Large wood (also known in the literature as large woody debris or LWD) has long been recognized as having an important influence on physical and ecological processes in stream channels and watersheds of all orders and gradients [1
]. Large wood (LW) acts as an important structural element in aquatic habitats that supports high levels of abundance and diversity of fish and invertebrates by providing cover and increasing habitat complexity [2
]. LW can also provide a strong physical linkage between terrestrial and aquatic ecosystems [9
]. The presence of LW also allows for the accumulation and storing of coarse organic matter, which serves as a vital ecological energy base and enhances biogeochemical cycling of nutrients and other constituents in watershed ecosystems [3
The presence of LW in stream channels influences pool formation, frequency, and morphology; increases flow resistance, sediment storage and sorting, and longitudinal variation of channel depth and width; and promotes bank stabilization and the creation of bars [3
]. Furthermore, by directing flow into the streambed, LW can increase hyporheic exchange, which is known to enhance water quality [15
]. When clustered into jams, LW influences streamflow and channel pattern more than individual wood pieces [17
]. For instance, jams can initiate lateral channel migration by increasing the elevation of the streambed and water surface. Additionally, the presence and frequency of jams is thought to be an indicator of stream wood transport capacity [3
With respect to recruitment, LW in low-gradient watersheds enters stream channels primarily through bank erosion and blow down of whole trees or branches and is cycled over time throughout the channel, banks, and floodplain [6
]. Subsequent to their formation in the channel, LW accumulations or jams can be deposited on the floodplain through flooding events or abandoned through lateral channel migration; when floodwaters or the meandering channel eventually return (LW jams themselves promoting both floodplain inundation and lateral channel migration), remnants can re-enter a stream and contribute to new LW accumulations [6
]. LW remnants can also remain situated on the floodplain for decades and provide habitat for a variety of terrestrial organisms [6
]. Those remnants remaining stable over long time periods may represent a sizable carbon reservoir [4
] and aid in replenishing supplies of new key members (larger pieces of LW which serve as an impetus for the accumulation of LW in stream channels and the formation of jams, and subsequently as a jam anchor) by protecting developing riparian forests from erosion long enough for trees to grow large [6
As the degree of development/urbanization of many watersheds has increased over time, the potential for LW generation has been reduced due to widespread deforestation and direct channel modifications [3
]. In many cases LW and LW jams are directly removed from the stream channel to improve navigation, protect bridges and other infrastructure, recover marketable timber, and eliminate perceived barriers to fish migration [6
]. Often the production and recruitment of LW is also restricted by stream management projects aimed at reducing channel migration (e.g., riprap, gabions, stream bank armoring, etc.). As a consequence of these LW-limiting practices, habitat simplification and reduced stream ecological function can become rampant [4
The composition and age of watershed forests also stands to affect LW recruitment potential. Comparisons of contemporary forest composition with lake sediment pollen analysis (as a proxy for historic forest composition) show that human disturbance of forests has resulted in a decrease in abundance of larger late-successional species and an increase in abundance of smaller early-successional species [23
]. Such comparisons indicate that, despite regional natural reforestation in some areas, forests are not yet returning to their pre-settlement composition, and that contemporary forests are often a fragmented patchwork of young regrowth stands [23
]. This type of forest composition in riparian zones severely limits the production and recruitment into stream channels of LW, especially key members. Such impacts on the age and composition of watershed forests therefore can be considered a legacy effect which compounds that of contemporary land use/urbanization in limiting potential for LW generation. Indeed, even when riparian forests are restored, full recovery of processes that allow for sufficient LW recruitment can lag forest regeneration by centuries [7
Although the impact of urbanization, land use, and forest composition on LW recruitment and LW sizes has been noted in the literature [1
], few studies document the character and extent of these impacts at the watershed scale. Fewer still document such impacts on watersheds in the eastern United States where urbanization and land use in general is relatively intensive and spatially pervasive, has a longer history, and has broadly shifted over the course of the past century from almost completely agricultural to a mix of urban, suburban, and agricultural. It stands to reason that the limiting of LW is a major way such patterns of spatially, temporally, and categorically heterogeneous land use—which is ubiquitous across not just the eastern U.S. but much of the developed world—contribute to the degradation of water resources and aquatic ecosystems. Therefore, in order to combat the problem, it is important to assess and document the phenomenon spatially on the broad scale at which it exists. Active and passive efforts toward restoring LW sources informed by such knowledge could represent a significant opportunity for effective restoration of watershed function and ecosystems in regions that have experienced complex, heterogeneous land use change.
With this study we aim to demonstrate how this could be done by using (Geographic Information System) GIS as a novel approach to determine the current distribution of functional LW recruitment zones—areas of forested land cover within 30 m of a stream channel [27
]—in the Niagara Subbasin, a 2051 km2
catchment comprising six watersheds located in western New York, a region dominated by the complex, heterogeneous patterns of land use change described hereinabove. In addition, in order to show the extent that LW is limited by urbanization and land use impacts in the study area, in-stream LW loads will be approximated based on field surveys of LW pieces and LW accumulations or “jams” and compared to similar surveys in other regions. The implications of our findings for stream restoration and watershed management in regions with heterogeneous land use will be discussed.
The Ellicott Creek and Murder Creek watersheds where in-stream LW surveys were conducted are in many ways representative of others in the study area (Table 1
) and present lower LW and LW jam densities than other land use-impacted watersheds where similar studies have been conducted. For instance, LW densities in New England streams have been found to be 17.4 pieces per 100 m and LW jam frequency to be 4.4 per 100 m on average [3
]. Similarly, streams in West Virginia and Pennsylvania have been found to average of 16.1 pieces of LW per 100 m [25
]. Moreover, in the western U.S. where deforestation has been less rampant and some old growth forests remain (e.g., Alaska, Washington, Montana, Oregon, California), densities of in-stream LW are frequently reported to be greater than 30 pieces per 100 m [25
], roughly three times those found in this study.
In addition to being tied to impacts of urbanization/land use, it is likely the low density of LW in the study watersheds is also tied to forest composition. LW recruitment may lag forest regeneration by centuries [7
], and the presence of large mature trees in LW recruitment zones to provide key members is critical for in-stream LW accumulation and especially LW jam development [6
]. According to forest composition surveys conducted in the region, forests in the study watersheds are composed of a fragmented patchwork of young regrowth stands [23
]. Indeed, field observations confirmed the virtual absence of large mature trees in the LW recruitment zones of the study watersheds, suggesting that forest age and composition also play a role in the low LW and LW jam density observed. Further, the sizes of in-stream LW measured at Site 1 at Ellicott Creek and Site 2 at Murder Creek are consistent with a scarcity of key members and LW size distribution being dominated by pieces too small to form into large jams that span the width of the channel (15-25 m) (Figure 2
). Ultimately, the slow regeneration of forests observed in the region [23
] and the lag known to exist between forest regeneration and LW recruitment [7
] suggests that without intervention the streams of these watersheds are likely to lack LW and suffer consequential impacts to habitat, water quality, and physical function long into the future.
In terms of distribution of potentially functional LW recruitment zones (forested areas within 30 m of the stream channel [27
]) among the study watersheds, only between 7% and 28% of LW recruitment zones are forested (Table 3
). Moreover, only two watersheds—Upper Tonawanda Creek and Murder Creek—have a “Good” Aquatic Habitat Viability Rating [28
]. This suggests that for much of the area studied watersheds have been severely impacted and that potentially productive LW recruitment zones are lacking.
Urban and suburban land uses are known to often result in the degradation of riparian areas as in many instances roads, parking lots, buildings, lawns, etc. encroach toward streams [34
]. Accordingly, it may be expected that urbanization—high-density development as well as low-density suburban development—is partly responsible for the observed patterns as the most urbanized watersheds in the study area tend to have low potential for LW recruitment (Table 1
and Table 3
, Figure 3
and Figure 4
). However, correlation analysis indicates that although levels of development, population density, and percent impervious surface are indeed negatively correlated to total area of forested LW recruitment zone, the correlation is not statistically significant. Instead, forested LW recruitment zone area is only significantly correlated to total forest land cover. Although this positive correlation is unsurprising since forested areas are needed for LW generation, the correlations seen illustrate the complex relationships between urbanization/land use/land cover and LW recruitment. Indeed, these findings suggest, perhaps unexpectedly, that extent of urbanization alone is not a best predictor of functional LW recruitment zones in watersheds with heterogeneous land use/land cover. Likewise, albeit also not statistically significant, the positive correlation seen between the percentage of land used for agriculture and occurrence of forested LW recruitment zones in the study watersheds further stresses the complex relationship between LW recruitment and land use. Indeed, unlike other regions where agriculture is more intensive, industrial, and monocultural, most agricultural uses in the study area are relatively small farm operations with varied practices (poultry, legumes, dairy cows, corn, etc.) where wooded lots and wooded riparian areas can often be found.
In terms of Aquatic Habitat Viability, urbanization seems to have the most drastic effect as the three watersheds with the highest degrees of such development have the lowest Aquatic Habitat Viability Ratings (Table 1
and Table 2
). On the opposite end of the spectrum, the two watersheds with the highest Aquatic Habitat Viability Rating in the study area (Good) have a relatively high percentage of land used for agriculture (>50%) and relatively high forest cover (25–34%) but low urbanization (4–8%) (Table 1
and Table 2
). Although agricultural land use has been tied in many cases to watershed degradation owing to, among other things, excessive nutrient (nitrogen, phosphorus) and pesticide losses to streams [36
], our results suggest that, in contrast to urban/suburban land use, agricultural land use as it is in the study area may not necessarily negatively impact LW recruitment nor Aquatic Habitat Viability Rating.
Although the occurrence of forested LW recruitment zones is greatly limited in the Niagara Subbasin, there are significant areas with a high density of forested and therefore potentially functional LW recruitment zones (Figure 5
). From a management standpoint these areas may be where land conservation in which development, timber harvesting/clearing (especially extraction of key members), agricultural uses, etc. are prohibited or severely limited through zoning, easements, acquisitions, etc. would be the most suitable approach for LW generation enhancement. However, in most of the study area, especially in the most urbanized watersheds, there is little that land conservation practices can do to restore LW production and recruitment because there are very few forested LW recruitment zones to protect (Figure 5
). Instead, at these locations, which are generally in the western part of the study area as seen in Figure 4
and Figure 5
, active stream restoration practices such as the addition of root wads, cross vanes, j-hooks, engineered log jams, floodplain reconnection/bank re-grading, and/or side channel addition may be required. Some of the known effects of these restoration practices include enhanced habitat for fish and macro-invertebrates through the creation of eddies, pools, and enhanced hyporheic flow; improvement in stream water quality with the creation of denitrification hot spots; and stream bank erosion reduction [38
]. All these effects are known benefits of natural LW and LW jams [4
Through a detailed investigation of the occurrence of forested LW recruitment zones and in-stream LW density in a series of watersheds with varying degrees of urbanization and types of land use, this study further constrains the relationship between urbanization/land use/land cover and LW recruitment and in-stream loads. Although the results linked high population density, elevated percentage of impervious surface, and high degree of urbanization to a lack of potentially functional recruitment zones for LW, they also reveal that agricultural land use in the study area is positively correlated to the occurrence of forested riparian areas at the watershed scale. In addition, although LW recruitment zones are generally considered to be restricted to areas within 30 m of stream channels [27
], total forested land cover appears to be highly correlated to forested LW recruitment zones at the watershed scale. However, in the study area, like in many regions with significant urbanization and spatially, temporally, and categorically heterogeneous land use, forests are often fragmented and young which strongly limits LW recruitment (especially that of key members) and, as a result, in-stream LW loads.
Using a combination of in-stream surveys and GIS analysis, the hybrid methodology of this study offers a way in which LW may be evaluated over larger areas toward informing watershed management, regional land use planning, and land conservation. With the exception of select areas with the highest density of forested LW recruitment zones where land conservation is susceptible to being a successful approach to enhance the number and size of in-stream LW, for the long run, in places where forested LW recruitment zones are scarce, active LW restoration through the addition of root wads, cross vanes, j-hooks, engineered log jams, and/or floodplain reconnection/bank re-grading may be more suitable than conservation for LW management and achieve many of the benefits (increased habitat, enhanced water quality, stream bank stabilization) normally provided by LW and LW jams.