Limited impacts of extensive human land use on dominance, specialization, and biotic homogenization in boreal plant communities

Global loss of biodiversity is a major concern, and human land use is among its chief
causes 1]. At local scales, species loss is less frequently observed, and the spread of weedy
or exotic species often instead results in increases in richness 2]. Focusing on increased richness alone can however give the misleading impression
that land use has negligible or positive effects on biodiversity, masking declines
in species at larger scales and oversimplifying “biodiversity” to mere species counts,
ignoring the composition of species and broader ecology of communities.

Communities may change in important ways with little impact on richness or diversity
3]-5]. In some communities with expanded human land use, generalists may replace specialists,
dominance by fewer species may increase, and communities may become more functionally
homogeneous 6]-9]. For instance, in metacommunity simulations comparing indices of biodiversity, specialists
declined rapidly with disturbance intensity while species richness and diversity were
far less sensitive to changes in disturbance 3].

We therefore explore two sets of complementary elements of community structure to
test for impacts of anthropogenic disturbance. First, we explore distributions of
occupancy (via ranked species occupancy curves), which indicate the relative occurrence
of species across communities. Some species are present in many sites, while others
occupy few sites. The distribution of species by occupancy is an important pattern
in community ecology that, it has been suggested, has predictable shapes for disturbed
and undisturbed communities 10]. Second, we investigate patterns of species on the specialist-generalist continuum,
and biotic homogenization (the similarity of species across communities). A core ecological
concept is that species vary not only in the conditions in which they exist, their
niche, but in the range of conditions over which they exist, the breadth of their
niche 11]. Theory suggests that disturbance poses greater threats, including extinction risk,
to species with narrower niches (specialists) 12]-14], and empirical evidence generally supports this expectation 3],7],15]-17]. As a result, individual local communities tend to lose specialist species and retain
or gain generalist species, leaving communities of more similar species composition
following disturbance 7].

Motivation to better understand these aspects of biodiversity change with land use
is three-fold. One, we seek to identify the ecological risks facing communities. Specialists,
for example, play important roles in ecosystems and functional homogenization can
reduce ecosystem function and resilience or resistance to further disturbance 7]. Low relative occupancy may contribute to extinction risk. Two, we seek to assess
the potential utility of community metrics for ecological monitoring applications.
If widespread, these patterns may serve as powerful local signatures of human land
use on biodiversity not evident from species richness patterns, with utility as community-level
metrics for ecological monitoring. And three, we test the generality of conceptual
ideas of biodiversity change driven by disturbance. Disturbance, anthropogenic or
natural, may have general predictable outcomes for ecological communities.

We investigate these patterns in a region unlike those in which these patterns are
typically investigated: the boreal vascular plant communities of northern Alberta,
Canada. In this large region, rapid expansion of human footprint, ongoing regional
land use planning and heightened interest in broad scale environmental monitoring
have deepened the importance of effective evaluation of how communities change with
human land use at a regional scale. Also, spatial heterogeneity and species richness
are both relatively low, and relative to previous studies of biotic homogenization,
there are no major elevation or climate gradients.

Occupancy

Among the most fundamental characteristics of communities is the relative abundance
or occupancy among species. Akin to species abundance distributions, ranked species
occupancy curves (RSOC)s provide detailed descriptions of the prevalence and rarity
of species where abundance data are lacking 10]. RSOCs also retain species identity information and avoid arbitrary frequency binning
that is characteristic of occupancy frequency distributions 10].

Jenkins 10] built on the intermediate disturbance hypothesis 18] and studies of species occupancy 19]-22] to develop hypotheses for RSOC shapes, suggesting those shapes should vary along
a successional gradient, and as a result, on a spatial gradient in disturbance. He
suggested that with high disturbance, recruitment limitation should cause an exponential
RSOC where some disturbance-adapted species are prevalent but most others rare. At
intermediate disturbance, Jenkins predicted a sigmoidal (S-shaped) RSOC, where moderate
regional recruitment limitation is accompanied by moderate local niche-based filtering.
In this intermediate scenario, many species are expected to be widespread. With little
disturbance, competitive species are expected to be dominant and widespread, while
other species exhibit low occupancy. By contrast, with high disturbance, ruderal species
adapted to quick regrowth and dispersal are expected to be widespread, with few late
seral specialists. We previously demonstrated that communities in the study region
were richest in species at intermediate disturbance 23], so in the current study we adopt Jenkins’ 10] predictions for sigmoidal RSOCs among intermediately disturbed communities, and exponential
RSOCs among either low or high disturbed communities. If RSOC shapes are predictive
of disturbance, they may be useful indicators of human impacts for management and
conservation purposes.

Specialism and homogenization

The shapes of RSOC’s described above (with some species widespread and others rarely
occurring) imply that species vary in the range of conditions under which they may
occur. That is, they imply that some species are specialists, and others generalists.
Many studies have suggested that human disturbance and invasive species are more likely
to negatively affect specialist than generalist species 6],7],9],15],24], including plants 25]-27]. Hanski 28] warned that the boreal faces an “imminent wave of extinctions of specialist forest
species”. Indeed, specialist species have declined globally at a greater pace than
other species 7]. ‘Weedy’, ‘invasive’, ‘generalist’ species are expanding in some areas such that
even with declines in specialists, richness may be maintained, if not greater, than
in the past 6]. Further, success of introduction and establishment by exotic species is strongly
related to generalism 29].

These observations are consistent with theory predicting that generalists should be
favored over specialists if human land use acts as disturbance to disrupt spatial
homogeneity and/or temporal stability. Niche evolution theory predicts that with less
heterogeneity across space, and more stable conditions over time, specialists should
be favored 12]-14],30],31]. In contrast, with greater spatial and temporal variation, generalists are thought
to benefit. Marvier et al. 12] showed this expectation holds despite lower competitive abilities of generalists
in any given environment than specialists under the same conditions. These theoretical
expectations are explained in part because specialists are thought to be more temporally
variable in abundance due to the changing environment 32] and variability in abundance can contribute to extinction 33].

Disturbance can thus be expected to act as a filter for specialist rather than generalist
species. However, this hypothesis contrasts expectations from the intermediate disturbance
hypothesis, for which generalists are expected to be most prevalent at intermediate,
rather than at high disturbance. We predicted that either a) with increasing disturbance,
specialism of species would decline assuming that disturbance filters species in favour
of those more successful, for example, in higher light conditions and drier soils;
or b) that the heterogeneity of disturbance itself influences composition, such specialism
should be higher at both low and high disturbance, but generalism higher at intermediate
disturbance.

Specialization is usually estimated by either i) laboratory based experimentation
on species responses under ranging environmental conditions, or ii) field based observation
of niche characteristics and generation of habitat suitability models 7]. Delineating species as specialist, generalist, or in between is in practice challenging
with field data because it involves defining the niche of each species in an unknown
number of dimensions 34]. Although these dimensions can sometimes be reliably estimated, we instead employ
a recently developed index of specialization that bypasses the need to determine niche
width directly, and instead uses the diversity of co-occurring species as an indicator
of habitat specialization 34]. Specialists are expected to co-occur with similar sets of species wherever they
occur, whereas generalists are expected to co-occur with a diverse array of other
species. Fridley et al.’s 34] specialization index was reviewed in 7],35] and is related to that of 36]. The index is further explained under Methods. At the community level, we used 34]’s related index as an indicator of functional biotic homogenization, the decline
in functional diversity that results from the replacement of specialists by generalists.
We answer Clavel et al. 7]’s call that these be used to assess human disturbance on ecological communities.