Resisting-Accepting-Directing: Ecosystem Management Guided by an Ecological Resilience Assessment.

As anthropogenic influences push ecosystems past tipping points and into new regimes, complex management decisions are complicated by rapid ecosystem changes that may be difficult to reverse. For managers who grapple with how to manage ecosystems under novel conditions and heightened uncertainty, advancing our understanding of regime shifts is paramount. As part of an ecological resilience assessment, researchers and managers have collaborated to identify alternate regimes and build an understanding of the thresholds and factors that govern regime shifts in the Upper Mississippi River System. To describe the management implications of our assessment, we integrate our findings with the recently developed resist-accept-direct (RAD) framework that explicitly acknowledges ecosystem regime change and outlines management approaches of resisting change, accepting change, or directing change. More specifically, we developed guidance for using knowledge of desirability of current conditions, distance to thresholds, and general resilience (that is, an ecosystem's capacity to cope with uncertain disturbances) to navigate the RAD framework. We applied this guidance to outline strategies that resist, accept, or direct change in the context of management of aquatic vegetation, floodplain vegetation, and fish communities across nearly 2000 river kilometers. We provide a case study for how knowledge of ecological dynamics can aid in assessing which management approach(es) are likely to be most ecologically feasible in a changing world. Continued learning from management decisions will be critical to advance our understanding of how ecosystems respond and inform the management of ecosystems for desirable and resilient outcomes.

Conceptualizing alternate regimes in a large floodplain-river ecosystem: Water clarity, invasive fish, and floodplain vegetation.

Regime shifts - persistent changes in the structure and function of an ecosystem - are well-documented for some ecosystems and have informed research and management of these ecosystems. In floodplain-river ecosystems, there is growing interest from restoration practitioners in ecological resilience, yet regime shifts remain poorly understood in these ecosystems. To understand how regime shifts may apply to floodplain-river ecosystems, we synthesize our understanding of ecosystem dynamics using an alternate regimes conceptual framework. We present three plausible sets of alternate regimes relevant to natural resource management interests within the Upper Mississippi River and Illinois River. These alternate regimes include: 1) a clear water and abundant vegetation regime vs. a turbid water and sparse vegetation regime in lentic, off-channel areas, 2) a diverse native fish community regime vs. an invasive-dominated fish community regime, and 3) a regime characterized by a diverse and dynamic mosaic of floodplain vegetation types vs. one characterized as a persistent invasive wet meadow monoculture. For each set of potential alternate regimes, we review available literature to synthesize known or hypothesized feedback mechanisms that reinforce regimes, controlling variables that drive regime transitions, and current restoration pathways. Our conceptual models provide preliminary support for the existence of alternate regimes in floodplain-river ecosystems. Quantitatively testing hypotheses contained within the conceptual model are important next steps in evaluating the model. Ultimately, the synthesis and evaluation of alternate regimes can inform the utility of resilience concepts in restoration and management on the Upper Mississippi River and Illinois River and improve our understanding of ecosystem dynamics in other large, heavily managed floodplain-river ecosystems.

Modelling moose-forest interactions under different predation scenarios at Isle Royale National Park, USA.

Loss of top predators may contribute to high ungulate population densities and chronic over-browsing of forest ecosystems. However, spatial and temporal variability in the strength of interactions between predators and ungulates occurs over scales that are much shorter than the scales over which forest communities change, making it difficult to characterize trophic cascades in forest ecosystems. We applied the LANDIS-II forest succession model and a recently developed ungulate browsing extension to model how the moose population could interact with the forest ecosystem of Isle Royale National Park, USA, under three different wolf predation scenarios. We contrasted a 100-yr future without wolves (no predation) with two predation scenarios (weak, long-term average predation rates and strong, higher than average rates). Increasing predation rates led to lower peak moose population densities, lower biomass removal rates, and higher estimates of forage availability and landscape carrying capacity, especially during the first 40 yr of simulations. Thereafter, moose population density was similar for all predation scenarios, but available forage biomass and the carrying capacity of the landscape continued to diverge among predation scenarios. Changes in total aboveground live biomass and species composition were most pronounced in the no predation and weak predation scenarios. Consistent with smaller-scale studies, high browsing rates led to reductions in the biomass of heavily browsed Populus tremuloides, Betula papyrifera, and Abies balsamea, and increases in the biomass of unbrowsed Picea glauca and Picea mariana, especially after the simulation year 2050, when existing boreal hardwood stands at Isle Royale are projected to senesce. As a consequence, lower predation rates corresponded with a landscape that progressively shifted toward dominance by Picea glauca and Picea mariana, and lacking available forage biomass. Consistencies with previously documented small-scale successional shifts, and population estimates and trends that approximate those from this and other boreal forests that support moose provide some confidence that these dynamics represent a trophic cascade and therefore provide an important baseline against which to evaluate long-term and large-scale effects of alternative predator management strategies on ungulate populations and forest succession.

Changes in aquatic vegetation and floodplain land cover in the Upper Mississippi and Illinois rivers (1989-2000-2010).

Quantifying changes in the cover of river-floodplain systems can provide important insights into the processes that structure these landscapes as well as the potential consequences to the ecosystem services they provide. We examined net changes in 13 different aquatic and floodplain land cover classes using photo interpreted maps of the navigable portions of the Upper Mississippi River (UMR, above the confluence with the Ohio River) and Illinois River from 1989 to 2000 and from 2000 to 2010. We detected net decreases in vegetated aquatic area in nearly all river reaches from 1989 to 2000. The only river reaches that experienced a subsequent recovery of vegetated aquatic area from 2000 to 2010 were located in the northern portion of the UMR (above navigation pool 14) and two reaches in the Illinois River. Changes on the floodplain were dominated by urban development, which increased in nearly every river reach studied from 1989 to 2000. Agricultural lands declined in most river reaches from 2000 to 2010. The loss of agricultural land cover in the northern UMR was accompanied by increases in forest cover, whereas in the lower UMR and Illinois River, declines in agriculture were accompanied by increases in forest and shallow marsh communities. The changes in aquatic vegetation occupied between 5 and 20% of the total aquatic area and are likely associated with previously reported regional improvements in water clarity, while smaller (1-15% of the total floodplain area) changes in anthropogenic land cover types on the floodplain are likely driven by broad-scale socio-economic conditions.

Effects of flood frequency and duration on the allometry of community-level stem size-density distributions in a floodplain forest.

PREMISE OF THE STUDY: Consistent with the self-thinning law of plant population ecology, Niklas et al. in 2003 proposed that stem size-density distributions (SDDs) of multispecies forest communities should change in very predictable ways as a function of the effects of past disturbances on average tree size. To date, empirical tests of this hypothesis have not been pursued in floodplain settings. METHODS: SDDs were constructed using tree stem-size and density data from forest plots positioned along a flood frequency and duration gradient in the Upper Mississippi River floodplain. KEY RESULTS: As flooding (both frequency and duration) increased, several small tree species were eliminated from forest plots and the persistent species increased in their size. Consistent with the predictions of Niklas et al., these changes corresponded with overall decreases in stem density, increases in average stem size, and reductions in both the Y-intercept and slope terms of the community-level SDDs. CONCLUSIONS: This study adds to a growing list of examples suggesting that theories related to forest community composition and biomass must account for both the broader effects of disturbances as well as the underlying biochemical processes that regulate plant growth. Further study is needed to fully address the role different disturbance frequencies play in determining plant density, diversity, average size, and associated size\frequency distributions.