Clifford E. Kraft

PREDICTION OF LONG‐DISTANCE DISPERSAL USING GRAVITY MODELS: ZEBRA MUSSEL INVASION OF INLAND LAKES

Gravity models are commonly used by geographers to predict migration and interaction between populations and regions. Even though rarely used by ecologists, gravity models allow estimation of long-distance dispersal between discrete points in heterogeneous landscapes. We developed a production-constrained gravity model to forecast zebra mussel (Dreissena polymorpha) dispersal into inland lakes of Illinois, Indiana, Michigan, and Wisconsin (USA) based on the site and location of lakes and the number and location of boats within 364 counties. A deterministic form of this model was used to estimate best-fit parameters for distance coefficient, Great Lakes boat-ramp attractiveness, and colonization cutoff threshold. A stochastic model thus developed from these parameters allows for random changes in colonization likelihood. The results of our model are highly correlated with the actual pattern of colonized lakes in southern Michigan and southeastern Wisconsin at the end of 1997. Areas of central Wisconsin and...

Resilience and Resistance of a Lake Phosphorus Cycle Before and After Food Web Manipulation

Parameters of a phosphorus cycling model were estimated for two configurations of a lake ecosystem. The piscivore-dominated configuration had one more trophic level than the planktivore-dominated configuration. We derived four main conclusions from analysis of the model. (1) Results support the argument of DeAngelis et al. that turnover rate of a limiting nutrient is directly related to ecosystem resilience. (2) Results support the hypothesis of Pimm and Lawton that longer food chains are less resilient. (3) Inputs of phosphorus to the pelagic system derived from inshore feeding by fishes were a large flux, which is comparable to inputs from physical-chemical fluxes. (4) Algal (seston) standing crops, unlike all other compartments, were less sensitive to phosphorus inputs in the piscivore-dominated system. Consistent with the trophic cascade hypothesis, the piscivore-dominated system had higher herbivore standing crops and lower algal standing crops than the planktivore-dominated system. Changes in trophic structure that derive from trophic cascades can be viewed as changes in the phosphorus cycle driven by fishes.

MATURE AND OLD-GROWTH RIPARIAN FORESTS: STRUCTURE, DYNAMICS, AND EFFECTS ON ADIRONDACK STREAM HABITATS

Riparian forests regulate linkages between terrestrial and aquatic ecosystems, yet relationships among riparian forest development, stand structure, and stream habitats are poorly understood in many temperate deciduous forest systems. Our research has (1) described structural attributes associated with old-growth riparian forests and (2) assessed linkages between these characteristics and in-stream habitat structure. The 19 study sites were located along predominantly first- and second-order streams in northern hardwood-conifer forests in the Adirondack Mountains of New York (U.S.A.). Sites were classified as mature forest (6 sites), mature with remnant old-growth trees (3 sites), and old-growth (10 sites). Forest-structure attributes were measured over stream channels and at varying distances from each bank. In-stream habitat features such as large woody debris (LWD), pools, and boulders were measured in each stream reach. Forest structure was examined in relation to stand age using multivariate techniques, ANOVA, and linear regression. We investigated linkages between forest structure and stream characteristics using similar methods, preceded by information-theoretic modeling (AIC). Old-growth riparian forest structure is more complex than that found in mature forests and exhibits significantly greater accumulations of aboveground tree biomass, both living and dead. In-stream LWD volumes were significantly (alpha = 0.05) greater at old-growth sites (200 m3/ha) compared to mature sites (34 m3/ha) and were strongly related to the basal area of adjacent forests. In-stream large-log densities correlated strongly with debris-dam densities. AIC models that included large-log density, debris-dam density, boulder density, and bankfull width had the most support for predicting pool density. There were higher proportions of LWD-formed pools relative to boulder-formed pools at old-growth sites as compared to mature sites. Old-growth riparian forests provide in-stream habitat features that have not been widely recognized in eastern North America, representing a potential benefit from late-successional riparian forest management and conservation. Riparian management practices (including buffer delineation and restorative silvicultural approaches) that emphasize development and maintenance of late-successional characteristics are recommended where the associated in-stream effects are desired.

When can efforts to control nuisance and invasive species backfire

Population control through harvest has the potential to reduce the abundance of nuisance and invasive species. However, demographic structure and density-dependent processes can confound removal efforts and lead to undesirable consequences, such as overcompensation (an increase in abundance in response to harvest) and instability (population cycling or chaos). Recent empirical studies have demonstrated the potential for increased mortality (such as that caused by harvest) to lead to overcompensation and instability in plant, insect, and fish populations. We developed a general population model with juvenile and adult stages to help determine the conditions under which control harvest efforts can produce unintended outcomes. Analytical and simulation analyses of the model demonstrated that the potential for overcompensation as a result of harvest was significant for species with high fecundity, even when annual stage-specific survivorship values were fairly low. Population instability as a result of harvest occurred less frequently and was only possible with harvest strategies that targeted adults when both fecundity and adult survivorship were high. We considered these results in conjunction with current literature on nuisance and invasive species to propose general guidelines for assessing the risks associated with control harvest based on life history characteristics of target populations. Our results suggest that species with high per capita fecundity (over discrete breeding periods), short juvenile stages, and fairly constant survivorship rates are most likely to respond undesirably to harvest. It is difficult to determine the extent to which overcompensation and instability could occur during real-world removal efforts, and more empirical removal studies should be undertaken to evaluate population-level responses to control harvests. Nevertheless, our results identify key issues that have been seldom acknowledged and are potentially generic across taxa.

Stream Fish Assemblages in Relation to Landscape Position and Local Habitat Variables

Abstract The relative influence of local habitat variables and stream network position on fish assemblages was evaluated in this study of first-order through third-order streams within the Beaverkill–Willowemoc watershed in New York. We compared fish distribution and abundance over local and landscape scales by surveying 69 randomly selected tributaries within this 775-km2 watershed. We used watershed-level metrics of stream link magnitude, branch link, confluence link, downstream link, and stream order to evaluate the importance of stream network position upon fish assemblages. Results of canonical correspondence analysis indicated that six factors significantly influenced fish species abundance in our study watershed. The proportion of fine substrate, canopy cover, in-stream vegetation, and water temperature were the four local habitat factors related to the abundance of fish species in this watershed; confluence link and stream order were the stream network position measures with the greatest influence...

Physical influences of landscape on a large-extent ecological disturbance: the northeastern North American ice storm of 1998

The 1998 ice storm was a large-extent ecological disturbance that severely affected the eastern Adirondack forests of northern New York. Ice damage produced widespread breakage of limbs and trunks in susceptible trees. Although ice storms are common within northeastern North American forests, the magnitude and extent of the 1998 storm far exceeded damage caused by typical ice storms in the recent past. While plot and stand-scale ecological impacts of ice storms have received attention insofar as tree species vulnerability, stand age susceptibility, and microhabitat alterations, larger-extent damage patterns have not been previously evaluated. The normalized difference vegetation index (NDVI) was employed to assess forest vigor and canopy density in atmospherically corrected Landsat Thematic Mapper (TM) satellite imagery of the Adirondacks. Digital change analysis of the baseline forest condition (1990 NDVI data), and the condition encountered in a post-storm image (1998 NDVI data) was conducted. Forest damage was separated from natural variations in canopy reflectance by employing a generalized linear model that incorporated in situ measurements. A robust empirical variogram analysis revealed that locations of tree damage were significantly correlated for distances up to 300 meters. Intensely-damaged forest exhibited greater spatial dependence, but over a smaller distance. Canopy damage was not greater proximate to stream and forest boundaries, and did not follow our hypothesis of decreasing damage with distance from the boundary. Overall, we show that local topography (elevation and aspect), forest composition (deciduous or coniferous), and the meteorological characteristics of the disturbance event acted together to determine the spatial extent of ice storm damage.

Landscape patterns of an aquatic invader: Assessing dispersal extent from spatial distributions

Assessing the spatial distribution of organisms across landscapes is a key step toward determining processes that produce observed patterns. The spatial distribution of an invasive aquatic mollusk, the zebra mussel (Dreissena polymorpha), was examined in two lake-rich areas (Belarus and midwestern United States) with contrasting invasion histories. Spatial distribution patterns of invaded lakes were determined using Ripleys K. Aggregation of invaded lakes was found at similar spatial extents ( 120 km in Belarus. The observed spatial extent of aggregation likely reflected the scale of secondary geographic spread, whereas the scale of long-distance dispersal events was reflected by the spatial extent of segregation. Isolated Belarus lakes were less likely to be invaded than those connected by waterways. Although one-dimensional aggregation of invaded lakes along connected Belarus waterways was not observed, nearest neighbor analysis indicated that zebra mussel dispersal occurred at distances <15 km within these waterways. Based on observed spatial pattern, we concluded that zebra mussels have not yet saturated European and North American lake landscapes, including many suitable lakes. Similar distribution patterns of invaded lakes in Belarus and North America suggest that similar processes have influenced zebra mussel spread in both landscapes.

Development of spatial pattern in large woody debris and debris dams in streams

The spatial distribution of large woody debris (LWD) in streams was evaluated using Neighbor K statistics, following extensive wood deposition from an ice storm in the eastern Adirondack Mountains (New York). Two years after wood deposition, we surveyed individual pieces of LWD in one stream and surveyed debris dam locations in eight streams within the ice storm area. To examine the linear pattern of debris dams within a stream, we used a one-dimensional version of Ripleys K, a second-order statistic that evaluates the spatial pattern of points within a landscape. Both aggregated and segregated (regularly spaced) distributions of wood were identified. Individual pieces of LWD were aggregated at spatial extents ranging from 0 to 40 m and were segregated at spatial extents ranging from 80 to 100 m. In two streams, we found that debris dams were segregated at distances ranging from 100 to 300 m relative to randomly chosen locations, but debris dams showed no significant spatial pattern in six other study streams. Previous studies of wood distribution in streams have not observed segregated distribution patterns. Spatial segregation of debris dams in the study area likely occurred in response to regularly spaced stream features or processes that allow movement of individual pieces of LWD toward more stable accumulation points. Neighbor K statistics can be used to identify and describe spatial pattern in large woody debris, and such patterns can be used to help evaluate and identify processes responsible for their generation.

Overcompensatory response of a smallmouth bass (Micropterus dolomieu) population to harvest: release from competition?

An intensive seven-year removal of adult, juvenile, and young-of-the-year smallmouth bass (Micropterus dolo- mieu) from a north temperate lake (Little Moose Lake, New York, USA) resulted in an increase in overall population abundance, primarily due to increased abundance of immature individuals. We developed a density-dependent, stage-struc- tured model to examine conditions under which population control through harvest could result in the increase of a tar- geted species. Parameter values were derived from a 54-year data set collected from another north temperate lake (Lake Opeongo, Ontario, Canada) smallmouth bass population. Sensitivity analyses identified the demographic conditions that could lead to increased abundance in response to harvest. An increase in population abundance with harvest was most likely to occur when either (i) per capita recruitment at low levels of spawner abundance was large, juvenile survivorship was high, and maturation of age-4 and older juveniles was moderately high or (ii) per capita recruitment at low levels of spawner abundance was slightly lower, yet the maturation rate of age-3 juveniles and adult survivorship were high. Our modeling results together with empirical evidence further demonstrate the importance of overcompensation as a substantial factor to consider in efforts to regulate population abundance through harvest.

Partitioning of food resources by sculpins in Lake Michigan

SynopsisResource partitioning was studied in two benthic Lake Michigan fishes, the deepwater sculpin,Myoxocephalus thompsoni, and the slimy sculpin,Cottus cognatus, that exhibit nearly disjunct distributions along a hypolimnetic depth gradient. Fish were collected in an area of sympatry over two 24 h periods. These sculpins exhibited food segregation—slimy sculpins ate primarilyPontoporeia affinis, deepwater sculpins ate bothP. affinis andMysis relicta — and their depth segregation was associated with the reported abundance of these prey. Different feeding behaviors may be responsible for this association, since slimy sculpins consumed intermediate size ranges ofPontoporeia and deepwater sculpins consumedPontoporeia in a pattern similar to the ambient size distribution. Neither fish showed a daily activity cycle, and both appear food-limited.