regression analysis of the effect of habitat area and habitat type
on restricted vascular-plant species richness in northeastern Iowa fen and algific talus slope
habitats (see Eq. 1). Restricted species are those present only in fens or algific talus slopes.
Habitat N Value 1 SE t P
Algific talus slope sites 76
Intercept, ␤0
Slope, ␤1
2.335
0.542
0.102
0.090
22.881
6.003
Ͻ0.0005
Ͻ0.0005
Difference between algific slopes
and fen sites 160
Intercept difference, ␤2
Slope difference, ␤3
Ϫ0.476
Ϫ0.428
0.115
0.097
Ϫ4.139
Ϫ4.421
Ͻ0.0005
Ͻ0.0005
Notes: N ϭ no. of observations. There were 160 extant fens and 76 extant algific talus slopes
in the Blanding Formation outcrop region.
TABLE 3. Summary statistics for restricted vascular-plant
species richness vs. area and isolation in northeastern Iowa
fen and algific talus slope habitats.
Exponen-
tial
decay
coefficient
Algific talus slopes
P
Additional
r2
Fens
P
Additional
r2
c ϭ 1.0
c ϭ 0.5
c ϭ 0.25
c ϭ 0.1
c ϭ 0.05
0.150
0.266
0.497
0.802
0.778
0.019
0.012
0.005
0.001
0.001
0.027
0.008
0.003
0.003
0.008
0.029
0.041
0.053
0.050
0.041
TABLE 2. Significance and r2 for the restricted vascular-plant
species richness vs. habitat-area relationship in northeastern
Iowa fen and algific talus slope habitats.
Habitat
No. of
observations P r2
Fens
Algific talus slopes
160
76
0.001
0.000
0.063
0.326
Note: Significance of the correlation-coefficient homogeneity
test: P ϭ 0.006.
difference was found to be highly significant (P ϭ
0.006).
Prediction 2: Species–isolation relationships will be
stronger in fens as compared to algific talus slopes
Isolation was found to be a significant additional
predictor of vascular-plant species richness in fens for
each of the calculated isolation-index values (P ranging
from 0.003–0.027; Table 3). The amount of additional
variance explained by isolation in fen habitats varied
between 0.026 and 0.053, or between 46% and 84% of
the variance explained by log-transformed habitat area.
However, for algific talus slopes, isolation was never
a significant additional predictor of species richness at
any of the isolation-index scores (P ranging between
0.150 and 0.802).
Prediction 3: Distance–decay patterns will be more
pronounced in fens as compared to
algific talus slopes
In fens, significant (P ϭ 0.0001) distance–decay relationships
were observed between the restricted species
floras at all grain sizes (Table 4). At the largest
grain (80 ϫ 80 km regions), this relationship explained
over one third of the variance in regional restricted
vascular-plant species composition (Table 5). At smaller
grains, distance accounted for Ͻ5% of observed variance
in floristic similarity. In algific slopes, only the
distance–decay relationship at the 20 ϫ 20 km sample
grain size was found to be significant at the 0.05 level.
At this scale, 23% of the observed variance in vascularplant
assemblage similarity was accounted for by intersample
distance.
Prediction 4: The proportional frequency of species
with good or poor dispersal abilities will vary
between the algific talus slope and fen flora
The frequencies of dispersal classes differed significantly
(P Ͻ 0.0005) between the two habitats (Table
6). Species with berries, fleshy fruits, or nuts were ϳ15
times more frequent in algific slopes than fens, whereas
species with plumose seeds were ϳ7 times more frequent
in fens than algific slopes.
DISCUSSION
Distinguishing among alternative explanations
In each case the results obtained agree with the a
priori hypotheses, which predicted differences between
the ecological patterns of paleorefugia and neorefugia
and suggest that extinction and immigration have
played differing roles in algific talus slope and fen
communities (Table 7). However, it is impossible to
control for all confounding factors in a natural experiment
such as this. By explicitly stating what some of
these factors may be, the preliminary nature of these
investigations, and potential avenues for future research,
may be made more clear.
Two potential confounders stem from the different
source pools for the flora of these two habitats, and
from the different landscape dynamics of the primary
disturbances that effect them. Although the flora of
each habitat is a subset of the boreal wetland flora, the
species of algific slopes have greater affinity with conifer
swamps while the species of fens have greater
2470 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
TABLE 4. P values for the distance–decay relationships of
the restricted vascular-plant floras of northeastern Iowa fen
and algific talus slope habitats at differing grain sizes.
Habitat
Grain size (km)
80 40 20 10
Individual
sites
Fen
Algific slope
0.0002
···
0.0001
0.6245
0.0001
0.0213
0.0001
0.2832
0.0001
0.2339
Note: P values were calculated using Mantel tests on distance
and similarity matrices for comparisons between each
region or site.
TABLE 5. The r2 values for the distance–decay relationships
of the restricted vascular-plant floras of northeastern Iowa
fen and algific talus slope habitats at differing grain sizes.
Habitat
Grain size (km)
80 40 20 10
Individual
sites
Fen
Algific slope
0.335
···
0.014
0.082
0.044
0.226
0.034
0.005
0.027
0.001
Note: The r2 values were calculated using linear regression
on log-transformed data.
TABLE 6. Frequency of four dispersal classes among restricted
vascular-plant species in northeastern Iowa fen and
algific slopes. The numbers in parentheses represent the
percentage of the total number of species recorded in each
habitat contained in each dispersal class.
Habitat
Dispersal class†
1 2 3 4
Fen
Algific
slope
42 (52.5%)
26 (51.0%)
1 (1.2%)
10 (19.6%)
17 (21.3%)
13 (25.4%)
20 (25.0%)
2 (3.9%)
Note: Chi-square statistic for differences in frequencies for
all dispersal classes ϭ 20.998; P Ͻ 0.0005.
† Dispersal classes: 1 ϭ macroscopic, non-plumose seeds
in dry capules; 2 ϭ berry/fleshy fruits; 3 ϭ spore or microscopic
seeds; and 4 ϭ plumose seeds.
similarity with open peatlands (e.g., Curtis 1959). Differences
in the composition, coexistence patterns, and
dispersal strategies of species in these pools could have
contributed to, or created, observed differences. Additionally,
algific slopes are occasionally impacted by
small-scale talus dislodgement from game trails or individual
tree-falls, whereas fens are more often affected
by larger-scale events such as fires and regional
droughts. Consequently, algific slopes seem more likely
than fens to be in a state of successional quasiequilibrium
(Shugart 1984, Turner et al. 1994). This
difference could potentially account for the stronger
species–area relationship in algific slopes, or for the
predominance of vagile species in fens.
Additional research will be necessary to tease apart
the effects of these and other factors from effects of
colonization history. For example, it would be instructive
to extend these analyses of algific slope and fen
vascular-plant floras to additional taxonomic groups
found on the same sites that possess varying dispersal
abilities, such as bryophytes, birds, lepidoptera, or terrestrial
gastropods. Such analyses would help document
whether the patterns observed for vascular plants
in this investigation are general in nature, and whether
species groups with different dispersal abilities will
respond to identical levels of geographic isolation and
patterns of habitat creation in the same way. It is possible
that poorly dispersing taxa (snails, for example)
may be more influenced by extinction and that morevagile
species (lepidoptera, for example) may be more
influenced by immigration processes on the same sites.
Important additional insights may also be garnered
if paleorefugia and neorefugia are studied over a more
diverse set of environmental situations and landscapes.
For instance, comparing identical habitats that have
formed at different times in different landscapes (creating
a neorefugia in one and a paleorefugia in the
other) may be a good way to control for the confounding
effects of species source pools or disturbance dynamics.
It may also be interesting to study the contrast
between these refugia types across a series of landscapes
in which sites vary from highly to less isolated.
Neorefugia and paleorefugia of very ancient origin may
also demonstrate different biogeographic patterns. For
instance, some ancient oceanic islands represent neorefugia
(like the Hawaiian archipelago, which formed
through volcanic eruptions) whereas others represent
paleorefugia (like New Zealand and New Caledonia,
which were created through fragmentation of continental
land masses). As time has been sufficient for
speciation to occur, a strong species–area relationship
should develop for both. However, the fact that dispersal
was originally more important for ancient oceanic
neorefugia should allow for stronger species–isolation
relationships, and different dispersal-strategy
frequencies in their contemporaneous biota as compared
to ancient oceanic paleorefugia. Hypotheses regarding
paleorefugia and neorefugia are also amenable
to some forms of manipulative experimentation for
those processes that operate over limited spatial and
temporal extents, or to computer simulations for largerscale
patterns. The future research priorities regarding
paleorefugia and neorefugia should include more taxa,
more habitats, more landscapes, and more times of origin.
Through such additional work a clearer understanding
of the role of colonization history on contemporaneous
ecological pattern can be achieved.
Implications for ecology, biogeography, and
conservation
The central question of this paper is whether the
historical contingency of habitat formation can have a
long-lasting impact on community patterns and processes.
The answer to this question has important implications
for ecological debates surrounding equilibrium
vs. non-equilibrium dynamics, vicariance vs.
long-range dispersal, and the delineation of optimum
reserve designs for maintenance of biological diversity.
December 1999 2471NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
TABLE 7. Summary outcomes of analysis: observed and predicted differences between algific talus slopes (paleorefugia)
and fens (neorefugia) in northeastern Iowa.
Factor
Response
Neorefugia Paleorefugia Result†
Relative steepness and strength of species–area relationship Less Greater Y
Strength of species–isolation relationship High Low Y
Change in richness over time Increasing Decreasing ···
Strength of distance-decay relations High Low Y
Dispersal ability of species in habitat Good Poor Y
Relative importance of geography in predicting species occurrence Strong Weak ···
Strength of community–environment correlation Weak Strong ···
Relative number of co-competitors in landscape More Less ···
† Y, relationship found statistically significant and in agreement with a priori predictions; ···, not addressed in this analysis.
Equilibrium vs. non-equilibrium dynamics.—Although
much classical ecological theory (e.g., MacArthur
and Wilson 1967, MacArthur 1972, Tilman
1988) assumes existence of equilibrium conditions, the
data presented above suggest that differences in colonization
history may still imprint biotas 5000–10 000
yr following habitat development. If ‘‘transient dynamics’’
continue to exist after such time periods, equilibrium
conditions cannot be assumed to dominate biogeographic
pattern. These results support paleoecological
work (Davis 1984) that suggests that disequilibrium
between climate and species range may persist
over relatively long temporal extents.
Vicariance vs. long-distance dispersal.—The relative
importance of vicariance vs. long-distance dispersal
in the development of biogeographic pattern has
been debated since the times of Wallace, Schimper,
Engler, and Hooker (Cain 1944). Many ecologists view
dispersal as so efficient that no barriers to movement
exist for any species within continental areas, making
biogeographic pattern simply the result of physiological
limitations and competitive interactions (e.g.,
Krebs 1985). Saur (1988) has summarized this belief
as Beijerinck’s Law: everything is everywhere but the
environment selects. However, some cladistic biogeographers
have stated that dispersal either does not occur
between isolated habitats, or is so unpredictable that it
cannot be studied. For these researchers, only the process
of vicariance can parsimoniously explain disjunct
distributions (e.g., Humphries and Parenti 1986, Brudin
1988).
Analysis of paleorefugia and neorefugia suggest that
such absolute positions may be fundamentally flawed
as both vicariance and dispersal can operate contemporaneously
in the same landscape. Instead of asking
which of these processes is responsible for observed
biogeographic patterns, it seems more productive to
determine how much each has contributed to that pattern.
If additional research confirms generality to the
patterns attributed above, it may be possible to use
similar analyses in additional landscapes to document
the occurrence of neorefugia and paleorefugia, and thus
the frequency of vicariance and long-distance dispersal.
Conservation of biological diversity.—These preliminary
analyses suggest that algific talus slopes are floristically
more self-contained than fens, with immigration
playing a relatively minor role in the determination
of species occurrence. Protection of the vascular flora
of these habitats should thus center on those sites that
harbor maximum diversity and the internal site dynamics
that are responsible for that diversity.
However, these preliminary analyses also suggest
that fens are not as self-contained, with immigration
from surrounding sites strongly influencing vascularplant
composition. Diversity in these communities varies
conversely with isolation, and community composition
is strongly influenced by the proximity of source
areas. To ensure adequate protection of the processes
that gave rise to site diversity in such systems, reserves
may have to be chosen to protect not only internal site
dynamics, but also the immigration potential between
sites. This will be possible only if many sites are protected
across the landscape, including those that may
not currently support diverse floras. Consequently, neorefugia
reserve design may require an integrated, regional
strategy.
Unfortunately, the reality of fen destruction in Iowa
may make such goals unrealistic. Since European settlement,
93% of northeastern Iowa fens have been destroyed
(Nekola 1994; Fig. 5). Even if all remaining
sites could be saved, immigration rates cannot be maintained
at presettlement levels. The ultimate effect of
this will be severely lowered species-immigration rates
compared to presettlement times, and low probabilities
that species will be able to reestablish themselves following
local extinction. If extinction rates are high,
strategies to guard fen biodiversity may require not
only protection of all remaining sites, but also the physical
transport of propagules between sites to reestablish
presettlement immigration levels. If this approach is
eventually deemed necessary, it would require creation
of stochastic immigration models to guide land managers
in their attempts to simulate ‘natural’ immigration
between sites.
ACKNOWLEDGMENTS
This research was supported by a National Science Foundation
Graduate Fellowship and by the McElroy Foundation
of Waterloo, Iowa. I wish to thank John and Gretchen Bray-
2472 Ecology, Vol. 80, No. 8JEFFREY C. NEKOLA
Concepts&Synthesis
ton, Bob Moats, Richard Baker, Willard Hawker, Louise
McEarchen, John Nehnevaj, Dean Roosa, Arlan Kirchman,
Dennis Schlicht, Bill Thomas, Robert Thomson, Dave Wendling,
and Herb Wilson for providing me with locality information
for extant fen sites. Terrence Frest was instrumental
in instructing me in the geology and ecology of algific talus
slopes, and initially located many of the Blanding outcrop
sites inventoried in this analysis. Peter White, Michael Palmer,
Susan Wiser, James Hazen Graves, Robert Howe, Clifford
Kraft, Ed Connor, Paul Keddy, three anonymous reviewers,
and especially Robert K. Peet read and made useful comments
regarding earlier drafts of this manuscript.
LITERATURE CITED
Anderson, W. A. 1943. A fen in northwestern Iowa. American
Midland Naturalist 29:787–791.
Austin, M. P., A. O. Nicholls, and C. R. Margules. 1990.
Measurement of the realized qualitative niche: environmental
niches of five Eucalyptus species. Ecological Monographs
60:161–177.
Bennett, D. G., and O. Gade. 1979. Geographic perspectives
in migration research. Department of Geography Studies
in Geography, Number 12. University of North Carolina,
Chapel Hill, North Carolina, USA.
Bierrgaard, R. O., Jr., T. E. Lovejoy, V. Kapos, A. Augusto
dos Santos, and R. W. Hutchings. 1992. The biological
dynamics of tropical rainforest fragments. BioScience 42:
859–866.
Brown, J. H. 1971. Mammals on mountaintops: non-equilibrium
insular biogeography. American Naturalist 105:
476–478.
. 1995. Macroecology. University of Chicago Press,
Chicago, Illinois, USA.
Brown, J. H., and A. Kodric-Brown. 1977. Turnover rates
in insular biogeography: effect of immigration on extinction.
Ecology 58:445–449.
Brudin, L. Z. 1988. Phylogenetic biogeography. Pages 343–
370 in A. A. Meyers and P. S. Giller, editors. Analytical
biogeography. Chapman & Hall, New York, New York,
USA.
Burrough, P. A. 1986. Principles of geographical information
systems for land resources assessment. Oxford University
Press, New York, New York, USA.
Bush, M. B., and R. J. Whittaker. 1993. Non-equilibrium in
island theory of Krakatau. Journal of Biogeography 20:
453–457.
Cain, S. A. 1944. Foundations of plant geography. Harper
& Row, New York, New York, USA.
Carlquist, S. 1974. Island biology. Columbia University
Press, New York, New York, USA.
Coffin, B., and L. Pfannmuller. 1988. Minnesota’s endangered
flora and fauna. University of Minnesota Press, Minneapolis,
Minnesota, USA.
Conner, E., and E. D. McCoy. 1979. The statistics and biology
of the species–area relationship. American Naturalist
113:791–833.
Cox, G. W., and R. E. Ricklefs. 1977. Species diversity,
ecological release, and community structuring in Caribbean
land bird faunas. Oikos 29:60–66.
Curtis, J. T. 1959. The vegetation of Wisconsin. University
of Wisconsin Press, Madison, Wisconsin, USA.
Davis, M. B. 1984. Quaternary history and the stability of
forest communities. Pages 132–151 in D. C. West, H. H.
Shugart, and D. B. Botkin, editors. Forest succession.
Springer-Verlag, New York, New York, USA.
Diamond, J. M. 1973. Distributional ecology of New Guinea
birds. Science 179:759–769.
. 1975. The island dilemma: lessons of modern biogeographic
studies for the design of nature reserves. Biological
Conservation 7:129–146.
Fernald, M. L. 1925. Persistence of plants in unglaciated
North America. Memoirs of the American Academy of Arts
and Sciences 15:241–342.
Fotherinham, A. S. 1981. Spatial structure and distance-decay
parameters. Annals of the Association of American
Geographers 71:425–436.
Frest, T. J. 1981. Final report, project SE-1-2 (Iowa Pleistocene
Snail). Iowa Conservation Commission, Des
Moines, Iowa, USA.
. 1982. Final report, project SE-1-4 (Iowa Pleistocene
Snail). Iowa Conservation Commission, Des Moines, Iowa,
USA.
. 1983. Final report (contract number 30181-1259)
northern Driftless Area survey. U.S. Fish and Wildlife Service,
Fort Snelling, Minnesota, USA.
. 1984. Final report, project SE-1-6 (Iowa Pleistocene
Snail). Iowa Conservation Commission, Des Moines, Iowa,
USA.
. 1986a. Final report, project SE-1-6 number 2 (Iowa
Pleistocene Snail). Iowa Department of Natural Resources,
Des Moines, Iowa, USA.
. 1986b. Final report, project SE-1-7 (Iowa Pleistocene
Snail). Iowa Department of Natural Resources, Des
Moines, Iowa, USA.
. 1987. Final report, project SE-1-8 (Iowa Pleistocene
Snail). Iowa Department of Natural Resources, Des
Moines, Iowa, USA.
. 1990. Final report (contract number 65-2454). Field
survey of Iowa spring fens. Iowa Department of Natural
Resources, Des Moines, Iowa, USA.
. 1991. Summary status reports on eight species of
candidate land snails from the Driftless Area (Paleozoic
Plateau), Upper Midwest. Final report (contract number
301-01366). U.S. Fish and Wildlife Service, Fort Snelling,
Minnesota, USA.
Frest, T. J., and J. R. Dickson. 1986. Land snails (Pleistocene-recent)
of the Loess Hills: a preliminary survey. Proceedings
of the Iowa Academy of Science 93:139–157.
Hall, S. A. 1971. Paleoecological interpretation of bison,
mollusks, and pollen from the Hughes peat bed, Linn County,
Iowa. Thesis. University of Iowa, Iowa City, Iowa, USA.
Hallberg, G. R., T. E. Fenton, G. A. Miller, and A. J. Lutenegger.
1978. The Iowan Erosion Surface: an old story,
an important lesson, and some new wrinkles. Pages 2.1 to
2.94 in R. Anderson, editor. Geology of East-central Iowa,
42nd annual tri-state geological field conference. Iowa
Geological Survey, Iowa City, Iowa, USA.
Harper, J. L. 1977. Population biology of plants. Academic
Press, New York, New York, USA.
Hedges, J. 1972. Expanded joints and other periglacial phenomena
along the Niagara Escarpment. Biuletyn Peryglacjalny
21:87–126.
Hilborn, R., and M. Mangel. 1997. The ecological detective.
Princeton Monographs in Population Biology number 28.
Princeton University Press, Princeton, New Jersey, USA.
Howe, R. W., M. J. Huston, W. P. Pusateri, R. H. Laushman,
and W. E. Schennum. 1984. An inventory of significant
natural areas in Iowa. Iowa Conservation Commission, Des
Moines, Iowa, USA.
Humphries, C. J., and L. R. Parenti. 1986. Cladistic biogeography.
Clarendon Press, Oxford, UK.
Jackson D. A., and K. M. Somer. 1989. Are probability
estimates from the permutation model of Mantel’s test stable?
Canadian Journal of Zoology 67:766–769.
Kleinbaum, D. G., L. L. Kupper, and K. E. Muller. 1988.
Applied regression analysis and other multivariate methods.
PWS-Kent Publishing, Boston, Massachusetts, USA.
Krebs, C. J. 1985. Ecology. Harper & Row, New York, New
York, USA.
Krukeberg, A. R., and D. Rabinowitz. 1985. Biological as-
December 1999 2473NEOREFUGIA AND PALEOREFUGIA
Concepts&Synthesis
pects of endemism in higher plants. Annual Review of
Ecology and Systematics 16:447–479.
Manley, B. F. J. 1992. RT—a program for randomization
testing. Version 1.02. The Centre for Applications of Statistics
and Mathematics, University of Otago, Dunedin,
New Zealand.
MacArthur, R. H. 1972. Geographical ecology. Harper &
Row, New York, New York, USA.
MacArthur, R. H., and E. O. Wilson. 1967. The theory of
island biogeography. Princeton University Press, Princeton,
New Jersey, USA.
McDonald, K. A., and J. H. Brown. 1992. Using montane
mammals to model extinctions due to global change. Conservation
Biology 6:409–415.
Moran, R. C. 1981. Prairie fens in northeastern Illinois: floristic
composition and disturbance. Pages 164–168 in R.
Stuckey and K. J. Reece, editors. Proceedings of the sixth
North American Prairie Conference. The Ohio State University,
Columbus, Ohio, USA.
Nekola, J. C. 1990. Rare Iowa plant notes from the R. V.
Drexler Herbarium. Proceedings of the Iowa Academy of
Science 97:55–73.
. 1994. The environment and flora of northeastern
Iowa fen communities. Rhodora 96:121–169.
Nekola, J. C., and P. S. White. In press.The distance decay
of similarity in ecology and evolution. Journal of Bioge-
ography.
Ode, D. J. 1985. Rediscovered plants of fen habitats of South
Dakota. Proceedings of the South Dakota Academy of Science
64:217.
Okubo, A., and S. A. Levin. 1989. A theoretical framework
for data analysis of wind dispersal of seeds and pollen.
Ecology 70:329–338.
Peet, R. K., and N. L. Christensen. 1988. Changes in species
diversity during secondary forest succession on the North
Carolina Piedmont. Pages 233–245 in H. J. During, M. J.
A. Werger, and J. H. Willems, editors. Diversity and pattern
in plant communities. SPB Academic Publishing, The
Hague, The Netherlands.
Pielou, E. C. 1979. Biogeography. John Wiley & Sons, New
York, New York, USA.
Preston, F. W. 1962. The canonical distribution of commonness
and rarity: part II. Ecology 43:410–432.
Prior, J. C. 1991. Landforms of Iowa. University of Iowa
Press, Iowa City, Iowa, USA.
Roughgarden, J., S. D. Gaines, and S. W. Pacala. 1987. Supply-side
ecology: the role of physical transport processes.
Symposium of the British Ecological Society 27:491–518.
Saur, J. D. 1988. Plant migration. University of California
Press, Berkeley, California, USA.
Shmida, A., and S. Ellner. 1984. Coexistence of plant species
with similar niches. Vegetatio 58:29–55.
Shugart, H. 1984. A theory of forest dynamics. SpringerVerlag,
New York, New York, USA.
Simberloff, D. S., and E. O. Wilson. 1969. Experimental
zoogeography of islands: a model for insular conservation.
Ecology 50:296–314.
Sokal, R. R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman,
San Francisco, California, USA.
Stebbins, G. L., and J. Major. 1965. Endemism and speciation
in the California flora. Ecological Monographs 35:1–
35.
Thompson, C. A., and E. A. Bettis III. 1994. Age and development
of Iowa fens. Proceedings of the Iowa Academy
of Science 101:73–77.
Thorne, R. F. 1964. Relict nature of the flora of White Pine
Hollow Forest Reserve, Dubuque County, Iowa. State University
of Iowa Studies in Natural History 20:1–33.
Tilman, D. 1988. Plant strategies and the dynamics and structure
of plant communities. Princeton University Press,
Princeton, New Jersey, USA.
Tobler, W. R. 1970. A computer movie simulating urban
growth in the Detroit region. Economic Geography 46:234–
240.
Turner, M. G., W. H. Romme, and R. H. Gardner. 1994.
Landscape disturbance models and the long-term dynamics
of natural areas. Natural Areas Journal 14:3–11.
Van Zant, K. L., and G. R. Hallberg. 1976. A late-glacial
pollen sequence from northeastern Iowa: Sumner Bog revisited.
Iowa Geological Survey Technical Informational
Bulletin Series number 3.
Zobel, M. 1997. The relative role of species pools in determining
plant species richness: an alternative explanation
of species coexistence. Trends in Ecology & Evolution 12:
266–269.